Subject: Re: CVS commit: othersrc/bootstrap-pkgsrc
To: None <email@example.com>
From: Jan Schaumann <firstname.lastname@example.org>
Date: 02/27/2003 10:37:00
Julio Merino <email@example.com> wrote:
> On Thu, 27 Feb 2003 10:16:50 -0500
> Jan Schaumann <firstname.lastname@example.org> wrote:
> > Ah, yes, very good. Still, the question remains how far we take this
> > approach. What if the next Linux sticks stuff in /foo/bin -- do we then
> > test those locations, too? Might get messy...
> Yep, you are right... maybe we could genearte defs.Linux.mk on the fly
> or automatically add stuff to mk.conf to determine paths in Linux.
mk/defs.* should, IMHO, not be generated on the fly. In it, we should
supply default values that hold for most versions of that operating
system. If the version at hand deviates, the user can (and should) set
the paths by hand in /etc/mk.conf.
What we _can_ do, is check during bootstrap-time what's going on and if
it deviates from the norm:
check if binaries are found in a different location
if yes, add FOO=/found/bin/foo to mk.conf.example
if not, or binaries are not found, add FOO="NOT FOUND - FILL IN" to
/etc/mk.conf and set a flag.
At the end of bootstrapping, if some binaries are not found display a
warning to the user.
This one's tricky. You have to use imaginary numbers, like eleventeen...