Put back the swept content that reached main

[thedavidmeister]

Five comment-sweep commits reached main and took prose with them that was not
a restatement. #102 merged without 845e591, the restoration commit on
hash-68, and 6e24ffd had only its src half put back. This is the full
sweep of main, not just the four known losses.

How the set was found

Every non-merge commit reachable from main since 2026-09-13 was checked for
comment-only deletions, and every deleted // / /// line was grepped against
the current tree. Six sweep commits are on main: 65499d5, 7566d57,
8532c38, 926dc2b, 9e711a8, c380644, plus 6e24ffd's test half. Lines
that came back under different wording (the Foo citation now in
test/lib/LibFooOracle.sol, testCombineHashesGas, the src title block) are
already restored and are not touched here. ccc3720, ea03286, f1f7ded,
8e963c6 and 899aff8 deleted comments as part of substantive rewrites and
are left alone.

Restored

test/src/lib/AbiEncodeGas.t.sol — 6e24ffd cut the title-block claims in
src/lib/LibHashNoAlloc.sol and this file's docstrings in one commit. The src
half is back on main; this half never was, so the file that exists to measure
the cost bullet no longer says which bullet, and the ABI length arithmetic in
each assertEq has nothing naming its terms. The suite docstring, the two test
docstrings and the one naming encodeGas's unnamed tuple returns come back
verbatim; the bullet they quote is on main word for word.

test/lib/LibHashSlow.sol — the title block is the only place "Slow" is
defined as allocation rather than speed, and the only statement that
hashBytesSlow is a value oracle with nothing to measure against, which is why
testHashBytesCostsTheSameAsBuiltin asserts equal gas where its siblings assert
cheaper. Restored with the hashBytesSlow docstring that says why.

test/src/lib/LibHashNoAlloc.t.sol — the suite docstring. Every other test
contract in the repo carries one; 9e711a8 cut this one and #96 did not put it
back. Restored to the suite as it now stands rather than verbatim: the
empty-input HASH_NIL identities it used to list moved to the cross-type suite
with ccc3720.

test/src/lib/HashPattern.t.sol — testBytesTrueLength's docstring, the
hex"01" / hex"0100" collision the test is about, which left it the only
undocumented test in its file. And the // Hash C and D, already in scratch, to produce the final hash E step comment, the last of the README's five steps and
the only one left uncommented.

test/src/lib/MemoryLayout.t.sol — testDeeplyNestedStructIsOnePointerWordPerLevel
in the corrected "2 words at each of the two outer levels" form 7566d57 cut,
not the "each level" form it had replaced, and
testNewFooArrayAllocatesListThenElements, which carries the allocation claim
ea03286 split out of testFooListIsWordList.

test/src/lib/HashPatternFold.t.sol — testFoldPrefixIsStepwiseCombine's
mapping of its locals onto the README's N, A, B, C, D, which nothing in the code
carries, and testFoldSingletonIsNotItem's "Nil hash prefix" citation, the one
its sibling testFoldEmptyIsNilHash kept.

test/src/lib/LibHashNoAlloc.crossType.t.sol — struct TwoBytes, both
pointer-only-struct tests, and the keccak256(s, 0) comment, which is the only
thing saying that a zero-length hash is deliberate.

Left cut

These were restatements and the sweep was right about them:

• 65499d5's sign-extension comment: the docstring directly above it already
  states the sign-extension claim, and the line below is the lint directive.
• 926dc2b's take docstring and testLeafNodeCollision docstring: the first
  paraphrases a three-line body, the second restates an assertion that
  hashBytes's NatSpec already states as a property.
• 9e711a8's per-test one-liners in LibHashNoAlloc.t.sol and the per-function
  abi.encodePacked notes in LibHashSlow, which the restored title block
  covers.
• 926dc2b's allocation paragraph on testFooListIsWordList: that test no
  longer measures allocation. The claim is restored on
  testNewFooArrayAllocatesListThenElements, which does.

QA

• Discriminating tests: n/a - comments only. The diff is 55 inserted lines and
  zero deleted, every inserted line a /// or // comment, so there is no
  behaviour for a test to discriminate and nothing that could fail on base.
• Mutations applied: n/a - a docs-only diff has no executable line to mutate.
  The existing suite is untouched and is the gate this branch stays green
  against.
• Oracle: the source each restored claim makes a claim about, read as it stands
  today rather than as it stood when the line was cut. The AbiEncodeGas
  docstrings were checked against the cost bullet currently in
  src/lib/LibHashNoAlloc.sol and against each assertEq on encoded.length;
  LibHashSlow's title block against which tests use it as a gas oracle and
  which as a value oracle; testBytesTrueLength against README.md line 368 and
  its own three assertions; the fold letters against README "Handling pointers"
  N-A-B-C-D and the five locals they name; the nested-struct docstring against
  fmp1 - midPtr == 0x40 and fmp2 - outermostPtr == 0x40, which is two outer
  levels and not three; testNewFooArrayAllocatesListThenElements against
  fmpAfter - ptr == 0x20 + n * 0x20 + n * 0x80 and its w2/w3 zero-slot
  reads. Two blocks were deliberately not restored verbatim because the
  verbatim text is now false: the LibHashNoAlloc.t.sol suite docstring, whose
  empty-input identities left that suite in ccc3720, and the nested-struct
  docstring, where 8532c38's "each level" form is the one docs: correct five inaccurate statements in the README and test docs #102 existed to
  correct.
• Category check: the task names LibHashSlow's title block,
  testBytesTrueLength, the Yul step comment and the nested-struct correction,
  and asks for a full check of 926dc2b and 8532c38 rather than those four
  alone. Covered: all four, the remainder of both named commits, the remainder
  of 7566d57, 9e711a8 and c380644, and 6e24ffd's test half, with
  65499d5 checked and deliberately left cut. The hash-55 / README: state each hash function's preimage as its API #88 Releases
  section is out of scope under the ruling that release-lifecycle documentation
  lives in rainix only, and is untouched.

🤖 Generated with Claude Code

https://claude.ai/code/session_01V8ViHcKLVk2YoS2joH4HdN

[coderabbitai]

Warning

Review limit reached

Next included review available in 31 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Advanced

Run ID: f2bf1b60-5ce5-4a28-9dcd-fac22fa392b5

📥 Commits

Reviewing files that changed from the base of the PR and between ca72f31 and 1281a90.

📒 Files selected for processing (7)

• test/lib/LibHashSlow.sol
• test/src/lib/AbiEncodeGas.t.sol
• test/src/lib/HashPattern.t.sol
• test/src/lib/HashPatternFold.t.sol
• test/src/lib/LibHashNoAlloc.crossType.t.sol
• test/src/lib/LibHashNoAlloc.t.sol
• test/src/lib/MemoryLayout.t.sol

---

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.}