On Wed, Apr 18, 2001 at 05:52:26PM +0200, Marcus Brinkmann wrote:
> On Wed, Apr 18, 2001 at 02:09:32AM -0500, Joey Hess wrote:
> > 2. in the main archive
> > Well, er, define "main".
> "unstable/main" of some architecture is the content of
> ROOT/dists/unstable/main/binary-ARCH/Packages
That's not particularly true, and is certainly the wrong way around.
dists/*/main/binary-*/Packages{,.gz,.bz2} is the list of packages that are
available to be installed on a Debian system.
Once upon a time, that used to be all of main, and all that Debian provided.
We're now providing (or soon will be, depending on your perspective)
a really minimal package based distribution for doing installs of the
real Debian system. These packages aren't suitable for installation into
a real Debian system, and they're not meant to be; so putting them into
main/binary-*/Packages* is just asking for users to break their systems
in random ways. This has been discussed on this list in the past.
If you want to get all the packages that can be built from main/sources
for an architecture, you have to cat these Packages files together, and
cope with the fact that some of them may be udebs. If we ever introduce
another distribution that's completely disjoint from the above two,
you might have to cat a third Packages file together. *shrug*
> I want something that can determine the full list of Packages and Sources
> record entries from DIST/ARCH pairs that works for past, present and future
> Debian archives. Note that "past" and "future" in this sentence prevents the
> simple "add debian-installer if DIST is unstable/main".
It also fails to work with pre-hamm versions of the distribution.
> People often say that Debian strives fro technical excellence. I can't see
> technical excellence in the introduction of the debian-installer
> subdirectory. I raised this issue because I think we can do better.
Sure we can. But the problem here isn't that the structure of the archive
is wrong, it's that the autobuilders haven't been particularly well
organised and consistent across architectures. Ryan Murray's working
on fixing that, and most architectures (alpha, arm, hppa, i386, ia64,
mips, mipsel, and sparc) are already using the centralised wanna-build
setup on ftp-master. If you'd like to get hurd-i386 working with that
too, I'm sure Ryan would be most happy to setup the extra group and so
forth for you. If you want to keep maintaining hurd separately, you'll
have to adapt yourself to the archive rather than it getting done for
you along with all the other architectures.
Cheers,
aj
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.
``_Any_ increase in interface difficulty, in exchange for a benefit you
do not understand, cannot perceive, or don't care about, is too much.''
-- John S. Novak, III (The Humblest Man on the Net)
Attachment:
pgpMmDwRhBZz0.pgp
Description: PGP signature