This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch, toplevel] --enable-multilib*S* is an easy typo to recogni


aoliva@redhat.com (Alexandre Oliva)  wrote on 15.02.04 in <orfzdbok6f.fsf@livre.redhat.lsd.ic.unicamp.br>:

> On Feb 15, 2004, Phil Edwards <phil@jaj.com> wrote:
>
> > On Sun, Feb 15, 2004 at 06:25:12PM -0300, Alexandre Oliva wrote:
>
> >> Why should an unrelated package conform to GCC's spelling?
>
> > Because it's going in our tree?
>
> That's one of the scenarios.  The scenario described in the autoconf
> docs is more general, and I actually referenced that as well: consider
> that I create my own top-level configure with which I intend to build
> not only GCC, but also other packages.  autoconf is designed so as to
> enable this scenario to work, as long as you don't use configure
> options that have different meanings for different packages in your
> tree.

That's fine for autoconf, but seems rather irrelevant to what is in the  
gcc/src toplevel.

Actually, IMO it's not quite so fine for autoconf, either - the right  
thing to do would have been a way for subconfigures to communicate the  
switches they recognize to parent configures (for example by having a  
standard file with them, or a standard command line argument to echo them,  
or whatever else would seem useful to autoconf developers), not to close  
the eyes and accept all typoes.

MfG Kai


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]