also, now that twitter isn't consuming new releases of pants, do we have any performance regression testing for pants as a whole? if i was to unify glob matching in a later PR, i would want to make sure that i don't introduce any regressions to pants's filesystem traversal code. #9779 introduces criterion benchmarks for several snapshot operations -- is that kind of benchmarking sufficient to cover the use cases that e.g. pantsbench inside twitter did (if we continue to expand it)? cc @witty-crayon-22786
it's probably good that we're not having to worry about slowing down twitter's mostly-v1 use cases now, so it's not really a bad thing to be in the situation of not having existing perf testing to lean on, i think