twuebi opened a new pull request, #1299:
URL: https://github.com/apache/iceberg-go/pull/1299

   compacting a pyiceberg created table fails because of this
   
   Summary
   - coerce LIST/LARGE_LIST records to the variant useLargeTypes requests in 
arrowProjectionVisitor.List, converting offsets rather than relabeling
   - add a regression covering large_list->list downcast and list->large_list
   
   Why
   ReadTasks returns a writer schema built with useLargeTypes (small list by 
default), but the visitor copied the reader's list variant verbatim. The 
arrow-go bump in #1238 taught getNestedFactory to reconstruct a parquet file's 
embedded ARROW:schema, so list columns from pyiceberg-written files (which 
declare large_list) now read back as large_list instead of list. The projected 
records no longer match the small-list writer schema, so the parquet file 
writer rejects every batch with "record schema does not match writer's" -- 
surfacing in compaction as a group that writes no files and a table that never 
gets its replace snapshot. Primitives dodge this because castIfNeeded already 
downcasts large_string/large_binary.
   
   Testing
   - go test ./table -run TestToRequestedSchema -count=1
   - go test ./table -count=1


-- 
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