Avoid using to-be-deprecated Data.List.NonEmpty.unzip

[mixphix]

For CLC ticket #86

---

Please include the following checklist in your PR:

• Patches conform to the coding conventions.
  [N/A] Any changes that could be relevant to users have been recorded in the changelog.
  [N/A] The documentation has been updated, if necessary.
  [N/A] Include manual QA notes if your PR relates to cabal-install.

Bonus points for added automated tests!

[ulysses4ever]

Can't we use Data.Functor.unzip? It doesn't have a since annotation, so I'm unsure if it fits into our support window. But that would be much preferable if it's possible imo.

[mixphix]

We'd have to restrict the bounds on base very heavily. I could add a comment about it in the source code, but I hesitate to touch more files than necessary.

[Kleidukos]

As said in the CLC ticket (haskell/core-libraries-committee#86 (comment)) I'd be more comfortable if we had an automated way to prevent it from coming back, preferably with an HLint or Retrie rule.

[ulysses4ever]

@mixphix a comment nearby the code you added would be reasonable, I think.

[ocharles]

The use of Data.List.NonEmpty.unzip is entirely appropriate here, no? Is it just the warning that's problematic?

[mixphix]

The warning is causing the GHC pipeline to fail, see here

[ocharles]

Right, but the code is right so what do you think about turning the warning off, or downgrading it's severity (not sure if the pipeline is looking for -Werror or something else)

[mixphix]

The pipeline has at least -Werror=deprecations. Disabling warnings is a module-wide option that may hide future issues.

[ocharles]

Yea, you'd probably want to downgrade that selectively with CPP.

[mixphix]

You want me to use CPP to avoid a one-line local definition?

[ocharles]

It's not a project I contribute to, so I don't want you to do anything. I'm just observing what people who use unzip are meant to do, and want to understand what the options are. That said, I do think it's a shame to lose code clarity by now needing to inline the definition

[ulysses4ever]

Hey @mixphix! Thanks a lot for the contribution! I see you requested my review but i already approved some time ago and that's about all i can do: the PR needs a second reviewer. We could tag some folks who are involved with Cabal but focused on other things and so didn't respond to this PR. E.g. @Kleidukos, @ffaf1, folks, I hope you don't mind me tagging you and perhaps you could find time to review this tiny improvement.

[Kleidukos]

With ndmitchell/hlint#1500 I am now much more comfortable.

[ulysses4ever]

@mixphix do you have permissions to add labels, and if so, are you ready to apply the merge-me label to hand it over to the merge bot?

[mixphix]

I don't have label permissions, but please go ahead!

[Kleidukos]

I gotchu

[Bodigrim]

What's up with Mergify here?

[Kleidukos]

Permissions stuff apparently, but let's merge this. :)

[mixphix]

How do I get the GHC MR to include a Cabal version with this patch?

[ulysses4ever]

@mixphix the usual way with submodules, I think, is to cd into the submodule directory, do git pull, cd back, git add --all and commit...