[vtkusers] AWT + SWT + Event Loops (was Re: VTK6 and Java Cone Example 5)

[Ric Wright]

So did some more spelunking and it looks like this is just a basic
incompatibility for SWT and AWT ­ in the specific case where AWT is trying
to start their own event loop once SWT has started.  Previously, in my other
work, I handled events on my own ­ I didn't ask or expect AWT to start or
manage it's own events. I used AWT basically to host the OpenGL canvas
(based on JOGL).  But when one invokes the vtkRenderWIndowInteractor an
event loop IS spawned and the causes various weird problems.

There is a fairly long summary of some of the issues here:

http://marc.info/?l=openjdk-awt-dev&m=135818602009136&w=2

So the question that comes to my mind is how dependent is the rest of VTK on
starting its own event loops?  I.e. Can I use the rest of VTK and its
pipeline to perform graphics, manage datasets, etc.  I am willing to give up
the Interactor and its mimicking of trackball if everything else is likely
to work.  But if spawning its own event loop(s) is intrinsic to how most of
VTK works then I'm probably dead in the water.

Thoughts or suggestions?

Thanks
Ric


From:  Ric Wright <rkwright at geofx.com>
Date:  Tuesday, August 13, 2013 12:59 PM
To:  Sebastien Jourdain <sebastien.jourdain at kitware.com>
Cc:  Sean McBride <sean at rogue-research.com>, vtkusers <vtkusers at vtk.org>
Subject:  Re: [vtkusers] VTK6 and Java Cone Example 5

An update on this.  I ported over some of my other Eclipse work and adapted
it to VTK.  It is pushed to my account on github and there is a writeup
here:  http://rkwright.github.io/geofx-pages/vtkview.html.  There are 3
articles in there.  The latest rev of the code is here:
https://github.com/rkwright/vtkscene in the develop branch.

Mostly, it works great.  However, when I tried the fifth cone example, which
has the problematic vtkRenderWindowInteractor, I ran into similar problems
as before.  I instantiated it with a valid vtkPanel.RenderWindow but it
still goes off into the ozone, sometimes crashing the native vtkOpenGL
dylib.

I'm going to skip the interactor for now and move on to see if I run into
more problems. If not, I can probably live without that widget but the
problems there might be symptomatic of deeper problems on OSX 64 bit
systems.

Any further suggestions would be welcome.

Thanks
Ric


From:  Sebastien Jourdain <sebastien.jourdain at kitware.com>
Date:  Tuesday, August 6, 2013 8:56 AM
To:  Ric Wright <rkwright at geofx.com>
Cc:  Sean McBride <sean at rogue-research.com>, vtkusers <vtkusers at vtk.org>
Subject:  Re: [vtkusers] VTK6 and Java Cone Example 5

In Java you should use the Java specific classes anyway like
vtkPanel/vtkCanvas or the new version of them used in the vtk.sample package
of the vtk.jar...

The purpose of those example was to show the possible match between the
different languages...

Nothing more I believe,

Seb


On Tue, Aug 6, 2013 at 8:52 AM, Ric Wright <rkwright at geofx.com> wrote:
> Ah, OK.  Well, I don't know what to do here.  Is this happening to others?
>  Should I file a bug?  Just try to work around it by not using
> RenderWindowInteractors (sounds kind of lame).  Or?
> 
> Suggestions?
> 
> Thanks
> Ric
> 
> 
> On 8/5/13 4:15 PM, "Sean McBride" <sean at rogue-research.com> wrote:
> 
>> >On Mon, 5 Aug 2013 16:10:55 -0500, Ric Wright said:
>> >
>>> >>Well, a little playing around didn't seem to help, but I am frankly not
>>> >>an
>>> >>expert on threads.  However, it seems to me that the source of the
>>> >>problem
>>> >>is that the vtkRenderWindowInteractor gets initialized then the start()
>>> >>method creates an event loop that is monitored by the
>>> >>vtkRenderWindowInteractor.  I suspect that start() method spawns a new
>>> >>thread to listen to the events and it is the spawning of that thread that
>>> >>causes the exception.
>> >
>> >VTK's Cocoa classes never create any threads.
>> >
>> >Cheers,
>> >
>> >--
>> >____________________________________________________________
>> >Sean McBride, B. Eng                 sean at rogue-research.com
>> >Rogue Research                        www.rogue-research.com
>> <http://www.rogue-research.com>
>> >Mac Software Developer              Montréal, Québec, Canada
>> >
>> >
> 
> 



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20130814/2aa4f775/attachment.htm>