[mythtv] [mythtv-commits] mythtv commit: r18379 by awithers

Anduin Withers awithers at anduin.com
Tue Sep 23 07:32:03 UTC 2008


> Err, un-noticed from the point of view of there being
> multiple SIGCHLD handlers? (QProcess and main thread)

Well there is only one handler at a time, the Qt 4 QProcess one tries to chain (calls whichever handler it replaced), the Q3Process one does not. If the Q3Process one did chain it would probably just be worse, our best solution is to just convert the few remaining Q3Process uses.

> I only ask because I have had a few occurances of
> QProcess never terminating, and I am at a loss.

Yeah, should be fixed now, the output you have seen is exactly what made me look at this, you should also see the frontend freeze later as the QProcess is destroyed in the event loop (from the deleteLater) as it waits for it to terminate.

I'll try to get to eliminating the other Q3Processes soon, this one was just the best starting place (it was stopping the frontend build I did this weekend from starting).

On an unrelated note, 18300 introduced use of udevinfo which seems to have been on its way out (in the udev TODO, seems to be gone on Gentoo's udev-126 for example).

--
Anduin Withers 



More information about the mythtv-dev mailing list