Alena0704 commented on PR #1961: URL: https://github.com/apache/cloudberry/pull/1961#issuecomment-5573818508
> It looks like something wrong with subselect.out > > ``` > --- /__w/cloudberry/cloudberry/src/test/regress/expected/subselect.out 2026-09-04 05:52:26.139317294 -0700 > +++ /__w/cloudberry/cloudberry/src/test/regress/results/subselect.out 2026-09-04 05:52:26.176315671 -0700 > @@ -2076,41 +2071,44 @@ > ); > QUERY PLAN > ___________ > - Gather Motion 3:1 (slice4; segments: 3) > + Gather Motion 3:1 (slice1; segments: 3) > Output: tl1.a, tl1.b, tl1.c, tl1.d > -> Hash Join > Output: tl1.a, tl1.b, tl1.c, tl1.d > + Inner Unique: true > Hash Cond: ((tl1.b = "Expr_SUBQUERY".csq_c1) AND (tl1.c = "Expr_SUBQUERY".csq_c0)) > - -> Redistribute Motion 3:3 (slice1; segments: 3) > + -> Seq Scan on public.tl1 > Output: tl1.a, tl1.b, tl1.c, tl1.d > - Hash Key: tl1.c > - -> Seq Scan on public.tl1 > - Output: tl1.a, tl1.b, tl1.c, tl1.d > -> Hash > Output: "Expr_SUBQUERY".csq_c1, "Expr_SUBQUERY".csq_c0 > - -> Subquery Scan on "Expr_SUBQUERY" > + -> Broadcast Motion 3:3 (slice2; segments: 3) > Output: "Expr_SUBQUERY".csq_c1, "Expr_SUBQUERY".csq_c0 > - -> HashAggregate > - Output: tl2.b, max((max(tl2.a))) > - Group Key: tl2.b > - -> Redistribute Motion 3:3 (slice3; segments: 3) > - Output: tl2.b, (max(tl2.a)) > - Hash Key: tl2.b > - -> HashAggregate > - Output: tl2.b, max(tl2.a) > - Group Key: tl2.b > - -> Hash Join > - Output: tl2.b, tl2.a > - Hash Cond: (tl4.d = tl2.d) > - -> Broadcast Motion 3:3 (slice2; segments: 3) > - Output: tl4.d, tl4.a > - -> Seq Scan on public.tl4 > - Output: tl4.d, tl4.a > - -> Hash > - Output: tl2.b, tl2.a, tl2.d > - -> Seq Scan on public.tl2 > - Output: tl2.b, tl2.a, tl2.d > -GP_IGNORE:(36 rows) > + -> Subquery Scan on "Expr_SUBQUERY" > + Output: "Expr_SUBQUERY".csq_c1, "Expr_SUBQUERY".csq_c0 > + -> Finalize GroupAggregate > + Output: tl2.b, max(tl2.a) > + Group Key: tl2.b > + -> Sort > + Output: tl2.b, (PARTIAL max(tl2.a)) > + Sort Key: tl2.b > + -> Redistribute Motion 3:3 (slice3; segments: 3) > + Output: tl2.b, (PARTIAL max(tl2.a)) > + Hash Key: tl2.b > + -> Streaming Partial HashAggregate > + Output: tl2.b, PARTIAL max(tl2.a) > + Group Key: tl2.b > + -> Hash Join > + Output: tl2.b, tl2.a > + Hash Cond: (tl4.d = tl2.d) > + -> Broadcast Motion 3:3 (slice4; segments: 3) > + Output: tl4.d > + -> Seq Scan on public.tl4 > + Output: tl4.d > + -> Hash > + Output: tl2.b, tl2.a, tl2.d > + -> Seq Scan on public.tl2 > + Output: tl2.b, tl2.a, tl2.d > +GP_IGNORE:(39 rows) > ``` Thanks for catching this - the failure is in the expected output, not in the fix itself. This patch was cherry-picked from open-gpdb (a Greenplum 6 fork on a PG 9.4-era core), and I forgot to regenerate it against Cloudberry (PG 14.8). The two planners simply print a different plan for the same query. I updated it and now it is fine. -- 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]
