Ragnar Groot Koerkamp

> Well, you might just log what you're doing, since you're already logging anyway. Something like "100 shards, keys will be re-read 100 times (switch to offline sharding for single-pass)"....

I pushed some more commits. 1. `sort_by_cached_key` sorts pairs `(key, idx in vec to be sorted)`, but in this case it's more efficient to directly sort `(key, suffix array idx)`,...

Did some statistics on the LCP of consecutive suffix array elements. Looking at this, bucketing by 9 elements twice followed by comparison based sorting may be promising. That way most...

oh right; that may be sufficient indeed. Would be good to document this.

Hmm right; doesn't really matter I suppose, so 0 or 1 I'd say. This becomes relevant though once you provide `extend` functionality. In my case, I don't know the right...

I have this (maybe ugly) pattern where I have ```rust struct PTHash { n: usize, ..., data: P } impl PTHash { pub fn new(n: usize) -> Self { let...

One possible solution to this may be to publish things as multiple creates with a smaller scope, so that you don't have to commit to a version for everything. That...

Re simd, that's also one of the big reasons A*PA uses unstable. There actually was a small backwards incompatible change recently which was kinda annoying. But it's so convenient to...

Hey! Thanks for your interest in ptrhash :) I'm not currently planning to support this, but if you have I compelling usecase you may be able to convince me otherwise....

If the lookup is also at compile time, a vector of `TypeId` with linear search should for sure be fine. In case you'll also do queries at runtime I could...