Repository navigation
T27 OS tools: registries, tri golden (own git), skills (Closes #11358) - #11359
Merged
Merged
Conversation
…oject summary a BENCH tab draws (Closes #11264) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ings, every agent widget group and course listed, scrolling Closes #11260 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ts, last activity, 7-day pulse, status lamp (Closes #11275) Hooks could not run here (core.hooksPath=/dev/null); gates are t27b test and t27c test-report, both 0 failures. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…s: every gallery widget assigned by domain (Closes #11275) Also assigns the 31 HUD views/tri screens and 34 pages of app.t27.ai/game to a letter. File is specs/t27os/agent_parts.t27 (module T27osAgentParts). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ule lives, build and run, traps, plan (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…on C; what each console needs (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…field (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…, so domain_pulse sees the project (Closes #11275) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…us feeders, mandatory, refused by the widget factory without them (Closes #11275) Hooks could not run (core.hooksPath=/dev/null). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…, tri commands, MCP tools, skills (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…sources and gaps (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…t with its agents, provenance of every number, GOD MODE evolves inside the same laws (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ority, guardrails, one lane per task (Closes #11275) Hooks could not run (core.hooksPath=/dev/null). queen.t27 gains two Queen controls (she started a lane, she asked the owner) for the RECENT log. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ckpit of nested widgets (dash.t27) with pinned agents, live widget pages and course maps (hooks skipped: no t27c binary here) (Closes #11275) - specs/t27os/dash.t27: widget tree (SCREEN > HOME > HERO/GAUGE, CONSOLE > INSTRUMENT, COURSE_MAP > LESSON), one packer for every level, hit tests, sidebar of letters by home, gallery.t27 parse, card cache paths, curl argv, live page rect, agents_of_widget stub (k > 0 none) - specs/t27os/agent_parts.t27 taken as-is from 7aa0724 (letter per widget, app surfaces) - glass_window: draw_home is the letter cockpit; card.png via NSImage/CGImage cache, background curl fetch, one WKWebView for the live page, CHAT / CONSOLE switch on petal C - homes.t27: petals are letters (HOME_PETAL gone); petal_menu WORLD_NAMES = letter + tool name; job_kind test re-pinned - pty_sys: sys_spawn (posix_spawn) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
# Conflicts: # specs/t27os/agent_parts.t27
…status + homes dash) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> (Closes #11275)
…0% of widgets; one tool per letter, t27k kanban, the work loop (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…uded; black in full screen (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…s are the titles, t27k KANBAN board of widgets (Closes #11275) TOOL_NAMES/PAGES/ARTIFACTS + tool_of_letter in agent_parts (each artifact exists); petal_menu WORLD_NAMES = the 27 titles, held equal by a test; kanban.t27 = BOARD > COLUMN > CARD > CHIP/BADGE, column rule, drag rules, WIP 3, dash packer layout. Hooks could not run here: committed with --no-verify. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…hrough the TRI CLAW App; sign-in starts the game (Closes #11275) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…olden head, one GitHub push per batch; skill own-git-landing (Closes #11275) Hooks could not run in this sandbox (--no-verify). t27b 28/28, test-report 0 failed 0 vacuous, reseal saved. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…h an empty chain; BENCH cards, busdev read from the scan, DETECT/FLASH/UART commands (Closes #11264) Hooks could not run; committed with --no-verify. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…top-10 gaps with why-not, lamp refresh rule, self-improvement meter, safety strip with STOP, nested layout (Closes #11275) Hooks could not run in this sandbox (--no-verify). t27b 13/13, test-report 0 failed 0 vacuous, seal saved. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…s data (actors.t27 used), resource-aware agent_cap (8 GB gives 3), shedding newest lane first, bounded mailboxes, kernel view layout; evolve takes the cap as an input (why_not_cap) Hooks could not run in this sandbox (--no-verify). t27b 10/10, test-report 0 failed 0 vacuous, seals saved. (Closes #11275) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ding, a thin host over golden_chain.t27; gc_land inlines gc_since so the generated Rust compiles (hooks cannot run here: --no-verify) (Closes #11275) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
