plusplusjiajia opened a new pull request, #10033: URL: https://github.com/apache/paimon/pull/10033
## Problem Two legitimate queries crash under query authorization. A mask whose transform reads a column outside the projection. The read type carries only the projected columns, so the masking reader raises `Column masking refers to field 'salt' which is not present in output row type`. Java widens the read with what the rules need (`ReadTransform.requiredFields` → `authFields`); Python widened only for filter operands. A nested ROW projection combined with a rule on an unprojected column. The widened read type gains a field while `nested_name_paths` does not, and the outer extraction fails with `name_paths length 2 does not match output_fields length 3`. ## Change `get_extra_fields_for_filter` becomes `get_extra_fields` and widens for mask inputs too, following Java's worklist: a mask whose target is unreadable is inert and widens nothing. The widened columns are plain top-level ones, so `_build_split_read` gives each a one-segment name path and passes the extended paths on to the top-level widening. -- 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]
