Trim RELEASE.md to only repo-specific content

[jnasbyupgrade]

Summary

• The org now has a shared release process doc at ../ai/RELEASE.md (Add shared RELEASE.md for PGXN-extension repos ai#8), covering the generic pgxntool make tag/make dist, versioning, and stable pseudo-version steps.
• Trims this repo's RELEASE.md down to a short pointer to the shared doc plus the one genuinely repo-specific gotcha: the CAT_TOOLS_GIT_REF dependency-override guardrail (PGXN's published cat_tools doesn't yet satisfy this distribution's declared floor, so CI builds cat_tools from a git ref — cutting a release while that's still set would produce a zip nobody could actually build). Verified this is still live: ci.yml still sets CAT_TOOLS_GIT_REF: 0.3.0, and sql/extension_drop.sql still calls cat_tools.routine__parse_arg_types_text.
• Everything else in the old RELEASE.md (ongoing-development/HISTORY.asc steps, the numbered cutting-a-release steps, "Future: CI automation", and the generic pgxntool/CI notes-and-gotchas) is now covered generically by the shared doc, so it's removed rather than kept for symmetry.
• Updated Makefile's comment cross-reference to RELEASE.md's "Ongoing development" section, since that section now lives in ../ai/RELEASE.md instead.

Depends on Postgres-Extensions/ai#8 landing first (that PR adds ../ai/RELEASE.md, which this repo's trimmed RELEASE.md links to and is not yet merged to ai's main).

Test plan

• Confirm CI passes (lint + test matrix + pg-upgrade-test) — no functional code changed, just RELEASE.md content and one Makefile comment.
• Land Add shared RELEASE.md for PGXN-extension repos ai#8 before/alongside merging this, so the ../ai/RELEASE.md link resolves.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 3be7a70d-8f1b-40db-8b68-20b4bed4af21

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Claude finished @jnasbyupgrade's task in 1m 55s —— View job

---

Claude's Review of PR #45

• Gather context (diff, CLAUDE.md, related files)
• Understand the request (code review of PR Trim RELEASE.md to only repo-specific content #45)
• Review RELEASE.md changes
• Review Makefile changes
• Verify the repo-specific gotcha claims (CAT_TOOLS_GIT_REF, cat_tools usage)
• Post final review

Summary: This is a clean, well-scoped doc trim. I verified the two live-repo claims the PR body makes:

• .github/workflows/ci.yml:43 still sets CAT_TOOLS_GIT_REF: 0.3.0.
• sql/extension_drop.sql:38 still calls cat_tools.routine__parse_arg_types_text.

Both check out, so keeping the CAT_TOOLS_GIT_REF guardrail as the one repo-specific section in the trimmed RELEASE.md is correct. The "stale 2017 release" characterization of PGXN's published cat_tools also matches the existing ci.yml header comments (lines 12, 38-39), so no new claim was introduced that isn't already substantiated elsewhere in the repo.

I also checked every other reference to RELEASE.md in the repo (ci.yml:41, Makefile:17, Makefile:111) to make sure none pointed at a section that got removed:

• ci.yml:41's reference to "RELEASE.md's guardrail against cutting a release while this is still set" — still valid, the guardrail section is kept (just reworded to point at the shared doc for the generic part).
• Makefile:17's reference is a general pointer to RELEASE.md/HISTORY.asc, not to a removed section — unaffected.
• Makefile:111 was updated in this PR to point at ../ai/RELEASE.md's "Ongoing development" section, matching where that content actually moved.

Minor, non-blocking nit: the Makefile comment rewrap (lines 108-115) leaves a short orphan line, # again), so, on its own line before the next sentence starts. Purely cosmetic — not worth a round-trip.

Dependency risk (already called out in the PR description, not new): RELEASE.md:3/:10 link to ../ai/RELEASE.md, which doesn't exist yet in this checkout — confirmed by the PR body's own note that it depends on Postgres-Extensions/ai#8 landing first. Nothing to fix here, just confirming the stated blocking dependency is real and this repo's link will 404 until that PR merges.

No functional/SQL code changed — this is docs + one comment only, consistent with the PR's stated scope. No issues found that need changes before merge, contingent on the ai#8 dependency landing.

[github-actions]

Claude finished @jnasbyupgrade's task in 1m 31s —— View job

---

Claude's Review of PR #45

• Gather context (diff, CLAUDE.md, related files)
• Review RELEASE.md changes
• Review Makefile changes
• Verify repo-specific claims (CAT_TOOLS_GIT_REF, cat_tools usage, link target)
• Post final review

Summary: Clean, well-scoped doc trim. No functional/SQL code touched — just RELEASE.md content and one Makefile comment, matching the PR's stated scope.

Verified the live-repo claims:

• .github/workflows/ci.yml:29 still sets CAT_TOOLS_GIT_REF: 0.3.0.
• sql/extension_drop.sql:38 still calls cat_tools.routine__parse_arg_types_text.
• META.in.json:82 still declares the cat_tools floor as 0.3.0, matching the new RELEASE.md's description of why PGXN's published cat_tools can't satisfy it.

Keeping the CAT_TOOLS_GIT_REF/CAT_TOOLS_SKIP_INSTALL guardrail as the one repo-specific section in the trimmed RELEASE.md is correct, and the new wording still tells a release-cutter exactly what to check (ci.yml's actual value, not just the Makefile's support for the override).

Checked every other RELEASE.md reference in the repo to make sure none point at removed content:

• .github/workflows/ci.yml:27 ("RELEASE.md's guardrail...") — still valid, the guardrail section is kept.
• Makefile:110 — correctly updated in this PR to ../ai/RELEASE.md's "Ongoing development" section, matching where that content actually moved.
• bin/test_existing:21 — a generic "see HISTORY.asc/RELEASE.md" pointer, not tied to a removed section — unaffected.

Pre-existing, unrelated to this PR: Makefile's cat_tools target still falls back to pgxn install 'cat_tools>=0.2.1' for the plain-install path, while META.in.json declares the floor as 0.3.0 (the same floor this PR's RELEASE.md text describes as unsatisfiable via plain pgxn install). Worth a look in a separate PR, but not introduced or worsened here — just flagging since it's adjacent to the exact guardrail this doc trim discusses.

Dependency risk (already called out in the PR description, not new): the ../ai/RELEASE.md link this file points to doesn't exist in the ai repo yet — confirmed by the PR body's own note that it depends on Postgres-Extensions/ai#8 landing first. Nothing to fix on this side; just confirming the stated blocking dependency is real.

No issues found that need changes before merge, contingent on the ai#8 dependency landing.