Skip to content

fix(release-verify-rc): align step 6 binary-exclude list with find - #987

Merged
potiuk merged 1 commit into
apache:mainfrom
Shaurya2k06:fix/release-verify-rc-step6-binary-exclude-list
Jul 30, 2026
Merged

potiuk merged 1 commit into
apache:mainfrom
Shaurya2k06:fix/release-verify-rc-step6-binary-exclude-list

Conversation

@Shaurya2k06

Copy link
Copy Markdown
Contributor

Summary

  • Align release-verify-rc Step 6 so prose and the emitted find agree:
    the scan always uses a fixed baseline of prohibited patterns; read
    release-build.md § Binary-exclude list (same file Step 4 uses for RAT)
    and append any extra globs beyond that baseline before emitting the recipe.
  • Document the same split in projects/_template/release-build.md and
    projects/magpie/release-build.md (additional prohibited globs vs
    known-accepted exceptions).
  • Refresh step-6 eval fixtures / output-spec.md to match.

Type of change

  • Skill change (.claude/skills/<name>/) — eval fixtures updated below
  • Tool / bridge contract (tools/<system>/*.md)
  • Python package (tools/*/ with pyproject.toml)
  • Groovy reference impl
  • Cross-cutting (RFC, AGENTS.md, sandbox, privacy-LLM)
  • Documentation (docs/, README.md, CONTRIBUTING.md)
  • Project template (projects/_template/)
  • CI / dev loop (prek, workflows, validators)
  • Other:

Test plan

  • uv run --project tools/skill-and-tool-validator --group dev skill-and-tool-validate passes
  • PYTHONPATH=tools/skill-evals/src python3 -m skill_evals.runner tools/skill-evals/evals/release-verify-rc/step-6-binary-exclusion/ (print mode; fixtures extract cleanly)
  • prek run --files on touched paths passes (includes check-placeholders, lychee, skill-and-tool-validate)
  • For skill behaviour changes: step-6 fixtures / output-spec.md updated for fixed-baseline + additions wording

RFC-AI-0004 compliance

  • HITL — any new mutation is gated on explicit user confirmation
  • Sandbox — no new unrestricted host access; network reach declared in the adapter
  • Vendor neutrality — placeholders used in all skill / tool prose
  • Conversational + correctable — agentic-override path documented if behaviour is adopter-tunable
  • Write-access discipline — no autonomous outbound messages; drafts only, sent on confirmation
  • Privacy LLM — private content does not reach a non-approved LLM; redactor invoked where needed

Linked issues

Closes #940

State that the paste-ready find uses a fixed baseline and show how
release-build.md supplies additional globs (RAT-style), so prose and
command agree on where the exclude list comes from.

Generated-by: Cursor Grok 4.5
Signed-off-by: shaurya2k06 <shaurya2k06@gmail.com>

@potiuk potiuk left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM — and this is a more valuable fix than "align the docs" suggests. The old wording wasn't merely ambiguous, it was inverted in a way that could have let a non-compliant source release pass its own pre-flight.

What was actually broken

Two documents disagreed about what the Binary-exclude list means:

  • projects/_template/release-build.md asked the adopter to "list any binary content the source artefact must NOT contain", and its example listed *.class, *.jar — "Java compiled output never ships in source."
  • skills/release-verify-rc/SKILL.md:462 then said: "A file that appears in the binary-exclude list of release-build.md is a known-and-accepted binary; surface it as EXPECTED-BINARY."

So an adopter who followed the template's own example would have configured Step 6 to classify compiled Java as EXPECTED-BINARY — i.e. accepted — switching off the prohibition for precisely the files the template told them never to ship. An artefact containing .class/.jar would have come back PASS. In a skill whose job is to catch exactly that before a release vote, that's the failure mode worth fixing.

Splitting the section's two jobs — additional prohibited globs vs known-accepted exceptions — removes the inversion, and rewriting the template example so *.class/*.jar are no longer list entries removes the trap that led to it.

The other drift you caught

The baseline find in Step 6 already scanned *.pyc and __pycache__, but three places disagreed with it:

  • the prose immediately below said "The default prohibited extensions are .class, .jar, .so, .dylib, .dll, .exe" — silently dropping both;
  • both eval fixtures' paste_recipe omitted them, so the fixtures contradicted the command the skill actually emits.

Making the baseline explicit and mandatory ("Do not drop baseline predicates when the Binary-exclude list is empty or only restates the baseline") is the right guard, given .pyc presence is the tell that an artefact was zipped from a working tree rather than exported from the tag — which the existing comment already explained well.

I checked the change is complete rather than partial: case-1, case-2, and output-spec.md are all updated, and those are the only step-6 fixtures carrying a paste_recipe.

Smaller observations

Neither blocking; both are follow-up material rather than anything to change here.

  • The two purposes now share one flat list, distinguished only by trailing prose (— additional prohibited glob / — known-accepted exception). An agent reading an adopter's file has to infer the category from free text. Given the bug being fixed was a semantic ambiguity in this exact section, two sub-headings (or a two-column table) would make the distinction mechanical rather than interpretive. Worth a follow-up once this lands.
  • assets/vendor/**/*.min.js isn't find syntax. The instruction says to "translate each into a -name or -path predicate", which is right, but ** has no meaning to find — the shaped form is -path '*/assets/vendor/*.min.js' (find's * already crosses /). One parenthetical showing that translation would save the next reader a wrong-but-plausible -path 'assets/vendor/**/*.min.js'.

This review was drafted by an AI-assisted tool and
confirmed by an Apache Magpie maintainer. The maintainer
approving this PR has read the findings and signed off. If
something feels off, please reply on the PR and a maintainer
will follow up.

More on how Apache Magpie handles maintainer review:
CONTRIBUTING.md § Opening a pull request.

@potiuk
potiuk merged commit bd024f7 into apache:main Jul 30, 2026
10 checks passed
@potiuk potiuk added capability:triage Sweep + classify + propose disposition family:release-management release-* skills family:setup setup-* skills labels Nov 1, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

capability:triage Sweep + classify + propose disposition family:release-management release-* skills family:setup setup-* skills

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Show how release-verify-rc loads the binary-exclude list in step 6

2 participants