[Smtk-developers] SMTK code review, more Remus debugging

Robert Maynard robert.maynard at kitware.com
Mon Nov 10 12:13:26 EST 2014


Okay that makes sense.

After you are finished the demo / SC we both should have time to sit
down and figure out the remus zombie issue.

On Mon, Nov 10, 2014 at 12:03 PM, David Thompson
<david.thompson at kitware.com> wrote:
>> Overall the branch looks good to me. I might have missed it but when a
>> bridge starts a server how do you capture and advertise what the
>> endpoint back to that server is?
>
> It's easy to miss and only minimally functional, but when RemusBridgeConnection::connectToServer is followed by a call to bridgeNames(), the BridgeRegistrar is told of new bridges; the smtk::function that the BridgeRegistrar is given to create a bridge will create a RemuteRemoteBridge configured to that endpoint. (I use smtk::function there so that I can pass in a functor that knows the proper endpoint and worker info to use.)
>
> The problems I'm having now are
>
> 1. The unit test works from a command line but not from ctest (unitRemusBridgeConnection starts a process-local server -- which finds a remus-worker file, starts the worker specified by the test, and runs a "read" operation on a test file via the worker). CTest fails to import things properly and I haven't figured out what's different about the environment variables that's causing the issue
>
> 2. The following Python script now runs on the branch but hangs because the worker turns into a zombie process immediately.
>
>   import smtk
>   rbc = smtk.bridge.remote.RemusBridgeConnection.create()
>   rbc.addSearchDir('/usr/local/var/smtk/workers')
>   rbc.connectToServer('local', 50505)
>   print smtk.model.Manager.bridgeNames()
>   sess = mgr.createSession('smtk::model[cgm{OpenCascade}]')
>
> I haven't had a chance to debug in detail and think the SC demo is more important at the moment.
>
>
>> In smtk/common/Paths.cxx Have you thought about simplifying the OS
>> includes, maybe to something like: ...
>
> I like that it's simpler but prefer to keep the headers ordered with highest sanity at the top and no repetition. :-)
>
>         Thanks,
>         David


More information about the Smtk-developers mailing list