Art Cancro [Wed, 29 Jul 2009 03:51:47 +0000 (03:51 +0000)]
* Eliminated the 'gotonext' problem with the bbs view drop-down selector by changing it to 'readgt' and keeping track of the range of messages which was returned from the previous operation. I'm assuming that this 'longvector' stuff is a set of general purpose registers; if that's not the case then I broke something.
Matt [Sat, 25 Jul 2009 02:06:34 +0000 (02:06 +0000)]
Comment out call to StrBufCutLeft, this was cutting the first character off the roomname causing groupdav delete to goto 'alendar' instead of 'Calendar' first
Art Cancro [Fri, 24 Jul 2009 20:51:52 +0000 (20:51 +0000)]
* Added new readloop command 'readgt' which will produce messages with numbers greater than the value supplied in the 'gt' variable. This of course maps to the 'MSGS GT|xxx' server command. For my next trick, I will modify the msgomatic selector to use this command, but first, I must find a way to export to the templating engine the value of the first msgnum we retrieved during the *current* operation.
* check more carefully the target buf-size, since we're converting to utf-8 on char may become 6
* if we've got a non encoded header, decrease len by it.
Art Cancro [Thu, 23 Jul 2009 16:06:24 +0000 (16:06 +0000)]
* serv_migrate.c: message encoding buffer is now static in order to avoid endlessly abusing malloc/free, and hopefully reduce the chances of heap corruption.
Matt [Tue, 21 Jul 2009 04:01:33 +0000 (04:01 +0000)]
* Fallback to eval() method for JSON loading if the native JSON object (FF 3.5, IE8) does not accept our data.
* Silence any harmless errors caught in IE
* create struct, so we can transport hostlists in and errormessages out
* report failed notifications to aide _after_ sending all notifications of this chunk.
Art Cancro [Fri, 3 Jul 2009 05:27:48 +0000 (05:27 +0000)]
* Temporarily disabled the 'Drafts' functions. I like the idea, but it doesn't work, and I'd like to shuffle some things around. But in the meantime, we need to get 7.60 out the door. The code is still there but the buttons are commented out, except for the VIEW_DRAFTS view, which I removed completely. Since the 'Drafts' room will end up being either a mailbox view or a BBS view, chosen by the user, we should implement some other way of determining whether we're in the Drafts room, and only offer the 'Edit' button in that case. Currently, when I click the 'Edit' button, it opens a message entry screen containing an error about a missing template.
Art Cancro [Tue, 30 Jun 2009 23:53:24 +0000 (23:53 +0000)]
* Repaired the code that decides whether it should not bother converting character sets to UTF-8. Not only was there a logical AND where there should have been a logical OR, but the whole thing was in the wrong place. As a result, it only made that decision some of the time. There may be other bugs in this, but I was unable to find them.
Art Cancro [Tue, 30 Jun 2009 23:29:40 +0000 (23:29 +0000)]
* Removed the RSS output code (which was still in TECH_PREVIEW anyway). There's a lot of duplicate code in there, and the underlying framework has changed enough since this was written that when we do finally get anonymous logins working, we're going to approach this in a completely different way.
* fix groupdav handling
* add common logic for outputting 401; centralize it, register it as handler.
* remove params to groupdav functions, since WC gives them all information they need.
* use the general post handler for groupdav too.
* since /groupdav/ isn't left in the path by the current logic anymore, we need to count less when parsing.
* add support for reading the UID via getsockopt from unix domain socket conneciotns in begin_session()
* use the UID like bzed suggested in debian Bug#528760 when sending mails via LMTP
* add configure options to detect ucread; borowed from http://git.musicpd.org/cgit/master/mpd.git/plain/m4/ucred.m4; thanks to Max Kellermann to point me there.
a further good read is http://www.wsinnovations.com/softeng/articles/uds.html ; it indicates that this might not be portable to *bsd / Solaris; configure should protect us from problems here though.
Art Cancro [Fri, 19 Jun 2009 20:57:10 +0000 (20:57 +0000)]
* locate_user_vcard_in_this_room() lost the ability to return the msgnum of the vcard it found during one of the recent overhauls. Fixed. The 'update your contact information' link wasn't working at all.
Art Cancro [Fri, 19 Jun 2009 18:26:30 +0000 (18:26 +0000)]
* do_edit_vcard() contained a 'StrBuf *Buf' which was unused in the entire function except for a call to free it. WebCit no longer crashes upon attempting to edit a vCard.
* Also performed some style cleanup while I was in here.
Art Cancro [Thu, 18 Jun 2009 17:17:45 +0000 (17:17 +0000)]
* Fleshed out the ctdlsh interface a bit. Right now it's just a simple shell that logs in to the server, authenticates as an internal program, and allows the user to enter protocol commands. In the future we'll probably add some other stuff to it. Or I might get bored and never touch it again. Who knows?
Art Cancro [Wed, 17 Jun 2009 21:26:18 +0000 (21:26 +0000)]
* ParsePref() was returning if it encountered a blank line in the preferences data, leaving the rest of the server transaction unfinished and causing the protocol to get out of sync. Fixed. This is almost certainly the cause of the 'numeric room name upon login' bug.