DeviousCardi opened a new pull request, #26350: URL: https://github.com/apache/datafusion/pull/26350
## Which issue does this PR close? Closes #26221. ## Rationale for this change `sum(x)` and `sum(CAST(x AS FLOAT))` get the same schema name, so using both in one query fails even with separate aliases. The collision happens in the aggregate node before the projection applies those aliases. ## What changes are included in this PR? Use the display name for colliding aggregates when it differs from their schema name. Expression rebasing then resolves the original expression to the renamed aggregate column. This keeps the naming change local to SQL aggregation. Queries without distinct output aliases still get the existing duplicate projection name error. ## What is the testing strategy for this PR? SQLLogicTest cases cover both argument orders, multiple casts, `TRY_CAST`, `COUNT`, grouping, `HAVING`, and `ORDER BY`. The existing `approx_distinct` regression now checks successful execution, and an `EXPLAIN` case checks the aggregate names. ## Are there any user-facing changes? Queries that aggregate a column and a cast of it can now plan successfully when their output names are distinct. AI assistance: Claude Code generated the patch and tests; Codex read the diff and prepared this description. The scope assumes the fix belongs in SQL aggregate naming rather than changing cast schema names globally. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
