lsp--capf-cached-items: dont group on prefix anymore #1774
Conversation
|
Can you do this PR against plists branch? We should focus on it otherwise it will never be merged. |
kiennq
force-pushed
the
bug/dont-group-onprefix
branch
from
20e366c to
93b1111
Compare
|
Done |
| (when (string-match fuz-query cand) | ||
| (setq cand (copy-sequence cand)) | ||
| (put-text-property 0 1 'match-data (match-data) cand) | ||
| (put-text-property 0 1 'completion-score (lsp--fuzzy-score query cand) cand) |
There was a problem hiding this comment.
can we calculate the completion-score here and remove the OR bellow? Also, access completion-score directly instead of using the get-text-property.
| (plist-put fuz-queries start-point s)) | ||
| s)))) | ||
| (when (string-match fuz-query cand) | ||
| (setq cand (copy-sequence cand)) |
There was a problem hiding this comment.
why this is needed? We should strive to reduce memory usage in this critical section. Note that dart language server might return 20k+ completion items.
There was a problem hiding this comment.
I modified the cached candidates to adding match-data and stuff so just copy the candidates to make sure.
| (put-text-property 0 1 'completion-score (lsp--fuzzy-score query cand) cand) | ||
| cand))) | ||
| items)) | ||
| (-sort (-on #'> (lambda (o) |
There was a problem hiding this comment.
we should remove the usage of -sort in favour of sort in completion hotpath.
| (put-text-property 0 1 'completion-score (lsp--fuzzy-score query cand) cand) | ||
| cand))) | ||
| items)) | ||
| (-sort (-on #'> (lambda (o) |
There was a problem hiding this comment.
-on would introduce some additional funcall - can we replace it with lambda?
|
And there is one more thing that we should do - use |
Good point, I will do that |
Hm, we should be careful, because this might not work, e. g. when user does company select next item we should not cancel the calculations... |
2 participants
Since we're using
lsp--fuzzy-score, there's no need for grouping candidates prefix anymore.