Subject: Re: CVS commit: pkgsrc/devel/gmake
To: None <tech-pkg@NetBSD.org>
From: Alistair Crooks <email@example.com>
Date: 02/09/2004 09:08:35
Content-Type: text/plain; charset=us-ascii
On Mon, Feb 09, 2004 at 12:40:59AM +0100, Thomas Klausner wrote:
> On Sun, Feb 08, 2004 at 10:21:42PM +0000, Alistair Crooks wrote:
> > As long as it's well-enough documented (in
> > pkgsrc/devel/gmake/Makefile), I'd go along with it, but... why bother
> > complicating things unnecessarily? These circular dependencies are a
> > right PITA to find and squish, and none of this does anything to help
> > that.
> I'm not sure I understand -- I thought circular dependencies are
> quite easy to find.
Maybe direct ones are. But indirect ones?
> Also, as Michal Pasternak wrote, bzip2 seems to compile fine
> with our native make and the one from bootstrap-pkgsrc.
I'm sure it does. But the bzip2 package might include info
documentation, and this require texinfo to format the documentation.=20
And texinfo might require gmake for its own Makefiles.
> I don't think it's a good idea to limit ourselves because of
> possible problems that could appear in the future, if it is
> not very probable that they will really appear, and that doesn't
> seem to be case here.
We are hardly limiting ourselves by using a distfile that is less
than 300K larger than another distfile. But we have also been, by
nature, conservative in the things we do in pkgsrc, and going out
and inviting trouble has never been one of the things that I do
on behalf of all the pkgsrc users out there.
Like I said, as long as it is documented well in gmake's Makefile
what the situation is, I can live with it, but I much prefer that
it hadn't happened.
Alistair Crooks <firstname.lastname@example.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (NetBSD)
-----END PGP SIGNATURE-----