Re: [Volker Borchert <address@hidden>] Re: Announcing Autoconf 2.49d

[Nicolas Joly]

On Wed, Mar 21, 2001 at 02:47:47PM +0100, Akim Demaille wrote:
> 
> Thanks Volker, but please, keep this public.
> 
> People, it seems that AC_FUNC_VFORK can produce SEGVable programs.
> IMNSHO, the test suite must not be robust to `core' and it's good that
> it is caught.  The alternative, as I see it, is
> 
> (i) AC_TRY_RUN should remove cores
> 
> (ii) AC_TRY_RUN doesn't, it is the corresponding macros which should
>      handle their specials cases.  Maybe AC_FUNC_VFORK is really doing
>      bad things, but one can imagine tests which will try to have the
>      test dump core.

I got similar behaviour on my Tru64 v5.1 station, but with the new
AC_FUNC_STRERROR_R. I had a look in macro definition and find this
comment:

      # strerror_r seems not to work, but now we have to choose between
      # systems that have relatively inaccessible declarations for the
      # function.  BeOS and DEC UNIX 4.0 fall in this category, but the
      # former has a strerror_r that returns char*, while the latter
      # has a strerror_r that returns `int'.
      # This test should segfault on the DEC system.

-- 
Nicolas Joly

Informatique Scientifique
Institut Pasteur, Paris.