sql: assign locking strength and wait policy in ZigzagJoinerSpec_Side #81350
Conversation
This was missed in a8b951b. I didn't notice until trying to backport. What would be the right level to add testing for this plumbing?
|
This change is |
There was a problem hiding this comment.
Nice catch!
Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status:complete! 0 of 0 LGTMs obtained (waiting on @nvanbenschoten)
-- commits line 6 at r1:
Hm, I think the best option would be to include locking information into DistSQL diagrams (in execinfrapb/flow_diagram.go), and then write an execbuilder logic test using EXPLAIN (DISTSQL). This would probably mean adding the locking info for all four processors (TableReader, JoinReader, InvertedJoiner, ZigzagJoiner) for consistency, and it might clutter the diagrams, but probably it seems ok. Another option is having an end-to-end test (i.e. a regular logic test) where a query encounters contention because of this locking, but the diagram option seems cleaner.
This was missed in a8b951b. I didn't notice until trying to backport.
What would be the right level to add testing for this plumbing?