mihaibudiu commented on code in PR #4706:
URL: https://github.com/apache/calcite/pull/4706#discussion_r2647247006
##########
core/src/main/java/org/apache/calcite/rex/RexSimplify.java:
##########
@@ -3034,6 +3038,71 @@ private static class VariableCollector extends
RexVisitorImpl<Void> {
}
}
+ private static final Set<SqlKind> CONSTANT_VALUE_CONSTRUCTOR_KINDS =
+ EnumSet.of(
+ SqlKind.ARRAY_VALUE_CONSTRUCTOR,
+ SqlKind.MAP_VALUE_CONSTRUCTOR,
+ SqlKind.MULTISET_VALUE_CONSTRUCTOR);
+
+ private static @Nullable Pair<RexNode, RexNode> constantEquality(RexCall
call) {
+ final RexNode o0 = call.getOperands().get(0);
+ final RexNode o1 = call.getOperands().get(1);
+ if (RexUtil.isReferenceOrAccess(o0, true) && isConstant(o1)) {
+ return Pair.of(o0, o1);
+ }
+ if (RexUtil.isReferenceOrAccess(o1, true) && isConstant(o0)) {
+ return Pair.of(o1, o0);
+ }
+ return null;
+ }
+
+ private static boolean constantsEquivalent(RexNode node1, RexNode node2) {
+ if (Objects.equals(node1, node2)) {
+ return true;
+ }
+ if (!(node1 instanceof RexCall) || !(node2 instanceof RexCall)) {
+ return false;
+ }
+ final RexCall call1 = (RexCall) node1;
+ final RexCall call2 = (RexCall) node2;
+ if (call1.getKind() != call2.getKind()) {
+ return false;
+ }
+ switch (call1.getKind()) {
+ case MULTISET_VALUE_CONSTRUCTOR:
+ return multisetLiteralEquals(call1, call2);
+ case MAP_VALUE_CONSTRUCTOR:
+ return mapLiteralEquals(call1, call2);
+ default:
+ return false;
+ }
+ }
+
+ private static boolean multisetLiteralEquals(RexCall left, RexCall right) {
+ return HashMultiset.create(left.getOperands())
+ .equals(HashMultiset.create(right.getOperands()));
Review Comment:
we just had a conversation about comparing maps for ordering, the rule that
you implemented there is not compatible with this one: this can claim equality
while ordering can claim inequality.
The question is: are maps ordered or not? I think the answer depends on the
dialect.
--
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]
