* Set a maximum number of messages which may be indexed before we force

[citadel.git] / citadel / ChangeLog

Sat Oct 22 22:55:49 EDT 2005 ajc
* Set a maximum number of messages which may be indexed before we force
  a cache flush to disk.  Currently 2500.

Sat Oct 22 00:48:47 EDT 2005 ajc
* msgbase.c: allow multipart/alternative part-swapping only at the top
  level, because if we recursed into another one then it's going to stomp
  all over our preferred part.  (even more uuuuunnnhhhhh...)

Sat Oct 22 00:46:52 EDT 2005 ajc
* mime_parser.c: recurse into message/rfc822 parts as if they were
  multipart, because we may need to extract attachments from the embedded
  submessage, etc.  (uuuuunnnhhhh...)

Fri Oct 21 15:12:45 EDT 2005 ajc
* MSG4 (and CtdlOutputMsg() as well) now accepts an optional MIME part
  specifier, allowing the client to fetch an encapsulated message
  attached as message/rfc822 instead of the top-level message.

Thu Oct 20 17:55:12 EDT 2005 ajc
* ft_wordbreaker.c: added a list of "noise words" to ignore.  This is
  admittedly EN/US specific, so if anyone wants to contribute noise words
  for other languages...

Wed Oct 19 22:55:19 EDT 2005 ajc
* serv_calendar.c: registered a fixed output hook for text/calendar.

Wed Oct 19 13:30:16 EDT 2005 ajc
* New type of server hook: CtdlRegisterFixedOutputHook().  This is for
  extending the fixed_output() function for arbitrary new MIME types.  The
  usefulness of this for end users is limited, since no clients use MSG0
  anymore.  The real purpose of converting various MIME types to text is to
  make them visible to the full text indexer.
* serv_vcard.c: registered a fixed output hook for text/x-vcard.

Tue Oct 18 22:46:41 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* msgbase.c, messages.c, html.c: conversion of HTML to plain text now accepts
  a maximum source length.  MSG0 output of multipart messages was running
  right past the end of the HTML and into, for example, a subsequent base64
  encoded attachment, which was then output -- or worse, indexed.
* Since everyone's fulltext indices probably contain a lot of base64 junk, the
  FT_WORDBREAKER_ID has been bumped.  This will force an index rebuild on any
  sites that are upgraded.

Tue Oct 18 13:11:38 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* serv_fulltext.c: index records are now cached in memory.  This
  significantly speeds up indexing of an existing message base.  We
  flush the cache to disk when finished.

Mon Oct 17 22:10:57 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Easy Install now uses db-4.3.29

Mon Oct 17 10:59:55 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* THIS IS 6.60

Sun Oct 16 00:23:19 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* msgbase.c: alias() now handles addresses such as 
  Display Name <user@host.org>
  ...when user@host.org resolves to a user on the local Citadel server
  or network, without trying to loop out and back in again via SMTP.

Sun Oct 16 00:11:16 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Handle email addresses with commas inside quotes, such as
  "Cancro, Art" <ajc@uncensored.citadel.org>
  ...without thinking that the comma is a separator between addresses

Thu Oct 13 00:02:23 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* citadel.c: don't crash when <.R>ead <U>serlist results in an empty
  list.  The ClientIPC API returns NULL if the search returned no results,
  which we have to handle.  This resolves Bugzilla issue #154.

Wed Oct 12 23:30:18 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* newinstall.sh: once again, prefer wget over curl.  At least one user
  reported a problem where the -O option to curl was not available.  Also
  changed -O to --remote-name in the hope that this works better.

Wed Oct 12 17:32:05 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* imap_search.c: when fulltext index is enabled, avoid doing a slow search
  on each message when BODY is requested.  Messages are already qualified
  by the indexer.

Wed Oct 12 10:30:22 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* html.c: added support for some additional character entity references.

Tue Oct 11 23:13:53 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Default for autologin is now DISABLED!  If you are upgrading an existing
  site and are using this function, you must --enable-autologin
* newinstall.sh: --enable-autologin if upgrading an existing installation
  that we think is set to autologin.  It checks for chkpwd to find out.

Tue Oct 11 12:55:43 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* The "set the flags" portion of IMAP COPY now sets the flags of all messages
  in bulk, instead of one message at a time.  Big performance boost.

Tue Oct 11 01:09:59 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Fixed a problem with replication.  I believe we've got it now but it needs
  another round of testing.

Tue Oct 11 00:45:02 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Replication checks and EUID indexing are now only enabled for rooms whose
  default view is set to a groupware type of room.  This speeds up the saving
  and moving of messages for message and mail rooms.

Mon Oct 10 00:22:49 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* IMAP STORE now calls CtdlSetSeen() with an entire list of message numbers.

Sun Oct  9 22:37:44 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
 * CtdlSetSeen() now accepts a list of message numbers instead of just one.

Sat Oct  8 18:40:50 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Replaced the various "socket is broken" messages with the message
  "Client disconnected: ending session."  The previous messages were concerning
  some site operators that there might have been something wrong with
  the system.  (Bugzilla #164)

Fri Oct  7 23:07:38 EDT 2005 Art Cancro <ajc@uncensored.citadel.org>
* Makefile.in: change "CVS" references to ".svn" to avoid errors
  during install.

Revision 655.25  2005/10/06 19:47:00  ajc
* Change to EUID command syntax: returned msg num is now guaranteed to
  exist.

Revision 655.24  2005/10/06 19:16:31  ajc
* Added the EUID command to search for a message by EUID

Revision 655.23  2005/10/06 17:14:41  ajc
* newinstall.sh: prefer curl over wget

Revision 655.22  2005/10/06 04:09:19  ajc
* THE DREADED AUTO-PURGER now purges euid index records which point to
  messages that no longer exist.

Revision 655.21  2005/10/06 03:36:05  ajc
* Changed the format of the euidindex record to contain the record's key.
  This will allow us to auto-purge stale records later.

Revision 655.20  2005/10/04 16:38:17  ajc
* CtdlOutputPreLoadedMsg() calling syntax has changed.  It no longer needs
  the message number, because it is being supplied a preloaded message.
* msgbase.c: fixed a problem where HEADERS_NONE mode was broken when
  outputting a message in RFC822 format.  This was breaking IMAP commands
  such as xx FETCH nn BODY[TEXT]

Revision 655.19  2005/10/02 04:40:58  ajc
* The EUID index is now built, and replication checks are being performed
  using it.  It is much faster now because we don't have to scan the entire
  room anymore.  We still need to do two things:
  1. Write a server command to fetch messages by EUID instead of msgnum
  2. Find a way to purge stale EUID index records.

Revision 655.18  2005/10/01 05:18:57  ajc
* Began writing code to index messages by euid per room

Revision 655.17  2005/09/27 04:18:45  ajc
* Auto-add *recipient* addresses to Contacts.  This is done asynchronously
  because we do have to scan the address book to make sure we don't
  already have the address recorded.

Revision 655.16  2005/09/26 21:46:08  ajc
* Attempt to save *outgoing* email addresses to the address book.

Revision 655.15  2005/09/21 20:07:18  ajc
* Set the To: field for digests as well

Revision 655.14  2005/09/21 16:56:17  ajc
* The "To: line of a mailing list message is now set to the address of the
  list instead of the address of the recipient.

Revision 655.13  2005/09/21 13:21:31  ajc
* Don't harvest incoming addresses.  A few hours of operation of this in a
  production environment proved that it's stupid.  This feature has been
  disabled; we will harvest outgoing addresses instead.

Revision 655.12  2005/09/21 04:27:34  ajc
* When reading messages in MT_MIME mode from a Citadel client, don't
  list MIME parts that are part of a multipart/alternative group.

Revision 655.11  2005/09/18 21:51:44  ajc
* File the auto-generated vCards into the users' address books.

Revision 655.10  2005/09/18 20:33:13  ajc
* Now harvesting addresses, converting them to vCards, and storing them in
  the Aide> room.  All that's left to do now is file the messages in the
  appropriate users' address books.

Revision 655.9  2005/09/18 19:34:26  ajc
* When submitting a message, harvest non-local addresses for potential
  inclusion in a user's Collected Addresses book.  Note: we don't actually
  do anything with these addresses yet.  That comes next.

Revision 655.8  2005/09/18 17:50:05  ajc
* serv_network.c: use a stat() call to determine the mtime of spoolin, and
  skip the scan if it hasn't been touched since the last time we looked.
* serv_network.c: don't create network/systems/ directory.  We haven't used
  that in ages.
* serv_network.c: only attempt to create directories at startup, not at
  every queue run.  Also, chown() them to the citadel user.

Revision 655.7  2005/09/16 20:40:44  ajc
* CC: support for message creation, and IMAP.  Not tested.

Revision 655.6  2005/09/16 20:21:38  ajc
* CC: and BCC: delivery are working (tested using message submittal
  from WebCit) -- still missing the insertion of Y (CC) header field, and
  the handling of CC in IMAP.

Revision 655.5  2005/09/16 04:23:21  ajc
* Cc: and Bcc: support.  Not finished yet.

Revision 655.4  2005/09/15 21:37:06  ajc
* Restructured cmd_auto() to be able to search vCards in more than one room.
  For now we are using the Global Address Book as the second room, but in
  production that might be a bit too slow.

Revision 655.3  2005/09/15 18:36:29  ajc
* serv_smtp.c: Allow the use of *any* RFC822-compliant address format,
  including Name <user@node>, or user@node (Name), by stripping down the
  address before doing RCPT To: in the SMTP client.

Revision 655.2  2005/09/15 03:31:09  ajc
* cmd_auto() (used by WebCit address autocompletion) now queries the
  fn, n, and all email fields in the vCards in user's Contacts room.

Revision 655.1  2005/09/14 03:48:32  ajc
* Bumped internal version number to 6.56
* Checked in an initial but incomplete version of the AUTO command
  (to be used for address autocompletion)

Revision 655.0  2005/09/13 14:00:12  ajc
* THIS IS 6.55

Revision 654.27  2005/09/13 03:56:37  ajc
* Don't crash when user hits <.A>ide <P>ost

Revision 654.26  2005/09/10 18:46:48  ajc
* Functions such as imap_fetch_internaldate() now return quietly with no
  output, instead of crashing, if passed a NULL message pointer.

Revision 654.25  2005/09/09 19:44:21  ajc
* New bmstrcasestr() function to perform very fast case-insensitive
  substring searches using the Boyer-Moore algorithm.  Like its predecessor,
  it is based on the one written by Urs Jannsen; unlike its predecessor, it
  actually works.

Revision 654.24  2005/09/08 03:25:09  ajc
* messages.c: break up long lines (preferably by substituting spaces with
  newlines) to avoid sending messages to the server containing lines with
  more than 1024 characters, to avoid having the server truncate those
  lines.

Revision 654.23  2005/09/07 03:08:06  ajc
* When rejecting a message due to RBL, give the alleged spammers the
  relevant TXT record if there is one.  Hopefully.  This needs testing.

Revision 654.22  2005/09/02 03:50:42  ajc
* serv_network.c: fixed a bug in digest generation that was causing the
  tops of some messages to be cut off.

Revision 654.21  2005/09/01 22:07:08  ajc
* citserver.c: change to usage of strdup() to work on RH9 where it's
  a macro

Revision 654.20  2005/08/29 20:49:50  ajc
* imap_fetch.c: fixed a bug in the IMAP FETCH BODY code that was causing the
  "most recently fetched message" cache to be burned even when it shouldn't
  have been.  This was causing abominally slow message load time when a message
  contains attachments and the MUA is a client such as Thunderbird that does
  partial fetches.

Revision 654.19  2005/08/23 04:00:01  ajc
* Mailing list messages are now customized with a To: header for each
  recipient.  This uses more overhead but makes delivery more reliable.
  We also prepend [List name] to the subject.
* The undocumented client-side mailing list participation option received
  a similar update.  We now set an explicit To: header in there as well,
  because some lists demand it.

Revision 654.18  2005/08/21 19:40:50  ajc
* techdoc/protocol.txt: document that unless otherwise specified,
  everything in the Citadel system is UTF-8.

Revision 654.17  2005/08/21 06:16:37  ajc
* html.c: don't transform <32 or >126 characters to question marks.  We
  like foreign character sets now.

Revision 654.16  2005/08/15 16:36:46  ajc
* Added a "MIME-Version:" header in a couple of places

Revision 654.15  2005/08/12 18:00:29  ajc
* Bounds checking in CtdlDirectoryLookup()

Revision 654.14  2005/08/12 15:18:27  ajc
* updated the roadmap

Revision 654.13  2005/08/12 02:12:31  ajc
* Added more room name macros

Revision 654.12  2005/08/11 23:12:26  ajc
* debian/citadel-server.postinst: remove reference to '?' file
  (patch sent in by W. Goesgens)

Revision 654.11  2005/08/11 13:55:26  ajc
* Attempt to fix broken build of our replacement strcpy() on gcc2.96 by
  issuing an #undef strcpy (it's a macro in that environment)

Revision 654.10  2005/08/10 21:47:41  ajc
* Added "instant expunge" and "allow spoofing" site config options.

Revision 654.9  2005/08/10 21:10:54  ajc
* moved "instant expunge" to imap_do_store() so that it gets called from
  both STORE and UID STORE commands.

Revision 654.8  2005/08/10 15:39:35  ajc
* Experimental "instant expunge" option.  For now, you have to put
  -DINSTANT_EXPUNGE into the CFLAGS line of your Makefile.  If we decide to
  keep this it will become a site config option.

Revision 654.7  2005/08/10 02:44:56  ajc
* Applied changes to debian files sent in by WG

Revision 654.6  2005/08/09 21:58:57  ajc
* Removed the ./help/? hack; replaced with actual "show directory" code
  in the server.

Revision 654.5  2005/08/09 16:28:32  ajc
* setup.c: added "exim4" to the list of non-Citadel MTA's which can be
  disabled during setup.

Revision 654.4  2005/08/09 14:13:36  ajc
* newinstall.sh: hunt for both 'make' and 'gmake', but this time make sure
  that the make in question is actually GNU Make, regardless of its name

Revision 654.3  2005/08/08 14:57:19  ajc
* Added a new "with header summary" mode to the MSGS command.

Revision 654.2  2005/08/08 14:28:41  ajc
* CtdlForEachMessage() -- avoid scanning v_seen sequence set for each message
  during a MSGS_ALL fetch, because the results don't matter.

Revision 654.1  2005/08/05 21:31:01  ajc
* Any "delete message" operation which is synchronous to a client is now
  deferred.  This is accomplished by copying the message pointer to
  the __CitadelDeletedMessages__ room, which keeps the reference count at
  least 1.  THE DREADED AUTO-PURGER can sweep it up later.

Revision 654.0  2005/08/05 16:22:38  ajc
* THIS IS 6.54

Revision 653.18  2005/08/04 04:46:42  ajc
* Bumped internal version number to 6.54
* removed some tracing messages

Revision 653.17  2005/08/04 04:22:09  ajc
* imap_fetch.c: removed a spurious ')' character from the output
  of the 'charset' field.

Revision 653.16  2005/08/02 03:40:15  ajc
* Removed instances of hard-coded CTDLDIR; this breaks -h

Revision 653.15  2005/08/02 03:05:31  ajc
* Added in Wilfried Goesgens' debian/ build directory.

Revision 653.14  2005/08/02 02:49:12  ajc
* Applied Wilfried Goesgens' dirconfig patch for more install targets

Revision 653.13  2005/08/01 20:45:26  ajc
* citadel_ipc.c: RUN_DIR not RUNDIR in line 2944

Revision 653.12  2005/07/29 03:50:52  ajc
* Temporarily disabling network_purge_spoolout() because it may be
  removing files it shouldn't.

Revision 653.11  2005/07/29 01:50:58  ajc
* Create network/ directory hierarchy if it doesn't exist

Revision 653.10  2005/07/28 03:22:19  ajc
* Added a "macintosh readme" written by Mathew McBride

Revision 653.9  2005/07/26 02:49:25  ajc
* Applied a patch sent in by Wilfried Goesgens which allows the various
  program and data directories to be set to any location on the host
  system.  This will allow packagers to do FSSTND-type configurations.

Revision 653.8  2005/07/25 17:37:36  ajc
* citadel_ipc.c: when performing a MSG4 command, don't return the
  charset as part of the content-type string.

Revision 653.7  2005/07/21 17:20:55  ajc
* <.R>ead <U>ser-list now takes advantage of the server-side string match
  if available.  It still filters on the client side as well, in case the
  server is older and sent back the whole list.

Revision 653.6  2005/07/21 15:02:27  ajc
* Server-side LIST command now accepts a search string.

Revision 653.5  2005/07/19 20:04:31  ajc
* MSG4 command now outputs content type *and* charset

Revision 653.4  2005/07/19 17:30:13  ajc
* Fixed a bug in the MIME parser that was causing it to prematurely go
  out of scope when binary parts are included.  (8-bit MIME is not yet
  explicitly supported in the Citadel server, but WebCit uses this to
  handle data coming in via the POST method.)

Revision 653.3  2005/07/19 14:21:51  ajc
* mime_parser.c: don't default to Content-type: text/plain; charset=us-ascii
  because there are places where those fields actually need to be empty (in
  WebCit, actually, but I don't want to fork the mime parser).

Revision 653.2  2005/07/19 04:10:01  ajc
* Updated the MIME parser API to include the "charset" portion of
  the content type.

Revision 653.1  2005/07/09 11:30:18  ajc
* Removed trailing space at the end of the list of messages returned by
  an IMAP SEARCH command.

Revision 653.0  2005/07/06 21:41:52  ajc
* THIS IS 6.53

Revision 652.1  2005/07/06 21:41:17  ajc
* Fixed bug #149 (incorrect sequence set optimization, resulting in
  messages being marked as read/unread incorrectly during IMAP sessions)

Revision 652.0  2005/07/06 02:28:38  ajc
* THIS IS 6.52

Revision 651.8  2005/07/06 02:28:20  ajc
* citadel.lsm: removed.  Nobody uses the Linux Software Map anymore.
* Updated internal version number to 6.52

Revision 651.7  2005/07/06 01:53:34  ajc
* imap_fetch.c: additional self-check to avoid attempting to fetch messages
  with UID's lower than 1.  Hopefully this fixes bug #150.

Revision 651.6  2005/07/01 22:07:27  ajc
* CtdlSetSeen() -- when new vset overflows its size, trim it in such a way
  so that it doesn't mark the oldest messages as new.

Revision 651.5  2005/06/28 02:58:19  ajc
* docs update

Revision 651.4  2005/06/26 22:19:20  ajc
* auth.c: applied fleeb's patch to validpw() to clean up FreeBSD compatibility

Revision 651.3  2005/06/22 03:45:17  ajc
* Documented the auto-log-cull as it relates to backup strategies

Revision 651.2  2005/06/22 03:03:34  ajc
* Automatic deletion of committed database logs is now a site-definable
  setting.

Revision 651.1  2005/06/16 02:42:58  ajc
* There is now a dedicated thread for doing database checkpoints.

Revision 651.0  2005/06/12 03:46:30  ajc
* THIS IS 6.51

Revision 647.42  2005/06/12 03:31:33  ajc
* ChangeLog: restored automatic prepending of CVS commit log messages
  to the top of this file.  Somehow it got lost somewhere around 647.20

Revision 647.41  2005/06/12 03:30:20  ajc
* Test

revision 647.40
date: 2005/06/12 03:24:15;  author: ajc;  state: Exp;  lines: +0 -1
* When delivering list digests, put the name of the room in [brackets]
  in the subject line.  This makes lots of other software happy.  (We
  need to do this for non-digest list subscribers too.)

revision 647.39
date: 2005/06/12 01:15:33;  author: ajc;  state: Exp;  lines: +1 -0
* Provide separate filtered and unfiltered LMTP sockets.

revision 647.38
date: 2005/06/09 20:10:02;  author: ajc;  state: Exp;  lines: +0 -1
* Easy Install requires gmake

revision 647.37
date: 2005/06/09 03:35:58;  author: ajc;  state: Exp;  lines: +1 -0
* Allow IMAP DELETE of a zapped/forgotten/unsubscribed room

revision 647.36
date: 2005/06/09 03:20:21;  author: ajc;  state: Exp;  lines: +0 -1
* Do not log IMAP/POP/SMTP password commands

revision 647.35
date: 2005/06/07 21:45:14;  author: ajc;  state: Exp;  lines: +1 -0
* bounce messages need subjects

revision 647.34
date: 2005/06/06 23:50:01;  author: ajc;  state: Exp;  lines: +0 -1
* Made some changes to the calendar/uuid logic to fix a bug that caused
  duplicate entries for an event to appear in some situations

revision 647.33
date: 2005/06/03 22:26:03;  author: ajc;  state: Exp;  lines: +1 -0
* When saving a vCard to a dedicated contacts room, always set the subject
  to the name in the vCard.

revision 647.32
date: 2005/06/03 22:22:36;  author: ajc;  state: Exp;  lines: +0 -1
* Removed old Aethera hacks

revision 647.31
date: 2005/06/03 04:01:27;  author: ajc;  state: Exp;  lines: +1 -0
* Documentation update

revision 647.30
date: 2005/06/02 19:32:30;  author: ajc;  state: Exp;  lines: +0 -1
* Set the correct flags in the target room after an IMAP COPY command.

revision 647.29
date: 2005/06/02 16:09:32;  author: ajc;  state: Exp;  lines: +1 -0
* tools.c: generated uuid's no longer contain "{" and "}" characters.

revision 647.28
date: 2005/06/02 03:39:44;  author: ajc;  state: Exp;  lines: +0 -1
* Do not turn the initial thread into a worker thread after initialization.
  Its stack size is too small, which could cause crashes.

revision 647.27
date: 2005/06/01 22:32:57;  author: ajc;  state: Exp;  lines: +1 -0
* Implemented a workaround for the IMAP "expungebob bug."  Specifically, since
  we don't store the \Deleted flag persistently (and instead auto-expunge
  folders when they are de-selected) we were not advertising \Deleted as
  a PERMANENTFLAGS flag.  This was causing some clients (particularly
  Thunderbird) to misbehave -- they were simply electing not to transmit the
  flag at all.  As a workaround, \Deleted is now advertised as a
  PERMANENTFLAGS flag, even though it technically isn't.

revision 647.26
date: 2005/06/01 18:31:50;  author: ajc;  state: Exp;  lines: +0 -1
* serv_vandelay.c: updated the export format to include some of the config
  items we missed.

revision 647.25
date: 2005/06/01 05:23:26;  author: ajc;  state: Exp;  lines: +1 -0
* The full text indexer now runs in its own dedicated thread instead of
  in the housekeeping thread.  The main indexer loop now also has the ability
  to save its place and bail out early when it discovers that the server is
  trying to shut down.  The main server loop will pthread_join() the indexer
  thread and patiently wait for it to complete before exiting.  These changes
  all put together mean that citserver will not hang when it is terminated
  during an indexing operation.

revision 647.24
date: 2005/05/27 23:46:57;  author: ajc;  state: Exp;  lines: +0 -1
* Removed OpenLDAP from Easy Install, because it was just too problematic.

revision 647.23
date: 2005/05/26 04:25:29;  author: ajc;  state: Exp;  lines: +1 -0
* Applied xmlns and etag patches sent in by Johannes Schneider that improve
  GroupDAV support.
* Located and fixed a MIME Content-type bug that I accidentally created
  while removing a temporary hack that was in place during the last KDE beta.

revision 647.22
date: 2005/05/23 19:33:56;  author: ajc;  state: Exp;  lines: +0 -1
* database_sleepycat.c: cdb_truncate() no longer encapsulated in a
  transaction.  Truncating a database in Citadel is always synchronous,
  and with big tables (such as when the full text indexer is switched off
  or reinitialized) it was running out of memory.

revision 647.21
date: 2005/05/23 19:31:52;  author: ajc;  state: Exp;  lines: +5 -1
* test

 Revision 647.20  2005/05/23 19:26:04  ajc
 * Move the location of the "enable full text index" configuration item to
   a new location.  Reusing an old location was a stupid idea because old
   clients can inadvertently switch it on.

 Revision 647.19  2005/05/23 14:07:39  ajc
 * serv_imap.c: improve SELECT time by fetching the msglist directly out of
   the database instead of doing a CtdlForEachMessage() loop.

 Revision 647.18  2005/05/22 16:12:25  ajc
 * Full text indexer is now switchable on/off

 Revision 647.17  2005/05/20 20:02:50  ajc
 * The IGnet map is now rewritten to disk only when it changes.
 * When processing inbound network spool, ignore "." and ".." instead of
   failing on them.

 Revision 647.16  2005/05/20 16:22:33  ajc
 Fixed some errors in the ChangeLog

 Revision 647.15  2005/05/20 16:14:43  ajc
 * Dramatically improved the time it takes to goto (or select) a room which
   contains a very long and complex seen/unseen list.  Our test folder,
   containing 359 new of 3162 messages, formerly took 22 seconds to select;
   now it takes 1 to 2 seconds.

 Revision 647.14  2005/05/20 02:37:17  ajc
 * Performance-optimized the full text indexer.

 Revision 647.13  2005/05/20 01:20:24  ajc
 * Cull logs immediately after a successful db checkpoint instead of only
   once every 24 hours.  During big db write operations (such as building
   the full text index) they were just piling up too much.

 Revision 647.12  2005/05/19 21:10:03  ajc
 * Altered the full text indexer to output messages as text before running
   through the wordbreaker.  This prevents the inclusion of encoded base64
   strings in the index, and also allows legitimate text encoded inside
   base64 to be decoded and then indexed.

 Revision 647.11  2005/05/19 03:42:29  ajc
 * Bound the full text index to IMAP search

 Revision 647.10  2005/05/18 22:09:01  ajc
 * Finished the indexer *and* deindexer!  The search API is now working
   flawlessly too.  Now all we have to do is glue it to IMAP and other
   user-facing functionality.  (The SRCH command is for testing only.)

 Revision 647.9  2005/05/18 04:02:54  ajc
 * Completed the "search for all of these words" functionality.  All we need
   to do now is generalize its calling syntax so it can be called from the
   IMAP service.

 Revision 647.8  2005/05/18 03:22:27  ajc
 * Finished the indexer and the first part of the search function...

 Revision 647.7  2005/05/17 20:36:48  ajc
 * Indexer is completed; also began work on the search function itself.
   Still need to add de-indexing so deleted messages are removed from index.

 Revision 647.6  2005/05/17 16:25:23  ajc
 * Completed the wordbreaker for the full text indexer.

 Revision 647.5  2005/05/17 04:04:46  ajc
 * Began some glue code for the full text indexer.

 Revision 647.4  2005/05/16 20:03:33  ajc
 * definition of struct CitControl moved from citadel.h to server.h

 Revision 647.3  2005/05/16 18:48:45  ajc
 * Don't prompt the user for screen dimensions anymore.  Nobody is using
   dialup terminals with arbitrary screen sizes anymore; nearly everyone (or
   more likely, *absolutely* everyone) is now using networked displays which
   can be automatically queried for their screen dimensions.  For now, we'll
   keep the screen size properties in the database and in the protocol, but
   the prompts have been commented out.

 Revision 647.2  2005/05/16 18:25:56  ajc
 * Avoid re-creating the default-named baseroom (Lobby) upon subsequent
   startups after it's been renamed to something else.

 Revision 647.1  2005/05/16 16:59:39  ajc
 * Default expire policy is now 'manual' (no automatic expiry of messages
   under any circumstances).  Implemented as per David Given's suggestion
   that we should operate using the element of least surprise.

 Revision 647.0  2005/05/12 16:54:10  ajc
 * THIS IS 6.47

 Revision 646.1  2005/05/12 16:53:18  ajc
 * Fixed a minor bug discovered in SMTP at one site

 Revision 646.0  2005/05/11 17:06:04  ajc
 * THIS IS 6.46

 Revision 645.20  2005/05/09 22:17:30  ajc
 * Interactive room deletions are now deferred.  The server reconfigures the
   room to be a mailbox owned by a nonexistent user and immediately returns
   control back to the client.  Later, THE DREADED AUTO-PURGER will see the
   dangling room and erase its contents.

 Revision 645.19  2005/05/08 03:43:22  ajc
 * Set the internal version number to 6.46 in preparation for a release

 Revision 645.18  2005/05/02 20:52:42  ajc
 * More IMAP optimizations, including the application of our latest "don't
   fetch the message body" trick to ENVELOPE and INTERNALDATE fetches, to
   make MS-Outbreak spread email viruses faster.

 Revision 645.17  2005/05/02 16:09:52  ajc
 * serv_imap.c: implemented the RFC 3501 suggestion that the initial
   greeting, and the response to the LOGIN command, include an output
   of the server's CAPABILITY string.  I find this to be gratuitous but
   the UW IMAP server does it, so we are also doing it in order to better
   interoperate with clients which make assumptions...

 Revision 645.16  2005/04/29 22:02:20  ajc
 * More IMAP tuning

 Revision 645.15  2005/04/29 20:47:45  ajc
 * More complex cache handling for IMAP fetch operations -- now we can
   fetch/cache "just the headers" and remember whether we did so, so we can
   burn the cache if the client then comes around and requests something
   that requires the body.  Still needs some testing and tuning.

 Revision 645.14  2005/04/29 16:50:03  ajc
 * Significantly reduced the memory footprint of struct CitContext.

 Revision 645.13  2005/04/29 16:26:00  ajc
 * Removed the CtdlRedirectOutput() API, as we are no longer using it.
   (Oh happy day; no more temp files!)

 Revision 645.12  2005/04/27 19:22:57  ajc
 * .ASG command: made better use of memory to avoid crashes

 Revision 645.11  2005/04/23 04:38:26  ajc
 * Substantially improved the appearance of mailing list digests.
   Superfluous RFC822 headers no longer appear, and all messages are
   converted to plain text using the preferred_formats framework.
 * CtdlOutputPreLoadedMsg() -- fixed bug that caused Citadel protocol
   headers to appear for MIME prefix, suffix, etc. even when do_proto is 0.

 Revision 645.10  2005/04/23 02:04:11  ajc
 * serv_imap.c: don't respond "OK DELETE completed" until after the room
   delete operation actually completes.

 Revision 645.9  2005/04/23 01:59:58  ajc
 * Big performance optimization on CtdlSetSeen()

 Revision 645.8  2005/04/22 04:26:34  ajc
 * is_msg_is_mset() has been renamed to is_msg_in_sequence_set() because
   "sequence set" is now the official terminology as of RFC3501.
 * imap_set_seen_flags() no longer calls is_msg_in_sequence_set() for each
   message and for each flag.  It's just too expensive.  We now parse each
   flag's sequence set manually, marking the relevant messages as we go.

 Revision 645.7  2005/04/22 00:52:03  ajc
 * small fix to previous commit

 Revision 645.6  2005/04/21 17:28:51  ajc
 * msgbase.c: when committing a new message to the store, save its RFC822
   length to the metadata record immediately.  This will eliminate the need
   to calculate it later during a fetch operation.

 Revision 645.5  2005/04/14 15:53:55  ajc
 * Variable names, comments, documentation, etc...  removed the acronym 'BBS'
   in places where functionality is not specific to the use of Citadel as
   a BBS platform.

 Revision 645.4  2005/04/13 20:42:44  ajc
 * citmail.c: changes to citmail to make it usable as a /usr/sbin/sendmail
   replacement; i.e. /bin/mail calls /usr/sbin/sendmail which sends mail
   through Citadel.

 Revision 645.3  2005/04/13 17:03:07  ajc
 * Reverted database changes because the cull_logs function wasn't working
   with the separate log directory.
 * citmail.c: started some fixes to make it able to work from

 Revision 645.2  2005/04/12 21:19:52  ajc
 * 'make install' now installs citadel-openldap.schema

 Revision 645.1  2005/04/11 16:31:57  ajc
 * Database logs are now kept in the "data_logs" directory instead of in
   the "data" directory.  If no "data_logs" directory is found, a symlink
   to "data" will be created, in order to preserve access to any existing
   log files -- a savvy sysadmin (or a storage management wizard script)
   will know what to do if a different location is desirable.

 Revision 645.0  2005/04/01 03:02:44  ajc
 * THIS IS 6.45

 Revision 641.34  2005/03/31 04:31:42  ajc
 * Bumped the internal version number to 6.45 in preparation for
   an upcoming release.

 Revision 641.33  2005/03/24 22:52:40  ajc
 * More extermination of the dreaded SIZ moby-buffers.

 Revision 641.32  2005/03/24 22:13:56  ajc
 * extract_token() now expects to be supplied with the size of the
   destination string buffer.  This, along with the elimination of other
   unbounded functions like strcpy(), will allow the removal of the
   "all string buffers are of size SIZ" assumption (a process which I have
   already begun), which will hopefully reduce stack consumption.

 Revision 641.31  2005/03/22 21:53:48  ajc
 * Oops, forgot to initialize some data structures...

 Revision 641.30  2005/03/22 16:49:29  ajc
 * Fixed a couple of memory allocation bugs

 Revision 641.29  2005/03/20 22:55:58  ajc
 * Logging to stderr no longer uses syslog()
 * Reworked the way dead sessions are purged.  More efficient and more
   reliable now.

 Revision 641.28  2005/03/18 21:40:36  ajc
 * Minor bugfix to previous checkin

 Revision 641.27  2005/03/18 21:25:06  ajc
 * Finished removing all the "dynamic session data" stuff in order to
   boost reliability, improve performance, and reduce complexity.

 Revision 641.26  2005/03/12 05:42:35  ajc
 * Trying to fix a memory bug somewhere.
 * While working on the above, noticed that the way we did the per-session
   dynamic symbols loses badly in terms of performance.  Began moving to
   a less modular but better performing way of doing the same.

 Revision 641.25  2005/03/10 03:36:25  ajc
 * Silenced a compiler warning
 * Our graceful cleanup handler no longer gracefully cleans up after
   receiving SIGSEGV or its friends.  Unfortunately we need the core dump.

 Revision 641.24  2005/03/10 03:11:07  ajc
 * Altered the algorithm by which the doubly-linked session list is
   amended and culled.  Decided that performance is better than cute
   session numbers (which we don't display to the users anymore anyway)
   and we now assign a session number (CC->cs_pid) starting with 1 when the
   server starts and incrementing indefinitely.  Need to test this more.

 Revision 641.23  2005/03/07 04:08:07  ajc
 * vcard.c: realloc fix

 Revision 641.22  2005/03/05 22:31:01  ajc
 * Allow the use of chained certificates for crypto

 Revision 641.21  2005/03/04 20:04:00  ajc
 * IMAP FETCH RFC822.SIZE now honors the cached rfc822 size in each
   message's metadata, using it if present and storing it for next time
   otherwise.
 * Fixed bug in POP3 server that was causing it to fail to store the
   correct RFC822 size in metadata.  Fortunately it was failing in a way
   that merely affected performance rather than corrupting the data on
   existing installations.

 Revision 641.20  2005/03/04 02:24:31  ajc
 * SMTP: only offer TLS on the MSA port (587), not on the MTA port (25).
   There seem to be some other MTA's out there that make things go haywire
   when TLS is offered.

 Revision 641.19  2005/03/04 02:18:46  ajc
 * config.c: default setting for "maxmsglen" is now 10 megabytes, not INT_MAX

 Revision 641.18  2005/03/03 18:10:27  ajc
 * sysdep.c: change the algorithm for dynamically expanding the redirect
   buffer's size.  Large writes were overflowing it.

 Revision 641.17  2005/03/03 17:56:53  ajc
 * Finished removing the use of temp files in IMAP.
   WARNING: DON'T USE THIS ON A PRODUCTION SYSTEM, IT HAS NOT BEEN
   FULLY TESTED WITH VARIOUS IMAP CLIENTS YET!

 Revision 641.16  2005/03/03 05:14:06  ajc
 * Realized that memreadline() does NOT return a NULL pointer when it hits
   a null character, and adjusted all the code I wrote yesterday under that
   assumption.
 * Removed the use of a temp file for IMAP FETCH BODYSTRUCTURE on a
   non-RFC822 message.

 Revision 641.15  2005/03/02 17:33:03  ajc
 * msgbase.c: fixed a buffer overflow error
 * imap_fetch.c, serv_imap.c, serv_imap.h: migrated imap_fetch_rfc822() to
   use the new in-memory message buffering.  BODY comes next...

 Revision 641.14  2005/03/02 03:35:18  ajc
 * serv_smtp.c: removed use of temporary file for SMTP transmission

 Revision 641.13  2005/03/02 03:01:18  ajc
 * serv_spam.c: use redirect_buffer instead of redirect_sock
 * Removed redirect_sock from the API.  redirect_fp is next!

 Revision 641.12  2005/03/02 02:42:06  ajc
 * Finished moving the POP3 server to the new redirect_buffer semantics

 Revision 641.11  2005/03/02 02:16:10  ajc
 * Toned down some of the hostility in the code's comments :)

 Revision 641.10  2005/03/01 22:03:35  ajc
 * Began implementation of a third RedirectOutput mode -- one which writes
   to a memory buffer.  This will replace the other two.
 * serv_pop3.c: began migration to the new redirect mode.

 Revision 641.9  2005/03/01 04:24:52  ajc
 * When saving an RFC822 message, use a less expensive algorithm to
   search for the Content Type

 Revision 641.8  2005/02/28 20:32:50  ajc
 * IMAP APPEND command now accepts the message text using one big
   client_read() call instead of a bunch of smaller ones.

 Revision 641.7  2005/02/28 02:25:29  ajc
 * added pid file writing

 Revision 641.6  2005/02/27 15:36:28  ajc
 * database_cleanup.sh: look for /usr/local/ctdlsupport/bin/db_dump and
   prepend that directory to the PATH if found.
 * database_cleanup.sh: add more warnings about how dangerous this tool is

 Revision 641.5  2005/02/26 16:27:54  ajc
 * setup.c: added more items to the list of non-Citadel MTA's which can be
   disabled during setup.

 Revision 641.4  2005/02/25 17:59:12  ajc
 * Ok, now we accept *all* the flags transmitted during an IMAP APPEND, not
   just the first one.

 Revision 641.3  2005/02/25 05:51:52  ajc
 * Started working on a fix for the IMAP APPEND "flags" thing.  It works
   but only for the first flag sent.  Will finish tomorrow.

 Revision 641.2  2005/02/24 18:14:24  ajc
 * Removed excessive trace messages that were slowing the server down
 * Prepared for handling of setting flags in IMAP APPEND

 Revision 641.1  2005/02/23 04:24:57  ajc
 * When auto-creating Mail>, Sent Items>, etc... set view to VIEW_MAILBOX

 Revision 641.0  2005/02/21 21:59:33  ajc
 * THIS IS 6.41

 Revision 640.14  2005/02/21 21:40:28  ajc
 * Updated internal version number to 6.41

 Revision 640.13  2005/02/20 21:10:59  ajc
 * Repaired access control problem in IMAP SUBSCRIBE

 Revision 640.12  2005/02/17 16:13:59  ajc
 * Removed "Log Hooks."  This enabled the removal of a buf[SIZ] in lprintf,
   where it can potentially blow lots of stacks.

