{"slug": "natural-programming-just-dropped", "title": "Natural Programming Just Dropped", "summary": "A new programming language called llang has been released, designed to be written by AI and read as structured, explicit English that behaves like Python. llang ships with an interpreter requiring only Python 3.10+, a WebAssembly backend for the integer/function subset verified by executing its output in Node, a self-hosted lexer written in llang and checked token-for-token against the bootstrap lexer, and a porting tool that converts Python, C and JavaScript repositories to llang with coverage reports and a JSONL cleanup queue. The language trades terse human-era syntax for labeled arguments, explicit block terminators and phrase-named builtins so reviewers can verify AI-written code, and includes native domain libraries for math, datascience (NumPy/pandas semantics) and machinelearning (scikit-learn semantics).", "body_md": "A programming language that reads like structured English and behaves like Python.\n\n**llang is built to be written by AI.** It is deliberately highly interpretable\nand direct: every construct announces itself, every block has an explicit\nterminator, every argument is labeled, and nothing hidden or implicit can slip\npast a reviewer. The goal is not clever code — it is code you can read once and\nunderstand, so humans can review what an AI wrote with confidence.\n\nHuman-era compromises (terse syntax, abbreviations, operator soup, clever implicit behavior) are pure downside for a model, so llang trades them for verbose, explicit, regular constructs. Natural-language phrasing is the point, not a novelty — llang is structured pseudocode you can actually run.\n\n```\nset scores to a map with \"Ada\" mapped to 91, \"Grace\" mapped to 98\n\nrepeat for each student in (call keys of with collection = scores)\n    show student plus \": \" plus (call text of with value = item student of scores)\nend repeat\n```\n\nThe language is implemented and usable. The primary execution path is the interpreter; a WebAssembly backend exists for the integer/function subset and is verified by executing its output in Node. For rewrites, \"compiles\" means the program parses, lints clean, and produces output identical to the original.\n\n- **Values:** numbers, text, booleans,`nothing` , lists, maps, functions\n- **Statements:**`set` /`let` ,`show` /`display` /`print` ,`require` ,`add` ,`set item` ,`if` /`otherwise` ,`repeat while` ,`repeat for each` ,`stop the loop` ,`skip to the next item` , functions with aliases, labeled`call` ,`fail` /`expect` problem handling,`use` modules\n- **Expressions:** strictly left-to-right with parentheses for grouping, word operators with closed synonyms (`floor divided by` included), postfix predicates, collection literals, indexing, phrase-named builtins,`apply` for calling function values\n- **AI ergonomics:** definitions may use phrase names (`define function find all ...` ), and`@` escapes reserved words anywhere a name is expected (`taking @end` ,`call f with @end = 1` )\n- **Domain libraries:**`advanced_math` (math/statistics),`datascience` (NumPy/pandas semantics),`machinelearning` (scikit-learn semantics) — all native llang, with self-describing results and loud validation\n- **Libraries:** native llang modules in`lib/` (`math` ,`text` ,`lists` ,`shell` ,`shlex` ,`fnmatch` ,`textwrap` ,`csv` ,`json` ,`re` ,`ast_contract` ,`pylev` ,`humanize` ), plus an explicitly labeled opt-in foreign bridge (`json` ,`math` ,`shell` )\n- **Objects:** minimal classes with explicit`this` , single inheritance-free construction via`call new ...` , and`ask instance to method ...`\n- **Compiler:** a WebAssembly backend for the integer/function subset —`llang-compile file.ll -o file.wasm` , verified by executing the output in Node\n- **Self-hosting:** the language's own lexer is written in llang (`lib/llang_lexer.ll` ) and verified token-for-token against the bootstrap lexer\n- **Universal porting app:** point it at a repository and it converts supported source files (Python, C, JavaScript) to readable llang, preserving structure, with coverage reports, verification, and a JSONL cleanup queue for AI-assisted completion —`llang-port-repo <repo> -o <out>`\n- **Tooling:** source-located diagnostics, canonical formatter, pre-run linter, LSP language server, Python-to-llang converter, property-based differential fuzzer, single-command verification, GitHub Actions CI\n- **Ecosystem:** native`json` module, JSON package manifests with a registry, editor grammars for VS Code and tree-sitter, and a browser playground\n\nThe interpreter needs Python 3.10+ only. The full verification suite additionally\nuses Node.js (for WebAssembly checks) and the Python data libraries used as\noracles: `numpy`, `pandas`, `scikit-learn`. Install them with\n`pip install -e '.[dev]'` before running `tools/check.py`.\n\n```\npython run.py examples/hello.ll          # run a program\npython repl.py                           # interactive shell\npython tools/check.py                    # every check: tests, lint, rewrites, links\npython tools/port.py my.py -o ported.ll --report report.txt\npython run.py --lint examples/hello.ll\npython run.py --format examples/hello.ll\nnode tools/playground/server.js           # browser playground at localhost:8080\npython -m llang.portapp.cli <repo> -o <out>  # port a whole repository to llang\n```\n\nOr install the package for console commands:\n\n```\npip install -e .\nllang examples/hello.ll\nllang-port my.py -o ported.ll\nllang-compile my.ll -o my.wasm\npython -m llang.lsp            # language server over stdio\n```\n\n**Surface flexibility, canonical precision.** The surface language has a closed vocabulary of synonyms; the parser normalizes every spelling to one canonical AST node, and the interpreter only ever sees that node. Synonyms add readability without adding ambiguity — the failure mode that sank AppleScript and Inform 7.