Remove submodule-era residue from .soldeerignore

[thedavidmeister]

Closes #553

What changed

Three lines removed from .soldeerignore. That is the whole diff:

--- a/.soldeerignore
+++ b/.soldeerignore
@@ -7,7 +7,6 @@
 .git
 .github
 .gitignore
-.gitmodules
 .pre-commit-config.yaml
 .prettierignore
 .soldeerignore
@@ -18,9 +17,7 @@ CLAUDE.md
 /dependencies
 /flake.lock
 /flake.nix
-/foundry.lock
 /foundry.toml
-/lib
 /out
 /remappings.txt
 /slither.config.json

The property being applied

.soldeerignore is a publish filter, not a description of the tree. An entry naming a file that is absent today is a forward-looking rule — it keeps that file out of the soldeer package if anyone ever adds it — so "this path does not exist" is not on its own a reason to delete an entry. Deleting one on that basis removes protection.

These three qualify on a different property: each names part of the git-submodule dependency layout, which this repo does not use and cannot go back to. The rule they encode can never fire again, so there is no protection to lose.

• .gitmodules and /lib are the submodule vendoring layout itself. Rainix CI runs a no-submodules check that fails on a root .gitmodules or any committed gitlink, so no future commit can reintroduce either path for these entries to filter. Ran that exact check against this branch: no-submodules: clean, and git ls-files --stage | grep ^160000 is empty. Separately, foundry.toml sets libs = ['dependencies'], so forge resolves under dependencies/ and never creates lib/ on its own either. The /lib pattern is rooted and never matched src/lib/, which is untouched.
• /foundry.lock is Foundry's git-submodule lockfile: it records the commit each dependency vendored under lib/ is pinned to, so forge install / forge update can restore identical submodule revisions. It is only meaningful alongside that same banned layout. soldeer.lock is the live lockfile here.

Not touched, on purpose

Every other entry stays, including the ones with no file on disk today, because absence is not the test:

• .DS_Store and /deployments are absent from a clean checkout by design — OS junk and deploy-script output that IS present when soldeer push runs. They still filter. Likewise .pre-commit-config.yaml, /out, /cache and /dependencies, which are generated at forge soldeer install / forge build time.
• Audited the remaining entries for the misspelling case (an entry that was meant to protect a real file but typos its name, where the fix is to correct it in place, not drop it). There are none here: every other entry resolves to a real path, and the only two absent are the two above, which are exempt by design rather than by typo.
• REUSE.toml is the opposite kind of file — its annotations describe files that actually exist — but it needs no change either: it annotates none of the three removed paths, and every path it does annotate resolves.
• CLAUDE.md makes no stale claim that dependencies are submodules under lib/. Left alone.
• The same residue in other rainlanguage repos is out of scope per the issue. flow and rain.tier.interface genuinely still use submodules and must keep their entries.

QA

• Discriminating tests: forge soldeer push --dry-run packaging-equivalence check, run in sol-shell at the CI-pinned rainix SHA. Packed the repo with the new .soldeerignore and with main's, then compared every zip entry by name + uncompressed size + CRC32. 525 entries both ways, listings byte-identical. Publishing is the only observable .soldeerignore has — it decides nothing but what soldeer push uploads, and rainix-static soldeer-gate hashes exactly that packed content — so an identical listing is proof the removal is inert today. There is no code in this diff, so no test fails on base: the correct expected result is a null, and the control below is what gives that null meaning.
• Mutations applied: no executable code in the diff, so no source line to mutate. The equivalent — a positive control proving the equivalence check is not vacuously green — was applied: appending /README.md (an ignore for a file that does exist) to .soldeerignore dropped the packed set from 525 to 524 entries, losing exactly README.md 4458 eed4a9dd. The harness does detect a live ignore line; it reported no difference for these three because their rule is dead, not because it cannot see one.
• Oracle: for the forward-looking half of the claim — can the filtered path ever come back? — the oracle is rainix's own no-submodules binary plus foundry.toml's libs = ['dependencies'], both external to this repo's .soldeerignore. For the inert-today half, the filesystem and git index (ls, git ls-files --stage). Expected packed contents derive from what is on disk, never from the ignore file under test.
• Category check: the issue asks for (a) .soldeerignore lines 10, 21 and 23 removed, (b) no reference to .gitmodules, lib/ or foundry.lock anywhere in the tree outside dependencies/, (c) CI green on rainix-sol test / static / legal. Covered a, b and c. (b) verified by grepping the whole tree excluding .git/ and dependencies/: the only two hits before this change were .soldeerignore:10 and .soldeerignore:21; every lib hit is src/lib/ or a dependency's src/lib/, neither of which the rooted /lib pattern ever matched. Zero hits after.

Verification

Run locally in sol-shell at the CI-pinned rainix SHA 53e96a7d0a97d7c7c75c3b2412521324776fdac6, and in CI:

• legal — reuse lint: 469/469 files with copyright and license information, 0 missing, compliant with REUSE 3.3. Green locally and in CI. This was the job most at risk from the change, since dropping a REUSE.toml annotation is what breaks it — hence leaving REUSE.toml untouched.

• static — green in CI. Locally: forge fmt --check clean, slither . clean, no-submodules clean.

• test — CI red, but pre-existing on main and identical, not caused by this change. main at a6b7ad73 (this branch's merge base) fails the same 5 tests with the same counts:

  	main a6b7ad7	this branch
  failing	5	5
  passing	1589	1589
  tests	testProdDeployArbitrum, testProdDeployBase, testProdDeployBaseSepolia, testProdDeployFlare, testProdDeployPolygon — all [FAIL: Interpreter not deployed]	identical

  These are fork tests asserting the prod interpreter is live on each chain; they fail because it is not deployed there, which no .soldeerignore edit can affect. rainix-sol has been red on main on every run since at least 2026-07-03. Zero test delta from this PR. This is a real pre-existing defect and is not fixed here — it needs a deployment, not a code change, and is outside the scope of Remove the submodule-era residue — .soldeerignore still names foundry.lock, .gitmodules and /lib #553.

  (A local forge test additionally fails on unset <NETWORK>_RPC_URL env vars, which CI's rpc-preflight step supplies; CI is authoritative for this job.)

🤖 Generated with Claude Code

Summary by CodeRabbit

• Chores
  • Updated project ignore settings so .gitmodules, foundry.lock, and the lib directory are no longer excluded from version control.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: f8809c46-148f-4b63-a861-55191307c9d3

📥 Commits

Reviewing files that changed from the base of the PR and between a6b7ad7 and a87eafc.

📒 Files selected for processing (1)

• .soldeerignore

💤 Files with no reviewable changes (1)

• .soldeerignore

---

📝 Walkthrough

Walkthrough

The change removes obsolete submodule-era entries from .soldeerignore. The /foundry.toml entry remains ignored.

Changes

Soldeer ignore cleanup

Layer / File(s)	Summary
Remove obsolete ignore entries .soldeerignore	Removes /foundry.lock and /lib from the ignore list. Retains /foundry.toml.

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

Merge Risk: ⚪ Minimal · up to a87ea

This change removes obsolete submodule-era filter entries without changing the published package contents; no actionable merge-blocking risk remains after normal checks.

Possibly related issues

• rainlanguage/raindex#2834 — Removes the same stale submodule-era entries during Soldeer cleanup.
• rainlanguage/rain.tofu.erc20-decimals#30 — Removes obsolete .soldeerignore entries for the same migration.
• rainlanguage/rain.flare#208 — Covers removal of stale submodule-era ignore entries.
• rainlanguage/rain.verify#30 — Removes obsolete .gitmodules, foundry.lock, and lib references.
• rainlanguage/rain.extrospection#47 — Addresses the same .soldeerignore cleanup.
• rainlanguage/rain.datacontract#26 — Removes the same obsolete ignore entries.

Possibly related PRs

• rainlanguage/rainlang#500 — Both PRs modify .soldeerignore, but they change different entries.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 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 obsolete submodule-era entries from .soldeerignore.
Linked Issues check	✅ Passed	The PR removes all three obsolete .soldeerignore entries required by issue #553 and preserves valid ignore entries.
Out of Scope Changes check	✅ Passed	The PR changes only .soldeerignore and removes exactly the entries identified as out of date by issue #553.
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 553-soldeerignore-submodule-entries

---

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]

🤖 ai:producer
Blocked-on: repo not migrated to the split release lifecycle: rainix-sol / test is red on the five testProdDeploy fork tests (testProdDeployArbitrum/Base/BaseSepolia/Flare/Polygon), each asserting 'Interpreter not deployed' against the single current deployment pin in src/lib/deploy/LibInterpreterDeploy.sol. This PR's entire diff is 3 deletions in .soldeerignore, a file read only by 'forge soldeer push' and never by forge build or forge test, so the diff structurally cannot move a test result. The red is the pre-split premerge deploy shape, not this change, and it clears when the repo adopts per-release deploy snapshots.
blocked-by #545