On Wed, 2012-11-28 at 04:25 -0500, Zdenek Pavlas wrote: > > One downside to this is that you need to deal with with all the > > possible deadlock issues. Eg. you now have: > > .. > > ...and if we ever block on one side of a worker, while it's blocked > > on the other side we deadlock. > > I've cheated a bit there, parent reads from worker's stdout only. > Stderr is not redirected, it goes directly to console/whatever, > so worker never blocks on stderr.write()..
Ahh, that explains why I couldn't see where you were reading from that then :). On the other hand, I'm pretty sure that technically breaks the API as there'll be no way to create a graphical createrepo and capture the stderr logging now? > I still don't get why some messages are printed to stdout and some > to stderr.. do we really need to log both? It's all to do with their logging level etc., I thought ... but for the workers I'd guess everything they print (apart from the actual data) could/should be an error. > Could be fixed either by select()ing, or updating worker.py > to always print errors on stdout (would prefer that) That we can't do due to things like ./run_cmd > /dev/null ... which should print errors. _______________________________________________ Yum-devel mailing list [email protected] http://lists.baseurl.org/mailman/listinfo/yum-devel
