Re: Debian home page -> Download link broken:
On Sun 11 Jun 2023 at 19:23:02 (-0400), songbird wrote:
> David Wright wrote:
> > songbird wrote:
> ...
> >> except that is a misconception for those who are running
> >> testing. we're not upgrading to a new release.
> >
> > I don't understand. Suite testing was codenamed bookworm until today,
> > and now testing is codenamed trixie. Why is that not a new release?
>
> testing is still testing is it not?
No, it was bookworm, it's now trixie. Different repository
operating under different rules.
> they didn't delete it and then create it again.
Trixie? no. Testing? Yes. AIUI, they delete the symlink that
pointed to bookworm, and recreate it, pointing to trixie.
When we used to connect to the Debian site by ftp, you could
see the symlinks themselves in directory listings. I presume
that it's the use of http access that disguises that fact by
displaying testing and trixie as if they were directories of
equal standing.
> they just created a new directory structure with
> the codename and put links to the packages that were
> the same as testing. it is like taking a snapshot
> but you don't destroy the original directory.
There's no evidence that anything like that happens,
but only symlink shuffling. It would be a great way
of wasting time and resource, and make the repository
pass through inconsistent states, rather than making
the atomic changes of moving syslinks.
> after that point testing and stable diverge as
> changes are made (under the rules and procedures of
> the release team and the various software gatekeepers,
> security team, etc.).
The rules change at the instant the symlinks move,
on Release Day. So stable doesn't diverge, it's
Testing that evolves.
> you could say that as soon as the first change
> happens that trixie is underway and i wouldn't
> argue too much about that at all, but i don't
> consider it anything other than testing and a
> release candidate for trixie. it's not officially
> a stable release for another 24-?? months and as
> such it isn't really named by me, but others can
> consider it what they want. it's only the view
> of the release team that really counts (and their
> established procedures and tools).
Think of it however you want, but the facts are that
the codename is the one point of stability in the
whole Debian edifice. If you track a codename, you
know you are using the same repository for the whole
of its life cycle.
> it's like the chicken and egg problem applied to
> making a cake. at some point you start with an
> empty bowl and then put in ingredients and then at
> some future point (when the baking is done) you
> have a cake (when it is released from the pan or
> even taken from the oven - as some people do eat
> the cake directly from the pan). flour alone
> isn't the cake. so let's just say that testing is
> the bowl which holds the ingredients of the next
> potential stable release, you can call it what you
> want but it isn't an official release until the
> release team kicks it out the door with the
> codename (or not as perhaps some year we run out
> of codenames or Debian stops producing official
> images of any kind or ...).
That's just plain wrong. What was added to bookworm,
the current stable release, on Release Day was a an
official number (12 in this instance). Please stop
trying to sow confusion about codenames.
Cheers,
David.
Reply to: