sorbet: Convert some hidden-definitions into RBI files

[issyl0]

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your changes? Here's an example.
• Have you successfully run brew style with your changes locally?
• Have you successfully run brew typecheck with your changes locally?
• Have you successfully run brew tests with your changes locally?

---

• Inspired by @dduugg's recent Sorbet work, I thought I'd start paring down the srb rbi hidden-definitions file by adding RBI files for dynamically generated methods.
• The srb rbi hidden-definitions command is deprecated and we've tried removing it before, but the outcome was that we needed to write more RBI files and it wasn't a massive priority. Tonight I had (a little) desire to start this, with the help of a hacky script.

[BrewTestBot]

Review period will end on 2023-02-17 at 01:06:39 UTC.

[dduugg]

Neat!

One suggestion, if we're going to start building out large rbi files, would be a test that ensures they stays in sync with their source. Maybe something like (pseudocode-ish):

methods_before = Homebrew::EnvConfig.methods
require_relative `Library/Homebrew/env_config.rbi`
methods_after = Homebrew::EnvConfig.methods
expect(methods_after - methods_before).to be_empty

wdyt?

[Bo98]

One suggestion, if we're going to start building out large rbi files, would be a test that ensures they stays in sync with their source.

tbh we should probably just auto gen the rbi files. I can have a look given I did the one for attr_predicate and attr_rw.

[MikeMcQuaid]

Agreed with @Bo98: it'd be really nice to autogenerate these.

[issyl0]

Yeah, no idea how to do that that's not my super hacky script. I'm assuming there's a more Sorbet-native way. Can @Bo98 or @dduugg teach us?

[Bo98]

Now that 4.0.0 is out I can have a look at this.

[Bo98]

Will have something to show by Monday. I would point to the existing sorbet/parlour/attr.rb, but parser code isn't always the easiest to follow so I'm looking to see if I can write that as a runtime generator rather than a static generator. Hardly anything we use supports Ruby 2.6 anymore (last Rubocop update is in about 6 weeks) so will need to see what is and isn't available for us on the versions we have.

[issyl0]

Can we merge this as-is (since it's identical to any automatic generation) and work on the autogeneration separately? I recognise the manual process is non-ideal, but it was vaguely fun last night and whatever we write next can be done by the Sorbet action or whatever, and replace my fingers.

[MikeMcQuaid]

Can we merge this as-is (since it's identical to any automatic generation)

Will this still be automatically updated when we e.g. change env_config.rb? If so: yes, if not: no 😁

[issyl0]

Will this still be automatically updated when we e.g. change env_config.rb? If so: yes, if not: no 😁

Oh. No. 😁 But brew typecheck will fail and someone will add the method to the RBI file (same as with the current https://github.com/Homebrew/brew/blame/master/Library/Homebrew/cli/args.rbi#L339-L340, for example).

[MikeMcQuaid]

Sorry @issyl0, this feels like a step back if it's adding more manual work that was previously done by CI 😭

[issyl0]

I'd argue that it's still moving us forward because rbi hidden-definitions is deprecated and moving off it in any way (while adding better type information than just Sorbet's "don't know, shrug") is better. But OK, I'd like to learn how to write RBI generators for dynamic methods anyway so this is a good excuse.

[BrewTestBot]

Review period ended.