Performance

[theHamsta]

Describe the bug

Current search for textobjects is too slow and could be more flexible (counts, textobjects in same line).

Solution

Avoid iter_prepared_matches in find_best_match. Use find_best_matches to allow the use of v:count

[theHamsta]

We could also do partial invalidations of the cache

[ilan-schemoul]

Is there any update on this front ?
In c files used at work (usually between 5K-15k lines) using textobjects freeze the computer for 15 seconds.
If there's some low hanging fruit I can maybe take a shot but in the same time I know you're waiting for TS 1.0 before accepting some PR

[ilan-schemoul]

I was curious what was happening also with swapping being slow. I tried swapping nodes on a 5K line c file and what take all the time is swap_nodes() which itself calls vim.lsp.util.apply_text_edits({ edit1, edit2 }, bufnr, "utf-8") which is very slow.
I didn't know that what was called was the lsp server. I thought it was pure treesitter. But anyway not this package fault for swap...
But still very annoying to have my computer freeze for 10 seconds whenever I forget not to use this package