chore(release): v0.8.0

ARCHITECTURE.md

@@ -36,12 +36,14 @@ The unified core library containing all AR functionality:
     -   `kpm::ref_data_set` — `.fset3` reference data I/O and compression modes.
     -   `kpm::types` — KPM data structures and constants.
     -   `kpm::freak` — FREAK descriptor math, homography, and matching utilities (pure Rust port of `WebARKitLib/lib/SRC/KPM/FreakMatcher`):
+        -   `freak::visual_database` — top-level per-frame query orchestrator (M9-1): builds the query keyframe, then runs the two-pass pipeline (feature match → Hough voting → homography → inlier filter, then a homography-guided re-match) across stored reference keyframes. Exposes `query` / `query_from_keyframe` and facade-parity accessors on the database + `FeatureStore` slices (#147, #148).
         -   `freak::math` — linear algebra (matrix operations, linear solvers) and Padé matrix exponential.
         -   `freak::homography` — homography estimation and refinement pipeline.
         -   `freak::hough` — Hough similarity voting (4D bin discretization over translation × angle × scale) for filtering matches by transformation consistency.
         -   `freak::clustering` — K-Medoids partitioning + Binary Hierarchical Clustering (BHC) vocabulary tree for fast approximate-NN search on 96-byte FREAK descriptors. Hamming distance via 24×32-bit bit-magic. Includes a byte-identical port of C++ `vision::FastRandom` / `vision::ArrayShuffle` so the BHC tree topology matches the C++ baseline given the same seed.
         -   `freak::matcher` — `FeatureStore` (points + flat descriptor buffer) and `FeatureMatcher` with three match variants: brute force, BHC-indexed (fast path), and homography-guided (spatial filter via 3×3 inverse + `tr` radius). All three apply the C++ ratio test (default 0.7) and filter by `FeaturePoint::maxima`.
         -   `freak::image_pyramid` — image pyramid construction via `BoxFilterPyramid8u` (box filtering for 8-bit grayscale) and `BinomialPyramid32f` (32-bit floating-point binomial pyramid with scale-space interpolation).
+        -   `freak::gaussian_pyramid` — Gaussian scale-space pyramid driving the DoG detector. Hot binomial-filter passes have NO-FMA SSE4.1 / AVX2 / wasm SIMD paths (runtime-detected, bit-exact against the scalar fallback for dual-mode parity) and rayon-parallelized rows (v0.8.0, #200–#207).
         -   `freak::feature_extraction` — keypoint detection via Difference-of-Gaussians (DoG), dominant orientation assignment via circular gradient voting, and FREAK descriptor computation with native Rust implementation of the FREAK binary pattern and bit-pair comparison.
 -   **Types** (`types`): core data structures (`ARHandle`, `ARParam`, etc.).
 
@@ -50,6 +52,13 @@ The unified core library containing all AR functionality:
 WASM wrapper and JavaScript/TypeScript glue code for browser targets.
 Depends only on `webarkitlib-rs` (the core crate).
 
+Since **v0.8.0 (#161)** it exposes the pure-Rust NFT pipeline to the
+browser: a `WasmKpmHandle` binding for KPM detection (load reference data
+from bytes, detect), `console_log` wiring via `ar_log_init_wasm()`, a dual
+**standard + SIMD** build pipeline (`npm run build:wasm`), and an
+end-to-end browser NFT demo under `crates/wasm/www`. The `cc`/`bindgen`
+build dependencies are optional so the wasm target builds pure-Rust.
+
 ## Feature Flags
 
 | Feature        | Description |
@@ -58,6 +67,7 @@ Depends only on `webarkitlib-rs` (the core crate).
 | `simd`         | Umbrella: enables all SIMD sub-features |
 | `simd-wasm32`  | WASM SIMD128 intrinsics |
 | `simd-x86-sse41` | x86 SSE4.1 intrinsics |
+| `simd-x86-avx2` | x86 AVX2 intrinsics (runtime-detected; NO-FMA for bit-exact scalar parity) |
 | `log-helpers`  | Enable logging infrastructure (installs `env_logger` for desktop/tests, `console_log` for WASM) |
 | `ffi-backend`  | **Opt-in** — compile the C++ FreakMatcher library and generate FFI bindings. Used for cross-validation and the regression-test suite; not required for production tracking. |
 | `dual-mode`    | Enables FFI-based parity tests that validate pure-Rust ports against the live C++ baseline (M6 math/solvers/homography, M7 BHC/matcher, PRNG). Transitively enables `ffi-backend`. Run in CI on Linux/macOS/Windows. |

CHANGELOG.md

@@ -1,8 +1,76 @@
-# Changelog - webarkit/webarkitlib-rs
+# Changelog - webarkit/webarkitlib-rs
 
 All notable changes to this project will be documented in this file.
 
-## [0.7.0] - 2026-06-05
+## [0.8.0] - 2026-06-26
+
+## Milestone — Pure-Rust completeness, validation & polish
+
+### ⚙️ Miscellaneous Tasks
+
+- *(ar2)* Allow too_many_arguments on SIMD get_similarity variants (#180)
+- Clear strict clippy lints in examples/tests/benches (#180)
+- *(codecov)* Exclude examples/benches from coverage targets (#180)
+- *(coverage)* Fix tarpaulin tests/ glob and mirror in codecov (#180)
+- Tighten clippy gate to --all-targets --all-features (#180)
+- Relocate strict clippy gate to kpm-build (ubuntu) job (#180)
+- *(miri)* Add Miri UB validation for pure-Rust code paths (#182)
+- *(miri)* Drop -Zmiri-strict-provenance (trips on third-party deps) (#182)
+- *(miri)* Switch to Tree Borrows to avoid crossbeam-epoch false positive (#182)
+- *(miri)* Enable -Zmiri-disable-isolation for tempfile-using tests (#182)
+- *(miri)* Enable -Zmiri-ignore-leaks for rayon thread pool (#194)
+- *(miri)* Promote to required gate, split workflow, add README badge (#182)
+- *(miri)* Skip docs-only changes via paths-ignore + same-name shim
+- *(benchmarks)* Harden against flaky external C-library downloads (#204)
+
+### ⚡ Performance
+
+- *(ci)* Scope Miri job to unsafe boundaries via cfg_attr(miri, ignore) (#194)
+- *(ci)* Annotate remaining real-image tests (#194 follow-up)
+- *(kpm)* Gaussian scale-space pyramid — benchmark + rayon parallelization (#209)
+- *(kpm)* Box-filter Pyramid downsample — criterion benchmark + SIMD (#211)
+
+### 🐛 Bug Fixes
+
+- *(core)* Clean up trivial clippy lints under --all-targets --all-features (#180)
+- *(kpm)* Tighten dual-mode FFI extern cfg to match caller cfg (#180)
+- *(kpm)* Replace unaligned transmute in hamming_distance_96 (#192)
+- *(ci)* Disable cargo-bin cache on kpm-build to stop stale rustup-init shim (#134)
+
+### 💼 Other
+
+- *(wasm)* Make cc/bindgen optional, fix wasm dead-code (#161 goal 1)
+
+### 📚 Documentation
+
+- *(benches)* Refresh BENCHMARKS.md for criterion 0.8
+- *(agents)* Codify #180 clippy conventions and un-ignore .agents/
+- *(kpm)* Mark box-filter Pyramid as kept-but-unused reference (#203)
+- *(wasm)* Note .fset3 / KPM detection in the NFT demo description (#161)
+- Update README + ARCHITECTURE for the v0.8.0 milestone
+
+### 🕸️ WebAssembly & Emscripten
+
+- *(wasm)* Wire console logger + clean SIMD wasm build; add .fset3 asset (#161 goals 2,3)
+- *(wasm)* KPM detection binding + browser NFT demo (#161 goal 4)
+
+### 🚀 Features
+
+- *(kpm)* Facade-parity convenience accessors on VisualDatabase + FeatureStore slices (#148)
+- *(example)* Nft_marker_gen .fset3 via RustFreakMatcher — drop ffi-backend (#179)
+
+### 🚜 Refactor
+
+- *(ar)* Convert Default::default() + field reassign to struct-init (#180)
+- *(kpm)* Factor VisualDatabase::query matching loop + add query_from_keyframe (#147) (#217)
+
+### 🧪 Testing
+
+- *(core)* Add unit tests for relocated math/pattern helpers (#180)
+- *(homography)* Widen cauchy_cost tolerance to 1e-5 for Miri (#194)
+- *(kpm)* Raise M9 coverage — exclude examples + edge tests (#177) (#218)
+- *(ci)* Verify ffi-backend tests + example link on macOS (#119)
+## [0.7.0] - 2026-06-05
 
