cid(review): PASS — codec decode optimization verified, issue resolved

Author: CID Agent

.claude/agent-memory/review/MEMORY.md

@@ -26,17 +26,27 @@ Review patterns, quality gate knowledge, and common issues accumulated across CI
 
 ## Review Shortcuts
 
+- Rust-only internal refactors (no public API changes, no binding crate changes):
+    `cargo test -p   iscc-lib` + `cargo clippy -p iscc-lib -- -D warnings` + `mise run check` is
+    sufficient — skip Maven/Go/Node/WASM tests
+
 - For Java conformance test reviews: verify vector count matches expected (46 total:
     16+5+3+5+3+2+4+3+5), check `mvn test` output for 0 failures, compare structure against Node.js
     conformance tests in `crates/iscc-napi/__tests__/conformance.test.mjs`
+
 - For Go conformance test reviews: same 46 vector count, check `go test -v` output shows all
     subtests pass, verify memory helpers handle empty/nil inputs correctly
+
 - Clippy workspace check is fast (~2s) after initial build — always run it
+
 - Documentation-only changes (READMEs, markdown): `mise run check` + clippy is sufficient — no need
     to run full test suites since no code was modified
+
 - Python-only changes: `mise run check` + `pytest` is sufficient; skip `cargo test` and `mvn test`
     unless Rust/Java code was also modified
+
 - Go-only changes: `mise run check` + `mise exec -- go test ./...` is sufficient
+
 - Full test suite (157 tests) runs in \<1s — always run it for Python changes
 
 ## Gotchas

.claude/context/handoff.md

@@ -1,29 +1,31 @@
-## 2026-02-25 — Optimize codec header decoding with direct bitwise operations
-
-**Done:** Replaced `Vec<bool>` intermediate allocation in `decode_header` and `decode_varnibble`
-with direct bitwise extraction from the byte slice. Added `get_bit` and `extract_bits` helpers that
-operate on `&[u8]` + bit position. Gated `bytes_to_bits` and `bits_to_u32` with `#[cfg(test)]` since
-they're only used by test helpers after the rewrite.
-
-**Files changed:**
-
-- `crates/iscc-lib/src/codec.rs`: Added `get_bit`, `extract_bits`, `decode_varnibble_from_bytes`
-    functions; rewrote `decode_header` to use byte-based decoding; gated `bytes_to_bits` and
-    `bits_to_u32` with `#[cfg(test)]`; updated `test_varnibble_roundtrip` to use new function; added
-    `test_extract_bits_basic` and `test_decode_varnibble_from_bytes_boundary_values` tests
-
-**Verification:** All 261 tests pass (208 unit + 31 algo primitives + 22 text utils). Clippy clean
-(workspace-wide `--all-targets`). All 14 pre-commit hooks pass. Benchmark shows 27% improvement for
-`gen_mixed_code_v0` (1.34µs → was ~1.87µs). Benchmark runs without error.
-
-**Next:** The codec issue (`[normal] Codec header parsing expands bytes to Vec<bool>`) is now
-resolved and can be deleted from issues.md. Consider addressing `DataHasher::update` buffer
-allocation (`[normal]`), iscc-ffi video frame allocation (`[normal]`), or lower-priority items.
-
-**Notes:** The old `decode_varnibble` function was removed entirely (no production or test callers
-remain after updating `test_varnibble_roundtrip`). Clippy suggested two improvements: collapsing the
-nested `if` in the padding check into a single condition, and using `bit_pos.div_ceil(8)` instead of
-`(bit_pos + 7) / 8`. Both applied. The tail extraction behavior is slightly cleaner for invalid data
-(non-zero padding) — it now rounds up to the next byte boundary instead of reconstructing mid-byte
-tail data through `bits_to_bytes`, but this difference is invisible for valid ISCC data where
-padding is always zero.
+## 2026-02-25 — Review of: Optimize codec header decoding with direct bitwise operations
+
+**Verdict:** PASS
+
+**Summary:** Replaced `Vec<bool>` intermediate allocation in `decode_header` and `decode_varnibble`
+with direct bitwise extraction from `&[u8]`. Implementation is clean, minimal, and matches the
+next.md specification exactly. All 261 tests pass, clippy is clean, and all 14 pre-commit hooks
+pass.
+
+**Verification:**
+
+- [x] `cargo test -p iscc-lib` passes — 261 tests (208 unit + 31 algo + 22 text), including 2 new
+    bitwise extraction tests
+- [x] `cargo clippy -p iscc-lib -- -D warnings` clean
+- [x] `bytes_to_bits` gated with `#[cfg(test)]` — no longer in production code path
+- [x] `Vec<bool>` only in `encode_varnibble` and test-gated helpers, NOT in `decode_header` or any
+    decode function
+- [x] `cargo bench -p iscc-lib -- gen_mixed_code_v0` runs without error — 1.35µs median
+
+**Issues found:**
+
+- (none)
+
+**Next:** The codec `Vec<bool>` issue is resolved. Consider addressing `DataHasher::update` buffer
+allocation (`[normal]` in issues.md), iscc-ffi video frame allocation (`[normal]`), or one of the
+`[low]` issues. Alternatively, if there are remaining target.md gaps, those take priority.
+
+**Notes:** The old `decode_varnibble` function was fully removed (not just gated). The roundtrip
+test was properly updated to use the new `decode_varnibble_from_bytes` via `bits_to_bytes`
+conversion. Benchmark shows ~1.35µs for `gen_mixed_code_v0` — consistent with the advance agent's
+reported 27% improvement from baseline.

.claude/context/issues.md

@@ -91,22 +91,6 @@ Fix: add `if vec.len() < 380 { return error }` guard at function entry.
 
 **Source:** [human]
 
-## [normal] Codec header parsing expands bytes to `Vec<bool>`
-
-In `crates/iscc-lib/src/codec.rs:121-127`, `bytes_to_bits` expands every byte into 8 `bool` values
-and returns `Vec<bool>`. This is used in `decode_header` (line 245) which is called on every codec
-operation (decompose, mixed-code hashing, conformance checks, etc.).
-
-Headers are small (typically 2-8 bytes) so the per-call cost is modest, but the pattern is avoidable
-— varnibble decoding can work directly on bytes with bitwise operations, eliminating the
-intermediate allocation entirely.
-
-Fix: replace `bytes_to_bits` + index-based varnibble decoding with direct bitwise extraction from
-the byte slice. After fixing, re-run `cargo bench -p iscc-lib` and compare `gen_mixed_code_v0` and
-`iscc_decompose` timings against the baseline.
-
-**Source:** [human]
-
 ## [normal] `DataHasher::update` copies input data on every call
 
 In `crates/iscc-lib/src/streaming.rs:88-93`, every `update()` call either copies the input via

.claude/context/iterations.jsonl

@@ -270,3 +270,7 @@
 {"ts":"2026-02-25T05:06:26.035498+00:00","iteration":16,"role":"define-next","status":"OK","turns":21,"cost_usd":1.028739,"duration_s":164.6}
 {"ts":"2026-02-25T05:10:42.550926+00:00","iteration":16,"role":"advance","status":"OK","turns":36,"cost_usd":1.27611,"duration_s":256.5}
 {"ts":"2026-02-25T05:14:00Z","iteration":16,"role":"review","status":"PASS","summary":"Java NativeLoader — clean implementation, all 6 criteria pass, 49 Maven tests pass, 14 hooks clean"}
+{"ts":"2026-02-25T05:13:22.856083+00:00","iteration":16,"role":"review","status":"OK","turns":29,"cost_usd":0.949717,"duration_s":160.3}
+{"ts":"2026-02-25T05:36:11.677620+00:00","iteration":17,"role":"update-state","status":"OK","turns":20,"cost_usd":0.736132,"duration_s":213.0}
+{"ts":"2026-02-25T05:39:33.281710+00:00","iteration":17,"role":"define-next","status":"OK","turns":25,"cost_usd":1.111233,"duration_s":201.6}
+{"ts":"2026-02-25T05:46:04.062564+00:00","iteration":17,"role":"advance","status":"OK","turns":38,"cost_usd":2.126975,"duration_s":390.8}