#1312 Duplicate sphinx_reredirects packages
Closed: fixed 9 months ago by tibbs. Opened 9 months ago by jjames.

I maintain the python-sphinx-reredirects package, which was added to Fedora in August 2022. Recently, a review request for a python-sphinx_reredirects package, with an underscore instead of a hyphen, was approved and built for Rawhide and all stable releases. I have taken the following steps to get the new maintainer to realize that the new package is a duplicate:
- I replied to the Fedora Rawhide compose report where the new package was reported.
- I commented on the review request bug, noting the duplication
- I left negative karma on the updates for:
- F37
- F38
- F39
- I sent email to the maintainer describing the situation

The maintainer of the new package did not reply to any of my contact attempts, and pushed all of the updates stable in spite of the negative karma. We now have duplicate packages in F37+. I don't know what else to do about the situation, so I am turning to the packaging committee for help resolving it.


Hm, looks like you left negative karma for the review bug, but not the actual update (which would have caused bodhi to stop the auto push). Regardless, this is obviously not a good situation, especially with getting no responses from the maintainer. They should have received 6 emails if I count correctly :(

Not sure what we as the packaging committee can do here, other than say "this is not good" :frowning: There are other things you could do, including non-responsive maintainer process, or an un-sponsoring request for FESCo (since they apparently don't fulfil the requirements for a packager).

That's really not good. Sadly the karma that was left on the updates wasn't actually negative (there was just a negative reaction to the included bug) so it didn't stop the push to stable.

Note that the maintainer of the "new" package is generally an active one, and I wouldn't assume any malice here. Maybe they would have noticed negative karma since they would have needed to override it to push. Whether they noticed and ignored the negative comment, I can't say but we should assume they didn't.

Unfortunately, the packaging committee, being the body that just talks about how to package, really has not much power here. I technically have enough privileges to retire the packages and then lock down the repo on src.fedoraproject.org but I'm reluctant to go that far unilaterally and I don't have the time to check both packages to make sure I'm not missing something. I also have no ability to actually remove the packages from the distribution (that would take a releng ticket to get them fully blocked).

So I really dislike passing the buck here, but I think the proper body for arbitration here would be FESCo.

Edit: 100% ninja'd. Oof.

The proper move here is to retire the package from rawhide, add Obsoletes to python3-sphinx-reredirects across all branches, and probably to start the Non-responsive Maintainer process if we still can't reach @jamatos. Beyond being a duplicate, the new package does not follow the Python Naming Guidelines :(.

Yeah, adding the obsoletes would kind of do the trick.

I want to say that we really shouldn't have this kind of drama between packagers, but on second thought there's no real drama here since there's been no actual discussion. This is a super weird situation where we have all of these processes where people could have stopped this from happening, but things just didn't work out somehow.

  • Packager should have caught it.
  • Reviewer should have caught it.
  • The maintainer of the other package did catch it, but didn't click the right box to keep the package from being pushed.
  • Packager somehow didn't see the negative comment, the comment on the bugzilla ticket or personal mail and clicked the button anyway. (Or am I wrong and was the push to stable automatic?)

Obviously somehow the process failed but we sure don't want any more process in the way. So I'm kind of at a loss.

  • Packager somehow didn't see the negative comment, the comment on the bugzilla ticket or personal mail and clicked the button anyway. (Or am I wrong and was the push to stable automatic?)

All stable pushes were automatic; the maintainer probably didn't even realize anything was wrong.

I opened PRs to add the appropriate Obsoletes and Provides to python3-sphinx-reredirects.

┏━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Target  ┃ URL                                                                         ┃
┡━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ rawhide │ https://src.fedoraproject.org/rpms/python-sphinx-reredirects/pull-request/1 │
│ f39     │ https://src.fedoraproject.org/rpms/python-sphinx-reredirects/pull-request/2 │
│ f38     │ https://src.fedoraproject.org/rpms/python-sphinx-reredirects/pull-request/3 │
│ f37     │ https://src.fedoraproject.org/rpms/python-sphinx-reredirects/pull-request/4 │
└─────────┴─────────────────────────────────────────────────────────────────────────────┘

Thank you, everybody. I don't believe there was any malice either. I have had positive interactions with the other packager in the past. I think he just wasn't paying attention. I have made another attempt at contacting him, hopefully with more success this time. I have also merged the PRs noted in the previous comment and am running through the builds.

And now I understand the difference between leaving negative karma on the review bug and negative karma on the update itself, so I learned something from all this. :-)

We discussed this during today's FESCo meeting, and since it's pretty clear that this package was added by error, it should be retired. That the package's maintainer is not responsive to bugmail / bodhimail / direct messages is an orthogonal problem.

I checked and it looks like it was retired this morning by the maintainer (not by anyone in FPC or FESCo). So it does appear the message was received at some point, and there's nothing further for us to do.

Metadata Update from @tibbs:
- Issue close_status updated to: fixed
- Issue status updated to: Closed (was: Open)

9 months ago

Log in to comment on this ticket.

Metadata