 ## Milestone 9 — VisualDatabase & pure Rust backend
 

Cargo.toml

@@ -9,7 +9,7 @@ exclude = [
 resolver = "2"
 
 [workspace.package]
-version = "0.7.0"
+version = "0.8.0"
 authors = ["kalwalt <github@kalwaltart.it>"]
 edition = "2021"
 description = "A high-performance, memory-safe Rust port of WebARKitLib (ARToolKit) for native and WASM."

README.md

@@ -14,7 +14,7 @@
 This project aims to provide a pure-Rust implementation of the core ARToolKit algorithms, targeting both **native** systems and **WebAssembly (WASM)** for high-performance augmented reality in the browser.
 
 > [!NOTE]
-> This project is currently a **Work in Progress**. Core marker detection and pose estimation are functional. KPM/NFT (Natural Feature Tracking) is in an early stage with partial Rust porting -- a fully idiomatic Rust KPM pipeline with a working example is the primary goal for v1.0.0.
+> Core marker detection and pose estimation are functional, and as of **v0.8.0** the **KPM/NFT (Natural Feature Tracking) pipeline is fully ported to pure Rust** — end-to-end keypoint matching, marker generation (`.iset`/`.fset`/`.fset3`), and runtime tracking all run with **no C++ toolchain**, natively and in the browser via WebAssembly. The project remains a **Work in Progress** toward v1.0.0 (broader API docs, multi-marker support, and a dedicated KPM benchmark are the remaining focus areas).
 
 ## 🌟 Key Features
 
@@ -40,7 +40,7 @@ Add `webarkitlib-rs` to your `Cargo.toml`:
 
 ```toml
 [dependencies]
-webarkitlib-rs = "0.7"
+webarkitlib-rs = "0.8"
 ```
 
 ### Pure Rust tracking (no C++ compiler required)
@@ -66,7 +66,7 @@ need to do anything special:
 
 ```toml
 [dependencies]
-webarkitlib-rs = "0.7"  # default backend is pure Rust
+webarkitlib-rs = "0.8"  # default backend is pure Rust
 ```
 
 ```bash
@@ -83,7 +83,7 @@ NFT tracking**:
 
 ```toml
 [dependencies]
-webarkitlib-rs = { version = "0.7", features = ["ffi-backend"] }
+webarkitlib-rs = { version = "0.8", features = ["ffi-backend"] }
 ```
 
 > When installing from crates.io, no extra setup is required — the C++
@@ -248,7 +248,7 @@ Enable the `log-helpers` feature and call the bundled initializer once in your b
 
 ```toml
 [dependencies]
-webarkitlib-rs = { version = "0.7", features = ["log-helpers"] }
+webarkitlib-rs = { version = "0.8", features = ["log-helpers"] }
 ```
 
