Re: [PATCH] run-command: detect finished children by closed pipe rather than waitpid

From: Stefan Beller <sbeller@google.com>
Date: Nov 11, 2015, 20:37 UTC
Message-ID: <CAGZ79kaZ0W5q8=iowbSTJY_mDNNYc6qdTszcWDDDCtYfZyeK_Q@mail.gmail.com>
In-Reply-To: <563DBDDA.2000106@kdbg.org>

including the list and all others this time.

Show 8 quoted lines

>>                         if (code < 0) {
>>                                 pp->shutdown = 1;
>> -                               kill_children(pp, SIGTERM);
>> +                               kill_children(pp, -code);
>
>
> I'll see what this means for our kill emulation on Windows. Currently, we
> handle only SIGTERM.

So currently we only pass in SIGTERM from the callers, and I certainly only intend to use that signal. I just thought special casing the SIGTERM signal would do no good in terms of design here.

So maybe that was not the right thought and we do have to special case SIGTERM here?

I worked with another large proprietary parallel program (spawning child processes) before. If you pressed CTRL-C, it would omit a message like:

    Sending no further input to the children

If you press Ctrl-C again, it would print:

    Sending SIGTERM to children, to shut them down gracefully

If you pressed once again, you'll get a

    User seems to be angry, shutting down hard, sending SIGKILL
    $ # that was fast now.

I guess we don't want to play around like that, but if the parent process gets a SIGTERM, we relay that and other signals we ignore for now?