sunchao commented on PR #2276: URL: https://github.com/apache/iceberg/pull/2276#issuecomment-1285799673
@aokolnychyi so in the case of non-storage partition join, the random values of `c3` in file `T_A` will be used if we have a join condition like `t.c3 = s.c3`, instead of `null`, is that correct? seems in this case the column `c3` of the target table evolved from a non-partition column to a partition column. Also for file `T_A`, the method `ContentFile.partition` will return `(part 1 = A, part2 = A)`? > It seems we can report KeyGroupPartitioning to Spark only on columns that were present in all partition specs that are being queried. Any thoughts? I think this probably is a safe solution. We can also derive the common columns after Iceberg has done partition pruning, so in that case if the file `F_A` is not selected as result of pruning, we can still report `c1`, `c2` and `c3` to Spark. -- 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]