 ```rust
@@ -429,12 +429,16 @@ The workspace contains two crates:
     - Cross-platform / cross-stack matcher determinism — Rust `HashMap` → `BTreeMap` and upstream C++ `unordered_map` → `std::map` ([WebARKitLib#39](https://github.com/webarkit/WebARKitLib/pull/39)).
     - Hand-annotated absolute corner-error regression gate (browser-based annotation tool + 5 fixtures + Linux CI gate). Finding: pure-Rust backend is more accurate than C++ on `pinball-demo` (5.27 px vs 18.79 px max corner error).
     - Cross-stack parity gate against `@webarkit/jsartoolkit-nft@1.10.0` ([jsartoolkitNFT#584](https://github.com/webarkit/jsartoolkitNFT/pull/584)) — guarantees native Rust pose matches what production WASM consumers see.
+- **v0.8.0 — Pure-Rust completeness, validation & polish**: Closed out the remaining gaps on top of M9:
+  - **Full WASM/NFT support** ([#161](https://github.com/webarkit/WebARKitLib-rs/issues/161)): `WasmKpmHandle` KPM-detection bindings, `console_log` wiring, clean dual (standard + SIMD) wasm builds, and an end-to-end **browser NFT demo** of the pure-Rust pipeline.
+  - **Pure-Rust `.fset3` marker generation** ([#179](https://github.com/webarkit/WebARKitLib-rs/issues/179)): `nft_marker_gen` now produces `.iset` + `.fset` + `.fset3` entirely via `RustFreakMatcher` — the `ffi-backend` is no longer needed for marker creation.
+  - **`VisualDatabase` ergonomics** ([#147](https://github.com/webarkit/WebARKitLib-rs/issues/147), [#148](https://github.com/webarkit/WebARKitLib-rs/issues/148)): factored the per-frame query loop, added `query_from_keyframe`, and facade-parity accessors on `VisualDatabase` + `FeatureStore` slices.
+  - **Gaussian scale-space pyramid performance** ([#200](https://github.com/webarkit/WebARKitLib-rs/issues/200)/[#201](https://github.com/webarkit/WebARKitLib-rs/issues/201)/[#207](https://github.com/webarkit/WebARKitLib-rs/issues/207)): criterion benchmark, NO-FMA SSE4.1/AVX2/wasm SIMD binomial filter with bit-exact scalar parity, and rayon-parallelized filter passes.
+  - **Validation & CI hardening**: raised M9 patch coverage ≥90% ([#177](https://github.com/webarkit/WebARKitLib-rs/issues/177)), Miri UB gate ([#182](https://github.com/webarkit/WebARKitLib-rs/issues/182)), strict `--all-targets --all-features` clippy gate ([#180](https://github.com/webarkit/WebARKitLib-rs/issues/180)), benchmark download hardening ([#204](https://github.com/webarkit/WebARKitLib-rs/issues/204)), and macOS/Windows build + cache fixes ([#134](https://github.com/webarkit/WebARKitLib-rs/issues/134), [#119](https://github.com/webarkit/WebARKitLib-rs/issues/119)).
 
 ### 🎯 Short-term Goals (toward v1.0.0)
 - **KPM-specific benchmark** ([deferred from #142](https://github.com/webarkit/WebARKitLib-rs/issues/142)): Add a dedicated `kpm_bench.rs` Criterion bench so we can verify the pure-Rust backend stays within 20% of C++ on `pinball-demo`. The existing `marker_bench` only measures barcode marker detection.
-- **WASM browser examples for KPM/NFT** ([#161](https://github.com/webarkit/WebARKitLib-rs/issues/161)): End-to-end runnable browser demo of the pure-Rust NFT pipeline.
-- **Raise M9 patch coverage** ([#177](https://github.com/webarkit/WebARKitLib-rs/issues/177)): Lift M9 modules from 84.76% → ≥90% patch coverage with no file under 85%.
-- **Dependency upgrades**: Criterion 0.5 → 0.8 ([#174](https://github.com/webarkit/WebARKitLib-rs/issues/174)).
+- **Live-camera WASM demo + pose parity** ([#215](https://github.com/webarkit/WebARKitLib-rs/issues/215)): Extend the browser NFT demo with a live camera feed and jsartoolkitNFT pose parity.
 - **Enhanced Documentation**: Expand API reference with complete module-level docs, integration walkthroughs for JS/TS, and detailed usage examples.
 - **WASM Memory Management**: Improve resource cleanup when switching engines or markers in long-running browser sessions.