3.x: Upgrade bundled orafce from 4.9.2 to 4.16.13
Les mainteneurs répondent en général sous 2 jours
Évaluation
- Difficulté
- 5/5
- Temps estimé
- Plus d'une semaine
- Accessibilité débutants
- 25/100
Piste de recherche
Start with gpcontrib/orafce/README.gp and the current regression schedule in gpcontrib/orafce; identify which of the 21 SQL tests can run in MPP before expanding coverage. Then compare the local patches with upstream tag VERSION_4_16_13 and preserve applicable changes. Done means the update chain, fresh install, upgrade and pg_upgrade paths pass, licensing checks stay green, and release notes describe user-visible changes.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Description
Summary
The bundled gpcontrib/orafce is 4.9.2 (META.json:5; orafce.control default_version = '4.9', imported 2024-07). Propose upgrading it to 4.16.13 — the current release of the upstream series that supports PostgreSQL 16 — as part of 3.0.
orafce is the Oracle-compatibility layer many migration workloads depend on, so refreshing it is user-visible content for a major release, not just dependency hygiene.
Current state
- 4.9.2 is roughly seven minor releases and 2+ years behind upstream (4.10 → 4.16.x). 4.16.13 is currently packaged for PG16 by PGDG and distributions.
- The module is imported as a git subtree of
orafce/orafce(workflow documented ingpcontrib/orafce/README.gp), so the upgrade is agit subtree pullto tagVERSION_4_16_13, with local patches applied explicitly afterwards. - PG16 compatibility today is fine: there are 9
#if PG_VERSION_NUM >= 160000guards in the code, and CI runsgpcontrib/orafce:installcheckas part of the requiredic-gpcontribcheck. This issue is about refreshing content, not fixing a build. - Test coverage is thin:
Makefile:76,92runsREGRESS = orafceplus--schedule=parallel_schedule, but the schedule only listsinit,dbms_pipe_session_A,dbms_pipe_session_B— that is 4 of the 21 SQL test files (~19%). Thedbms_alerttests are commented out as MPP-unsuitable. - Local delta that must survive the upgrade:
regexp_not_like()added toorafce--4.8--4.9.sqlandorafce--4.9.sql(7bda69676d5);- MPP test adaptations (
gp_distributed_by,gp_partition_by,nlssort) and rebranding (b64d461e37c); - locale handling change in
others.c(4af1cf5a09b).
Why this belongs in 3.0
- A major release is the natural window to refresh a bundled dependency; doing it now avoids turning this into a multi-year jump.
- orafce is user-facing: accumulated upstream fixes and compatibility improvements can be summarized as a concrete 3.0 highlight in the release notes.
- It keeps future refreshes cheap — the next one can target a handful of minor releases instead of seven.
Work items
- Prerequisite: expand orafce regression coverage — move as many of the 21 test files into the schedule as can run single-node/MPP, keeping
dbms_alertskipped with an explicit reason. Without this, behavior changes from the upgrade are invisible. -
git subtree pullorafce/orafceat tagVERSION_4_16_13(perREADME.gp). - Re-apply the local delta:
regexp_not_like()(check whether upstream 4.16 has adopted it — drop the patch if so), MPP test adaptations,others.clocale handling, rebranding. - Ship the full upgrade chain from 4.9 to 4.16.13 (
orafce--4.9--4.10.sql…orafce--4.16.13.sql) plus the new base install script; bumporafce.controldefault_versionandMETA.json. - Verify the extension update path:
CREATE EXTENSION orafce VERSION '4.9'; ALTER EXTENSION orafce UPDATE;, and apg_upgradefrom a 2.x cluster (pg_upgrade'supdate_extensions.sqlemitsALTER EXTENSION orafce UPDATE). - Check licensing: orafce is 0BSD (
COPYRIGHT.orafce), registered underLICENSE:542+licenses/LICENSE-orafce.txt; confirm copyright/NOTICE text is still accurate after the update and RAT stays green. - Run the required
ic-gpcontribcheck plus broader ICW; prepare a short "user-visible changes" section for the 3.0 release notes.
