Bug#683142: unblock: bdii/5.2.12-1
On 2012-07-31 10:49, Mattias Ellert wrote:
> sön 2012-07-29 klockan 12:46 +0200 skrev Niels Thykier:
>> On 2012-07-29 06:47, Mattias Ellert wrote:
>>> Package: release.debian.org
>>> Severity: normal
>>> User: firstname.lastname@example.org
>>> Usertags: freeze-exception
>>> unblock bdii/5.2.12-1
>>> The bdii package was removed from testing due to an RC bug, together
>>> with the packages that depends on it. The 5.2.12-1 update fixes the RC
>>> bug (bug #663444). I would like to request a freeze exception for this
>>> update to allow the bdii package and the packages depending on it to be
>>> part of the release.
>> Why did you include a new upstream release in this? It makes it harder
>> for us to review and reduces the chance for you to get the unblock?
>> Does this upstream release have important bug fixes, if so what are they?
> I had been preparing an update to a new upstream release for a long time
> before finally making the upload. On several occasions I have completed
> a potential update and then looked at the BTS and thought that I should
> fix that RC bug before doing the upload. Since fixing the RC bug was not
> trivial this always ment that I held off doing the upload. I finally did
> fix the RC bug. The fixed package compared to the last package I
> prepared and did not upload was really just fixing the RC bug.
My problem is; this RC bug was reported in March. It receives no
(public) reply from you until you close it with the 5.2.12-1.
While you do have a point that conffile handling is not trivial, the
reporter gave you a link to http://wiki.debian.org/DpkgConffileHandling
to assist you. That page has both the "old school" handling and plenty
of references to the dpkg-maintscript-helper tool. Failing that, an
email to d-mentors could easily have given you the answer within a day.
Sorry, but I do not buy the 4 month delay for fixing this.
> The changes in the package between the previous upload and the new one
> are very minor. It is true that if you list the files changed the list
> is not short, but most of the changed files are in the debian directory.
> These changes are there to do the fix of the RC bug, fix some lintian
> warnings and update the copyright file to the new recommended format.
> The changes to the patches are just dropping the parts of the patches
> that were accepted upstream and rebasing the remaining parts.
These changes I get and I can (mostly) ignore. My primary concern is
actually the upstream changes. Admittedly I am not too pleased with the
dh_pysupport -> dh_python...
> For the changes to the upstream itself, i.e. the files outside the
> debian directory. These are mainly changes to the default configuration
> to reduce the memory consumption and to add support for IPv6.
Those changes sound nice to have, especially lack of IPv6 is in fact
starting to be a bit sad. The problem here is timing; during the freeze
we have to manually review this stuff.
>> I haven't read the full diff, so there are possibly more issues lurking
>> in it. In its current state, I am not inclined to grant an exception.
>> PS: urgency=high is no effect when the package is not in testing (in
>> case you weren't aware of it)
> I was not aware. However, the package was in testing until 2 days before
> I did the upload. The fact the package was removed made the update very
> urgent - and then the urgency is ignored because it was removed....
> Well... I don't make the rules.
Urgency describes how important it is for people to upgrade their
package. If bdii had still been in testing, the urgency would have made
(partly?) sense... Anyway, it is hardly a problem, so just a FYI. :)
> I can make another update using the dpkg-maintscript-helper script
> instead of my own not-so-great fix. If you truly do not want to take
> advantage of the fixes for memory usage and IPv6 support I could also
> upload a version where I backport the fix for the RC bug to the 5.2.5
> version. But I personally think using the new version would be better.
> Let me know what you think is petter.
Please do not get me wrong; I like improvements as much as every one
else, but I don't like having to manually review them.
I believe the RC bug fix on 5.2.5-2 should be reasonable sane and lets
take that as a starting point.