Re: [sqlite] Many bugs with threads (errors and corruption)

2004-01-10 Thread D. Richard Hipp
Guillaume Fougnies wrote: 2) Disk image corruption (with indexes): After a time of running, i get this error in return: "database disk image is malformed". I have tracked this problem down to what appears to be a bug in Linux threads. For details see http://www.sqlite.org/cvstrac/tktview?tn=56

Re: [sqlite] Many bugs with threads (errors and corruption)

2004-01-09 Thread Guillaume Fougnies
Fri, Jan 09, 2004 at 09:08:37AM -0500: D. Richard Hipp wrote: > Guillaume Fougnies wrote: > Probably the SQLITE_SCHEMA error is being incorrectly converted > into an SQLITE_ERROR somewhere as the call stack unwinds. I'll > fix it - but I consider this a minor problem. If you can give > me more hi

Re: [sqlite] Many bugs with threads (errors and corruption)

2004-01-09 Thread D. Richard Hipp
Guillaume Fougnies wrote: CFLAGS : -O2 -ggdb -DTHREADSAFE=1 -DNDEBUG=0 \ -DOS_UNIX=1 -DOS_WIN=0 -DHAVE_USLEEP=1 For debugging, it is best, I think, to omit the -DNDEBUG altogether. Unless I'm badly mistaken, any definition of the NDEBUG macro will disable the assert()s, even a definition of

[sqlite] Many bugs with threads (errors and corruption)

2004-01-09 Thread Guillaume Fougnies
I'm embedding sqlite using pthread. I have multiple tables creation table on the fly and many concurrent accesses. Sqlite version : 2.8.9 OS : Linux Debian Cc : gcc version 3.3.2 CFLAGS : -O2 -ggdb -DTHREADSAFE=1 -DNDEBUG=0 \ -DOS_UNIX=1 -DOS_WI