\n\n| Decision | Rule | \n|---|---|\n| Semantics | Grounded in authoritative docs of existing languages — re-express, never invent | \n| Blocks | Explicit `end if` /`end repeat` /`end function` /`end expect` ; indentation ignored | \n| Evaluation | Strictly left-to-right; parentheses for grouping | \n| Types | Dynamic; strict operations, no implicit coercion | \n| Arguments | Labeled roles ( `call f with name = x` ), never bare position | \n| Extension | Libraries extend values and types, never syntax | \n| Diagnostics | Source line/column and expected-token hints are part of the spec | \n\nStart here:\n\n- [Tutorial](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/TUTORIAL.md) — learn the language from zero\n- [Reference](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/REFERENCE.md) — complete authoritative reference\n- [Grammar](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/GRAMMAR.md) — compact grammar that fits in an LLM prompt\n- [Examples](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/examples) — runnable programs with pinned output\n\nDesign and process:\n\n- [Design](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/DESIGN.md) — decisions and rationale\n- [Semantic decisions](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/SEMANTIC-DECISIONS.md) — tricky choices and why\n- [Method](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/METHOD.md) — the repeatable process for adding a language surface\n- [Family plan](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/FAMILY-PLAN.md) — tiered IR, converter, libraries, build order\n- [Roadmap](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/ROADMAP.md) — honest scope and deferred features\n- [Rewrites](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/REWRITES.md) — open-source Python software rewritten and verified in llang\n- [Research](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/docs/research) — grounded matrices: Python semantics, precedents, stdlib, demo, terminal, test plan\n\nProject:\n\n- [Changelog](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/CHANGELOG.md)\n- [Contributing](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/CONTRIBUTING.md)\n- [License](https://github.com/fraliv50/llang-natural-language-programming-/blob/main/LICENSE) — MIT\n\n- `examples/report_card.ll` — flagship program: a graded report card that checks its own result\n- `examples/rubiks.ll` — Rubik's cube simulator that writes a browser viewer (`rubiks.html` )\n- `lib/shlex.ll` ,`lib/fnmatch.ll` ,`lib/textwrap.ll` ,`lib/csv.ll` ,`lib/json.ll` ,`lib/re.ll` — CPython modules rewritten in llang and verified against the stdlib\n- `lib/pylev.ll` ,`lib/humanize.ll` ,`lib/box2d.ll` — external open-source repos cloned and rewritten, verified against the originals (1,562 cases);`box2d` is a headless physics engine subset with a working 60-step simulation\n- `examples/box2d_visual.ll` — writes`physics.html` , a live browser animation of that simulation (all frames computed in llang)\n- `lib/llang_lexer.ll` — llang's own tokenizer, written in llang (self-hosting stage 2)\n- `lib/advanced_math.ll` ,`lib/datascience.ll` ,`lib/machinelearning.ll` — domain libraries grounded in Python math/statistics, NumPy/pandas and scikit-learn (330 cases)\n- **2,546 differential cases total, 0 mismatches**\n- `src/llang/compiler.py` — WebAssembly backend;`tools/playground/` — browser playground\n- `src/llang/lsp.py` — language server;`src/llang/packages.py` — package manifests and registry\n- `editors/` — VS Code and tree-sitter grammars;`docs/SITE.md` — documentation site guide\n- `tools/port.py` — automated Python → llang converter with an honest coverage report\n- `src/llang/portapp/` — universal repository porting app; see`docs/PORTING-APP.md`\n- `tools/cleanup_prompt.py` — turns a cleanup-queue item into a ready-to-paste AI prompt\n- `tools/check.py` — one command for tests, lint, rewrite verification, and link checks\n\n```\nsrc/llang/    lexer, parser, canonical AST, interpreter, synonyms, errors,\n              formatter, linter, converter, WASM compiler, LSP, packages,\n              CLI entry points, foreign allowlist\nlib/          standard library written in llang\nexamples/     runnable programs\ntests/        unit, differential, diagnostics, end-to-end, compiler\ndocs/         reference and design documentation\ndocs/research/ grounding documents from authoritative external sources\ntools/        port.py, check.py, compare_*.py, fuzz_semantics.py, playground,\n              fixtures, pilots\neditors/      VS Code TextMate grammar and tree-sitter grammar\n.github/      CI workflow and issue templates\nrun.py        development CLI entry point\nrepl.py       interactive shell\n```\n\n", "url": "https://wpnews.pro/news/natural-programming-just-dropped", "canonical_source": "https://github.com/fraliv50/llang-natural-language-programming-", "published_at": "2026-11-22 21:37:49+00:00", "updated_at": "2026-11-22 22:14:48.237887+00:00", "lang": "en", "topics": ["ai-tools", "developer-tools", "ai-agents", "large-language-models"], "entities": ["llang", "Python", "WebAssembly", "Node.js", "NumPy", "pandas", "scikit-learn", "GitHub Actions"], "also_reported_by": [], "takeaway": "llang's explicit, labeled, terminator-delimited syntax is intended to make AI-generated code reviewable, and its porting tool converts Python, C and JavaScript repositories to llang with coverage reports and a JSONL cleanup queue for AI-assisted completion.", "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/natural-programming-just-dropped", "markdown": "https://wpnews.pro/news/natural-programming-just-dropped.md", "text": "https://wpnews.pro/news/natural-programming-just-dropped.txt", "jsonld": "https://wpnews.pro/news/natural-programming-just-dropped.jsonld"}}