New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
DOC: Formalize LTS backport policy #11429
Conversation
👋 Thank you for your draft pull request! Do you know that you can use |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I approve this message.
******************* | ||
|
||
Starting from astropy 5.0, backports to Long-Term Stable (LTS) releases | ||
will only cover: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
After the standard support life cycle.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Cadair , can you please elaborate on this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Well why the LTS version is both "the LTS" and the latest release the backport policy is the same as normal?
I think this is well stated, thank you. |
After some thought I tend to disagree, or at least that not what LTS means for me. As a comparison, Python releases have bugfix releases during 2 years, and then security/critical fix releases when needed during 3 more years. Changing LTS to security/critical/on-demand fixes only is a regression imo, and not what LTS was intended for, and also not what we have been doing since LTS started. |
@saimn I agree with your points overall, and it's how I feel as well. But I didn't really want to push back on this because I know LTS maintenance has been a big burden for all involved, so I'm sympathetic to the desire to change the policy. I know I just wrote earlier that I think this policy is well-said, you have me wanting to backtrack a bit. I think the backport procedure usually worked pretty well in the past but over time became more complicated for various reasons (mostly to do with CI I think). Now that we are experimenting with meeseeksbot and also have CI a little better straightened out for the 4.0.x branch (#11123) I think we should wait and see if it becomes any easier. Maybe what we could do is soften this slightly with wording like:
Where we can define "trivial to backport" as either "automated backports merge / pass CI without any additional intervention" or it would take me at most 5 minutes to resolve any minor merge conflicts. And by "me" I mean I'll volunteer. |
Indeed, the backport workflow with the bot should help by backporting just after PRs are merged (so changes would still be fresh, and it would be up to the sub-package maintainer to see if it's worth fixing the conflicts) and by running CI on the backport branches more regularly. |
So, uh, how do we proceed? Should I just close this without merge? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some wording suggestions based on some out-of-band discussion.
My two cents is that with the backport bot things should be a bit easier, so "critical" is not mandatory. But it is good to have this here nonetheless to make sure everyone understands that the purpose of LTS and the scope of fixes.
LTS Backport Policy | ||
******************* | ||
|
||
Starting from astropy 5.0, backports to Long-Term Stable (LTS) releases |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Cadair , you mean like this?
Starting from astropy 5.0, backports to Long-Term Stable (LTS) releases | |
After the standard support life cycle, backports to Long-Term Stable (LTS) releases |
Update:
|
I don't know--if that's the policy starting v5.0 I don't see why it shouldn't be in the stable docs. I think we want to defend this policy as "stable" and not be pressured into changing it. If it absolutely must be changed, I think any change to the policy should coincide with a new release. But as it is I'm pretty sure this covers all the bases, including the "pipelines that absolute need it can have backports but they must be willing to help". |
docs/lts_policy.rst
Outdated
* Critical security fixes. | ||
* Bug fixes where backporting is straightforward (e.g., no conflicts). | ||
* More complex fixes are permissible only as resources allow, ideally with | ||
backports done by the original author or from institutions who are on LTS |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How about "users"? In most cases, those users are institutions, but I don't think that we should write a policy that explicitly excludes non-institutional users.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I added "users". Look better now?
Hmm looks like I need to rebase to get rid of failures. |
as Stuart Mumford suggested. [ci skip]
[ci skip] Co-authored-by: Erik Tollerud <erik.tollerud@gmail.com>
Apply suggestion from hamogu [ci skip] Co-authored-by: Hans Moritz Günther <moritz.guenther@gmx.de>
We can further word-smith this in follow-up PR(s). Thanks for the reviews! |
Description
This pull request is to formalize LTS backport policy.
We have to decide where to best place such a policy and hammer out the wordings.
Fixes #11405