Are there any existing tests that cover the expected behaviour of the old hook?

This restores the behavior we had earlier until the bug in kbar library was discovered: #61278.

it looks like they swapped out for fuse.js as the fix? not ideal 😅. it's slow for many items :(

https://leeoniya.github.io/uFuzzy/demos/compare.html?libs=uFuzzy,fuzzysort,QuickScore,Fuse&search=super%20ma

hopefully they have the correct fuse.js settings that don't return nonsense results, too.

https://github.com/leeoniya/uFuzzy?tab=readme-ov-file#search-quality

btw, if you're looking to simplify, we now have a fuzzySearch function that implements all the necessary conditions: https://github.com/grafana/grafana/blob/main/packages/grafana-data/src/utils/fuzzySearch.ts

btw, if you're looking to simplify, we now have a fuzzySearch function that implements all the necessary conditions: https://github.com/grafana/grafana/blob/main/packages/grafana-data/src/utils/fuzzySearch.ts

Yeah, that was my original commit, but then I saw a TODO in useMatchers, so decided to revert that change instead.

the reason i wrote uFuzzy was because of my frustration with Fuse (both startup/indexing time and mem/cpu perf and result quality). ufuzzy also has typo tolerance, which i dont think fuse supports. maybe you can see why i'm not thrilled about the change 😅. your call, tho. i'd suggest at least testing it with 10k-20k items and compare results/ordering.

/0.02

the reason i wrote uFuzzy was because of my frustration with Fuse (both startup/indexing time and mem/cpu perf and result quality). ufuzzy also has typo tolerance, which i dont think fuse supports. maybe you can see why i'm not thrilled about the change 😅. your call, tho. i'd suggest at least testing it with 10k-20k items and compare results/ordering.

/0.02

I did some quick tests and it seems like the fuzzySearch function is way faster, so let's use that.