On Wed, Aug 31, 2005 at 07:57:36AM -0700, Ben Pfaff wrote:
> demon ~ % uname -a
> IRIX64 demon 6.5 01062343 IP35
> demon ~ % cat tst.c
> #ifdef FOO
> # error "FOO is defined"
> #endif
> int main(void)
> {
> return 0;
> }
> demon ~ % cc -DFOO tst.c && echo OK
> cc-1035 cc: WARNING File = tst.c, Line = 2
> #error directive: "FOO is defined"
>
> # error "FOO is defined"
> ^
>
> OK
> demon ~ %
In Autoconf itself, we often eschew #error in favor of `choke me' inside the
#ifdef, which yields a syntactically invalid compilation unit. Autoconf users
can do the same in their own tests like this one, and that should avoid the
problem with this compiler.
Perhaps you could post the version string from that compiler, so we can document
this problem with a specific reference?
> i.e. could the implementation decide that the translation is not
> successful because of the above warning? Anyway, one of the goals
> of the configure script is to detect when things are wrong on the
> current platform. So, I think that autoconf should cope with this
> problem.
It would be nice to handle this automatically, but I doubt we could do so
without breaking something elsewhere. Ralf has more experience with matters
like this one, and he mentioned some of the major obstacles to a reliable
implementation.
_______________________________________________
Autoconf mailing list
Autoconf@xxxxxxx
http://lists.gnu.org/mailman/listinfo/autoconf
