Re: Parcellfun leaving defunct processes

> Parcellfun (from the General octave-forge package) is really
> promising, but is not quite behaving.
>
> Using octave 3.1.55 and the current download of parcellfun (no version
> number on that sadly) under Ubuntu 9.04...
>
> Something as simple as:
> a = parcellfun(4,@(x)horzcat(1,x),num2cell([1:10]),'UniformOutput',false);
> leaves a bunch of defunct octave processes laying around.
>
> These get cleared when octave exits, but that is the only way I find
> to remove them.  Not a problem for small jobs, but I'm literally
> running into the wall of maximum number of processes.
>
> It has been a while since I did any parallel programming outside of
> mpi, so I'm not sure what the problem really is.
> IIRC though, doesn't the parent process have to catch SIG_CHILD?  I
> don't see that happening in the script.
>
> BTW: If this problem gets sorted, I have a very minor bugfix and
> improvement to submit.  They are pointless if the major bug doesn't
> get fixed first though.
>
> Travis Collier
> UCLA biology grad student
>