2.4.3 - not ready

[Phil Longstaff]

On Thu, 2011-02-24 at 19:40 -0800, John Ralls wrote:
> On Feb 24, 2011, at 6:16 PM, Phil Longstaff wrote:
> 
> > I tagged and built 2.4.3.  It passes make check and make distcheck.
> > However, when I run it, I get a segv when I close gnucash.  The problem
> > seems to be during the call to gnc_dbi_unlock().  While looking for the
> > current process pid in the lock table, the sqlite3 error fn is called.
> > According to valgrind, when setting the error code, this is an invalid
> > write, past the end of valid memory.
> > 
> > I think this should be tracked down and solved before 2.4.3. is
> > released.
> 
> Odd that it popped up now; that code hasn't been touched since 2.3.16. Untag 2.4.3 and I'll take a look at it tomorrow.
> 
> Oh, are there any pertinent messages at the end of your gnucash.trace?

I've noticed for a while that I get:
* 13:42:37 DEBUG <gnc.backend.dbi> [enter
gnc-backend-dbi.c:gnc_dbi_session_end()]
* 13:42:37  CRIT <gnc.backend.dbi>     [sqlite3_error_fn()] DBI error:
* 13:42:37  CRIT <gnc.backend.dbi>     [sqlite3_error_fn()] DBI error:
* 13:42:37 DEBUG <gnc.backend.dbi> [leave gnc_dbi_session_end()]

Real helpful.

Phil