 Revision 640.11  2005/02/17 03:52:16  ajc
 * setup.c: clarify some of the messages

 Revision 640.10  2005/02/16 19:03:38  ajc
 * master_cleanup() now passes along an exit code from its caller to the OS.

 Revision 640.9  2005/02/16 18:48:39  ajc
 * Try to reach our cleanup routine when SIGSEGV is caught.  Hopefully
   we'll get there and we can close the databases cleanly.

 Revision 640.8  2005/02/16 17:37:16  ajc
 * Be more aggressive about shutting down when told to.  Getting the
   databases closed is the highest priority.

 Revision 640.7  2005/02/16 04:08:42  ajc
 * newinstall.sh: put the checksum files server-side to avoid
   unnecessary downloads

 Revision 640.6  2005/02/16 03:02:28  ajc
 * Applied patches submitted by Kevin Kilbride for TCP_CORK and for
   turning CitContext into a doubly-linked list.

 Revision 640.5  2005/02/13 04:55:18  ajc
 * Did the server port hooks etc. for SSL listeners.  For some reason it
   doesn't work.  :(

 Revision 640.4  2005/02/13 04:23:59  ajc
 * Added server and client configuration settings to specify port
   numbers for IMAPS, POP3S, and SMTPS (SSL-encrypted services that start
   SSL upon connect instead of using a STARTTLS command).  These services
   are not yet implemented, only the port number settings are.

 Revision 640.3  2005/02/13 03:39:21  ajc
 * IMAP flag twiddling with STORE was broken because of the \Flag leading
   backslash getting stripped out.  Located problem and worked around.  Said
   many nasty things about IMAP's gratuitous complexity, particularly the
   moronic idea to put backslashes in a place where they were completely
   unnecessary.

 Revision 640.2  2005/02/12 16:58:36  ajc
 * Changes to lprintf() and start_daemon() submitted by Kevin Kilbride
   for more "true unix" approaches to each.

 Revision 640.1  2005/02/11 03:51:51  ajc
 * <.A>ide <U>ser-edit is now <.A>ide <U>ser <E>dit
 * Added a new <.A>ide <U>ser <D>elete command, because it is unintuitive
   to tell people "delete a user by setting their access level to 0"

 Revision 640.0  2005/02/10 16:52:18  ajc
 * THIS IS 6.40

 Revision 630.16  2005/02/10 16:52:03  ajc
 * Updated documentation etc. for 6.40 release (and for 2005 copyright)

 Revision 630.15  2005/02/08 03:33:49  ajc
 * client_gets(char *buf) has been replaced by
   client_getln(char *buf, int maxbytes)

 Revision 630.14  2005/02/05 22:56:31  ajc
 * More reliable handling of conversion of vCard UID to Citadel Extended ID
   (necessary for GroupDAV URL's to be generated properly in WebCit)

 Revision 630.13  2005/02/05 04:13:34  ajc
 * msgbase.c: replication checks for Exclusive-ID no longer cause a save
   operation to fail when the existing message is newer.  This was causing
   groupware clients to go haywire because of datestamps being set to
   event start times, etc.

 Revision 630.12  2005/02/03 04:36:56  ajc
 * serv_calendar.c: beforesave hook now looks for the UID inside nested
   VTODO components, in addition to VCALENDAR components.

 Revision 630.11  2005/02/01 23:11:46  ajc
 * new ENT0 syntax now also outputs EUID

 Revision 630.10  2005/02/01 19:46:12  ajc
 * Added an SMTP-like calling syntax to ENT0 to allow a confirmation message
   to be sent back to the client after a message is transmitted.

 Revision 630.9  2005/02/01 03:33:22  ajc
 * Changed CtdlRoomAccess() calling syntax in order to return both
   the access bits and the current view
 * All "list rooms" commands now return the view for each room.

 Revision 630.8  2005/01/27 22:05:21  ajc
 * Renamed the "Extended message ID" field to "Exclusive message ID"
   (nothing changes except documentation and internal variable names)

 Revision 630.7  2005/01/27 21:59:22  ajc
 * serv_vcard.c: changed the logic a bit.  When saving a vCard to a user's
   "My Citadel Config" room, force-feed the vCard a site-compliant UID and
   FBURL.  When saving a vCard to *any* address book room, set the Citadel
   EUID to the vCard UID.

 Revision 630.6  2005/01/27 17:33:52  ajc
 * The before-save hook in serv_calendar.c now runs for any room with a
   view of VIEW_CALENDAR or VIEW_TASKS, not just Calendar> and Tasks>.

 Revision 630.5  2005/01/26 23:04:22  ajc
 * When saving calendar items, if the vCalendar object does not contain
   a UUID, generate one on the fly.  In order to support GroupDAV we need
   persistent URL's, and I plan to base those on the UUID.

 Revision 630.4  2005/01/25 20:58:07  ajc
 * serv_pop3.c: minor change to logging output

 Revision 630.3  2005/01/25 15:02:12  error
 * Updated SSL/TLS code to actually use the TLS cipher suites

 Revision 630.2  2005/01/22 04:15:32  ajc
 * Increased per-thread stack size again, this time to 1MB

 Revision 630.1  2005/01/22 03:14:20  ajc
 * Each message's metadata now has the ability to cache the length of
   the message when output in RFC822 format.  The POP3 service populates
   this field the first time it sees each message, and fetches the length
   from cache on subsequent visits.
 * Because of this optimization, we no longer need to keep the entire POP3
   mailbox stored in open temp files during a session.  Each message is
   opened again when it is fetched.

 Revision 630.0  2005/01/21 20:25:08  ajc
 * THIS IS 6.30

 Revision 629.10  2005/01/21 20:24:34  ajc
 * Internal version number is now 6.30

 Revision 629.9  2005/01/21 20:20:06  ajc
 * Increase per-thread stack size from 128k to 256k.  This fixes the
   crashing IMAP service on 64-bit Linux.

 Revision 629.8  2005/01/19 20:49:46  ajc
 * Minor fixenbugs after running with Valgrind

 Revision 629.7  2005/01/19 03:03:36  ajc
 * Minor and/or cosmetic changes made during x64 troubleshooting

 Revision 629.6  2005/01/18 17:58:38  ajc
 * Trying to fix IMAP brokenness on Linux-x64.  Committing changes even
   though there are a lot of annoying trace messages in this tree, because
   I did clean up a bunch of stuff here and there.

 Revision 629.5  2005/01/14 17:08:30  ajc
 * Applied a configure script fix sent in by David Given for more reliable
   detection of the installed libical version.

 Revision 629.4  2005/01/13 05:43:16  ajc
 * Initial support for having a room participate in a remote mailing list

 Revision 629.3  2005/01/09 03:20:41  ajc
 * Applied a patch sent in by David Given for handling of the "&" character
   plus Unicode characters in IMAP folder names.

 Revision 629.2  2005/01/07 15:56:14  ajc
 * msgbase.c: fixed an incorrect variable name in the bigmsg-handler

 Revision 629.1  2005/01/07 03:05:04  ajc
 * serv_imap.c: fail with a "BAD" error response if the LOGIN command is
   called with the wrong number of parameters (otherwise it goes boom)
 * user_ops.c: CtdlLoginExistingUser() was calling strproc() on the supplied
   name, which killed the ability to log in with your email address (because
   it stripped the @ symbol).  Changed it to striplt() instead.

 Revision 629.0  2005/01/04 03:57:32  ajc
 * THIS IS 6.29

 Revision 628.2  2005/01/04 03:57:21  ajc
 * newinstall.sh: save tarball checksums after successful install of each
   component; skip subsequent installs for components which have not changed
 * Bumped internal version number to 6.29

 Revision 628.1  2005/01/04 02:06:28  ajc
 * sysdep.c: if we have enough information to do so, adjust maximum file
   descriptors to avoid exceeding FD_SETSIZE.  This is done to circumvent
   a rare but specific vulnerability.

 Revision 628.0  2004/12/14 03:16:19  ajc
 * THIS IS 6.28

 Revision 627.12  2004/12/12 17:33:24  error
 * sysdep.c: lprintf(): Bug 124: Fix remote format string vulnerability
   (thanks to coki@nosystem.com.ar)

 Revision 627.11  2004/12/01 16:28:58  ajc
 * tools.c: don't crash when striplt() is called with z zero length string

 Revision 627.10  2004/11/26 22:44:08  ajc
 * Added a sooper-seekrit way to spool network messages to a remote node
   where the room has a different name
 * Internal version number is now 6.28, so that WebCit doesn't b0rk the
   netconfigs of older Citadel servers

 Revision 627.9  2004/11/19 02:31:56  ajc
 * Added developer ID # 177

 Revision 627.8  2004/11/16 23:04:18  ajc
 * setup.c: added "dovecot" to the list of "other email services" which
   the setup program can helpfully disable for you.  :)

 Revision 627.7  2004/11/12 03:51:42  ajc
 * serv_expire.c: auto-purge any user whose user number is less than 1.

 Revision 627.6  2004/11/11 16:21:12  nbryant
 * database_sleepycat.c: make cdb_rewind enforce that cursors must be
   closed before they can be opened again. (prevents cursors being held
   open for longer than they should be and holding database locks.)

 Revision 627.5  2004/11/11 14:56:16  ajc
 * Further adjustments to ldap setup

 Revision 627.4  2004/11/11 04:38:36  ajc
 * setup.c: fix bug that kept creating new inittab entries for slapd

 Revision 627.3  2004/11/10 21:33:47  ajc
 * Go to Berkeley DB 4.3.21 in Easy Install

 Revision 627.2  2004/11/10 19:59:23  nbryant
  * configure.ac: detect /usr/local/BerkeleyDB.4.3

 Revision 627.1  2004/11/10 03:03:14  ajc
 * Added a few wonderful linebreaks to >80 column lines

 Revision 627.0  2004/11/03 20:18:58  ajc
 * THIS IS 6.27

 Revision 626.18  2004/11/03 20:18:46  ajc
 * Internal version number is now 6.27

 Revision 626.17  2004/11/02 22:59:20  ajc
 * setup.c: fixed an incorrect variable name that was causing inittab
   twiddling to malfunction.

 Revision 626.16  2004/11/02 02:47:10  ajc
 * Easy Install: put the log directly in /tmp instead of in the build
   directory so it doesn't get deleted on a failed install.

 Revision 626.15  2004/10/23 14:33:44  ajc
 * serv_notes.c: finished.  When saving a message in a room whose view
   is set to "notes," look for an X-KOrg-Note-Id: header, and if one is
   present, set both the Extended ID and the Subject to that.  This is for
   Aethera compatibility.

 Revision 626.14  2004/10/22 14:49:25  ajc
 * newinstall.sh: Be more careful about locating make/gmake on the host
   system.  Also now tries to use curl if wget is not available.

 Revision 626.13  2004/10/22 02:53:25  ajc
 * serv_notes.c: added skeleton module
 * serv_newuser.c: corrected internal comment
 * techdoc: renamed session.txt to protocol.txt because it really isn't
   a session layer protocol, it's more of an application layer protocol.

 Revision 626.12  2004/10/13 02:11:00  ajc
 * Repaired a completely broken <;Z>ap floor command.

 Revision 626.11  2004/10/12 11:24:42  ajc
 * Tracing a problem with ;Z command

 Revision 626.10  2004/10/12 02:48:11  ajc
 * Reworked shutdown sequence to avoid thread deadlock

 Revision 626.9  2004/10/12 02:17:49  ajc
 * Cleaned up some things that generated compiler warnings
 * crypto keys directory is now relative to the server's working directory,
   not to the compiled-in CTDLDIR
 * Re-ordered the security checks in CtdlAccessCheck()

 Revision 626.8  2004/10/06 21:23:21  error
 * Fixup a few more compiler warnings from icc

 Revision 626.7  2004/10/05 01:44:20  ajc
 * Changed a bunch of localtime() calls to localtime_r(), for great justice.

 Revision 626.6  2004/10/04 21:40:29  error
 * configure.ac: Add CFLAGS for icc Intel Compiler

 Revision 626.5  2004/10/04 21:39:35  error
 * Fix a couple of minor compiler warnings

 Revision 626.4  2004/10/03 04:25:23  ajc
 * Removed some vestiges

 Revision 626.3  2004/10/03 04:11:48  ajc
 * sysdep.c: in the main server loop, when time_to_die is detected, return
   immediately instead of falling through to the end of the function.

 Revision 626.2  2004/10/03 03:57:32  ajc
 * Added pthread_attr_destroy() in the appropriate location (thanks fleeb!)

 Revision 626.1  2004/10/01 15:58:08  ajc
 * Configure "notes" view and auto-create Notes> room

 Revision 626.0  2004/09/28 16:11:16  ajc
 * THIS IS 6.26

 Revision 625.33  2004/09/28 16:09:58  ajc
 * Update documentation and config files for 6.26 release

 Revision 625.32  2004/09/28 02:18:12  ajc
 * Improve output of RECENT flag in IMAP

 Revision 625.31  2004/09/23 03:02:31  ajc
 * MyContext() no longer declared INLINE

 Revision 625.30  2004/09/23 02:54:46  ajc
 * in MyContext(), reduced the number of calls to pthread_getspecific()
   from two to one

 Revision 625.29  2004/09/23 00:22:18  error
 * rooms.c: create_floor(): Fix the check for whether a floor name was given

 Revision 625.28  2004/09/21 02:09:30  ajc
 * Tweaks to above

 Revision 625.27  2004/09/21 01:43:23  ajc
 * imap_search.c: when search criteria permit, do not fetch messages.

 Revision 625.26  2004/09/17 16:54:13  ajc
 * Updated documentation to include information about setting up MSA 587 port.

 Revision 625.25  2004/09/17 04:14:18  ajc
 * serv_network.c: added a missing \n in log output

 Revision 625.24  2004/09/17 03:54:47  ajc
 * citadel-openldap.schema: replaced.  Dunno why it was missing.
 * serv_calendar.c: fixed a return with no value for a function returning
   int; this was causing a before-save hook to abandon messages

 Revision 625.23  2004/09/16 01:46:40  ajc
 * CRE8 command: allow setting default view during room creation

 Revision 625.22  2004/09/15 03:02:47  ajc
 * Add an SMTP MSA listener (separate port, requires auth)

 Revision 625.21  2004/09/13 15:51:59  ajc
 * newinstall.sh: updated from the working version at easyinstall.citadel.org

 Revision 625.20  2004/09/11 03:13:04  error
 * Remove calls to the broken flush_output() while I try to figure out what's
   wrong with it.  Replaced with unbuffer_output() which does work.

 Revision 625.19  2004/09/10 02:54:26  ajc
 * Added flush_output() calls to IMAP modules.  (Do we need them?  Are we
   buffering IMAP output?  Should we?)

 Revision 625.18  2004/09/09 02:26:45  ajc
 * Completed (I think) the 'dialog' mode in setup

 Revision 625.17  2004/09/08 04:16:07  ajc
 * setup.c: initial changes to use a 'dialog' based setup (yes, it's back,
   because the b0rken version is now long gone and it appears to be more
   portable than newt)

 Revision 625.16  2004/09/07 04:15:35  error
 * msgbase.c: cmd_msgs(): Call unbuffer_output() when a message list is
   requested using a search template.

 Revision 625.15  2004/09/06 01:11:35  error
 * msgbase.c: cmd_ent0(): Fix another unbuffer_output()

 Revision 625.14  2004/09/06 00:59:01  error
 * file_ops.c: cmd_writ(): Add an unbuffer_output() that I forgot.

 Revision 625.13  2004/09/05 17:39:09  error
 * Buffered output needs to be flushed in several places.  Added calls to
   flush_output().  (basically anywhere where we send a response and then
   wait for the client, except chat, where we turned it off entirely)

 Revision 625.12  2004/09/05 15:41:45  error
 * Network optimizations - buffer output server-side for better network
   utilization; one client-side optimization

 Revision 625.11  2004/09/05 15:20:41  error
 * sysdep.c: unbuffer_output(): Split the writing part to a new function
   flush_output() for more precise control

 Revision 625.10  2004/09/03 04:34:30  ajc
 * setup.c: when creating an inittab entry for slapd, use "-d 0" instead
   of "-d 1" to avoid spewing messages to the console, while still remaining
   in the foreground.

 Revision 625.9  2004/09/03 04:19:17  ajc
 * setup.c: Changes to inittab-twiddling to make it a bit more reliable
 * ipc_c_tcp.c: don't send SIGHUP to children; it makes them very angry

 Revision 625.8  2004/09/01 15:34:52  ajc
 * serv_smtp.c: fix build error on hosts with no SSL support

 Revision 625.7  2004/08/31 17:31:55  ajc
 * Calendar: store UUID+comment instead of comment as Subject (this was done
   at the request of Eugen Constantinescu for Aethera optimizations)

 Revision 625.6  2004/08/30 02:45:33  ajc
 * setup.c: when running in Newt mode, make the size of the dialogs dynamic
   to the size of the text in them.

 Revision 625.5  2004/08/29 15:18:41  error
 * newinstall.sh: fix detection of gmake/make

 Revision 625.4  2004/08/28 14:21:25  error
 * Change the logging level of some messages and add a couple of new ones to
   show when a network node is connecting.

 Revision 625.3  2004/08/28 02:42:44  ajc
 * setup.c: when run from the Easy Install script, auto-configure slapd
   and put it into /etc/inittab

 Revision 625.2  2004/08/28 01:56:38  ajc
 * Generalize some of the inittab-tweaking stuff so that we can use it
   for slapd, too.

 Revision 625.1  2004/08/27 21:39:33  ajc
 * Fleshed out contemplate_ldap() a bit

 Revision 625.0  2004/08/27 21:06:30  ajc
 * THIS IS 6.25

 Revision 624.8  2004/08/27 20:16:51  ajc
 * Update internal version number to 6.25

 Revision 624.7  2004/08/26 04:13:55  ajc
 * newinstall.sh: updated with some new goodies
 * setup.c: when run from Easy Install, offer to set up LDAP.  (Not finished.)

 Revision 624.6  2004/08/23 21:41:26  error
 * html.c: Try to something sane with lists

 Revision 624.5  2004/08/23 21:40:04  error
 * No longer accept MD5 as a hash when encrypting

 Revision 624.4  2004/08/23 21:34:28  error
 * serv_network.c: Suppress Invalid node name for "." and ".."

 Revision 624.3  2004/08/16 04:19:14  ajc
 * newinstall.sh: more intelligent auto-run of setup programs
 * setup.c: change inittab question to be more newbie-friendly

 Revision 624.2  2004/08/15 05:04:36  ajc
 * newinstall.sh: various tweaks to make it more reliable

 Revision 624.1  2004/08/14 03:59:43  ajc
 * newinstall.sh: shuffled around the environment variables to prevent
   "C compiler cannot generate executables" error.

 Revision 624.0  2004/08/12 13:47:47  ajc
 * THIS IS 6.24

 Revision 623.16  2004/08/11 04:09:14  ajc
 * Replaced all "Citadel/UX" references with "Citadel"

 Revision 623.15  2004/08/05 03:19:48  ajc
 * html.c: handle the <BLOCKQUOTE> tag properly

 Revision 623.14  2004/08/05 02:02:33  ajc
 * CtdlHostAlias() now accepts "localhost" as a localhost address.
 * working_ignetcfg is now loaded during incoming NETP commands, which
 should theoretically take care of the missing config problem.

 Revision 623.13  2004/08/02 02:51:03  ajc
 * utilsmenu: removed.  Because most of the admin functions have been moved
   into Citadel proper, what's left of this menu is now pathetic.
 * techdoc/build.txt: removed verbage that referred to the old build system.

 Revision 623.12  2004/07/30 03:18:21  ajc
 * user_ops.c: limit the length of the name which can be supplied
   to CtdlLoginExistingUser() to avoid crashing it.

 Revision 623.11  2004/07/28 04:09:58  ajc
 * serv_network.c: we had ignetcfg and working_ignetcfg, but it turns out we
   no longer use the former anywhere, so it has been removed (along with the
   aftersave hook that detected changes)

 Revision 623.10  2004/07/25 16:20:04  error
 * msgbase.c: CtdlCheckInternetMailPermission(): Do not allow twits to send
   Internet mail (they shouldn't be sending ANYTHING at all)

 Revision 623.9  2004/07/25 16:18:43  error
 * citadel.spec: update version number

 Revision 623.8  2004/07/25 16:17:37  error
 * stress.c:  Slightly more verbose output

 Revision 623.7  2004/07/25 16:16:19  error
 * citadel_decls.h: fixed the new logoff() macro

 Revision 623.6  2004/07/24 05:11:52  ajc
 * citadel.c: temporarily commented out the signal() call for SIGHUP because
   it is causing the client not to work at all on Linux 2.6 (or at least on
   Fedora Core 2) when citadel is called as the login shell -- a SIGHUP is
   mysteriously received at the username prompt and the program exits.

 Revision 623.5  2004/07/15 02:54:49  ajc
 * serv_network.c: grab and use a networker-local copy of the network
   configuration, ensuring that it doesn't accidentally get clobbered while
   the network is running.

 Revision 623.4  2004/07/11 18:41:39  ajc
 * Updated roadmap document slightly

 Revision 623.3  2004/07/11 03:59:02  ajc
 serv_network.c: small fix to above

 Revision 623.2  2004/07/11 03:50:31  ajc
 * serv_network.c: removed a stray end_critical_section()
 * serv_network.c: purge invalid ignet_push_share entries during
   netconfig load instead of during rewrite

 Revision 623.1  2004/07/10 02:51:01  ajc
 * Changed "policy set" message to "policy has been updated" to make it
   clearer.

 Revision 623.0  2004/07/06 18:15:04  ajc
 * THIS IS 6.23

 Revision 622.7  2004/07/06 18:09:13  ajc
 * When writing a message body to the CDB_BIGMSGS table instead of the main
   message base, increase data length by one byte in order to include the
   NULL terminator.  (Doh!)

 Revision 622.6  2004/07/03 04:09:37  ajc
 * Check the "Permission to send Internet mail" access controls when
   messages are submitted via Authenticated SMTP.

 Revision 622.5  2004/06/28 17:09:20  ajc
 * Properly re-terminate messages posted via IMAP APPEND.  The newline
   conversion process was shortening the text but not adjusting string
   termination, causing extra message snippets to appear.

 Revision 622.4  2004/06/24 20:38:06  ajc
 * updated serv_vandelay.c to include new "c_ip_addr" config

 Revision 622.3  2004/06/24 20:33:37  ajc
 * Bumped internal version number to 6.23
 * Added "Server IP address" to config (server, client, docs)

 Revision 622.2  2004/06/24 15:26:33  ajc
 * Updated the ig_tcp_server() function to allow binding to a single IP
   address.  This is not yet used by anything.

 Revision 622.1  2004/06/24 02:34:39  ajc
 * serv_imap.c: when an IMAP socket breaks while a folder is selected,
   auto-expunge the folder before closing it.  Fixes the "mysteriously
   reappearing deleted messages" for clients that just drop the session
   without issuing a CLOSE when the user closes the window.

 Revision 622.0  2004/06/19 03:20:22  ajc
 * THIS IS 6.22

 Revision 621.19  2004/06/19 03:13:26  ajc
 * Updated internal version number to 6.22

 Revision 621.18  2004/06/19 02:35:16  ajc
 *** empty log message ***

 Revision 621.17  2004/06/19 02:33:54  ajc
 * Removed the configuration option "Allow Aides access to all mailboxes."
   This functionality *must* always be enabled for administrative create/edit
   of vCards, which live in the user's My Citadel Config> room.

 Revision 621.16  2004/06/16 04:17:43  ajc
 * More scheduler changes.  Removed the rescan pipe again, and also
   removed the mutex wrapper around select().  In my initial testing I am
   getting reliable, fast service, but further testing is needed.

 Revision 621.15  2004/06/16 03:13:02  ajc
 * Scheduler fix ... added the rescan pipe back in

 Revision 621.14  2004/06/15 20:42:42  ajc
 * Store the body of any large (>1K) message in a separate database.  This
   will allow fast headers-only retrieval later.

 Revision 621.13  2004/06/15 03:05:01  ajc
 * Bumped internal version number to 6.22
 * Added a new faster headers-only mode that excludes enumeration of
   MIME parts.

 Revision 621.12  2004/06/14 19:09:26  ajc
 * serv_vandelay.c: re-inserted a missing line of code for setting tempfile
   names, which was causing exports to crash.

 Revision 621.11  2004/06/11 01:26:02  ajc
 * Optimized the scheduler some more.  The S_I_WANNA_SELECT mutex now
   wraps only the select() call itself.

 Revision 621.10  2004/06/09 03:54:07  ajc
 * The scheduler can now "wake up" a session to deliver async messages.
 * Renumbered the "instant msgs waiting" message and changed its usage

 Revision 621.9  2004/06/07 22:12:19  ajc
 * Removed network/filterlist network/mailinglists network/rnews.xref
   because none of these are used anymore.

 Revision 621.8  2004/06/07 16:41:28  ajc
 * Changed "express message" to "instant message" everywhere in the code
   and documentation, to reflect the now industry-standard terminology for
   this function.  (Obviously, the server command names have not changed,
   because that would break everything.)

 Revision 621.7  2004/06/06 22:30:10  ajc
 * New session scheduler.  All sessions which select() marks for activity
   are now handled before select() is called again.

 Revision 621.6  2004/06/03 02:49:14  ajc
 * html.c: allow parsing of tags even when they're qualified
           (i.e. <TAG foo=bar> instead of just <TAG> )
 * html.c: handle escaped decimal characters (such as &#39; for an apostrophe)

 Revision 621.5  2004/06/03 02:28:16  ajc
 * citadel_ipc.c: SETR command was missing defaultview and flags2.  Added.

 Revision 621.4  2004/05/31 15:19:14  ajc
 * Reworked the security checks for move/copy commands.  Theoretically it's
   performing all the same checks, but the code is far more readable.

 Revision 621.3  2004/05/26 18:13:15  nbryant
 * stress.c: silence warning

 Revision 621.2  2004/05/26 16:53:31  nbryant
 * configure.ac: check for /usr/local/BerkeleyDB.4.2

 Revision 621.1  2004/05/24 01:59:16  ajc
 * serv_network.c: when processing incoming IGnet spool, give the <C> field
   priority over the <R> field.  This allows messages emailed to a network
   room to be shared properly instead of attempting to deliver via email a
   second time.

 Revision 621.0  2004/05/21 01:58:16  ajc
 * THIS IS 6.21

 Revision 620.35  2004/05/20 16:14:08  ajc
 * Changed internal version number to 6.21
 * Added missing VIEW_* defines in citadel.h (they're used only by WebCit
   for now, which is why Citadel didn't need them, but they should be there
   anyway for completeness)
 * Removed the last vestiges of the built-in spam filter.  It had been
   commented out, but it's definitely never coming back now, so away it goes.
 * Updated some of the docs

 Revision 620.34  2004/05/15 14:19:52  error
 * Scheduler: when next_session is to be deleted, make it point to some
   other session which isn't being deleted (or NULL if there aren't any).

 Revision 620.33  2004/05/14 03:09:54  ajc
 * When a session kills itself (for example, due to a broken socket),
   force the dead_session_purge() to run immediately.  This avoids
   thousands of error messages for up to the next five seconds while it
   waits for the next purge.
 * For the main select() loop, we can now recover from EBADF by jumping
   back to the code that scans for valid descriptors.  (Yeah, I used a
   goto.  It's more readable that way, so STFU if you have a problem with
   it.)

 Revision 620.32  2004/05/11 15:21:45  nbryant
 * configure.ac: check for <sys/prctl.h>
 * server_main.c: call prctl(PR_SET_DUMPABLE, 1) if we are dropping root
   permissions.
 * sysdep.c: no longer call setrlimit(RLIMIT_CORE, ...); this can be handled
   from a shell script.

 Revision 620.31  2004/05/10 01:47:11  error
 * messages.c: Fix crash when trying to use m<Y> next

 Revision 620.30  2004/05/07 20:27:46  ajc
 # citadel-openldap.schema: updated to include RFC2739 objects & attributes

 Revision 620.29  2004/05/07 19:31:40  ajc
 * More RFC2739 compliance: when converting vCard to LDAP, include
   the calFBURL attribute.

 Revision 620.28  2004/05/07 19:09:07  ajc
 * Implemented partial RFC2739 compliance (calendar free/busy URL in vCard).
   Still need to add it to LDAP.

 Revision 620.27  2004/05/07 18:17:26  ajc
 * Scrawled some notes in roadmap.txt

 Revision 620.26  2004/05/06 03:38:26  ajc
 * Never reduce the size of the thread pool, only increase it (still
   observing the max ceiling, of course)
 * Don't explicitly call RemoveContext() when a session exits.  The
   subsequent call to dead_session_purge() will handle it.

 Revision 620.25  2004/05/03 00:12:27  ajc
 * Found and removed a bug that caused network rooms to be un-shared with
   all nodes instead of only with nodes that no longer exist.
 * New target "make upgrade" (to be consistent with other projects)

 Revision 620.24  2004/04/29 02:37:58  ajc
 * Updated the roadmap

 Revision 620.23  2004/04/26 15:11:17  ajc
 * Enable core dumps regardless of system ulimit setting
   (maybe temporary until we fix BOOM)

 Revision 620.22  2004/04/23 17:27:50  ajc
 * user_ops.c: do not attempt to save the loaded user record into the
   supplied buffer, when the supplied buffer is NULL.  (This calling
   syntax is used to check for the existence of a user without storing it.)

 Revision 620.21  2004/04/21 03:44:50  ajc
 * serv_crypto.c: small fix to make self-signed certs no longer invalid

 Revision 620.20  2004/04/21 03:00:06  ajc
 * Replace ctdl_install_certificate() with convenience functions found
   in the OpenSSL library.

 Revision 620.19  2004/04/20 02:42:54  ajc
 * techdoc/binaries.txt : updated, now includes WebCit instructions
 * setup.c: detect when setup is run from within the Citadel Ridiculously
   Easy Installer and skip the directory prompt; the installer sets it.

 Revision 620.18  2004/04/14 18:43:51  nbryant
 * citadel.spec: BuildRequire redhat-rpm-config

 Revision 620.17  2004/04/14 16:16:47  nbryant
 * citadel.spec: don't BuildRequire newt-devel

 Revision 620.16  2004/04/14 03:42:01  ajc
 * ICAL FREEBUSY output now includes ORGANIZER, DTSTART, and DTEND fields.
   (Required for Kolab compatibility.)

 Revision 620.15  2004/04/13 23:53:00  nbryant
 * citadel.spec: BuildRequire bison

 Revision 620.14  2004/04/13 22:44:35  nbryant
  * citadel.spec: commented out Icon line and the openldap-servers dependency.
  * citadel.spec: added several BuildRequires. Some of these are, strictly
    speaking, compile time options, but the goal is to document the runtime
    dependencies and make sure our RPM's are always built consistently.

 Revision 620.13  2004/04/13 21:09:20  error
 * citadel.spec: Updated for 6.20p1

 Revision 620.12  2004/04/13 18:31:40  ajc
 * serv_calendar.c: ICAL FREEBUSY command now tries the supplied name not
   only as a screen name, but as an email address, and then as an
   unqualified email address in every hosted domain.  (For Kolab compat)

 Revision 620.11  2004/04/13 02:45:07  ajc
 * Don't display the "Ending SSL/TLS" log message unless TLS is actually present

 Revision 620.10  2004/04/12 20:51:40  error
 * citadel.lsm: Update LSM for 6.20p1; uploaded to ibiblio.org

 Revision 620.9  2004/04/09 23:11:01  error
 * serv_chat.c: Differentiate when a user does not exist when paging users.

 Revision 620.8  2004/04/09 23:05:52  error
 * user_ops.c: fix unused variable compiler warnings

 Revision 620.7  2004/04/03 15:42:35  ajc
 * stress.c: include <stdlib.h> in order to get RAND_MAX
   (Submitted by Thomas.Lotterer@cw.com, Cable & Wireless / OpenPKG)

 Revision 620.6  2004/04/01 04:41:55  ajc
 * Put some blankety-blank values in the default generated vCard so it's
   more acceptable to LDAP conversion

 Revision 620.5  2004/03/31 02:19:09  ajc
 * Change the IMAP folder delimiter from "|" to "/" because more than a few
   client programs don't follow RFC2060 strictly enough to work with
   non-"/" delimiters.  Actual slashes in room names appear as "|" in IMAP.

 Revision 620.4  2004/03/30 03:11:07  ajc
 * control.c: when config.c_maxsessions is unset or negative, set it to 0
   instead of setting it to 1 when it is 0 or negative.  The default should
   be unlimited sessions.

 Revision 620.3  2004/03/29 16:05:02  ajc
 * begin_critical_section() -- bypass transaction checking for S_FLOORCACHE
   sections, to avoid crashing the db layer

 Revision 620.2  2004/03/29 02:33:19  error
 * citadel.spec: major modifications for 6.xx (hasn't been updated since 5.xx)

 Revision 620.1  2004/03/27 23:04:59  error
 * citadel.lsm: update version number, sending to ibiblio

 Revision 620.0  2004/03/27 02:33:26  ajc
 * THIS IS 6.20

 Revision 614.105  2004/03/27 02:33:10  ajc
 * Bumped the version number to 6.20
 * Corrected session.txt writeup for CONF command (three of the five LDAP
   related fields were missing)
 * Updated the roadmap documentation slightly

 Revision 614.104  2004/03/27 02:21:30  ajc
 * Completed documenting the LDAP Connector

 Revision 614.103  2004/03/26 05:41:25  ajc
 * Began writing up the LDAP Connector for Citadel in the documentation

 Revision 614.102  2004/03/26 05:40:47  ajc
 *** empty log message ***

 Revision 614.101  2004/03/26 05:13:15  ajc
 * citadel-slapd.conf: comment out the reference to the Citadel schema
   and add "schemacheck off" to make it easier to get started with the
   Citadel LDAP connector.  Also made the backend in the sample conf
   ldbm instead of bdb, because that's what RH9 ships with.

 Revision 614.100  2004/03/24 21:23:50  ajc
 * msgbase.c: repaired a memory leak

 Revision 614.99  2004/03/24 17:07:11  ajc
 * Put a new memory leak checker into the server because it turns out that
   the third-party ones all suck.  :)  This one doesn't involve renaming
   all of the malloc() related functions, though.

 Revision 614.98  2004/03/24 15:04:06  ajc
 * stress.c: fix "wrong password" race condition by giving the very
   first worker thread time to finish creating the user account before the
   other threads start using it.  (Ok, I fixed it by creating a different
   race condition, but if your Citadel takes more than three seconds to
   create an account, your problems can't be fixed with the stress tester.)

 Revision 614.97  2004/03/24 03:46:40  ajc
 * Reorg header stuff to make it more compatible with leak checkers

 Revision 614.96  2004/03/24 03:25:19  ajc
 * Removed the built-in memory leak checker.  It wasn't threadsafe and
   there now exist third-party utilities that do this job better.

 Revision 614.95  2004/03/24 02:59:19  ajc
 * After initializing the database, chown and chmod all files in the data/
   directory correctly to avoid EPERM errors later on when we drop root privs

 Revision 614.94  2004/03/22 19:37:28  error
 * sysdep.c: worker_thread(): Make scheduling a little more fair to higher
   sessions
 * sysdep.c: lprintf(): Enable microsecond display in trace file
 * stress.c: worker(): Sleep for random amount of time as per specification

 Revision 614.93  2004/03/21 22:51:54  error
 * Fix a few remaining lprintf(9, ...) to lprintf(CTDL_DEBUG, ...)

 Revision 614.92  2004/03/21 22:35:20  error
 * stress.c: Wait before posting instead of after posting.

 Revision 614.91  2004/03/21 22:34:41  error
 * database_sleepycat.c: emit a panic message when Berkeley DB wants us to
   run recovery; should provide a little more detail on the actual error

 Revision 614.90  2004/03/21 22:32:24  ajc
 * room_ops.c: increased the thread safety of cgetfloor()
 * housekeeping.c: check floor reference counts in two passes instead of
   trying to manipulate multiple tables in O^2

 Revision 614.89  2004/03/21 17:14:46  error
 * stress.c: Fixed.  Now properly spawns threads and stresses out your
   favorite Citadel server by simulating large numbers of really active
   users.  (It's configurable on the command line, too, to vary the stress
   level.)  Don't use it against a production system!  You've been warned!

 Revision 614.88  2004/03/21 17:13:15  error
 * configure.ac: Fix for defines for pthreads not being included properly

 Revision 614.87  2004/03/21 06:15:06  ajc
 * docs/citadel.html: documented TLS support and administration procedures

 Revision 614.86  2004/03/20 22:32:06  ajc
 * citadel.rc: changed the comments around local_screen_dimensions= to
   reflect its legacy type of situation

 Revision 614.85  2004/03/16 22:44:15  ajc
 * Better conversion of friendly name to fully qualified vcard name

 Revision 614.84  2004/03/16 21:09:37  ajc
 * Eliminate EVT_OUTPUTMSG server extensions (don't need them anymore)
 * Add EVT_NEWUSER server extensions
 * EVT_NEWUSER and EVT_PURGEUSER server extensions now specify a
   struct ctdluser * instead of a username and usernum
 * serv_vcard.c: automatically create and submit a skeleton vCard when
   a new user is created.

 Revision 614.83  2004/03/16 19:05:50  error
 * Further changes to stress.c, Makefile.in to build it, .cvsignore
   (but it's still broken)

 Revision 614.82  2004/03/15 18:11:34  error
 * stress.c: various fixed/improvements (but it still isn't working)

 Revision 614.81  2004/03/15 17:22:14  error
 * First draft of the stress tester.  Isn't quite ready (it needs to be
   tested itself!).

 Revision 614.80  2004/03/15 16:48:22  error
 * Documentation update: update citadel.html for new syslog logging

 Revision 614.79  2004/03/15 16:47:04  error
 * msgbase.c: cmd_move(): Fix the security logic for non-aides trying to
   move or copy messages.  The logic should do the following:
   * Allow move/copy for Aides
   * Allow regular users to:
     * Move a message between two personal rooms
     * Copy a message from a Personal room to a room the user is subscribed to
     * Copy a message from a room the user is subscribed to, to a personal room
   * Prohibit other moves/copies

 Revision 614.78  2004/03/15 16:39:27  error
 * Multiple files: Set permissions on existing Citadel directories to prevent
   access to the database by local unix users to prevent direct database reads.