Acceptance criteria
\dx orafcereports 4.16.13; fresh install, update from 4.9, and a 2.x → 3.0pg_upgradeall succeed.- Existing orafce behavior is unchanged except for documented upstream fixes/features; the expanded test suite passes.
regexp_not_like()and the other Cloudberry-specific behavior remain available.- License/NOTICE and RAT checks stay green; CI
ic-gpcontribis green. - A user-visible changes section is ready for the 3.0 release notes.
Risks
- Compatibility-layer semantics changes may alter results for Oracle-migration workloads; needs release-note attention.
- Local SQL patches complicate the subtree merge and the upgrade-script chain.
- Expanding tests in an MPP environment may require per-test expected-output adaptation — scope the prerequisite before starting the merge.
References
- orafce 4.16.13 packaged for PG16 (PGDG): https://www.postgresql.org/message-id/E1xDjRr-0000000146B-230h%40atalia.postgresql.org
- Upstream project and tags: https://github.com/orafce/orafce (
VERSION_4_16_13) - Subtree workflow:
gpcontrib/orafce/README.gp
- Langage dominant
- C
- Étoiles
- 1.4k
- Forks
- 265
- Merge moyen
- 3 j 18 h
- PR mergées (30 j)
- 70
Préparer son environnement
- Aucun Dockerfile ni fichier Docker Compose
- Propose un modèle de pull request
- Lire le guide de contribution
Par où commencer
- Lisez l'issue en entier, puis le guide de contribution du projet.
- Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
- Forkez le dépôt et travaillez sur une branche.
- Ouvrez une pull request qui référence le numéro de l'issue.
Autres issues de apache/cloudberry
-
[Bug] Expand fails to recognize user-defined tablespacesPeut-être pris @nix-oss l’a pris il y a 76 jours. Ouvertetype: Bug
Difficulté 2/5 1-3 heures Accessibilité débutants 76/100
apache/cloudberry#1885 · 2 réactions ·
Les mainteneurs répondent en général sous 2 jours
-
gpexpand first stage fails: 'SyncPackages' object has no attribute 'ret' (extension sync never runs; regression)Peut-être pris @talmacschen-arch l’a pris il y a 137 jours. Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 86/100
apache/cloudberry#1825 ·
Les mainteneurs répondent en général sous 2 jours
-
Difficulté 3/5 Une demi-journée Accessibilité débutants 30/100
apache/cloudberry#2224 ·
Les mainteneurs répondent en général sous 2 jours
-
[Bug] GPORCA: “could not find MotionState” for correlated EXISTS with LIMIT and a scalar subqueryPeut-être pris @Alena0704 l’a pris il y a 4 jours. Ouvertetype: Bug
apache/cloudberry#2221 · 1 personne assignée ·
Les mainteneurs répondent en général sous 2 jours
-
Difficulté 4/5 3-5 jours Accessibilité débutants 42/100
apache/cloudberry#2214 · 2 commentaires ·
Les mainteneurs répondent en général sous 2 jours
Toutes les issues de apache/cloudberry
Issues similaires
-
examples/dnsdoctor: cleanup.sh stops at the first missing piece of a setup.sh that did not finishOuverteseverity: low
Difficulté 2/5 1-3 heures Accessibilité débutants 70/100
luainkernel/lunatik#2263 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 2/5 Une demi-journée Accessibilité débutants 72/100
vercel-labs/scriptc#2910 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 2/5 1-3 heures Accessibilité débutants 68/100
darktable-org/darktable#22939 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 1/5 Moins d'une heure Accessibilité débutants 82/100
espressif/idf-extra-components#903 ·
Les mainteneurs répondent en général sous 1 jour
-
os: input_password leaves terminal echo disabled when reading failsPeut-être pris @davlgd l’a pris aujourd’hui. Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 82/100
Les mainteneurs répondent en général sous 1 jour