<@U02KAN6061E> I put a proposal in a small interna...
# developmentl
late-advantage-75311
09/22/2023, 7:58 PM@bitter-ability-32190 I put a proposal in a small internal enhancement issue you raised in June https://github.com/pantsbuild/pants/issues/18856
b
bitter-ability-32190
09/22/2023, 8:20 PMI have a branch with the prototype of a really expressive syntax, but mypy is the killer there
bitter-ability-32190
09/22/2023, 8:21 PMAnd it's not worth uprooting our type-checker for 😔
l
late-advantage-75311
09/22/2023, 8:23 PM👍 do you think you'll keep going on it? it would be fun to see what you came up with.
b
bitter-ability-32190
09/22/2023, 8:31 PMMaybe eventually if I feel like mypy would work. It's been a few months so I don't remember the details 😭
bitter-ability-32190
09/22/2023, 8:32 PMBut the rule callsite had a type like
option_value: SubsystemName.option_namel
late-advantage-75311
09/22/2023, 8:38 PMInteresting, so it wouldn't be returning an
OptionsInfolate-advantage-75311
09/22/2023, 8:39 PMAbout how many occurrences of this pattern do you think we have? (there are two mentioned in the issue)
b
bitter-ability-32190
09/22/2023, 8:43 PMInteresting, so it wouldn't be returning an OptionsInfo, or maybe it would be returning some unique generated type that satisfies the OptionsInfo interface, ok
Yeah I think that's the special sauce
bitter-ability-32190
09/22/2023, 8:43 PMAnd idk about usage. All subsystems options could be migrated
l
late-advantage-75311
09/22/2023, 8:57 PM+1. I am certainly timid in this area, and can see how it could be challenging to get type-checking to work. I'll look at some other issue. (no shortage of them 😛)
b
bitter-ability-32190
09/22/2023, 9:02 PMThe benefit is less restarting of the daemon when people are changing options. So definitely a plus, but not insanely boosting