blocked-by: rain.math.float.deploy Soldeer publish — strip deploy + Rust/npm surface (library-only)

[thedavidmeister]

DRAFT — DO NOT MERGE. Blocked-by: rain.math.float.deploy Soldeer publish.

Strips the deploy surface and the Rust/npm surface from rain.math.float, leaving the pure Solidity math library that publishes only the rain-math-float Soldeer package. Executes the plan ruled in #262 (the migration-plan comment + the Hard Case A ruling). The concrete, deploy pins/snapshot, deploy scripts/tests, and the Rust/WASM/npm bindings now live in rain.math.float.deploy.

🔒 Blocked-by (ordering constraint)

Must not land until rain.math.float.deploy has published its rain-math-float-deploy Soldeer package, so external importers of LibDecimalFloatDeploy / the deployed addresses have somewhere to repoint. Tracker: rainlanguage/rain.math.float.deploy#1 (step 4 — sol-v* tag → Soldeer publish — is still pending). Source ruling + plan: #262.

Repointing other org consumers is out of scope here (2026-08-20 human narrowing).

Removed (77 files)

Deploy Solidity (Tier 1 + Tier 2, un-gated):

• src/concrete/DecimalFloat.sol, src/generated/LogTables.pointers.sol, src/lib/deploy/LibDecimalFloatDeploy.sol
• script/Deploy.sol, script/BuildPointers.sol, script/check-published-deploy-constants.sh, script/lib/LibEtchLogTables.sol
• test/src/concrete/*.t.sol (31), test/src/lib/deploy/*.t.sol (4), test/src/lib/table/LibLogTable.bytes.t.sol

Rust/npm surface + CopyArtifacts bridge:

• crates/** (15), Cargo.toml, Cargo.lock, package.json, package-lock.json, tsconfig.json, .prettierignore, test_js/**, scripts/*.js
• script/CopyArtifacts.sol, script/lib/LibCopyArtifacts.sol, test/script/CopyArtifacts.t.sol
• test/concrete/TestDecimalFloat.sol + its two wrapper tests (TestDecimalFloat.unpack.t.sol, DecimalFloat.packLossless.t.sol) — per Correction 1; direct packLossless/unpack coverage stays in test/src/lib/LibDecimalFloat.pack.t.sol
• Workflows: copy-artifacts.yaml, manual-sol-artifacts.yaml, rainix-rs-static.yaml, rainix-rs-test.yaml, rainix-rs-wasm.yaml

Kept

• Pure-math library: src/lib/{LibDecimalFloat, format, parse, implementation, table/LibLogTable} + src/error/*; its full test suite; the slow reference impls (test/lib/*).
• rain-datacontract dep (the test log-tables helper needs it).
• Lifecycle: package-release.yaml (now publishes onlyrain-math-float Soldeer), rainix-sol.yaml, pr-assessment.yaml.

Added (Hard Case C — mandatory)

test/abstract/LogTest.sol rewritten self-contained: rebuilds the combined log/anti-log tables purely from LibLogTable source (toBytes of the five tables + LOG_TABLE_DISAMBIGUATOR, byte-identical to the old frozen combinedTables()), deploys them via LibDataContract at a create address, and hands the address to the 8 transcendental tests. No Zoltu deploy pin, no src/generated.

Config edits

• foundry.toml: drop rain-deploy + rain-sol-codegen deps, [rpc_endpoints], [etherscan], fs_permissions, ffi (no retained test uses vm.ffi). Pragma stays ^, optimizer_runs untouched.
• flake.nix: drop test-wasm-build + test-js-bindings tasks.
• remappings.txt / soldeer.lock regenerated (deps: forge-std, @openzeppelin-contracts, rain-solmem, rain-string, rain-datacontract).
• .gas-snapshot regenerated for the stripped suite; CLAUDE.md / README.md / REUSE.toml / .soldeerignore / .gitignore library-only.

CI (green locally, nix develop -c)

• forge build → successful
• forge test → 444 passed, 0 failed, 0 skipped (48 suites); transcendental tests pass on the new helper
• slither . → 0 results (14 contracts)
• reuse lint → compliant (78/78 files)

QA

• Discriminating tests: the 8 transcendental tests that consume the new LogTest.logTables() helper — LibDecimalFloat.{log10,pow,pow10,sqrt}.t.sol and LibDecimalFloatImplementation.{internals,log10,lookupLogTableVal,pow10}.t.sol. They fail if the helper deploys the wrong table bytes (transcendental lookups extcodecopy from the tables address; garbage tables give wrong results, and the packed-vs-unpacked and slow-reference cross-checks in these files diverge). Verified: 444 passed / 0 failed / 0 skipped on the stripped tree; before the strip these tests reached the tables via the Zoltu-pinned etch, and they now reach byte-identical tables via the self-contained helper (same 8 tests, same assertions, still green).
• Mutations applied: n/a — this PR removes the deploy/Rust surface and rewrites a test helper; it introduces no new production logic to mutate. The one behavioral risk it does carry — that the rebuilt-from-source tables differ from the deployed tables — is pinned by structural equivalence (below), not left to chance: change the helper's concatenation order or drop the disambiguator and the 8 transcendental tests fail.
• Oracle: independent of the new helper. The deleted test/src/lib/table/LibLogTable.bytes.t.sol asserted LibLogTable.toBytes(logTableDec()) == LOG_TABLES (and the same for the other four tables) against the frozen src/generated/LogTables.pointers.sol snapshot — exactly the five toBytes(...) terms the new helper concatenates. So the helper's output is provably byte-identical to the old LibDecimalFloatDeploy.combinedTables() it replaces. Transcendental expected values are additionally cross-checked against the slow reference implementations in test/lib/.
• Category check:Split rain.math.float into a library half and a new rain.math.float.deploy #262 keep/remove list is the spec. Covered — removed = Tier 1 + Tier 2 (un-gated) Solidity deploy surface + the ruling's Rust/npm addition + TestDecimalFloat (Correction 1); kept = the pure-math surface + tests + rain-datacontract; added = the mandatory Hard Case C helper. No consumer repointing (out of scope per the 2026-08-20 narrowing).

Closes the library-strip step of #262 once unblocked.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: f5a9514d-fd5f-4f72-96df-7af517216f26

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

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.}