De-deduplicate completion items + improve sorting #2638
Conversation
* Remove completion items that are duplicates for all attributes other than `sortText` (which is always different but is just some machine-readable hash for sorting by). * Continue to sort abbreviations below full matches. * Sort matching `filterText`/`labels` so they appear next to each other. * Otherwise use `sortText` given by the language server.
|
The main motivation for this PR was to remove the duplicates coming back from However, I found that the sort order of results was odd (which makes filtering duplicates more expensive, and just looks strange). Perhaps they make some sense straight from the language server, but Oni then filters as the user types, so retaining the sort order from the language server makes even less sense. Hence why I thought it reasonable to "fix" the sort order for filtered completions. |
| @@ -3,6 +3,7 @@ | |||
| * | |||
| * Selectors are functions that take a state and derive a value from it. | |||
| */ | |||
| import * as _ from "lodash" | |||
There was a problem hiding this comment.
@feltech if you import just the specific functions you need rather than _ from lodash this will help reduce our bundle size which is something we should aim for but probably don't get right in many situations
e.g.
import uniq from "lodash/uniq"
// or if type issues
import * as omit from "lodash/omit"
Codecov Report
@@ Coverage Diff @@
## master #2638 +/- ##
==========================================
+ Coverage 45.43% 45.51% +0.07%
==========================================
Files 361 361
Lines 14563 14591 +28
Branches 1912 1919 +7
==========================================
+ Hits 6617 6641 +24
- Misses 7721 7723 +2
- Partials 225 227 +2
Continue to review full report at Codecov.
|
cquery(C++) completions.filterText/labelsso they appear next to each other.sortTextgiven by the language server.