David DeSimone writes:
> Lars Hecking <[EMAIL PROTECTED]> wrote:
> >
> > Is this the standard, bundled compiler that comes with HP-UX 10.20
> > (/usr/bin/cc -> /usr/ccs/bin/cc)? mutt requires an ANSI C compiler.
>
> If you read a few lines higher up, you would see the use of the "cc -Ae"
> switch, and thus know that it is indeed the ANSI compiler. He would
> never have got that far with just the bundled compiler.
Hhm, you're right. Should've wondered then why the error didn't occur
earlier ...
> > What HP ships for a compiler these days can only be considered an
> > insult. You're infinitely better off installing gcc or egcs.
>
> I am sorry that you feel insulted by HP's failure to give away software.
> However, since I routinely build Mutt using the HP ANSI compiler, I
> don't agree that gcc is required.
I don't care whether they ship an ANSI compiler or not. There's always
a way to get a compiler, even to buy it. If it's a pre-ANSI compiler
(almost ten years after C became an ISO standard, I might add), fine,
SunOS cc is K&R too.
But what really p*sses me off is that cc on HP-UX 10.20 is crippled.
You cannot use -g. Even this was possible with HP-UX 9.x.
> > Also, HP's make exhibits some strange behaviour, ie. rebuilding files
> > even when it should not.
>
> I have never experienced this; I'd be interested in a demonstration of it.
The m4/Makefile.am.in trick found in some distributions. For some reason,
HP's make will always try to rebuild Makefile.am. This fails if $< is used,
because $< can only be used in inference rules.
Check out the end of the first attachment of the mail that started this
thread.