Re: [PATCH 15/15] Hurd signals: Use POSIX sigcodes

[Samuel Thibault]

Jérémie Koenig, le Wed 29 Jun 2011 23:17:49 +0000, a écrit :
> On Wed, Jun 29, 2011 at 10:59 PM, Samuel Thibault
> <samuel.thibault@gnu.org> wrote:
> > Jérémie Koenig, le Wed 29 Jun 2011 22:55:48 +0000, a écrit :
> >> The old !SA_SIGINFO case was a non-standard (BSD-based?) extension. Is
> >> there really anything using it?
> >
> > You can never know.  And thus probably have to keep it.
> 
> Would mapping the old values to the new ones in the trampoline code be
> an acceptable compromise?

It is probably better than duplicating the mapping everywhere the values
are generated, indeed

> Should I rename the constants so that programs using them would fail
> to compile, and therefore be noticed and, hopefully, updated to the
> more portable SA_SIGINFO interface?

I don't think so.  We don't want to unnecessarily break something that
can still work.

> Would it still be ok to introduce POSIX sigcodes where 0 was used before?

I'm not sure about it. I'd say when !SA_SIGINFO, better reset it to 0.

Samuel