Prio per snippet and distance to cursor ?

[MarcWeber]

The keyword snippet is the most often used word.
Yet it doesn't show up first.

If you have code like

snippet is_cross
	stdenv.hostPlatform != stdenv.buildPlatform

and you write one line below starting with i it should find is_cross first
because local or recently defined stuff is very very very very likely to be used again.

Yes, I found function(...) return cmp_buffer:compare_locality(...) end,
but doesn't seem to work. Like its not proposing words nearby but 60 lines apart when starting with chars of a word or is it caused by fuzzy matching ?

So it might even be useful to track recently added texts and take that into account.

Changing = true to + 1 is easy.
What I didn't find is the weight per entry. I only find label = "foo"

Is it already supported that completion entries from a source can have their own weights ?

I nvim-cmp I only find ---@Class lsp.CompletionItem
which doesn't have prio field ?

preselect seems to be closest.

The docs talk about orig_score which seems to be gone in the code.

Anyway cmp/cmp-buffer are very helpful and speed up a lot.