Open treeowl opened 1 year ago
@jaspervdj The biggest question for me is how to benchmark this. I don't understand the benchmark suite of this package; can you give me some guidance?
@treeowl It's been some years since I looked at that myself! It looks like this is really a benchmark suite to compare different implementations, not really to look for regressions and speedups for the same implementation. That means we can't easily benchmark things like split
if they don't have e.g. a HashPSQ
implementation. Should we add a new simple benchmark suite? We can probably reuse a bunch of the existing code.
That sounds reasonable to me.
Start work on adding set/map-like functions. Currently,
split
,splitMaybe
, andunion
.