Yes. We are limited by the parallelism of the stream underlying the generator as far as producing elements, but we could definitely check multiple samples in parallel. I think we just need to implement a variant of checkStream like so:

def checkStreamPar[R, R1 <: R, E, A](stream: ZStream[R, Nothing, Sample[R, A]], maxShrinks: Int = 1000)(
  test: A => ZIO[R1, E, TestResult]
): ZIO[R1, E, TestResult] =
  stream.zipWithIndex
    // use mapMPar instead of mapM to perform test effects in parallel
    .mapMPar(???) { 
      case (initial, index) =>
        initial.foreach { input =>
          test(input).traced
            .map(_.map(_.copy(gen = Some(GenFailureDetails(initial.value, input, index)))))
            .either
        // convert test failures to failures to terminate parallel tests on first failure
        }.flatMap(sample => sample.value.fold(_ => ZIO.fail(sample), _ => ZIO.succeed(sample)))
    }
    // move failures back into success channel for shrinking logic
    .catchAll(sample => ZStream.succeed(sample))
    // shrinking logic unchanged and could be factored out into a separate combinator

I think the biggest challenge here is just method names for all these variants. We probably want to allow the user to control the degree of parallelism here and we are already using the N suffix to denote the number of samples to check. 😂

Uh oh!

zio-test: Parallel checkM #3688

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions