Re: testing, unstable, and dependencies
On Thu, 6 Feb 2003, Marcelo E. Magallon wrote:
> On Thu, Feb 06, 2003 at 11:14:07AM +0100, guenter geiger wrote:
> > But generally it would be better attacked if developers would care
> > for flexibility. Packages should at least be compilable on testing
> > and unstable whenever possible. I see tendencies going in the other
> > direction, and I consider this as really bad.
> Build-Depends: foo (>> 4)
> And testing has foo 3.
> I guess you are talking about things like debhelper... funny, me too.
> Are you proposing that developers should hold back/wait until the
> version with the required features hits testing? I see only limited
Not really. It is not only debhelper that creates the problems.
If I have a package that build depends on libfoo3-dev for example,
but it is known to build with libfoo2-dev as well (the configure
script in most software detects this and acts accordingly).
Most of the time, only libfoo3-dev will be in the build depends,
taking away all the flexibilty that the package has, and making
it impossible to build it on testing (which only has libfoo2-dev)
Most of the time this scenario can be extended to woody.
If the package maintainer takes this into account, we could add
several new applications to woody, for example, by just autobuilding
them. This would make the stable distribution much more up to date,
and give the users the advantage of having a stable system with the
> Packages in testing should be buildable with stuff in testing, but I
> don't think that's what you meant.
No, what I am really talking about is creating the awareness of building
flexible packages. This flexibility is common in the software we package,
we should not restrict it.
> PS: Yes, I know, there is a debhelper (>> 4) in testing. That's not
> the point.
> To UNSUBSCRIBE, email to email@example.com
> with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org