… z.ai and OpenRouter, key-free config by env/dir, not-installed state), OpenRouter free ids as measured (lightning dropped, super-vl token floor 64), traffic light per pane (green paid, yellow free, red no balance) in the listing (Closes #11275, Closes #11314) Hooks cannot run in this worktree, so this commit uses --no-verify. Seals of keys, pty_logic and their importers are NOT re-saved: the local t27c is 0.5.2 and stale. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…th/count/live limits, global cap, waiting queue), splits a Boundary into disjoint path sets with sub-issue skeletons, gathers RESULT messages with a bounded-retry join landed by gc_land, and the lane-tree + inspector view model (hooks cannot run here: --no-verify) (Closes #11275) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…tterns (Closes #11275) Hooks could not run in the authoring sandbox. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…pecs/t27os/swarm_pulse.t27 (Closes #11348) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gHashTag
enabled auto-merge (squash)
October 20, 2026 12:17
![tri-claw[bot]](https://avatars.githubusercontent.com/in/5166765?s=60&v=4){kind=small}
| assert agent_domain(0, b[0..64]) == 0; | ||
| assert agent_domain(27, b[0..64]) == 18; | ||
| } | ||
|
|
Contributor
This was referenced Oct 20, 2026
…er master, follow master after a squash merge (Closes #11361) (#11379) Blocker 1: the SEAL stage saves a seal that does not match again (t27c seal --save, then --verify) once the tests are clean. Only .trinity/seals/ may change (git add -A, git diff --cached --name-only). The link lands as REGEN, on a commit of its tip. A failing test or a spec that does not compile still refuses the link. Links whose parent is not the head are CARRIED by merge-tree; a merge-tree conflict is CONFLICT and the pass goes on with the next link. Blocker 2: gc_rebase resets refs/golden/head and refs/golden/pushed to origin/master when nothing is unpushed and master lost the head, or carries the unpushed links onto master as one commit (merge-tree with the pushed mark as the base). Both moves are update-ref with the old value. The host only runs the commands the spec names (no hand lines in cli/tri). Measured on 16 open PRs with tri golden land --plan: 0 of 16 landed before, 12 of 16 after. Co-authored-by: Claude <claude@anthropic.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
| assert agent_domain(0, b[0..64]) == 0; | ||
| assert agent_domain(27, b[0..64]) == 18; | ||
| } | ||
|
|
Contributor
PR DashboardGenerated at: 2026-10-20 12:57:41 UTC
Summary
Seal Status
|
…loses #11358) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
gHashTag
added
the
owner-approved-foreign
Contributor
PR DashboardGenerated at: 2026-10-20 13:21:33 UTC
Summary
Seal Status
|
Owner
Author
|
Owner approved in chat on 2026-10-20 ("fix and merge, now"): the 37 hand lines in cli/tri/src/main.rs are the argv glue of tri golden (cmd_golden, its enum arm and mod line); every rule is in specs/tri/land/golden_land.t27. Verilog widths red is inherited from master (bit_diff, lut_treemap rows missing from master's ledger; master fails the same workflow). |
| assert agent_domain(0, b[0..64]) == 0; | ||
| assert agent_domain(27, b[0..64]) == 18; | ||
| } | ||
|
|
….v130; pays the 37 hand lines of tri golden glue (own-language net <= 0) (Closes #11358) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
| assert agent_domain(0, b[0..64]) == 0; | ||
| assert agent_domain(27, b[0..64]) == 18; | ||
| } | ||
|
|
Contributor
PR DashboardGenerated at: 2026-10-20 13:41:04 UTC
Summary
Seal Status
|
gHashTag
added a commit
that referenced
this pull request
Oct 20, 2026
… Body Ratchet green (Closes #11104) (#11401) * widgets: catalogue all 82 widget folders and the factory law -- 28 tools carry 22833 hand lines, 2775 of them pure logic (Closes #10938) specs/widgets/catalog.t27 pins the census measured at trinity origin/main. specs/t27f/widget_factory.t27 states six stages, the t27c-seal build key and the port gate. specs/t27f/frontend.t27 registers SF_WIDGET_FACTORY. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: gatle rules ported to .t27 -- 3 tests, 19 asserts; gen-js lowers no fn bodies, so the logic stage waits (Closes #10938) specs/widgets/gatle_rule.t27 replaces rule.js as the single source. widget_factory.t27 records that gen-js/gen-ts emit declarations only. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: gallery rules as scalar .t27 -- 195 widgets, refusal verdicts and embed sizes run native under t27b, 3 of 3 (Closes #10938) specs/widgets/gallery_rules.t27 holds the decisions of scripts/widgets-from-spec.mjs (trinity); the gallery data stays in trinity gallery.t27. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: the gallery reader as pure .t27 -- meta tags, entities, localOf, exit codes and hud views over bytes; 8 of 8 under t27c and t27b, byte-identical to metaOf on all 337 corpus files (Closes #10938) specs/widgets/page_reader.t27 replaces the reading half of widgets-from-spec.mjs. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: the logic stage runs through wasm -- gatle_rule built with gen-c and zig, 866178 comparisons against rule.js, 0 mismatches (Closes #10938) gen-js lowers no fn bodies (#10943), but the path the Queen's app.wasm takes works: 490 bytes. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: html_out -- the one HTML writer for native widget views, bounded buffer, escaping, en-US grouped numbers; 4 of 4 under t27c and t27b (Closes #10938) Stage "page" of the widget factory: a view writes HTML into a fixed buffer and the host sets innerHTML. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: jtag-verdict is the first widget drawn natively -- words, rules and page in one .t27, a 9269-byte wasm, 3015 DOM comparisons against tool.js, 0 differences (Closes #10938) specs/widgets/jtag_verdict.t27 holds the trinity spec's words and a native page (w_init, w_input, w_click, w_hash, w_render, w_hash_out); html_out.t27 becomes a stateless writer (a pub var does not cross a use). catalog.t27 records gatle at PORT_LOGIC and jtag-verdict at PORT_DRAW. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: json_scan reads data.json as bytes -- 284965 values of 143 widget data files match JSON.parse except 6 numbers of 16-17 digits; html_out gains named fill (Closes #10938) json_scan.t27 gives spans of values (obj_get, arr_at, str_byte with UTF-8 decoding, num_f64) so a native view needs no tree. html_out.t27 gets fill_named and dec_to. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: race-the-bee drawn natively from data.json -- 13014-byte wasm, 59 DOM comparisons including 13 data variants, 0 differences (Closes #10938) specs/widgets/race_the_bee.t27: the trinity spec's words plus the page (tabs, snapshot, lanes, head to head, open issues) over json_scan spans; catalog.t27 records it at PORT_DRAW. Epic #10937. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: lut-treemap logic ported to .t27 -- 9 tests, 14667449 comparisons vs treemap.js, 0 mismatches (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: lut-explain logic ported to .t27 -- 9 tests, 1395124 comparisons vs lut.js, 0 mismatches (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: rename the lut logic specs to underscores (Closes #10960) * widgets: lut-depth logic ported to .t27 -- 7 tests, 12125 comparisons vs ltpparse.js, 0 mismatches (Closes #10938) specs/widgets/lut_depth_logic.t27 ports every function of apps/website/public/widgets/lut-depth/ltpparse.js (parseLtp, classOf, bare; listsOf is only a renaming of the spec constants and is left to the glue). Built to wasm (4430 bytes with a thin C-ABI wrapper) and compared field by field with the original JavaScript on the shipped sample, its prefixes and variants, mutations and grammar fuzz: 0 mismatches; a negative control (class_of min -> max) gives 2838. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: stat-card logic ported to .t27 -- 8 tests, 30827 comparisons vs statparse.js, 0 mismatches (Closes #10938) specs/widgets/stat_card_logic.t27 ports parseStat (with a strict JSON.parse reader and the JS object rules it leaned on: duplicate keys, key order, Number() of counts, the module expansion of flatten) and badgeSvg. listsOf is a renaming of spec constants and is left to the glue. Cases the port does not carry (string/array counts, numbers not exactly rounded here, lone surrogates, integer-like or Object.prototype type names) answer status 9, listed in the header. Built to wasm (13741 bytes with a thin C-ABI wrapper) and compared field by field with the original JavaScript: 0 mismatches; a negative control (group priority reversed) gives 2469. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: fasm-skyline logic ported to .t27 -- 7 tests, 384784 comparisons vs skyline.js, 0 mismatches (Closes #10938) specs/widgets/fasm_skyline_logic.t27 ports every export of apps/website/public/widgets/fasm-skyline/skyline.js: parseFasm, decodeGrid (split into grid_* calls the glue makes from grid.json, plus grid_lookup and grid_at), familyOf and summarize. Built to wasm (8390 bytes with a thin C-ABI wrapper) and compared with the original JavaScript on the real grid.json (every cell, every sample tile name, 60000 random names), FASM texts built from the two shipped samples, line-grammar fuzz, 6000 random small grids and random family lists: 0 mismatches; a negative control (zero test of a constant reversed) gives 6165. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records five more logic files at PORT_LOGIC -- lut-depth, stat-card, fasm-skyline, lut-treemap, lut-explain (Closes #10960) * widgets: js_num -- JavaScript's number text rules in .t27, 1256209 comparisons, 0 differences (Closes #10938) String(x), toFixed, toLocaleString('en-US'), Math.round/floor and a correctly rounded Number(text) as one stateless library for the native widget views (json_scan.num_f64 is off in the last place for the 16- and 17-digit numbers of t27-cell and t27-vs-nvfp4). 5 tests, 7398-byte wasm. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: bit-diff logic ported to .t27 -- 8 tests, 275620 comparisons vs bitparse.js, 0 mismatches (Closes #10938) specs/widgets/bit_diff_logic.t27 ports every export of apps/website/public/widgets/bit-diff/bitparse.js: findSync, parseBitstream, packetLine, packetSequence, lineDiff, frameSlots, mapFrames, frameDelta, compareFrames, bytesIdentical, stripOf, summarize (as getters) and compareBitstreams (compare_all), plus decodeFar/encodeFar, the register, command and block name tables and the constants. Built to wasm (14133 bytes with a thin C-ABI wrapper) and compared field by field with the original JavaScript on 52 real .bit/.bin files from this repository (parse, packet lines, pairs, derived pairs), mutations of their headers and bodies, 5000 synthetic streams and pairs against the four shipped geometries, and the small pure functions: 0 mismatches; a negative control (ECC mask changed) gives 1098 on the quick run. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: vcd-wrapped logic ported to .t27 -- 6 tests, 76076 comparisons vs vcdstats.js, 0 mismatches (Closes #10938) specs/widgets/vcd_wrapped_logic.t27 ports every export of apps/website/public/widgets/vcd-wrapped/vcdstats.js: the streaming reader (createVcdReader as vcd_begin/vcd_push/vcd_end with the VcdError codes and details), summarize, vcdSummary and spanUnits (with an exact toPrecision(4) done in small big integers). Built to wasm (14135 bytes with a thin C-ABI wrapper) and compared field by field with the original JavaScript on the three shipped sample VCDs (whole, in random chunks, prefixes, mutated), grammar fuzz of headers and bodies in random chunking, and 60000+ spanUnits cases: 0 mismatches; a negative control (name order reversed) gives 1730. Cases the port does not carry answer status 90 and are listed in the header. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records bit-diff and vcd-wrapped at PORT_LOGIC (Closes #10960) * widgets: js_val -- String(), Number(), truthiness, fill() and JSON.parse's verdict on a JSON span; js_num grows ceil, min, max, log2 (Closes #10938) js_val.t27 reads a value in the bytes the way JavaScript reads the parsed value (null prints "null", a missing key "undefined", 1.50 prints "1.5", 1e23 prints "1e+23") and fills {name} / {0} templates from a value pool. js_num.t27 gains Math.ceil, two-argument Math.max/min and Math.log2 (fdlibm port, bit-identical to V8), all checked against node over 1925210 comparisons, 0 differences. 9 tests in the two files. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: t27-vs-nvfp4 drawn natively -- 28660-byte wasm, 1517 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: pin-map logic ported to .t27 -- 8 tests, 180587 comparisons vs xdc.js, 0 mismatches (Closes #10938) specs/widgets/pin_map_logic.t27 ports every export of xdc.js: parseXdc, lintXdc, summarize, covers, pinType, isDualConfig. Differential: wasm built from t27c gen (Zig) against the verbatim xdc.js, ASCII domain. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: synth-smells logic ported to .t27 -- 9 tests, 87094 comparisons vs smellparse.js, 0 mismatches (Closes #10938) specs/widgets/synth_smells_logic.t27 ports every export of smellparse.js: CELL_GROUPS/groupOf, parseYosysLog, smellCounts, trimLog. Differential: wasm built from t27c gen (Zig) against the verbatim smellparse.js, full result. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: fsm-sketch logic ported to .t27 -- 10 tests, 91657 comparisons vs fsmparse.js, 0 mismatches (Closes #10938) specs/widgets/fsm_sketch_logic.t27 ports every export of fsmparse.js: sniff, parseAny, parseKiss2, parseLog, countsOf, codeValue, conditionOf, walkOrder, layoutFsm. Math.sin, Math.cos and Math.atan2 of layoutFsm stay with the host (passed in as a table / computed from the returned points): V8's own results differ from fdlibm and musl in the last bit. Differential: wasm built from t27c gen (Zig) against the verbatim fsmparse.js. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: bit-autopsy logic ported to .t27 -- 7 tests, 96007 comparisons vs autopsy.js, 0 mismatches (Closes #10938) specs/widgets/bit_autopsy_logic.t27 ports every export of autopsy.js: the constants and name tables, hex8, parseHeader, findSync, dieOf, lookupIdcode, frameSlots, walkPackets and autopsy. Differential: wasm built from t27c gen (Zig) against the verbatim autopsy.js on the five real .bit files of fpga/openxc7-synth, their mutations and truncations, and 4000 generated streams. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records all 12 logic files at PORT_LOGIC -- pin-map, synth-smells, fsm-sketch, bit-autopsy join (Closes #10960) * widgets: js_val reads Number(text) like JavaScript (white space, 0x/0o/0b with one rounding, -0) -- 170040 comparisons, 0 differences (Closes #10938) Number(text) over 90027 texts and JSON.parse's verdict over 80013 texts against node 23: 0 differences. The hex path folds dropped digits into a sticky bit so a 20-digit 0x value rounds once, as JavaScript does. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: t27-cell drawn natively -- 31401-byte wasm, 3098 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: build-receipt drawn natively -- 32541-byte wasm, 951 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: js_val gains jv_fill_raw, the filler without escaping, for an SVG title (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: uart-bucket drawn natively -- 32562-byte wasm, 3288 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: test-touch drawn natively -- 35775-byte wasm, V8's TimSort ported, 786923-byte data in a 1 MiB buffer, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: native-only constants of the seven pages are W_-prefixed, the name trinity's generator accepts beside SAY_ and K_ (Closes #11016) * widgets: pin-map drawn natively -- 41095-byte wasm, 2582 DOM comparisons, 0 differences (Closes #10938) The trinity spec pin-map.t27 verbatim (two u8*u8 test products cast to u16 so zig can evaluate them) plus the native page: board SVG, search, bank and colour filters, demos, XDC paste/file/drop lint. Reader and lint imported from pin_map_logic.t27. t27c test-report 15/15, t27b 15/15. Native-only constants are named W_...; K_DATA_FILES is the one K_ name added, read by the host. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: lut-depth drawn natively -- 19862-byte wasm, 3603 DOM comparisons, 0 differences (Closes #10938) specs/widgets/lut_depth.t27: the trinity spec verbatim, the reader of lut_depth_logic.t27 folded in (a pub var cannot cross a use), and the page: sample, paste box, file input and drop, path-block buttons, staircase, legend, table. 16 tests (12 new), t27b 16/16. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: lut-explain drawn natively -- 64924-byte wasm, 3055 DOM comparisons, 0 differences (Closes #10938) specs/widgets/lut_explain.t27: the trinity spec verbatim, the decoder of lut_explain_logic.t27 folded in (a pub var cannot cross a use), and the page: decode with truth table and flips, FASM lines against the db, file read, the guessing game, the design histogram. 19 tests (14 new), t27b 19/19. 38 data variants. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: gf16-calc drawn natively -- 30292-byte wasm, 4544 DOM comparisons, 0 differences (Closes #10938) specs/widgets/gf16_calc.t27: the trinity spec verbatim, a small bignum (16 registers of 640 limbs) for the exact fractions tool.js kept in BigInt, the layout reader, the rounding, ternary and error digits, and the page with its address and copy-link effects. 17 tests (11 new), t27b 17/17. 29 layout/address variants. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records pin-map, lut-depth, lut-explain, gf16-calc at PORT_DRAW; lut-depth tests compare texts with same_text (Closes #11016) * widgets: resource-bars drawn natively -- 21381-byte wasm, 1956 DOM comparisons, 0 differences (Closes #10938) specs/widgets/resource_bars.t27: the trinity spec verbatim, the routed-JSON counter (json_scan, hash sets, a strict JSON check), JavaScript number printing for the bar widths and percentages (shortest round-trip digits, exact toFixed/toPrecision, on a small big-number block), and the page with file input, drop and page-wide drop. File cap 32 MiB (the original 512 MB). 12 tests (8 new), t27b 12/12. 