[brettheap]

Lane: openreposhape-1 (openRepoShape-1)
Claim: #931 (comment)

A v1.2.0 draft on main 64464d31. It follows #887 (#885), whose head lane 5 found READY with three NITs. Those NITs are #931. It changes tests/test_repo_hygiene.py only.

What

1. The message offers the exclusion set only where it works.
   • The failure text is now a module-level _one_line_install_failure(name).
   • Its "or, if it is not a release's notes, name 'X' in RELEASE_NOTES_WITHOUT_THE_ONE_LINE_INSTALL" clause goes only to a .md in docs/release-notes/ whose name does not match v*.md, on its own line after the indented install line.
   • Before, a mistyped line in a release's own notes (mutant M5) was told to name the file in the set, which the second test then refuses.
   • The comment above the set now says "a v*.md other than v1.0.0.md named here fails the second test below".
2. A test reads the message. test_the_one_line_install_message_offers_the_set_only_where_it_works asks it for docs/release-notes/v9.9.9.md, docs/release-notes/README.md and README.md. It holds that:
   • the install line is in all three;
   • the clause is in the second alone;
   • the line comes before the clause.
3. Every line of the notes that runs the install carries the one line. test_every_release_notes_line_that_installs_holds_the_one_line_install reads every v*.md but v1.0.0.md.
   • A line holding both curl and download/install.sh must contain ONE_LINE_INSTALL.
   • What follows it must be nothing, or the pinned form -s -- v<major>.<minor>.<patch>, which is install.sh's own usage, with only a closing backtick and punctuation after it.
   • Each failure names the file, the line number and the line.
   • The test also holds a count per file (v1.0.1.md at least 2, the others at least 1), so a rule that matches nothing fails.

M10. Lane 5 left it open whether to hold the text after the line. This takes the pinned form, measured first: the four lines today (v1.0.1.md:12 and :45, v1.1.0.md:49, v1.1.1.md:132) all match it, and | bash -s -- --yes or -s -- main does not.

Mutants

Round 1, at a6597405, run with -k one_line_install (11 tests then):

Mutant	Change	Result
M1	| bsh at v1.1.0.md:49	caught
M5	| bsh at v1.1.1.md:132	caught, and its message has no clause
M6	v1.1.1.md added to the set	caught
M7	| bsh at v1.0.1.md:12	caught (new)
M8	a scratch docs/release-notes/README.md	caught, and its message has the clause
M10	| bash -s -- --yes at v1.1.1.md:132	caught (new)
X3	the message's if comparison turned around	caught (new)
X3b, X3c, X4, X5	the clause's condition broken four more ways	caught
X6	the rule's curl test made to match nothing	caught
	-s -- main at v1.0.1.md:45	caught

Rounds 2 and 3 (a1649ef6, daf0376d), from r959's NITs:

• The per-line verdict is _one_line_install_fault(line), with tests of accepted tails, rejected tails and a bsh fault.
• A stray draft-v1.2.0.md gets the clause.
• The accepted tails include ;, :, ) and -s -- v10.20.30.
• The wording is fixed, and the count assert says what it found and wanted.
• Y1, Y1c, Y1d, Y4, Y5, Y3 and Y3e, and T-semicolon, T-colon, T-rparen and T-digits, now each fail. r959 found the PR READY at daf0376d (tests/test_repo_hygiene.py: the one-line-install check offers the exclusion set only to a .md in the notes directory that is no release's v*.md, a test reads its message, and every line of the release notes that runs the install holds the one line, with nothing or the pinned version after it (#931) #959 issuecomment-6253273039).

Tests

Run with /usr/bin/python3 -m pytest, a stub gh first on PATH, a bogus GH_TOKEN, an empty GH_CONFIG_DIR, GIT_CONFIG_GLOBAL=/dev/null and GIT_SSH_COMMAND=false, on main 64464d31:

• tests/test_repo_hygiene.py, tests/test_release_workflow.py and tests/test_workflow_concurrency.py: 170 passed at daf0376d (151 before);
• -k one_line_install: 28 passed at daf0376d (9 before).

No added line is over 79 columns, the change is ASCII, and there is no compound assert.

Known limit: the version in the pinned form is not read, so -s -- v9.9.9 passes in any release's notes.

Written by lane 1's writer w931 on #887's head 96422b98, then cherry-picked onto main once #887 landed. The patch-id is unchanged.

Closes #931

🤖 Generated with Claude Code

Summary by Sourcery

Strengthen repository hygiene checks for one-line installation commands and ensure their failure guidance is accurate and actionable.

Bug Fixes:

• Prevent release-note hygiene failures from suggesting an exclusion-list workaround that the same test rejects.
• Detect mistyped or malformed one-line install commands on every applicable release-notes line, including invalid trailing arguments.

Enhancements:

• Add focused validation of failure-message guidance and accepted install-command suffixes.
• Require release notes to contain the expected minimum number of install-command examples so the hygiene rule cannot pass without matching any lines.

Tests:

• Expand repository hygiene tests to cover message guidance, valid and invalid command endings, typo detection, and per-line release-note validation.

[sourcery-ai]

Reviewer's Guide

The PR hardens one-line-install hygiene tests by making failure guidance context-aware and adding mutation-resistant checks that every executable release-note install line contains the exact command with only an allowed pinned suffix.

File-Level Changes

Change	Details	Files
Centralize and correct the one-line-install failure guidance so the exclusion set is suggested only for eligible non-release notes.	Added a reusable failure-message builder with conditional exclusion guidance. Updated the exclusion-set comment and migrated the existing test to use the builder. Added coverage for release-note, non-release-note, and outside-directory message variants and clause ordering.	tests/test_repo_hygiene.py
Strengthen release-note hygiene checks to validate every executable install line, including permitted pinned arguments.	Scan all applicable v*.md files and identify lines containing both curl and download/install.sh. Require the canonical install command on each matching line and restrict trailing text to the pinned version form or punctuation. Report file, line number, and source line for violations, while enforcing per-file minimum match counts to prevent vacuous passes.	tests/test_repo_hygiene.py

Assessment against linked issues

Issue	Objective	Addressed	Explanation
#931	Offer the exclusion-set remedy only for Markdown files in the release-notes directory whose names are not release-style v*.md, and place it after the required install line in the failure message.	✅
#931	Add a test that reads the failure message and verifies the exclusion-set clause appears only for an eligible non-release notes file, with the install line preceding it.	✅
#931	Check every release-notes line that runs the install so it contains the exact one-line install, catching mistyped copies even when another correct copy exists; constrain any trailing text to the permitted pinned-version form.	✅

Possibly linked issues

• tests/test_repo_hygiene.py: the one-line-install check offers its exclusion set to a release's own notes, which the second test refuses, no test reads its message, and two mistyped copies still pass (follow-up to #887) #931: The PR directly addresses all three issue goals: usable exclusion advice, a message test, and validation of every install line.
• tests/test_repo_hygiene.py: the one-line-install check offers its exclusion set to a release's own notes, which the second test refuses, no test reads its message, and two mistyped copies still pass (follow-up to #887) #931: The PR directly covers the missing release-note checks, catching mistyped install lines in v1.1.0.md and v1.1.1.md.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[brettheap]

(lane openRepoShape-1, 2026-10-18T17:58:24Z) Independent review of #959 at a6597405, by lane 1's reviewer r959 (sonnet), who is not its writer: READY, with no MUST and no SHOULD. Lane 1 makes its three NITs before marking the PR ready.

• tests/test_repo_hygiene.py: the one-line-install check offers its exclusion set to a release's own notes, which the second test refuses, no test reads its message, and two mistyped copies still pass (follow-up to #887) #931's three points hold.

  1. The clause goes to a .md in docs/release-notes/ that does not match v*.md, and to no other file. It sits on its own line after the install line. With "README.md" added to the set, the 11 tests pass, so the clause's advice works.
  2. A test reads the message, for v9.9.9.md, a README.md in the notes directory and the root README.md. Every variant of the clause's condition that r959 broke was caught (X3, X3b, X3c, X3d).
  3. Every notes line: the rule picks exactly four lines (v1.0.1.md:12 and :45, v1.1.0.md:49, v1.1.1.md:132), and all pass. | bsh at each of them, -s -- --yes and -s -- main are all caught, and the message gives the file, the line number and the line.

• Tests: the three files (test_repo_hygiene.py, test_release_workflow.py, test_workflow_concurrency.py): 153 passed, against 151 with main's file. -k one_line_install: 11 passed.

• NIT 1: the pinned-form rule's rejecting side has no case of its own. All five of these survive:

  • Y1, the regex loosened to -s -- .*;
  • Y1c and Y1d;
  • Y4, fullmatch changed to match;
  • Y5, the "does not hold the line" branch switched off.

  The fix r959 tried is a per-line _one_line_install_fault(line), with tests of accepted tails, rejected tails and bsh. It gives 24 passed and kills all five.

• NIT 2: the v*.md boundary is tested by one stray name only, so a "contains v" test (Y3, Y3e) survives. Fix: assert the clause for docs/release-notes/draft-v1.2.0.md.

• NIT 3, wording:

  • "a release's v*.md" does not say "other than v1.0.0.md";
  • it says "punctuation" where the regex allows six characters;
  • one antecedent is unclear;
  • a bare assert all(...) prints a dict and assert False.

• Known limit, not a defect: the rule fails, with a clear message, on prose after the closing backtick on the install line's own line, and on a one-line curl ... download/install.sh -o install.sh. Today's verify blocks avoid it by wrapping.

Lane 1's writer w959n has made NITs 1 to 3 at a1649ef6.

• -k one_line_install: 24 passed. The three files: 166 passed.
• Y1, Y1c, Y1d, Y4, Y5, Y3 and Y3e now each fail.

r959 re-checks that head now. Then lane 1 marks #959 ready and gates it.

Lane: openreposhape-1 (openRepoShape-1)

[brettheap]

(lane openRepoShape-1, 2026-10-18T18:03:14Z) Re-check of #959 at daf0376d, by r959 (sonnet), who is not its writer: READY, with no MUST, SHOULD or NIT left.

• Round 2 (a1649ef6) made NITs 1 to 3:

  • the per-line verdict, _one_line_install_fault(line), with tests of its own;
  • the stray draft-v1.2.0.md;
  • the wording.

  All seven mutants that had survived (Y1, Y1c, Y1d, Y4, Y5, Y3, Y3e) now fail.

• Round 3 (daf0376d) makes r959's last two asks:

  • The accepted tails gain ;, :, ) and -s -- v10.20.30. So a trailing class that drops one of them fails, as does \d in place of \d+.
  • The docstring's antecedent is fixed.

• Tests: -k one_line_install: 28 passed. The three files: 170 passed.

• Style: the widest line is 79 columns, the text is ASCII only, and git diff --check is clean.

Lane 1 marks #959 ready for review now. Once its tests are green, lane 1 asks for a turn, gated over main as it then stands.

Lane: openreposhape-1 (openRepoShape-1)

[sourcery-ai]

Sorry @brettheap, you've used your own review budget of 250,000 diff characters for the last 7 days.

You can request another review in 1 day and 6 hours by commenting @sourcery-ai review. Upgrade to get a review now.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud