Release 2.5.3 crashes randomly

[John Ralls]

On Jul 12, 2013, at 2:47 AM, David Carlson <carlson.dl at sbcglobal.net> wrote:
> 
> The file that had the seemingly random crashes was one that has about 20
> (actually 64 in the .gcm) accounts open in the old registers because it
> has been used a lot in stable releases 2.4.13 and earlier.  I am not
> intentionally mixing old and new, but among other things 2.5.3 is
> apparently not advanced enough to automatically switch open registers to
> the new type yet.  I do a lot of 'jumps' from account to account and a
> lot of editing of scheduled transactions which is _not_ working well in
> 2.5.3, but as a practical matter was impossible for me in 2.5.2 and
> earlier.  I have been maintaining a bug report about that.  Perhaps the
> 'jumps' do not always trigger the 'trying to open new register when old
> is open' warning reliably. 
> I do want to use the new register type and my primary test file is, i
> think, all new style, but my real work file is old register style.  That
> is why I have to switch back to 2.4.13 when things start crashing.  New
> style registers automatically get switched back to old style (I think)
> when going back.
> Maybe I should manually maintain two copies of the .gcm files, one for
> '2.4.13' and one for '2.5.x'.
> I do not use nightly builds because (a)they would be a major
> inconvenience to re-compile in Windows and (b) there is no easy to find
> changelog to tell me when one of my obstacles has been addressed.

Nightly builds are at http://code.gnucash.org/builds/win32/trunk/ though they haven't been
exactly nightly lately as there haven't been enough commits.

Changes are published on two mailing lists, gnucash-patches and gnucash-changes. For historical
reasons that were discussed here a couple of months ago, gnucash-changes includes the actual
patches and gnucash-patches has only the commit messages. You'll probably want gnucash-patches,
and I suggest the digest so that you get only one email a day.

Interesting that 2.5 opens "leftover" tabs in the old register if the account was last touched in
2.4. That you have a bunch of accounts open in the old register supports my hypothesis that
the problem might be that the crashes you're getting are from having the same account in both
an old and a new register.

So if you could, try repeating the procedure with which you started the thread on your test
database first with only new register windows open, then with only old register windows open.

Regards,
John Ralls