[Development] Changes to automatic cherry-picking

[Daniel Smith]

Hi all,

I'll be going live today with these changes to the automatic cherry-pick processes. I'll be monitoring the systems for the next few days at least to catch any issues. If you experience any oddities or unexpected behavior, please contact me at daniel.smith at qt.io directly with a report.

Reminder: You do not need to change the way that you work, but the resulting cherry-picks from your footers will now be created in order of newest-to-oldest branches. This means that cherry-picks will now block the propagation of your change to older branches if a merge conflict exists in a cherry-pick or the pick fails to integrate, so please try to take action on cherry-picks with issues ASAP.

Best regards,
-Daniel
________________________________
From: Daniel Smith <daniel.smith at qt.io>
Sent: Friday, July 28, 2023 12:31 PM
To: Tor Arne Vestbø <Tor.arne.Vestbo at qt.io>
Subject: Re: [Development] Changes to automatic cherry-picking

> I would recommend skipping this new footer. There’s nothing preventing manual cherry-picks, either locally via git, or via Gerrit’s UI, and doing those manualy seems more appropriate anyways for the exception when a change needs to be rushed into a branch.

I think I ultimately agree with this. It would be ideal for even hotfixes to go into dev first, and then be manually picked if a faster turnaround is needed for an impending release. It's worth noting that it would still be best to specify all the intended pick targets in the Pick-to footer, even if one intends to pick some of them manually. The bot will simply ignore any targets that already exist (and are merged).

In such cases where a pre-emptive pick is performed from an unmerged change on dev, we would need to accept that it won't be possible to reference a valid sha in the footer generated by cherry-pick -x. Gerrit would still show the correct relation in the UI, but the commit history would be a little less traceable. Perhaps for urgent issues in impending releases, a Fixes: footer referencing a P0 could be required in practice.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20230911/6f13690f/attachment-0001.htm>