Art Cancro [Mon, 26 Apr 2010 16:20:53 +0000 (16:20 +0000)]
* Executive decision. From now on, intr (ctrl-c) begins an attempt at a graceful shutdown, but quit (ctrl-backslash) forcibly quits the server. Signal handler for SIGQUIT has been disabled.
* add Remove button
* post_message(): use IKEY and Flathash, we just use numbers anyway.
* post_message(): add handler for removal of attachments; it seems as if 'filename' comes out URL-encoded, so we need to decode it.
We need to iterate and strcmp() so we find the right one :(
* post_message(): when adding a new item just using count+1 isn't sufficient anymore; we need to use the key value + 1 of the last item in the list instead
* DeleteEntryFromHash(): don't skip indirection via the lookuptable, else we free some random payload
* DeleteEntryFromHash(): NULL the end of the list; This is the entry we just removed.
* hashlist_test: add some tests with removing items from the list
* move policy.c into modules/expire/expire_policy.c, since it just controls this.
* remove all unneded includes to policy.h
* remove hardcoded reference to policy from mk_module_init.sh (hard to find...)
* add one more callback hook to the webcit DAV handler structure; this will be used to revalidate URLs, move or delete certain items referenced by URLs
* add handler vor EUID based namespaces; so far we just have the exists logic; this will be filled with the other required operations soon.
* finish REST URL parsing with 'best-guess'-room logic
* ReadPostData(): abort if we fail to read the whole content
* session_loop(): allocate buffers after reading post data, so we don't have to free them on error
* session_loop(): abort if reading post data fails
* change_start_page(), LoadStartpage(): check whether the startpage is allowed by rule
* InitModule_MSG() disallow postpart_download, postpart, confirm_move_msg, delete_msg, move_msg, post to become the startpage; they don't do usefull things there.
Art Cancro [Thu, 8 Apr 2010 04:33:45 +0000 (04:33 +0000)]
* Moved instant messaging into its own module -- serv_instmsg
* Removed serv_chat and the CHAT command
* Implemented serv_roomchat and the RCHT command, which implements a brand new chat protocol
* GetRESTFolder(): find the parts of the URL matching our rooms; since there may be uids/messageids/mimeparts keep a 'best guess' we return if theres no direct match
* GetRESTFolder(): collect sub-rooms of the current URL so we can list them later
* cdb_decompress_if_necessary(): memset 0 the buffer before accessing it
* cdb_decompress_if_necessary(): don't lean on cdb being big enough to fill our header
* string_append_c(): use another name for buf
* imap_roomname():
* 256 is double ROOMNAMELEN; calculate it here to show the relation in place
* extract_token() gives us -1 or the stringlength, use it instead of doing another strlen a line below
* use safestrncpy()
* do_imap_match():
* while lowercasing run till \0 rather than doing strlen above
* respect the size of the target buffer...
* For-loop: assign pointers here, since we start using them here; use IsEmptyStr all over the place (also some forgotten ones...)
* CtdlClientGetLine(): we don't support Error while SSL I/O; remove it.
* client_readline_sslbuffer() put in Pos from the outside
* client_readline_sslbuffer() rewrite contents with StrBufTCP_read_buffered_line_fast() as sample
* fix imap uid fetching:
- imap_strip_headers() : fix loading of the line from the buffer
- imap_strip_headers() : split evaluation for headers_not/headers so we can more easily determine and abort if we want to print the header
- imap_strip_headers() : if ptr IS StrBufNotNULL we're done reading headers
- imap_fetch_body() : partial has to be a StrBuf too if its source is; adjust processing accordingly
* add StrBufPook which blobs out a sub-part of the string
* add StrBufStripAllBut, which aims to replace stripallbut, whose functionality is not yet all clear to me.
* add NULL protection to StrBufSub
Art Cancro [Sun, 7 Mar 2010 05:01:27 +0000 (05:01 +0000)]
* Never mind ... I understand now. Messages which were copied *in* by code between r8300 and r8431 are corrupt on-disk and cannot be rendered. Backed out my last change.
Art Cancro [Sun, 7 Mar 2010 00:19:01 +0000 (00:19 +0000)]
* msgbase.c now contains both the new and old versions of CtdlOutputPreLoadedMsg() so that I can work on other things while the new version is debugged. define NEW_COPLM to get the new version, otherwise you get the one from r8422.
* fix convert_internet_message_buf(); we forgot to append a newline here
* calculate the body start pos while iterating, so we can react the same way we did before
Art Cancro [Fri, 5 Mar 2010 21:19:07 +0000 (21:19 +0000)]
* New parameter to CtdlSaveMsgPointersInRoom() to suppress the reference count adjustment. To be used ONLY when we're going to rewrite the metadata manually. Used ONLY by serv_migrate. Everyone else HANDS OFF.
Art Cancro [Wed, 3 Mar 2010 05:29:25 +0000 (05:29 +0000)]
* Removed code from serv_chat.c that checks to see whether the final character was a newline. Moved it to memfmout() instead. memfmout() is now guaranteed to output a newline as its final character *if* it outputs any data at all.
Art Cancro [Wed, 3 Mar 2010 05:17:28 +0000 (05:17 +0000)]
* Changed strings to 'First Name' and 'Last Name' in the vcard editor to avoid namespace conflict with the strings 'First' and 'Last' in the bbsview. Also changed these strings in all existing localizations in order to avoid losing them.