github-actions[bot] commented on code in PR #66206:
URL: https://github.com/apache/doris/pull/66206#discussion_r3681847860


##########
be/src/format_v2/column_mapper.cpp:
##########
@@ -2004,6 +2125,7 @@ Status 
TableColumnMapper::_create_mapping_for_column(const ColumnDefinition& tab
     mapping->global_index = global_index;
     mapping->table_column_name = table_column.name;
     mapping->table_type = table_column.type;
+    mapping->variant_access_paths = table_column.variant_access_paths;

Review Comment:
   [P2] Finish leaf projection for a Variant nested under a complex column.
   
   The parser now stores `{"x"}` on the nested `ColumnDefinition` for 
`info.payload['x']`, but recursive mappings are built through 
`_create_direct_mapping()`, which never copies that field (only this root 
constructor does). There are two further root-only gates: `add_scan_column()` 
calls `build_variant_projection()` only for the root mapping, while 
`build_complex_projection()` treats a child Variant with no semantic children 
as `project_all_children`; and `finalize_variant_leaf_projections()` only 
validates/counts top-level schema nodes whose kind is `VARIANT`. Consequently 
the new regression at `test_iceberg_variant_read.groovy:418-429` cannot make 
`VariantLeafProjections` positive for its sole nested predicate—the query stays 
correct only by reading the full Variant subtree. Please propagate the child 
paths, recurse into the Variant projection builder and Parquet finalizer, and 
add recursive mapper/reader coverage for STRUCT/ARRAY/MAP.



-- 
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]

Reply via email to