Bump rain-datacontract 0.1.0 -> 0.1.3 (and rain-solmem 0.1.26 it pins) - #13

What

Bumps rain-datacontract 0.1.0 -> 0.1.3, the latest published revision.

Why rain-solmem rides along

rain-datacontract 0.1.3 hard-pins rain-solmem-0.1.26/... import paths (LibDataContract re-exports LibPointer), and with recursive_deps = false those resolve through this repo's declared pins. So the transitive rain-solmem pin moves 0.1.3 -> 0.1.26 with it. Nothing else in the compile graph imports solmem: this repo's own sources never do, and neither the vendored rain-string 0.2.0 nor rain-math-float 0.1.7 sources reference it — the old foundry.toml comment claiming rain-string -> rain-solmem was stale and is corrected in this diff (mirrors rain.math.float#270, where the same pair moved together in the library repo).

Dropped API check

datacontract 0.1.3 dropped its write API, including the DataContractMemoryContainer type. Everything this repo actually calls survives: every call site is LibDataContract.contractCreationCode (DecimalFloatDeploySuites, LibEtchLogTables, the deploy tests). The only casualty is test/abstract/LogTest.sol's type import and its dead using LibDataContract for DataContractMemoryContainer; directive — removed, no behaviour behind them.

Deploy pins

contractCreationCode is byte-identical between 0.1.0 and 0.1.3 (only its DataTooLarge error declaration moved from the library into ErrDataContract.sol, same signature), and the solmem import is an unused re-export. forge script script/Build.sol was re-run after the bump: the src/generated/candidate/ snapshots did not move — same log-tables creation code, same DecimalFloat bytecode, same Zoltu addresses. The frozen src/generated/0_1_1/ record is untouched.

Scope

• foundry.toml + soldeer.lock: the two version pins, plus the corrected solmem-edge comment.
• Versioned rain-datacontract-0.1.0/ -> rain-datacontract-0.1.3/ import paths across src/abstract/DecimalFloatDeploySuites.sol, script/lib/LibEtchLogTables.sol and four test files.
• test/abstract/LogTest.sol: dead write-API references removed.
• No remappings.txt in this repo (gitignored, soldeer regenerates it), so nothing to prune there.

QA

• Discriminating tests: n/a — dependency pin bump with zero behavioural intent; no new behaviour exists to discriminate. The gate is the full existing suite plus the Build-script currency check that the candidates did not move.
• Mutations applied: n/a — the diff contains no executable logic to mutate (version pins, a comment fix, import paths, deletion of a dead using directive; the compiler kills any wrong path/symbol edit with a build failure).
• Oracle: the pre-bump suite on main (same tests, same assertions) plus rain-datacontract's 0.1.0..0.1.3 source diff, read in full: contractCreationCode byte-identical, read/readSlice doc-only changes, write API deleted (unused here), DataTooLarge moved files with an unchanged signature. Candidate pins re-derived by forge script script/Build.sol and byte-identical to main's.
• Category check: no issue — housekeeping dep bump per the stale-deps campaign; covered: pin bump, forced transitive solmem pin, dropped-API audit of every call site, candidate regeneration, full-suite verification.

Verification

Full suite locally in the flake shell: 71 passed, 6 failed — the 6 failures are exactly the live-network prod-deployment fork tests (testProdDeployment*, testSuitesLiveOnEverySupportedNetwork) failing at vm.createSelectFork: environment variable ... not found, which need the CI RPC secrets and fail identically on an unmodified checkout. CI is the gate for those.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Chores

  • Updated supporting package versions for improved compatibility and maintenance.
  • Refreshed related dependency references throughout deployment and testing workflows.
  • Expanded dependency documentation to clarify transitive package requirements.

• Refactor

  • Removed an unused data contract utility reference from test infrastructure without changing functionality.