cabal Text file, newlines at end of files

Fixes #9802 for *.hs and *.project files ~and adds a script taking a file glob pattern for doing this~, adding newlines at the ends of files where needed. Adds back the whitespace github workflow and uses fix-whitespace to fix non-conformance. Add a short note about this in the contributing readme.

[x] Patches conform to the coding conventions.

Mar 13 '24 21:03 philderbeast

@andreasabel created whitespace.yml, let us see what he has to say.

Mar 13 '24 22:03 ffaf1

@ulysses4ever I see you are a fix-whitespace contributor. Do you think we can get fix-whitespace and fourmolu to play nicely together?

Mar 14 '24 11:03 philderbeast

Thanks @andreasabel fix-whitespace-action is way less configuration. I reverted one of the corrections and it shows violations in the output.

Mar 15 '24 10:03 philderbeast

Label merge+no rebase is necessary when the pull request is from an organisation.

Mar 15 '24 10:03 philderbeast

I reverted one of the corrections and it shows violations in the output.

@ulysses4ever I checked and it shows violations.

Mar 15 '24 10:03 philderbeast

Very cool, thanks!

Mar 15 '24 10:03 ulysses4ever

@ulysses4ever, funny thing about moving it to quick jobs is that it now doesn't run nearly as quickly (the ghcup, hackage and alex stuff takes a lot longer to run)!

Mar 15 '24 10:03 philderbeast

In terms of developer feedback, putting it alongside (at the end of) quick jobs is a big degradation.

Mar 15 '24 10:03 philderbeast

Mm, good point. My intention was not having it run quickly, though, and rather I dislike proliferation of standalone workflows. But maybe it’d be good to get an actually quick feedback from this job…

We need to fix quick-jobs to only download binaries and not build anything with ghc!

Mar 15 '24 10:03 ulysses4ever

I do think that jobs like this one and the docs should run near instantly otherwise the wait is annoying.

Mar 15 '24 10:03 philderbeast

@ulysses4ever are you OK if I revert the last change, putting whitespace into its own workflow?

Mar 15 '24 10:03 philderbeast

Yeah, up to you, approved. Thanks!

Mar 15 '24 10:03 ulysses4ever

@ulysses4ever I see there is something wrong with the cache in quick jobs;

Post job cleanup. Warning: Path Validation Error: Path(s) specified in the action for caching do(es) not exist, hence no cache is being saved.

Mar 15 '24 10:03 philderbeast

Interesting... Maybe worth a new ticket with the continuous-integration label.

Mar 15 '24 11:03 ulysses4ever

How much squashing should I be doing with this given that the advice is:

If you push a fix of a whitespace violation, please do so in a separate commit.

Mar 16 '24 11:03 philderbeast

Sorry @Mikolaj for the lack of squashing. I asked about squashing but forgot to come back to it. Noticed when I got the cabal.head email.

Mar 25 '24 17:03 philderbeast

Ouch. With no-rebase and no manual rebase nor squash just before merging, the 12 commits are strewn artistically across other PRs from the several last days. I guess we have to get used to git art. :D

Mar 26 '24 08:03 Mikolaj

For the sake of bisecting, a linear history is really a plus. Can't we enforce it on this repo? (On the Agda repo, I disabled merge commits.)

Mar 28 '24 06:03 andreasabel

Some contributors — and most importantly, some employers — do not fancy cabal to have write access their repos (e.g.).

At first we did some manual cherry-picking, but then we gave in and merge+no rebase was introduced.

Mar 28 '24 06:03 ffaf1

Can mergify do the rebase by copying the contributor branch to a new branch on this repo first? Can this be set up? This might be a technical solution.

If a PR cannot be rebased then it should be rejected by default. Otherwise, we'll not get a linear history.

Contributors with write access could be encouraged to push to a branch here rather than on their own repo.

Mar 28 '24 07:03 andreasabel

Contributors with write access could be encouraged to push to a branch here rather than on their own repo.

This could be difficult if the contributor is working on an employer's tree as part of their job.

Mar 28 '24 15:03 geekosaur

This requires a new issue for visibility and discussion.

Mar 28 '24 17:03 ffaf1

@mergify backport 3.12

May 13 '24 15:05 ulysses4ever

backport 3.12

✅ Backports have been created

#10007 Text file, newlines at end of files (backport #9804) has been created for branch 3.12 but encountered conflicts

May 13 '24 15:05 mergify[bot]

@philderbeast

Can you check that fix-whitespace is run with a flag that will show the actual violations?

I checked and it shows violations.

Apparently, we misunderstood each other. What I meant is that it should show not only the names of the files with whitespace issues but also the issues themselves. It could be done by the verbose: true option of the action:

https://github.com/andreasabel/fix-whitespace-action/blob/2a7678e2f1319ae6e9595acb06b8cfb46e38af23/action.yml#L17-L20

May 13 '24 18:05 ulysses4ever