adriangb commented on code in PR #20925:
URL: https://github.com/apache/datafusion/pull/20925#discussion_r2990369821
##########
datafusion/datasource-parquet/src/row_filter.rs:
##########
@@ -579,6 +595,59 @@ pub(crate) fn build_parquet_read_plan(
)))
}
+/// Builds a unified [`ParquetReadPlan`] for a set of projection expressions
+///
+/// Unlike [`build_parquet_read_plan`] (which is used for filter pushdown and
+/// returns `None` when an expression references unsupported nested types or
+/// missing columns), this function always succeeds. It collects every column
+/// that *can* be resolved in the file and produces a leaf-level projection
+/// mask. Columns missing from the file are silently skipped since the
projection
+/// layer handles those by inserting nulls.
Review Comment:
If we never expect to be called with columns missing from the file,
shouldn't we error if we are?
--
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]
