You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Bug] Deleted rows take full-text top-k slots in data-evolution tables #403
I searched in the issues and found nothing similar.
Paimon-cpp version
main at commit d7c5f607900ffd63b2e067800dc4b9097c24d888.
Minimal reproduce step
Create a data-evolution table with row tracking and deletion-vectors.enabled = true. Write 10 rows whose text column matches apple with different relevance, then build a full-text global index on that column.
Delete the three rows that match best, using deletion vectors.
Search for apple with limit = 3 and with_score = true through GlobalIndexScan::CreateReader → VisitFullTextSearch.
Pass the result to ScanContextBuilder::SetGlobalIndexResult, then scan and read.
What doesn't meet your expectations?
The index still ranks the deleted rows, so they take all three top-k slots. The read then drops them, because deletion vectors are applied only at read time (src/paimon/core/operation/data_evolution_split_read.cpp). The query returns 0 rows instead of 3. Nothing in the global index scan path builds an include set from deletion vectors.
Activity