github-actions[bot] commented on code in PR #67891:
URL: https://github.com/apache/doris/pull/67891#discussion_r4003738157
##########
fe/fe-core/src/main/java/org/apache/doris/nereids/rules/rewrite/ForeignKeyContext.java:
##########
@@ -142,21 +143,27 @@ public boolean isForeignKey(Set<Slot> key) {
}
public boolean isPrimaryKey(Set<Slot> key) {
- return primaryKeys.containsAll(
- key.stream().map(s ->
slotToColumn.get(s)).collect(Collectors.toSet()));
+ return !key.isEmpty() && activePrimaryKeySlots.containsAll(key);
}
void putSlot(SlotReference slot, TableIf table) {
if (!slot.getOriginalColumn().isPresent()) {
return;
}
Column c = slot.getOriginalColumn().get();
- slotToColumn.put(slot, new QualifiedColumn(table, c));
+ QualifiedColumn qualifiedColumn = new QualifiedColumn(table, c);
+ slotToColumn.put(slot, qualifiedColumn);
+ if (declaredPrimaryKeys.contains(qualifiedColumn)) {
+ activePrimaryKeySlots.add(slot);
Review Comment:
[P1] Validate scan semantics before activating PK slots
This treats every catalog scan as both complete and unique, but two
independent reachable cases violate those assumptions. First, `PARTITION`,
`TABLET`, `TABLESAMPLE`, and filtered direct-index restrictions live on
`LogicalOlapScan` rather than in a `LogicalFilter`; for a self-FK table with
rows `(1,1),(2,2)`, selecting a primary partition containing only `id=1` makes
the join return `1`, while eliminating that branch returns `1,2`. Second,
raw-version modes (`skip_delete_bitmap`, `skip_storage_engine_merge`,
`read_mor_as_dup_tables`) and a bare `@incr` MIN_DELTA scan can expose
duplicate versions of a declared key, so the join can emit one retained foreign
row twice while elimination emits it once. `LogicalOlapScan.computeUnique`
already suppresses uniqueness for those modes, but this direct constraint
activation bypasses that trait. Please activate PK slots only when the concrete
scan covers the full base relation and preserves declared-key uniqueness, and
add subset plus duplic
ate-version regressions.
##########
fe/fe-core/src/main/java/org/apache/doris/nereids/rules/rewrite/ForeignKeyContext.java:
##########
@@ -142,21 +143,27 @@ public boolean isForeignKey(Set<Slot> key) {
}
public boolean isPrimaryKey(Set<Slot> key) {
- return primaryKeys.containsAll(
- key.stream().map(s ->
slotToColumn.get(s)).collect(Collectors.toSet()));
+ return !key.isEmpty() && activePrimaryKeySlots.containsAll(key);
}
void putSlot(SlotReference slot, TableIf table) {
if (!slot.getOriginalColumn().isPresent()) {
return;
}
Column c = slot.getOriginalColumn().get();
- slotToColumn.put(slot, new QualifiedColumn(table, c));
+ QualifiedColumn qualifiedColumn = new QualifiedColumn(table, c);
+ slotToColumn.put(slot, qualifiedColumn);
+ if (declaredPrimaryKeys.contains(qualifiedColumn)) {
+ activePrimaryKeySlots.add(slot);
+ }
}
void putAlias(Slot newSlot, Slot originSlot) {
if (slotToColumn.containsKey(originSlot)) {
slotToColumn.put(newSlot, slotToColumn.get(originSlot));
+ if (activePrimaryKeySlots.contains(originSlot)) {
+ activePrimaryKeySlots.add(newSlot);
Review Comment:
[P1] Preserve predicate state across active aliases
Consider `Project(id#1 AS pk#3) -> Filter(parent_id#2 = 1) ->
Scan(self_ref)` joined to another `self_ref` scan on `pk#3 = f.parent_id#5`.
`addFilter` records the non-key restriction under `id#1`, while this branch
activates `pk#3` without creating a `slotWithPredicates` entry. Since the
project drops `parent_id#2`, predicate pull-up cannot expose the restriction at
the join, and `isPredicateCompatible` treats `pk#3` as unfiltered. With rows
`(1,1),(2,2)`, eliminating the primary branch changes the correct output `1` to
`1,2`. Please propagate and rewrite the complete origin predicate set through
every alias hop, or conservatively keep an alias inactive when its predicate
proof cannot be preserved; add the combined alias-plus-hidden-filter regression.
--
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]
