Skip to content

Conversation

@manaldush
Copy link
Contributor

fix #8310

PG vanilla commit ab16418ee0e changed StatisticsObjIsVisible logic, so output for temp objects in propogate_statistic test was changed. Fixed normalized.sed for this.

m3hm3t and others added 8 commits October 27, 2025 18:51
Pg18 beta conf file updated

(cherry picked from commit c36410c)

Update image suffix in build and test workflow

Update image suffix in build configuration

Update image suffix in build configuration

Update image suffix in build configuration

(cherry picked from commit 7dbb946)

Update image suffix in build_and_test.yml to reflect latest development version

Update PostgreSQL version to 18beta3 in Dockerfile and CI workflow
… WHERE … IS NOT NULL (PG15–PG18) (citusdata#8139)

DESCRIPTION: Stabilize multi_insert_select expected: accept unqualified
columns in WHERE … IS NOT NULL

fixes citusdata#8133 


**Context**

* With PG18, ruleutils adds a GROUP RTE and improves column-name dedup.
As a side-effect, Vars that point at the GROUP RTE print as unqualified
column names even when `varprefix` is true.
* In Citus’ vendored `ruleutils_18.c` we already flattened GROUP Vars in
`targetList` and `havingQual`, but not in `jointree->quals`.
* For queries like `INSERT … SELECT … GROUP BY …`, Citus injects an
implicit null-guard on the group key in the WHERE clause. Because that
Var was still referencing the GROUP RTE, the deparser emitted `WHERE
(user_id IS NOT NULL)` instead of `WHERE (raw_events_first.user_id IS
NOT NULL)`, causing regress diffs only in grouped SELECTs.
* Related upstream change: PostgreSQL commit
`52c707483ce4d0161127e4958d981d1b5655865e` (ruleutils column-name de-dup
/ GROUP RTE exposure).

**What changed**

* Added an alternative expected file
`src/test/regress/expected/multi_insert_select_0.out` to keep CI green
across mixed environments where the qualified form may still be
produced.
The change in `merge_planner.c` fixes _unrecognized range table entry_
diffs in merge regress tests (category 2 diffs in citusdata#7992), the change in
`multi_router_planner.c` fixes _column reference ... is ambiguous_ diffs
in `multi_insert_select` and `multi_insert_select_window` (category 3
diffs in citusdata#7992). Edit to `common.py` enables standalone regress tests
with pg18 (e..g `citus_tests/run_test.py merge`).
Probably Colm's commits on `GROUP BY RTE` fixes have fixed the test
diffs we used to have.
PG vanilla commit ab16418ee0e changed StatisticsObjIsVisible logic, so
output for temp objects in propogate_statistic test was changed. Fixed
normalized.sed for this.
@m3hm3t m3hm3t force-pushed the m3hm3t/pg18_beta_confs branch 6 times, most recently from 5565920 to e302318 Compare November 10, 2025 08:19
@m3hm3t m3hm3t force-pushed the m3hm3t/pg18_beta_confs branch 4 times, most recently from 8c37f62 to d4fcdcb Compare November 20, 2025 08:21
@m3hm3t m3hm3t force-pushed the m3hm3t/pg18_beta_confs branch 5 times, most recently from bdd4c24 to 206fc6e Compare December 5, 2025 13:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants