On Thu, 2004-07-29 at 21:37, Andrew Bennetts wrote:
On Thu, Jul 29, 2004 at 09:21:41PM -0700, Dave Peticolas wrote:
On Thu, 2004-07-29 at 20:52, Andrew Bennetts wrote:
On Thu, Jul 29, 2004 at 08:39:14PM -0700, Dave Peticolas wrote: [...]
This is on a Redhat 7.2 system using python2.3 and a newish version of twisted. I don't know why the process seems to get lost, but it would be nice if Twisted would at least notice the ECHILD and signal process termination (or lost, or something).
What version of Twisted, exactly? There's been some changes to the reaping code in recent releases (e.g. between 1.2.0 and 1.3.0), which might be relevant.
I believe it is 1.2.0.
Ok, then I think it's likely that 1.3.0 fixes your bug. 1.2.0 had a nasty hack in its child reaping that went away in 1.3.0.
Yes, I see the diff. That seems to have fixed it, thanks for the info!
dave