Hi Dennis, >> Of course I suspected some error in my program - but there aren't any. >....There aren't *ANY*. > I highly doubt that. :-) All programs of any significant size contain > errors. Hehehe - of course there might be one or two logical errors ;) But regarding memory leaks I cannot find any (analysed by hand (quite a job - but had to use the time between crashed to something), with BoundsChecker and with SciTech Memory Profiler). I have compiled my own SQLite.dll - both with NDEBUG (for release) and with DEBUG - so I would expect the asserts to fire if their conditions was met in DEBUG-mode. /Michael
________________________________ From: Dennis Cote [mailto:[EMAIL PROTECTED] Sent: Fri 2006-05-26 20:19 To: [email protected] Subject: Re: [sqlite] Possible bug in sqlite3_prepare Michael B. Hansen wrote: > > Of course I suspected some error in my program - but there aren't any. > ....There aren't *ANY*. I highly doubt that. :-) All programs of any significant size contain errors. > > I have run SQLite in both runtime and debug - and know that the > sqlite3_prepare has an assert that checks for "mallocHasFailed", but I > haven't been able to verify the value of this because of the nature of > the bug. The assert isn't fired - although this might be because it > resides in a separate DLL and VS.NET doesn't catch it, although I debug > in both native and managed code and is able to step through the > sqlite3_prepare(). Are you saying you have built your own version of the SQLite DLL compiled without NDEBUG set (i.e. a version with the asserts active)? The DLLs available for download are all production builds with the asserts disabled for maximum performance. Dennis Cote
