Completion ordering by client #473
Open
+40
−10
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit introduces a new ranking algorithm: all the completion candidates are first ranked by the weight of the client and only then withing each specific client they are ranked according to a more sophisticated metric. This change allows one to have guaranteed consistent client ordering in the completion results. This can be achieved simply by assigning appropriate `weight_adjust` values to the clients. For example, setting `weight_adjust = 2.0` next to the LSP client will guarantee that the LSP results will be always on top.
ms-jpq
force-pushed
the
coq
branch
4 times, most recently
from
0265a02
to
7783671
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Hey!
This PR introduces an alternative way to order the completion results. This idea was already suggested by @elkowar in #92 and by me in #241. I've read the discussion in #92 and while I agree with you that giving unconditional preference to some clients is, generally speaking, not a good idea, I find that for the languages I personally use LSPs give me good enough suggestions that I would love to see them on top all the time. I've tried to play with the client weights, but no matter what I do I cannot achieve this.
This PR implements this unconditional client-based ranking approach -- aliased in the code as
stratified-- as an alternative to the current one -- aliased asuniform. Withstratifiedapproach we first sort the completion results by client (usingweight_adjustas the client priority) and then withing the client using all the metrics used before. The choice of a specific ranking strategy is determined by a new settings parameterranking.The implementation at the moment is a bit ad hoc and it can be properly generalized in future. Also, there are neither documentation nor tests for that change, but I will be happy to add them in case you decide to accept the PR.