3.22 release

It looks like there are a lot of changes slated for the pulpcore 3.22 release but perhaps the biggest change (the domain work) is still yet to be merged. We need a few of the features in the 3.22 but my concern is about the stability of 3.22 if the domain work is merged right before 3.22 is cut. So I’m wondering if it would make sense to do a smaller 3.22 release in December, then merge the domain work and do a 3.23 release in January or so? This would allow the domain work some time to soak in main before it’s released. Thoughts?

We talked about this in the pulpcore and go/no-go meetings. The net of the discussion was, until a release happens, there’s very little actual “soaking” - devs work with main, but the actual users (which are the way most edge-cases are tripped over) won’t; the feature won’t actually get any soaking until it’s been in a release.

What do you think?

Thanks for the response. I do remember discovering bugs with features before they were released but then again, I agree that a majority of bugs are discovered by users after the feature is released.

Aside from this point though, I guess the two questions I have is first, is the next release being held up by the domain work? Ideally for us we’d like the release sooner (ie December) rather than later. And secondly, is there any idea of how the domain feature might impact the stability of the release if the user has it disabled?

It’s more accurate to say “the domain feature is driving core/3.22”, in response to another stakeholder’s request.

You’re the first user/group that’s asked specifically for a core/3.22 in December, other than the domain-request. Can you put yourself on the Open Floor agenda for tomorrow so we can discuss there?

As far as “how risky is it to have the domain-code without turning it on” - I believe it should be “safe”, but I’ll defer to @gerrod to respond more here. Gerrod?


Added a topic to tomorrow’s open floor. I’ll be there to discuss. Thanks @ggainey!

1 Like

In theory the domain work should be completely optional and not impact other plugins if it is not turned on. I’ll do some testing this week to ensure that all other plugin’s testing passes with domains in pulpcore.


Outstanding, thank you!

At today’s go/no-go meeting on #pulp-meeting, we determined:

  • The domains work has outstanding changes requested and cannot be merged within the time that existing 3.22 features are available. It has been moved to the 3.23 blockers list.
  • The 3.22 blockers list has only one blocker. It’s nearly ready to merge as it is in ready for re-review stage.
  • This is the last release that will use pulp_installer. Future releases will either be “built by hand Python deployments” or container based. Upgrade instructions into the official containers are available and tested already. See the thread ^ for more info.
  • Names for releases were a fun idea that was suggested. This one could be “winter solstice” as it is close to the winter solstice.

“Winter Solstace” aka 3.22 is planned to be released on Monday, Dec 19th.