[osg-users] [Any feedback welcome] lack of robustness of the VAO path

[Julien Valentin]

Arg it doesn't crash with Mesa 18.0.5/intel chipset!?


mp3butcher wrote:
> Hi
> I think the first step to fix this issue is first to reproduce it....
> I crash osg3.6 with a GT440 on Ubuntu NVIDIA driver 390.87
> https://github.com/openscenegraph/OpenSceneGraph/issues/697
> I'll check if mesa fails too tonight....
> But  if anyone have the time to confirm it, I would be very glad
> Thanks
> 
> 
> 
> mp3butcher wrote:
> > I admit the fix is kind of hacking :/ Haven't find a proper way to fix it yet
> > but you keep closing issues pointing the bug...
> > I debugged a lot to find this issue so please respect that.
> > I reopened an issue https://github.com/openscenegraph/OpenSceneGraph/issues/697....may this one is clean enough not to be treated as garbage
> > 
> > 
> > 
> > 
> > robertosfield wrote:
> > > On Mon, 21 Jan 2019 at 10:56, Julien Valentin
> > > <> wrote:
> > > 
> > > > If you have any real argument against this change, please expose it publically.
> > > > 
> > > 
> > > The Issue code your wrote is a mess of bad programming practices.  How
> > > would we know where a problem is when the Issue code has so many
> > > problems of it's own.  Should these be fixed first?  Would the problem
> > > still exist once they are fixed?
> > > 
> > > The PR is idiotic.  Changing a osg::State pointer to the current VAS
> > > just because there is a "chance" that a VAS might be the same one as
> > > one that is deleted and invalidate it is just crazy .  It's such a
> > > bad "solution" I can't believe that you asking for a public correction
> > > of it.   To see how silly it is just think how often you "fix" would
> > > call the same reset of the osg::Sate::CurrentVertexArrayState() when
> > > you have multiple VAS that need deleting.  If we had a 1000 to delete
> > > then it's call reset 1000 times and even then not one of these 1000
> > > might not be the same as the CurrentVertexArrayState().
> > > 
> > > I'm not rejecting these PR's out of spite, they simply aren't good
> > > fixes, they are hacks.
> > > 
> > > Robert.
> > > _______________________________________________
> > > osg-users mailing list
> > > 
> > > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
> > > 
> > >  ------------------
> > > Post generated by Mail2Forum
> > 
> 


------------------------
Twirling twirling twirling toward freedom

------------------
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=75526#75526