On 11/12/2019 00:58, Stephen Oberholtzer wrote:
Harald, Interesting; I had a different take on it: https://github.com/Stevie-O/dash/commit/792fee674accdbf01767a6f3549f66e807bdf7c9 (I specifically avoided adding any loops.)
Nice work. I have not checked in detail, but the approach of keeping track of the right-most command that has already failed seems like a decent one.
I do not see a particular reason for avoiding loops: you do still have a check for each command in the pipeline, you just put it in a function that is already called once for each command in the pipeline. This does not suggest either approach over the other to me; both approaches should work equally well.
Cheers, Harald van Dijk
