[Bf-python] Python context state bug?

[Bassam Kurdali]

thanks sergey, I'll see if I can dig deeper and find out what exactly
happens.
On Tue, 2016-11-01 at 11:48 +0100, Sergey Sharybin wrote:
> Hi,
> Multiple handlers should not cause problems, they are executed sequentially. Doing stuff from drivers in new dependency graph might still have some threading issues.
> 
> in any case, hard to guess something. Just minimize the case and make a report i'd say.
> 
> 
> > On Mon, Oct 31, 2016 at 7:17 PM, Bassam Kurdali <bassam at urchn.org> wrote:
> > can this happen from having multiple update handlers/ property updaters etc. in your scene? or is that guaranteed to execute in a sane way?
> > If the former I can expect to see these errors just from installing multiple addons, rather than a bug in a single addon, which feels wrong.
> > 
> > On Mon, 2016-10-31 at 17:10 +0100, Sergey Sharybin wrote:
> > > Hi,
> > > AFAIR, such issues are usually happening when context is used from multiple threads. This is something you are not supposed to do actually, at least without knowing exactly what you're doing (such access is required for render engines to work but there we have special code which guarantees stability in the python bindings).
> > > 
> > > 
> > > > > > On Thu, Oct 27, 2016 at 7:57 PM, Bassam Kurdali <bassam at urchn.org> wrote:
> > > > Hi folks, I've been using animation nodes addon recently - overall,
> > > > it's great, and quite ambitious, given blender's limitations, and while
> > > > 
> > > > it mostly works great, there are a few problems (lockups due to some
> > > > 
> > > > race condition I haven't figured out, occasional DAG errors, maybe more
> > > > 
> > > > things). I'd like to help debugging this addon since it is so darn
> > > > 
> > > > useful.
> > > > 
> > > > 
> > > > 
> > > > One message I've been seeing a lot of recently on the console:
> > > > 
> > > > 
> > > > 
> > > > ERROR: Python context internal state bug. this should not happen!
> > > > 
> > > > 
> > > > 
> > > > So this seems to be connected to a sort of counter - py_call_level,
> > > > 
> > > > which is a global variable in bpy_interface.c it gets incremented, or
> > > > 
> > > > decremented in two functions - the latter in bpy_context_clear() in the
> > > > 
> > > > same file. The error happens when the decrement goes less than zero.
> > > > 
> > > > 
> > > > 
> > > > Any idea why/how this could be happening? I've only the fuzziest idea
> > > > 
> > > > of what is going on, but it seems something in the addon, under some
> > > > 
> > > > conditions, is managing to deepen the call level without having the
> > > > 
> > > > counter increment?
> > > > 
> > > > 
> > > > 
> > > > If anyone has a clue, it would be interesting to figure this one out.
> > > > 
> > > > I'm a bit out of my depth currently.
> > > > 
> > > > _______________________________________________
> > > > 
> > > > Bf-python mailing list
> > > > 
> > > > Bf-python at blender.org
> > > > 
> > > > https://lists.blender.org/mailman/listinfo/bf-python
> > > > 
> > > > 
> > > 
> > > 
> > > 
> > > -- 
> > > With best regards, Sergey Sharybin
> > > 
> > > 
> > > 
> > > 
> > > _______________________________________________
> > > Bf-python mailing list
> > > Bf-python at blender.org
> > > https://lists.blender.org/mailman/listinfo/bf-python
> > 
> > 
> > 
> > _______________________________________________
> > 
> > Bf-python mailing list
> > 
> > Bf-python at blender.org
> > 
> > https://lists.blender.org/mailman/listinfo/bf-python
> > 
> > 
> 
> 
> 
> -- 
> With best regards, Sergey Sharybin
> 
> 
> 
> 
> _______________________________________________
> Bf-python mailing list
> Bf-python at blender.org
> https://lists.blender.org/mailman/listinfo/bf-python
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.blender.org/pipermail/bf-python/attachments/20161109/4e308a84/attachment.html>