On Sat, Feb 08, 2003 at 10:44:28PM +0100, Ronald Bultje wrote: > Hi Brian,
Hi Ronald, > On Fri, 2003-02-07 at 14:17, [EMAIL PROTECTED] wrote: > > Should be. But it might lag while it waits for disk i/o or something. > > Could be... But this never happens in practice actually. Something happens. Every now and then (before I commented the kill() out entirely), I got a "Killed" in my shell script (that starts glav) and the part of the script after the glav was not executed, so something was lagging. > Anyway, I see > your point and it's valid. That's good. > The idea is that there is no issue in lavplay that triggers this. But if glav goes on doing the kill -9 silently, if and when a problem does come up in lavplay that this kill() actually deals with (hacks around IMHO), nobody (using glav) will know about it. > > c. alternatively change the first argument of kill() to -1 or > > Erm... Bah! Wrong one. I meant change the first argument of kill() to -n where n is the pgrp id. Messy. > I personally would love to try this out, but I don't think end users > will appreciate it. Ohhhh. Bit of a BOFH too huh? :-) > I'm all in for changing the first argument to pid. OK. I can live with that. > Changing to SIGINT > isn't actually that much of a good idea, since lavplay grabs SIGINT to > close down. Right. That is the point of using SIGINT. To let lavplay know that things are going down and give it it's chance to clean up. > SDL actually does the same thing which even complicates > things more. If something went wrong (and that's what happens if this > one is called), you don't actually want the application to grab the > signal and exit gracefully because that's probably the whole reason why > this function was called in the first place - it will just hang. That is what would need to be fixed if this were the case then. > It > should die. So I'd suggest maintaining SIGTERM, but using kill(pid, 9) > instead of kill(0, 9). 9 is (usually and in context of glav/lavplay) SIGKILL (and in fact the macro SIGKILL should be used in glav rather than the digit 9 for portability). SIGTERM is (usually) 15. > Patch committed to CVS. Excellent! Thanx. At least any shells/scripts that call glav won't get mysteriously slaughtered now. :-) b. -- Brian J. Murrell
msg00670/pgp00000.pgp
Description: PGP signature
