[Citadel Development] Citadel commit log: revision 8713

2010-07-24 Thread dothebart

r8713 | dothebart | 2010-07-24 06:05:03 -0400 (Sat, 24 Jul 2010) | 1 line
Changed paths:
   M /trunk/libcitadel/lib/stringbuf.c

* StrBufSanitizeEmailRecipientVector () fix null pointer accesses



[Citadel Development] Re:Citadel commit log: revision 8712

2010-07-24 Thread dothebart


 

Fr Jul 23 2010 16:23:20 EDT von  "ajc" a...@uncensored.citadel.org  Betreff: Citadel commit log: revision 8712
 r8712 | ajc | 2010-07-23 16:23:18 -0400 (Fri, 23 Jul 2010) | 1 line Changed paths: M /trunk/citadel/context.c M /trunk/citadel/context.h M /trunk/citadel/sysdep.c  * Rewrote vCtdlLogPrintf() to NOT use any buffers at all, only v*printf() type functions.  We do this by writing the log data and metadata in separate calls.  In practice, this will cause them to appear on the same line even when other threads are concurrently writing because we are using buffered I/O.  Unfortunately, this does NOT fix the server crash that occurs when saving instant message transcripts to disk.  It points us in the right direction, though.  

seems to me as if you've nuked logging to syslog?
 
I think we should call another log function for mailtraffic logging anyways, which wraps syslog and is independend of what we do.
 
afaik the state as it was before i've added a buffer was exactly because of lines were intermixed:
r5416 | davew | 2007-08-23 19:43:29 + (Thu, 23 Aug 2007) | 1 linePatches from Matt with slight mod from me to do without strlen Also modified lprintf to be atomic when logging to screen.
 




[Citadel Development] Citadel commit log: revision 8714

2010-07-24 Thread dothebart

r8714 | dothebart | 2010-07-24 06:30:45 -0400 (Sat, 24 Jul 2010) | 1 line
Changed paths:
   M /branches/stable-78x/libcitadel/lib/stringbuf.c

* StrBufSanitizeEmailRecipientVector () fix null pointer accesses