LuciferYang commented on PR #58333: URL: https://github.com/apache/spark/pull/58333#issuecomment-5450984044
Both addressed. **`toSQLId(Seq(...))`**: switched, and both consequences reproduce. `parseAttributeName` splits on `.`, so a column literally named `x.y` rendered as `` `x`.`y` ``, and a name holding a backtick reaches `throw e` at `AttributeNameParser.scala:50`, so the user got `INVALID_ATTRIBUTE_NAME_SYNTAX` with ``name -> z`w`` while this message was being built. Those names do arrive here unparsed: `CatalogUtils.normalizePartCols` and `normalizeBucketSpec` match against `schema.map(_.name)` by plain resolution. Added a test with columns `x.y` and ``z`w`` pinning `` `x.y` `` and `` `z``w` ``; run against the `String` overload it fails with exactly the error above. Existing assertions unchanged, as you expected. **Parameter names**: took the point and went a bit further. The three keys are now `bucketColumn`, `sortColumn` and `partitionColumns`, the singular and plural of one noun. Going by what is published rather than what is in the file: `<partitionColumnNames>` does appear twice, but both are legacy keys (`_LEGACY_ERROR_TEMP_1232`, `_LEGACY_ERROR_TEMP_2182`), and `reservedColumnNames` is the only published `...ColumnNames`, while 19 named conditions use `...Columns` for a list, `CANNOT_UPDATE_PARTITION_COLUMNS` with `requestedPartitionColumns` / `existingPartitionColumns` among them, and five use `...Column` for a single column. This also drops `...Col`, which had no published precedent outside this PR, so it is a better answer than the one I gave earlier when I said I would keep it. Say the word if you would rather match `BucketSpec` field for field with `partitionColumnNames`. -- 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]
