Subject: Re: problems with choosing a Berkeley DB
To: Todd Vierling <email@example.com>
From: Julio M. Merino Vidal <firstname.lastname@example.org>
Date: 01/05/2005 21:43:29
On Wed, 5 Jan 2005 14:59:25 -0500 (EST)
Todd Vierling <email@example.com> wrote:
> ...which may even be db1 on some platforms, depending on pkgsrc leadership's
> opinion. That still means that PKGREVISION bumps are needed whenever db4
> changes in this scenario, of course.
Yeah, well. What I mean is that, regardless what the default is, we must
have a default and stick to it. If the user is not happy and want to
change it, fine, but then he can't expect pre-built binary packages work
fine for him.
> This also happens for pgsql and other packages, and it really is annoying
> because of the use-whatever-is-installed scheme. Perhaps this is an
> indicator that we should change the policy to make these bl3 files use
> whatever the user prefers, or if the preference is not supported, default to
> a fixed version based on the package acceptable list, never checking which
> version is installed already.
Agreed. I'm all for consistency (one of the strong points of pkgsrc I'd
say, compared to other systems).
Just as a note: I'm only a bit familiar with python, but since I set
"PYTHON_VERSION_DEFAULT= 23pth" in my mk.conf, I only have _one_
version of it installed. IIRC, I used to have several in the past for no
These kind of settings should be defaults (maybe differing depending on
the platform, I don't know), as you outlined in the previous paragraph.
> I should add that removing the use-whatever-is-installed logic to achieve
> this consistency will help my derecursion effort in spades. That's because
> it means far less build-time pkg_info and .if exists() checks in the
> Makefile goop. :)
Julio M. Merino Vidal <firstname.lastname@example.org>
The NetBSD Project - http://www.NetBSD.org/