ClickHouse / ClickHouse/pg_clickhouse
Push down all JOIN expressions as JOIN expressions
- Dominant language
- C
- Stars
- 283
- Forks
- 21
- Avg merge
- 2d 6h
- Merged PRs (30d)
- 11
Description
Currently, `foreign_join_ok()` uses only [JOIN key expressions](https://clickhouse.com/docs/sql-reference/statements/select/join#on-section-conditions) in INNER JOIN and SEMI JOIN expressions: that is, only expressions that reference both the right and left keys using the equality operator. The rest are appended to remote conditions for use in the `WHERE` clause.
I suspect that ClickHouse used to allow only join key expressions but opened it up to additional expressions later. Today at least one must be a join key expression but other expressions are allowed.
Would it make sense to allow all expressions to be pushed down as JOIN expressions, rather than filtering some to remote conditions (the WHERE clause)? Seems like there might be some efficiencies in that, although maybe the ClickHouse optimizer adjusts appropriately to save cycles and iops?
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by tracing foreign_join_ok() and the code that moves non-key expressions into remote conditions. Compare how INNER JOIN and SEMI JOIN conditions are handled, then identify existing tests or query-planning checks for these paths; the issue is complete only when the intended pushdown behavior and its efficiency implications are established.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- c, postgresql, sql
- Domain
- backend, databases
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100