(As you have probably seen, the

test

and

fmt/lint/fix

goals have custom batching logic in their respective APIs.)

The pattern to likely use is to define request / response types for each batch (with a rule to compute the batch). Decompose the work into those batch requests. Then just use

MultiGet

/

concurrently

(for rule by name) to compute them concurrently.

Of course the problem in your case is whether you have all of the targets "at once" to be able to partition.

And this will depend on which goal it is and whether that goal's API feeds your rules one target at a time or all of the targets in a single request.

Which goal did you have in mind to do in batches?

this makes sense to me

i was overcomplicating it in my head with the partition stuff

but just MultiGet on some set of the targets is good enough

thank you!