lib/attrsets: add mapCartesianProduct function

[gvolpe]

Description of changes

The cartesianProductOfSets function was added some time ago (aka transpose or sequence in other functional languages), while deprecating the crossLists function. I have come across the deprecation message "use cartesianProductOfSets instead" without further details, which motivated me to improve the situation by adding a new convenient function. Example:

λ lib.mapCartesianProduct ({a,b}: a+b) {a=[1 2]; b=[3 4];}
[ 4 5 5 6 ]

It's the equivalent of using the former followed by map, which seems to be quite a common pattern to justify a single function:

λ builtins.map ({a,b}: a+b) (lib.cartesianProductOfSets {a=[1 2]; b=[3 4];})
[ 4 5 5 6 ]

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 24.05 Release Notes (or backporting 23.05 and 23.11 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/lib-crosslists-is-deprecated-use-lib-cartesianproductofsets-instead/41824/4

[infinisil]

aka transpose or sequence in other functional languages

I don't think I've ever seen this operation be called that? Transposing usually means diagonally mirroring a Matrix.

cartesianProductOfSets is a bit too long imo, I think we could shorten it to cartesianProduct though, otherwise it seems like a fitting name.

And then I think mapCartesianProduct would be best because it's the same as map f . cartesianProduct a.

[gvolpe]

I don't think I've ever seen this operation be called that? Transposing usually means diagonally mirroring a Matrix.

Oh I got the transpose function completely wrong, you're right it doesn't apply to this use case 😅

Though, sequence is what it's called in functional languages like Haskell and Scala, e.g.

λ sequence [[1,2,3],[4,5,6]]
[[1,4],[1,5],[1,6],[2,4],[2,5],[2,6],[3,4],[3,5],[3,6]]

It is more generic than this, but it results in the Cartesian product of every item when dealing with lists of lists.

cartesianProductOfSets is a bit too long imo, I think we could shorten it to cartesianProduct though, otherwise it seems like a fitting name.

I agree. This would mean deprecating the existing cartesianProductOfSets for this new one, right?

And then I thin mapCartesianProduct would be best because it's the same as map f . cartesianProduct a.

Definitely makes sense, would be happy to adjust to the suggested names, thanks for reviewing!

[infinisil]

Though, sequence is what it's called in functional languages like Haskell and Scala, e.g.

λ sequence [[1,2,3],[4,5,6]]
[[1,4],[1,5],[1,6],[2,4],[2,5],[2,6],[3,4],[3,5],[3,6]]

It is more generic than this, but it results in the Cartesian product of every item when dealing with lists of lists.

I'd say that's half of a coincidence, because it relies on the Applicative instance (see the more generic sequenceA) for lists to be the cartesian product, but there's also another Applicative implementation on lists that would make sense :D

I agree. This would mean deprecating the existing cartesianProductOfSets for this new one, right?

Yeah I guess that would be best.

[infinisil]

Please also separate each change into a separate commit, e.g.:

• Renaming cartesianProductOfSets
• Tree-wide replacement
• Introducing mapCartesianProduct and introducing its tests
• Tree-wide replacement
• Applying lints

[gvolpe]

Should've looked at your profile before, we could discuss ZipLists another day 😄

Applying lints

Can you elaborate on what this means? I guess it's running some kind of formatter, but I'm not familiar with the current process.

[gvolpe]

@infinisil could you have a look at the remaining open questions when you get a chance? Thanks

[infinisil]

Other than the commits needing cleanup this looks good to me!

[gvolpe]

Other than the commits needing cleanup this looks good to me!

@infinisil done ✅

[Mindavi]

LGTM except for the 'apply lints' commit that doesn't seem very related

[gvolpe]

LGTM except for the 'apply lints' commit that doesn't seem very related

It is not related indeed, but I had to make changes on that file and noticed a lot of things that can be removed. I don't think anybody would care to submit a PR to remove unused bindings to be honest, but happy to adjust the commit message if there are better suggestions.

[Mindavi]

I think it's ok on second thought, thanks!