Re: the exquisite pain of signal handling in shell scripts

[G. Branden Robinson]

At 2020-05-02T10:07:20+0100, Ralph Corderoy wrote:
> Hi Branden,
> 
> > It seems like the main difference between our signal handlers is that
> > you do not reset the handler, and rely on exiting with a status
> > greater than 127 to notify the shell that the process was killed with
> > a signal.  I'm not sure that is reliable, or supposed to be.
> 
> Do you mean distinguish these two cases from a shell script using `$?'?
> 
>     $ sh -c 'kill $$'; echo $?
>     Terminated
>     143
>     $ sh -c 'exit 143'; echo $?
>     143
>     $

Yes.  I suppose, without checking any Unix's implementation details,
that these may differ because the fact that a process has been signaled
could be in a part of the wait-status (which is an int, wider than a
byte) that is not exposed as part of the exit integer.

I suppose that in practice it may be that bit 7 of wstatus is typically
the bit returned by the WIFSIGNALED() macro, but I wouldn't want to
stake my life on it.

The more of the Austin Group's time I see taken up with bizarre
implementation differences in the shell, the more paranoid and less
certain I get about what is reliable.

Regards,
Branden

signature.asc
Description: PGP signature