The inside story of building the largest pure-Rust sovereignty stack
159 posts
OxiBLAS 0.2.0 is a major step up: cache-oblivious recursive and parallel factorizations, batched BLAS, runtime auto-tuning, multifrontal sparse solvers, mixed-precision refinement, NUMA-aware allocation, and no_std support — with the Fortran FFI retired in favor of a fully pure Rust workspace.
A minimal DSL + IR that compiles logical rules (predicates, quantifiers, implications) into optimized einsum graphs. Full neurosymbolic integration with differentiable fuzzy/probabilistic semantics — now available as 0.1.0-rc.1.
OxiLean 0.1.0 is the debut of a memory-safe Interactive Theorem Prover written entirely in Rust, inspired by Lean 4 — a zero-dependency kernel implementing the full Calculus of Inductive Constructions, a complete tactic framework, and a WASM REPL that runs in the browser.
SciRS2 is a pure-Rust SciPy/NumPy/scikit-learn replacement, and 0.3.0 is the biggest release yet — transformers, GNNs, diffusion, MoE/RLHF, Gaussian processes, MCMC, survival analysis, radar/compressed sensing, LOBPCG/AMG, plus new Julia and Python bindings. 19,644 tests, ~2.59M lines of Rust. No C, no Fortran.
XSL-FO 1.1 to PDF (and SVG/text) pipeline in pure Rust. 10–1200x faster than Java FOP, zero JVM, full Knuth-Plass line breaking, 294 properties supported. Production-grade document generation is now memory-safe and sovereign.
A complete distributed mail server in pure Rust. Full SMTP/IMAP/JMAP/POP3 support, Mailet-based processing pipeline (Apache JAMES port), AI integration via OxiFY, and AmateRS-backed storage. 10–50 MB memory footprint, no JVM, no GC pauses.
From a single monocular video → consistent 3D Gaussian avatar in pure Rust. Full differentiable Gaussian Splatting (wgpu) + Multi-View Diffusion (Candle) + FLAME binding. 796 passing tests, zero C/Fortran, production-ready digital twin reconstruction.
The first public release of OxiGDAL: a 100% Pure Rust reimplementation of GDAL with zero C/C++/Fortran in default features. 11 format drivers (GeoTIFF/COG, GeoParquet, Zarr, NetCDF, HDF5, GRIB, JPEG2000, ...), a Pure Rust PROJ with 211+ EPSG definitions, SIMD raster/vector algorithms, cloud-native async I/O, and WASM + Python bindings — across 68 workspace crates.
Drop-in PyTorch replacement in pure Rust. Full SciRS2 integration (18 crates), SIMD CPU backend, autograd, and native sharding support. 2—3× faster inference, 50% less memory, single-binary deployment — no Python, no CUDA required.
SciRS2 is the pure-Rust SciPy/NumPy replacement. 0.2.0 restores the entire workspace to zero compile errors, makes Pure Rust OxiFFT the default FFT backend (rustfft now optional), rebuilds the neural stack, and ships the first WebAssembly bindings. ~11,400 tests, zero warnings. No C. No Fortran.
OxiArc 0.2.1 brings security and scale: traditional ZipCrypto encryption for password-protected ZIPs, plus async I/O, SIMD-accelerated CRC, and memory-mapped I/O in the core. Also LZ4 dictionary support and LZHUF streaming. Pure Rust archive and compression — no C, no zlib, no libarchive.
Pure-Rust SciPy/NumPy replacement. 0.1.5 is a polish release — workspace version alignment, strict scirs2-core import policy, correct GPU feature wiring, and a leaner dependency tree. OxiFFT stays the Pure Rust default. No C, no Fortran, no system dependencies.
Pure-Rust SciPy/NumPy replacement. This release: SIMD phase 60-69 (8 new AVX2/NEON modules), modular Bowyer-Watson Delaunay (2D/3D/ND + constrained), OxiFFT now the 100% Pure Rust FFT default, and autograd optimizer fixes. 11,400+ tests, 27 crates. No C, no Fortran.
OxiArc 0.2.0 adds full LZW compression for TIFF and GIF — MSB/LSB bitstreams, variable bit widths — bringing the workspace to 10 pure-Rust crates. Now on crates.io, with a default test suite cut 76% faster (137s → 32s) and 427 passing tests. Pure Rust archive and compression with no C, no zlib, no libarchive.
A complete, high-performance Satisfiability Modulo Theories solver written entirely in Rust. Full CDCL(T) architecture, comprehensive theory support, proof generation, and Z3-level performance — with zero C/C++ dependencies and full memory safety.
Pure-Rust legal statute engine. 0.1.4 adds 5 new jurisdictions (Korea, Mexico, Malaysia, Russia, Saudi Arabia) for 23 total, 6 production examples proving one generic engine handles Civil/Common/Supranational law, real-time self-healing verification on the OxiZ SMT solver, and auto-generated TypeScript/Python SDKs. 14,705 tests passing.
SciRS2 0.1.3 expands Python (PyO3) bindings across autograd, datasets, graph, io, metrics, ndimage, neural, sparse, text, transform and vision; adds PCHIP linear extrapolation; better manylinux wheels and an Adam scalar fix. 100% Pure Rust on OxiBLAS + OxiFFT.
A polish patch for QuantRS2, the pure-Rust quantum computing framework — Python-binding bugfixes, refreshed docs, dependency-compatibility bumps, and the removal of legacy C++ SymEngine tooling now that symbolic math is 100% Rust.
Legalis-RS 0.1.3 doubles jurisdiction coverage from 8 to 18 — adding the UAE, Australia, Brazil, Canada, China, India, Indonesia, Laos, Thailand, Vietnam and South Africa — across 41 crates, 863,282 lines and 13,083 passing tests.
OxiRS 0.1.0 is the first stable release of a Rust-native Semantic Web platform: complete SPARQL 1.1/1.2, a Fuseki-compatible HTTP server, GraphQL over RDF, SHACL validation, and industrial IoT bridges — a JVM-free alternative to Apache Jena + Fuseki.
OxiZ 0.1.2 adds PyO3 Python bindings (oxiz-py), a Pure-Rust BLAS + multi-precision (MPFR-style) math layer, an SMT-COMP benchmark suite, and a vastly expanded CLI (dashboard/server/distributed/TPTP/interpolation). A Pure Rust Z3 replacement with zero C/C++.
QuantRS2 is a pure-Rust quantum computing framework — type-safe circuits, state-vector/stabilizer/tensor-network simulators, Grover/QFT/VQE/QAOA, and IBM/Azure/AWS/D-Wave hardware integration. This is the first stable release.
The first patch for QuantRS2, the pure-Rust quantum computing framework. Fixes conditional compilation and feature gating for cloud backends (IBM, Azure, AWS) and photonic modules so the modular workspace builds cleanly with any feature set.
Kizzasi 0.1.0 is the first release of an Autoregressive General-Purpose Signal Predictor (AGSP) in Pure Rust — neuro-symbolic prediction for audio, sensors, robotics, and video that fuses State Space Models (Mamba/RWKV/S4) with TensorLogic constraints.