Re: [Denemo-devel] lily 2.16, 2.18 zombies on darwin when ran via glib g_spawn

[Richard Shann]

On Fri, 2014-01-17 at 05:34 +0100, David Kastrup wrote:
> Jeremiah Benham <address@hidden> writes:
> 
> > I am having an issue running lilypond via glib when cross compiled for 
> > darwin.  
> >
> > This is when cross compiled for darwin using the same gub snapshot:
> > Version/Status
> > 2.14.2 works fine
> > 2.16.2 created pdf and .ps but goes into a zombie state on exit
> > 2.18.0 does not create anyting and immediately goes zombie
> 
> Are you using "zombie state" in the official meaning here (a process
> that has terminated without its parent waiting for it, listed in a "ps
> a" listing as Z)?  Or just as a colloquial description?
> 
> > Now all of the above versions are executable. I inserted a printf of
> > the lilypond arguments before g_spawn.... was launched. Then I copied
> > and executed it at the command line. Each time it was a success on all
> > three tested versions. Why is it not working via g_spawn...? I have
> > cross compiled all the above for mingw and have never had this
> > problem. I am begining to diff through versions 2.14 and 2.18. Does
> > anyone have any idea as to what is causing this?
> 
> At some point of time, the terminal output was not routed through
> whatever Guile considered to be its current-error-output but rather
> through file descriptor 2.  If you have a blocking pipe or similar on
> that 

is "that" here file descriptor 2 or current-error-output ?

The "Guile" you mention here is guile invoked by LilyPond I assume.
Denemo asks GLib to capture the stdout and stderr of the spawned
process ...

Richard


> in your process environment, this might be good for problems.
>