> Your "pipeline" isn't actually connecting the processes to each other. You first connect the processes to each other, and only then add them to a "pipeline",

> Your "pipeline" isn't actually connecting the processes to each other. You first connect the processes to each other, and only then add them to a "pipeline", whose primary (only?) responsibility is to ensure that each process runs in a different goroutine. Exactly! > This much doesn't really need parametric polymorphism, let alone generalized algebraic data types. However, calling it "pipeline" is kind of misleading, because the actual "piping" Ah, yes, didn't think about that, but yeah, but that's a good point! I should consider a better name for this ... Thanks for pointing it out!