Remove the dead submodule-era entries from .soldeerignore

[thedavidmeister]

Closes #208

What changed

Three lines removed from .soldeerignore, and nothing else in the tree:

• .gitmodules — there is no .gitmodules here, and git ls-files --stage | grep ^160000 is empty, so there are zero committed gitlinks. It cannot come back either: rainix CI runs a no-submodules check in the static job that fails on a root .gitmodules or any gitlink.
• /foundry.lock — Foundry's git submodule lockfile, recording the commit each dependency vendored under lib/ is pinned to, so forge install / forge update can restore identical submodule revisions. Meaningful only in a repo that vendors dependencies as submodules. Already deleted here.
• /lib — foundry.toml sets libs = ['dependencies'], so forge resolves everything under dependencies/ and never creates lib/.

Why only three, when the issue lists ten

The issue also nominated seven template-copied entries (.cargo, .coderabbit.yaml, .devcontainer.json, .envrc, .prettierignore, /audit, /deployments) on the grounds that those paths do not exist here. That reasoning does not hold, and they are kept.

.soldeerignore is a publish filter, not a description of the files that exist. An entry naming an absent path is a forward-looking rule, not residue: delete it and that file gets published into the soldeer package the moment anyone adds it. Absence is therefore not evidence for removal — the only entries genuinely dead are the ones tied to a mechanism this repo can no longer use, and submodules are exactly that.

The seven kept entries all name real, correctly spelled filenames, so each still does its job if the file ever appears. Verified against repos that actually have them rather than from memory — .cargo, .coderabbit.yaml, .devcontainer.json, .envrc and .prettierignore all exist verbatim at the root of rainlanguage/raindex and rainlanguage/rainlang. No entry in this file is misspelled, so there was nothing to correct in place either.

The issue's own out-of-scope list (.DS_Store, .vscode, .pre-commit-config.yaml, /out, /cache, /dependencies, /remappings.txt) is untouched for the same reason, plus its stated one.

CLAUDE.md:15 already reads "Soldeer (not git submodules)" — correct, explicitly out of scope, left alone. REUSE.toml is the opposite kind of file (its annotations describe files that must actually exist) but needs no change here: it annotates .soldeerignore, which still exists, so no annotation was removed and reuse lint has nothing new to resolve.

After this, the only remaining mention of .gitmodules / foundry.lock / lib/ outside dependencies/ is that one correct CLAUDE.md sentence.

Branch history has two commits — the first implemented the issue as filed, the second restores the seven after the filter-vs-description correction. Net diff is exactly three deletions.

QA

• Discriminating tests: soldeerignore-diff.sh — a behavioural differential, not a grep. It mirrors the real post-forge soldeer install, post-forge build tree (1634 paths, symlinks included — exactly the state soldeer push packages from) into a throwaway git repo and evaluates both .soldeerignore revisions under gitignore semantics, which is what soldeer's ignore crate implements. Result: 1566 paths excluded under origin/main, 1566 under this branch, diff empty. So removing the three lines provably changes nothing about what gets published today. It fails on base in the only sense a no-op change permits — point it at a revision that drops a live entry and the diff is non-empty, which is the mutation ledger below.

• Mutations applied: negctl.sh drops each kept entry one at a time and re-evaluates, proving the differential can see a real change rather than being a zero-match filter that always says "identical":

  mutant	excluded	delta	verdict
  baseline	1566	—	—
  drop /dependencies	297	−1269	KILLED
  drop /out	1291	−275	KILLED
  drop /cache	1565	−1	KILLED
  drop /remappings.txt	1565	−1	KILLED
  drop .pre-commit-config.yaml	1565	−1	KILLED
  drop .vscode	1566	0	survives — path absent in this checkout
  drop .DS_Store	1566	0	survives — path absent in this checkout

  The two survivors are the point of this PR's scope correction, not a gap: a .soldeerignore entry whose path is absent is unobservable by construction, which is precisely why survivorship cannot be used as evidence for deleting one. .vscode and .DS_Store are absent on a CI checkout but present on a real developer machine at soldeer push time. The same is true of all seven entries the issue asked to delete, and it is why they are kept.

  Harness honesty note: the first run reported .pre-commit-config.yaml as SURVIVED. That was a lying harness, not a finding — find -type f skipped it because git-hooks.nix installs it as a symlink into the nix store. Fixed to \( -type f -o -type l \), after which it kills; every count above is post-fix.

• Oracle: the filesystem, foundry.toml, and other repos' actual contents — all independent of the .soldeerignore text being edited. Each of the three removed paths was checked for existence, for a .gitignore entry that could produce it, and for any producer at all: libs = ['dependencies'] for /lib; zero gitlinks from git ls-files --stage | grep ^160000 plus rainix's no-submodules gate for .gitmodules. A positive control closes the loop the other way — each removed pattern was re-added alone against a probe tree that does contain its path, and all three correctly excluded it, so "diff empty" is caused by path absence and not by a dead evaluator. The keep decision's oracle is the GitHub contents API for raindex and rainlang, which is what establishes the seven filenames are real and correctly spelled.

• Category check: the issue asks for lines 10, 21, 23 (submodule-era) and 2, 3, 4, 5, 12, 16, 29 (template-copied). Covered 10, 21, 23. Deliberately not covered: 2, 3, 4, 5, 12, 16, 29 — the premise behind that half of the issue (absent path ⇒ dead entry) is wrong for a publish filter, reasoned above; deleting them would silently widen what soldeer push publishes. The issue's "no reference to .gitmodules, lib/ or foundry.lock outside dependencies/" is met — grep leaves only CLAUDE.md:15, which the issue itself rules correct. CI green below. The issue should be closed as done on the submodule half and the template half retracted; if the maintainer disagrees, this is the line to push back on.

CI

Full rainix-sol suite run locally in nix develop github:rainlanguage/rainix/53e96a7d0a97d7c7c75c3b2412521324776fdac6#sol-shell — the exact RAINIX_SHA the reusable workflows pin — all green on the final tree:

job	command	result
legal	reuse lint	compliant with REUSE 3.3, exit 0
static	slither .	92 contracts, 98 detectors, 0 results, exit 0
static	forge fmt --check	exit 0
static	rainix-sol-single-contract	exit 0
test	forge test	16 suites, 85 passed, 0 failed, 0 skipped

legal is the job at risk from touching this repo's metadata files and it is clean. FLARE_RPC_URL was supplied so the fork suites (FlareInterfacesProd, LibFtsoV2LTS, LibSceptreStakedFlare, LibDineroFlrEth, LibOpFtsoCurrentPricePair, …) actually ran against Flare rather than erroring out — the 85 includes them.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Chores
  • Updated project tracking configuration so previously ignored dependency and lock files can now be included.

[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: 8aa0dd97-fff6-4514-a077-0ff4843e6da7

📥 Commits

Reviewing files that changed from the base of the PR and between b634aad and fc81ab1.

📒 Files selected for processing (1)

• .soldeerignore

💤 Files with no reviewable changes (1)

• .soldeerignore

---

Walkthrough

The change updates .soldeerignore by removing obsolete /foundry.lock and /lib entries. /foundry.toml remains ignored.

Changes

Soldeer ignore cleanup

Layer / File(s)	Summary
Update ignored paths .soldeerignore	Removed /foundry.lock and /lib from the ignored paths. /foundry.toml remains ignored.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Merge Risk: ⚪ Minimal · up to fc81a

This change removes obsolete submodule-era ignore rules without changing the current Soldeer package contents, so no actionable merge-blocking risk remains after normal checks and review.

Possibly related issues

• rainlanguage/rainlang#553 — Removes the same obsolete submodule-era entries from .soldeerignore.
• rainlanguage/raindex#2834 — Addresses the same cleanup pattern for .soldeerignore.
• rainlanguage/rain.tofu.erc20-decimals#30 — Removes stale submodule-related ignore entries after migration to Soldeer.
• rainlanguage/rain.datacontract#26 — Targets the same obsolete .soldeerignore references.
• rainlanguage/rain.intorastring#20 — Removes obsolete .soldeerignore entries for submodule-era paths.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Linked Issues check	⚠️ Warning	The PR removes the three submodule-era entries but leaves the seven template-derived paths required by issue #208.	Remove the seven template-derived entries listed in issue #208, then verify the required reference checks and CI checks pass.

✅ 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 describes the removal of the three obsolete submodule-era entries from .soldeerignore.
Out of Scope Changes check	✅ Passed	The changes are limited to removing obsolete entries from .soldeerignore, which matches the linked issue scope.
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 208-remove-submodule-era-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.}