Le Sun, 8 Jul 2012 17:20:01 +0000 (UTC) Thorsten Glaser <[email protected]> a écrit:
> >BTW, I recently submitted on the dash ML the idea that kill could > >reach "%" jobs independently from the state of the job control flag. > > If you really need that, you’re welcome to submit a patch ☺ If was able to do that, be sure it would have been a patch attached to my first mail... but my C is... yes even worse than that. ☺ What is weird is that it does not work even with $!, which is a PID. This never ends: set -m cat /dev/zero >/dev/null & kill -s "TERM" $! wait (... sorry, I torture you... ;) > >As a matter of fact, if it's not a requirement, POSIX seems to > >encourage it: > > > > The jobs utility is not dependent on the job control option, as are > > the seemingly related bg and fg utilities because jobs is useful for > > Right. The jobs, fg, bg commands *are* available in scripts and > other places where full job control isn’t enabled, unless job > control is not compiled in due to OS bugginess. Is it the case with Linux? > [ signal numbers ] > >Not all of them, indeed... but as far as it makes sense I try to > >stick with POSIX shell for scripting. :) > > Note that, at least for the trap builtin, signal numbers are > not accepted by POSIX, only signal names. (I’m not in the mood > to check it for kill right now as I have a headache.) > > trap 'exit 1' 1 2 3 13 15 # not POSIX! Well, it should works, but only with the POSIX signals: http://pubs.opengroup.org/onlinepubs/009695399/utilities/trap.html I agree though that using signal numbers is not a very good practise... ++ Seb
