Re: confusion
On Sat, Dec 02, 2006 at 10:56:04PM -0800, Thomas Bushnell BSG wrote:
> On Sat, 2006-12-02 at 22:45 -0800, Russ Allbery wrote:
> > Thomas Bushnell BSG <tb@becket.net> writes:
> > > It would be better to have a halfway modern lilypond on 32-bit archs and
> > > nothing at all on 64-bit archs, than to have a medieval lilypond of
> > > them.
> > So, if that's the case, wouldn't the next step be to file a bug with
> > ftpmaster requesting the removal of the 64-bit builds of the old version
> > of lilypond from testing, after which the new lilypond would propagate to
> > testing automatically since it would no longer be breaking those
> > architectures?
> You're missing that better than this is to have a version skew between
> the 32-bit archs and the 64-bit archs.
I disagree. Our users expect that a Debian release will provide a
consistent platform across architectures, with the exception of packages
that aren't ported to one architecture or another. Shipping multiple
versions of a package across different architectures violates this
expectation, bloats the archive and complicates security support (if two
different source packages are used across architectures, that's twice as
many source updates that need to be made for a security fix; if there is
*not* a separate source package for the out-of-date archs, no security
support is possible for those archs).
> > > Better still would be to allow a version skew as I described, hinting
> > > lilypond 2.8 for the 32 bit archs. This involves no regressions. It is
> > > certainly not the ideal thing, but, news flash: the ideal thing is not
> > > likely to be possible.
> > You don't believe it's possible to port guile to 64-bit architectures in
> > the etch release timeframe, but the new lilypond absolutely depends on the
> > new version of guile?
> No, I didn't say that. I don't have control over guile-1.8, so I can't
> make any statements about how fast anything there will happen. However,
> Rob Browning did just upload a new version of guile-1.8 which seems to
> be successfully building on the 64-bit archs. So the only question now
> is whether it will be allowed in before freeze. I hope so!
Yes, it was always my intention to make allowances for a lilypond 2.8 update
after the freeze given that the version in unstable is a candidate for
testing now except for the guile-1.8 64-bit issues.
BTW, guile-1.8 seems to still have an ia64-specific failure which is holding
things up.
I still recommend asking for the out-of-date ia64 binary to be removed from
unstable pending resolution of the guile-1.8 bug, as I had said before. If
you elect to reintroduce lilypond 2.6 for ia64 as a separate source package,
that's ultimately between you and the ftp team to decide; but I don't think
it's warranted to allow an ia64 lilypond package into etch that doesn't
correspond to the current sources.
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
vorlon@debian.org http://www.debian.org/
Reply to: