code wiki / js
topic: js
94 modules sharing the js name family (derived from the tree's prefix discipline).
The sovereign browser's JavaScript lane: parser, evaluator with arena allocation and fuel budgeting, and the JS2 engine that took real-site fetching from hangs to working renders (a 205 KB Wikipedia page in ~half a second). The doctrine is js-last-mile-only — JS exists to make the browser and the served pages work, never as an implementation language for organs. Current known frontier: generator support, the blocker for the heaviest bundle-driven sites.
narrated overview -- maintained by the narration lane, module links verified against this wiki.
| module | description | lines | funcs |
|---|---|---|---|
| nx_js_arena_gate.nx | escape-aware ARENA RECLAMATION. Proves (A) call-heavy code that USED to exhaust | 49 | 3 |
| nx_js_arena_probe.nx | isolate the escaped-closure crash: interpreter vs JIT. expect_exit: 0 license_tier: ORIGINAL | 14 | 3 |
| nx_js_autoglobal_hoist_gate.nx | 36 | 3 | |
| nx_js_binopfuse_gate.nx | DIFFERENTIAL correctness gate for the JIT const-operand BINOP fusion peephole | 119 | 6 |
| nx_js_bom_gate.nx | 58 | 5 | |
| nx_js_bundle_probe.nx | name the NEXT construct blocking real bundles (2026-07-27). | 52 | 4 |
| nx_js_callapply_gate.nx | 32 | 3 | |
| nx_js_callcost_probe.nx | differential CALL-COST breakdown for the last >5x kernel (recursion 6.86x). | 48 | 4 |
| nx_js_compile.nx | the bytecode-VM PARITY GATE (rungs 2..5c): every chk() runs the SAME source | 213 | 1 |
| nx_js_conformance.nx | WB-JS-001 conformance HARNESS (test262-lite, no-wave | 233 | 9 |
| nx_js_consumer_swap_gate.nx | CONSUMER SWAP: route the renderer's page scripts through the VM/JIT | 162 | 7 |
| nx_js_dom_demo.nx | INDEPENDENT proof of the JS<->DOM binding (R-JS-DOM1/2/3): seed a realistic | 63 | 3 |
| nx_js_earley_diag.nx | run the earley DIAGNOSTIC bundle (head+raw+diagfoot): replays test(k) | 30 | 3 |
| nx_js_error_gate.nx | Error/TypeError/RangeError constructors + Array/Error called WITHOUT `new`, | 74 | 7 |
| nx_js_es_census.nx | MEASURE the modern-ES surface of the sovereign JS engine (present vs absent), so | 62 | 6 |
| nx_js_eval.nx | R-JS-EVAL (WB-JS-001 rung 2): the ECMAScript tree-walking | 4641 | 268 |
| nx_js_evalorder_gate.nx | 49 | 3 | |
| nx_js_event_loop_gate.nx | 71 | 5 | |
| nx_js_extends_gate.nx | 70 | 5 | |
| nx_js_f64.nx | JS Number (f64) <-> i64 / decimal-string converters for the sovereign JS engine. | 136 | 5 |
| nx_js_f64_test.nx | gate for the f64 converter library nx_js_f64.nx (R-JS-F64-0/1). | 83 | 6 |
| nx_js_f64lit_probe.nx | check 0.1 parses to the correctly-rounded IEEE bits. expect_exit: 0 license_tier: ORIGINAL | 14 | 3 |
| nx_js_feedback_gate.nx | P6 rung 1: TYPE FEEDBACK (Maglev's "runtime metadata collected during | 66 | 3 |
| nx_js_fetch_gate.nx | 63 | 5 | |
| nx_js_floatfuse_probe.nx | SELF-ASSERTING perf-regression GATE for the const-fusion float fallback. | 52 | 4 |
| nx_js_fnprop_diag.nx | pinpoint Splay's runtime gap: fn-static-property + String(). Minimal calls (avoid arena flake). expect_exit: 0 | 27 | 4 |
| nx_js_fuel_gate.nx | non-vacuous tooth for the GLOBAL JS execution budget (debt 1785032369). | 41 | 1 |
| nx_js_gc_gate.nx | prove the non-moving mark-sweep collector (conservative word-scan made EXACT by the | 78 | 5 |
| nx_js_h2h_bench.nx | HEAD-TO-HEAD: our sovereign VM/JIT vs V8 (node v22) on the SAME compute kernels, | 155 | 7 |
| nx_js_h2h_probe.nx | probe the limits the h2h benchmark hit (fib depth/call-count, array capacity). expect_exit: 0 | 24 | 4 |
| nx_js_headless_render.nx | RUNG 5 CONSUMER + API-CAPTURE: fetch a REAL site over sovereign TLS, fetch its | 121 | 5 |
| nx_js_headless_render_gate.nx | proves rung 5 (headless hydration): the ASYNC rungs (event loop / | 75 | 5 |
| nx_js_hydrate_index_gate.nx | CLOSES THE LOOP: a page whose INLINE JS builds content -> hydrate | 97 | 8 |
| nx_js_indel_gate.nx | the `in` operator + `delete` operator, V8-pinned, both tiers. `in` = property existence | 41 | 4 |
| nx_js_instanceof_gate.nx | 30 | 3 | |
| nx_js_jit_gate.nx | P5 BASELINE JIT: prove it (A) is CORRECT vs the VM and the tree-walker on the | 241 | 4 |
| nx_js_jit_keystone.nx | P5 BASELINE-JIT KEYSTONE (SOTA ladder, after P4 ICs; mirrors how the VM | 119 | 2 |
| nx_js_jitgc_gate.nx | prove JIT + GC COEXISTENCE: the P5 native JIT runs WITH GC enabled, a collection can | 49 | 4 |
| nx_js_lex.nx | R-JS-LEX (WB-JS-001 rung 0): the ECMAScript TOKENIZER, the | 653 | 30 |
| nx_js_missingarg_gate.nx | 36 | 3 | |
| nx_js_navier_diag.nx | diagnose NavierStokes runtime gap. Minimal calls (tree+VM). expect_exit: 0 license_tier: ORIGINAL | 24 | 4 |
| nx_js_objchurn_probe.nx | diagnose the object-churn gap: is it the CTOR CALL, the property SETs, or obj_new? expect_exit: 0 | 29 | 4 |
| nx_js_objproto_gate.nx | 31 | 3 | |
| nx_js_octane_crypto.nx | run the REAL Octane Crypto benchmark (unmodified Google source: jsbn RSA-1024, | 43 | 3 |
| nx_js_octane_deltablue.nx | run the REAL Octane DeltaBlue benchmark (unmodified Google source, harness | 44 | 3 |
| nx_js_octane_earley.nx | run the REAL Octane EarleyBoyer benchmark (unmodified scheme2js source, harness | 62 | 3 |
| nx_js_octane_memcensus.nx | QUANTIFY the GC debt: run each real Octane benchmark ONCE and report the | 44 | 4 |
| nx_js_octane_navier.nx | run the REAL Octane NavierStokes (unmodified, harness stubbed). Self-validates in | 51 | 3 |
| nx_js_octane_next_diag.nx | PARSE-FIRST diagnostic for the next Octane targets (DeltaBlue/RayTrace/Crypto). | 69 | 4 |
| nx_js_octane_next_run.nx | attempt to RUN the three staged next-Octane benchmarks (all 3/3 parse-clean). | 44 | 4 |
| nx_js_octane_raytrace.nx | run the REAL Octane RayTrace benchmark (unmodified, harness stubbed). Renders a | 43 | 3 |
| nx_js_octane_regexp.nx | run the REAL Octane RegExp benchmark (126KB unmodified V8 source, harness | 61 | 3 |
| nx_js_octane_regexp_ck.nx | AUTHORITATIVE Octane RegExp verification: run the real 126KB V8 regexp.js | 44 | 2 |
| nx_js_octane_richards.nx | MILESTONE: run the REAL Octane Richards benchmark (unmodified Google source, | 85 | 3 |
| nx_js_octane_splay.nx | run the REAL Octane Splay benchmark (unmodified Google source, harness stubbed) on | 56 | 3 |
| nx_js_or_diag.nx | localize the ||-SEGV: tree-walker first (prints), then VM. expect_exit: 0 license_tier: ORIGINAL | 15 | 3 |
| nx_js_parse.nx | R-JS-PARSE (WB-JS-001 rung 1): the ECMAScript PARSER, riding | 2485 | 106 |
| nx_js_pending_fetch_gate.nx | 77 | 5 | |
| nx_js_poly_ic_gate.nx | P4 POLYMORPHIC inline caches: prove (A) a K-way IC gives IDENTICAL results to | 142 | 5 |
| nx_js_probes_nojit.nx | to isolate JIT bugs from VM-core | 23 | 2 |
| nx_js_probes_run.nx | run knowledge/octane/jsprobes.js on the VM/JIT and print its result string (differential vs node). expect_exit: 0 | 26 | 2 |
| nx_js_probes_vm.nx | run knowledge/octane/jsprobes.js on the VM/JIT ONLY (tree-walker skipped: call-heavy probes SIGTERM it), | 22 | 3 |
| nx_js_promise_gate.nx | 65 | 5 | |
| nx_js_proto_gate.nx | built-in prototype EXTENSION on the tree tier: user code adds methods to | 33 | 4 |
| nx_js_proto_probe.nx | probe prototype-based OOP on tree-walker vs VM. expect_exit: 0 license_tier: ORIGINAL | 21 | 4 |
| nx_js_proto_vm_gate.nx | built-in prototype EXTENSION on the FAST tiers (VM interpreter BC_CALLM + | 48 | 5 |
| nx_js_random_probe.nx | verify Math.random(): runs, in [0,1), deterministic (seeded), matches tree-walker. expect_exit: 0 | 44 | 3 |
| nx_js_realbench.nx | CAPSTONE of the JIT arc: realistic BUNDLE-SHAPED programs (the composition of | 102 | 5 |
| nx_js_regalloc_probe.nx | validates the register-allocation eligibility DETECTOR (jit_intloop_eligible), | 63 | 4 |
| nx_js_regex_features_gate.nx | regex-engine features added during the Octane RegExp arc (2026-07-12), | 67 | 6 |
| nx_js_regex_gate.nx | 81 | 5 | |
| nx_js_render_emit.nx | read _offc/nxjs_in.html, run js_render_page (execute the page's inline scripts | 33 | 2 |
| nx_js_render_gate.nx | gate for R-JS-DOMWRITE: the sovereign JS engine now BUILDS rendered content. | 78 | 6 |
| nx_js_rich_diag.nx | probe Richards' runtime operations to find the gap. expect_exit: 0 license_tier: ORIGINAL | 25 | 4 |
| nx_js_rt_memprobe.nx | measure the pool allocation of a REDUCED RayTrace render (20x20, checkNumber assert disabled) to quantify | 22 | 3 |
| nx_js_run_demo.nx | END-TO-END COMPOSITION proof for the sovereign JS engine | 19 | 1 |
| nx_js_safepoint_cost.nx | measure the PURE per-back-edge safepoint overhead on a hot JIT loop. GC-off = no | 34 | 4 |
| nx_js_semantics_gate.nx | JS semantics pinned during the EarleyBoyer root-cause hunt (2026-07-11), | 81 | 7 |
| nx_js_shape_gate.nx | P3 SHAPES / hidden classes: prove (A) shape-keyed ICs give IDENTICAL results | 167 | 5 |
| nx_js_sota_bench.nx | JS-ENGINE SOTA BENCHMARK (operator 2026-07-07: "we want to get to state of | 259 | 7 |
| nx_js_sota_scorecard.nx | HONEST placement of the sovereign Nishi JS engine vs the 2026 SOTA engines | 137 | 5 |
| nx_js_speed_bench.nx | the BENCHMARK-FIRST baseline for the bytecode-VM arc: measure the tree-walker's | 28 | 4 |
| nx_js_splay_diag.nx | diagnose which Splay feature is missing. expect_exit: 0 license_tier: ORIGINAL | 25 | 4 |
| nx_js_spread_gate.nx | 44 | 4 | |
| nx_js_strcoerce_gate.nx | 37 | 3 | |
| nx_js_this_test.nx | isolate why method-shorthand+this isn't firing. Prints rc/kind/val per probe. | 27 | 5 |
| nx_js_unicode_probe.nx | ISOLATE the construct that blocks real bundles (2026-07-27). | 48 | 4 |
| nx_js_update_gate.nx | SINGLE-EVALUATION of the reference in ++/-- update expressions on BOTH tiers. | 68 | 7 |
| nx_js_urlfold.nx | SOVEREIGN JS constant/template URL folder (operator 2026-07-08: "get to state of the art | 258 | 11 |
| nx_js_urlfold_gate.nx | 62 | 4 | |
| nx_js_vm.nx | BYTECODE-VM rungs 2+3+4a+4b: the AST->BYTECODE COMPILER + a TAGGED-VALUE stack VM. | 4436 | 111 |
| nx_js_vm_bench.nx | KEYSTONE of the bytecode-VM arc: a minimal stack-based BYTECODE VM (tight dispatch | 92 | 6 |
| nx_js_xhr_gate.nx | 59 | 5 | |
| nx_js_ytsig_gate.nx | 50 | 5 |