On Monday 03 November 2008 02:58:15 Giambattista Bloisi wrote: > Michael Homer wrote: > > On Sunday 02 November 2008 18:32:39 Lucas C. Villa Real wrote: > >> On Sat, Nov 1, 2008 at 7:42 PM, <[EMAIL PROTECTED]> wrote: > >>> Author: mwh > >>> Date: 2008-11-01 14:42:16 -0700 (Sat, 01 Nov 2008) > >>> New Revision: 3704 > >>> > >>> Modified: > >>> trunk/Scripts/bin/GetAvailable > >>> Log: > >>> Hopefully fix the GetAvailable hang with patch from Giambattista Bloisi > >> > >> I believe this deserves a new release.. > > > > Let's find out whether it works first, though. > > > > There do seem to be two distinct bugs playing into it as well, which > > explains why there were two different sets of symptoms. > > PythonUtils.bash(), or any alternative implementation that ultimately > > relies on the subprocess module, is inherently not thread-safe and will > > occasionally leave zombie processes. At this point I'm not sure whether > > those processes complete their work and return or not, but there are > > spare /bin/bash processes living on. > > > > I'm also not sure how to fix it without it amounting to a > > reimplementation of subprocess.Popen. If this patch gets rid of the hang > > entirely it isn't so important, but we're not sure yet. > > In this moment I'm able to reproduce the bash problem even after the > PythonUtils.bash change taken from svn. Yes, so am I. I think it does come down to "reimplement Popen", because the subprocess module has some downright stupid behaviour where multiple threads are concerned. I would really like to avoid that, though. The other alternative is to make GetAvailable.downloadFiles work sequentially; I don't think that's optimal either.
I will take another look at it tomorrow. -Michael
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ gobolinux-devel mailing list gobolinux-devel@lists.gobolinux.org http://lists.gobolinux.org/mailman/listinfo/gobolinux-devel
