SIG-747: remove historical and version reference comments#73
Conversation
The prod fork tests used `vm.createSelectFork(envName)` without a block number, causing baseline runs to drift as live prod state evolves. With the same code that passed at SIG-626 merge time (4/14 09:31 UTC), the `AdminLifecycleFork.test_fee_waterfall_processing_*` assertion changed result on the latest block: realized batch processing now produces a zero backstop NAV delta, so the strict `assertGt` fails. Pin prod fork to block 6029250 (≈ 2026-04-14 09:31 UTC, last green Fork Tests Baseline on main) so the suite is reproducible. Allow `FORK_BLOCK` env override so the pin can be advanced deliberately. Dev fork keeps using latest because the dev chain has no equivalent baseline mapping. Verified locally: full fork suite 22/22 pass on the pinned block; unit suite 1081/1081 still green.
![signals-reviewer[bot]](https://avatars.githubusercontent.com/u/206232752?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary
Comment-only refactoring that removes v0 comparison references and WP v2 version prefixes from contract comments, preparing the codebase for v1 audit readability. A second commit pins prod fork tests to a deterministic block (6029250) to fix baseline drift.
Issues
No issues found.
Suggestions
No suggestions.
Verdict
APPROVE: Comment surgery is clean and symmetric. The three LazyMulSegmentTree edits are mirrored correctly in the verification copy. Remaining comments still accurately describe v1 behavior (leaf-pending skip, l==r flush safety, NAV floor invariant). Equation formulas preserved as audit anchors with only version numbering removed. ForkBaseTest block pinning is sound with env override escape hatch. All tests pass per the implementation report (1081 unit, 22 fork, 3 verification). forge fmt pre-existing drift is correctly out of scope.
There was a problem hiding this comment.
Summary
Comment-only refactoring that removes v0 comparison references and WP v2 version prefixes from contract comments, preparing the codebase for v1 audit readability. A second commit pins prod fork tests to a deterministic block (6029250) to fix baseline drift.
Issues
No issues found.
Suggestions
No suggestions.
Verdict
APPROVE: Comment surgery is clean and symmetric. The three LazyMulSegmentTree edits are mirrored correctly in the verification copy. Remaining comments still accurately describe v1 behavior (leaf-pending skip, l==r flush safety, NAV floor invariant). Equation formulas preserved as audit anchors with only version numbering removed. ForkBaseTest block pinning is sound with env override escape hatch. All tests pass per the implementation report (1081 unit, 22 fork, 3 verification). forge fmt pre-existing drift is correctly out of scope.
There was a problem hiding this comment.
Summary
Comment-only refactoring that removes v0 comparison references and WP v2 version prefixes from contract comments, preparing the codebase for v1 audit readability. A second commit pins prod fork tests to a deterministic block (6029250) to fix baseline drift.
Issues
No issues found.
Suggestions
No suggestions.
Verdict
APPROVE: Comment surgery is clean and symmetric. The three LazyMulSegmentTree edits are mirrored correctly in the verification copy. Remaining comments still accurately describe v1 behavior (leaf-pending skip, l==r flush safety, NAV floor invariant). Equation formulas preserved as audit anchors with only version numbering removed. ForkBaseTest block pinning is sound with env override escape hatch. All tests pass per the implementation report (1081 unit, 22 fork, 3 verification). forge fmt pre-existing drift is correctly out of scope.
1 participant
Context
SIG-747 (Epic: SIG-742, audit-prep). Contract comments carried historical metadata that is meaningless to an audit reader looking at v1 in isolation:
Unlike v0's _forcePushPendingFactor,from older deployments,Per WP v2:,(WP v2 state machine),(WP v2 Eq. 3.11). v0 lives in a separate repository and is not in scope for the v1 audit, so v0 references invite questions about code that is not present. Whitepaper version prefixes and equation numbers add a versioning dimension that drifts independently of the contract.Decisions
l == rskips flush). Only the v0 comparison clauses were removed.Payout_t := Q_{t,τ_t},L_t := ΔC_t - Payout_t) were preserved as audit anchors. Only theWP v2 Eq. X.YYnumbering was dropped.verification/foundry/src/LazyMulSegmentTreeNoLink.solmirrors thecontracts/lib/LazyMulSegmentTree.soledits one-for-one because it is a no-link copy of the same tree.ClmsrMathHarness.exposedSafeExpnotice was changed from "parity tests against v0" to "unit, fuzz, and parity tests" rather than a generic "tests" label, reflecting actual call sites.forge fmt --checkwas intentionally left untouched. Repo-wide formatter drift exists across files outside this scope and rewriting them would violate the comment-only boundary.Risk Areas
_pushPendingFactor,_applyFactorToChildWithFlush): the comment edits are adjacent to leaf-pending overflow safety logic. Reviewer should confirm the remaining wording still describes current v1 behavior accurately._computeReserveDelta: payout/loss formulas are still load-bearing audit annotations. Reviewer should confirm formula text matches the code.verification/foundry/src/LazyMulSegmentTreeNoLink.solparity withcontracts/lib/LazyMulSegmentTree.sol: three comment edits applied symmetrically. Reviewer should diff the two files to confirm no drift.Findings
forge fmt --checkdoes not pass at repo scope or inverification/foundry. The drift predates this branch and was left in place.