On Thu, 24 Jun 2010 23:59:23 +0200, Jeroen van Meeuwen <kanarip(a)kanarip.com> wrote:
Scott Henson wrote:
> At one point in the 1.x series we had limited support for Debian based
> distros, but no one was using/testing that support, so with 2.0, it was
> removed. If you are willing to put the work in, we would love to
> support Debian based distros. I'm not entirely sure how much work it
> would be to test it though.
>
I know of at least 3 parties that would be interested, myself not included.
Then there's some parties interested in (Open)SUSE, and then some in Windows.
I may have mentioned this, but we also have someone working on vmware
and a bsd of some sort. I would love to see cobbler support more
distros from a deployment stand point.
> The first step would be to package cobbler on Debian. This
should be a
> lot easier now that we have cleaned up the packaging on the Red Hat
> side. The setup.py should put things together in such a away that the
> dpkg should be pretty straight forward. If you want to try that and let
> us know how it works, that would be great.
>
Actually, the first step would be to undo the commenting out of lines of code.
Whether one deploys Cobbler on Debian, or deploys Debian with Cobbler are two
different things.
I mentioned install because right now we are working on cleaning up the
way cobbler installs. I am not opposed to starting elsewhere, that was
just my suggestion.
As for uncommenting the code, if you can test that the current code
works, I would be fine accepting a patch to do that. I just can't
verify that the code works, so I can't really uncomment it.
FWIW, one important detail that I'm trying to tackle is
accounting for the
major difference between the layout of release media (.iso) and the contents
of a mirror.
What I would like to do is move all of the breed specific parts into
modules. That should make the import (and other places) code cleaner,
and make it easy to see what code needs to be written to add a new
breed.
The one part I am not clear on is how to detect which breed. I'm
thinking a file based regular expression of some kind. Something that
can be passed to fnmatch.filter or some such.
For the redhat breed, I would say look for RPM-GPG-*. That seems to
mark the beginning of all the Fedoras and Red Hat ELs. Is there
something similar for other breeds? Does another breed also have
RPM-GPG-* files (looking SUSE)?
For clarity sake, a breed is a group of distros that can all be treated
the same by import. For instance, Fedora and Red Hat Enterprise Linux
are both in the redhat breed. I would assume that Debian and Ubuntu
would both be in the debian breed. So on and so forth.
--
Scott Henson
Red Hat CIS Operator
WVU Alum BSAE/BSME