Radix 0.6.0

Install this version#

No prebuilt archives were published for this version. It is listed here because its release notes are part of the record.

Release notes#

Major release spanning 313 non-merge commits (v0.5.0..v0.6.0). This release adds Python and C++23 codegen targets, a complete YAML-based cross-target test framework (~2500+ tests), a unified stdlib registry system with explicit allocator plumbing, the Fab (canonical Faber source) codegen target, and dozens of new language features including regex literals, variant matching, streams, collection filtering DSL, bitwise operators, array destructuring, spread/rest, and break/continue. The event system and WASM codegen are removed.

Scale#

Major tracks#

New codegen targets: Python and C++23, Fab (canonical Faber)#

• Python codegen target with full type system, enums, preamble infrastructure, break/continue (rumpe/perge), lista (list), tabula (dict), copia (set) methods, async bindings (figendum/variandum), nexum (reactive fields), spread/rest, and lambda handling (2931d5266, 34f33f3e8, d06d22d92, 6f8f78725, 36dcf3104, 181dbdad4, b5baa4b7c)
• C++23 codegen target with === operator support, collection stdlib, and auto-merge constructor (f8e65b354, cfe0f07fe)
• Fab codegen target for emitting canonical Faber source, with full test expectations across all test cases (1dd33f8a2, 2247115b2)
• TypeScript codegen marked complete — all generated targets reach parity (cf49a58c9)
• Python test coverage milestones: 85% (125 tests), then 95% with async bindings and nexum (1191ce716, d08175e06, 5aa1de505)
• Modular codegen architecture: extracted per-node handler files into generator classes for TS, Zig, C++, Rust (ac7f52e09, 89bff39f2, 36189a28f, e538186e2)
• Added fabrica/ demo projects for TS, Zig, C++ and fabrica/ISSUES.md documenting codegen limitations (61572652d, 0ed270b7c)

Stdlib refactor: unified registries with explicit allocators#

Three-phase refactor replacing per-target ad-hoc collection implementations with unified registries:

1. Phase 1: Unified Lista registry with explicit allocator plumbing (a1fff7fbd)
2. Phase 2a/2b: Unified Tabula and Copia registries (ff2d79324, ef52d8187)
3. Phase 3: Wire all targets to the unified registries — preamble generation uses external files and RequiredFeatures (2e0605dd3, a48767ed2)

Supporting changes:

• Preamble rework design document (d76a63ff4)
• Zig native runtime architecture proposal (1ca9ec9fd)
• External build.zig and mod.zig files (85a59a0a5)
• aleator (random) module for Zig, Rust, C++ (15d8ee381)
• tempus (time) module for Python, C++, Rust, Zig (66e6ece9e, c80a2487d)
• mathesis (math) module for Rust, Zig, C++ (2b29f0426, 87bf924cd)
• I/O intrinsics for all targets (96d8c5a47)

Entry points: initium → incipit, incipiet (async entry)#

• Renamed initium to main across all codegen targets, exempla, parser, AST, and lexicon (bd221ca2b, 58608f896)
• Added incipiet (async entry point) with per-target codegen and ergo chaining form (e79e5a126)
• Added initium entry point statement (pre-rename) (5e57ce509)
• Tests for incipit and incipiet (ef6e2968e)
• Incipit facade examples for all codegen targets (dc626d1f3)

Resource management (cura/curator)#

• Redesigned cura syntax with explicit curator kinds: arena and page (eb4a5ac6f, 0a739faf1)
• Curator auto-injection for Zig target: semantic analyzer propagates arena allocator through call chains (7e8b824e3, f5a367c95)
• Require explicit cura blocks for Zig allocating operations (d25da9092)
• Arena/page allocator tests and semantic analyzer fixes (e7cb39dbd)
• Design docs: allocators, cura/curator/curatum roles (92874360b, 417c14eb9, 7ffb12c65, 67d2ffc4f)
• Updated exempla to use incipit ergo cura arena for Zig compatibility (506431554)

Language features#

Regex literals: Implemented sed /pattern/flags syntax with per-target codegen support and design doc (8092193b2, 921f43750, ea2048316, 10ca81cf2). Removed g flag from regex design (854d08a05).

Database DSL: Renamed tabularium to arca for database DSL (505e032a3)

Variant matching: match statement for variant matching, paired with discretio for tagged union patterns (d4c820d98, ee0ecbdb9)

Streams (flumina): Three-phase flumina protocol — fiunt (multi-value streams), fiet/fient (async streams), TS codegen for fit functions, streams-first design doc (a6167aec1, a0eb16e4a, a99915709, f3ac0f68e)

Dispatch (ad): ad statement for dispatch with design doc and per-target codegen (584cc71b9, 7ffb12c65)

Collection filtering DSL: ab expression for collection filtering with design doc (b16d5a794, a46303860)

Sequence operators:

• Compound assignment (+=, -=, *=, /=, &=, |=) (809e5a829)
• Bitwise operators (&, |, ^, ~, <<, >>) (606021a55)
• ante/usque range operators for explicit exclusive/inclusive semantics (9a6f1efce)
• Change range expressions to exclusive-end semantics (17528cdd0)
• Negative indices and slicing support for all targets (bea7c1799)
• Array destructuring support for all targets (d1447c146)
• sparge (spread) and ceteri (rest) operators (0fa8b9b40)
• vel nullish coalescing operator and CLI stdin support (f41b74a46)
• ?. optional chaining and !. non-null assertion (3d7eb61ff, 910d694a7)
• Hex literal support (0xFF, 0xFFn) (00d40b94a)
• : shorthand for lambda expressions (pro x: expr) (027408abf)
• Lambda return type annotation syntax (pro x -> Type: expr) (df4d3ac4e)
• Async lambda syntax: fiet x: expr, fit x: expr (84289ddb5)
• Default parameter values with vel (83b36e535)
• Dual parameter naming with as alias (efcc1e99e)
• Separate arrow and verb syntax for function return types (373f4aed7)
• break/continue (break/continue) for all targets, including Zig codegen (da17263ac, 806f44d1e)

Compile-time evaluation: prae compile-time evaluation (prae typus, praefixum) with design doc (b772b0898, fc4fde49e). Graceful degradation for TS/Python (3374ff7ed).

Type system#

• unio<A, B> union types for TS, Python, and Zig (7c43c4217, af02493f9)
• discretio (tagged unions) with pattern matching (6ad3e15fc)
• magnus (bigint) type for TS and Python (13f55efca, 6208cca00)
• ignotum (unknown) type mapping for TS and Python (2fd03da4e)
• numquam (never) type mapping for TS and Python (6c2f3eba1)
• is type checking operator, typus RHS typeof, null/nonnihil operators (660cd4b66)
• as (type cast) operator for type assertions (a13efa238)
• supra() as Latin equivalent of super() (c76140dfb)
• Design docs: new type features (cd2570cb6), aperit keyword for index signatures (5255afeba)

Preposition system (praepositiones)#

• Unified Latin preposition system with praepositiones.md design doc (e89cebe47)
• Refined roles: ad for dispatch, pro for binding; removed ad from params (425d84f5d)
• qua replaces as for type assertions, as for aliasing (d1a151b8f, 33a4fdffb, 95907fdd1, b3597aede)
• ref/mut ownership prepositions for Rust and Zig codegen (670a069ed, 82f2738dd)
• Removed cum keyword entirely; mut for mutation blocks, ref for for-in loops (84be77479, 05a46950a)
• Ban cum from the language (33a4fdffb)
• per replaces ad for property shorthand in collections DSL (dabdf8297)
• Wildcard import alias support (5b29247cc)

Semantic analysis#

• Validation for cede/figendum/variandum context (a8543d18b)
• Semantic analysis for enum, class, and pactum declarations (10c36c02c)
• Fix semantic analyzer issues and exempla variable collision (d8676f0bd)
• scriptum() for cross-target string formatting (f796021fb)
• @target filter syntax and implementation stats (adffc28fc)

Removals#

Design docs and consilia reorganization#

• Reorganized consilia/: moved 10 completed docs to consilia/completa/, 3 planned docs to consilia/futura/, merged fasciculus.md into solum.md (7cbef99ae)
• Moved stdlib designs to norma/ subdirectory (bf4a4277c)
• Moved and compressed flumina.md to main consilia directory (9f7243319)
• Archived eventus.md design doc (2398e79c7)
• Added design docs: Zig async via Responsum state machines (37e27a5f5), comment preservation (73189f77d), Faber codegen target (579d8eac1), TS-to-Faber transpiler (d06be1cf8, 4075e9b26), two-pass compilation (565f112a2), array destructuring (00ebefc41), operators (4aa09c071), test syntax (ae03ba520)
• Nucleus design: micro-kernel runtime for unified async/I/O (2eec2c5e1), Monk OS analysis (0040e3931), Responsum protocol (707d1ea2e)
• Updated binario.md final design decisions (1eb9d655e)
• Rewrote operators.md with complete reference (a6c158417)
• Updated consilia/clausura.md with target-by-target implementation status (aa82c1e78)

Test infrastructure (proba/)#

• YAML-based cross-target codegen tests: shared format running same cases against multiple targets, 250+ tests across 8 domains (557f97e9c)
• Complete YAML test migration with per-target expectations for TS, Zig, Python, C++ (f21f8072b, e38b36e0a, 7d21cd9e6, 417c83dc7, 0275afdb3)
• Comments on every YAML test describing each case (e0b660c11)
• Recursive test structure with modern format support (300d4c32a)
• Expression test files in modern format (c901c2c6e)
• Comprehensive statement codegen tests — 816 tests (617bf6de4)
• 200+ edge case tests across all codegen test files (502b9bb28)
• Errata test support for expected compilation errors (a08647b5e)
• Coverage reports and strict mode (48ff2b4cf, 113f9aa00)
• Rust test expectations and validation pipeline (e30570bd1)
• Zig test expectations and dead code removal (5d4d4fa3c)
• Move codegen tests to proba/ at project root (77cf944d0)
• proba/README.md documenting test framework (8a2355996)
• Updated proba README with norma/ directory structure (abd0e245f)
• Moved stdlib tests to proba/norma/ (2863e6918)
• Expect-failure tests for invalid syntax patterns (a355ba852)
• TS codegen tests split into modular domain files (15d9bda69)
• Zig expectations added to cura tests (6ea7e0a40)

Zig codegen#

• Error handling: iace → error unions, mori → panic (5acd4a68d)
• Collection methods (lista, tabula, copia) (b16b28142, 955130e1d)
• For-range native syntax (8fb7a169d)
• Remove auto-generated main() and m_ prefix (300b75283)
• Preamble emits imports instead of inlining stdlib (44fba8526)
• Emit compile error for objectum/anytype return types (df4d3ac4e)
• String switch detection by checking case literals (5beb90408)
• If-else chains instead of switch (1f27b2aca)
• Division/modulo operators and exempla script fix (a062f3feb)
• Unused alloc warning and stepped range redeclaration fix (cd6e6397d)
• Array literals: [_]T{} when element type is known (dd543633c)
• All 44 exempla passing (100%) (6b1e7e495)
• Refactored expressions to guard clauses with tests (e71c24572)
• Module constants prefix with m_ to avoid parameter shadowing (72c48eb54)
• Map === to == and !== to != (43152756e)
• Zig codegen failure categories documented in consilia/codegen/zig.md (d125a9c38)

Rust codegen#

• Full collection methods: lista (Vec), tabula (HashMap), copia (HashSet) (9bb34cc9f, 80e12df45, 0eb491aab, 0bdef6f2b, 70f250a84)
• ref/mut preposition handling for ownership (670a069ed)
• Lambda return types and numerus type mapping fix (a57a98e8d)
• Main wrapper, println format, test updates (dbf88f526)
• Fix Rust lista RsGenerator type signature (5999e79ed)

CLI and tooling#

• Default to stdin with flexible option order (10e24783d)
• faber check -t zig target validation (f6e976734)
• verify:exempla script for quick pass/fail per target (e709f6d48)
• --force flag to continue despite test failures (9fc897b06)
• Unbundled CLI for exempla script (59ae3328a)
• prettier for formatting (20432d549, 502b9bb28)

Rivus#

• Added rivus/ directory as a Faber language compiler with lexicon files: nomina, typi, typi_constructi, verba, verba_clavium (1479 lines) (1e460c3dc)

Agent and documentation#

• AGENTS.md symlink (300a44a2c)
• Rewrote CLAUDE.md for accuracy and conciseness, trimmed from 235 to 163 lines, added Latin phrase guidance and agent delegation guidance (09958958e, 2cb19d9e8, 8b9203714, e6b3b3ef9)
• Added pragmatic simplicity to project philosophy (6fd506610)

Exempla restructuring#

• Relocated ex-pro to from (ec8f8b63d)
• Added exempla/expressions/ with buildable program files (a9f425c5f)
• Added exempla/statements/ with comprehensive buildable files for all statements (4f62a5f86, 496f17719, 06a57d1f5)
• Added exempla/errores/ for 1:1 mapping with grammatica/ (af46095c1)
• Added grammatica/ directory and GRAMMAR.md for language reference (8edf0ec60)
• Cleaned up exempla/ files and added ego.fab (1d40b3fdb)
• Fixed exempla compilation errors (74c80073a)

Grammar updates#

• Updated GRAMMAR.md and grammar extraction (efd059b1b, 12248482e, 8463039eb, 89bf930fe, 96f582057, 8bc31e952, e7b45cca1, 2362e59ac, c0b8fd46e)
• Fix grammar extraction to preserve whitespace correctly (f8865a523)
• Renamed AST node types to Latin-facing names (91b28aeca)
• Renamed parser functions to match Latin keywords (9897368c7)
• Added target mappings to AST node documentation (90f5fd7e4)

README documentation#

• Added implementation completion summary table (58ee671d4)
• Fixed feature coverage metric: use feature coverage instead of exempla pass rate (9626d173c)
• Updated README implementation status with comprehensive corrections and additions (dcc5e5d72)
• Multiple progress updates: Python milestones, Rust stdlib, C++ mathesis (48f5e8d9c, 18b785355)

Other changes#

• Make keywords and types case-sensitive (2690e5310)
• Document indexed iteration syntax and per property shorthand (8111ec987)
• Fix 73 failing tests across tokenizer, parser, and codegen (0e0ffc039)
• Fix keyword collision in member expressions (09e1d6208)
• Fix TS strict mode errors in source files (302f704d6, 639197134, e0d4ecd64)
• Fix parser infinite loop on invalid syntax (b5baa4b7c)
• Fix custom type parsing (5b29247cc)
• Fix method handler architecture to preserve argument boundaries (2dc8abdc7)
• Use Latin keywords in parser error enum names and messages (33a4fdffb)
• Comment pass-through in codegen (97502682a)
• Arrow binding for direct codegen escape hatch (d74a23a80)
• Document borrowed return types with from source specifier (41c68dd82)
• Reframe project as LLM intermediate language (a334deae4)

Breaking / author-visible#

---

All releases · Install the current release