[vtk-developers] LONG backtrace of segfault in Geovis-TestCoincidentGeoGraphRepresentation2D on bubbles OSMesa build

[David Partyka]

I checked in a fix for this stack overflow yesterday and it seems to have
fixed all the failing tests. It might even make views faster as they are not
not initializing (which internally calls render) every time you call render,
and thus not double rendering ;-)

Thanks to Utkarsh, Rob and Aashish for their advice for toward the final
fix!

On Tue, Jan 4, 2011 at 7:23 PM, David Partyka <david.partyka at kitware.com>wrote:

> Coming back to this. It is definitely a stack overflow. What's mind blowing
> is that this exact code was written in 1998 and 2005 and unfortunately it
> appears the dashboard just pruned itself of September/October's dashboard
> results (grumble). Further more, this particular failures only appears to
> happen on osmesa. I guess I'll see if git bisect can be of assistance.
>
> On Tue, Nov 2, 2010 at 2:12 PM, Kevin H. Hobbs <hobbsk at ohiou.edu> wrote:
>
>> On 11/02/2010 12:28 PM, David Cole wrote:
>> > Why are we doing this? (line 328 and surroundings of
>> Views/vtkRenderView.cxx)
>> >
>> >   if (caller == this->GetInteractor() && eventId ==
>> vtkCommand::RenderEvent)
>> >     {
>> >     vtkDebugMacro(<< "interactor causing a render event.");
>> >     this->Render();
>> >     }
>> >
>> > We should not call Render in repsonse to a RenderEvent unless we are
>> > 100% confident that calling Render will not throw another RenderEvent.
>> > I think this is one of the definitions of an infinite loop. Am I
>> > missing something?
>> >
>>
>> Dropping this render ( and adding a render in each of
>> Charts-TestMultipleScalarsToColors, TestMultiTexturing, and
>> TestMultiTexturingTransform mentioned in another thread ) gets rid of
>> all of my segfaults.
>>
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.vtk.org/mailman/listinfo/vtk-developers
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20110106/fde63aaa/attachment.html>