docs(ci): clarify Postgres support - #329
Conversation
NikolayS
left a comment
There was a problem hiding this comment.
REV rubric review
This is a manual execution of the repository's five applicable REV rubrics because the automated ultrareview quota was unavailable. It is not bot output. SOC2 was omitted as required.
- Bug hunter: the stable 14–18 support statement and separate 19 beta compatibility lane match the CI matrix.
- Security reviewer: no security-sensitive behavior changes.
- Test analyzer: YAML parsing and support-string checks are adequate for the prose/config delta.
- Docs reviewer: the production-support distinction is clear and uses the repository's preferred Postgres terminology.
- Guidelines checker: CI is 17/17 green and commits are within the repo-specific type/length rules.
Blocking findings: none in the patch.
Nonblocking: the PR title uses Postgres while the main commit subject uses PostgreSQL; the generic commit standard asks titles to match the main subject. Lifecycle still requires a real-user evidence comment before merge. This COMMENT review is not an approval.
Real-user verification evidenceVerified exact head Commands used: export PAGER=cat
psql --no-psqlrc \
--set=ON_ERROR_STOP=1 \
--dbname="$PGQUE_TEST_DSN" \
--file=sql/pgque.sql
psql --no-psqlrc \
--set=ON_ERROR_STOP=1 \
--dbname="$PGQUE_TEST_DSN"In the second command I executed the README sequence verbatim: create the Key output: The supported Postgres wording, badge, matrix, and roadmap statements were |
samorev Code Review Report
REVIEW FINDINGS (7)HIGH
CRITICAL
MEDIUM [docs] Roadmap table row switched from the table's ✅ convention to the bare word "Yes", making a completed item render as inconsistent with — and visually weaker than — every other row in the same "Done" column.
MEDIUM [tests] Reclassifying PG19 as an unsupported beta lane leaves the upgrade job with no coverage of the newest supported major. The matrix endpoints are 14 and 19beta1, so under the new definition the upgrade path is exercised on the oldest supported version and on a version the project explicitly does not support — PG15–18 upgrades are untested, including the newest production major users are told to run.
MEDIUM [docs] The support-claim correction is only applied to three files, and the PR description scopes its own verification to "the changed support surfaces". Support ranges in this kind of project typically also appear in docs pages, SPEC, the
LOW [guidelines] The new CLAUDE.md sentence makes the beta lane optional ("may be added") while the PR's stated intent is that the PG19 beta CI coverage stays in place. As written, a contributor could delete the 19beta1 jobs without violating rule 5, since the mandatory list is only 14–18.
LOW [guidelines] The docs now say beta coverage "is not a production-support claim", but the beta job remains an ordinary blocking matrix leg. Upstream beta churn (a
Summary
Note:
Review metadatasamorev-assisted review (AI analysis by Tanya301/samorev) |
What changed
Fixes #324.
Why
The repository tests
postgres:19beta1, but the badge shortened that to PostgreSQL 19 support. Beta compatibility is valuable evidence, but it is not the same commitment as support for a final production major.Validation
.github/workflows/ci.ymlwith Ruby YAML;git diff --check.This is intentionally a draft and has not been merged.