[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Please give a test to cron package, in experimental



Thank you for bringing this up @Georges !

I would like to add, that generally it is a bit confusing that you don't restrict merge requests against master on https://salsa.debian.org/ to maintainers, since the preferred way for contributors seems to be a patch/diff on top of a already patched repository.

Having a discussion about a patch/diff can be much more cumbersome than having a discussion in a MR, directly on the related code lines.

So I wonder why you don't just provide an additional branch "master-patched" against which merge requests can be opened without the risk of causing conflicts with existing patches.

IMO that could ease the life of contributors and maintainers (You don't need to press "merge" when done ..  gitlab always provides */merge_requests/408.patch, so you still could rely on patches)

(And ofc it would be great to use gitlab issues, instead of bugs.debian.org ... though that's a different story)

Sorry, it might be a bit off-topic, just wanted to throw in my 2cent, from a contributors POV.

Cheers, Alexander Schwinn

Am 08.10.23 um 13:54 schrieb Georges Khaznadar:
Hello,

I began to maintain cron (https://tracker.debian.org/pkg/cron) a few
months ago, and lowered the bug report count by a few dozens, began also
to write a few autopkgtests for it, and made it easier to test cron, by
providing a possibility to run cron tasks immediately (not after a one
minute wait)

While discussing about bug report #830821
(https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=830821), it appeared
that Alexabner Schwinn wanted to contribute, but was not at ease with
the packaging scheme which exists currently with cron package, in our
distributions stable, testing and unstable : we are maintaing a big heap
of patches to apply to Vixie's cron source version 3.0 (this latter is no
longer available from Internet).

The difference of difficulty is between:
---------------------8<---------------------------
... hack something
$ git commit -m "my awesome contribution"
$ git diff HEAD^
-------------------- and -------------------------
$ export QUILT_PATCHES=debian/patches # if you're no debian maintainer
$ quilt push -a
$ quilt new my_awesome_contribution.patch
$ quilt add <some_file>, ...
... hack someting
$ quilt refresh
$ quilt pop -a
$ git commit -m "my awesome contribution"
$ git diff HEAD^
$
---------------------8<---------------------------

Everyone can see that the second sequence is rather tricky for a
volunteer contributor, and results probably from the history of cron's
maintenance, which began far before our usage of version control
systems (Vcs).

So, I am trying to create a new upstream cron package, which is made
from Vixie's cron version 3.0, with all of the debian patches applied,
and packaging it; which results in a debian release 4.0-1, now uploaded
to experimental. All of the comments which came with the headers of
every patch are now included in Git's log, which preserves authorship
information.

If you have some free time, please can you give a try, or just share
your thoughts about this change ? I intend to push version 4.0 to
debian/sid shortly if we can agree.

Best regards,			Georges.


Attachment: OpenPGP_signature
Description: OpenPGP digital signature


Reply to: