chore: remove the submodule-era residue — the dead foundry.lock and its references

[thedavidmeister]

Closes#35.

Configuration only. No Solidity source, no deployed bytecode, no audited artifact.

What was dead, verified on a fresh clone of main (af4e5a9)

foundry.lock is Foundry's git submodule lockfile. This repo takes no submodules:

• no .gitmodules, no lib/ directory
• git ls-files --stage | awk '$1=="160000"' → empty (zero gitlinks)
• foundry.toml sets libs = ['dependencies']; soldeer.lock is the live lockfile, and it holds exactly one package, forge-std 1.16.1

Its two pins contradicted each other and the build:

• lib/forge-std → 1801b054, which is forge-std v1.14.0, while the build resolves 1.16.1
• lib/rain.math.binary → a dependency that appears in neitherfoundry.toml [dependencies]norsoldeer.lock

Not silent either. forge build --force on the clean clone, before this change:

Warning: Dependency 'lib/rain.math.binary' not found at expected path
Warning: Dependency 'lib/forge-std' not found at expected path

Both are gone after it. The one remaining Warning: Found unknown config section in foundry.toml: [package] is pre-existing, is forge deprecating the section soldeer itself uses, and is not this issue.

Submodules also cannot return: rainix-sol-static runs no-submodules, which fails on a root .gitmodules or any committed gitlink.

The references go with the file

None of these fail anything on their own — reuse lint tolerates annotation paths that do not exist, and a .soldeerignore line for an absent path is a no-op. That is exactly why they survived the first pass, so they are removed in the same change rather than left for a third.

• REUSE.toml — dropped ".gitmodules" and "foundry.lock". "src/lib/LibCodeGen.sol" is kept: that is first-party source that exists, not a lib/ submodule path.
• .soldeerignore — dropped .gitmodules, /foundry.lock and /lib for the same reason, plus three more absent paths found while checking:
  • .coderabbitai.yaml — no such file, and not gitignored, so nothing can produce it
  • CLAUDE.md — no such file
  • /target — this repo has no Rust: no Cargo.toml, no crates/, target is not in .gitignore, and nothing under nix develop produces one (checked git status --porcelain --ignored after a devShell build: the only ignored-but-present paths are .pre-commit-config.yaml, cache/, dependencies/, out/)

Deliberately kept:.DS_Store, .vscode, .pre-commit-config.yaml, /out, /cache, /dependencies, /docs. Those are absent from a clean checkout by design — OS junk, local developer files, or artifacts generated at forge soldeer install / forge build / devShell-entry time, and therefore present when soldeer push runs. .pre-commit-config.yaml in particular is written into the tree by entering the rainix devShell, which the run above confirms.

What ships is unchanged

forge soldeer push --dry-run on main and on this branch produce identical 18-entry file lists. Removing the .soldeerignore lines adds nothing to the package, because none of the paths they named exist.

QA

• Discriminating tests: n/a — this diff contains no executable code, so there is no behaviour a Solidity test could discriminate on. The equivalent discriminating evidence is the build itself: nix develop -c forge build --force on the unmodified clone emits Warning: Dependency 'lib/rain.math.binary' not found at expected path and Warning: Dependency 'lib/forge-std' not found at expected path, and the same command on this branch emits neither. That pair fails on base and passes here. Alongside it, nix develop -c reuse lint is compliant with REUSE 3.3 (42/42 files, 0 missing/unused/bad) after the two annotation paths are dropped, nix develop -c forge fmt --check is clean, and nix develop -c forge test is 84 passed / 0 failed — byte-identical to the pre-change baseline measured on main at af4e5a9, with no test added, edited or weakened.
• Mutations applied: n/a — the three files are a deleted JSON lockfile that nothing in the tree reads, a REUSE annotation path list, and a packaging ignore list. None contains a statement a mutant could be applied to or survive in. The nearest analogue was run instead, as a whole-file deletion probe on the packaging config: forge soldeer push --dry-run on main and on this branch produce identical 18-entry package listings, so no removed ignore line was load-bearing.
• Oracle: the tree and the tooling, not the issue text. Every claim was re-derived on a fresh clone rather than taken from Remove the submodule-era residue — a dead foundry.lock pinning a dependency this repo no longer takes #35 — git ls-files --stage | awk '$1=="160000"' for the absence of gitlinks, ls -a + git ls-files for the absence of .gitmodules, lib/, CLAUDE.md, .coderabbitai.yaml, Cargo.toml and crates/, soldeer.lock for the one live dependency (forge-std 1.16.1) against foundry.lock's v1.14.0 pin, git status --porcelain --ignored after a devShell build for which ignored paths actually materialise, and rainix's no-submodules — confirmed wired into rainix-sol-static — for the claim that submodules cannot return.
• Category check: Remove the submodule-era residue — a dead foundry.lock pinning a dependency this repo no longer takes #35 asks for six things — delete foundry.lock; drop REUSE.toml lines 8 and 19 while keeping line 18; drop .soldeerignore lines 2, 7, 11, 18, 20 and 25; no Dependency ... not found warning from forge build; no .gitmodules / lib/ / foundry.lock reference left outside dependencies/ and first-party src/lib/; CI green. All six covered. The last was re-verified by grep after the change: outside dependencies/ there is now no gitmodules and no foundry.lock string anywhere, and every surviving lib/ hit is a first-party src/lib/ path — the REUSE.toml annotation the issue explicitly carves out, plus script/ and test/ imports of src/lib/*.sol.

One consequence to note

REUSE.tomlis in the published package (it appears in the 18-entry dry-run listing above), so the rainix-autopublish soldeer content gate will read this as a content change: merging publishes 0.1.7 and bumps [package].version to 0.1.8. That is correct — the packaged REUSE.toml genuinely changed — but it means a configuration-only PR cuts a release.

Summary by CodeRabbit

• Chores
  • Updated project file handling and licensing annotation configurations.
  • Removed obsolete entries from package and compliance metadata lists.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: a043050c-d3cf-4585-a76c-97a51903baee

📥 Commits

Reviewing files that changed from the base of the PR and between af4e5a9 and e7e947d.

⛔ Files ignored due to path filters (1)

• foundry.lock is excluded by !**/*.lock

📒 Files selected for processing (2)

• .soldeerignore
• REUSE.toml

💤 Files with no reviewable changes (2)

• REUSE.toml
• .soldeerignore

Included review availability: Your plan includes up to 1 review per rolling hour; 0 remain after this review.

---

Walkthrough

The change removes obsolete .soldeerignore entries and removes .gitmodules and foundry.lock from REUSE.toml. No public declarations or source code change.

Changes

Repository metadata cleanup

Layer / File(s)	Summary
Remove stale metadata entries .soldeerignore, REUSE.toml	.soldeerignore no longer ignores obsolete configuration, submodule, lockfile, library, and target paths. REUSE.toml no longer annotates .gitmodules or foundry.lock.

Estimated code review effort: 1 (Trivial) | ~3 minutes

Merge Risk:⚪ Minimal · up to e7e94

This configuration-only cleanup removes obsolete dependency and packaging references without changing Solidity code or published package contents; no actionable merge-blocking risk remains after normal checks and review.

Possibly related issues

• rainlanguage/rain.vats.flare#2 — Tracks removal of the same stale .soldeerignore and REUSE.toml entries.
• rainlanguage/rain.vats#334 — Covers removal of obsolete .gitmodules and foundry.lock metadata.
• rainlanguage/rain.math.float#258 — Covers removal of obsolete submodule-era metadata entries.
• rainlanguage/rain.sol.binmaskflag#20 — Targets the same stale .gitmodules, /lib, and foundry.lock references.
• rainlanguage/rain.math.fixedpoint#29 — Targets the same submodule-era residue.
• rainlanguage/rain.math.binary#33 — Targets removal of obsolete .gitmodules, /lib, and foundry.lock references.
• rainlanguage/rain.factory.deploy#12 — Targets removal of stale .gitmodules and foundry.lock entries.
• rainlanguage/rain.factory#52 — Covers the same metadata cleanup.
• rainlanguage/rain.solmem#111 — Targets removal of stale submodule-era references.
• rainlanguage/rain.merkle#17 — Covers removal of obsolete .gitmodules and foundry.lock entries.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Linked Issues check	❓ Inconclusive	The reviewed configuration removals match issue #35, but deletion of foundry.lock and validation results cannot be verified because foundry.lock is excluded by !**/*.lock.	Provide reviewable evidence for foundry.lock deletion and the required build, reference-scan, and CI results; the file is excluded by !**/*.lock.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the main change: removing the obsolete foundry.lock file and related submodule-era references.
Out of Scope Changes check	✅ Passed	The changes remove only stale entries explicitly listed in issue #35 and do not modify Solidity source, bytecode, or audited artifacts.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch 2026-08-16-remove-submodule-residue

---

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.}

[thedavidmeister]

Reviewed e7e947d: ready — Closes #35. Three files, 16 deletions, no source and no test touched.

The evidence behind it is what makes it landable rather than plausible. Both Warning: Dependency 'lib/...' not found at expected path warnings reproduce on a clean clone at af4e5a9 and are gone after the change, so the residue was doing something rather than merely sitting there. forge soldeer push --dry-run on main and on the branch produce identical 18-entry package listings, which is the proof the .soldeerignore edits ship nothing new — an ignore-list change is exactly where a quiet addition to a published package would hide. rainix-static no-submodules is wired into rainix-sol-static, so submodules cannot come back unnoticed.

forge test 84 passed, identical before and after.

The consequence is stated in the body rather than left to be discovered: REUSE.toml ships in the package, so autopublish's content gate reads this config-only PR as a content change and will publish 0.1.7 on merge, bumping the manifest to 0.1.8. That is the lifecycle working — #24's reasoning, posted today, is that this repo's next-version slot is self-maintaining and correct for a pure source emitter.

All five checks green, and CodeRabbit actually reviewed this one — Review completed, not the rate-limit banner the coverage PRs carried — with 0 unresolved threads checked via GraphQL. So that zero is coverage rather than vacuous.