On Sat, Feb 21, 2004 at 10:36:55PM +0100, Julio M. Merino Vidal wrote:
> On Saturday 21 February 2004 21:50, you wrote:
> > On Sat, 21 Feb 2004, Julio M. Merino Vidal wrote:
> > > Any comments about it will be welcome.  Testing could be good too! ;-)
> >
> > Have you tested this?
> 
> Yes.
> 
> > When I tried the idea of using a site-wide config.cache some time ago, it
> > didn't work. IIRC the reason was that later runs detected that some
> > pkg-related setthings had changed ``Woah, but that pkg wasn't installed
> > before! Aborting...'' (Something like that; try & see)
> 
> Yeah, it may bring problems.  This is why the configure script installed by
> the package just checks for things that are unlikely to change over time
> (like functions in libc, system headers, size of standard types, etc).  Of
> course the user might add a check to cache if gtk2 is present, for example,
> but that will surely bring problems when updating and/or removing the package.
> 
> I don't think remember things like "maximum command length", "readdir_r is
> not present", "err.h exists" can bring problems.  And if it does, it's really
> easy to rebuild the cache with up-to-date results.

Sounds great, I look forward to testing this on my old laptop where autoconf
takes 3-4 minutes each time. I wonder if this could be integrated into pkgsrc,
in a simlar manner as security/audit-packages et al...

-- 
This life is a test.  It is only a test.  Had this been an actual life,
you would have received further instructions as to what to do and where
to go.