Re: RFC: Autom4te: defining M4 sets

[Tim Van Holder]

On 27 Aug 2001 13:08:29 +0200, Akim Demaille wrote:
> 
> I don't understand well how it would restrict anything.  It does
> promote some naming conventions, which is indeed something I'm pushing
> for, but --mode will always be stronger than patterns.

OK - guess I was overreacting.

> | > arguments: --include /usr/local/share/autoconf
> | 
> | Hmm - that means you can't really have two concurrent versions of
> | Autoconf, with different prefixes.  
> 
> I don't understand what problem you are referring to here, as two
> different Autoconves means two autom4te hence twice that file.

For the globally installed one autom4te.cfg, that's correct - but with
~/.autom4te.cfg that is not the case.

> | Maybe this should always get
> | autom4te's $prefix (or $datadir) prepended?
> | Which would probably mean using
> | 
> | includedir: share/autoconf
> | or
> | includedir: autoconf
> 
> Hm...  Why not.
> 
> But I think you are contradicting yourself here, as having the full
> path would allow you to have ~/.autom4te defining the modes
> autoconf-2.13 and autoconf-2.50 and let the magic happen.

I suppose, but that would require using '--mode autoconf-2.13', while
(automake-)generated rules would probably just use '--mode autoconf'.