[IMPALA-11168] Add pseudo-columns for Impala

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Fix Version/s: Not Applicable
• Component/s: Backend, Frontend
• Epic Color: ghx-label-4
• Assignee: Unassigned
• Reporter: Zoltán Borók-Nagy
• Created: 08/Mar/22 15:15
• Updated: 05/Dec/22 09:55
• Resolved: 19/Sep/22 11:23

Description

Hive supports a few pseudo-columns (virtual column):
https://cwiki.apache.org/confluence/display/hive/languagemanual+virtualcolumns

It would be useful to add support for a few pseudo-columns for Impala as well.
E.g. to implement support for Iceberg position-based deletes we'd need the file name and ordinal position for each row.

Hive has INPUT__FILE__NAME which returns full URI of the data file. Unfortunately it doesn't have a pseudo-column for the ordinal position of rows. So we need to introduce a new pseudo-column for that, probably with the name ORDINAL__POSITION__INSIDE__FILE, or stg. like that.

Issue Links

• duplicates: IMPALA-11350 Add virtual column FILE__POSITION
• duplicates: IMPALA-801 Add function or virtual column for file name