-
-
Notifications
You must be signed in to change notification settings - Fork 159
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
[RFC 0106] Nix release schedule #106
Conversation
Nixpkgs/NixOS releases have mechanisms that encourage stability as well as ensuring it gets done. Concepts like release managers, ZHF, release channels (along with all the corresponding testing), backports policy, wiki about release process, and so forth help drive progress. I don’t have a solid idea formulated yet, but I think some analogous concept would be good to include in this RFC. |
I have no strong opinions on semver vs. calver. Thoughtful semver is useful, but not so great that it'd be worth regular debates over whether a release qualifies as an x/y/z bump instead of something a script generates. I guess I'd prefer semver if it's a decision being made within 24h of release by a small, decisive group with good heuristics/razors. If it takes longer than that, it's probably better for everyone to just automate it. This is a little far afield, but I've daydreamed about "test-driven versioning." I can't really extoll its virtues--it's just something I vaguely intend to give a try someday--but it struck me while writing this that some of Nix's ~contracts (like store, CLI, and daemon compatibility) might make interesting candidates for test-enforced version promises. |
rfcs/0106-nix-release-schedule.md
Outdated
[unresolved]: #unresolved-questions | ||
|
||
* Should we still do maintenance releases (like 2.3.x)? Should there | ||
be a long-term stability release (like 2.3 is now, de facto)? |
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 don't see the point of this. Ideally, the only backward-compatible changes that happen in Nix are in the interface, not the language and building mechanisms. Because of this, supporting a new Nix release in a project will likely never require large changes.
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.
Dunno for the long-term releases, but I think we’ll have to do maintenance releases anyways, if only because we might have to do some bug fixes on the latest release before the end of the 6w cycle
a way that they can be changed or removed, while still getting | ||
feedback from adventurous users. So long as experimental features | ||
don’t cause breakage in stable features, it’s fine to merge them into | ||
master and include them in a new release. |
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.
While this is true, a lot of smaller changes don't get an experimental feature, e.g. NixOS/nix#4922. Perhaps there should be a generic experimental feature for enabling unstable minor changes such as this, so that we don't have to add an experimental feature for each?
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 don’t think having a arbitrary (reasonable) number of experimental features is too much of an issue. People shouldn’t be expected to just blindly buy into them, and having each feature guarded by an experimental flag gives more flexibility to rework/abandon them.
(Anyways, I think the usage of experimental features should be refined/standardized, but that’s out of scope for this RFC)
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/nix-release-schedule-and-roadmap/14204/18 |
I think the most critical thing is not necessarily release branches, but upgrading. E.g. if Nixpkgs goes CA-drvs default, making sure the previous n stable Nixpkgs releases have a new enough Nix. |
I'd like to nominate:
as shepherds. |
Sure! |
I'm certainly interested, but not sure if I have the capacity in the next time for another duty. I should have an answer to this by the beginning of next week, then I can accept or decline the nomination, hope that's OK :) |
Yes, I can definitely shepherd that one 👍 |
Sorry, not my cup of tea. |
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/installer-test-suite-small-project-s-high-leverage-help-wanted/13662/5 |
So, I'd be up for it if everyone else involved is OK with that :) |
I can assist with this. I've been focusing on Nix itself for a bit. I'll self-nominate. I've got an updated flake fork of the nix-install-matrix lying around that can track and test various scenarios - throw it up on a "Nix installer status page"? |
Please update the RFC with the following shepherds: @Ericson2314, @regnat, @Ma27, @tomberek (leader). Thank you! |
Plan of Actioncreated a room at #nix-rfc-106:matrix.org for any discussion My guess is that the schedule and intent is not controversial. Focus on establishing some policy and guidance the community can depend on.
|
So based on struggles I am having updating a client from 2.3 to 2.4, one thing I would like to see more of is testing of mixed clients/daemons/sshes as part of release-readiness. We've made NixOS/nix#5572 to test the new client with the old daemon, but I would also like to test the old client with the new daemon, and it is a but unclear what the best way to do that is. |
Regarding updating, I updated the daemon to 2.4pre and my services ran 2.3. After a few day I was confronted with NixOS/nix#5299 . I spend hours to fix the broken entries in the nix store. |
On 1: That's out of scope for this RFC IMHO. We should probably have another RFC to decide on an experimental features policy. On 2: Releasable state means that https://hydra.nixos.org/jobset/nix/master must be green. This is a superset of On 3: That's me currently since I have the GPG signing key. On 4: We should update the PR template to ask for release notes / documentation updates, if appropriate. |
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/nix-2-4-and-what-s-next/16257/1 |
a way that they can be changed or removed, while still getting | ||
feedback from adventurous users. So long as experimental features | ||
don’t cause breakage in stable features, it’s fine to merge them into | ||
master and include them in a new release. |
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 don’t think having a arbitrary (reasonable) number of experimental features is too much of an issue. People shouldn’t be expected to just blindly buy into them, and having each feature guarded by an experimental flag gives more flexibility to rework/abandon them.
(Anyways, I think the usage of experimental features should be refined/standardized, but that’s out of scope for this RFC)
rfcs/0106-nix-release-schedule.md
Outdated
[unresolved]: #unresolved-questions | ||
|
||
* Should we still do maintenance releases (like 2.3.x)? Should there | ||
be a long-term stability release (like 2.3 is now, de facto)? |
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.
Dunno for the long-term releases, but I think we’ll have to do maintenance releases anyways, if only because we might have to do some bug fixes on the latest release before the end of the 6w cycle
Co-authored-by: Théophane Hufschmitt <regnat@users.noreply.github.com>
Update Plan of Action
@NixOS/rfc-steering-committee: recommend moving this to FCP after RFC edits mentioned above. Shepherd team has general consensus and is recommending adoption. |
@tomberek I've updated the RFC to define "releasable state" and mention the issue template. |
@tomberek do you think this RFC is now ready for FCP since the last commit? |
Yes. |
When it comes to back-porting fixes, how many versions back would receive backports with this new more frequent releases? Will the recommended way be to move forward to next version instead? Also should the nix versions automatically be updated in nixpkgs as well? I notice now we do have a release of Nix 2.5.0. But nixpkgs for example have yet to be updated with this new release. |
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/rfc-106-fcp-nix-release-schedule/17096/1 |
Sorry, we forgot to announce the RFC on Discourse. I've done that now. |
What about my previous questions, are those things that should be part of the RFC? E.g.
|
We currently don't have a formal back-porting policy, so this RFC doesn't really change anything in that respect. On the Nixpkgs release process, we probably want to maintain (i.e. back-port to) the default version used in the latest NixOS release. |
The RFC has been merged. Thanks everybody! |
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/nix-2022-what-was-great/24410/5 |
* Nix release schedule * Fix link * Update rfcs/0106-nix-release-schedule.md Co-authored-by: Théophane Hufschmitt <regnat@users.noreply.github.com> * Add shepherds * Update to reflect latest meeting Co-authored-by: Théophane Hufschmitt <regnat@users.noreply.github.com>
This RFC proposes doing new Nix releases every six weeks, as previously proposed on Discourse.
Rendered