On Sat, 2009-01-31 at 02:25 +0000, Ben Hutchings wrote: > > Just as a remark regarding your intended fix: I haven't yet seen the > > place where xine does a wait() or waitpid() for those forked childs to > > clean up the zombies. If there is a place, ok, I just havn't seen it > > yet. > > I added signal(SIGCHLD, SIG_IGN) which should make that unnecessary. > > Also, the change I actually made to xine-ui *does* include a loop to > close fds 3-255 (sadly there is no easy way to find out what the last > open fd is). So unless xine is opening a *huge* number of files I can't > understand this bug.
OK, now I see what's happened. The correct fix for #374644 is in sid, but not in lenny. Because the new version of xine-ui in sid changed library dependencies, it could not propagate to lenny and I had to upload separately to testing-proposed-updates. I mistakenly uploaded an earlier version of the changes which has this bug. I will upload a new version to t-p-u. Ben.
signature.asc
Description: This is a digitally signed message part