 Revision 614.77  2004/03/15 16:36:50  error
 * Multiple files:  Convert most remaining client code to use new Citadel IPC
   functions.  A few bits remain and will be converted when the chat system
   is rewritten.

 Revision 614.76  2004/03/14 06:35:46  ajc
 * Cache IMAP "BODY" fetches in an already-converted format.  This speeds
   up clients like Mozilla and Thunderbird that request big messages in
   chunks -- we don't have to load and convert the message every time
   another chunk is requested.
 * Fixed a potential memory leak in fetch

 Revision 614.75  2004/03/13 23:26:10  ajc
 * imap_fetch.c: don't fetch the message from disk at all for UID and FLAGS
   fetch items.  (It's rare, but sometimes IMAP optimizations are actually
   possible, despite IMAP's brain-dead design!)

 Revision 614.74  2004/03/13 22:52:05  ajc
 * Documented the GNET and SNET commands  :)

 Revision 614.73  2004/03/12 19:35:13  error
 * routines2.c: Fix two typos

 Revision 614.72  2004/03/12 19:32:09  error
 * routines2.c: Convert multiple functions to use new IPC code (almost done!)

 Revision 614.71  2004/03/12 19:30:06  error
 * citadel.c: main(): Fix error response for RCHG, HCHG and UCHG commands.

 Revision 614.70  2004/03/12 19:28:04  error
 * Implement GNET/SNET commands in IPC code; provide a CtdlIPC_delete();
   emit warnings when client code uses CtdlIPC_getline() or CtdlIPC_putline()
   (These are reserved and should not be used by client code.)

 Revision 614.69  2004/03/10 04:50:04  ajc
 * serv_expire.c: auto-purge any Citadel account that is associated with
   a Unix account that no longer exists.

 Revision 614.68  2004/03/09 20:46:33  error
 * Add a Linux Software Map file

 Revision 614.67  2004/03/08 05:26:49  ajc
 * When the last argument of an IMAP APPEND command is a binary literal
   (i.e. all the time), the client will still be sending a CRLF after the
   literal.  Added an extra client_read() after the message input to absorb
   that extra CRLF, to keep client and server from getting out of sync.

 Revision 614.66  2004/03/06 05:09:04  ajc
 * serv_smtp.c: don't offer the STARTTLS extension if TLS is already
   active.  Done for to-the-letter compliance with RFC2487.

 Revision 614.65  2004/03/06 02:03:39  ajc
 * serv_smtp.c: do not offer the PIPELINING extension when TLS is
   running.  Doing so causes sessions to hang unexpectedly.

 Revision 614.64  2004/03/05 23:21:25  ajc
 * Added some more debugging to serv_smtp.c to try to figure out the
   problem with Postfix TLS hanging during send

 Revision 614.63  2004/03/05 03:36:03  ajc
 * Remove the automatic population of A and N fields we recently added
   when found to be NULL.  It was confusing the output functions into
   thinking that Internet messages were local.

 Revision 614.62  2004/03/03 05:43:29  ajc
 * Generate better output when a user is kicked out of chat

 Revision 614.61  2004/03/01 22:36:14  ajc
 * Implement the NAMESPACE extension of IMAP (RFC 2342)

 Revision 614.60  2004/03/01 17:47:28  error
 * Fix missing arg in m<Y> next rewrite.

 Revision 614.59  2004/03/01 17:33:30  error
 * Rewrite m<Y> next function; now uses CtdlIPC code.

 Revision 614.58  2004/03/01 04:08:34  ajc
 * Revoke access to room when /kicked

 Revision 614.57  2004/02/29 23:26:48  ajc
 * Added /kick command to chat.  /ban coming soon.

 Revision 614.56  2004/02/28 16:37:41  ajc
 * setup.c: removed an unused temp file name variable

 Revision 614.55  2004/02/27 17:23:21  error
 * Use syslog-compatible logging levels in lprintf(); the loglevel chosen
   is passed directly to syslog().  The available levels are docuemnted in
   sysdep_decls.h.

 Revision 614.54  2004/02/27 04:29:00  ajc
 * Fixed a bug in PLAIN authentication

 Revision 614.53  2004/02/26 22:03:24  error
 * ;Goto floor now goes to first KNOWN room, or first room only if no known
   rooms on the floor

 Revision 614.52  2004/02/26 20:30:36  error
 * Populate IPC into a few more places to avoid a nasty crash at MORE prompt

 Revision 614.51  2004/02/26 20:17:29  error
 * Client stability and enhancements:
   * CtdlServInfo structure moved inside CtdlIPC; eliminates unnecessary
     global and makes IPC more self-contained
   * Removed redundant serv_ from variable names in CtdlServInfo struct
   * Send SIGHUP to process group when connection_died().  Kills self and
     children (e.g. external editor).

 Revision 614.50  2004/02/25 01:35:34  error
 Missing parameter to a printf() function.  No idea where it went, but I
 put it back.

 Revision 614.49  2004/02/24 05:09:06  ajc
 * Better validation of incoming network messages.

 Revision 614.48  2004/02/23 16:10:47  nbryant
 --disable-pie by default.

 Revision 614.47  2004/02/22 04:55:15  ajc
 * CSR code more or less in its final form for now.

 Revision 614.46  2004/02/22 04:42:05  ajc
 * Initial version of function to automatically generate a Certificate
   Signing Request if one is not present

 Revision 614.45  2004/02/20 19:29:05  error
 * Fix bug 112: .Goto allowed partial room match on forgotten rooms

 Revision 614.44  2004/02/20 02:55:18  ajc
 * Fixed one of the prompts (bugzilla #118)

 Revision 614.43  2004/02/19 04:12:56  ajc
 * Instead of doing the silly "Kolab reserved folder names" thing, instead
   implemented the Cyrus-compatible style of forcing all personal mailbox
   folders to be subfolders of INBOX.  But only for IMAP.

 Revision 614.42  2004/02/18 14:46:05  ajc
 * smtp: EHLO response now includes both "AUTH " and "AUTH=" output, due to
   the requirement to interoperate with brain-dead Microsoft shitware that
   doesn't follow the RFC.

 Revision 614.41  2004/02/17 16:56:51  ajc
 * During SSL initialization, create the "keys" directory if it does not
   exist ... generate a private key if that does not exist ... more code
   coming soon to generate CSR and self-signed cert.  Hard-coded pathnames
   have been moved to sysconfig.h.

 Revision 614.40  2004/02/17 04:47:22  ajc
 * Support PLAIN auth method in SMTP

 Revision 614.39  2004/02/17 03:53:11  ajc
 * New ICAL subcommand "sgi" to enable or disable automatic server
   generated invitations.  (WebCit wants sgi's but Kontaqt doesn't.)

 Revision 614.38  2004/02/16 21:54:22  ajc
 * Support POP3 over TLS

 Revision 614.37  2004/02/16 21:45:43  ajc
 * Implement RFC 2487 - SMTP Service Extension for Secure SMTP over TLS

 Revision 614.36  2004/02/16 21:02:28  ajc
 * IMAP and Citadel protocols now use the same code path for TLS negotiation

 Revision 614.35  2004/02/16 20:55:47  ajc
 * Genericized the Citadel API for TLS-enabling protocols

 Revision 614.34  2004/02/16 18:16:39  error
 * Remove some unnecessary and possibly hazardous debugging code leftover
   from debugging IMAP STARTTLS

 Revision 614.33  2004/02/16 18:14:00  error
 * Fixed IMAP STARTTLS; trouble was in client_read_ssl the whole time.
   It should now be possible to implement SSL/TLS for any protocol.

 Revision 614.32  2004/02/16 18:13:10  error
 * Log session IDs in syslog as well as stderr

 Revision 614.31  2004/02/16 15:06:44  error
 * Add specific error codes for every command on the wire protocol, so that
   clients can more easily determine what went wrong.  Partially updated
   session.txt (will finish it later).  This lets clients more easily
   determine what, if anything, went wrong with a particular command.

 Revision 614.30  2004/02/15 06:06:49  ajc
 * More work on IMAP TLS.  Still not working correctly.  :(  Added in
   support for server-side certificates.  Now instead of failing it hangs.

 Revision 614.29  2004/02/14 04:41:55  ajc
 * STARTTLS attempt #2.  Still disabled because it's broken.

 Revision 614.28  2004/02/13 20:51:13  error
 * Reset screen attributes before fork() so that external programs do the
   right thing

 Revision 614.27  2004/02/12 04:16:38  ajc
 * Support for PLAIN logins in IMAP (need to test!)

 Revision 614.26  2004/02/11 03:50:49  ajc
 * The groupware folders "Calendar", "Contacts", "Notes", and "Tasks" are
   now presented as subfolders of INBOX when using IMAP.  This is for the
   purpose of Kolab compatibility.

 Revision 614.25  2004/02/09 22:40:57  error
 * oops...also close the client on EOF/error on tty OUTPUT

 Revision 614.24  2004/02/09 22:26:17  error
 * Exit the citadel client when EOF reached on input - should prevent
   a process from hanging around when a telnet connection dies unexpectedly

 Revision 614.23  2004/02/08 05:29:34  ajc
 * Corrected various small syntax conversion problems that were keeping
   some vCards out of the LDAP directory.

 Revision 614.22  2004/02/07 05:28:10  ajc
 * Automatically purge files from ./network/spoolout which were intended
   for nodes which no longer exist

 Revision 614.21  2004/02/07 04:59:21  ajc
 * Cache the IGnet config in memory, avoiding a db fetch for every single
   operation which references it.  Invalidate the cache when a new config
   is uploaded.
 * When doing network spool for a room, automatically remove shares for
   Citadel nodes which no longer exist.

 Revision 614.20  2004/02/07 04:22:12  ajc
 * serv_network.c: fixed a problem with an uninitialized data structure

 Revision 614.19  2004/02/06 04:40:50  ajc
 * Suppress posting of messages to Aide> when the loopzapper catches dupes.
 * New command NSYN to sync the entire contents of a room to a specified
   network node.  This will be used to bring new nodes in sync with existing
   nodes.

 Revision 614.18  2004/02/06 03:58:10  ajc
 * Added some comments to the sample citadel-slapd.conf

 Revision 614.17  2004/02/06 03:52:46  ajc
 * Changed the domain name and password in the sample slapd.conf

 Revision 614.16  2004/02/05 05:20:20  ajc
 * Delete a user's LDAP entry when deleting the vCard.  NOT TESTED.

 Revision 614.15  2004/02/05 03:54:14  ajc
 * Completed the per-user initialization of LDAP entries.

 Revision 614.14  2004/02/04 18:27:25  error
 * Don't delete temp file when composing message until after server writes
   it successfully (or an error occurs in the client itself).

 Revision 614.13  2004/02/04 04:07:56  ajc
 * More work on LDAP connector

 Revision 614.12  2004/02/01 06:19:22  ajc
 * policy.c: fix bug in GetExpirePolicy() that was causing the default
   mailbox purge policy to be misinterpreted if it was set to "use
   system default"
 * serv_vcard.c: always set the Global Address Book room's expire policy
   to "never expire" and its default view to "address book"

 Revision 614.11  2004/01/31 05:44:29  ajc
 *** empty log message ***

 Revision 614.10  2004/01/27 19:56:29  ajc
 * serv_smtp.c: HELO and EHLO responses now reply with the detected
   IP address and reverse DNS lookup of the connecting host

 Revision 614.9  2004/01/24 05:47:50  ajc
 * Got Citadel talking to LDAP.  Still requires manual creation of schema
   and container entries, which I don't like; we'll have to fix that.  It
   also does not yet populate all fields.

 Revision 614.8  2004/01/19 21:01:15  error
 * Clear out the autom4te.cache when bootstrapping to avoid autoheader problems

 Revision 614.7  2004/01/19 04:44:11  ajc
 * Further work on creating LDAP entries.

 Revision 614.6  2004/01/18 21:04:40  ajc
 * Reworked vCard etc. functions for addition of new vCard data to LDAP

 Revision 614.5  2004/01/17 22:57:54  ajc
 * Added "Base DN" "Bind DN" and "Bind DN password" config options
 * serv_ldap.c: added.  (New module implementing the LDAP Connector)
 * Upon startup, connect to LDAP directory and bind to it.  Unbind at exit.

 Revision 614.4  2004/01/17 04:23:28  ajc
 * Sysconfig commands for specifying where the LDAP server lives

 Revision 614.3  2004/01/17 03:26:17  ajc
 * Changed a diagnostic in msgbase.c to a higher logging level so it doesn't
   make as much noise.
 * Setup now adds "-x3 -llocal4" to the default inittab line.

 Revision 614.2  2004/01/17 03:17:34  ajc
 * New server command-line option "-l" to send log output to the host
   computer's syslog facility instead of to a trace file.

 Revision 614.1  2004/01/02 22:13:59  ajc
 * Link to OpenLDAP client library (-lldap) and set HAVE_LDAP if present.

 Revision 614.0  2003/12/30 04:13:40  ajc
 * THIS IS 6.14

 Revision 613.12  2003/12/30 04:13:31  ajc
 * Commented out one of the supplied .RU commands in citadel.rc

 Revision 613.11  2003/12/24 05:02:49  ajc
 * Allow a separate default message expire policy for mailbox rooms.  The
   floor default setting is now ignored for mailboxes.  Updated client and
   documentation to match.

 Revision 613.10  2003/12/23 03:53:49  ajc
 * Expanded the size of all string fields in "struct CtdlIPCMessage" to
   the standard SIZ length, in order to accomodate "weird" messages
   without crashing.  Also converted many strcpy()'s to safestrncpy()'s
   while populating this data structure, to avoid overruns.

 Revision 613.9  2003/12/22 16:49:36  error
 * Add Preview for Mac OS X as an image viewer option in citadel.rc

 Revision 613.8  2003/12/22 15:30:53  error
 * Fork image viewer process in background (works now); suppress
   stdout/stderr from the image viewer

 Revision 613.7  2003/12/21 20:02:28  error
 * When viewing images, append the image filename to the temp filename so
   that less-intelligent viewers can deal with the images

 Revision 613.6  2003/12/21 19:07:28  error
 * Start a new thread and fork() for image viewer (broken/disabled; for
   refernce only)

 Revision 613.5  2003/12/21 01:23:12  nbryant
 added some additional comments to citadel.rc concerning possible image viewers

 Revision 613.4  2003/12/21 00:44:19  error
 * Fix for idle_threshold being ignored when reading citadel.rc

 Revision 613.3  2003/12/21 00:37:54  error
 * Fix for crash when downloading multiple attachments

 Revision 613.2  2003/12/21 00:19:41  error
 * Image viewer code for the text client.  Hit 'I' to view an image
   attached to a file.  Hey aahz, this code actually works!

 Revision 613.1  2003/12/19 04:33:52  ajc
 * Changes to auto-expunge algorithm to support correct behavior in
   some IMAP clients while moving messages.

 Revision 613.0  2003/12/15 16:37:01  ajc
 * THIS IS 6.13

 Revision 612.14  2003/12/15 16:33:47  ajc
 * Added support for "Cc" and "Bcc" IMAP SEARCHes
 * Bumped the version number to 6.13

 Revision 612.13  2003/12/11 04:21:23  ajc
 * Removed a "FIXME this needs blah blah" comment because blah blah has
   since been completed

 Revision 612.12  2003/12/11 04:06:55  nbryant
 configure.ac: add support for position-independent executables. can be disabled
 with --disable-pie

 Revision 612.11  2003/12/11 03:44:18  nbryant
 domain.c: include <arpa/nameser_compat.h> if present
 configure.ac: check for <arpa/nameser_compat.h>

 Revision 612.10  2003/12/10 03:58:26  ajc
 * Completed a more accurate implementation of the ENVELOPE fetch in IMAP.

 Revision 612.9  2003/12/09 06:39:19  nbryant
 did a minimal amount of ANSIfication without changing anything that would
 affect the compiler output in any way. I was just trying to get enough warnings
 enabled to be able to tell if there are any nested functions lurking about...

 If we get rid of nested functions we can get some big security gains under
 Fedora.

 Revision 612.8  2003/12/09 05:12:49  ajc
 * Don't give away anonymous poster information in IMAP or RFC822

 Revision 612.7  2003/12/09 04:50:16  ajc
 * Painstakingly combed through IMAP headers and responses for two hours in
   an attempt to figure out why the horrendous pile of crap called Microsoft
   Outlook counts the correct number of messages but wasn't displaying any
   of them.  Eventually determined that it was ONE EXTRA SPACE after the
   closing paren of the ENVELOPE output that caused Outlook to totally ignore
   all the messages.  This only reinforces my belief that anything designed
   in the state of Washington is utter and complete CRAP.
 * Corrected the above glitch.  Outbreak works properly now.

 Revision 612.6  2003/12/09 03:56:29  ajc
 * user_ops.c: missing string declaration or something?

 Revision 612.5  2003/12/08 17:41:01  ajc
 * serv_smtp.c: validated and documented compliance with a big pile of RFC's
   which are in one way or another related to SMTP.

 Revision 612.4  2003/12/07 19:59:13  error
 * Minor enhancements to a few of the trace file entries

 Revision 612.3  2003/12/07 19:57:48  error
 * Added some useless comments to some of the techdocs/ files

 Revision 612.2  2003/12/07 19:56:24  error
 * CREU command now allows specification of an initial password

 Revision 612.1  2003/12/04 04:20:08  ajc
 * Incoming RFC822 messages get the To: field translated directly to
   a Citadel <R> field.

 Revision 612.0  2003/12/01 04:11:48  ajc
 * THIS IS 6.12

 Revision 611.10  2003/11/30 03:43:34  ajc
 * Support IMAP \Answered flag

 Revision 611.9  2003/11/22 03:30:14  ajc
 * Commented out the call to the 'high speed download' function and
   uncommented the call to the original 'chatty' download function.  Some
   users were reporting lockups.

 Revision 611.8  2003/11/14 20:15:44  ajc
 * Reversed the previous change because it made things b0rken

 Revision 611.7  2003/11/14 20:04:28  ajc
 * IMAP FETCH ENVELOPE now outputs more fields correctly.

 Revision 611.6  2003/11/14 03:49:54  ajc
 * MUCH faster implementation of rfc822_fetch_field()

 Revision 611.5  2003/11/13 04:25:38  ajc
 * mime_parser.c: handle multipart *much* more efficiently now.  Instead
   of scanning line by line, we snag the boundaries using the Boyer-Moore
   algorithm.

 Revision 611.4  2003/11/12 04:28:22  ajc
 * Replaced bmstrcasestr() with a more generic function, bmstrstr(), which
   can be supplied with *any* strncmp()-compatible compare function.

 Revision 611.3  2003/11/08 06:46:08  ajc
 * IMAP FETCH BODY<xxx.yyy> was outputting the offset and length in the
   wrong order.  Fixed.

 Revision 611.2  2003/11/08 06:29:47  ajc
 * IMAP FETCH FLAGS ... removed extra trailing space after last flag

 Revision 611.1  2003/11/06 04:25:30  ajc
 * For certain IMAP outputs (such as BODYSTRUCTURE), buffer all output until
   we've got the whole thing, then spew it all at once to avoid lots of
   tiny TCP packets.
 * New API calls buffer_output() (to start buffering) and unbuffer_output()
   (to dump the buffer and stop buffering) are available to do this.

 Revision 611.0  2003/10/21 04:20:41  ajc
 * THIS IS 6.11

 Revision 610.22  2003/10/17 02:18:40  ajc
 * Replaced the DEFAULT_ENTRY definition in sysconfig.h with a new option
   "reply_with_external_editor" in citadel.rc.

 Revision 610.21  2003/10/14 03:09:47  ajc
 * routines2.c: fixed an off-by-one error in filename sanitization for
   client file uploads

 Revision 610.20  2003/10/10 05:43:57  ajc
 * citadel.c:  Reworded one of the prompts

 Revision 610.19  2003/10/07 15:56:17  ajc
 * Better handling of incoming Internet addresses consisting of an address
   in angle brackets with no name outside the brackets.

 Revision 610.18  2003/09/21 04:10:56  ajc
 * messages.c: don't crash when a message contains more than MAXURLS of
   embedded URL's.  Omit them instead.
 * citadel.rc: updated the sample browser remoting command for Macintosh
   based on a Mr.T suggestion
 * techdoc/session.txt: redid the writeup for the MOVE command (fixed a
   mistake and made the writeup clearer)

 Revision 610.17  2003/09/17 21:40:13  ajc
 * html.c: don't truncate messages when wacky characters are
   encountered.  Substitute '?' characters instead.  Now the Citadel
   community can complain about b0rken MS ASCII.

 Revision 610.16  2003/09/13 02:53:04  ajc
 * More intelligent selection of directory to save attachments

 Revision 610.15  2003/09/12 15:48:55  ajc
 * serv_smtp.c: remember whether we're SMTP or LMTP across a RSET command.

 Revision 610.14  2003/09/09 02:42:28  ajc
 * Documentation update for LMTP

 Revision 610.13  2003/09/09 01:47:02  ajc
 * Suppress domain forgery check when using LMTP

 Revision 610.12  2003/09/08 18:48:14  ajc
 * The unix domain socket used for citmail to talk to citserver now
   implements LMTP (RFC2033).  Other LMTP client implementations (such as
   Postfix) can also use the socket, eliminating the need for citmail.

 Revision 610.11  2003/09/07 03:29:25  ajc
 * serv_network.c: aggressively poll any node for which we have data

 Revision 610.10  2003/09/06 02:04:54  ajc
 * Don't reset "last poll" time when not doing the "full" net processing

 Revision 610.9  2003/09/03 03:29:01  ajc
 * syslog messages now say session started/ended instead of ended/ended

 Revision 610.8  2003/09/03 03:24:40  ajc
 * client: remove rooms from the march list when zapping them (or the floors
   they're on)

 Revision 610.7  2003/09/02 15:05:34  ajc
 * Repaired an incorrect adjustment of floor reference counts when a room was
   moved to a different floor.  (Thanks to Winzlo for reporting this.)

 Revision 610.6  2003/09/02 14:49:05  ajc
 * Improve client chat mode reliability and performance by assuming that all
   server output will be in complete lines.  (This is a safe assumption.)

 Revision 610.5  2003/09/02 04:04:39  ajc
 * Don't try to directly spool to non-neighbor nodes

 Revision 610.4  2003/08/29 08:21:57  ajc
 * Do network runs more proactively.  The inbound queue is now scanned
   every minute instead of every poll frequency.

 Revision 610.3  2003/08/21 23:26:06  ajc
 * setup.c: offer to disable "exim" if found

 Revision 610.2  2003/08/20 03:51:08  ajc
 * Bumped the internal version number to 6.11

 Revision 610.1  2003/08/20 03:36:13  ajc
 * Create a "Contacts" room for personal address book use

 Revision 610.0  2003/08/17 03:00:37  ajc
 * THIS IS 6.10

 Revision 609.2  2003/08/17 02:58:46  ajc
 * Changed internal version number to 6.10

 Revision 609.1  2003/08/17 02:46:37  ajc
 * Don't fail on "data files too old to be upgraded" for NEW installations

 Revision 609.0  2003/08/14 03:40:55  ajc
 * THIS IS 6.09

 Revision 608.21  2003/08/13 18:08:24  ajc
 * Removed the last vestiges of the now-obsolete global networking password

 Revision 608.20  2003/08/13 14:36:04  ajc
 * "make install" now installs the contents of docs/ as well as README.txt

 Revision 608.19  2003/08/13 14:30:47  ajc
 * Temporarily disabled IMAP TLS support due to the discovery of some
   compatibility problems.  It will be re-enabled when they are fixed.

 Revision 608.18  2003/08/12 00:39:35  ajc
 * setup now knows about lots of other mail programs it can disable (various
   flavors of Cyrus and Courier, for example)
 * setup now ensures that it really did startup/shutdown the Citadel service,
   and displays an error message when that fails.
 * "make install" now instructs the user run "setup"
 * Removed an old GDBM reference from the documentation

 Revision 608.17  2003/08/05 03:06:58  ajc
 * Added a README.txt file because some Joker kept bugging me about it
 * Reloaded the code that extracts embedded URL's to a place where it'll
   get picked up on *every* message, not just the old variformat stuff

 Revision 608.16  2003/08/03 17:51:52  ajc
 * Clear out all masqueraded wholist fields when logging out, in case another
   user logs in without reconnecting.

 Revision 608.15  2003/07/30 20:36:18  ajc
 * Allow users to authenticate with either their display name or any valid
   e-mail address which belongs to them.  Applies to all protocols.

 Revision 608.14  2003/07/30 03:54:34  ajc
 * Fixed a small client bug in the purge hour setting

 Revision 608.13  2003/07/30 03:47:53  ajc
 * Eliminated the EXPI command
 * Completed the configuration item of "purge hour"
 * Auto-purger now runs as a scheduled job

 Revision 608.12  2003/07/30 00:26:50  ajc
 * Removed the "weekly" script and began installing a facility to allow
   database maintenance to happen automatically.  (One Step Install can't
   require the sysadmin to know cron)

 Revision 608.11  2003/07/28 04:17:24  ajc
 * migratenet.c: removed.  Adjusted Makefile.in accordingly.

 Revision 608.10  2003/07/27 21:15:23  ajc
 * Implemented all IMAP date-based search criteria.  Note that Citadel does
   not record an "internal date" of a message, so the "sent date" and "internal
   date" searches perform identically.
 * Date search comparisons available: "before," "on," or "on or after."  Yet
   another example of IMAP's gratuitous complexity.

 Revision 608.9  2003/07/26 04:49:40  ajc
 * Implemented a bunch of the IMAP SEARCH keywords

 Revision 608.8  2003/07/26 04:28:44  ajc
 * tools.c: added bmstrcasestr(), a Boyer-Moore, case-insensitive string search
 * imap_search.c: implement BODY criterion in search command

 Revision 608.7  2003/07/24 04:57:53  ajc
 * Began implementation of the IMAP SEARCH command.  Basically just got the
   parser working and stubbed all the search criteria.  Next step is to
   implement the functionality of each criterion keyword.
 * Implemented the functionality for the ALL (wow!) and UID criteria.

 Revision 608.6  2003/07/23 03:57:05  ajc
 * Optimized the updating of visit records in several places by checking to
   see if they've changed and aborting the db update if they haven't.
 * Functions which read/write user records are now slightly faster due to
   an improvement of the algorithm used to generate the database keys.

 Revision 608.5  2003/07/22 03:07:50  nbryant
 fix build for Solaris

 Revision 608.4  2003/07/20 20:46:06  ajc
 * build system: link Berkeley DB only to the server, not to the client & utils

 Revision 608.3  2003/07/20 03:51:46  ajc
 * setup.c: offer to disable sendmail, postfix, and qmail if found (only if
   using the /etc/init.d type of startup scripts)

 Revision 608.2  2003/07/20 03:08:22  ajc
 * setup.c: offer to hack /etc/xinetd.d/telnet

 Revision 608.1  2003/07/15 14:54:09  ajc
 * Killed off CtdlGetDynamicSymbol() and just put all the symbols in server.h

 Revision 608.0  2003/07/15 04:12:52  ajc
 * THIS IS 6.08

 Revision 607.23  2003/07/14 17:26:42  ajc
 * strcpy() --> safestrncpy() in a few other random places

 Revision 607.22  2003/07/14 17:12:28  ajc
 * Replaced strcpy() with safestrcpy() in cmd_setp()

 Revision 607.21  2003/07/14 03:07:05  ajc
 * ipgm_secret is no longer set during setup.  Now it is set at server startup
   and automatically changes whenever an IPGM command is executed.
 * Upon an IPGM authentication failure, wait 5 seconds before displaying an
   error, then disconnect the session.

 Revision 607.20  2003/07/13 04:58:35  ajc
 * Allow connect on unix domain sockets to Citadels in other directories
 * sendcommand now uses unix domain sockets instead of the network
 * Do not allow IPGM command to run on the network -- unix domain sockets only

 Revision 607.19  2003/07/11 22:33:02  ajc
 * Ignore comments in public_clients file

 Revision 607.18  2003/07/11 22:30:19  ajc
 * Automatically add 127.0.0.1 and config.c_fqdn to public_clients

 Revision 607.17  2003/07/11 22:19:49  ajc
 * is_public_client() now caches the IP addresses of all hosts in
   the public_clients file, at server startup and whenever the file is
   modified.  This eliminates the need to make lots of calls to the
   resolver library every time.

 Revision 607.16  2003/07/10 05:51:46  ajc
 * Added cs_addr field to struct CitContext -- holds a dotted quad string
   of the user's source IP (if applicable).  It's big enough to hold other
   types of address strings in the future (such as IPv6).
 * locate_host() populates cs_addr when on a network connection.
 * serv_smtp.c now saves the IP address in the proper place in
   the Received: header.
 * is_public_client() no longer accepts a hostname.  It just looks at
   CC->cs_host instead.

 Revision 607.15  2003/06/29 19:54:39  ajc
 * Renamed "struct user" to "struct ctdluser"
 * Renamed "struct room" to "struct ctdlroom"

 Revision 607.14  2003/06/29 04:13:32  ajc
 * Renamed:
   S_QUICKROOM to S_ROOMS
   S_USERSUPP to S_USERS
   CDB_QUICKROOM to CDB_ROOMS
   CDB_USERSUPP to CDB_USERS

 Revision 607.13  2003/06/29 04:07:11  ajc
 * Globally renamed "struct quickroom" and "struct usersupp" to
   "struct room" and "struct user".

 Revision 607.12  2003/06/29 04:06:41  ajc
 *** empty log message ***

 Revision 607.11  2003/06/29 03:58:41  ajc
 * Renamed fields in 'struct config' which are no longer relevant (but kept
   them in place to avoid corrupting everyone's data files).  Removed them
   entirely from serv_vandelay.c.

 Revision 607.10  2003/06/28 05:12:56  ajc
 * Bump internal version number to 6.08
 * Use (-1) instead of CTDLUID as the uid of user records which exist only
   in Citadel and not in the system password database.  serv_upgrade also
   changes this setting for ALL such users, the first time the server is run
   after upgrading to 6.08

 Revision 607.9  2003/06/27 22:19:31  error
 * routines2.c: Quick and dirty fix for Mail> becoming new after entering a
   mail message

 Revision 607.8  2003/06/26 01:50:14  ajc
 * Compiler warning fix for building setup without newt

 Revision 607.7  2003/06/25 15:40:57  ajc
 * setup now uses "newt" instead of "curses" to draw its pretty screens.
   Changes were made to setup.c as well as all the autoconf stuff.

 Revision 607.6  2003/06/25 02:40:05  ajc
 * Per-user and global flags calling for validation need to be set in the
   vCard upload thingie, not in the wrapper function that emulates the
   deprecated REGI server command.  Moved it to the correct place.

 Revision 607.5  2003/06/22 20:49:31  error
 * citadel.c: display idle times in wholist up to 999 days

 Revision 607.4  2003/06/19 04:12:30  ajc
 * domain.c: use qsort() to sort MX records by preference.  Why have a
   custom function when the operating system provides one for free?
 * serv_smtp.c: accept mail from "<>" (empty sender).  RFC1123 5.2.9 demands it.

 Revision 607.3  2003/06/19 03:55:22  ajc
 * Fixed a subtle but ugly bug in the SMTP sender that was causing it to
   ignore all successful connections except the last one.  Now when it gets
   a connection it uses it.

 Revision 607.2  2003/06/17 22:05:11  ajc
 * New format for wholist.

 Revision 607.1  2003/06/16 04:39:45  ajc
 * support.c: removed an unused variable
 * citadel.c: sort wholist by idle time (most recently active users first)

 Revision 607.0  2003/06/10 04:15:39  ajc
 * THIS IS 6.07

 Revision 606.10  2003/06/04 21:46:13  error
 * citadel.c: Added a missing #ifdef __CYGWIN__

 Revision 606.9  2003/06/04 21:39:27  error
 * Numerous fixups needed for Windows port:
   * cmd_mesg() rewritten to use the stat() syscall to find files
   * really_do_keepalive() does not try to send if we are not connected
   * Misc. fixups; added Windows readme file

 Revision 606.8  2003/06/04 21:30:39  error
 * messages.c: When file attachment has no name, use message subject as name

 Revision 606.7  2003/06/02 23:11:00  error
 * Checkin installer script citadel.nsi for NSIS Windows installer

 Revision 606.6  2003/06/02 03:01:22  ajc
 * citserver.c: added another trace message to is_public_client().  Uncensored
   has crashed twice in this function, for no apparent reason.
 * internet_addressing.c: don't attempt to look up the specified Internet
   address in a Citadel directory, if the supplied address has no hostname
   portion.  (Fixes a crash in the calendar service when alias() attempts to
   look up potentially nonexistent attendees during event validation)

 Revision 606.5  2003/05/28 03:08:38  ajc
 * msgbase.c: validate_recipients() now rejects Internet addresses in
   domains belonging to the local system or an attached Citadel network,
   when it fails to translate to a Citadel address.

 Revision 606.4  2003/05/26 05:36:25  ajc
 * citadel.h: changed internal version number to 6.07
 * file_ops.c: #include <ctype.h>
 * migratenet.c: #include <ctype.h>

 Revision 606.3  2003/05/26 05:30:18  ajc
 * serv_calendar.c: fixed a bug in the "ICAL freebusy" subcommand which
   caused it to always fetch the logged in user's freebusy times instead
   of the user specified.

 Revision 606.2  2003/05/22 13:34:30  ajc
 * database_sleepycat.c: added a couple of diagnostic messages to help
   troubleshoot problems with db-4.1.25 on Red Hat Linux 9
 * room_ops.c: call mkdir() instead of system() to create a directory

 Revision 606.1  2003/05/20 01:28:50  error
 * citadel.c: don't truncate roomname when user is idle, in long who list;
   display idle time in short who list

 Revision 606.0  2003/05/16 15:17:11  ajc
 * THIS IS 6.06

 Revision 605.50  2003/05/16 14:59:08  ajc
 * Removed the old rec_log() facility because the utilities which read its
   data are crufty, disused, and produce incorrect output.
 * Updated the documentation for the forthcoming 6.06 release.
 * public_clients: removed hostnames which crept in from development servers

 Revision 605.49  2003/05/16 04:07:37  nbryant
 make configure.ac automagically do the right thing on redhat 9
 remove crappy tolower() hack that was breaking the build on redhat 9
 (because the kerberized openssl includes ctype.h)

 before anyone yells at me, i did a little cvs archaeology on the tolower
 thing, and it appears to have been in the Citadel source base since,
 roughly, the Epoch. I doubt that the hack is needed any longer, and using
 the OS's tolower() will handle extended charsets.

 Anyone who is stupid enough to be using an OS with such a horribly broken
 tolower() can just scratch their head and figure things out for
 themselves. Citadel needs higher barriers to entry, there are too many
 weirdos on the IGnet lately. :-)

 Revision 605.48  2003/05/13 03:22:04  ajc
 * Implemented the IMAP STARTTLS command as specified in RFC 2595.

 Revision 605.47  2003/05/06 03:38:59  ajc
 * IMAP folder create and rename commands -- don't allow backslashes in
   folder names.

 Revision 605.46  2003/05/03 16:50:52  ajc
 * Change folder delimiter in room names from / to \

 Revision 605.45  2003/05/02 04:02:47  ajc
 * setup.c: allow specification of the Citadel system account by either
            username or uid
 * setup.c: tell init to re-read /etc/inittab by sending a SIGHUP to pid 1
            instead of hunting around for the correct init or telinit command
 * docs/citadel.html: documented the above change, and also rewrote some
                      other stuff to be less BBS-specific

 Revision 605.44  2003/04/30 16:16:13  ajc
 * Minor fix to ESMTP greeting (missing '-' screwed up pipelining)

 Revision 605.43  2003/04/30 15:46:27  ajc
 * Added the ENHANCEDSTATUSCODES extention to ESMTP.

 Revision 605.42  2003/04/29 04:41:22  ajc
 * Fix an off-by-one error in the mime parser

 Revision 605.41  2003/04/29 04:13:58  ajc
 * mime parser: look for the "name=" subfield tacked onto either
   Content-type or Content-disposition, wherever it finds it.

 Revision 605.40  2003/04/28 16:56:51  ajc
 * Added a site-configurable setting to suppress the automatic correction of
   forged From: lines from authenticated SMTP users, for those who prefer
   strict RFC compliance instead of common sense.

 Revision 605.39  2003/04/26 21:55:15  ajc
 * Anytime a socket connect() fails, CLOSE THE SOCKET before erroring out.
   Not doing so causes a file descriptor leak.

 Revision 605.38  2003/04/26 04:22:51  ajc
 * Removal of individual "remote" lines from message delivery instruction sets
   was corrupting the end of the set, rendering the "bounceto" line unusable
   and preventing bounces from being delivered.  Fixed this by explicitly
   adding a trailing newline when remove_token() strips it.

 Revision 605.37  2003/04/25 18:28:47  ajc
 * When rejecting spam, use SMTP error code 550, not 552

 Revision 605.36  2003/04/23 03:18:44  ajc
 * Allow IMAP STORE of more than one flag at a time (Mail.app from MacOS
   does this).

 Revision 605.35  2003/04/15 02:44:02  ajc
 * Do not allow incoming network polls while an outbound network processing
   session is in progress.
 * tools.c: removed collapsed_strcmp() as it is no longer used anywhere

 Revision 605.34  2003/04/08 04:03:37  ajc
 * Implemented the ".SILENT" protocol option in IMAP STORE.  Certain apps seem
   to want to use it.
 * Optimized the IMAP operations which scan for expunged and added messages.
   These loops no longer make multiple traversals through the message list.

 Revision 605.33  2003/04/07 05:02:23  ajc
 * Reworked all the "list rooms" operations so that they only require one
   pass through the database.
 * Repaired the "create floor" operation which was broken by the switch
   to the new IPC libray

 Revision 605.32  2003/04/02 13:33:28  ajc
 * Fixed output of "-0500" vs. "+0500" type of timezone stamps in RFC822.
   (I think they were reversed.)

 Revision 605.31  2003/04/01 05:01:08  ajc
 * sysdep.c: optimized MyContext() a bit, and inlined it.  A little profiling
   revealed that this function was getting used super heavily.
 * tools.c: removed the older, slower string tokenizer functions that
   weren't getting used anymore.

 Revision 605.30  2003/03/31 04:55:58  ajc
 * Repaired the formatting of text/plain messages with blank lines.

 Revision 605.29  2003/03/30 06:16:52  ajc
 * Optimized CtdlReadMessageBody() and also gave it an option to store
   messages with CRLF newlines instead of LF.  This option is used when
   reading SMTP in order to keep Pine from barfing on LF-terminated newlines
   while decoding quoted-printable.  *sigh*

 Revision 605.28  2003/03/26 05:17:12  ajc
 * Downloading of attachments was completely broken by the change to the new
   protocol library.  Located and fixed bugs.
 * html.c: increased the conversion buffer size

 Revision 605.27  2003/03/24 03:42:14  ajc
 * Add the ability to quit out while validating users.  Also added a help
   option listing available access levels.  This closes Bugzilla #37.

 Revision 605.26  2003/03/22 05:38:23  ajc
 * During startup, display the version string from the Berkeley DB library.
 * When deleting a mailbox, don't reveal the namespace prefix to the user.

 Revision 605.25  2003/03/20 08:37:04  error
 * connection_died(): Don't crash before printing message, crash afterward.
   (Bugzilla id 111)

 Revision 605.24  2003/03/19 11:48:06  error
 * entmsg(): Don't corrupt the msg_arr[] when replying to a message in a room

 Revision 605.23  2003/03/19 04:04:56  ajc
 * Dropped support for GDBM and for versions of Berkeley DB prior to 4.1

 Revision 605.22  2003/03/18 05:15:05  ajc
 * Applied a patch sent in by Clint Adams <schizo@debian.org> to handle
   building on Debian's unstable with Berkeley DB 4.1, compiled with
   the --with-db-uniquename when errno is a macro (as in glibc 2.3.1).

 Revision 605.21  2003/03/17 11:56:35  error
 * Fix <G>oto in alternate_semantics - it actually marks messages read now.

 Revision 605.20  2003/03/17 04:03:14  ajc
 * Supply a NULL argument as the second argument to accept() instead of an
   empty buffer whose contents we never even look at.

 Revision 605.19  2003/03/16 05:07:12  ajc
 * Fixed the bug that caused freebusy publishes to crash.

 Revision 605.18  2003/03/15 23:05:08  ajc
 * Rewrite of calendar-to-freebusy conversion functions.  This one creates
   freebusy in the correct format, instead of what I mistakenly assumed it
   was supposed to look like.  :(

 Revision 605.17  2003/03/15 22:31:06  ajc
 * When scanning a user's Calendar> room for calendar events, search for
   MIME parts of type text/calendar instead of assuming that the event will
   always appear as part "1"

 Revision 605.16  2003/03/14 05:08:25  ajc
 * Clone calendar subcomponents before encapsulating in case they already
   happen to have a parent.

 Revision 605.15  2003/03/13 05:48:33  ajc
 * Move towards storing calendar objects as fully encapsulated VCALENDAR
   components instead of just VEVENT subcomponents.

 Revision 605.14  2003/03/12 03:33:54  ajc
 * More changes to the handling of RFC822 headers with regard to
   splitting up the headers and body.  (Blank lines and such.)

 Revision 605.13  2003/03/11 06:23:50  ajc
 * More accurate handling of IMAP FETCH xx BODYSTRUCTURE command.  This should
   make Pine happier with multipart messages.

 Revision 605.12  2003/03/10 05:38:21  ajc
 * Tweaks to msgbase.c and imap_fetch.c to fix slightly incorrect byte counts
   reported in the numerous variations of IMAP FETCH.  This silences a number of
   error messages reported by Pine.

 Revision 605.11  2003/03/10 03:40:08  ajc
 * Fixed bug that caused segv when <R>eplying to certain messages

 Revision 605.10  2003/03/07 17:39:45  ajc
 * Remove the "modules" directory

 Revision 605.9  2003/03/06 04:58:11  ajc
 * When entering a message into Citadel with authenticated SMTP, stamp the
   room name as MAILROOM (Mail>) so it doesn't end up with an ugly moniker
   like "0000058008.Sent Items"

 Revision 605.8  2003/03/05 04:55:32  ajc
 * serv_smtp.c: when authenticated, do not log out when a RSET command is
   issued.  This breaks clients which carelessly issue RSET before each
   message, such as Pine.
 * serv_smtp.c: when authenticated, ignore envelope FROM:<xxx> sender name
   and replace with address of user who authenticated.  Also fixes Pine.
 * serv_smtp.c: when authenticated, not only do we continue our glorious
   RFC-violating condition of replacing the From: header with the user who
   is logged in, but we also stamp it with their preferred Internet address
   for outbound mail (which is kept in CC->cs_inet_email).

 Revision 605.7  2003/03/03 04:09:21  ajc
 * When creating a user, hide the "My Citadel Config>" room of the user being
   created, not the user currently logged in.

 Revision 605.6  2003/03/01 18:18:03  ajc
 * commands.c: alternate_semantics might have been getting set where it
   should not.  Fixed.

 Revision 605.5  2003/03/01 17:25:33  ajc
 * citadel.rc default for alternate_semantics = no

 Revision 605.4  2003/02/23 05:34:45  ajc
 * When the <N>ew command is used to read new messages, the next "lazy"
   command (space bar) now always does <G>oto, regardless of the last one.

 Revision 605.3  2003/02/21 04:24:51  ajc
 * Completed the "ICAL freebusy" subcommand, which fetches free/busy times
   for any requested user (output in stripped-down VCALENDAR format)

 Revision 605.2  2003/02/20 04:51:17  ajc
 * Began framing up a command to view other users' free/busy times

 Revision 605.1  2003/02/19 22:38:43  ajc
 * Changed socklen_t to int (to allow compile on Macintosh)

 Revision 605.0  2003/02/19 03:46:01  ajc
 * THIS IS 6.05

 Revision 601.134  2003/02/17 05:23:20  ajc
 * Fixed a small bug in the GDBM backend (deprecated, but the bug was very
   obvious and a kind user pointed it out)
 * Removed vestiges of setjmp/longjmp from the client
 * When doing .TS, don't get caught in a constant-logout loop

 Revision 601.133  2003/02/14 16:12:04  ajc
 * Added support for any standard RBL

 Revision 601.132  2003/02/13 22:33:41  ajc
 * Fixed algorithm for reporting "last login"

 Revision 601.131  2003/02/13 21:52:13  ajc
 * IMAP service: fixed bug that was causing the wrong messages (usually *all*
   messages) to be flagged when a STORE or COPY operation was performed.

 Revision 601.130  2003/02/13 20:13:51  ajc
 * Applied "multi editor" patch supplied by georbit

 Revision 601.129  2003/02/12 04:51:44  ajc
 * More docs update

 Revision 601.128  2003/02/10 04:50:33  ajc
 * More documentation updates

 Revision 601.127  2003/02/09 03:38:28  ajc
 * Updated the documentation some more

 Revision 601.126  2003/01/25 07:37:15  error
 * whobbs.c: Print error message if we can't connect to server

 Revision 601.125  2003/01/23 05:50:59  ajc
 * citadel.c: hitting enter without entering a room name in the <.G>oto and
   <.S>kip commands now does nothing instead of exhibiting undefined behavior

 Revision 601.124  2003/01/23 04:42:33  ajc
 * Began reviewing, updating, and HTML-izing the documentation

 Revision 601.123  2003/01/22 03:53:24  ajc
 * More reliablility hacks for autologin mode, and better quarantine of
   autologin code when it's disabled.

 Revision 601.122  2003/01/21 04:42:12  ajc
 * Moved login-to-screenname code ... no longer part of CtdlTryExistingUser(),
   it's now part of getuser() so it always gets called, even when looking up
   addresses for mail and stuff.

 Revision 601.121  2003/01/19 08:59:02  error
 * Remove irrelevant file hpsux.h (was part of dynloader)

 Revision 601.120  2003/01/19 05:55:59  ajc
 * serv_calendar.c: allocate dynamic symbol at module startup, not during
   each session startup.  This error was causing crashes (of course).

 Revision 601.119  2003/01/18 06:18:20  ajc
 * whobbs.c: line up the columns better
 * routines2.c: when doing <.ASI> command, entering a blank hostname causes
   it to abort instead of adding a new host.  Also strip l/t whitespace.

 Revision 601.118  2003/01/17 22:36:56  ajc
 * This commit is a figment of your imagination.

 Revision 601.117  2003/01/17 21:50:27  ajc
 * Began making some build changes for Cygwin compatibility

 Revision 601.116  2003/01/17 10:06:54  error
 * fmout():  Ignore \r (carriage return) characters in messages

 Revision 601.115  2003/01/17 04:36:04  ajc
 * whobbs.c: fix bad call to CtdlGetServInfo() so we can get our session ID
   and the name of the Citadel site

 Revision 601.114  2003/01/16 21:16:23  error
 * CtdlIPCGetMessages(): Fix message read failing on first attempt to read
   messages

 Revision 601.113  2003/01/16 10:04:03  error
 * ENT0 command: changed post-as username from arg 4 to arg 5; 4 was already
   used as the message subject
 * Convert message entry and reading to new IPC code
 * Minor bugfixes throughout IPC code

 Revision 601.112  2003/01/16 04:17:02  ajc
 * citadel_ipc.c: when issuing a SPEX command, send the string value for
   room/floor/site instead of the number.  citserver wants a string.
 * ipc_c_tcp.c: don't call SSL_shutdown() from connection_died().  This just
   causes it to segfault because there's no valid socket.

 Revision 601.111  2003/01/16 03:41:45  ajc
 * Renamed "dynloader" to "serv_extensions" globally.  We don't want people
   to think we have a dynamic loader, do we?  :)
 * serv_*_init() is now declared in serv_extensions.h.  This is definitely the
   wrong place for it.  It's there temporarily until we decide on a new
   convention for server extensions.
 * bootstrap: more reliable detection of Red Hat Linux 7.3 -- more specifically,
   fewer fireworks when the host system is *not* Red Hat Linux.

 Revision 601.110  2003/01/15 16:34:53  ajc
 * bootstrap: don't check /etc/redhat-release if it doesn't exist (which is
   the case on all non Red Hat Linux systems)

 Revision 601.109  2003/01/15 05:57:09  ajc
 * First cut at static-linking the citserver.  Ripped out libtool and
   replaced the dynloader paradigm with "server extensions" paradigm (static
   linked, but still very loosely coupled by way of using the API's to
   register callbacks and commands etc.)   Needs more testing.

 Revision 601.108  2003/01/14 04:12:26  ajc
 * Set default view for new user Calendar> rooms to 3 (calendar)
 * Set default view for new user Tasks> rooms to 4 (tasks)

 Revision 601.107  2003/01/13 17:04:19  ajc
 * Add a sample URL view command for Macintosh in the included citadel.rc

 Revision 601.106  2003/01/13 02:55:07  ajc
 * Do calendar support only if libical header *and* library are present

 Revision 601.105  2003/01/12 23:07:40  ajc
 * configure.ac: Attempt to check for libical 0.24 or newer

 Revision 601.104  2003/01/12 05:01:58  ajc
 * Start marking things for 6.05 release

 Revision 601.103  2003/01/12 04:48:23  ajc
 * Display message subjects in magenta instead of white

 Revision 601.102  2003/01/06 09:17:10  error
 * Remove non (yet) existent sstring.c dependency from Makefile.in

 Revision 601.101  2003/01/05 07:35:03  error
 * Add an AllFloors parameter to the citadel IPC API

 Revision 601.100  2003/01/01 08:07:46  error
 * commands.c: bugfix for reading from FILE pointers; affected quote, print
   and external editor.

 Revision 601.99  2002/12/28 05:33:29  ajc
 * ical_dezonify: set is_utc=1 even if we didn't convert from some arbitrary
   timezone.  Presumably this means the time was already UTC, and we really
   need that "Z" to get slapped on it.

 Revision 601.98  2002/12/28 05:12:06  ajc
 * Yet another attempt at making ical_dezonify() send outgoing calendar items
   in UTC format.  (This one will probably work -- the timestamps have the
   "Z" after them which signifies UTC instead of local time.)

 Revision 601.97  2002/12/27 04:40:40  ajc
 * Another attempt at fixing timezones :(

 Revision 601.96  2002/12/25 23:17:07  ajc
 * ical_dezonify.c: shuffle around #includes and #ifdef's

 Revision 601.95  2002/12/25 21:46:19  ajc
 * Another shot at timezone handling (ugggghh)

 Revision 601.94  2002/12/25 07:05:26  ajc
 * set up ical_dezonify() to be called at appropriate times
   (but it doesn't seem to work correctly yet)

 Revision 601.93  2002/12/25 06:41:44  ajc
 * ical_dezonify.c: added (function to strip localized timestamps out of
   a component and replace them with UTC timestamps)

 Revision 601.92  2002/12/19 04:51:49  ajc
 * database_cleanup.sh: added

 Revision 601.91  2002/12/18 04:08:16  ajc
 * Don't attempt to send out invitations when there are no attendees.

 Revision 601.90  2002/12/15 10:53:51  error
 * Final touches on the new message formatter.

 Revision 601.89  2002/12/15 09:42:37  error
 * Converted more routines to new IPC code.

 Revision 601.88  2002/12/14 23:21:29  ajc
 * When saving a calendar event, set the message Subject to the event
   summary string, and the message Date/time to the event start time.

 Revision 601.87  2002/12/10 23:36:20  ajc
 * Fixed a bug in the allocation of per-session dynamic data for calendar module

 Revision 601.86  2002/12/09 06:07:29  ajc
 * Finished the code to accept incoming calendar REPLY messages and
   merge/save the updated event in the user's calendar.

 Revision 601.85  2002/12/08 06:01:48  ajc
 * More work on the reply handler.  Wrote functions to locate the message
   containing the invitation being replied to.  Just need to write the
   merge and resave functions now.

 Revision 601.84  2002/12/08 00:23:08  ajc
 * Began adding a calendaring subcommand to accept a reply to a meeting
   invitation and update the event in the user's calendar.  (Not finished.)

 Revision 601.83  2002/12/04 05:01:18  ajc
 * When sending out invitations, encapsulate the VEVENT component inside a
   VCALENDAR component, along with proper metadata such as product ID and
   vcalendar version.  Move the METHOD:REQUEST property to the encapsulating
   component rather than the event itself.  (Still need to add a VTIMEZONE).

 Revision 601.82  2002/12/03 04:49:15  ajc
 * Send out meeting requests!  (Need to test with various clients.)

 Revision 601.81  2002/12/02 08:09:00  error
 * Improvements to the new message formatter, including URL support and
   doing the Right Thing for text/plain messages.

 Revision 601.80  2002/12/01 11:02:57  error
 * New experimental message formatter - try it, you'll like it!

 Revision 601.79  2002/12/01 04:48:24  ajc
 * The code to check for sending invitations needs to happen *after* save

 Revision 601.78  2002/12/01 00:59:54  ajc
 * Reliably detect when the user saving an event is listed as the meeting
   organizer.  (Currently calls a stub function.  Still need to write the
   function to generate and send invitations.)

 Revision 601.77  2002/11/30 21:35:30  ajc
 * just some trace messages

 Revision 601.76  2002/11/30 05:39:28  ajc
 * Started banging out some code to determine when a saved vEvent needs to
   send out meeting invites.

 Revision 601.75  2002/11/29 16:24:59  ajc
 * When calling ical*_remove_*() routines, the caller then owns the object
   which is removed.  Added ical*_free() calls to free the memory.

 Revision 601.74  2002/11/29 15:44:41  ajc
 * CC->cs_inet_email is now a string buffer, not a pointer
 * Initialize CC->cs_inet_email with a default address, so it always
   contains something even when the directory doesn't
 * Augment CHEK command to return the user's preferred Internet e-mail addr.

 Revision 601.73  2002/11/27 21:05:31  nixo
 added dot ungoto functionality.
 It allows you to enter a room to ungoto, it will look
 through the ungoto list and goto there at the room position
 you last goto-ed it from.
 Don't blame me, this was Gary's idea.

 Revision 601.72  2002/11/25 05:10:07  ajc
 * Post a notification to Aide> when a user is manually deleted
   (resolves Bugzilla enhancement request #73)

 Revision 601.71  2002/11/25 04:57:43  ajc
 * Allow set/clear of per-user Internet mail privilege
 * Honor that flag

 Revision 601.70  2002/11/23 13:35:47  error
 * Makefile.in: Alphabetized all the source files

 Revision 601.69  2002/11/21 05:38:13  ajc
 * Added support for a "New User Greetings" room.  See docs/install.txt

 Revision 601.68  2002/11/15 11:24:40  error
 * serv_rwho.c: cmd_rwho(): move logged_in from param 8 to param 11
   (see RWHO in techdoc/session.txt)

 Revision 601.67  2002/11/13 17:20:57  ajc
 * When in curses mode, call beep() instead of putc(7, stdout) to make it beep.

 Revision 601.66  2002/11/12 04:30:16  ajc
 * Completed the invitation accept/decline code.  It now sends back a
   correctly formated reply!  (Tested with Evolution)

 Revision 601.65  2002/11/11 06:21:55  error
 * Fixed minor bugs with color handling; added explicit support for handling
   the background color e.g. for transparent terminals

 Revision 601.64  2002/11/11 04:17:24  ajc
 * More work on reply to meeting invitation

 Revision 601.63  2002/11/10 09:21:57  error
 * messages.c: fixed <H>eader command

 Revision 601.62  2002/11/10 09:19:38  error
 * Convert some more routines to new IPC code

 Revision 601.61  2002/11/10 09:14:16  error
 * citadel.spec: updated for post-6.01

 Revision 601.60  2002/11/10 09:12:06  error
 * citadel.c: When terminating, don't mark messages new when using new
   message semantics

 Revision 601.59  2002/11/08 05:28:54  ajc
 * Reply to VEVENT invitations: generate reply by cloning the request,
   stripping out non-me attendees, setting the partstat to accept/decline,
   and extracting the address of the organizer.   (Still not finished: right
   now it does everything except actually sending the reply.)

 Revision 601.58  2002/11/07 04:40:07  ajc
 * Repaired the "remember username/password" functionality of the client
   (it was broken during the transition to the new client protocol library)

 Revision 601.57  2002/11/06 05:03:01  ajc
 * No changes; just testing CVS permissions

 Revision 601.56  2002/11/04 21:49:53  ajc
 * Don't run the wait indicator when in an external editor

 Revision 601.55  2002/11/01 04:57:56  ajc
 * Fixed .AM and .AE commands

 Revision 601.54  2002/10/29 20:14:56  error
 * rooms.c: don't allow a blank floor name when creating a floor (in client)

 Revision 601.53  2002/10/26 06:15:09  ajc
 * cmd_cflr() - don't allow zero length floor names

 Revision 601.52  2002/10/26 06:01:19  ajc
 * Increase size of temp file name buffer in struct CitContext

 Revision 601.51  2002/10/25 09:38:38  error
 * Fixed broken .Goto by partial match

 Revision 601.50  2002/10/25 09:26:26  error
 * Fix the status line not updating with the new wait indicator; rearrange
   the code so network_status_cb is a member function of ipc.

 Revision 601.49  2002/10/25 04:39:38  ajc
 * When in fullscreen mode, display an "X" icon in the lower right corner of
   the screen while waiting for the server.

 Revision 601.48  2002/10/24 09:04:53  error
 * acconfig.h: add DISABLE_CURSES template

 Revision 601.47  2002/10/23 04:06:41  ajc
 * "conflicts" check now also tells the client whether a conflicting event
   is really just an older version of the same object.

 Revision 601.46  2002/10/23 03:55:21  ajc
 * Finished: when saving an object of type text/calendar to the Calendar> room,
   set the Citadel "extended message ID" to the UID of the VEVENT object.  This
   causes the message base to automatically delete any existing instance(s) of
   that object: automatic replacement.

 Revision 601.45  2002/10/23 03:07:43  ajc
 * First hack at UID-to-EMSGID mapping (for auto replacement)

 Revision 601.44  2002/10/21 20:00:41  ajc
 * List UID and SUMMARY properties of conflicting events

 Revision 601.43  2002/10/21 17:14:42  ajc
 * More work on conflict detects

 Revision 601.42  2002/10/20 21:42:54  ajc
 * More conflict checking stuff

 Revision 601.41  2002/10/20 20:21:27  ajc
 * Put in a skeleton "hunt for conflicts" code

 Revision 601.40  2002/10/20 08:01:03  error
 * docs/install.txt: add Mac OS X build instructions

 Revision 601.39  2002/10/19 21:35:34  ajc
 * Moved some of the calendar logic over from WebCit

 Revision 601.38  2002/10/19 08:30:03  error
 * dynloader.c:  Darwin places underscores in front of symbol names, just
   like OpenBSD.

 Revision 601.37  2002/10/19 08:18:06  error
 * Disable any use of curses on Darwin (this is temporary until I beat
   Apple's bizarre linker into submission)

 Revision 601.36  2002/10/18 10:33:09  error
 * More signed/unsigned fixes (for the new progress gauge)

 Revision 601.35  2002/10/18 07:56:30  error
 * fix self-service account creation at login prompt (accidentally called
   CREU instead of NEWU).

 Revision 601.34  2002/10/18 03:44:13  ajc
 * Fixed a bug in CtdlRenameRoom() that caused the old room record not to be
   deleted when there was a mailbox namespace attached.
 * Reworked imap_rename() and imap_rename_backend() to not use nested functions

 Revision 601.33  2002/10/17 12:56:44  error
 * Bug fixes:  Fix numerous char array size mismatches, signed/unsigned
   mismatches, and const correctness problems (though not nearly all)

 Revision 601.32  2002/10/17 11:13:27  error
 * Convert room listings and some aide functions to new IPC code

 Revision 601.31  2002/10/16 13:46:19  ajc
 * Remove some trace messages

 Revision 601.30  2002/10/16 08:59:41  error
 * Don't assume pointer to time_t is compatible with pointer to
   struct timeval.tv_sec (it isn't on darwin)

 Revision 601.29  2002/10/16 02:49:55  ajc
 * CtdlWriteObject() encode in memory instead of on disk (not tested)

 Revision 601.28  2002/10/15 17:41:20  ajc
 * Numerous warning fixes and cleanups for compile on Linux for IBM S/390
 * Name temp files with source code location of who created them

 Revision 601.27  2002/10/15 06:47:11  error
 * IPC support for resumable downloads

 Revision 601.26  2002/10/15 06:44:55  error
 * configure.ac check for malloc.h header file (Mac OS X needs this)

 Revision 601.25  2002/10/15 03:56:33  ajc
 * Run without crashing on Berkeley DB 4.1  :)

 Revision 601.24  2002/10/15 03:30:52  ajc
 * Build ok on Berkeley DB 4.1

 Revision 601.23  2002/10/14 08:27:40  error
 * Mac OS X build fixes (everything but serv_imap now builds)

 Revision 601.22  2002/10/08 09:44:52  error
 * Build environment:  cvs ignore .diff/.patch files

 Revision 601.21  2002/10/07 18:59:37  ajc
 * Build fixes for Solaris box without OpenSSL installed

 Revision 601.20  2002/10/07 09:40:49  error
 * Don't try to compile non-existent client_crypto.c (oops!)

 Revision 601.19  2002/10/06 18:46:30  error
 * Move (nearly) all IPC-related code to citadel_ipc.[ch].

 Revision 601.18  2002/10/05 04:48:29  ajc
 * Change MAXSETUP from 4 to 3 (bug reported by mavherzog, fix suggested by IO)

 Revision 601.17  2002/10/04 07:11:56  error
 * Fix bootstrap for Red Hat 7.3

 Revision 601.16  2002/10/02 04:07:27  error
 * status_line() now shows "new mail" only when there is new mail

 Revision 601.15  2002/10/01 04:00:13  ajc
 * html.c: speed improvement in html-to-ascii converter
 * messages.c: MASSIVE speed improvement in message output

 Revision 601.14  2002/09/30 08:07:11  error
 * ipcdef.h: add extern "C" for linking to C++ programs

 Revision 601.13  2002/09/30 07:32:09  error
 * ipcdef.h: remove prototype for connection_died(), it doesn't belong there

 Revision 601.12  2002/09/30 07:20:35  error
 * Fix configure mis-detecting libical and zlib and a missing semicolon in
   tools.h

 Revision 601.11  2002/09/29 04:55:13  error
 * File and attachment downloads now use the new IPC code.
 * Implemented high-speed pipelined file download in IPC code.

 Revision 601.10  2002/09/29 04:41:43  error
 * file_ops.c: cmd_read() now returns a short read at end-of-file instead of
   4096, this prevents trailing garbage on the downloaded file; also it now
   succeeds if the requested number of bytes is > 4096; it simply returns
   only 4096 bytes.

 Revision 601.9  2002/09/27 06:53:20  error
 * Allow multiple simultaneous IPC connections.  All changes necessary for
   the client to use the new code are necessarily included.

 Revision 601.8  2002/09/22 02:48:04  ajc
 * axdefs.h: add "Notes" view

 Revision 601.7  2002/09/17 03:18:15  ajc
 * Create the Calendar> and Tasks> rooms at login time, and set their default
   views to calendar and tasklist views.  Renamed "My Calendar>" to "Calendar>"
   to be in sync with the KDE groupware project.

 Revision 601.6  2002/09/15 03:24:28  ajc
 * Renamed decode_base64() to CtdlDecodeBase64()
 * Renamed decode_quoted_printable() to CtdlDecodeQuotedPrintable()
 * These changes are to avoid namespace conflict with libical

 Revision 601.5  2002/09/13 15:41:39  ajc
 * serv_calendar.c: #include "serv_calendar.h", not "serv_ical.h"

 Revision 601.4  2002/09/11 03:17:18  ajc
 * [EXPERIMENTAL] Link in the "libical" calendar library if present
 * Rename serv_ical to serv_calendar (because the resulting module would be
   named "libical" which conflicts with the external library by that name)

 Revision 601.3  2002/09/10 08:07:38  error
 * update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00* update citadel.spec for 6.00

 Revision 601.2  2002/09/09 03:03:18  ajc
 * When a user logs in, create the Mail> room if it doesn't exist.  (Should
   never happen, but one site had a situation where this patch came in handy.)

 Revision 601.1  2002/09/08 04:15:28  ajc
 * Also fixed a bug in <.T>erminate <S>tay which caused the terminal to
   become wacky on second and subsequent sessions.

 Revision 601.0  2002/09/08 03:51:30  ajc
 * Let's call this 6.01, shall we?

 Revision 600.2  2002/09/08 03:48:48  ajc
 * Fixed a crashy crashy bug in the AGUP protocol function in the client

 Revision 600.1  2002/09/07 03:24:23  ajc
 * When displaying messages using MSG4, enumerate the attachments on the
   client side, because the server won't add their descriptions to the
   message text.

 Revision 600.0  2002/09/04 03:06:58  ajc
 * Committing the current code base as version 6.00

 Revision 591.109  2002/09/03 14:09:04  ajc
 * setup.c: automatically stop/start services

 Revision 591.108  2002/09/03 04:42:15  ajc
 * Added MSG4 support to client-side IPC
 * Moved HTML-to-text conversion to the client side

 Revision 591.107  2002/09/02 20:40:34  ajc
 * Automation work for setup

 Revision 591.106  2002/08/31 04:36:26  ajc
 * IMAP APPEND no longer forces messages to appear as from the logged-in user
   when appending to a mailbox room.  (Allows migration to Citadel from some
   other type of server without losing all the sender names.)

 Revision 591.105  2002/08/31 04:12:39  ajc
 * Set version number to 6.00 in documentation and header files.
 * Convert RFC822 newlines (CRLF) to Unix/Citadel newlines (LF) when
   performing an IMAP APPEND command.

 Revision 591.104  2002/08/28 03:18:06  ajc
 * Make reply_to and reply_subject global (otherwise they don't work!)

 Revision 591.103  2002/08/27 04:01:03  ajc
 * Added the ability to run an external command from the client when new mail
   arrives.  (Ok, I admit it: I just discovered http://www.dailywav.com and I
   just *had* to have Bear in the Big Blue House tell me I've got mail)

 Revision 591.102  2002/08/26 02:55:09  ajc
 * Small bugfix for memory bounds checking in the MIME parser

 Revision 591.101  2002/08/25 21:23:49  ajc
 * Hide the SMTP queue and the sysconfig rooms.

 Revision 591.100  2002/08/24 05:58:53  ajc
 * Bugfixes and cosmetic changes to listsub system

 Revision 591.99  2002/08/23 03:36:52  ajc
 * Finished all of the subscribe/unsubscribe/confirm stuff!

 Revision 591.98  2002/08/23 02:30:02  ajc
 * Delete unconfirmed pending subscribe/unsubscribe requests after three days
 * Don't allow multiple subscriptions of the same address to a list

 Revision 591.97  2002/08/22 03:55:11  ajc
 * Fixed bug in above

 Revision 591.96  2002/08/22 03:40:00  ajc
 * Add "unsubscribe" command (but not yet "confirm unsubscribe")

 Revision 591.95  2002/08/21 21:58:00  ajc
 * Completed self-service list subscription via web.
   (Still need to implement unsubscribe.)

 Revision 591.94  2002/08/16 21:04:56  ajc
 * Add LPRM command

 Revision 591.93  2002/08/14 20:21:52  ajc
 * Extend GETR/SETR to allow twiddling of bits in the QRflags2 bucket
 * New flag and config options to allow/disallow self-service list subscription
   and unsubscription to a room

 Revision 591.92  2002/08/14 02:36:05  ajc
 * WebCit list subscription integration

 Revision 591.91  2002/08/13 18:48:46  ajc
 * Generate and send subscription confirmation requests.
   (Still need to finalize the URL format)

 Revision 591.90  2002/08/13 17:19:11  ajc
 * Added email capability to quickie_message()

 Revision 591.89  2002/08/13 04:02:06  ajc
 * <R>eplace string is now case sensitive

 Revision 591.88  2002/08/13 03:46:31  ajc
 * More work on the self-service subscribe/unsubscribe infrastructure

 Revision 591.87  2002/08/12 03:00:20  ajc
 * Fixed bug in <R>eplacestring which caused it to lock up in certain conditions

 Revision 591.86  2002/08/12 00:09:05  ajc
 * Compress VISIT records using zlib if available.  This reduces the object
   size from over 4k to about 70 bytes.  Experimental.
 * Automatically uncompress database records found to be compressed (duh)

 Revision 591.85  2002/08/08 22:01:35  ajc
 * autoconf to look for zlib (for an upcoming experiment)

 Revision 591.84  2002/08/08 03:25:56  ajc
 * Bugfix for above
 * Started writing some infrastructure for sub/unsub

 Revision 591.83  2002/08/08 02:49:12  ajc
 * serv_network.c: retain unknown commands in netconfigs and write them back
 * techdoc/netconfigs.txt: specify commands for subscribe/unsubscribe pending
 * serv_listsub.c: added (currently a stub)

 Revision 591.82  2002/08/06 03:12:14  ajc
 * Fixed and clarified "new mail has arrived" messages.

 Revision 591.81  2002/08/06 03:02:29  ajc
 * ;AE command -- Make it clearer which floor the user is editing

 Revision 591.80  2002/08/06 02:58:01  ajc
 * Cleaned up the output of digest mode.  It's usable now, but the self-service
   subscribe/unsubscribe functions are not there yet.

 Revision 591.79  2002/08/05 14:38:11  ajc
 * rooms.c: corrected a typo reported by Campagnolo (Mike Poulin)

 Revision 591.78  2002/08/03 19:48:40  ajc
 * NULLify 'roomrec' before attempting partial match in <G>oto
   (fixes heap corruption when this command is executed)

 Revision 591.77  2002/08/03 04:13:46  ajc
 * citadel_ipc.c: Attempted to clean up memory allocation

 Revision 591.76  2002/08/01 05:41:53  ajc
 * Completion of (most of) digest mode.  Still needs some polish.

 Revision 591.75  2002/07/31 04:01:57  ajc
 * Began implementing "digest mode" for listserving.  (Not complete)

 Revision 591.74  2002/07/28 20:33:13  ajc
 * Augmented the "headers only" functionality of the message reading API (as
   well as the options of the server commands which expose it) to allow
   headers+body, headers only, or body only.
 * Adjusted message output of IMAP command FETCH BODY[1] when message is in
   legacy Citadel format.  This fixes a bug exposed by SquirrelMail.

 Revision 591.73  2002/07/23 04:00:06  ajc
 * Completed the MSGP and MSG4 commands to set the client's preferred MIME
   content types, and fetch messages with MIME content types.

 Revision 591.72  2002/07/21 22:29:46  ajc
 * Added in some infrastructure to switch the client's message reading from
   conventional MSG0 to a MIME-download type thing, like WebCit does.

 Revision 591.71  2002/07/21 15:45:07  error
 * Remove duplicate call for server time in who_is_online(); IPC does this now

 Revision 591.70  2002/07/21 15:43:57  error
 * citadel_ipc.c: Fix for segfault on empty messages

 Revision 591.69  2002/07/20 04:05:10  ajc
 * Updated hack.txt with some fresh new information

 Revision 591.68  2002/07/19 02:28:11  ajc
 * citadel_ipc.c: changed various buffer lengths from 256 to SIZ in order to
   accomodate long lines which often spew out (usually from spam unfortunately)

 Revision 591.67  2002/07/18 21:29:37  nbryant
 added a note about GNU make to the configure script output

 Revision 591.66  2002/07/18 20:21:18  ajc
 * Eliminated the 512-message limit in the client for reading messages.

 Revision 591.65  2002/07/13 04:12:40  ajc
 * Don't lie about format types in MSG0 anymore.  We no longer need to.

 Revision 591.64  2002/07/13 03:24:32  ajc
 * CtdlOutputMsg() caches the most recently fetched message in memory.  This
   eliminates the need to do multiple database fetches when we go back for
   additional MIME parts, etc.

 Revision 591.63  2002/07/11 03:40:51  ajc
 * When outputting a multipart MIME message, supply the client with "pref="
   and "suff=" lines in addition to the "part=" lines.

 Revision 591.62  2002/07/02 16:27:06  error
 * prototypes in citadel_ipc.h updated

 Revision 591.61  2002/06/29 23:26:55  error
 * fixed some memory leaks

 Revision 591.60  2002/06/29 18:36:31  error
 * tools.c: Massively faster versions of extract_token() and remove_token().
   These functions can now handle multi-kilobyte sized strings with hundreds
   of tokens (or more?).

 Revision 591.59  2002/06/29 15:55:07  error
 * Convert whobbs, serv_info, misc. functions to new IPC code

 Revision 591.58  2002/06/27 03:10:04  error
 * rooms.c: converted most functions to new IPC code

 Revision 591.57  2002/06/25 15:38:52  error
 * fixed bug in dotgoto() causing free() to segfault

 Revision 591.56  2002/06/25 15:13:27  error
 * gotonext() now uses new IPC code to retrieve room listing

 Revision 591.55  2002/06/25 14:21:35  error
 * remove a debugging trap I accidentally left in the code

 Revision 591.54  2002/06/24 20:17:43  error
 * Fixed code for client segfault at login

 Revision 591.53  2002/06/24 20:11:00  error
 * Buffer overflow fixes and minor cleanup in IPC code

 Revision 591.52  2002/06/24 16:07:42  error
 * Message reading and other functions which use the fmout() screen
   formatter now use the new IPC code.

 Revision 591.51  2002/06/22 20:09:16  error
 * Use a new IPC API (in citadel_ipc.c).  Partially converted citadel.c to
   use the new API.

 Revision 591.50  2002/06/21 12:32:48  ajc
 * Minor documentation update

 Revision 591.49  2002/06/19 21:52:13  ajc
 * Support a transient room create as well

 Revision 591.48  2002/06/19 21:42:57  ajc
 * Added support for "transient goto" which allows entry into a private and/or
   zapped room without putting the room [back] on your known rooms list.  This
   is useful for the new administrative functions in WebCit (updating a user's
   vCard without the Aide suddenly having that user's config room on their
   list).

 Revision 591.47  2002/06/18 16:34:06  error
 * room_ops.c: Fix for old room record not being deleted when renaming
   baseroom or aideroom

 Revision 591.46  2002/06/16 21:01:11  ajc
 * Allow Aides to create rooms in other users' namespaces (if global access
   controls allow)

 Revision 591.45  2002/06/15 20:48:50  ajc
 * Fixed small bug in the e<X>pert mode toggle

 Revision 591.44  2002/06/15 20:14:55  ajc
 * Fixed a memory allocation bug in the vCard parser

 Revision 591.43  2002/06/15 17:53:59  error
 * citserver.c: MESG command can now send a different system message based
   on the developer and client ID of the connected client

 Revision 591.42  2002/06/15 04:52:26  ajc
 * SpamAssassin connector is now configurable in <.A>ide <S>ysconfig <I>nternet.
 * Allow more than one SA server (it'll try 'em all)
 * Don't run SA for logged in users

 Revision 591.41  2002/06/14 20:42:56  ajc
 * Discovered that spamd works even without the Content-length: command, so I
   was able to redo the spam checker to work without a temp file.

 Revision 591.40  2002/06/14 20:37:03  ajc
 * Disabled the spam strings checker I wrote a few days ago.
 * When receiving SMTP, check to see if spamd (the SpamAssassin daemon) is
   running on the local machine.  If yes, run the message through it and
   reject if spam.

 Revision 591.39  2002/06/12 03:42:21  ajc
 * "Suppress message prompts" has been changed to "Prompt after each message"
   and of course the effect has been reversed.
 * "Be unlisted in userlog" has been moved to the end of the list of
   configuration prompts, so it doesn't interrupt the flow of thought regarding
   the prompting questions.

 Revision 591.38  2002/06/10 22:25:25  ajc
 * Configuration for spam filter

 Revision 591.37  2002/06/09 23:59:38  ajc
 * Started working on the spam filter

 Revision 591.36  2002/06/07 22:10:51  ajc
 * Added a new message function hook type EVT_SMTPSCAN which permits modules to
   register hooks that can scan incoming SMTP messages and elect to reject them
   (due to virus or spam content, for example).

 Revision 591.35  2002/06/07 03:22:13  ajc
 * Added a module "serv_mrtg" which allows activity reporting to MRTG
   (http://www.mrtg.org) -- this will replace our stats program.

 Revision 591.34  2002/06/02 16:42:17  error
 * Support for Ctrl-W to erase a word when editing or at a prompt

 Revision 591.33  2002/05/31 18:34:04  nbryant
 better curses compatibility, and a couple makefile/configure tweaks

 Revision 591.32  2002/05/28 13:59:02  ajc
 * Removed the 'netsetup' and 'dnetsetup' utilities (obsolete)

 Revision 591.31  2002/05/24 19:58:13  ajc
 * Fixed the "idle timeout during paginator prompt" bug by reintroducting the
   concept of a "half keepalive" and sending them during paginator prompts.

 Revision 591.30  2002/05/23 03:33:21  ajc
 * Added a GTSN (GeT list of SeeN messages) command

 Revision 591.29  2002/05/20 14:29:59  ajc
 * commands.c: fixed bug in the scan for idle_threshold= which didn't include
   the = sign and ended up always setting this value to 0

 Revision 591.28  2002/05/17 03:57:30  ajc
 * When doing fixed_output() of converted HTML, output the whole block of
   data at once instead of one character at a time

 Revision 591.27  2002/05/16 04:44:58  ajc
 * Reduce the number of socket writes when doing fixed_output() to avoid
   getting killed by overhead.  (Thanks to IO for the insight)

 Revision 591.26  2002/05/14 15:25:34  error
 * room_ops.c: clarified aide message when room aide is removed

 Revision 591.25  2002/05/14 15:18:43  error
 * rooms.c: Print name of room aide if any when doing <i>nfo

 Revision 591.24  2002/05/14 01:33:18  error
 * Fixed some incredibly silly typos

 Revision 591.23  2002/05/14 01:27:18  error
 * Minor cosmetic changes, extraneous double spaces etc.

 Revision 591.22  2002/05/14 01:15:54  error
 * Idle threshold on the who list is now customizable in the citadel.rc

 Revision 591.21  2002/05/14 01:09:57  error
 * citadel.c: spacebar won't read New if there are no new messages

 Revision 591.20  2002/05/12 23:00:11  ajc
 * Removed an unused variable

 Revision 591.19  2002/05/12 22:57:04  ajc
 * Removed the unfinished moderation system.
 * CtdlForEachMessage() - only fetch metadata when hunting for messages with
   a specified Content-type.  Serious performance boost.

 Revision 591.18  2002/05/05 17:33:09  error
 * screen.c: fix lack of beeps in curses mode

 Revision 591.17  2002/05/04 02:58:16  ajc
 * Documented a small protocol change for the STEL command

 Revision 591.16  2002/04/30 03:13:59  ajc
 * When sending a page that results in the receiver's Sent/Received Pages>
   room to be created, don't automatically grant the sender access to that room.
 * Added a parameter to create_room() to implement the above

 Revision 591.15  2002/04/23 13:38:08  ajc
 * Minor docs update

 Revision 591.14  2002/04/21 21:28:06  ajc
 * Create the My Citadel Config> room along with the user account

 Revision 591.13  2002/04/20 03:00:01  ajc
 * One more fix for the RENAME command wrt namespaces.

 Revision 591.12  2002/04/18 18:44:36  nbryant
 don't use libtool -avoid-version for libcitserver - this might help for
 OpenBSD

 Revision 591.11  2002/04/17 04:35:06  ajc
 * Finished the IMAP RENAME command.  (I ended up using nested functions
   because it made the task much, much easier.  We can fix it later or convert
   it to C++ if we find this becomes a problem.)

 Revision 591.10  2002/04/15 13:25:47  ajc
 * Add error responses to imap_rename() and set up subfolder framework

 Revision 591.9  2002/04/14 22:42:49  ajc
 * Began implementing RENAME command in IMAP.

 Revision 591.8  2002/04/14 22:27:05  ajc
 * Added access control checking to CtdlRenameRoom()

 Revision 591.7  2002/04/14 22:11:22  ajc
 * New back end function CtdlRenameRoom() which is used to rename a room and/or
   move it to a different floor.
 * cmd_setr() now uses CtdlRenameRoom() to do part of its work

 Revision 591.6  2002/04/10 03:58:40  ajc
 * Began work on IMAP RENAME

 Revision 591.5  2002/04/09 14:26:43  ajc
 * Allow INBOX to have subfolders.  There's no longer any reason not to.

 Revision 591.4  2002/04/05 22:31:59  error
 * Send time-of-day with pages sent via IMAP

 Revision 591.3  2002/04/05 14:34:02  ajc
 * Minor tweak to nested folder algorithm to handle nesting in mail root

 Revision 591.2  2002/04/05 04:25:56  ajc
 * Support nested folders in IMAP.  We might want to change the delimiter.

 Revision 591.1  2002/04/04 23:25:30  ajc
 * Experimental hacking to handle subfolderization in IMAP.  Seems to work ok
   but it makes Mozilla mail hang.  Will investigate more later...

 Revision 591.0  2002/04/01 05:13:20  ajc
 * Tagged everything for 5.91 release

 Revision 590.168  2002/04/01 05:12:57  ajc
 * Prep for 5.91 release

 Revision 590.167  2002/03/29 04:43:15  ajc
 * Removed the BMBX command.  Mailbox security update now runs automatically when the
   server starts and it sees data files version <5.91.

 Revision 590.166  2002/03/26 05:58:35  nbryant
 don't check for a database driver until after the openssl check is
 complete. if the ld paths aren't set up for the db installation this
 can interfere with the openssl check because it tries to run a program
 linked against the libraries we've been finding.

 Revision 590.165  2002/03/26 05:20:46  nbryant
 removed last vestiges of "#ifdef CIT_CLIENT" conditional compilation,
 which has been rendered unusable by the build system simplifications

 Revision 590.164  2002/03/26 05:13:32  nbryant
 fix monstrous shell script syntax in configure.ac

 Revision 590.163  2002/03/26 04:38:31  nbryant
  * support autoconf 2.53
  * make a note that people should be using libtool 1.4d
  * cut some of the more pointless bells and whistles out of the configure
    script in an effort to make it more maintainable
  * modularized ipc_c_tcp.c and client_crypto.o so that they're not tied
    to the curses stuff.
  * timezone/daylight and other FreeBSD fixes
  * more effort in the configure script to detect the common
    bastardizations (ahem, ports) of Berkeley DB without falling down

 Revision 590.162  2002/03/25 03:09:39  nbryant
 fix warning on platforms where pthread_t may be a pointer

 Revision 590.161  2002/03/25 00:01:50  nbryant
  * compatibility with Berkeley DB < 3.3
  * squished symbol clashes with the OK symbol from curses in certain *n[iu]x
 distributions. this is kind of a pain in the ass, but I had to rename our
 OK to CIT_OK :-(

 Revision 590.160  2002/03/22 04:35:38  ajc
 * Handle vCard registration updates for users other than the one currently
   logged in.  (Allows administrative editing of contact information.)

 Revision 590.159  2002/03/20 19:03:27  ajc
 * Don't re-declare timezone variables ('cuz FreeBSD chokes on that)

 Revision 590.158  2002/03/19 04:34:42  ajc
 * mime_parser.c: minor changes for easier porting between Citadel and WebCit

 Revision 590.157  2002/03/19 04:19:33  ajc
 * Saw what IO did with strchr() and did the same in a few more places

 Revision 590.156  2002/03/17 00:08:02  error
 * mime_parser.c: more robust parsing of Content-Type header

 Revision 590.155  2002/03/16 05:22:59  ajc
 * Post an error message to Aide> when unlink() is unable to delete old
   database log files.

 Revision 590.154  2002/03/14 04:35:26  nbryant
 avoid symbol clash with curses' "timeout" function (which may be a macro
 in some versions)

 Revision 590.153  2002/03/14 04:24:20  nbryant
 support window resizing in curses mode

 Revision 590.152  2002/03/13 04:11:11  nbryant
 fix up minor gotcha introduced by fgets change

 Revision 590.151  2002/03/13 03:58:29  ajc
 * Site-configurable option "Allow system Aides to gain access to mailboxes"

 Revision 590.150  2002/03/13 03:34:38  nbryant
 /* when running in curses mode, the scroll bar in most
    xterm-style programs becomes useless, so it makes sense to
    pause after a screenful of pages if the user has been idle
    for a while. However, this is annoying to some of the users
    who aren't in curses mode and tend to leave their clients
    idle. keepalives become disabled, resulting in getting booted
    when coming back to the idle session. but they probably have
    a working scrollback in their terminal, so disable it in this
    case:
  */
 if (!is_curses_enabled())
        lines_printed = 0;

 Revision 590.149  2002/03/12 23:34:37  nbryant
 use ncurses in preference to curses if it's installed; it handles
 background colors properly on dtterm, has a larger color pair palette,
 and has a bigger terminal database than solaris curses

 Revision 590.148  2002/03/12 22:47:17  nbryant
 curses fix: map our normal color pairs into the 0-7 range instead of 1-8,
 in order to make our pairs fit on terminals such as dtterm where COLOR_PAIRS=8.
 map the white/blue color pair onto 8 instead of 9, but only if that slot
 is available; fall back on white/black otherwise.

 it seems there may be an off-by-one error in the color pair manpages for
 the various curses packages (?) if not, our 0 entry is unusable, but that's
 the DIM_BLACK color and we don't use it anyway.

 Revision 590.147  2002/03/12 22:17:20  ajc
 * Give mailbox owners access to "who knows room" command

 Revision 590.146  2002/03/12 21:08:03  nbryant
 support color under Solaris curses

 Revision 590.145  2002/03/12 19:59:40  ajc
 * Access control change: do not treat mailboxes as guessname rooms for Aides.
   Open up INVT/KICK commands to non-Aides for their mailboxes.

 Revision 590.144  2002/03/12 04:30:52  nbryant
 if a filesystem node exists at /var/run/egd-pool, try to connect to it as
 the EGD (Entropy Gathering Daemon) or PRNGD (pseudorandom number
 generator daemon) socket and seed OpenSSL's RNG.

 this is necessary on solaris and other systems which lack /dev/urandom.

 Revision 590.143  2002/03/12 03:43:26  nbryant
 squished the last remaining calls to sprintf

 Revision 590.142  2002/03/12 03:36:55  nbryant
 replace calls to gets with fgets

 Revision 590.141  2002/03/12 03:19:09  nbryant
 more sprintf bashing. now the only ones left are in mime_parser

 Revision 590.140  2002/03/12 01:33:42  nbryant
  - pass -Wcast-qual to gcc
  - more sprintf bashing

 Revision 590.139  2002/03/12 00:03:43  nbryant
 more sprintf removals

 Revision 590.138  2002/03/11 06:00:21  nbryant
 use <db.h> before <db3/db.h> or <db4/db.h>
 this is the only way i can think of to make it work everywhere; people on
 systems like FreeBSD where the ports work the other way around will have to
 specify an extra -I flag in their CPPFLAGS variable.

 Revision 590.137  2002/03/11 05:42:46  nbryant
 removed all references to sprintf from several files (not all files yet)
 and replace with snprintf

 Revision 590.136  2002/03/11 04:16:20  nbryant
 warning fixes on sparc-sun-solaris2.8 with gcc 3.0.4, mostly for *printf
 format strings

 Revision 590.135  2002/03/11 03:55:24  nbryant
  - fixes for building without OpenSSL
  - setenv doesn't exist on all systems, use putenv instead
  - support Solaris' curses implementation

 Revision 590.134  2002/03/09 22:52:04  ajc
 * Applied a patch submitted by <xperc@hotmail.com> to fix a potential buffer
   overflow problem in lprintf().  I also did the same fix to cprintf().

 Revision 590.133  2002/03/09 16:47:57  ajc
 * Added BMBX to fix a problem

 Revision 590.132  2002/03/09 06:18:37  ajc
 * one more tweak

 Revision 590.131  2002/03/09 05:22:29  ajc
 * this should do it.

 Revision 590.130  2002/03/09 05:02:20  ajc
 * Attempts to fix the access control crap

 Revision 590.129  2002/03/08 05:42:02  ajc
 * Patch to allow invitations and admin access to mailbox rooms.  NEEDS TESTING!

 Revision 590.128  2002/03/07 04:30:37  ajc
 * Force recipient only in Mail>, not in all mailbox rooms

 Revision 590.127  2002/03/05 22:45:40  error
 * Autoconf fixes for DB4 support

 Revision 590.126  2002/03/05 04:47:49  ajc
 * vcard.c: another API update

 Revision 590.125  2002/03/04 05:29:39  ajc
 * Made a small API change to vcard.c for WebCit, brought the change over here too
   in order to keep vcard.c identical everywhere.

 Revision 590.124  2002/03/03 06:48:25  ajc
 * Client and server options to disable self-service user account creation

 Revision 590.123  2002/03/03 06:31:58  ajc
 * Added password starred-out entry to newprompt() and strprompt()
 * Applied the above setting to password set/change in <.A>ide <U>seredit

 Revision 590.122  2002/03/03 06:18:45  ajc
 * Implemented the CREU server command to administratively create user accounts
 * Added the ability to create new user accounts to <.A>ide <U>seredit

 Revision 590.121  2002/03/03 06:05:16  ajc
 * Split up some of the code in order to prepare for user accounts to be
   administratively created without logging in to them.

 Revision 590.120  2002/03/02 05:56:48  ajc
 * Properly implemented the network filter list.  Finished the server module and
   did a client-side <.A>ide <S>ysconfig <F>ilterlist command.

 Revision 590.119  2002/03/01 04:24:20  ajc
 * Cosmetic change to Received: line

 Revision 590.118  2002/03/01 04:16:22  ajc
 * CtdlReadMessageBody() -- fixed bug that caused the prepend buffer to be
   discarded instead of prepended.  "Received:" lines now work.

 Revision 590.117  2002/02/23 19:20:51  ajc
 * Do the use table purge in two phases to avoid crashy crashy

 Revision 590.116  2002/02/20 22:42:19  ajc
 * Started adding better management of source IP addressses in SMTP service

 Revision 590.115  2002/02/15 04:28:57  ajc
 * Wrote the expire/purge routine for the new use table

 Revision 590.114  2002/02/15 04:05:08  ajc
 * Began implementation of a networker use table that doesn't chew up oodles
   of CPU time.  (It uses a cdb instead.)

 Revision 590.113  2002/02/15 03:40:06  ajc
 * Stu's changes (which he checked in without making any ChangeLog comments,
   bad Stu!) didn't build properly without curses.  Added #ifdef's.

 Revision 590.112  2002/02/13 22:15:10  ajc
 * That was stupid.

 Revision 590.111  2002/02/13 22:04:11  ajc
 * added vcard_to_html() function

 Revision 590.110  2002/02/13 15:48:55  ajc
 * Allow the READ command to return packets bigger than 1 byte.  (ooops!)

 Revision 590.109  2002/02/12 20:15:25  ajc
 * Threw in a few more #ifdef's so the client build doesn't barf on
   non-curses systems
 * Added rc_prompt_control (<N>ext/<S>top active at paginator: on/off/user)

 Revision 590.108  2002/02/11 15:52:10  ajc
 * Don't crash when deleting "purge this vCard" messages

 Revision 590.107  2002/02/10 22:36:41  nbryant
  - replace cdb_trunc with a complete version of the code i had been
    working on; fallback code for db < 3.3.x needed
  - change 'can't connect to host.port' to 'can't connect to host:port'

 Revision 590.106  2002/02/08 22:39:08  ajc
 * If there's already a Subject line in memory, display it below the usual
   headers when the user hits <E>

 Revision 590.105  2002/02/08 22:36:23  ajc
 * Changed the logic for printing RFC822 addresses (again)
 * Implemented cdb_trunc() in database_sleepycat.c, using db_truncate()
   (We need to either provide an alternative implementation or require DB >=3.3)
 * Automatically Re: subject line in the client where appropriate

 Revision 590.104  2002/02/08 19:02:25  ajc
 * Added client and server side support for entering Subject lines in
   messages when not using RFC822.

 Revision 590.103  2002/02/08 18:10:07  ajc
 * When outputting a message in non-RFC822 format, don't display an Internet
   address if the user is local.

 Revision 590.102  2002/02/07 04:42:49  ajc
 * Silently refuse to add directory entries for Internet addresses already
   belonging to other users.
 * cdb_trunc() for CtdlDirectoryInit: implemented for GDBM, stubbed for DB

 Revision 590.101  2002/02/05 05:05:53  ajc
 * Don't crash when posting if the user doesn't have an Internet directory address

 Revision 590.100  2002/02/03 15:29:03  error
 * fixed a silly oversight in serv_crypto.c when removing the ETLS command

 Revision 590.99  2002/02/03 15:21:48  error
 * Remove the ETLS command, it is no longer needed

 Revision 590.98  2002/02/02 21:44:04  ajc
 * If a user has at least one valid Internet directory address, stamp it onto
   any outgoing messages.

 Revision 590.97  2002/02/01 05:11:26  ajc
 * Added a QDIR protocol command to do quick-and-dirty queries of the directory
 * In the client, check the directory for conflicts when selecting email addr.

 Revision 590.96  2002/01/31 05:13:44  ajc
 * When deleting a vCard from the Global Address Book room, remove the
   corresponding address in the directory.  (Not tested.)

 Revision 590.95  2002/01/30 19:03:41  ajc
 * Added a new DeleteFunctionHook type of thing.  These get called when a
   message is being deleted from a room.
 * When deleting messages from a room, do the AdjRefCount() calls (and now,
   the PerformDeleteHooks() calls) in a second pass.  This keeps that stuff
   outside of the S_QUICKROOM critical section.

 Revision 590.94  2002/01/27 06:39:45  error
 * file_ops.c: fixed bug in cmd_read() which could cause server to report
   the wrong number of bytes for the client to download

 Revision 590.93  2002/01/26 21:33:38  ajc
 * More internet addressing and global directory stuff.  I think it's all working now
   except for the purging of old entries.

 Revision 590.92  2002/01/26 11:02:37  error
 * citadel.spec cleaned up

 Revision 590.91  2002/01/26 09:23:40  error
 * setup now has a silent running mode (-q option) where it silently sets
   defaults for everything.  This is intended for use in scripts such as
   the RPM packages, eliminating a step from the installation process.

 Revision 590.90  2002/01/26 09:19:16  error
 * citadel.spec has been completely overhauled, we can make RPMs now!

 Revision 590.89  2002/01/26 04:59:57  ajc
 * smtp FROM command now validates sender using the validate_recipients() loop
   (yeah, it's good for that too) making it directory-aware

 Revision 590.88  2002/01/26 04:01:10  error
 * Formatter now uses more of the available screen width

 Revision 590.87  2002/01/26 03:57:30  error
 * Revised status_line() display, it's much more compact now

 Revision 590.86  2002/01/26 03:50:26  error
 * Better error reporting in connection_died(), kills curses before printing
   error message, prints last errno.

 Revision 590.85  2002/01/25 05:19:03  ajc
 * Greatly simplified the logic for validating recipient addresses for incoming
   SMTP.  This logic destroys the whitespace mangling for local names; I will
   fix this tomorrow.

 Revision 590.84  2002/01/25 04:36:35  ajc
 * fixz to allow incoming vCards in the address book to actually get processed

 Revision 590.83  2002/01/24 06:52:54  error
 * citadel_decls.h: fix unresolved extern errors

 Revision 590.82  2002/01/23 05:04:05  ajc
 * Add vCards from incoming network messages in the GAB to the directory.

 Revision 590.81  2002/01/23 03:39:32  ajc
 * Added a new hook type for handling incoming network messages
 * Wrote a skeleton module for net filtering

 Revision 590.80  2002/01/22 10:46:25  error
 * read_message() and fmout() now accept a FILE to which to send their
   output; this fixes quoting in the fullscreen client

 Revision 590.79  2002/01/20 08:03:43  error
 * curses client: use the status line as "input" line in chat mode

 Revision 590.78  2002/01/20 07:43:07  error
 * serv_chat.c: Server no longer crashes when CHATLOG can't be opened

 Revision 590.77  2002/01/20 05:22:07  error
 * curses client:  allow goodbye message to be seen on some terminals

 Revision 590.76  2002/01/19 16:56:31  error
 * Fixed color support, now works when rc_ansi_color is on or auto

 Revision 590.75  2002/01/19 15:10:25  error
 * Cosmetics for the client status line

 Revision 590.74  2002/01/19 11:59:33  error
 * A real status line for the text client

 Revision 590.73  2002/01/19 10:08:43  error
 * fix link for libcitserver.so to tools.o which I broke (oops!)

 Revision 590.72  2002/01/19 09:59:08  error
 * Full-screen curses support for Citadel text client

 Revision 590.71  2002/01/17 20:11:05  nbryant
 remove lock.c/lock.h; don't need them for what i was planning after all

 Revision 590.70  2002/01/17 10:48:36  error
 * cosmetic fixes in the new trace file functionality

 Revision 590.69  2002/01/17 10:32:14  error
 * lprintf() now logs the session ID for each log entry within a session.
   Also SMTP, IMAP, POP3 and Citadel protocol commands are differentiated.

 Revision 590.68  2002/01/17 10:16:09  error
 * migratenet.c: cygwin fix: include limits.h

 Revision 590.67  2002/01/17 10:13:31  error
 * serv_ical.c: set expire policy for My Calendar> to manual

 Revision 590.66  2002/01/17 07:18:11  ajc
 * Changed all "free software" references to "open source" in order
   to piss off Richard Stallman

 Revision 590.65  2002/01/17 00:22:35  nbryant
 added lock.[ch]: recursive read/write locking support. (not actually used yet)

 Revision 590.64  2002/01/15 12:41:53  error
 * Implement alternate_semantics (see comments in citadel.rc file)

 Revision 590.63  2002/01/15 11:07:51  ajc
 * vcard.c: updated vCard "object methods" to handle multiple instances of
   the same key name when necessary.

 Revision 590.62  2002/01/15 06:38:39  error
 * Update citadelapi.txt with CtdlUnregister* calls, LogHook calls

 Revision 590.61  2002/01/15 06:20:18  error
 * Modules can now unregister any of their hooks (though none yet take
   advantage of this).

 Revision 590.60  2002/01/14 08:49:13  error
 * Fixed bug in cmd_cre8() causing protocol to get out of sync when creating
   a new room

 Revision 590.59  2002/01/13 04:46:31  ajc
 * Allow incoming SMTP to relay to other Citadel nodes for whom we are
   providing directory service.

 Revision 590.58  2002/01/13 04:06:33  ajc
 * Repaired the problems I created when moving the_mime_parser()'s variables
   from the stack to the heap.  (Hint: sizeof(char *) is 4, not 4096)

 Revision 590.57  2002/01/11 15:46:57  error
 * Allow users to move/copy messages between personal rooms

 Revision 590.56  2002/01/11 04:59:00  ajc
 * Finished most of the work for the Global Address Book.

 Revision 590.55  2002/01/11 04:37:03  ajc
 * More code for the Global Address Book

 Revision 590.54  2002/01/11 02:57:35  error
 * Don't print **** when sending a page or mail from an anonymous-only room

 Revision 590.53  2002/01/10 21:22:37  ajc
 * Minor changes for global directory service

 Revision 590.52  2002/01/10 04:29:28  ajc
 * Minor updates for directory service

 Revision 590.51  2002/01/09 23:12:40  ajc
 * Allow users to zap mailbox rooms

 Revision 590.50  2002/01/09 04:37:32  ajc
 * Finished the callback stuff for vCard address extraction

 Revision 590.49  2002/01/09 04:05:53  ajc
 * Began writing code to harvest Internet e-mail addresses from vCards, and
   hacked together a temporary version (and writeup) of the IGAB command.

 Revision 590.48  2002/01/08 16:34:22  ajc
 * serv_vcard.c: cosmetic cleanup

 Revision 590.47  2002/01/06 22:44:21  error
 * Enable/disable encryption in client from command line and/or citadel.rc

 Revision 590.46  2002/01/06 21:25:26  ajc
 * sysdep.c: in client_write(), handle redirect_fp and redirect_sock *before*
   handling redirect_ssl, because these need to be done the same way regardless
   of client session crypto
 * serv_crypto.c: pasted a bunch of Nathan's #ifdef blocks from sysdep.c in
   order to gain greater portability (or even to get it to compile on splorph)

 Revision 590.45  2002/01/06 11:13:33  error
 * Enable SSL/TLS support in the client (again)

 Revision 590.44  2002/01/06 11:11:31  error
 * Enable SSL/TLS in the client

 Revision 590.43  2002/01/06 10:49:55  error
 * Add some #includes I apparently somehow missed

 Revision 590.42  2002/01/06 10:33:10  error
 * SSL/TLS support for the Citadel wire protocol

 Revision 590.41  2002/01/06 08:54:58  error
 * user_ops.c: fixed become_session() when calling EVT_LOGOUT session hooks

 Revision 590.40  2002/01/05 22:31:22  ajc
 * Removed some protocol commands and writeups that are no longer necessary
 * Began some of the framework for the Global Address Book

 Revision 590.39  2002/01/05 12:44:43  error
 * serv_chat.c: allow a session to be killed while in chat

 Revision 590.38  2002/01/05 12:31:04  error
 * user_ops.c: become_session() when calling EVT_LOGOUT session hooks

 Revision 590.37  2002/01/05 04:51:36  error
 * stats now sorts its top 20 lists properly

 Revision 590.36  2002/01/04 20:57:36  nbryant
 cygwin fix

 Revision 590.35  2002/01/04 20:46:26  nbryant
 Makefile fix for cygwin (fix migratenet linkage)

 Revision 590.34  2002/01/04 20:43:26  nbryant
 configure/genstamp: check for struct tm.tm_gmtoff

 Revision 590.33  2002/01/03 22:01:17  ajc
 * Fixed mail to "sysop"

 Revision 590.32  2002/01/03 21:35:07  ajc
 * I think this is the last of the fixes for the new submit queue.

 Revision 590.31  2002/01/03 12:27:35  error
 * Fixed my name in docs/copyright.txt, why didn't I notice that before?

 Revision 590.30  2002/01/03 12:21:02  error
 * Autoconf support for recognizing OpenSSL

 Revision 590.29  2002/01/03 04:52:28  ajc
 * serv_network.c: migrated deliveries and bounces to the new message
   submission subsystem.  NOT TESTED.

 Revision 590.28  2002/01/01 21:32:10  ajc
 * Finished the updates to serv_smtp.c, although I think there may be a
   problem with one-too-many reference counts when a message is submitted
   via SMTP.

 Revision 590.27  2001/12/31 20:15:13  ajc
 * Almost finished converting serv_smtp.c to the new message submission
   framework.  Still not done yet; don't use this.

 Revision 590.26  2001/12/30 06:20:46  error
 * More keys while reading messages:  Q or S same as Ctrl-C, N same as Ctrl-O.

 Revision 590.25  2001/12/30 05:50:46  error
 * Security:  Citadel now drops privileges when called from telnetd, also
   checks to make sure you didn't set the setuid/setgid bits.  No more
   loginwrapper.sh!

 Revision 590.24  2001/12/29 05:19:32  ajc
 * Minor cosmetic hack

 Revision 590.23  2001/12/29 04:21:22  nixo
 stupid me. I didn't realize that asking for the header did what I wanted
 so I changed my little 'y' hack to not bother reading the text of the
 message. a little saving on the bandwidth.

 Revision 590.22  2001/12/28 22:32:38  nixo
 Added a "read m<y> next" function in read mode. It will skip to the next
 message by the user in the current message list (whatever mode you're in
 be it read forward, last 50, whatever.)

 Revision 590.21  2001/12/28 11:06:53  error
 * More server support for hostnames up to 63 characters (oops I missed a spot)

 Revision 590.20  2001/12/28 09:39:10  error
 * Client support for hostnames up to 63 characters: truncated at 24 in
   short who list, full display in long who list.

 Revision 590.19  2001/12/28 09:28:04  error
 * Server support for hostnames up to 63 characters

 Revision 590.18  2001/12/26 05:01:30  ajc
 * Added a new developer ID for Anticlimactic Teleservices

 Revision 590.17  2001/12/23 10:00:43  error
 * Pages are once again formatted to the caller's screen width.

 Revision 590.16  2001/12/23 09:57:47  error
 * tools.c: added parameter to fmt_date() to allow for printing the seconds
   along with the time, e.g. 12:34 pm or 12:34:56 pm

 Revision 590.15  2001/12/20 04:54:26  ajc
 * If you paid for this software, someone is ripping you off.

 Revision 590.14  2001/12/18 08:24:56  nbryant
 more lovely configure tweaks (include paths for db)

 Revision 590.13  2001/12/18 06:04:08  ajc
 * Moved the buffers in the_mime_parser() from the stack to the heap, because
   it was crashing boxen with small stack sizes.

 Revision 590.12  2001/12/18 05:54:16  ajc
 * Added more load_floorlist() commands to the beginning and end of functions
   in the client that manipulate the floor list.  This fixes a bug in which
   new floors don't show up right away after being created.

 Revision 590.11  2001/12/17 08:14:26  nbryant
 restored the checks for /usr/include/db3 and /usr/local/include/db3
 in configure.  there are too many variations on db installation; this is
 getting messy :-(

 Revision 590.10  2001/12/17 08:00:45  nbryant
 set the pthreads stack size to 128K because FreeBSD's default of 64K
 seems too small. fixes crashes under FreeBSD.

 Revision 590.9  2001/12/16 00:50:14  error
 * Added usersupp.lastcall to the parameters returned from the PASS/PAS2
   commands in logged_in_response().

 Revision 590.8  2001/12/14 21:33:18  nbryant
 finally changed configure to complain if there's no database driver ;)

 Revision 590.7  2001/12/14 08:29:30  error
 * Security: trace file is now only readable by owner, since it contains
   plain text passwords.

 Revision 590.6  2001/12/14 07:04:24  ajc
 * Now you can send mail to yourself.  Hi from Stu.

 Revision 590.5  2001/12/14 06:58:12  ajc
 * Hi from Stu

 Revision 590.4  2001/12/13 22:36:30  nbryant
 make configure search for /usr/local/BerkeleyDB.4.0

 Revision 590.3  2001/12/13 22:29:57  nbryant
 make it compile with Berkeley DB 4.0.x

 Revision 590.2  2001/12/11 21:31:07  nbryant
  - test for -ldb3 before -ldb

 Revision 590.1  2001/12/11 20:04:41  nbryant
  - fix library flags, includes for portability
  - malloc.h is deprecated
  - fix genstamp, hopefully
  - fix size_t *printf handling for portability

 Revision 590.0  2001/12/08 03:31:41  ajc
 * THIS IS 5.90

 Revision 580.95  2001/12/08 03:30:37  ajc
 * Final changes to networking docs for 5.90

 Revision 580.94  2001/12/06 05:13:34  ajc
 * Added the documentation for room sharing and listserv

 Revision 580.93  2001/12/04 05:24:15  ajc
 * Added two more bytes to the possible length of shared secrets in networking
   due to some legacy support requirements.

 Revision 580.92  2001/12/04 05:16:19  ajc
 * mime_parser.c: change to memory allocation algorithm ... some badly done
   messages were crashing the server

 Revision 580.91  2001/12/03 22:48:16  ajc
 * ooops.  Look for the QR2_SYSTEM flag in QRflags2, not QRflags.

 Revision 580.90  2001/12/03 17:02:50  ajc
 * dynloader.c: fixed improperly done declaration and mallok()

 Revision 580.89  2001/12/03 04:28:02  ajc
 * mime_parser.c: now uses built-in functions to decode base64 and
   quoted-printable attachments, instead of piping data to outboard programs.

 Revision 580.88  2001/12/03 02:45:46  ajc
 * Began implementing some code to handle multiple recipients (but #define'd
   it all out because we're approaching a release)

 Revision 580.87  2001/12/03 01:50:17  ajc
 * When sending mail, copy to the sender's "Sent Items>" room instead of to
   the sender's "Mail>" room.

 Revision 580.86  2001/12/02 23:36:24  ajc
 * On a new system, set the default new user level to 4 instead of 1.

 Revision 580.85  2001/12/02 23:27:01  ajc
 * Removed references to the old networker from the documentation.  Did not
   write any new documentation, so what's there now is kind of sparse.

 Revision 580.84  2001/12/02 02:42:55  ajc
 * Implemented new room flag QR2_SYSTEM which supresses the room from all
   room listings, even for Aides (but it's still gotoable).  This will be used
   for rooms which hold system configuration and message queues.

 Revision 580.83  2001/12/01 19:23:26  ajc
 * clientsocket.c: implement socket timeouts for read operations

 Revision 580.82  2001/12/01 17:00:23  ajc
 * serv_smtp.c: when multiple MX's are the same preference, randomize them

 Revision 580.81  2001/12/01 07:18:28  ajc
 * Fixed an SMTP delivery problem that was causing certain classes of
   transient errors to cause a message to never be delivered.

 Revision 580.80  2001/12/01 05:26:01  ajc
 * Added a command "SMTP" to the Citadel protocol, to do some unimportant
   utility/diagnostic functions.

 Revision 580.79  2001/11/27 17:08:29  ajc
 * When calling an external editor, set the environment variable
   WINDOW_TITLE to an appropriate value.

 Revision 580.78  2001/11/26 03:27:08  ajc
 * new algorithm to load the use table

 Revision 580.77  2001/11/17 19:55:08  ajc
 * Updated some of the documentation

 Revision 580.76  2001/11/16 04:43:12  ajc
 * Eliminated the sock_puts_crlf() function and ensured that all SMTP client
   commands are sent out using a single sock_write() call.  There are broken
   SMTP server implementations that can't handle SMTP commands split across
   multiple writes.  (Thanks to Andru Luvisi and Ben Mehlman for the idea.)

 Revision 580.75  2001/11/15 04:11:30  ajc
 * hack.doc: updated to reflect Cit86Net compatibility fields removed from the
   file format (since we dumbed down the gateway software)
 * ipc_c_tcp.c: removed SOCKS4 support.  Nobody uses it anymore.
 * ipc_c_tcp.c: fixed a bug which caused the client to fall back to defaultPort
   if a numeric port number was specified instead of a service name

 Revision 580.74  2001/11/14 02:59:01  ajc
 * Network run frequency is now a site-definable setting

 Revision 580.73  2001/11/13 22:05:23  ajc
 * Re-introduced the ability to enter IGnet mail into the system.

 Revision 580.72  2001/10/29 22:59:22  ajc
 * Renamed "SuppMsgInfo" to "MetaData" because that's what it is

 Revision 580.71  2001/10/29 16:39:54  ajc
 * Finished the migratenet utility (finally).

 Revision 580.70  2001/10/28 05:18:51  ajc
 * migratenet almost finished

 Revision 580.69  2001/10/26 04:26:45  ajc
 * more work on the net migrator

 Revision 580.68  2001/10/23 03:37:33  ajc
 * Threw a few more lines of code into migratenet.c

 Revision 580.67  2001/10/20 18:10:50  ajc
 * migratenet.c: added (not even close to being finished)

 Revision 580.66  2001/10/17 21:07:20  nbryant
 further format string cleanups (for i686-linux type sizes)

 Revision 580.65  2001/10/17 20:41:07  nbryant
  - declare *printf format specifiers if gcc detected
  - format string fixes (compiles w/o warnings on alpha osf/1)

 Revision 580.64  2001/10/17 19:40:38  nbryant
 warning fixes and cleanups for 64-bit machines

 Revision 580.63  2001/10/16 20:47:37  nbryant
 - backed out -export-dynamic, it doesn't do anything and i've found the real
 problem
 - remove declaration for make_message

 Revision 580.62  2001/10/16 19:18:49  nbryant
 backed out that compiler detection change for Tru64. it's not incredibly
 important and results in broken autoconf macro expansions.

 Revision 580.61  2001/10/16 18:36:33  nbryant
 reinstate -export-dynamic for citserver in case libtool decides to build
 static libraries (why?)

 Revision 580.60  2001/10/16 18:21:53  nbryant
 add some explanatory text to bootstrap

 Revision 580.59  2001/10/16 17:43:53  nbryant
  - further configure tweaks for FreeBSD and Tru64 Unix
  - updated to latest libtool configure fragment
  - there are two functions named make_message. (?) so declare them both
    static.

 Revision 580.58  2001/10/16 01:48:55  nbryant
 - configury tweaks for a /usr/include/db3 goof and Digital/Tru64 Unix
 - #ifdef out inline on non-GCC compilers

 Revision 580.57  2001/10/15 19:50:50  ajc
 * Fixed a bug in the loopzapper that was corrupting the use table saved copy.
 * Post notification in Aide> when the loopzapper catches a message.

 Revision 580.56  2001/10/12 22:41:11  ajc
 * Wrote the rest of the use table code.  Finished except for a bug.

 Revision 580.55  2001/10/10 18:35:12  ajc
 * Comments & cosmetics for previous update

 Revision 580.54  2001/10/10 17:12:54  ajc
 * Bugfix for MSG0 command to properly handle multipart/alternative

 Revision 580.53  2001/10/06 21:32:29  ajc
 * Finished the concurrency check for network polling.  (Now works both for
   polling and being polled.  Severe UUCP deja vu.)

 Revision 580.52  2001/10/06 20:28:06  ajc
 * Began implementing some concurrency stuff for the networker

 Revision 580.51  2001/10/06 19:51:47  ajc
 * Stripped the build of obsolete parts of the old networker no longer in use.

 Revision 580.50  2001/10/03 20:05:50  ajc
 * serv_smtp.c: implement RFC2920 ESMTP "pipelining" extension on the server
   side.  (No changes required other than advertising the extension.)

 Revision 580.49  2001/10/03 03:15:16  ajc
 * Implemented BOUNCE BOUNCE BOUNCE

 Revision 580.48  2001/10/02 03:04:30  ajc
 * Allow non-Aides to terminate sessions belonging to them

 Revision 580.47  2001/09/24 18:55:13  ajc
 * Completed migrating the "netpoll" utility into the serv_network module.
   Removed this utility.

 Revision 580.46  2001/09/21 20:58:25  nbryant
 support different log_archive prototype in DB versions prior to 3.3

 Revision 580.45  2001/09/20 04:17:10  ajc
 * Inbound network authentication working.  Fixed a bug in the split-horizon
   algorithm.  Still need to move the 'netpoll' command into the server.

 Revision 580.44  2001/09/18 04:05:04  ajc
 * Added host/IP and port to node config (client side only)

 Revision 580.43  2001/09/17 23:55:45  ajc
 * Support for IGnet routing (not tested)

 Revision 580.42  2001/09/16 05:44:51  ajc
 * serv_smtp.c: instead of doubling delivery retry times unbounded, set a
   maximum retry time of SMTP_RETRY_MAX (currently set to 12 hours)

 Revision 580.41  2001/09/09 16:19:29  error
 * Updated PAM configuration file citadel.pam for Red Hat 7.x.

 Revision 580.40  2001/09/09 03:19:38  ajc
 * cdb_cull_logs() now removes log files as soon as the log_archive() function
   says it's ok to do so.

 Revision 580.39  2001/09/08 18:58:38  ajc
 * More changes to the new networker.  Added client command for room sharing.

 Revision 580.38  2001/09/07 04:05:27  ajc
 * You guessed it: still more code for the new networker.

 Revision 580.37  2001/09/06 05:47:29  nbryant
 check for /usr/include/db3 (for RedHat 6.2; others?)

 Revision 580.36  2001/09/06 05:23:14  nbryant
 #include fix for glibc 2.1.3

 Revision 580.35  2001/09/06 04:02:34  ajc
 * A few more updates to the networker

 Revision 580.34  2001/09/06 03:32:41  nbryant
 build fix for sparc-sun-solaris2.8; i think the dependencies should be
 set up properly for all platforms now.

