Closed elshize closed 9 months ago
All modified and coverable lines are covered by tests :white_check_mark:
Comparison is base (
994d101
) 93.21% compared to head (30ee2a7
) 93.23%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Right, I don't think thresholds should be any different than queries, but I'll give it another look.
Good idea about regression tests. There's a test docker image that I created for that, but haven't finished. Maybe it's a good idea to continue with it to make it easier to repeat in the future (or even automate).
I need to fix the conflicts, and after that, I'll run the docker that was just merged in the other PR to evaluate and see if there's any regression.
@JMMackenzie The regression test was successful. Are you ok merging it?
Great, let's merge!
Weights are now stored together with term IDs and resolved at construction time according to one of the policies. In our tools, we use the default policy that removes duplicates and sets the weight to the number of occurrences of the term in a query. Other policies are, for the time being, only available programmatically via the library API.
Some legacy code used to parse and process queries has been removed in favor of the text analyzer and the new query parser.
Because weights are resolved when a query object is created, I also refactored creating the cursors: now the weight is simply taken from the query.
Fixes #501