--- Chris Schirlinger <[EMAIL PROTECTED]> wrote: > On 6 Apr 2005 at 3:13, Dan Kennedy wrote: > > Date sent: Wed, 6 Apr 2005 03:13:58 -0700 (PDT) > From: Dan Kennedy <[EMAIL PROTECTED]> > Subject: Re: [sqlite] Access Violations in sqlite3_step when in > DLL > To: sqlite-users@sqlite.org, [EMAIL PROTECTED] > > > > > Exactly what are you calling sqlite3_free() on? > > Sorry, I am probably being a bit misleading there. That Free was on > any error message returned from SQlite3_ErrMsg which we call if the > result from SQLite3_Open isn't SQLITE_OK
It sounds like it's not the problem, but don't call sqlite3_free() on anything returned by sqlite3_errmsg(). This memory is managed by SQLite and calling sqlite3_free() on it will cause memory corruption somewhere down the line. Are you *sure* it's not being called? ;) Try defining -DSQLITE_MEMDEBUG=1 when you compile SQLite. If it is memory corruption this might catch the problem sooner rather than later, and maybe more consistently. If you can get it to occur in a consistent manner you can probably use a debugger to figure out who is colouring outside the lines. > In theory it's never called, we never get any errors out of > SQLIte_Open, it just AV's > > We call SQLite3_Close and pass the pointer that is returned from the > open call. I havn't seen any errors on close. We see errors on Open, > Step or Get_Table depending on what we use (ie, if we just Open and > close, the AV occurs eventually on one of the opens. If we try > Get_Table the AV will happen sooner and happen on a Get_Table) > > None of the code we used (and I even tried other componants and > wrappers in case the one we were using is suspect) has ever caused > any issues previously and we've been using it for months. Only when > called from this plugin are we seeing problems. > > I've tried 2 wrappers and one componant and still get the issue __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com