Closed sonalmahajan15 closed 3 months ago
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 87.63%. Comparing base (
9012b93
) to head (6fb5e7a
).
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
[!WARNING]
❌ NilAway errors reported on stdlib are different 📉.3275 errors on base branch (main, 9012b93) 3271 errors on test branch (006a70c)
This PR fixes the bug which made rich check effect handling sensitive to return statement ordering, resulting in false positives for certain cases. The effect of ordering of statements is illustrated in the example below.
The problem is with no entry being created in
inferredMap
for non-nil error return fromretErr()
inStoreImplication
, since we don't propagate non-nilness forward. This creates a problem when the anonymous function parameter inFilterTriggersForErrorReturn
analyzes for nilability of return sites, where it incorrectly assumes that absence ininferredMap
implies unknown nilability. This PR corrects that logic.