 Revision 580.33  2001/09/06 02:55:27  nbryant
 build fix for Linux

 Revision 580.32  2001/09/06 02:49:22  ajc
 * Fixed paste-post mode (<.E>nter <A>scii) to append instead of replace when
   the user hits <C>ontinue (bug reported by Stu Mark)

 Revision 580.31  2001/09/06 01:26:39  nbryant
  - port to Cygwin (DLL support, etc.)
  - don't build SMTP module if there's no resolver library (eg on Windows)

 Revision 580.30  2001/09/06 00:54:01  nbryant
 updated to libtool 1.4.1 and automake 1.5

 Revision 580.29  2001/08/29 02:51:25  ajc
 * More work on the new networker.

 Revision 580.28  2001/08/25 05:04:57  ajc
 * Worked a little more on the in-server replacement for netproc

 Revision 580.27  2001/08/22 04:18:17  ajc
 * Realized that there was lots of similarly broken code in
   process_rfc822_addr().  Wrote two new utility functions in tools.c
   stripout() and stripallbut() and used them where appropriate.  This should
   take care of all possible infinite loops.

 Revision 580.26  2001/08/22 03:43:11  ajc
 * internet_addressing.c: fixed a bug in process_rfc822_addr() that caused the
   server to jump into an endless loop when an e-mail address contained
   unbalanced angle brackets.

 Revision 580.25  2001/08/15 04:26:02  ajc
 * Added split horizon and delete-after-spool to the new networker

 Revision 580.24  2001/08/14 02:41:57  ajc
 * Began the migration of netproc into part of the serv_network.c module instead
   of a standalone program.

 Revision 580.23  2001/08/11 22:35:40  nbryant
 updated citadel-with-berkeley-db.txt.
  - updated build instructions
  - improved backup procedures to be safer and more space-efficient.

 Revision 580.22  2001/08/11 19:18:41  ajc
 * Realized that I am stupid and started implementing server commands to load
   and save network configurations, when I had already lovingly implemented the
   CONF GETSYS and CONF PUTSYS commands to store arbitrary configuration sets
   in the Local System Configuration> room.  Ripped the newer crap out.
 * Implemented a skeleton of <.A>ide <S>ysconfig <N>etwork on the client side.

 Revision 580.21  2001/08/11 03:51:56  ajc
 * Removed the idle timer from the client.  Dialup is dead.

 Revision 580.20  2001/08/06 21:33:29  nbryant
 made the client fall back on port 504 if there's no /etc/services entry

 Revision 580.19  2001/08/05 23:54:14  ajc
 * prep for new network node infrastructure

 Revision 580.18  2001/08/03 16:53:21  ajc
 * Added some more "break" statements to the main switch..case loop in
   citadel.c where they were needed.

 Revision 580.17  2001/08/03 16:43:53  ajc
 * database_sleepycat.c: when running txn_checkpoint(), handle DB_INCOMPLETE
   return code as a warning instead of an error worthy of aborting the server.
   See http://www.sleepycat.com/docs/api_c/txn_checkpoint.html for explanation.

 Revision 580.16  2001/07/30 03:46:14  nbryant
 made ForEachUser use a read-only cursor, too. there is now only one piece of
 code in Citadel proper (not the database driver) that needs transactions.
 that's check_ref_counts; in other words it's the only thing standing in the way
 of a clean implementation of retryable transactions.

 Revision 580.15  2001/07/29 22:24:04  nbryant
  - added a new function to the database interface, cdb_close_cursor().  always
 call this when you're finished with a traversal but didn't bother reading all
 the way to the end.

  - removed several cdb_begin_transaction()/cdb_end_transaction() calls that are
 no longer needed because of the read-only cursor support.

 Revision 580.14  2001/07/29 20:56:09  nbryant
 change ForEachRoom to use read-only cursors by default. it can be overridden to
 still use read/write cursors by doing:

  cdb_begin_transaction();
  ForEachRoom(...);
  cdb_end_transaction();

 the only place I found where it appears necessary to do so is check_ref_counts,
 so this checkin affects that function too.

 Revision 580.13  2001/07/29 20:06:33  nbryant
 generate symlinks to .libs in modules directory

 Revision 580.12  2001/07/28 00:02:50  nbryant
 implemented read-only cursors. one of the advantages to these is that
 transactions can be avoided; a cursor operation that occurs within a
 transaction will often acquire a read lock on every single database page.  in
 general, the Sleepycat documentation recommends avoiding transaction-protected
 read-only operations where practical. read/modify/write operations can still
 be transaction protected, of course.

 to use a read-only cursor, call cdb_rewind without a previous call to
 cdb_begin_transaction. the DB driver will notice this and prevent the current
 thread from modifying data or starting a transaction until the cursor is
 closed.

 Revision 580.11  2001/07/27 20:45:44  nbryant
 libtool has matured a lot since the last time i looked at it (years ago)
 so we now use it to handle the details of building shared libraries and
 the linker flags for the main executable.

 in theory this should bring a lot more portability to the dynloader
 subsystem and enable us to do things like transparently detect GNU vs Sun
 linkers on solaris, for example

 Revision 580.10  2001/07/27 03:29:04  nbryant
 missed one thing in the autoconf move

 Revision 580.9  2001/07/27 02:57:43  nbryant
 support one cursor per database rather than one global cursor

 Revision 580.8  2001/07/27 01:32:07  nbryant
 remove the automatic transaction demarcation on singleton read operations

 Revision 580.7  2001/07/26 21:43:46  nbryant
  - move to autoconf 2.52
  - random warning fix
  - check for db 3.3

 Revision 580.6  2001/07/24 13:17:54  ajc
 * New UI for mailing list setup
 * rooms.c: code cleanup
 * docs update

 Revision 580.5  2001/07/20 23:48:23  nbryant
 fix build on solaris, check default install location for db 3.2, and silence
 gcc 3.0

 Revision 580.4  2001/07/16 14:24:30  ajc
 * Silly cosmetic change to keep the wholist ordered by ascending session number

 Revision 580.3  2001/07/13 00:01:36  ajc
 * Shuffled around some of the housekeeping loop code

 Revision 580.2  2001/07/11 17:01:10  ajc
 * database_sleepycat.c: small changes to log messages

 Revision 580.1  2001/07/11 04:35:40  nbryant
 moved dret initialization in cdb_fetch, just in case

 Revision 580.0  2001/07/03 03:07:06  ajc
 * THIS IS 5.80

 Revision 573.143  2001/07/03 03:06:50  ajc
 * Last minute doco update for 5.80

 Revision 573.142  2001/07/01 15:44:32  nbryant
 configure.in: check for Berkeley DB first.

 Revision 573.141  2001/06/27 23:34:30  ajc
 * Added some verbage to messages/roomaccess to placate a user who keeps
   bitching about privacy policy.

 Revision 573.140  2001/06/19 03:41:04  ajc
 * Ooops... last_cull needs to be declared static

 Revision 573.139  2001/06/19 03:33:19  ajc
 * imap_fetch.c: download MIME parts without decoding first.  We like that.
 * database_sleepycat.c: added automatic culling of log files which have not
   been written to in five days.

 Revision 573.138  2001/06/17 19:42:23  nbryant
 fix all the <time.h> vs. <sys/time.h> issues, hopefully

 Revision 573.137  2001/06/07 03:28:37  ajc
 * More tweaks to the MIME parser

 Revision 573.136  2001/06/06 15:44:37  ajc
 * msgbase.c: output extra newline at end of RFC822 message if necessary to
   ensure that 000 termination string appears on a line by itself.

 Revision 573.135  2001/06/06 04:22:25  ajc
 * Moved memreadline() to tools.c
 * internet_addressing.c: fixed conversion of fields to (hopefully) never get
   into an active loop when encountering badly formed headers

 Revision 573.134  2001/05/27 05:23:03  ajc
 * Added a "no new messages" response in the client, displayed when a read
   command turns up a zero message count.

 Revision 573.133  2001/05/18 20:12:09  ajc
 * Fixed bug in mime_parser.c that caused parts to be dropped when the last
   boundary was the very last line of the message.
 * serv_smtp.c: toned down some of the command response verbage.

 Revision 573.132  2001/04/28 04:42:55  ajc
 * Updated some of the docs.  Bumped version number to 5.80 in anticipation
   of going into a release cycle soon.

 Revision 573.130  2001/04/26 03:31:00  ajc
 * Finished the implementation of per-message seen/unseen logic, both in the
   server proper and in IMAP.  Citadel protocol uses new "seen" command.

 Revision 573.129  2001/04/21 04:55:51  ajc
 * Began implementation of per-message seen/unseen attribute

 Revision 573.128  2001/04/20 03:39:54  ajc
 * IMAP LIST/LSUB: made it case insensitive.  Also minor IMAP code cleanup.

 Revision 573.127  2001/04/17 00:35:19  cough
 * Modified rooms.c in the client so that it would allow inviting into
   public rooms.  This is important since there is now a V_LOCKOUT
   flag which prevents users who have been kicked from rejoining
   a room unless/until you invite them back in.

 Revision 573.126  2001/04/16 19:21:14  cough
 * Fixed bug in room_ops.c that wasn't allowing aides to goto passworded
   rooms without knowing the password.

 Revision 573.125  2001/04/14 04:26:44  ajc
 * Fixed an unterminated string bug in IMAP APPEND.  Storing messages should
   work now.

 Revision 573.124  2001/04/10 01:04:10  ajc
 * Finished coding IMAP APPEND.  It works, but there's a bug in it somewhere
   that is corrupting the memory.

 Revision 573.123  2001/04/03 00:47:23  ajc
 * Began implementing IMAP APPEND

 Revision 573.122  2001/04/01 22:05:44  cough
 * *Actually* fixed a botched ChangeLog entry.

 Revision 573.121  2001/04/01 22:04:28  cough
 * Fixed a botched ChangeLog entry.

 Revision 573.120  2001/04/01 22:03:10  cough
 * Changed two fclose()s to pclose()s.

 Revision 573.119  2001/03/25 11:52:36  error
 * serv_pop3.c: Fixed APOP. Now logs in properly. Also cleaned up some non-
   RFC-compliant error messages.

 Revision 573.118  2001/03/21 05:47:49  ajc
 * Added the new IMAP mailbox string compare submitted by Daniel Malament.

 Revision 573.117  2001/03/20 01:33:55  ajc
 * Added the (\NoInferiors) tag to all rooms listed in IMAP.  This made Mozilla
   behave very nicely.

 Revision 573.116  2001/03/13 17:19:33  ajc
 * support (BODY[HEADER.FIELDS(BLAH BLAH BLAH)]) and HEADER.FIELDS.NOT

 Revision 573.115  2001/03/12 01:27:42  ajc
 * Implemented SUBSCRIBE and UNSUBSCRIBE commands

 Revision 573.114  2001/03/11 23:00:29  ajc
 * Mega sexy hack to deliver express messages THROUGH IMAP!  uber coolness!!

 Revision 573.113  2001/03/11 22:09:20  ajc
 * Replaced the "citlogin" binary wrapper with the "loginwrapper.sh" script.

 Revision 573.112  2001/03/11 20:06:53  ajc
 * Fixed bug that created incorrect roomnames when sending pages

 Revision 573.111  2001/03/11 19:23:32  ajc
 * IMAP DELETE command ... also split up access control for room delete cmds

 Revision 573.110  2001/03/10 17:29:07  ajc
 * Implement proper access control for deleting messages from IMAP

 Revision 573.109  2001/03/07 04:02:27  ajc
 * Fixed some small IMAP bugs

 Revision 573.108  2001/03/06 04:44:00  ajc
 * Probable completion of STATUS, COPY, STORE, and EXPUNGE commands in IMAP

 Revision 573.107  2001/03/06 03:31:58  nbryant
 database-related cleanups and paranoia tests;
 fixed a transaction-leak/deadlock problem in cdb_delete;
 solved the SIGPIPE mystery (GDB stops on SIGPIPE is all it was)

 Revision 573.106  2001/03/05 04:59:31  ajc
 * IMAP COPY

 Revision 573.105  2001/03/04 23:49:41  ajc
 * IMAP EXPUNGE responses -should- be correct now

 Revision 573.94  2001/02/20 00:02:56  ajc
 * IMAP: implemented the STATUS command (sort of).

 Revision 573.93  2001/02/19 22:24:41  ajc
 * IMAP server: added untagged, unsolicited server messages for newly arrived
   messages, and messages expunged by another session.

 Revision 573.92  2001/02/17 05:53:35  ajc
 * Repaired the creation of page log rooms in the wrong namespace when the
   recipient does not yet have his/her log room created.
 * Rewrite "EXPI messages" to run in two passes: one to gather messages to
   purge and the next to delete them.  Works better in transactionland.

 Revision 573.91  2001/02/14 08:11:27  error
 * citadel.rc: added RCS ID (it's about time!)

 Revision 573.90  2001/02/14 04:23:54  ajc
 * Fixed POP3 server responses ending in \n instead of \r\n as they should be.
   This was causing some clients (such as Pine) to lock up.

 Revision 573.89  2001/02/13 04:06:14  ajc
 * Worked out the remaining bugs in IMAP FETCH for the BODYSTRUCTURE and
   BODY[n] data items.  I think.  So much protocol crud, so little time...

 Revision 573.88  2001/02/13 01:18:44  ajc
 * imap fetch

 Revision 573.87  2001/02/12 04:31:34  ajc
 * sysdep.c ig_tcp_server() - use IPPROTO_TCP instead of getprotobyname()

 Revision 573.86  2001/02/08 04:45:58  ajc
 * Fixed namespace problems resulting from the automatic namespece prefixing
   added to create_room().  Also added the ability to specify "create a mailbox
   but I've already supplied the namespace prefix" for situations where the
   namespace isn't that of the logged in user.
 * Made the POP3 server response messages slightly less humorous

 Revision 573.85  2001/02/06 04:44:12  ajc
 * Added a floor listing (complete with \NoSelect flag) to LIST and LSUB

 Revision 573.84  2001/02/06 02:09:38  ajc
 * citadel.rc: changed the default for local_screen_dimensions to 1, since
   Internet users now outnumber dialup users 100 to 0.
 * room_ops.c: added a really_create option to create_room().  Also moved the
   generation of personal namespace into that function.  MODULE OWNERS PLEASE
   CHECK YOUR CALLS TO AVOID MULTIPLE NAMESPACING!!
 * room_ops.c: fixed a bug in cgetfloor() that left bad pointers around
 * serv_imap.c: finished the CREATE command (finally)

 Revision 573.83  2001/02/05 05:20:22  ajc
 * Made some changes to functions which translate between Citadel room names
   and IMAP folder names.  They're still buggy.

 Revision 573.82  2001/02/04 23:17:28  ajc
 * Implemented the IMAP CREATE command

 Revision 573.81  2001/02/04 02:40:07  ajc
 * more imap.  imap sucks.  die crispin die.

 Revision 573.80  2001/02/03 10:02:12  error
 * serv_ical.c: Verify that objects posted to My Calendar> are of type
   text/x-calendar or text/calendar; abort saving if not

 Revision 573.79  2001/02/03 09:30:46  error
 * serv_ical.c: now creates a My Calendar> personal room, sets attributes

 Revision 573.78  2001/02/03 08:21:00  error
 * serv_ical.c and serv_ical.h added; skeleton code for now

 Revision 573.77  2001/02/02 20:18:18  ajc
 * Changed the error message in cdb_delete() to actually *say* cdb_delete
   instead of cdb_store.  Useful to know which function failed...

 Revision 573.76  2001/02/01 04:08:03  ajc
 * IMAP minor change to mailbox name output
 * Increased size of buffer in lprintf()

 Revision 573.75  2001/01/28 09:50:02  error
 * sysdep.c: lprintf() now generates timestamps

 Revision 573.74  2001/01/28 07:35:04  error
  * serv_bio.c: RBIO now also returns stats about a user, see session.txt

 Revision 573.73  2001/01/16 04:03:13  ajc
 * yeesh ... more on the IMAP BODYSTRUCTURE

 Revision 573.72  2001/01/16 01:51:12  ajc
 * imap bodystructure

 Revision 573.71  2001/01/16 00:46:40  ajc
 * Changed the MIME parser API *again* because we now need the ability to
   supply callback functions to be executed before and/or after parsing a
   multipart.  (Need this for IMAP BODYSTRUCTURE output.)

 Revision 573.70  2001/01/15 23:59:26  ajc
 * user_ops.c: reject NULL password in CtdlTryPassword() instead of crashing

 Revision 573.68  2001/01/15 20:34:04  ajc
 * "Path:" removed for now because it confuses brain-dead Microsoft shitware
   into thinking that mail messages are newsgroup messages instead.  When we
   add NNTP support back into Citadel we'll have to add code to only output
   this field when appropriate.

 Revision 573.67  2001/01/15 16:30:31  ajc
 * temporary implementation of 901 asynchronous express messages

 Revision 573.66  2001/01/14 14:55:39  ajc
 * Changed the format of <.W>holist <L>ong

 Revision 573.65  2001/01/13 06:40:26  nbryant
 merged remaining changes from TRANSACTIONS (using cvs update -j TRANSACTIONS)
 which should now be considered closed.

 Revision 573.64  2001/01/13 06:12:15  ajc
 * Added the ASYN command

 Revision 573.63  2001/01/12 22:05:46  ajc
 * Fixed a bug that caused bogus wholist lines to be displayed when a non-aide
   reads a list containing stealth mode sessions.

 Revision 573.62  2001/01/09 05:39:45  ajc
 * Merged in code from the TRANSACTIONS branch for testing.

 Revision 573.61  2000/12/30 06:17:17  ajc
 * Still more work on IMAP.  Damn this is tedious.

 Revision 573.60  2000/12/30 04:55:05  ajc
 * more buffer size stuff

 Revision 573.59  2000/12/27 20:19:51  ajc
 * The size constant "256" which shows up everywhere as a buffer size has now
   been changed to SIZ.  And, SIZ has been defined now as 1024, not 256, because
   we need 1024 byte buffers for most Internet protocols.

 Revision 573.58  2000/12/27 05:09:58  ajc
 * Added a skeleton IMAP "SEARCH" command (based on the FETCH logic)

 Revision 573.57  2000/12/26 03:46:50  ajc
 * More IMAP tweaks

 Revision 573.56  2000/12/25 22:50:43  ajc
 * Added an API function to extract and unfold specific RFC822 fields.
 * imap-->fetch-->envelope-->in-reply-to now works
 * More robust checking and reporting of temp file errors in the client

 Revision 573.55  2000/12/25 20:43:24  ajc
 * imap_fetch.c: added support for fetch-->envelope-->from

 Revision 573.54  2000/12/20 04:09:24  ajc
 * A few memory handling fixes to netproc.

 Revision 573.53  2000/12/20 01:57:37  ajc
 * netproc.c: added bounds check to fpgetfield()

 Revision 573.52  2000/12/19 20:41:55  ajc
 * Fixed generation of unique file names for network uploads etc.

 Revision 573.51.2.11  2000/12/26 05:30:55  nbryant
 remove extraneous transaction around dynamic module initializations. this will
 fix the crash on database creation.

 Revision 573.51.2.10  2000/12/24 23:00:58  nbryant
 clean: also remove parsedate.c

 Revision 573.51.2.9  2000/12/20 01:38:42  nbryant
 require transactional cursors

 Revision 573.51.2.8  2000/12/20 00:30:01  nbryant
 release any stale db handles at the end of a server command
 (unfinished transactions will be aborted to annoy lazy programmers)

 Revision 573.51.2.7  2000/12/19 06:18:27  nbryant
 set sched_yield as sleepycat's yield function. this should improve locking
 performance.

 Revision 573.51.2.6  2000/12/19 02:22:29  nbryant
 added automatic transaction start/end on cdb_fetch, cdb_delete, and cdb_store

 Revision 573.51.2.5  2000/12/18 03:51:13  nbryant
 ditto S_USER_TRANS, S_CALLLOG, and S_HOUSEKEEPING, which are no longer used at
 all

 Revision 573.51.2.4  2000/12/18 02:49:17  nbryant
 removed all references to S_MSGMAIN critical section; it wasn't really needed.
 this should make things significantly more scaleable.

 Revision 573.51.2.3  2000/12/17 22:12:48  nbryant
 reworked shutdown sequence to wait for worker threads to terminate before
 checkpointing and closing databases. it is no longer safe to call
 master_cleanup() directly to force a shutdown; instead, just set
 time_to_die to a nonzero value

 Revision 573.51.2.2  2000/12/17 05:06:09  nbryant
 added deadlock detection and cleaned up messages

 Revision 573.51.2.1  2000/12/16 21:06:59  nbryant
 created TRANSACTIONS branch
 track cursor and transaction id's in thread-specific data

 Revision 573.51  2000/12/14 18:36:34  ajc
 * Fixed the "users not in chat" wholist display

 Revision 573.50  2000/12/12 18:06:46  ajc
 * Removed the transaction stuff (but left the log in).  It wasn't working.

 Revision 573.49  2000/12/12 06:19:55  ajc
 * Stabilize, dammit!!

 Revision 573.48  2000/12/12 04:20:03  ajc
 * Made the transaction open/close a global thing, in a frantic attempt to get
   Uncensored to stop crashing.  More fixes on the way...

 Revision 573.47  2000/12/11 06:08:41  ajc
 * Removed the housekeeper thread, moved terminate_idle_sessions() out to a
   timer event, and check_sched_shutdown() to the end of the worker thread
   loop.  Seems to have improved reliability (but why?)

 Revision 573.46  2000/12/11 03:22:11  ajc
 * Added server-side REQT command to issue client termination requests

 Revision 573.45  2000/12/11 02:19:26  ajc
 * Client now honors EM_GO_AWAY flag, used by the server to request that a
   client log off.  (The server doesn't support sending that flag yet, though)

 Revision 573.44  2000/12/09 06:20:06  ajc
 * A few final touches to the Sleepycat DB back-end

 Revision 573.43  2000/12/08 17:06:33  ajc
 * Wrap txn_begin and txn_end in S_DATABASE mutex

 Revision 573.42  2000/12/07 20:21:39  ajc
 * begin/end transaction in master_startup()

 Revision 573.41  2000/12/07 16:59:02  nbryant
 added --with-db and --with-gdbm options to configure

 Revision 573.40  2000/12/07 04:50:33  ajc
 * Wrap housekeeper and timer events in transaction open/close functions
 * Checkpoint the DB as an EVT_TIMER event instead of after each session
   (runs each minute, but actually limited by the parameters of the function)

 Revision 573.39  2000/12/06 04:44:36  ajc
 * Changed netproc to keep the use table in a flat file instead of a database

 Revision 573.38  2000/12/05 05:32:58  ajc
 * Added support for non-USA country identities in vCard and registration
 * User edit now asks whether it should prompt to change the password

 Revision 573.37  2000/12/03 04:12:21  ajc
 * Finished (mostly) the Sleepycat DB backend ... added transaction logging

 Revision 573.36  2000/11/30 03:23:17  ajc
 * Got the Sleepycat DB back end working, by opening the databases in a non
   shared, non threaded mode, and using Citadel's locking to serialize access.

 Revision 573.35  2000/11/29 05:00:02  ajc
 * I think the db stuff is ok, but my db library is fux0red...

 Revision 573.34  2000/11/27 14:12:09  error
 * commands.c: fixups to print_express() to make external command not print
   extraneous stuff to the terminal and make the displayed message consistent

 Revision 573.33  2000/11/27 10:41:14  error
 * print_express(): now uses GEXP instead of old PEXP; displays timestamps

 Revision 573.32  2000/11/27 10:29:59  error
 * serv_chat.c: fix send_express_message() to include timestamps

 Revision 573.31  2000/11/27 03:44:27  ajc
 * Initial checkin of database_sleepycat.c (doesn't work yet)

 Revision 573.30  2000/11/26 05:24:22  ajc
 * msgbase.c: Added new API function CtdlOutputPreLoadedMsg(), and
   re-implemented the existing CtdlOutputMsg() as a wrapper around it.
 * imap_fetch.c: used the above function to do all output pre-loaded

 Revision 573.29  2000/11/25 09:36:18  error
 * Added a bit of detail to syslog entries; now shows session id attached to
   client, hostname, and username, and time the session ended.

 Revision 573.28  2000/11/25 06:17:06  ajc
 * Minor IMAP tweaks.  It still doesn't work.  :(

 Revision 573.27  2000/11/23 07:22:21  error
 * citadel.spec: update version number

 Revision 573.26  2000/11/21 11:12:56  error
  * domain.h: changed the HP/UX compatibility code to use defines from
    typesize.h for integers of specific bit widths (needed for Solaris, etc)

 Revision 573.25  2000/11/12 04:20:49  ajc
 * Optimized server side input of message text

 Revision 573.24  2000/11/10 03:55:06  ajc
 * Ford's Fix for Faster Functionality (save position during reply)

 Revision 573.23  2000/11/09 04:48:50  ajc
 * tools.c: striplt() strips all whitespace, not just spaces

 Revision 573.22  2000/11/07 20:47:21  ajc
 * imap_fetch.c: added a skeleton "ENVELOPE" fetch.  Currently sends NIL's.

 Revision 573.21  2000/11/07 15:54:53  ajc
 * xx FETCH n:n BODY[pn.MIME] now works

 Revision 573.20  2000/11/06 05:10:01  ajc
 * Changed the mime_parser() API (again) to allow "don't decode" mode

 Revision 573.19  2000/10/29 18:11:07  ajc
 * Start numbering top-level MIME parts as 1, 2... not 1.1, 1.2...

 Revision 573.18  2000/10/28 14:14:19  error
 * msgbase.c: eliminated most gotos; a single goto in alias() remains because
   it actually makes sense to do it that way...

 Revision 573.17  2000/10/25 21:37:09  ajc
 * Implemented the AUTHENTICATE LOGIN command in IMAP

 Revision 573.16  2000/10/25 19:20:37  ajc
 * FETCH now works for ranges *and* sets, and with sequence numbers *and* UID's

 Revision 573.15  2000/10/24 20:39:59  ajc
 * Added RFC822, RFC822.HEADER, RFC822.SIZE, RFC822.TEXT fetch keys to IMAP

 Revision 573.14  2000/10/23 20:26:51  error
 * War on goto:  rewrote a few easy functions to eliminate unnecessary gotos

 Revision 573.13  2000/10/11 23:03:44  error
 * utilsmenu: obey $PAGER environment var, if any.  Default to more if neither
   $PAGER nor less is available.

 Revision 573.12  2000/10/11 22:55:25  error
 * citadel.c: when ansi_color=user, enable color at login, so Lobby> posts
   displayed at login are in color

 Revision 573.11  2000/10/11 22:47:51  error
 * domain.c: getmx() returns hostname as MX if no MX records found a la RFC 974

 Revision 573.10  2000/10/10 19:18:12  ajc
 * Added support of macros ALL, BODY, FAST, and FULL to the IMAP FETCH command

 Revision 573.9  2000/10/06 03:31:55  ajc
 * IMAP is a convoluted mess.

 Revision 573.8  2000/10/05 22:23:16  ajc
 * Slowly and painfully writing IMAP support

 Revision 573.7  2000/10/04 22:39:06  ajc
 * Added skeleton versions of the LIST and LSUB commands to the imap server

 Revision 573.6  2000/10/04 17:48:21  ajc
 * Allow Aides to zap rooms (site configurable setting)

 Revision 573.5  2000/10/03 01:45:00  ajc
 * Changed the <.A>ide <S>ysconfig <G>eneral command to explicitly allow the
   global page log room to be disabled (answer "no" to set the log room to a
   null string)

 Revision 573.4  2000/09/28 10:27:38  error
 * commands.c: changed stty_ctdl() to support HP/UX termios VMIN and VTIME

 Revision 573.3  2000/09/24 22:01:45  ajc
 * ipc_c_tcp.c: don't hardcode CTDLDIR path for unix domain sockets

 Revision 573.2  2000/09/21 04:16:44  ajc
 * Fixed logged_in_response() so it only displays responses during Citadel
   protocol sessions.  (This was affecting POP etc.)

 Revision 573.1  2000/09/11 22:05:04  ajc
 * citadel.c: accept -h <host> and -p arguments, so citadel can be called
   directly by telnetd, bypassing /bin/login.  It works, but not recommended at
   this time because it has to run as root.

 Revision 573.0  2000/09/05 18:35:22  ajc
 * Tagged everything for version 5.73 release

 Revision 572.39  2000/09/04 03:59:15  ajc
 IO's changes:
 
 revision 572.4
 date: 2000/09/03 06:36:01;  author: error;  state: Exp;  lines: +5 -1
 Added HP/UX linker flag for dynamic modules to work
 
 revision 572.3
 date: 2000/09/01 06:50:00;  author: error;  state: Exp;  lines: +8 -1
 Changed to use integer macros from typesize.h for specific bit widths

 Revision 572.38  2000/09/01 20:17:08  ajc
 * msgbase.c: cmd_opna() - increase desired_section buffer from 64 to 256 bytes

 Revision 572.37  2000/09/01 17:31:47  ajc
 * Fixed oopseth in control.c that might call fileno(NULL)

 Revision 572.36  2000/09/01 13:37:16  ajc
 * control.c: chown citadel.control to ctdluid when opening/creating as root

 Revision 572.35  2000/09/01 03:55:44  ajc
 * Fixed a few more references to the deprecated uncnsrd.mt-kisco.ny.us name

 Revision 572.34  2000/09/01 03:43:09  ajc
 * Added 'author' command-line arg to aidepost.  Closes enhancement request
   #71 on bugzilla.
 * Put the default SMTP and POP3 ports back to 25 and 110.  Now that the
   unix domain socket bug is fixed, it's ok if these binds fail.

 Revision 572.33  2000/08/31 23:02:15  ajc
 * ig_tcp_server() and ig_uds_server()  -  check to make sure queue length is
   always at least 5.  Zero-length queues can cause connection lockups.

 Revision 572.32  2000/08/31 21:32:44  ajc
 * Still trying to fix a socket connect bug

 Revision 572.31  2000/08/31 16:37:08  ajc
 * docs/import-export.txt: added.

 Revision 572.30  2000/08/28 19:51:51  ajc
 * messages.c: cosmetic cleanup (coding convention and comments)

 Revision 572.29  2000/08/26 20:23:18  ajc
 * Finished up the back end code for mailing list sends.  Sends now work!

 Revision 572.28  2000/08/24 02:48:18  ajc
 * Merged in IO ERROR's diffs to make Citadel work with HP/UX

 Revision 572.27  2000/08/22 02:31:47  ajc
 * nonce (for APOP-style auth) is now generated when a context is created
   instead of during protocol greeting functions.
 * Moved Citadel protocol nonce output from greeting to INFO

 Revision 572.26  2000/08/18 21:09:36  ajc
 * Added a little more mailing list code to serv_network.c

 Revision 572.25  2000/08/10 04:36:25  ajc
 * Fixed a bug in keyboard polling (in commands.c) which was causing the
   client protocol to get out of sync in certain conditions.

 Revision 572.24  2000/08/10 02:42:13  ajc
 * Changed all the "200 ok" responses to more descriptive strings
 * Added a *temporary* protocol sync check.  Remove this!

 Revision 572.23  2000/08/09 17:14:34  ajc
 msgbase.c: fixed a bug in
            remove_any_whitespace_to_the_left_or_right_of_at_symbol() that was
            causing the <R>eply function to fail on names with whitespace in
            certain parts of the string.  This closes Bug #56.

 Revision 572.22  2000/08/05 04:24:00  ajc
 * Added [idle] to client wholist display for sessions idle >15 minutes
 * Added a generic "void *userdata" field to CtdlForEachMessage()
 * More prep for mailing list handling in the server

 Revision 572.21  2000/07/30 04:36:12  ajc
 * Set up the SNET (Send NETwork config) and GNET (Get NETwork config) commands
   for the network overhaul.

 Revision 572.20  2000/07/29 05:29:19  ajc
 * Changed the format of RWHO output to provide non-masqueraded user/room/host
   names (to Aides only) as additional fields rather than an extra line of
   output.
 * Changed the client to display new RWHO fields, in the "long" wholist only.
 * Default SMTP and POP ports are now -1, not 25/110.  These services must now
   be activated manually.

 Revision 572.19  2000/07/24 00:39:13  ajc
 * Fixed a path problem when calling netmailer from mailinglist.c

 Revision 572.18  2000/07/22 03:44:17  ajc
 * Prepared infrastructure for the networker rewrite

 Revision 572.17  2000/07/17 02:38:08  ajc
 * Completed serv_vandelay.c (importer/exporter module)
 * sendcommand.c: fix behavior of SEND_LISTING mode
 * sysdep.c: client_gets() fill buffer with "000" terminator when returning -1

 Revision 572.16  2000/07/14 03:06:55  ajc
 * Added .ATN (DOWN) and .ATS (SCDN 1) commands to the client

 Revision 572.15  2000/07/10 23:36:08  ajc
 * Another attempt to fix the crashy crashy bug in serv_vcard
 * Did more work on the Art Vandelay module
 * Replaced all instances of sprintf(tempfile, tmpnam(NULL)) with strcpy()

 Revision 572.14  2000/07/10 04:01:12  ajc
 * added an unfinished serv_vandelay.c (Art Vandelay module - importer/exporter)

 Revision 572.13  2000/07/09 02:47:40  ajc
 * Overhauled the keepalive logic in the chat client.  Closes bug #20.

 Revision 572.12  2000/07/09 02:27:02  ajc
 * Eliminated the whole SIGINT/SIGQUIT based handling of Ctrl-O and Ctrl-C
   keyboard interrupts.  Replaced with a non-blocking check for keyboard input
   which sets the global variable 'sigcaught' if either key was pressed.
   fmout() and pprintf() switch to 'drain mode' if sigcaught is set.
   This closes Bug #18.

 Revision 572.11  2000/07/06 20:26:36  ajc
 * updated .Help SUMMARY

 Revision 572.10  2000/07/04 20:02:46  ajc
 * Fixed potential crashy crashy bug in serv_vcard.c

 Revision 572.9  2000/06/28 03:42:56  ajc
 * Changed the comments at the beginning of each file to a consistent format
 * Improved the parameterization of commands in the IMAP module

 Revision 572.8  2000/06/27 01:27:13  ajc
 * Coupla very small changes to get on the road to IMAP support

 Revision 572.7  2000/06/22 21:41:48  ajc
 * Made the ICQ stuff far more reliable ... by removing it!

 Revision 572.6  2000/06/21 03:46:20  ajc
 * IMAP is now legal but useless, supporting NOOP, LOGIN, and LOGOUT.

 Revision 572.5  2000/06/15 20:15:52  ajc
 * Inserted a skeleton IMAP module into the build.  IT DOES NOT WORK AT ALL.

 Revision 572.4  2000/06/04 02:30:56  ajc
 * CtdlForEachMessage() now returns the number of messages processed.  It also
   accepts the MSGS_EQ mode, for targeting a specific message number (useful
   for determining whether the specified message actually exists in a room).
 * Completed the server side of the moderation system (serv_moderate.c module
   which implements the MMOD command)

 Revision 572.3  2000/06/03 05:47:57  ajc
 * Replaced most of the very repetitive and very redundant access level checks
   in most commands with a single API call:  CtdlAccessCheck()
 * serv_moderate.c: added (not finished)

 Revision 572.2  2000/06/02 03:38:50  ajc
 * Bind unix socket prior to TCP socket for citadel protocol
 * Fixed bug in sendcommand.c which was causing it to crash on attach

 Revision 572.1  2000/05/26 19:27:51  ajc
 * Changed some of the rev-level sensitive stuff to look at the actual version
   of Citadel running, not the last version with which we ran setup
 * Added a moderation system.  Default filter level for new users is in the
   config file.  Per-user setting is in usersupp.  Moderation level of each
   message is in SuppMsgInfo.  Tweaked CONF, GETU, and SETU.  Read filter is
   working.  Moderate message up/down commands are not here yet.  See
   techdoc/moderation.txt for more info.

 Revision 572.0  2000/05/23 02:09:30  ajc
 * Updated docs and tagged everything for the 5.72 release

 Revision 571.7  2000/05/20 23:28:20  ajc
 * Fixed bug in client API that was causing netproc to crash

 Revision 571.6  2000/05/15 00:05:19  ajc
 * Double the retry interval for SMTP deliveries after each failed attempt.

 Revision 571.5  2000/05/11 03:08:47  ajc
 * serv_smtp.c: clear the relevant state buffers after an SMTP DATA command
   completes, allowing multiple messages in one session.  Closes bug #58.

 Revision 571.4  2000/04/24 03:36:43  ajc
 * Removed references to strucmp() and struncmp(), replaced them with
   strcasecmp() and strncasecmp() as we did in the server a while ago, and
   set up the config script to figure out whether they need to be compiled
   in.  Also moved them to tools.h
 * Wrote a password manager for the client

 Revision 571.3  2000/04/19 03:17:10  ajc
 * Don't ever expire the Local System Configuration> room

 Revision 571.2  2000/04/16 19:03:47  ajc
 * Minor changes to avoid host lookup for local clients

 Revision 571.1  2000/04/15 19:55:52  ajc
 * Fixed "Unvalidated users appear even with registration disabled" bug #36

 Revision 571.0  2000/04/13 02:43:24  ajc
 * Fixed a problem that crashed the client during <G>oto commands if a room
   existed with a name more than 32 characters (thanks to Magus for reporting
   this one).
 * Tagged everything and updated docs for the 5.71 release

 Revision 570.18  2000/04/10 01:47:22  ajc
 * More paginator changes.  Shuffled code around, added pagination to a bunch
   of other functions, and replaced the old, cumbersome pagination with the
   new, easy, API-based one in everything except message output.

 Revision 570.17  2000/04/09 17:51:18  ajc
 * Added pprintf() (paginated version of printf) to the client-side API.  Now
   any client side function can be paginated simply by changing all of the
   printf's to pprintf's.  I've already done this for the user list and wholist
   (removing the old style pagination) and for "read directory."

 Revision 570.16  2000/04/08 04:52:48  ajc
 * Another minor stoopid little time display fix

 Revision 570.15  2000/04/08 03:58:12  nbryant
  * backed out the previous changes since they didn't look good on some
    xterms with a blue bold mode. just set the background color to black
    instead. this makes black-on-white terminals actually readable, if not
    particularly good looking.

 Revision 570.14  2000/04/08 01:36:30  nbryant
  * citadel.c, commands.c: comment out cls() as this wasn't called if
    ansi_color=user anyway, and we have no way of knowing whether the user's
    terminal does background color erase, which was the reason for adding this
    in the first place. with the recent changes to color(), the display will get
    screwed up if they *do* have background color erase. perhaps this is a job
    for terminfo.

