-
Notifications
You must be signed in to change notification settings - Fork 697
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
3.12.1.0 — a minor release of Cabal and friends #9885
Comments
I added another checkbox in the very beginning: to release 3.12.0.0 because as Mikolaj suggests on Matrix, a lot is blocked on some of it. |
Possible blocker: |
Added a reference to #9729 by the "Release 3.12.0.0" todo. |
In the issue description above, I added a suggestion to add links to respective PRs when ticking out the checkboxes. |
I added another point that goes before the main list of steps: to check if any more backports are needed for 3.12.1.0. |
Seems like two PR's are worth of backporting:
|
Sounds good. @ffaf1, @Kleidukos? |
Is there an ETA when release will happen? The lack of GHC-9.10 compatible |
There's no ETA that I know of but thank you for signaling this concern: this will help us to prioritize things. |
We are going through backports slowly. As discussed on Matrix, backports will have to be limited to three things: bugfixes, documentation, CI (except bootstrap). Rationale: it's a slow process, and we have a growing demand for a cabal-install that can work with Cabal-3.12. |
How is it going? There is always going to be one more backport to do, but are they that critical? |
There were some critical CI fixes that we were backporting over last week. We want to have one last pass over the list of actual bugfixes, and then get to actually do the release. We still don't have a tentative date. All who's in need of a newer cabal-install (e.g. because they use Custom setups), should consider using the prerelease in the meantime: https://discourse.haskell.org/t/ann-cabal-3-12-0-0-released/9504#how-to-get-the-cabal-install-pre-release-3 One issue that we potentially want to be resolved, fixed on master and backported is #9917 and it's in very early stages... |
You might want to change things in future so that it does not happen again in time for GHC 9.12, but #9917 per se would be immediately resolved by releasing
But CI fixes are not a part of |
Well we consider green CI a prerequisite for a release.
We want to change this before 3.12 so that when Cabal-3.14 is release we don'tget into the same trouble. |
My opinion that it's important to have a release of |
@ulysses4ever: How about timeboxing #9917? Once all other backports are merged, give it 2 more days? (CC @andreasabel) |
Sounds good to me. |
I do not have time to work on #9917 now, emitting where and how to emit the warning needs more understanding of the code base. I am paid to work on Agda and I need a working Haskell infrastructure, but digging so deep into Cabal is somewhat out of scope. But also, there seems to be no one who is paid to do bugfixing work on Cabal, is that right? From our meetings, I get the impression that all is volunteer work, and @Mikolaj has a part-time job to orchestrate this work... |
@andreasabel: I can't speak for others (AFAIK, some other companies are permitting or encouraging their employees to contribute to cabal) but that's about right regarding my humble person and the key point is we crucially depend on contributions from cabal users. But I did not suggest you fix the problem. I CCd you, because you are an interested party and you did cabal a favour by coordinating this particular issue, so you should be aware of the constraints as they emerge. As soon as there's consensus about the issue, maybe we could advertise it to find an implementer? |
Discussion over at benz0li/ghc-musl#8 (comment) reminded me that it'd be great to update cabal-install so that it builds with GHC 9.10 before the 3.12.1.0 release. So #9914, which is currently stuck with some test failures due to 9.10. |
#9914 (support for GHC 9.10) has landed but its backport to 3.12 currently has some test failures: |
I've added some more info in the backport, but now somebody has to do the actual work. |
Hi @ulysses4ever, kindly reconsider #9917 for 3.12.1.0. It is wrecking thick havoc onto custom-setup packages with GHC 9.10.
👇 👇 👇
In effect, cabal just unconditionally overrides bounds set by user explicitly in Additionally, getting |
@ulidtko: hi! I'm afraid the ticket you linked is an issue, not a (doubly reviewed) PR. As such, it can't be considered by the release manager. On the bright side, we do plan to release |
There are two sticky points that I can see now:
I propose to pass on both of these for the release and get them for a future point release, hopefully. One thing that we probably want to get in before proceeding:
|
Changelogs PR: #10124 (second step in the release process, see the whole list of steps in the OP). |
Does anything remain to be done on the hackage-security side? Its GitHub repo is already updated, but I don't think it's on Hackage yet. |
Does bootstrapping work? |
No, but not for that reason.
Looks like we have a leak from Also, what's with the index-state warning?
Looks newer than the requested index-state to me…. (ETA: forgot to switch back from cabal 3.10.3.0, and doesn't happen with my 3.12 build) |
FWIW, git history is a bit difficult because the file was renamed, but it does indeed look like either the rename (which is faked by git, and determined post hoc by heuristics) or backporting #9565 inadvertently brought in |
This should not show up in latest cabal, it was demoted to a warning (and reworded unless I am misremembering). #8944 |
Bootstrap jobs seem to be passing with updated bootstrap JSONs. |
We're moving to the next step on the list — uploading release candidates. |
Is this point cleared up? If so, in which PR? |
@jasagredo what do you mean by "latest cabal"? None of the recent changes around index state have been released, I think. Do you mean |
3.10.3.0, because I'd forgotten to switch back after probing a different issue. When I switched to my usual cabal (dogfooding |
Good, thank you. We can proceed then. |
FTR we're currently blocked on #10136 which is breaking CI. (ETA: merged and backported, we can proceed.) |
We're done with the candidates:
We're about to move on to the next step -- creating binaries and uploading the artifacts (binaries and tarballs) to the website. (As a part of this step, we'll push the git tags.) But before going there we need to merge one last fix for how we create the release: |
Tags are up (as a part of the artifacts step), and they point at the same commit as usual: |
The release artifacts have been uploaded to https://downloads.haskell.org/~cabal/cabal-install-latest/ It's not an official announcement yet (see the checklist in the OP) but if all goes well, that should be final artifacts. |
Hackage released is up. Announcement on Discourse: https://discourse.haskell.org/t/cabal-install-3-12-1-0-and-accompanying-libraries-released/ |
All checkboxes appear to be ticked. Closing... |
The 3.12.1.0 release of Cabal and related packages, including
cabal-install
, is supposed to follow a (yet to be released) major release ofCabal
andCabal-syntax
-- 3.12.0.0. Below is a checklist for the release taken from the wiki page on releasing. /cc @Mikolaj @ffaf1 But first and foremost:base
bounds for desupported ghc 8.2 (backport #9964) #9980master
since March, and see if anything is worth of backports to3.12
but didn't make it (We use this search query.)When ticking out the boxes below, please, try to add links to respective PRs.
A.1 Before the release
A.2 After the release
git log --use-mailmap --pretty=format:%an Cabal-v3.6.1.0...Cabal-v3.8.1.0 | sort -u
and don't forget to mention "reviewers, QA testers, devops and others"ghcup install cabal --url https://downloads.haskell.org/~cabal/cabal-install-3.8.1.0-rc1/cabal-install-3.8.1.0-x86_64-linux-deb10.tar.xz 3.8.1.0
), changelogs, how great our community is and how we love and cherish collaborators, known important bugs and that if anybody is keen to see a particular bug go before 3.8.1.0, the window is now very narrow but we will try to help the interested person squash the bug)The text was updated successfully, but these errors were encountered: