[brettheap]

Change

Fix the mount failure when an adoption plan moves the source's .gitmodules into a leg with an existing dependency. After removing moved paths, create and stage a fresh assembly registration file only if it is absent. Retained assembly registrations are preserved.

The original .gitmodules, dependency gitlink and history remain in the extracted leg. Content verification is unchanged.

Verification

• Before the fix, the new regression reproduced Git's fatal: please make sure that the .gitmodules file is in the working tree refusal at the first new leg mount.
• python3 -m pytest tests/test_adopt_*.py -q: 90 passed, 1 skipped inside py-bench with git-filter-repo 2.47.0.
• Final regression run after adding explicit gitlink-mode and bootstrap assertions: 4 passed.
• New coverage exercises a populated dependency moved into code or spec, exact registration/blob/gitlink/history preservation, recursive clone and bootstrap, retained assembly registrations, and recovery from a failed mount into a fresh clone. Existing adopter tests cover sources without submodules.
• Disposable openRepoTools rehearsal at f5444c5e1f3c6c9ed43aeafcc000b0e78b10497a completed with this commit. Independent path/mode/type/object accounting: 94 original entries, 45 spec, 43 code, 6 root, zero missing or duplicated.
• Recursive clone, generated assembly bootstrap and validators passed. The nested openRepoShape dependency stayed at 39d5c986fcfac1a160474bfe91c5f1c37fccc72c, digest a44c0165beb9d3b6ddfce46925bf04b8d19934d5327b5a304b5f3f11af699b02.
• The rehearsal used disposable local bare remotes and file-only Git transport. The isolated source stayed clean on its original main commit. Live openRepoTools main, gitlink and dependency contract remain unchanged.

Delivery boundary

This is the upstream prerequisite for openRepoTools migration task T005. Consumer repinning waits for a reviewed commit on upstream main and a freshly computed digest. The rehearsal does not authorize real leg repository creation, migration cutover, seat movement or merging.

Summary by Sourcery

Preserve existing submodule registrations and dependency history when extracting adoption legs.

Bug Fixes:

• Preserve an existing dependency’s .gitmodules, gitlink, and history when adoption moves it into a leg.
• Allow leg mounting to recreate and stage .gitmodules safely, including repositories whose ignore rules hide dotfiles.

Enhancements:

• Retain assembly-level submodule registrations that are not moved by the adoption plan.

Tests:

• Add regression coverage for moved and retained submodule registrations, gitlink preservation, ignored .gitmodules files, mount failure recovery, recursive cloning, and bootstrap validation.

[sourcery-ai]

Reviewer's Guide

The adoption mount now stages a fresh empty .gitmodules only when extraction removed the source registration, while preserving registrations retained in the assembly; comprehensive local Git regression tests cover dependency/history fidelity, registration merging, validation/bootstrap, and recovery after failed mounts.

Sequence diagram for preserving gitmodules during leg extraction

sequenceDiagram
    participant Mount as AdoptionMount
    participant Assembly
    participant Git
    participant Leg as ExtractedLeg

    Mount->>Git: git rm -r -q path
    Git-->>Assembly: Source paths removed and staged
    Mount->>Assembly: .gitmodules exists?
    alt .gitmodules absent
        Mount->>Assembly: write_lf .gitmodules
        Mount->>Git: git add .gitmodules
        Git-->>Assembly: Fresh registration file staged
    else .gitmodules retained
        Git-->>Assembly: Existing registrations preserved
    end
    Mount->>Git: git submodule add work_root/spec spec_path
    Mount->>Git: git submodule add work_root/code code_path
    Git-->>Leg: Original .gitmodules, gitlink, and history remain

Loading

File-Level Changes

Change	Details	Files
Ensure leg mounting can recreate the assembly’s Git submodule registration file after the source registration is moved away.	Remove paths selected for extraction before mounting generated legs. Create and stage an empty .gitmodules only when no registration file remains, allowing subsequent git submodule add operations. Leave retained assembly registrations untouched instead of replacing the file.	adopt-project.py
Add regression coverage for preserving dependencies and mount recovery across adoption workflows.	Exercise moving .gitmodules and its dependency into both code and spec legs, including exact blob, gitlink mode, commit, history, recursive clone, validation, and bootstrap checks. Verify existing assembly registrations survive alongside generated leg registrations. Verify a failed mount does not mutate the source and a fresh clone can be mounted successfully afterward.	tests/test_adopt_submodules.py

---

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]

@codex review

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[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

[brettheap]

Review, and what changed since 91d5685

An adversarial review of head 91d5685 ran the new tests with the hunk reverted (3 of 4 fail with the quoted .gitmodules refusal; all 4 pass with it), ran tests/test_adopt_*.py on a local merge with #164 (90 passed, 1 skipped), and then built the cases the hunk did not test. Three findings are fixed on this branch, two are filed as issues, and one Sonar finding is taken.

Fixed here

• A regression on the common case. git add -- .gitmodules respects .gitignore. A source with NO submodules whose .gitignore is the usual .* + !.gitignore adopted cleanly before this PR and failed at the new git add after both legs had already been created and pushed, so a re-run met REFUSED leg-remote-exists. The fresh file is now written ONLY when the plan moved or dropped the source's .gitmodules, and staged with -f. A source that never had one is mounted exactly as before.
• Order of removal. A submodule whose path sorts before .gitmodules (.agents/, .claude/, .deps/, .github/ ...) made git rm -r stage an edit to .gitmodules, and the later git rm .gitmodules was refused. Pre-existing, but inside this hunk's remit: .gitmodules is now removed first.
• The tests. The module-wide skip kept the two mount-level tests off every machine without git-filter-repo, including the Windows job; it now sits on the end-to-end test alone. The failed-mount test accepted any CommandFailed (with the fix reverted it passed on the wrong failure) and asserted three things about a source the function never touches; it now matches the missing leg. The sys.modules registration the fixture never removed is gone. Mount-level tests cover the .* gitignore with and without a submodule, and the dot-directory ordering.
• Sonar S1192: .gitmodules is spelled once, as GITMODULES, the way scripts/family.py already spells it.

Filed, not fixed here

• adopt-project.py _commit_the_split stages with git add -A -- ., so a source .gitignore of .* silently drops .gitattributes and validate.yml from the split, execute still says adoption verified, and the assembly fails its first bootstrap #167 — _commit_the_split stages with git add -A -- ., so the same .* gitignore silently drops .gitattributes and validate.yml from the split while execute still says verified; pre-existing, surfaced by the re-verification at 6fa4e07.

• adopt-project.py: a plan that sends .gitmodules to one leg and a submodule it registers to another passes check and execute, says adoption verified, and the assembly's recursive clone then fails: plan should ask, check should find, execute should refuse #166 — a plan that sends .gitmodules to one leg and a gitlink it registers to another now passes check and execute and says adoption verified, and the recursive clone fails. Before this PR those plans crashed at the mount (after pushing two legs); the refusal never existed. The hunk's comment is narrowed to say so.

• adopt-project.py _verify skips every root gitlink and compares a kept root .gitmodules byte for byte, so any plan that keeps a source submodule or its .gitmodules in the root ends with exit 1 and adopt-lost #165 — _verify skips every root gitlink and byte-compares a root-kept .gitmodules, so any plan that keeps a source submodule in the root exits 1 with adopt-lost. Pre-existing; test_mount_preserves_registration_kept_in_assembly guards the mount, not that verification.

Branch brought up to main (1a9fc53) by merge, no rebase. Fix commits: 4b5591e (S1192), 6fa4e07 (the rest).

Re-verified at 6fa4e07: every fix put back in turn fails its own test; CI green on ubuntu, macOS and Windows (where the mount tests now run); SonarCloud reports zero open issues on the PR.

[brettheap]

@codex review

[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 16 hours and 11 minutes by commenting @sourcery-ai review. Upgrade to get a review now.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.