 Revision 570.13  2000/04/08 00:37:42  nbryant
  * commands.c: fix color support for black-on-white color terminals such as
    CDE's dtterm. this just avoids white-on-white text; some color combinations
    such as yellow-on-white still aren't very legible due to contrast. There's
    not much I can think to do about that without affecting the look on the
    white-on-black terminals which probably account for most users.

 Revision 570.12  2000/04/07 20:22:34  ajc
 * Fixed am/pm bug in time display

 Revision 570.11  2000/04/07 19:22:45  ajc
 * For services disabled by setting port number to -1, administratively skip
   the bind instead of just allowing it to fail.

 Revision 570.10  2000/03/31 04:31:02  ajc
 * Quick fix to 'stats' utility to purge records with bogus timestamps

 Revision 570.9  2000/03/31 02:10:52  ajc
 * Caved in to pressure and enabled <R>eply in public rooms.

 Revision 570.8  2000/03/28 03:55:53  ajc
 * Modified <.RU> to allow search for partial match

 Revision 570.7  2000/03/27 03:08:19  ajc
 * Third parameter of CtdlDeleteMessages() now takes "" instead of NULL to
   specify 'any MIME type'.

 Revision 570.6  2000/03/27 01:14:08  nbryant
  * Makefile.in: generate parsedate.c automatically
  * configure.in: check for bison/byacc/yacc. bison seems to build cleaner
    code.
  * parsedate.c: removed from CVS
  * parsedate.y: added #include <stdlib.h> so it builds with bison

 Revision 570.5  2000/03/27 00:46:10  nbryant
  * First cut at Solaris fixes. There may still be some *printf("%s", NULL)
    type of issues lurking in the shadows.

 Revision 570.4  2000/03/25 18:29:58  nbryant
 changed my email address

 Revision 570.3  2000/03/23 02:41:50  ajc
 *** empty log message ***

 Revision 570.2  2000/03/21 03:23:24  ajc
 * Experimental new linebreak mode

 Revision 570.1  2000/03/20 14:43:19  ajc
 * "Brown paper bag" fix for SMTP bug (incoming long fields crash logger)

 Revision 570.0  2000/03/19 23:42:34  ajc
 * This is the official 5.70 release.

 Revision 1.494  2000/03/19 23:04:08  ajc
 * Small cosmetic change to date/time output

 Revision 1.493  2000/03/19 05:02:39  ajc
 * SMTP hacks to deal with AOL braindamage

 Revision 1.492  2000/03/18 18:18:04  ajc
 * Support multiline responses from SMTP servers when sending mail

 Revision 1.491  2000/03/17 16:26:57  ajc
 * Set up a private "Sent/Received Pages" room for each user

 Revision 1.490  2000/03/17 04:11:24  ajc
 * Moved bio-related commands out to a loadable module

 Revision 1.489  2000/03/16 17:58:54  smw
 Created a docs directory.
 Moved install.txt to docs.
 Added inetmailsetup.txt (Citadel server side e-mail)
 Added inetmailsetupmx.txt (local mail AND Citadel e-mail
 Added inetsiteconfig.txt (describes the .asi command)
 Added siteconfig.txt (describes .asg)
 Added chat.txt (describes changes and new chat functionality)
 Made a couple of changes to install.txt (references to new documentation)
 Added Steve Williams to copyright.txt as the document writer.

Revision 1.488  2000/03/15 03:04:51  ajc
* Added DEXP server command to disable incoming express messages.
* <Q>uiet mode client side command to set/clear DEXP mode.
* <K>nown rooms list displays through the paginator.

Revision 1.487  2000/03/12 00:21:35  ajc
* Removed the semi-broken "chat room" functionality in the chat system, and
  replaced it with direct mapping to the actual Citadel room the user is in.
* Display masqueraded roomname in chat, if applicable.  Fixes bug #19.

Revision 1.486  2000/03/11 21:29:37  ajc
* SM_ flags for CtdlSaveMsgPointerInRoom() need to be a bit bucket, not an
  enum, since more than one can be passed.  Changed SM_DONT_BUMP_REF_COUNT
  from 3 to 4.  This also fixes bug #33.

Revision 1.485  2000/03/11 20:26:03  ajc
* Reworked the <R>eply logic in messages.c - fixes bug #34

Revision 1.484  2000/03/11 19:22:19  nbryant
 * commands.c: improved timing of background keepalives if connection is
               lagged

Revision 1.483  2000/03/11 05:08:48  nbryant
 * commands.c: oops, that mutex stuff wasn't necessary

Revision 1.482  2000/03/11 04:09:03  nbryant
 * new threaded client code for background keepalives

Revision 1.481  2000/03/10 21:40:04  ajc
* Changes to message base and networker to support Internet-style message
  ID's instead of the conventional Citadel style.

Revision 1.480  2000/03/08 03:36:37  ajc
* Shut off hostname resolution when dealing with Unix domain sockets
* Cleaned up the 'citmail' MDA tool
* Added POP3 and SMTP port numbers to global system configuration

Revision 1.479  2000/03/07 21:54:58  ajc
* Fixed the naming conventions and permissions for unix domain sockets.

Revision 1.478  2000/03/05 07:33:23  ajc
* Added support for protocols over Unix domain sockets.

Revision 1.477  2000/03/04 22:36:23  ajc
* Remove nulls appended to editor files during replace, edit, and print
  operations.  Truncate temp files during same operations.
  Closes bugs #6 and #7.

Revision 1.476  2000/03/04 05:29:18  ajc
* Relax restrictions on editing of base rooms.  Renaming is not allowed but
  all other attributes can be edited.  Closes feature request #21.
* Sending pages from the client now uses the same message editing functions
  as entering messages, allowing edit/abort.  Closes feature request #25.

Revision 1.475  2000/03/03 04:50:14  ajc
* Moved all of the wholist masquerading commands into the serv_rwho module

Revision 1.474  2000/03/03 04:12:37  ajc
* Finished the inbound side of gateway domain service

Revision 1.473  2000/02/27 04:55:51  ajc
* Added "keymenu()" generic menu-maker to commands.c
* Blocked non-numeric input to intprompt()  (fixes bug #16)

Revision 1.472  2000/02/27 03:57:35  ajc
* Completed 'fsck'-like reference count verifier (server and client)

Revision 1.471  2000/02/26 18:30:40  ajc
* Properly handle all aliases specified in network/mail.aliases for incoming
  SMTP mail (uses the alias() function, so if we replace that function with
  something that uses the same calling convention, it'll still work)

Revision 1.470  2000/02/26 05:15:38  ajc
* Fortified the message base and SMTP code so that misdirected bounce messages
  end up in the Aide> room instead of getting dereferenced
* Started writing a message reference count verifier ('fsck' for message base)

Revision 1.469  2000/02/25 06:14:05  ajc
* Modularized the RWHO (Read WHO is online) command, basically as a pilot
  for modularizing all "non-API" functionality.

Revision 1.468  2000/02/24 03:44:00  ajc
* Implemented holdoff time (15 minutes) for SMTP send retry.
* Implemented "try for 3 days and then give up" on SMTP send.

Revision 1.467  2000/02/24 00:51:48  ajc
* Client protocol synchronization check during exit from chat.
  This closes Bug #15.

Revision 1.466  2000/02/22 16:37:28  ajc
* Minor tweaks to RFC822 output to keep brain-damanged MS Outlook from dying

Revision 1.465  2000/02/22 04:17:56  ajc
* Got bounce messages working (mostly ... testers, please beat this up!)
* Changed 'FIX' comments to 'FIXME' (less conflict, plus vim highlights it!)

Revision 1.464  2000/02/18 22:29:18  ajc
* Coded up the "bounce" functions.  Still a coupla bugs.

Revision 1.463  2000/02/18 05:10:50  ajc
* Made the <.ASI> command a bit friendlier.
* SMTP sender now pays attention to "smarthost" entries in the system's
  Internet configuration, using them if one or more is present.

Revision 1.462  2000/02/17 05:27:39  ajc
* Got the "MAIL From:" command sending the correct data.  (unnnhhhhnnhhhh...)

Revision 1.461  2000/02/16 22:06:26  ajc
* Altered the display and conversion of RFC822 messages

Revision 1.460  2000/02/16 03:43:28  ajc
* Added the resolver library to the configure script

Revision 1.459  2000/02/16 01:19:39  ajc
* Vanquished the evil dn_expand() beast.  getmx() now seems to be working.

Revision 1.458  2000/02/14 04:36:14  ajc
* sysdep.c: added new event hook type EVT_TIMER.  Timer event hooks are called
  once per minute by any worker thread.
* msgbase.c: removed dependence on nested functions in CtdlOutputMsg() by
  replacing them with an API call CtdlRedirectOutput() in sysdep.c, which
  can temporarily redirect a session's output to an arbitrary file or socket.
* serv_smtp.c: implemented the purging of messages in the queue for which all
  deliveries have been completed.
* serv_smtp.c: removed temporary 'QQQQ' server command and replaced it with
  a timer event hook that runs the queue once per minute (this needs to be
  made more robust)

Revision 1.457  2000/02/08 21:00:47  ajc
* Implemented the deprecated "LAST" command in POP3.  Some clients need it.
* POP3 sessions now set the last-read pointer in Mail>.

Revision 1.456  2000/02/07 05:15:00  ajc
* Renamed CtdlLocalHost() to CtdlHostAlias() and worked it a little deeper into
  the message routing logic.  Still needs some work on the gateway-domain
  stuff.
* Twiddled CtdlOutputMsg() a bit for 'all Internet' situations.  Still needs
  some work to avoid printing dual headers when both Cit and RFC822 exist.

Revision 1.455  2000/02/03 03:57:35  ajc
* Formalized the 'Internet Configuration' logistics.  Added new API call
  CtdlLocalHost() to detect aliases for the local host.  Used in SMTP listener.

Revision 1.454  2000/01/31 02:13:05  ajc
* <.A>ide <S>ystem configuration <I>nternet  in the client (unfinished)

Revision 1.453  2000/01/26 02:41:27  ajc
* SMTP delivery is working but still *very* rough.

Revision 1.452  2000/01/25 04:45:50  ajc
* Wrote enough of the SMTP sender to get Patriot drooling over it, but not
  enough to complete the transmission of mail.

Revision 1.451  2000/01/23 21:25:45  ajc
* Temporary hack to ig_tcp_server() to listen on an arbitrary port if the
  one specified is not bindable (for development only)
* Added SM_DONT_BUMP_REF flag to CtdlSaveMsgPointerInRoom() to be used only
  in very specific and special situations
* Generate delivery instructions when outbound SMTP mail is created from
  within Citadel (as opposed to being from the SMTP module)

Revision 1.450  2000/01/23 05:22:41  ajc
* Coded up some more of the SMTP-sender (still not done)

Revision 1.449  2000/01/22 05:13:56  ajc
* Added some more functionality to the string tokenizer

Revision 1.448  2000/01/17 20:57:43  ajc
* CR to CRLF hacks (lose, lose, lose)

Revision 1.447  2000/01/17 18:30:27  ajc
* Completed POP3 server.  All RFC1939 commands except APOP are implemented.

Revision 1.446  2000/01/17 17:09:23  ajc
* Implemented LIST and STAT commands in the pop3 server

Revision 1.445  2000/01/17 05:38:14  ajc
* citserver.c: cleanup hook functions are now run under the proper context,
               even when initiated by the housekeeper thread
* serv_pop3.c: establish a place to hold the message list

Revision 1.444  2000/01/17 04:26:39  ajc
* Modified CtdlOutputMsg() to handle output to arbitrary sockets or files.
  This uses nested functions and may not be portable beyond GCC...

Revision 1.443  2000/01/15 18:29:15  ajc
* Added a generic (void *) parameter to the ForEachUser() and ForEachRoom()
  callback mechanisms, to allow callers and callbacks to pass arbitrary data
  between each other without requiring TSD variables.
* room_ops.c: eliminated the need for 'FloorBeingSearched' TSD variable
* internet_addressing.c: eliminated 'buffer1' and 'buffer2' TSD variables

Revision 1.442  2000/01/15 04:31:44  ajc
* Removed UI_DIALOG mode in setup.  Can't count on 'dialog' to be consistent.

Revision 1.441  2000/01/15 04:07:17  ajc
* Fixed the access rights on auto-created rooms (the same changes that were
  made to version 5.62 in the stable tree)

Revision 1.440  2000/01/13 03:32:36  ajc
* techdoc/delivery-list.txt: added (syntax for delivery lists)
* domain.*: added (will contain MX lookup code)
* internet_addressing.c, logging.c: fixed some buffer overflow bugs

Revision 1.439  2000/01/12 03:56:27  ajc
* sysdep.c: start the housekeeping thread *after* dropping root perms.

Revision 1.438  2000/01/09 19:03:16  ajc
* Removed the fifo-based protocol downloads and replaced it with a less
  elegant "download temporary file to client, then sx/sb/sz" because
  downloading through a fifo was confusing some software.

Revision 1.437  2000/01/08 22:19:44  ajc
* Completed spool to outbound delivery queue (still no queue sender implemented)

Revision 1.436  2000/01/08 05:00:09  ajc
* Reworked some of the data structures to handle multiple recipients
* Began implementation of the delivery queue
* Added CtdlReallocUserData()
* CtdlSaveMsg() now returns the local message ID in the database

Revision 1.435  2000/01/06 03:50:34  ajc
* Replaced citmail.c with a new one that simply SMTP-forwards to Citadel
* Started outbound SMTP queue work

Revision 1.434  1999/12/30 04:56:29  ajc
* Got initial SMTP delivery working in a very specific situation (delivery
  to a single, local user)

Revision 1.433  1999/12/29 04:44:00  ajc
* client_chat.c: display "No message sent" if a send page is aborted.
  Closes bug #2 in bugzilla.

Revision 1.432  1999/12/26 21:50:07  ajc
* serv_vcard: don't run hooks when not logged in (such as in SMTP sessions)
* serv_pop3: added.  This is the skeleton for a module implementing POP3.

Revision 1.431  1999/12/23 04:46:23  ajc
* "Finished" initial hack of RFC822 import

Revision 1.430  1999/12/22 04:46:34  ajc
* Fixed up the "Date:" headers to be RFC822-compliant

Revision 1.429  1999/12/13 05:30:57  ajc
* Removed our naive 'conv_date()' RFC822-to-unixtime conversion function
  and replaced it with the public domain 'parsedate()' function from UseNet

Revision 1.428  1999/12/10 23:58:25  ajc
* internet_addressing.c: added.  (Internet address to Citadel mapping)

Revision 1.427  1999/12/10 21:34:19  ajc
* serv_smtp: implemented RFC821 "VRFY" and "EXPN" commands

Revision 1.426  1999/12/09 05:01:14  ajc
* Split cmd_user() and cmd_pass() into frontend/backend functions
* serv_smtp: implemented AUTH LOGIN for client authentication

Revision 1.425  1999/12/09 00:22:58  ajc
* Finished the "arbitrary service" registration.
* Eliminated "special" master socket for Citadel protocol - just register it
  like any other protocol.
* Began initial implementation of native SMTP service.

Revision 1.424  1999/12/08 18:09:10  ajc
* Added CtdlRegisterServiceHook() and its data type, for implementing arbitrary
  TCP-based services directly in the Citadel server.  Not finished yet.

Revision 1.423  1999/11/29 17:39:07  nbryant
* citserver.c: Solaris lacks inet_aton; use inet_addr instead

Revision 1.422  1999/11/29 17:26:15  nbryant
* citserver.c: include <sys/types.h>; may help portability to Solaris

Revision 1.421  1999/11/22 00:27:42  ajc
* Added some temporary variables to OpenCmdResult().  Hopefully fixes
  bug #14 when running on FreeBSD.

Revision 1.420  1999/11/21 18:30:16  ajc
* Protected cmd_move() from buffer overrun (no longer crashes the server)
* cmd_chat() -- truncate input at 100 characters to prevent buffer overruns.
  Also handle broken client sockets properly.  (Thanks to DME for bug report)

Revision 1.419  1999/11/19 01:57:40  ajc
* Fixed a *serious* memory leak in the database function wrappers.
* Updated version number to 5.60 -- run setup when installing this version.

Revision 1.418  1999/11/18 03:29:20  ajc
* Changed the order of parameters in <.A>ide <S>ystem config into a more
  logical grouping.

Revision 1.417  1999/11/18 02:31:50  ajc
* Updated some of the documentation
* Brought the internal version number up to 5.60

Revision 1.416  1999/11/17 04:15:05  ajc
* Removed the session_count() function.  Instead, keep a reference count
  updated when sessions begin and end.
* Replaced fixed number of worker threads with lower and upper limits; current
  code now tries to make thread count == session count, within these limits

Revision 1.415  1999/11/15 03:17:39  ajc
* Put lockfile in /tmp instead of in /var/lock.   The latter is not guaranteed
  to exist, nor is it guaranteed to be writable by CTDLUID
  (Resolves bug #11 from the Bugzilla repository)

Revision 1.414  1999/11/15 03:07:24  ajc
* Fixed the network-wide vCard purge logic so that it (1) actually works,
  and (2) forces a netproc run immediately when a purge is entered

Revision 1.413  1999/11/09 21:20:44  nbryant
* configure.in: include <sys/types.h> when doing checks which require <utmp.h>
  (should fix Bug #10 on FreeBSD)

Revision 1.412  1999/11/05 03:53:47  ajc
* Issue 'cancel' messages for vCard when a user is deleted.
* Try to delete 'cancel' messages locally after they've been distributed.

Revision 1.411  1999/11/03 04:01:20  ajc
* Fixed buffer overrun problems in cmd_rchg(), cmd_hchg(), and cmd_uchg()
* Removed my email address as the feedback content from the docs; replaced
  it with a reference to the Citadel web site.

Revision 1.410  1999/11/02 19:51:23  ajc
* Fixed timeout problem for remote client sessions (all timeouts were set to
  1 second ... probably a temporary hack that was missed in the cleanup)

Revision 1.409  1999/11/02 03:03:27  ajc
* Several fixes to msgbase.c and netproc.c to prevent corrupted incoming
  network traffic from crashing the server.  Reject bad messages.

Revision 1.408  1999/11/01 04:21:34  ajc
* Fixed a concurrency bug which crashed the server when multiple sessions
  terminated simultaneously.

Revision 1.407  1999/11/01 00:54:02  ajc
* CtdlFetchMessage() - generate a "<no text>" message body if there's none
  on disk.  Too much stuff goes haywire if there's no M field.

Revision 1.406  1999/10/31 18:17:17  ajc
* Fixed buffer overrun in cmd_rchg()
* Call master_cleanup() when time_to_die==1 for proper shutdown

Revision 1.405  1999/10/31 16:26:55  ajc
* Fixed incorrect assignment of new session ID's

Revision 1.404  1999/10/31 04:17:17  ajc
* Fixed a bug which was crashing the server during very long message entry.

Revision 1.403  1999/10/29 01:48:45  ajc
* database.c: Removed arbitrary limit on maximum number of sessions

Revision 1.402  1999/10/29 01:03:03  ajc
* Debugged all possible ways for a session to terminate; do them cleanly.
* Assign session numbers in a more portable and less arbitrary way.

Revision 1.401  1999/10/28 19:50:55  ajc
* Fixed a problem where the client protocol would spit out two responses
  and therefore get out of sync if ASUP command set the access level to
  0 and therefore deleted the user (thanks to Eric McDonald)

Revision 1.400  1999/10/28 05:08:49  ajc
* Removed all of the thread cancellation cruft that is no longer necessary
* Moved the now non-system-dependent RemoveContext() out of sysdep.c (now
  it's part of cleanup() in citserver.c)
* Removed all references to pthread_* from all modules except sysdep.c

Revision 1.399  1999/10/28 03:20:17  ajc
* Fixed the problem of worker threads waking up prematurely.
* 'QUIT'-terminated sessions now exit properly.

Revision 1.398  1999/10/27 04:26:58  ajc
* Initial hack of worker-thread rearchitecture.  Right now it is successfully
  dispatching worker threads to active client sockets (and to the master
  socket too, of course).  Removing sessions is currently broken.

Revision 1.397  1999/10/26 20:20:29  ajc
* Removed the auto-reconnect stuff... it was locking the client in an active
  loop more often than it was reconnecting.

Revision 1.396  1999/10/26 13:59:11  ajc
damn bugs

Revision 1.395  1999/10/26 03:48:39  ajc
* Shuffled around the order of events when a thread is terminating.  All
  mutex operations now happen prior to the freeing of the CitContext structure,
  otherwise begin_critical_section() and end_critical_section() try to
  manipulate the context's mutex count when there isn't any context.

Revision 1.394  1999/10/26 03:21:16  ajc
* Changed a lot of strncpy() calls to safestrncpy() and replaced most of their
  hardcoded size arguments with 'sizeof' based arguments.

Revision 1.393  1999/10/24 19:22:51  nbryant
        * Makefile.in, configure.in: added --enable-icq flag; made checks for
          authentication libraries more intelligent.

Revision 1.392  1999/10/23 03:39:12  ajc
* Finished moving vCard functionality to the new message base functions.

Revision 1.391  1999/10/21 00:50:14  ajc
* Finished up the flags and replication checks in CtdlSaveMsgPointerInRoom().

Revision 1.390  1999/10/20 16:46:27  ajc
* More code shuffle.  Added some flags to CtdlSaveMessagePointerInRoom() and
  enabled the MOVE command to also do a "copy" operation (actually just
  creates a second link and bumps the ref count).  Implemented "<C>opy" in
  the client.

Revision 1.389  1999/10/20 16:07:48  ajc
* Wholist fixes for users who are in chat mode

Revision 1.388  1999/10/20 03:42:29  ajc
* In the wholist, only show <private room> if the user viewing the list
  doesn't know that room.  Otherwise show the name.

Revision 1.387  1999/10/20 02:59:22  ajc
* Code reorganization.  Making it easier to move/copy messages without
  duplicating existing code.

Revision 1.386  1999/10/17 02:25:18  ajc
* Discovered a huge design flaw in the replication algorithm.  Ripped it
  out and replaced it with something a bit more robust.

Revision 1.385  1999/10/16 05:30:17  ajc
* Changes to message replication code.  Don't do server-side hooks during
  an ENT3 command.  Also fixed a bug in cmd_whok() that caused crashes
  after a file format change.

Revision 1.384  1999/10/14 03:04:16  ajc
* Finished the netproc side of Z (zap/supersede) processing for replication

Revision 1.383  1999/10/13 04:24:18  ajc
* Added search-by-header-fields to CtdlForEachMessage(), and then to the
  server MSGS command.  This will have lots of uses.

Revision 1.382  1999/10/13 01:36:39  ajc
* Starting some work on network zap (supersede) mode for replication

Revision 1.381  1999/10/08 02:55:57  ajc
* More vCard-related debugging

Revision 1.380  1999/10/07 02:58:46  ajc
* Semi-broken vCard replacement implementation in place.
* Added "Z" (Zap, supersede) field to message format

Revision 1.377  1999/10/04 03:19:52  ajc
* We now have a housekeeping thread and a housekeeping queue.

Revision 1.376  1999/10/03 21:48:21  ajc
* Added serv_upgrade.h to automagically convert pre-5.55 format user records
  to 5.55 format user records and generate vCards.

Revision 1.375  1999/09/29 21:13:17  ajc
* CtdlWriteObject() can now store objects in personal rooms for any specified
  user -- rather than only the current user or non-personal rooms.

Revision 1.374  1999/09/29 17:26:56  ajc
* serv_vcard.c: fixed crashola bug in cmd_greg()
* tools.c: simplified and improved the string tokenizer.  Now it runs in a
           single pass with no intermediate buffer.

Revision 1.373  1999/09/28 03:27:37  ajc
* Fully migrated cmd_greg() and cmd_regi() into serv_vcard (still has bugs)

Revision 1.372  1999/09/27 03:33:40  ajc
* cmd_regi() is now in serv_vcard and writes to the vcard instead of to the
  usersupp file.  Still needs tweaking.

Revision 1.371  1999/09/24 03:32:19  ajc
* "read my vCard" and "write my vCard" are written and tested.

Revision 1.370  1999/09/24 02:54:17  ajc
* Worked a little more on the vCard stuff.  The serv_vcard module is now in
  place, and a "read my vcard" function is there; "write my" is next...

Revision 1.369  1999/09/23 03:07:56  ajc
* The vCard 'class' is now linked into the server, though it's not really
  functional yet.  Its constructors/destructors are debugged, though.

Revision 1.368  1999/09/19 21:28:33  ajc
* Finished off the message architecture stuff with a new class of hooks to
  enable future server-side handlers.

Revision 1.367  1999/09/19 15:57:06  ajc
* migrated cmd_ent3() to CtdlSaveMessage()

Revision 1.366  1999/09/19 05:13:57  ajc
* Debugged the new version of CtdlWriteObject()

Revision 1.365  1999/09/16 03:23:23  ajc
* Did most of the migration from save_message() to CtdlSaveMsg().  The
  latter builds a "struct CtdlMessage" (so we can run server-side handlers
  against it later on), then serializes it and stores to disk.
* BROKEN BUILD ALERT!!  cmd_ent3() and CtdlWriteObject() are still not
  migrated.  They are stubbed out and will MALFUNCTION if used right now!!

Revision 1.364  1999/09/07 01:42:42  ajc
* cmd_msg3() now uses serialize_message() for its output.  All message
  commands will eventually exist as a "struct CtdlMessage" at some point
  so that we can install server-side handler hooks.

Revision 1.363  1999/09/07 00:04:13  ajc
* netproc.c: put outgoing messages into the use table, too -- this prevents
  locally originated messages from showing up again if a remote system is
  misconfigured and spools them back to us.

Revision 1.362  1999/09/06 03:39:15  ajc
* citadel.c: run strproc() on new passwords

Revision 1.361  1999/09/03 17:50:26  playcow
For URL view, don't prompt user to select url if there is only one.  Display
<U>RL View prompt if message contains url(s). -Ben

Revision 1.360  1999/09/02 02:09:59  ajc
* msgbase.c: new function serialize_message() for future use

Revision 1.359  1999/09/01 21:09:25  ajc
* database.c: display the GDBM version string on startup

Revision 1.358  1999/09/01 02:36:34  ajc
* Actually _enforce_ the max msg len limit

Revision 1.357  1999/09/01 01:51:48  ajc
* Added the ability to handle embedded URL's from the text client

Revision 1.356  1999/09/01 01:02:47  ajc
* Implemented "maximum message length" in global system config

Revision 1.355  1999/08/31 00:57:17  ajc
* Handle multipart/alternative properly during legacy message outputs.
  Basically it just prints the first alternative and skips the rest.

Revision 1.354  1999/08/29 21:12:24  ajc
* Made some changes to the output of MIME (especially multipart) messages.

Revision 1.353  1999/08/29 19:56:43  ajc
* HTML updates

Revision 1.352  1999/08/24 02:01:03  ajc
* html.c: added.  This is an overly simplistic HTML-to-text converter.

Revision 1.351  1999/08/21 18:37:29  ajc
* Minor cosmetic cleanup.  No code changes.

Revision 1.350  1999/08/21 05:15:34  ajc
* mailinglist.c, netmailer.c: fixed to allow list submissions from all posters
  on a Citadel network rather than only on the local system.

Revision 1.349  1999/08/08 00:25:45  ajc
* Made one more byte available in locate_host() and in all the structs which
  its output gets written to (client can display 24 positions but we were only
  saving 23).

Revision 1.348  1999/08/07 16:34:38  nbryant
* serv_icq.c: warning fixes

Revision 1.347  1999/08/06 02:57:26  ajc
* locate_host.c: use strdoop() and phree() instead of strdup() and free()
* serv_icq.c: run learned IP's through Citadel's locate_host() & put in Wholist

Revision 1.346  1999/08/05 17:58:59  ajc
* RWHO command now returns express message code in postion 3

Revision 1.345  1999/08/04 02:21:45  ajc
* Fixed some bugs in the ICQ metaclient, and documented the new protocol cmds

Revision 1.344  1999/08/03 11:34:35  ajc
* Added client_icq.c and client_icq.h

Revision 1.343  1999/08/03 03:14:51  ajc
* Wrote the client side of the ICQ gateway.  Now on to other projects.  :)

Revision 1.342  1999/08/03 01:52:06  ajc
* Redesigned the client protocol commands for dealing with ICQ
* Implemented page function priority ordering to prevent pages from being
  simultaneously delivered over multiple IM systems
* Migrated serv_icq.* into the Makefile

Revision 1.341  1999/08/01 21:36:30  ajc
* EXTREME coolness.  The server side of the ICQ metaclient is now working.
  It is set up using ICQL and ICQA commands, it automatically logs the user
  onto ICQ along with Citadel, and displays all non-offline ICQ contacts
  in the Wholist.

Revision 1.340  1999/07/31 07:18:01  ajc
* Restructured the express message infrastructure, adding a class of function
  hooks for the addition of multiple paging modules with message routing

Revision 1.339  1999/07/30 22:20:19  ajc
* Applied bugfix patches contributed by Vaggelis Tsirkas:
  * rooms.c: buffer overrun fix
  * room_ops.c: cmd_rdir() now behaves better when directory doesn't exist

Revision 1.338  1999/07/30 03:32:24  ajc
* Added strdoop(), a leak-checked version of strdup()
* Small fixes to new API functions in msgbase.c
* ICQ metaclient stores/reads config using the message base API functions

Revision 1.337  1999/07/29 03:36:37  ajc
* msgbase.c: reorganized.  output_message() now uses CtdlFetchMessage(),
  cmd_msg3() now fetches directly from disk and spews to the client.

Revision 1.336  1999/07/28 04:02:37  ajc
* Server modules are now labelled with their RCS ID instead of a complex and
  manually-updated data structure.

Revision 1.335  1999/07/28 03:50:24  ajc
* serv_expire.c: expire-by-age now calls CtdlFetchMessage() instead of
  calling output_message() in MT_DATE mode.
* msgbase.c: removed MT_DATE mode ('twas a sleazy hack)

Revision 1.334  1999/07/27 22:47:26  ajc
* Implemented new data type "CtdlMessage" which will eventually be used as
  widely as possible to represent a message in memory.
* Implemented CtdlFetchMessage() which is intended to become the back-end to
  output_message() as well as a bunch of other things.

Revision 1.333  1999/07/27 20:00:24  ajc
Removed all references to CC->msglist and CC->num_msgs, and all utility
functions which relied upon them.  Citadel Is Now Better.

Revision 1.332  1999/07/27 19:32:22  ajc
Removed serv_upgrade.c and all references to it in Makefile.in
Reworked new-mail-count to not use MessageFromList() etc.

Revision 1.331  1999/07/25 02:59:37  ajc
Fixed reference count problem in cmd_move()

Revision 1.330  1999/07/24 22:50:38  ajc
Continued replacing references to [get|put]_msglist() with better code.
For some reason, cmd_msgs() still doesn't always work right.

Revision 1.329  1999/07/24 22:16:41  ajc
Experimenting with automatic updating of ChangeLog by CVS.  Simply twiddle
ChangeLog a bit (i.e. by adding or removing a newline from the end of the
file) before issusing "cvs commit", and the comments recorded by CVS will
automatically appear at the beginning of ChangeLog.

Revision 1.328  1999/07/24 22:14:21  ajc
cmd_move() now uses CtdlDeleteMessages().
WARNING: build is temporarily broken.  Currently removing all references
to the "msglist" kept in CitContext.  It's ugly and must die.

Thu Jul 22 22:26:50 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Moved message deletion into new API function CtdlDeleteMessages()
        * Added CtdlWriteObject() to store generic data in the msgbase
        * Fixed really dumb error that prevented network msgs from posting

Tue Jul 20 22:14:54 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Moved the actual work done in cmd_msgs() into a new API function
          called CtdlForEachMessage() which is supplied a callback function.

Mon Jul 19 23:24:18 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Keep the (unqualified) content-type in the SuppMsgInfo record.  We'll
          be using this shortly to search rooms for specific object types.

Sun Jul 18 14:53:16 EDT 1999 Art Cancro <ajc@uncensored.citadel.org> 
        * Changes to dynloader et al to handle ICQ module being written
        * serv_icq.c, serv_icq.mk: added (separate makefile is temporary)

1999-07-17 Nathan Bryant <bryant@cs.usm.maine.edu>
        * chkpwd.c: DELETED CVS REVISION 1.3 (backed out Art's last change)
          use 'cvs update -r 1.2 chkpwd.c; cvs update -A chkpwd.c' NOW to
          avoid problems with working directories.
        * Makefile.in: don't install chkpwd setuid if make install isn't
          being run as root.
        * citadel.spec: chmod u+s chkpwd during %install stage

Fri Jul 16 18:39:04 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * PEXP and GEXP no longer trip the idle time display
        * Fixed bug which duplicated incoming private mail to Trashcan
        * Improved auto-reconnect by NOT using SIGPIPE and longjmp(); also
          implemented a 15 second delay to wait for a crashed server to restart
        * Relaxed the security check in chkpwd.c a bit; it was just flat-out
          preventing logins on my system otherwise

Thu Jul 15 22:57:32 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * eliminate redundant "name" parameter in [l]putuser(), now uses
          usbuf->fullname to guarantee the correct name for the index

1999-07-12 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in, configure.in: link netproc with gdbm
        * netproc.c: only include gdbm.h if HAVE_GDBM_H
        * user_ops.c: warning fix

Mon Jul 12 19:51:30 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Mail is now saved in both sender and recipient mailboxes.  This is
          structured in a way that will allow a separate "outbox" room and/or
          multiple recipients in the future.

Sun Jul 11 18:46:48 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * netproc.c: msgfind() no longer uses the timestamp as a message-ID
          when no other message-ID is available (it screws up the loopzapper)
        * room_ops.c: eliminate room name parameter in putroom() and its ilk;
          get data from quickroom.QRname instead; prevents incorrect indexes
        * Tentative implementation of "personal rooms" (user-private namespace)
        * Added supplementary message info records for info that may change
          at some time later than when the message is saved (i.e. ref counts)
        * Implemented msg reference count increment/decrement; delete messages
          whose reference count reaches zero

Wed Jul  7 23:25:09 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * control.c: create citadel.control if it doesn't exist (yikes!)
        * serv_expire.c: purge mailbox rooms belonging to non-existent users
        * user_ops.c: don't delete user's mailbox at user-delete time

Mon Jul  5 17:01:29 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * utilsmenu: removed menu items for defunct utilities

Mon Jun 28 16:24:10 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Changed any remaining references to UUCP, to "Internet" instead.

Thu Jun 24 11:13:23 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * added server command line option "-f" to defrag databases on startup
        * control.c: better performance and reliability in [get|put]_control()
        * netproc.c: Finished the loopzapper

Mon Jun 21 00:04:15 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * netproc.c: started writing a vortex checker.  Not finished.

Wed Jun  9 23:34:25 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Replaced all code that generated temporary filenames with calls to
          tmpnam().  Rewrote using tmpfile() where possible.

Thu Jun  3 11:35:18 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * base64.c: mapped fi and fo to stdin and stdout using
          actual code rather than assignment at declaration time
          (several users of Red Hat Linux 6.0 reported problems)

Fri May 21 20:05:00 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Added function CtdlGetDynamicSymbol() for dynamic symbol allocation
        * server.h: Changed discrete #define's to enum's where appropriate
        * sysdep.c: Changed the startup message to give credit to the whole
          development team :)  Also made the message more GNU-ish.

Thu May 20 20:01:30 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * database.c: print log messages for file defragmentations
        * citserver.c: implemented CtdlAllocUserData() and CtdlGetUserData()
          for arbitrary per-session data storage (by modules etc.) without
          having to add fields to struct CitContext
        * msgbase.c: removed "desired_section" from struct CitContext and
          implemented it using CtdlGetUserData() as a test.

Wed May 19 19:30:28 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * commands.c, commands.h, routines.c: began color scheme changes

1999-05-15 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: Added untested support for BSDI 4.x.

1999-05-13 Nathan Bryant <bryant@cs.usm.maine.edu>
        * acconfig.h, configure.in, routines.c: fix for certain SYSV variants
          which lack utmp.ut_host
        * citadel.h, file_ops.c, msgbase.c, netproc.c, serv_chat.c: fix
          namespace collision with <sys/stream.h> on aforementioned SYSV
          variant
        * configure.in, getutline.c: check for paths.h
        * configure.in, Makefile.in: check for -lsocket and -lnsl

1999-05-11 Nathan Bryant <bryant@cs.usm.maine.edu>
        * chkpwd.c: fixed excessive paranoia; it used to refuse to run when
          invoked by root and CTDLUID != 0
        * Makefile.in: ignore errors while installing /etc/pam.d/citadel

1999-04-27 Art Cancro <ajc@uncensored.citadel.org>
        * file_ops.c: fixed NDOP to not crash the server if it has trouble
          with a download file
        * netpoll.c: upped default packet size

Mon Apr 26 22:06:57 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Repaired IGnorant security hole blunder re. citmail

Sun Apr 25 12:44:08 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * serv_chat.c: notify user of number of participants upon entering chat
        * Built the 5.53 distribution

Wed Apr 21 22:23:13 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * aidepost.c: add -r flag to allow posting to rooms other than Aide>
        * serv_expire.c: now posts transcripts of all auto-purged rooms/users

Tue Apr 20 12:45:55 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c: downloading more than MAX_MSGS messages now truncates
          (off the beginning of the list) rather than crashing.

Mon Apr 19 12:11:48 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * whobbs.c: auto-detect when being called from a webserver, and act
          as a CGI (print HTTP headers and HTML output)

1999-04-18 Nathan Bryant <bryant@cs.usm.maine.edu>
        * serv_chat.c: fixed some potential buffer overruns (thanks dme)

Wed Apr 14 21:32:28 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Makefile.in: chmod 4755 citmail to prevent citmail from aborting
          when called from sendmail due to citadel.config security check

1999-04-13 Nathan Bryant <bryant@cs.usm.maine.edu>
        * dynloader.c: OpenBSD places underscores in front of symbol names
        * Makefile.in: fixed a few sillies
        * aidepost.c, citmail.c, file_ops.c, logging.c, msgbase.c, netmailer.c,
          netproc.c, rcit.c, routines.c, serv_upgrade.c: fixed time_t handling
          (have to cast it to long for printf/scanf)

Mon Apr 12 22:13:26 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * aidepost.c: rewrote to unlink temp file before writing to it so that
          it will automatically go away if interrupted. Also ran indent -kr -i8

1999-04-12 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in, Makefile.in: taught it how to generate OpenBSD shared
          libraries
        * config.c, sysdep.c: fix -h option not setting proper modules dir
        * configure.in, citadel.spec: make --enable-chkpwd the default
        * setup.c: don't chown chkpwd

1999-04-11 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: another OpenBSD fix, but we're still not quite there yet
        * serv_chat.c: warning fix

Thu Apr  8 22:51:28 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * config.c: now requires a setup run for *any* rev level difference
        * Updated docs & confs for 5.53b1 release
        * setup.c: sets the 0600 permission bits on citadel.config that
          is checked for in config.c

1999-04-08 Nathan Bryant <bryant@cs.usm.maine.edu>
        * citserver.c: improved is_public_client(), also if a public_client
          only supplies a numeric address, attempt to resolve it
        * locate_host.c: verify that the forward DNS matches the reverse
        * locate_host.c, locate_host.h: more general interface
        * configure.in, acconfig.h: fixes for Digital UNIX

Wed Apr  7 21:36:16 EDT 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented "access level required to create rooms" (client & server)

1999-04-07 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: updated help messages, OpenBSD support
        * setup.c: if /etc/inittab doesn't exist, don't ask to create an
          entry in it
        * server.h, sysdep.c: fix a potential deadlock/data corruption bug
        * room_ops.c: fixed the 'gdbm error: Illegal data' message when
          deleting a room which had never been posted to
        * user_ops.c: include errno.h
        * dynloader.c: fix for OpenBSD

1999-04-06 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in, configure.in, getutline.c, sysdep.c:
          fixes/bug workarounds for FreeBSD

1999-04-03 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in, configure.in, chkpwd.c, acconfig.h: support for
          `chkpwd', a setuid helper program for machines which use shadow
          passwords (configure --enable-chkpwd)
        * Makefile.in, configure.in, auth.c, citadel.pam, user_ops.c: support
          for PAM or shadow passwords (configure --with-pam)
        * Makefile.in: made some messages simpler
        * citadel.spec: updated for 5.53; correct name of tarball; build with
          --enable-chkpwd and --with-pam; add defattr tags so rpm's can be
          built by non-root user
        * commands.c: cosmetic cleanup
        * config.c: (security/paranoia) check permissions on citadel.config
        * configure.in: check for ncurses if we can't find curses
        * dynloader.c: warning fix
        * sysdep.c: don't complain if initgroups() fails
        * citadel.c: fix systems with SYSV-style signal handling (e.g. libc5)
        * Makefile.in: New! Improved! Cleaner! Shinier!

Sun Mar 21 14:21:47 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c: cosmetic cleanups to message reading loop

Sat Mar 13 21:33:19 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * commands.c: use bright colors by default in color mode
        * citserver.c: initialize wholist fields with (not logged in) etc.

1999-03-08 Nathan Bryant <bryant@cs.usm.maine.edu>
        * sysdep.c: call DLoader_Init() with an absolute path so that gdb can
          find module symbols
        * database.c: bail out if opening databases fails

Sat Mar  6 01:55:55 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * serv_chat.c: use memfmout(), *not* cprintf() to transmit express
          messages.  Calling cprintf() on strings >256 bytes crashes the server
        * msgbase.c: minor logging fix in save_message()

1999-03-05 Nathan Bryant <bryant@cs.usm.maine.edu>
        * sysdep.c: add undocumented -r flag to citserver to prevent it from
          dropping root permissions.
        * sysdep.c: also drop supplementary groups

1999-03-04 Nathan Bryant <bryant@cs.usm.maine.edu>
        * config.c: error checking in put_config()
        * setup.c: chgrp files to the login group associated with CTDLUID
        * sysdep.c: copyright 1987-1999; drop root perms; load modules and call
          master_startup() after dropping perms

Wed Mar  3 00:00:55 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Prevent buffer overruns in lowercase_name in [get|put]user()
        * client_chat.c: use citedit() for page composition

Sat Feb 27 07:47:36 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * rooms.c: download_to_local_disk() prompts for a filename if a blank
          filename was supplied to it (for attachments without names)
        * mime_parser.c: strip leading whitespace in content_type & disposition

1999-02-24 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: improved check for pthreads
        * configure.in, routines.c, acconfig.h: check for ut_type in struct utmp
        * configure.in, Makefile.in: support for building server modules as
          relocatable objects for BSDI (which still uses a.out *gag* *choke*)
        * configure.in: compiler choice & flags for BSDI; check for libtermcap
        * database.c: don't use a critical section in open_databases()
        * housekeeping.c: use getfloor()/putfloor() instead of
          lgetfloor()/lputfloor() in check_ref_counts()
        * mime_parser.c: include <errno.h>
        * msgbase.c: include <limits.h>
        * sysdep.c: hacks for BSDI. use signals to fake thread cancellation;
          don't call master_cleanup() directly from signal handler.
        * routines.c: prototype getutline() if necessary
        * getutline.c: stupid bugfix
        * acconfig.h, configure.in, locate_host.c, server.h: work around
          nonreentrant gethostbyaddr() on BSDI

Mon Feb 15 22:59:00 EST 1999 Vaggelis Tsirkas
        * citadel.c: increased hostname buffer size to handle very big names

1999-02-15 Nathan Bryant <bryant@cs.usm.maine.edu>
        * sysdep.c(main): initialize alen before call to accept()

1999-02-04 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: improved check for cygwin
        * configure.in, routines.c: access utmp directly instead of calling
          `who' if getutline() is available.
        * configure.in, Makefile.in, getutline.c: replace getutline() on
          systems which don't have it
        * routines.c: now always access utmp directly unless we can't find
          utmp.h

1999-02-02 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Fixes for Cygwin:
          - ifdef out file download methods that require named pipes (client)
          - include pthread.h and gdbm.h only if they are present (fixes
            auto dependency generation)
          - include snprintf.h where needed
          - handle .exe suffixes for "make install"

Tue Feb  2 22:15:08 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Overhauled the express messaging system (again)

Mon Feb  1 19:48:04 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c: implemented client download of MIME attachments

Sun Jan 31 18:29:18 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Added qpdecode.c to the distribution (decodes quoted-printable)
        * Finished the MIME parser
        * Gave MSG0 a reasonable behaviour for MIME messages
        * Added the OPNA command for downloading attachments

Sat Jan 30 18:39:53 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Look for citadel.rc in current directory if not found elsewhere
        * More work on the MIME parser
        * Added base64.c to the distribution

1999-01-29 Nathan Bryant <bryant@cs.usm.maine.edu>
        * fixes for IRIX (thanks to wr and family for use of the Indy):
          - use memset()/memcpy() instead of bzero()/bcopy() in all cases
          - configure updates
          - handle `long' pid's
          - a few other little bits

Mon Jan 25 21:23:07 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed bug in save_message() which crashed the server on mail to sysop
        * Rewrote pop_march() to be smarter about <G>oto heuristics

Sat Jan 23 14:32:19 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Changed internal storage of express messages from a linked list to
          a single, resizable buffer.
        * Added a "room order" key to the room record, to allow some control
          over room listing order.
        * Made the room list commands aware of the room order key.
        * Overhauled <G>oto heuristics to pay attention to floor & room order

Wed Jan 20 19:21:51 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Added some more code to the unfinished MIME parser
        * Changed module loading path to simply "modules" because after calling
          get_config(), the cwd is guaranteed to be the correct Citadel directory.

Tue Jan 19 21:28:29 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed a bug in the user editing command (client side)
        * Started a rewrite of the MIME parser

Thu Jan 14 21:21:15 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Brought over the mime_parser from WebCit and began preliminary work
          on supporting MIME format messages.

Tue Jan 12 22:30:00 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Various changes to begin work on support for MIME messages
                - Defined format type 4 for MIME
                - msgbase.c: *temporary* hacks in output_message() for Type 4
                - citmail.c: added more robust header parsing, and support
                             for Type 4.  Also eliminated the crappy built-in
                             SMTP server.
                - Updated some of the technical documentation

Sun Jan 10 13:34:36 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed access to page log room

Fri Jan  8 12:35:09 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * control.c: include <limits.h> to fix PATH_MAX undefined
        * serv_chat.c: made the following changes to cmd_sexp() --
                * Send zero-length message to check only, don't send
                * Send "-" message on the command line to invoke
                  the SEND_LISTING transfer mode for a multi-line message
        * Added facilities to log all pages to a room (site configurable)

Tue Jan  5 23:24:52 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Replaced all occurances of malloc(), realloc(), and free() in the
          server and server-modules with mallok(), reallok(), and phree().
          Wrote macros and a set of leak-tracking functions.

Sun Jan  3 20:38:45 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Documentation changes

Fri Jan  1 01:01:45 EST 1999 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed security problem relating to private rooms

Wed Dec 30 20:10:52 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * sysdep.c: put a bigger string buffer into lprintf() to avoid overruns

Sat Dec 26 16:56:46 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * internetmail.config: commented this file more clearly

Wed Dec 23 20:42:49 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * citadel.c: added some experimental code to automatically reconnect
          to the server if the connection is broken.  For some reason, it only
          works once.

Wed Dec 23 18:47:12 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * sysdep.c: ignore SIGPIPE.  This keeps broken connections from
          crashing the whole server.
        * Tagged everything for the official 5.50 release.

Mon Dec 21 07:54:20 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * housekeeping.c: call kill_session() with session_to_kill,
          not ccptr->cs_pid (was crashing the server)

Sat Dec 19 13:57:48 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Added "-i" flag to netproc to make it skip the export phase.
          Updated other programs to call netproc in this way when appropriate.
        * Updated network.txt to reflect the usage for netproc (which has
          been wrong for several releases)

Thu Dec 17 00:17:04 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Started removing the word "beta" from the docs and code.
          Preparing for an actual release.
        * msgbase.c: generate an 'I' field when requested (i.e. on locally
          originating messages.  this was breaking parts of the network)

1998-12-15 Art Cancro <ajc@uncensored.citadel.org>
        * msgbase.c: remove extra call to alias() which was causing
          Citadel-to-Citadel mail to fall into the bit bucket.
        * msgbase.c: fixed tempfile naming problem that could cause a new
          outgoing netmail message to overwrite another if netproc had not been
          run in between

1998-12-14 Art Cancro <ajc@uncensored.citadel.org>
        * More session table stability nonsense

Sun Dec 13 17:40:08 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * sysdep.c, citserver.c: (hopefully) fixed a session table concurrency
          bug which was causing the server to occasionally crash.
        * removed serv_test.so from the default build

Fri Dec 11 18:50:00 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * setup.c: default value for maxsessions is now 0 (no limit)
        * room_ops.c: don't allow users to create a room called "Mail"
        * serv_expire.c: fixed "number of messages purged" display
        * commands.c: when using color, default to low-intensity colors

Tue Dec  8 07:58:16 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Beta 2

Sat Dec  5 01:24:03 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Added a new type of module hook for adding logging functions
        * Removed whitespace to left and right of '@' in netmail recipients
        * sysdep.c: read citadel.config _before_ initializing loadable modules
        * stats.c: fixed segfault resulting from extracting log lines
        * Ripped most of the "attachments" stuff out of both the client and
          server.  (Thought of a better way to handle it ... LATER.)

1998-12-03 Nathan Bryant <bryant@cs.usm.maine.edu>
        * setup.c: create citadel.config with mode 0600
        * Makefile.in: don't chmod sendcommand
        * serv_upgrade.c: don't create citadel.config if it doesn't already
          exist (paranoia)
        * setup.c: saner defaults for nodename and fqdn

Wed Dec  2 20:37:05 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * msgbase.c: modified AddMessageToRoom() and all functions that call it
          to use a more reliable/accurate method to set quickroom.QRhighest
        * weekly.in: don't sort/purge filedir where filedir doesn't exist

1998-12-02 Nathan Bryant <bryant@cs.usm.maine.edu>
        * weekly is now generated by configure

Mon Nov 30 19:48:52 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * room_ops.c: added sort_msglist() to move and save operations
        * sendcommand.c: added (also updated utils.txt, weekly, Makefile.in)
        * BETA 1

Sun Nov 29 23:57:39 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed cmd_regi() to not display a second result code after xfer
        * Makefile.in: Removed "chmod 4755 citmail netmailer"

1998-11-23 Nathan Bryant <bryant@cs.usm.maine.edu>
        * citadel.spec: added

1998-11-22 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: split install target into install-exec, install-data,
          and install-doc subtargets

Sat Nov 21 16:53:30 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Added a fourth color mode in the client: "user" which turns color
          on or off according to a per-user flag stored on the server.  Added
          server-side support for this too, of course.
        * import.c: removed
        * serv_expire.c: finished the code to purge stale visits
        * sysdep.c: strip trailing nonprintables in client_gets()
        * routines2.c: fixed <.AS> command; all configs now work properly
        * Moved num_parms() and all the extract() type functions into tools.c
          and removed them from all other files.  Linked in tools.[o|ro] there.
        * netproc.c: handled incoming file transfers to correct room directory
        * room_ops.c: fixed incorrect naming of files in info and images dirs

Fri Nov 20 20:29:07 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * setup.c: removed all prompts that can be configured from within the
          client in order to simplify the setup procedure

Thu Nov 19 23:28:33 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed serv_upgrade.c and export5.c (found elsewhere) to use a new
          export format which treats visits as a separate section
        * serv_expire.c: began writing functions to purge rooms and visits
          (neither work yet), and added a way to call defrag_databases()

Wed Nov 18 23:51:17 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Reimplemented 'visit' structs stored globally instead of as
          lists-per-user

Tue Nov 17 22:37:48 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Started implementing global room numbers.

Sun Nov 15 20:32:34 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * room_ops.c: mailbox rooms always appear on the main floor
        * made QR_MAILBOX rooms non-editable
        * cmd_setr() delete old room record when room name changes
          (This causes a big mess that exposes a flaw in the whole design.)
        * room_ops.c: users can delete messages from their mailboxes

Thu Nov 12 23:59:13 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Rewrote cmd_rchg() and also increased the size of the "fakename"
          buffer.  Overruns are probably what was causing the crashes.
        * Changed the way cmd_ent3() handles mail messages; the previous code
          rerouted all private mail to the trash.

Wed Nov 11 17:57:39 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * citserver.c: slight changes to cmd_rchg() and cmd_hchg() [crashes]
        * citserver.c, msgbase.c, user_ops.c: hide the owner-prefix of mail
          rooms in a couple more places: set_wtmpsupp() and make_message()
        * sysdep.c: added an fflush() to lprintf() for "tail -f"-able logs
        * serv_expire.c: purge ops are now a command instead of a cleanup
          function.  This is probably temporary as well.
        * citadel.c: fixed the way <.WL> parses the returned data from a
          TIME command.

1998-11-11 Nathan Bryant <bryant@cs.usm.maine.edu>
        * serv_upgrade.c: fix uninitialized variable

Wed Nov 11 00:47:32 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * msgbase.c: fixed a bug that was misrouting incoming network msgs
        * server.h, database.c: wrapped all GDBM calls in critical sections
          to avoid making those calls re-entrantly (gdbm fatal: lseek error)

1998-11-10 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: link in snprintf.o where needed

1998-11-09 Nathan Bryant <bryant@cs.usm.maine.edu>
        * client_chat.c: eliminate calls to sprintf()
        * commands.h, routines.c, routines2.c: warning fix
        * commands.c, control.c, cux2ascii.c, file_ops.c, import.c,
          ipc_c_tcp.c: eliminate sprintf() calls

Mon Nov  9 19:15:31 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * serv_upgrade.c: added all missing fields to export/import
        * serv_expire.c: support per-user purge time when purging users
        * user_ops.c: added per-user purge time to AGUP and ASUP commands
        * routines.c: added more stuff to <.A>ide <E>dit user

Sun Nov  8 22:56:53 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * serv_expire.c: created; moved message expiry from serv_test.c,
          moved user purge from userpurge.c
        * userpurge.c: deleted
        * routines2.c: finished <.AS> command
        * room_ops.c: fixed Aide room access (for some reason, the Aide
          room had the QR_MAILBOX flag set)

1998-11-08 Nathan Bryant <bryant@cs.usm.maine.edu>
        * useradmin.c: really removed (cvs remove)
        * aidepost.c, citadel.c: convert all sprintf() calls to snprintf()
        * sysdep.c: fix overrun in lprintf() described by dme/Dead Monkey
        * citmail.c, citserver.c: convert all sprintf() call to snprintf()

Sun Nov  8 13:19:36 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * useradmin.c: removed
        * utils.doc: removed references to useradmin and sysoputil

Fri Nov  6 20:22:20 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * citadel.h (and related files): removed defunct parameters,
          c_defent and c_msgbase (erase your test system)
        * Implemented CONF server command for site-global configs
        * Shuffled yesno() and yesno_d() from routines.c to commands.c
        * commands.c: implemented boolprompt()
        * routines2.c: started adding CONF questions to <.AS> command
        * room_ops.c: began a fix for the mysterious disappearing Aide room

1998-11-05 Nathan Bryant <bryant@cs.usm.maine.edu>
        * snprintf.c: warning fix propagated over from gcit

1998-11-04 Nathan Bryant <bryant@cs.usm.maine.edu>
        * added RCS Id keyword strings to sources
        * citmail.c: reverted to version 1.10

Wed Nov  4 10:53:13 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c: temporarily set screenwidth to a fixed value of 80
          during <P>rint operations.

Mon Nov  2 12:59:03 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * setup.c: looks for 'citadel', 'bbs', or 'guest' in /etc/passwd to
          try to get a default for ctdluid if it's currently set to 0
        * citmail.c: changed usersupp.eternal to usersupp.usernum (why didn't
          it complain about this before?)
        * serv_upgrade.c: began writing an "export" command to do sidegrades

Sun Nov  1 18:47:42 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * serv_upgrade.c: cosmetic changes
        * Implemented message expiry by date (this really needs to be moved
          out of serv_test.c, but where does it belong?)

1998-11-01 Nathan Bryant <bryant@cs.usm.maine.edu>
        * serv_upgrade.c: warning fixes
        * acconfig.h: remove ANSI_COLOR
        * Configure, Makefile.tmpl: removed

Sat Oct 31 20:48:44 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * More stoopid ANSI colour additions here and there.
        * ANSI colour selection is now in citadel.rc instead of an option in
          the configure script.  "on" "off" and "auto" are available.
        * added "build.txt" to the techdoc directory, with LS's build notes

1998-10-31 Nathan Bryant <bryant@cs.usm.maine.edu>
        * citadel.c, commands.c, commands.h: set background color to black
          before clearing screen, so that we can actually see text on
          black-on-white xterms.

1998-10-29 Nathan Bryant <bryant@cs.usm.maine.edu>
        * rooms.c: fix color of Mail>
        * citadel.c: send ANSI detect sequence after attach_to_server() so the
          terminal doesn't send the answerback sequence to the shell if we
          can't connect.

Wed Oct 28 20:20:14 EST 1998 Art Cancro <ajc@uncensored.citadel.org
        * citadel.c: Added a splash of colour to the Wholist

1998-10-28 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: added comments
        * mkinstalldirs: new file to be used for `make install'
        * aclocal.m4, missing: new files, also swiped from automake
        * citadel.c: don't mung the terminal if we can't find citadel.rc
        * Makefile.in: added `install' target, `clean' removes
          *.mo, rebuild configure when configure.in changes.
        * configure.in: check for install and autoconf
        * configure.in, Makefile.in: only pass -fPIC to gcc
        * policy.c: warning fix for OSF/1 (use memset() instead of bzero())

Tue Oct 27 22:25:42 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Unified the "unpacked database" format for both upgrades and
          sidegrades (also see export5.c elsewhere)
        * citserver.c: clear out cmdbuf before reading a command; some server
          commands were accidentally extracting parameters from previous cmds
        * rooms.c: removed the warning about the ineffectiveness of kicking
          users out of public rooms, because the new server can do lockouts

1998-10-27 Nathan Bryant <bryant@cs.usm.maine.edu>
        * sysdep.c (client_gets), ipc_c_tcp.c (serv_gets): improved handling
          of long lines. 
        * Makefile.in: partial support for VPATH builds, autodependency fix

Tue Oct 27 00:08:16 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * minor documentation changes

Sun Oct 25 14:57:40 EST 1998 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c: fixed an oversight that was allowing attachments even
          when disabled in citadel.rc

1998-10-25 Nathan Bryant <bryant@cs.usm.maine.edu>
        * routines2.c: warning fix

Sat Oct 24 22:07:56 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Client: added message expiration policy questions to room edit
        * Client: added <.A>ide <S>ystem configuration command

1998-10-24 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: Auto dependency generation (may require GNU make, but
          does at least function without this feature on OSF/1 make);
          Makefile is regenerated when Makefile.in changes (ditto);
          `realclean' is now known as `distclean';
          portability fixes for older Unix make utilities
        * citadel.c, citadel.h, commands.c: make client suspendable

Fri Oct 23 19:34:38 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * setup.c: default node name is now obtained from uname()
        * config.c: added put_config()
        * policy.c: added, moved GetExpirePolicy() from room_ops.c
        * policy.c: implemented cmd_gpex() and cmd_spex()

Wed Oct 21 22:24:48 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Mail rooms now hide their owner-prefix from the client.
        * proxy.c: added configurability and primitive message expiry

1998-10-20 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.tmpl: fix to build client with old Configure script
        * configure.in: autologin defaults to enabled if crypt() is available
        * room_ops.c: fix improper null-termination bug I introduced

Mon Oct 19 20:52:55 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Client ability to force display of prompts in Mail rooms, even when
          the user has prompting turned off (citadel.rc option)

1998-10-16 Nathan Bryant <bryant@cs.usm.maine.edu>
        * sysdep.c (cprintf): generate a newline on truncated buffer
        * room_ops.c: exploitable overrun fixes

Thu Oct 15 19:27:32 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * msgbase.c: reimplemented cmd_move()
          room_ops.c: wrote AddMessageToRoom() which is used for both entering
          and moving messages.
        * setup.c: system-default message expire policy of "number of
          messages, 150" is now a default configuration instead of a temp hack
        * proxy.c: cache dir create now dies on any error except EEXIST

Wed Oct 14 22:41:16 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Misc code cleanup

1998-10-13 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: don't check for -lcrypt unless autologin is enabled
        * file_ops.c: fix another overrun

Mon Oct 12 15:27:21 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Killed the "rooms" subdirectory (it isn't used anymore)
        * dynloader.c: Made dynamically added server commands case-insensitive
        * import.c is now serv_upgrade.c, a module
        * Removed most of the "level 9" trace messages no longer needed

1998-10-12 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: simplified to use pattern rules; files compiled with
          -D_REENTRANT go to *.ro to allow the same files to be used with both
          client and server
        * tools.c, tools.h: new files; misc routines used by both client and
          server go here. contains safestrncpy() at the moment.
        * rooms.c: fix several exploitable buffer overruns
        * sysdep.c: fix infinite loop when long lines are received from the
          client; fix exploitable buffer overrun in cprintf()
        * ipc_c_tcp.c: fix infinite loop on long line from server
        * serv_upgrade.sh: remove uncnsrd-dependent absolute path
        * .cvsignore: add *.ro

Sun Oct 11 23:17:48 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Built some more of the message expiry infrastructure

1998-10-11 Nathan Bryant <bryant@cs.usm.maine.edu>
        * citserver.c: fix two more overruns, one of which was preventing
          the "From Host" from showing up in the <W>ho listing.

Sun Oct 11 02:51:55 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Moved "struct visit" and its associated defs from citadel.h to
          server.h where they belong
        * Set up data structures for room policies (expiry, etc.)

1998-10-10 Nathan Bryant <bryant@cs.usm.maine.edu>
        * citserver.c: fix overrun which caused segv's on servers with long
          hostnames.

Fri Oct  9 18:34:06 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * user_ops.c: added PurgeStaleRelationships() to do processing at
          session logout time to remove visits for rooms which no longer exist
        * user_ops.c: implemented NewMailCount()

1998-10-09 Nathan Bryant <bryant@cs.usm.maine.edu>
        * serv_chat.c: fix buffer overrun that was resulting in segv's
        * serv_chat.c: fix another overrun that could cause sessions to hang,
          and cleaned up some other strncpy()-related stuff. DON'T FORGET TO
          NULL-TERMINATE DESTINATION BUFFERS AFTER STRNCPY CALLS.

Fri Oct  9 13:22:37 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented "lazy mode" traversal - pressing the space bar will do
          <N>ext messsage, <G>oto next room, or read <N>ew as appropriate.
        * room_ops.c: modify CtdlRoomAccess() to allow access to mailbox rooms
          only to their owners.

Thu Oct  8 17:13:27 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * messages.c, citadel.rc: added the ability to display message numbers
          in the header when reading messages.  I think this is butt ugly, but
          some of the DaveCode afficionados seem to like it...

Thu Oct  8 15:34:45 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * room_ops.c: Added is_noneditable() function to replace all of the
          duplicated code present in all functions which edit room parameters.

1998-10-08 Nathan Bryant <bryant@cs.usm.maine.edu>
        * lots of warning fixes; builds with -std1 on dec unix
        * aidepost.c, citadel.h, citmail.c, file_ops.c, msgbase.c, netmailer.c,
          netproc.c, rcit.c, server.h, stats.c, userlist.c: use time_t where
          needed
        * control.c, room_ops.c, serv_chat.c, sysdep.c: use memset() instead of
          bzero()
        * dynloader.c, dynloader.h, messages.c, server.h, sysdep.c,
          sysdep_decls.h: function pointer/prototyping fixes
        * rooms.c: use mkfifo(3) instead of system("mkfifo")

1998-10-07 Nathan Bryant <bryant@cs.usm.maine.edu>
        * snprintf.c, snprintf.h: new files
        * Makefile.in, configure.in, dynloader.c, sysdep.c: support for the
          above; citserver now builds and runs on Digital Unix 4.0d with the
          GNU-style configure script. there is a bug with the hostname display
          in the wholist.
        * netproc.c: sillyness fix
        * room_ops.h: prototype delete_room()
        * client_chat.c, commands.c, serv_chat.c, sysdep.c:
          use HAVE_SYS_SELECT_H macro

Mon Oct  5 17:01:32 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Began fixing the stuff I broke

Sun Oct  4 23:35:18 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Did the big migration to the new data structures.  Lots of stuff is
          now broken.  Basic moving from room to room works, but Mail is
          broken, and some of the administrative commands are unimplemented.

1998-10-02 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in: autologin now defaults to disabled

Fri Oct  2 00:04:31 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Finally removed all three usersupp.foo[MAXROOMS] elements, and
          migrated all the code that used them to use "struct visit" instead.

Thu Oct  1 23:02:20 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Rewrote [l][get|put]room() functions to use room names rather than
          room index numbers.  Temporarily prepended a "n" to these four
          function names until they are put to use.

Thu Oct  1 16:27:13 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Removed a few more references to usersupp.lastseen[]

1998-10-01 Nathan Bryant <bryant@cs.usm.maine.edu>
        * .cvsignore: add so_locations (generated by osf1 ld with shared libs)
        * Makefile.in: restructured variables for greater consistency, use
          @echo to print out notices during the make process, add so_locations
          to `cleaner'
        * configure.in, Makefile.in: configure checks for -rdynamic
        * ipc_c_tcp.c: fix DEC compiler warning wrt unsigned char
        * stats.c: add semicolon to placate DREC compiler
        * user_ops.c: define _POSIX_C_SOURCE, include <limits.h>
        * configure.in: pass -pthread to DEC compiler, don't check for
          libpthread[s] on DEC Unix

1998-09-30 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: new variable PTHREAD_DEFS for portability
        * aidepost.c, citadel.c, citmail.c, mailinglist.c, msgform.c,
          netmailer.c, netpoll.c, netproc.c, rcit.c, readlog.c, setup.c,
          stats.c, userlist.c, whobbs.c: return type of main() is int
        * citadel.c, commands.c, messages.c: use time_t properly
        * citserver.c: include <limits.h>
        * config.guess, config.sub, install-sh: new files
        * configure.in: don't use gcc on Digital Unix

Tue Sep 29 23:17:34 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * room_ops.c: modified usergoto() to look at the new data structures
          for counting new messages and such.

1998-09-29 Nathan Bryant <bryant@cs.usm.maine.edu>
        * user_ops.c: fix compiler warning and potential memory leak,
          include sysdep.h
        * configure.in, Makefile.in: only build the server if we find pthreads
        * Makefile.in: realclean removes config.{cache,log,status}

Tue Sep 29 13:20:14 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Removed code from some of the utilities which was still attempting
          to access the old non-gdbm data store.
        * housekeeping.c: rewrote check_ref_counts() to do a ForEachRoom()
          traversal instead of a MAXROOMS loop.
        * sysdep.c: set up a dummy CitContext record to be used during server
          startup, during which time there is no real context.

Mon Sep 28 23:51:51 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented the function ForEachRoom() to handle all-rooms traversal
          (this will work with both the old and new paradigms, because both
          use a GDBM database with one room per record).  Migrated all room
          list commands to use it.

Mon Sep 28 22:05:35 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented the function CtdlRoomAccess() to handle *all* of the
          user-access-to-rooms functionality.  Migrated all room list commands
          to use it.  Still need to migrate gotos.

1998-09-28 Nathan Bryant <bryant@cs.usm.maine.edu>
        * configure.in, acconfig.h: new files; partially functional GNU
          autoconf configure script. Run autoheader; autoconf; ./configure
          --prefix=`pwd` to test.
        * Makefile.tmpl: new file; this is what Makefile.in used to be. Used by
          Configure.
        * Makefile.in: modified to work with autoconf-style configure script
        * Configure: modified to use Makefile.tmpl and generate autoconf-style
          macros. Removed procfs detection. Pass -O2 to gcc, -O to other
          compilers. Removed mknod/mkfifo detection; code should use mkfifo(3).
        * citmail.c, msgform.c, netproc.c, routines.c, support.c, userlist.c,
          whobbs.c: use HAVE_STRERROR macro rather than NO_STRERROR
        * commands.c: use HAVE_TERMIOS_H macro rather than POSIX_TERMIO
        * netproc.c: remove procfs stuff. simply attempt to kill the target
          process with signal zero instead; this checks whether the process
          exists.
        * setup.c, useradmin.c: use HAVE_CURSES_H macro

Sun Sep 27 23:41:29 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * BOTH the old and new generation systems are being written to at
          this point.  Code that reads stuff is still using the old system.

Sun Sep 27 16:10:49 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Changed all "generation" variables from char to long, in preparation
          for removing MAXROOMS.  Generations for new rooms are now timestamps.
        * Defined "struct visit" to hold user/room relationships.
        * Removed some #define's from citadel.h that are no longer used.

Wed Sep 23 13:41:49 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * More changes to support attachments.  They mostly work, but only
          in fixed-format messages.

Mon Sep 21 21:19:17 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * msgbase.c: began laying the groundwork to support attachments.
          Purchased Rogaine(tm) in preparation for expected hair loss.

1998-09-21 Nathan Bryant <bryant@cs.usm.maine.edu>
        * msgbase.c: include dynloader.h
        * citadelapi.h: removed
        * dynloader.h: prototype CtdlRegisterUserHook()

Sun Sep 20 18:56:37 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Added a UserFunctionHook category to implement hooks which perform
          operations on various users or usernames

Fri Sep 18 21:14:41 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * citserver.c: removed cmd_extn() and related code

1998-09-18 Nathan Bryant <bryant@cs.usm.maine.edu>
        * user_ops.c: include dynloader.h
        * roomstats.{c,mk}: removed
        * Configure, Makefile.in: autodependency-related fixes

Thu Sep 17 22:55:29 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Various changes to allow "new messages" to work correctly with Mail

Thu Sep 17 22:21:45 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * server.h, dynloader.c, citserver.c, user_ops.c: reduced the number
          of hook types by inventing an EventType field to the Session hook.
        * proxy.c: added pre-fetching

1998-09-17 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: add SERV_MODULES and PROXY_TARGETS to `cleaner'
        * dynloader.[ch], serv_{chat,test}.[ch], sysdep.c: cleaned
          up the dynamic loader interface as follows:
          - all the symbol table stuff is gone.
          - modules are loaded once at server startup and never unloaded.
          - Added a new function CtdlRegisterProtoHook() to handle the stuff
            that was being done with the symbol tables.
          - Dynamic_Module_Init() now returns a pointer to a static struct
            DLModule_Info; this structure itself has been modified to use char*
            fields instead of fixed char arrays.
        * roomstats.c: include <stdarg.h> not <stdargs.h> (is this file still
          in use?)
        * Configure, Makefile.in: added autodependency support

Wed Sep 16 22:25:13 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented separate structs, lists, and functions for each type
          of server-side hook available.

1998-09-16 Nathan Bryant <bryant@cs.usm.maine.edu>
        * ipc_c_tcp.c: Fixed up some #include/prototyping stuff, call memcpy()
          instead of bcopy()
        * hooks.h: removed
        * sysdep.c, user_ops.c: removed reference to hooks.h

Wed Sep 16 11:42:42 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * ipc_c_tcp.c: Reversed any changes that have been made to this file,
          because something was causing abominally slow response time.
        * proxy.c: added.  This will eventually become a caching, pre-fetching
          multiuser proxy server for the Citadel protocol.

1998-09-15 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: remove support.o from serv_chat.so, add -fPIC to compile
          flags for serv_chat.o
        * dynloader.c: include "sysdep_decls.h", use RTLD_NOW not RTLD_LAZY
        * dynloader.h: prototype CtdlRegisterHook()
        * .cvsignore: added data

Mon Sep 14 20:49:08 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Tried my hand at adding the ability for server extensions to
          register various types of "hooks" in addition to just adding
          server commands.  This is probably not final.

Tue Sep  8 12:11:56 EDT 1998 Brian Costello <btx@calyx.net>
        * Added support for dynamic server modules.  Reworked serv_chat.c
          to be such a module.

Tue Sep  1 23:09:50 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * userpurge.c: rewrote using functions from the server core, rather
          than the now-defunct external API.  This'll be ready once the module
          loading code is done.  (I just had to commit _something_ tonight.)

Mon Aug 31 22:47:58 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Yanked the citadelapi.c module.  This wasn't working out well.
        * techdocs/citadelapi.txt - began documenting the new API to be used
          by modules which will be dynamic linked into the server - most of
          this API is existing server functions.
        * Added a ForEachUser() function with callback mechanism, and reworked
          cmd_list() to use it.

Sun Aug 30 21:52:43 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Moved all of the gdbm databases to a separate "data" directory.

1998-08-26 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: realclean removes Makefile, fixed `touch citadel.h'
          problem

1998-08-25 Nathan Bryant <bryant@cs.usm.maine.edu>
        * room_ops.c: include time.h
        * userlist.c, whobbs.c, serv_chat.c, user_ops.c, sysdep.c, stats.c,
          citadel_decls.h, commands.c, messages.h, routines.h, routines2.h:
          remove duplicated declarations

Mon Aug 24 23:45:01 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * setup.c: Removed yesno_s()
        * citadel.h, room_ops.c: added QRmtime field to struct quickroom,
          modified whenever a room is modified or posted in.
        * citadelapi.c: Added CtdlForEachRoom() function

Mon Aug 24 20:04:04 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: new target `cleaner' does the same as `realclean' 
          without removing sysdep.h
        * proto.h: is bad. eliminate. I've moved the prototypes into several
          header files, one per .c file

Mon Aug 24 00:45:55 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Added a CtdlGotoRoom() function to the CitadelAPI.
 
Sun Aug 23 21:47:00 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * sysoputil is finally dead!  Removed it from the build.
        * Added userpurge.c server extension (initial implementation)

Tue Aug 18 00:42:33 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: `clean' target no longer rm's sysdep.h; new target
          `realclean' removes everything clean does, plus sysdep.h, plus
          target binaries.
        * Configure: add -Wstrict-prototypes to CFLAGS for gcc systems
        * *.[ch]: protoized. Added several new header files containing
          prototypes and other external declarations; many duplicated
          declarations still should be moved to header files. proto.h must die
          as well, IMHO.

Mon Aug 17 23:52:13 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Implemented a bunch of user account related functions in the
          CitadelAPI library.

Mon Aug 17 20:01:18 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Fixed the crash problem.  It wasn't AGUP/ASUP, but rather a buffer
          overrun in getuser() (thanks, Nathan).  Implemented overrun checks
          in getuser(), getroom(), and getfloor() to prevent future problems.

Mon Aug 17 00:06:52 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Updated citmail.c with the latest stuff from the production system.
        * Implemented AGUP and ASUP commands, but AGUP crashes the server
          after its first successful use (user-not-found's don't affect it).

Thu Aug  6 19:25:01 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Got the CitadelAPI library to the point where the server can start
          up an extension, and the extension will connect to the server, do
          some initialization, call a user-supplied CtdlMain(), and exit.  Also
          hacked together a _temporary_ form of the new EXTN server command.
 
Wed Aug  5 23:02:22 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Second attempt at getting the server API started.  Now it runs
          outside of the server and builds a connection.

Tue Aug  4 18:33:06 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Modified the appearance of Internet addresses when they arrive on
          a Citadel system.
        * Removed the <E> field from the message format writeup in hack.txt.
        * Fixed-up citmail.c so that it doesn't try to do database lookups.

Mon Aug  3 23:01:37 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Started developing the server-side API.  This is in its very
          initial stages.  See serverapi.c and techdoc/api.txt

1998-08-02  Nathan Bryant  <bryant@cs.usm.maine.edu>
        * Makefile.in: added config_decls.h to dependencies

Sun Aug  2 21:09:09 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * config_defs.h: renamed to config_decls.h
        * config.c, sysoputil.c: updated to reflect the above

Sun Aug  2 18:52:05 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * config_defs.h: new file, contains external declarations from config.c
        * config.c: moved defs to config_defs.h, use PATH_MAX from <limits.h>
          for ctdl_home_directory
        * mailinglist.c, support.c: include <string.h>
        * sysoputil.c: include <string.h>, <limits.h>, "config_defs.h", remove
          duplicated defs, replace gets() call with fgets()
        * user_ops.c: define _XOPEN_SOURCE_EXTENDED

Sat Aug  1 18:32:52 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * ipc_c_tcp.c: fixed order of memcpy parameters after gethostbyname

Sun Jul 19 17:26:12 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * ChangeLog: reordered; the GNU standard is to add new entries to the
          top.
        * .cvsignore: added userlist

Sun Jul 12 20:58:59 EDT 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Finished migrating everything to the new data store.
        * Replaced the binary "calllog" with the ASCII "citadel.log"
        * Began converting broken utilities that depend on the old data store

Sat Jul 11 00:20:48 EDT 1998 Nathan Bryant <bryant@cs.usm.maine.edu>
        * Makefile.in: removed msgstats

Fri Jul 10 1998 Art Cancro <ajc@uncensored.citadel.org>
        * Initial CVS import