Re: Two suggestions for woody release
Juha Jäykkä wrote:
> > > to fix them later and upload them. Would it be possible to create an
> > > "official"(1) archive of these dropped packages?
> > We still have them in sid. Isn't that enough ?
> To make it short, no.
> To make it longer, yes we do, but possibly linked against newer
> libraries or otherwise uninstallable into a stable system. That is the
> whole point. Anyone can install XFree 4.1.0 from sid to potato, but that
> means upgrading about everything else, as well. It is no problem if you
> want to go to sid but I imagine there are many people who want to use
> stable (whether it be potato or, in the future, woody) for many good
> reasons but still want to install something not included in woody because
> it happened to have an RC bug at release time.
Speaking as a user, I do think Juha's put his finger on something here,
but it goes beyond release-critical bugs. The useability of stable has
long been hampered by the fact that AFAIK there's no ((semi-?)official)
way for a Debian developer to:
* Add a package (possibly held up prior by an RC bug)
* Apply a (possibly trivial) bug fix (non-security-related) to a package
and make the update available to users of stable. And library
dependencies *are* the problem with simply grabbing an updated package
from testing or unstable.
The idea of grabbing the source from testing and compiling is one I've
used myself, but it's inconvenient and time-consuming, and doesn't
always work. For instance, I've been trying to use the gs from testing,
but it seems to have some serious problems which aren't fixed by
I certainly don't think Debian developers should take time to work on
this issue now, but it might be something to look at after the next
release. An improvement would be if there were just some links from the
main or distribution page to unofficial debs that are supposed to work
with stable and its libraries, unless that's already there and I've
missed it. (I take it for granted that there are no guarantees on the
useability of such packages).
Maybe this would encourage a diversion of resources, but would it really
be more than the unofficial debs that are done already?