Re: Bug in signalmodule.c + FIX (probably a hack :-)
Guido.van.Rossum@cwi.nl
Sat, 17 Sep 1994 10:18:27 +0200
> Guido> Hmm, I'm still awfully afraid of using the POSIX signal
> Guido> interfaces. E.g. the SGI IRIX 5 man page for signals warns:
>
> Guido> The POSIX signal routines (sigaction(2), sigpending(2),
> Guido> sigprocmask(2), sigsuspend(2), sigsetjmp(3)), and the 4.3BSD
> Guido> signal routines (sigvec(3B), signal(3B), sigblock(3B),
> Guido> sigpause(3B), sigsetmask(3B)) must NEVER be used with
> Guido> signal(2) or sigset(2).
>
> Well, it is possible to simulate the BSD signal routines using the
> POSIX ones. That's what I would suggest doing. That's what I did
> when porting GNU Emacs to the DG AViiON. If you have Emacs source
> code lying around take a look in src/sysdep.c.
I think that would break if Python were ever to be embedded in another
application that requires use of the SGI signals, or if it were to
interface to a library that does so. I also want a portable solution.
Anyway, I'd like to wait until someone points out an application that
can't be written using the current signal module. It's not at all
clear to me that we need the full power of Unix or Posix signals,
given Python's different application area -- nor that implementing
Unix or Posix signals is necessarily enough!
--Guido van Rossum, CWI, Amsterdam <mailto:Guido.van.Rossum@cwi.nl>
<http://www.cwi.nl/cwi/people/Guido.van.Rossum.html>