[vtkusers] Autorelease pool issue with vtkCocoaRenderWindowInteractor

[David Cole]

Looks to me like this call...
  // Now that we are about to begin the standard Cocoa event loop, we can
get
  // rid of the 'pool of last resort' because [NSApp run] will create a new
  // pool for every event
  #ifndef __OBJC_GC__
  gEarlyCocoaSetup->DestroyPoolOfLastResort();
  #endif

...should be moved up to just before the lines that read:
  // Start the NSApplication's run loop
  [NSApp run];

...they used to be in exactly that order prior to revision 1.15 of that
file.

Unfortunately, that also means it looks like moving the entire
vtkEarlyCocoaSetup class above the ObjectiveC class implementations to
factilitate that call.

I've modified the file to see if it works for you. I will send it to you
separately as an attached file. Let me know if it works for you and I will
commit it to VTK after you let me know...


Thanks, (and hope this helps!)
David Cole
Kitware, Inc.


On 2/23/08, Corrie Allen <corrieallen at gmail.com> wrote:
>
> Hi David & VTKers,At the time my program runs, __OBJC_GC__ is undefined
> and I can't find anywhere in my project, the VTK build, or my environment
> setting where it is either defined  (or undefined).  So, if the definition
> isn't hiding away somewhere, my understanding of it is that the
> PoolOfLastResort "hack" should be setting up a pool that can be subsequently
> autoreleased during Cocoa operations.   I guess I need to now dig into the
> workings of CMake so as to understand what Sean is doing with his
> RogueResearch builds.  Any further suggestions (Like Hey Corrie, buy the
> Cmake book!) will certainly be appreciated.
> Corrie
>
> On Feb 22, 2008, at 9:55 AM, David Cole wrote:
>
> Look at the code around line 222 of
> VTK/Rendering/vtkCocoaRenderWindowInteractor.mm
>
> If you are getting that message, then the vtkEarlyCocoaSetup constructor
> is probably not being called. Maybe __OBJC_GC__ is defined, but there is a
> need for an autorelease pool anyhow....?
>
> Some of the RogueResearch mac VTK dashboards use this cache entry...
> VTK_REQUIRED_OBJCXX_FLAGS:STRING=-fobjc-gc
> ...to turn on objective C++ garbage collection.
>
> Are you doing that in some fashion for your build? Or is __OBJC_GC__
> defined for your build somehow?
>
>
> HTH,
> David
>
> On 2/21/08, Corrie Allen <corrieallen at gmail.com> wrote:
> >
> > Hi VTK gurus,
> >
> > I'm getting this snippet of output in my log window during the run of a
> > little program I've written.
> > "
> > Gonna initialize interactor next!
> > And now starting interactor
> > 2008-02-21 20:03:26.496 spawar2[1365:10b] *** _NSAutoreleaseNoPool():
> > Object 0x13e22460 of class vtkCocoaServer autoreleased with no pool in place
> > - just leaking
> > Stack: (0x90a3012f 0x9093cec2 0xda040 0xdaa3c 0x702b 0x6176)
> > "
> > The c++ code where this is occurring is
> > ....
> >    cout<<"Gonna initialize interactor next!"<<endl;
> >   iren->Initialize();
> >   cout<< "And now starting interactor"<< endl;
> >   iren->Start(); //where iren is an instance of
> > vtkCocoaRenderWindowInteractor
> > ....
> > The 3-D object is rendered fine after the log note appears and the
> > interactor works well. The log output,though, is worrisome.  Where can I
> > begin to look for the cause of this: just leaking with no pool in place
> > seems bad.
> >
> > Corrie
> >
> > MacBook Pro, OSX 10.5.1, CMake 2.4.8, VTK 5.1,
> > i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1 (Apple Inc. build 5465)
> >
> >
> > _______________________________________________
> > This is the private VTK discussion list.
> > Please keep messages on-topic. Check the FAQ at:
> > http://www.vtk.org/Wiki/VTK_FAQ
> > Follow this link to subscribe/unsubscribe:
> > http://www.vtk.org/mailman/listinfo/vtkusers
> >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20080223/dc79cf20/attachment.htm>