Re: testing d-i Release Candidate 1 and more release adjustments
On Thu, Nov 16, 2006 at 10:00:48AM -0800, Thomas Bushnell BSG wrote:
> > Release goals and etch-ignores
> > ==============================
> > After a mailing list discussion about release-criticality and policy, we've
> > decided to include "fixing bashisms" as a release goal. This means that if
> > packages have bugs that render them unusable with other implementations of
> > /bin/sh, such as dash, we encourage maintainers to consider NMUing those
> > packages under the same rules that govern release-critical bugfixes.
> > This does not mean that such bugfixes *are* release-critical (they're
> > severity: important unless there is another reason they should be release
> > critical), so such fixes will be included in Etch on a best-effort basis.
> I'm hoping for clarity here, and I can't quite tell what's the policy
> It sounds like you are saying that the issue is not whether something is
> "Posix-compatible" or a "Posix-feature", but rather the practical
> question "does it break with an different shell?" This is, it seems to
> me, a very good criterion.
> However, I still wonder: which shells? Your example is dash. But I
> find meeting a moving target hard; can we have a list of exactly which
> shells must work?
dash should work; posh is not required to work because nobody uses it as a
Those are the only three shells I know about that are even an issue, so I
can't specify beyond that -- and I'm not going to offer anything that could
be construed as a final word on the question of which shells Debian wants to
support while there's still an ongoing policy discussion on this point. For
now, the bottom line is that if someone has a shell from a Debian package
which implements POSIX sh installed as /bin/sh because they want to *use*
it, and something breaks, it's covered under this NMU policy.
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
 this excludes posh -- it's been pointed out multiple times that posh is
a better test suite than dash but a worse shell, and no one has ever
disputed this, and the discussion on -policy doesn't lead me to think
the policy maintainers are going to require posh compatibility in the