Hi Adam,
thanks for your quick reply.
On Mon, Aug 04, 2014 at 09:52:35AM +0100, Adam D. Barratt wrote:
> On 2014-08-04 9:19, Andreas B. Mundt wrote:
[...]
> >The package did not make it for the wheezy freeze and I provided it
>
> Many packages didn't make the freeze. That doesn't mean we should add them
> all to stable.
I absolutely agree; this was not meant as an argument.
> Looking at the package's upload history, the first upload reached the
> archive a few weeks before the wheezy release, with the first upload to
> unstable being on the day of the release. That's really not the same as "did
> not make it for the wheezy freeze", which had happened almost a year
> earlier.
Yes. But if you look at the complete history in git, you'll see that
there has been done a lot of work much earlier. I did not upload that
work because wheezy was still a moving target and the package would
not have worked without issues. The problem is, that it builds on the
finished distribution and therefore has to wait for all ingredients
being ready. (Comparable to the debian-edu-config package).
> >via backports after the release. However, with further improvements,
> >it got less and less justified to use backports, as the wheezy package
> >is not a strict 'backport' of the version in jessie, which is of
> >course targeted at jessie.
>
> Given it's a configuration package, the diffstat of
>
> 66 files changed, 390 insertions(+), 245 deletions(-)
>
> doesn't look crazy for a backport. Reading the unstable changelog, I have to
> admit that I'm not sure why you're maintaining the backport as a separate
> branch and wonder how big the diff would be if one took the current package
> from testing and then applied the changes required for wheezy.
OK, in my experience the setup is very fragile and it's hard to
oversee and test all modifications. Therefore, I chose the path to
only cherry-pick stuff from testing and keep the well-tested core from
the wheezy-version. (Comparable to the stable distribution where
usually only bugs are fixed.)
Concerning backports, I already had a discussion [1] about the issue;
we agreed that it's better to use a point release for debian-lan.
So I find myself caught between two stools now.
[...]
> It's not a question of "posing a risk", but rather what compelling reason
> there is to break the general policy of adding new packages to stable -
> particularly a package that was barely in the archive when stable released.
I hope I could explain the reason why the upload happened late above.
The situation in jessie looks slightly better (apart from the fact that
the package is already in the archive now) and I hope all needed stuff is
available before the freeze this time. However, it would be nice to
have the package in wheezy too.
Best regards,
Andi
[1] Thread:
<URL:https://lists.debian.org/debian-backports/2014/05/msg00049.html>
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
