[Zope-Coders] Zope 2.5 instability problems

[Martijn Faassen]

Matthew T. Kromer wrote:
> 
> On Monday, December 10, 2001, at 01:28 PM, Martijn Faassen wrote:
> 
> >Hm, then I hope the other traceback that gave Matt a clue *was*
> >correct. I've just run a plain gdb against Zope without trying a debug
> >build of either Python or the Zope modules. Should I try that? And
> >would this help? I have little experience in debugging Python on
> >this level, so some clues would be useful.
> 
> Martijn, I'm preparing a trap patch right now that you should be able to 
> apply that will force an error message and a segfault when the loop 
> condition is detected.
> 
> You should then be able to re-run your condition (still with gdb 
> attached) and get a better notion of what's going on.   If it pops (ie 
> the patch cause you to segfault like it should) I can replace the 
> segfault instruction with a python error raise so you can see the 
> traceback.  Just apply this diff at the top level of your Zope and it 
> will detect if you try to wrap an acquisition wrapper with itself.

Although it seems as it cAccessControl if not the direct cause of the
problem at least triggers this. Is this consistent with your diagnosis? 
Note that the second traceback as I recall was caused by me opening a
node in ParsedXML and *then* going to the management screens -- then it
crashed. I.e. something got into a bad state when I opened the node and caused
it to segfault later on. Though that may have involved wrapping a wrapper
with itself, of course. 

Just as a PS, I'll hang out on #zope IRC for a while now if anyone wants to
talk to me through that channel.

Regards,

Martijn