Re: autoconf 2.49c AC_CACHE_CHECK failure

[Nicolas Joly]

On Tue, Jan 30, 2001 at 02:10:19PM +0100, Akim Demaille wrote:
> >>>>> "Alexandre" == Alexandre Oliva <address@hidden> writes:
> 
> Alexandre> On Jan 25, 2001, Akim Demaille <address@hidden> wrote:
> >> Should we (i) make sure not to use config.site in the test suite,
> >> or (ii) have this test grep out this message?
> 
> Alexandre> (ii)
> 
> Since now the `loading config.site' is using AC_MSG_NOTICE which is
> subject to --quietism, this no longer can happen.

Good.

But i don't understand something in AC_CACHE_CHECK test


## -------------- ##
## AC_CACHE_CHECK ##
## -------------- ##

# make sure AC_CACH_CHECK is silent with -q
               ^^^^
           Small typo here

AT_SETUP([AC_CACHE_CHECK])

AT_DATA([configure.ac],
[[AC_INIT
AC_CACHE_CHECK([for nothing],[ac_nothing],[ac_nothing=found])
]])

AT_CHECK_AUTOCONF
AT_CHECK([./configure -q], 0)
AT_CHECK_CONFIGURE([], 0, [checking for nothing... found
])

Why do we need both AT_CHECK and AT_CHECK_CONFIGURE calls here ?

AT_CHECK_CONFIGURE([-q], 0, [])

should be enough, no ?

-- 
Nicolas Joly

Informatique Scientifique
Institut Pasteur, Paris.