This is an automated email from the ASF dual-hosted git repository.
alamb pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/datafusion.git
The following commit(s) were added to refs/heads/main by this push:
new 266daf8b3e doc: Add comments to clarify algorithm for `MarkJoin`s
(#16436)
266daf8b3e is described below
commit 266daf8b3ecaa489400dd0abe46af86a33b80367
Author: Jonathan Chen <chenleejonat...@gmail.com>
AuthorDate: Thu Jun 19 11:17:56 2025 -0400
doc: Add comments to clarify algorithm for `MarkJoin`s (#16436)
* doc: Add doc to clarify algorithm for `MarkJoin`s
* fix: fmt
* fix: NullEquality merge
---
datafusion/physical-plan/src/joins/symmetric_hash_join.rs | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/datafusion/physical-plan/src/joins/symmetric_hash_join.rs
b/datafusion/physical-plan/src/joins/symmetric_hash_join.rs
index d540b6d2a3..6dbe75cc0a 100644
--- a/datafusion/physical-plan/src/joins/symmetric_hash_join.rs
+++ b/datafusion/physical-plan/src/joins/symmetric_hash_join.rs
@@ -810,6 +810,21 @@ where
{
// Store the result in a tuple
let result = match (build_side, join_type) {
+ // For a mark join we “mark” each build‐side row with a dummy 0 in the
probe‐side index
+ // if it ever matched. For example, if
+ //
+ // prune_length = 5
+ // deleted_offset = 0
+ // visited_rows = {1, 3}
+ //
+ // then we produce:
+ //
+ // build_indices = [0, 1, 2, 3, 4]
+ // probe_indices = [None, Some(0), None, Some(0), None]
+ //
+ // Example: for each build row i in [0..5):
+ // – We always output its own index i in `build_indices`
+ // – We output `Some(0)` in `probe_indices[i]` if row i was ever
visited, else `None`
(JoinSide::Left, JoinType::LeftMark) => {
let build_indices = (0..prune_length)
.map(L::Native::from_usize)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@datafusion.apache.org
For additional commands, e-mail: commits-h...@datafusion.apache.org