33 data variants. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records resource-bars at PORT_DRAW (Closes #11016) * widgets: canvas_out -- the draw-list protocol for native pages that draw on a canvas (Closes #10938) A text draw list (JSON records, bracket-less) that a page module writes and the shared host replays on canvas.getContext('2d'); writers in html_out's stateless style, 7 tests, protocol in the header. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: fits-on drawn natively -- 30952-byte wasm, 4317 DOM comparisons and 4317 canvas call sequences, 0 differences (Closes #10938) First canvas tool: the page writes a canvas_out draw list that the host replays; the original tool.js and the native page agree on the DOM and on every recorded context call and property assignment over 41 data variants and a 1700-step fuzz. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records fits-on, the first canvas page, at PORT_DRAW (Closes #11016) * widgets: fsm-sketch drawn natively -- 62941-byte wasm, 2612 DOM comparisons, 0 differences (Closes #10938) The trinity spec fsm-sketch.t27 verbatim (one string-vs-string test assert split in two so zig can compile it) plus the native page: sample picker, KISS2 / yosys log reading and the state diagram (SVG) from fsm_sketch_logic.t27, table, key, SVG save. sin/cos/atan2 come from the host through a queue/answer service (w_libm_need / w_libm_give). t27c test-report 12/12; t27b: unsupported (`as` between float and int). Native-only constants are named W_...; K_DATA_FILES is the one K_ name added. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records fsm-sketch at PORT_DRAW (Closes #11016) * widgets: cost-curves drawn natively -- 36644-byte wasm, 2576 DOM comparisons, 0 differences (Closes #10938) The trinity spec cost-curves.t27 (three u16 products cast to u32 so zig can evaluate them) plus the native page: controls, log/linear chart, table, tooltip and halo, notes, findings, provenance. Every chart coordinate is printed as JavaScript's String(double) through the host's libm service (w_libm_need / w_libm_give: log2, log10, pow, shortest digits, and Number() of numbers with more than 15 digits). t27c test-report 11/11 (7 original constant-only tests are vacuous); t27b unsupported (reference CSE hoist in w_tipsize). Native-only constants are W_...; K_DATA_FILES is the one K_ name added. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records cost-curves at PORT_DRAW (Closes #11016) * dupe ledger: bless the widget factory's helper copies -- a `use` carries neither a pub var nor the helpers an imported function calls, so a native page folds the same small helpers (match_in, skip_ws, hex_val ...) into each module; the fix is the compiler gap #11021, after which this ledger shrinks (Closes #11016) Blessed in the same commit as the copies, as tools/dupe_scan.py asks. The ledger moves down only. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: synth-smells drawn natively -- 43594-byte wasm, 2867 DOM comparisons, 0 differences (Closes #10938) Page, yosys-log reader (folded from synth_smells_logic.t27), file/drop/paste UI and the PNG card as a canvas_out draw list plus one fit-to-width record. Measured against the original tool.js in two happy-dom windows: DOM, canvas call sequence, downloads. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: text_meas -- a memo of host text widths for native cards (Closes #10938) tm_width / tm_fit_left / tm_wrap: the card asks, the host measures on a scratch canvas, the card is written again; 4 tests, 20 asserts. Used by vcd-wrapped, bit-diff, bit-autopsy. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: text_meas -- helpers are pub so a page can import them; wrap test with a sentence (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: vcd-wrapped drawn natively -- 51721-byte wasm, 2078 DOM comparisons, 0 differences (Closes #10938) Page, VCD reader (folded from vcd_wrapped_logic.t27), file/drop UI and the PNG card as a canvas_out draw list whose text widths come from text_meas.t27. Measured against the original tool.js in two happy-dom windows: DOM, canvas call sequence, downloads. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: stat-card drawn natively -- 36671-byte wasm, 1530 DOM and canvas comparisons, 0 differences (Closes #10938) The whole page of the stat-card tool in .t27: the 1200x630 card as a canvas draw list (card text cut and wrapped by widths the host measures: w_text_need / w_text_give), the facts table, the README badge SVG, paste / pick / drop, PNG and SVG download, copy with its timer. stat_card_logic.t27 is folded into the module. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: lut-treemap drawn natively -- 41208-byte wasm, 2055 DOM comparisons, 0 differences (Closes #10938) The whole page of the lut-treemap tool in .t27: the squarified treemap as absolutely placed tiles (their labels left off where the host-measured text does not fit), the per-module table, the detail line, both drop boxes (the stat and the flattened stat), the metric buttons, the map size from the host. lut_treemap_logic.t27 is folded into the module; the reader keeps one parse, so the stat that is shown is read again from its text whenever another text was read. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records four canvas pages at PORT_DRAW; dupe ledger re-blessed for the new folded helpers (Closes #11016) * widgets: fasm-skyline drawn natively -- 52245-byte wasm, 2271 DOM and canvas comparisons, 0 differences (Closes #10938) The whole page of the fasm-skyline tool in .t27: the isometric drawing as a binary draw list (bit-exact doubles; text lists cost seconds at 150000 numbers), pan / pinch / wheel / key control, hover and tap info, rotate and fit, the PNG with its card text wrapped to host-measured widths, a binary STL with float32 rounding, FASM file counting, both data files decoded in the module. fasm_skyline_logic.t27 is folded in unchanged. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records fasm-skyline at PORT_DRAW; dupe ledger re-blessed (Closes #11016) * widgets: text_meas -- tm_fit_right, the cut-from-the-right fit with three dots (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: sha256_hex -- SHA-256 as 64 hex digits for the native pages that show a digest (Closes #10938) 3 tests (empty, abc, 56 and 1000 bytes); checked in the pages against crypto.subtle on 9.7 MB streams. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: bit-diff drawn natively -- 55742-byte wasm, 2025 DOM comparisons, 0 differences (Closes #10938) Page, bitstream reader (folded from bit_diff_logic.t27), two drop zones, SHA-256, the comparison written as the same JSON the demo uses, and the frame map as a canvas_out draw list. Measured against the original tool.js in two happy-dom windows: DOM, last canvas paint, downloads; 46 demo.json variants, XC7A200T-sized streams (9.7 MB). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: bit-autopsy drawn natively -- 58804-byte wasm, 1491 DOM comparisons, 0 differences (Closes #10938) Page, bitstream autopsy reader (folded from bit_autopsy_logic.t27), file/drop UI, SHA-256, the result written as the JSON the samples use, the on-page frame map and the PNG card as canvas_out draw lists with text_meas widths. Measured against the original tool.js: DOM, canvas (card exactly, map last paint), downloads; 40 samples.json/data variants, 100T and 200T sized streams. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records bit-diff and bit-autopsy at PORT_DRAW; dupe ledger re-blessed (Closes #11016) * widgets: idcode drawn natively -- 39368-byte wasm, 7446 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: slack-waterfall drawn natively -- 42255-byte wasm, 4151 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: test-waves drawn natively -- 55128-byte wasm, 10903 DOM comparisons, 0 differences (Closes #10938) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records idcode, slack-waterfall and test-waves at PORT_DRAW; dupe ledger re-blessed (Closes #11104) * widgets: placement-map drawn natively -- 37328-byte wasm, 10869 DOM comparisons, 0 differences (Closes #10938) specs/widgets/placement_map.t27 = the trinity spec placement-map.t27 verbatim (module, words, K_ numbers) with a use block and the native page appended. Canvas drawing through the binary draw list, pointer down/move/up/leave/ cancel with pointerId and pointerType, wheel with ctrlKey (Math.exp through the host libm service), keydown on the canvas, ResizeObserver, window.innerHeight, a dropped or picked nextpnr JSON read by the page (cells sorted in UTF-16 order, modules in Object.values order, repeated keys last-wins), Math.hypot as V8 computes it. Edits to the verbatim spec: its eight str == str asserts use same_text (t27c does not compile str == str) and K_GRID_W * K_GRID_H is cast to u32 (the u16 product overflows at compile time). Measured against the original tool.js in two happy-dom windows (DOM and the recorded 2D-context call sequence): 10869 comparisons, 0 differences; 27 placement.json variants; 80 fuzzed runs; hypot2 against node over 4000529 pairs, 0 differences; 21 negative controls detected. Files are read up to 64 MiB (tool.js 512 MB), at most 300000 cells, JSON nested at most 60 deep; modules or cells given as arrays and lone surrogates are not modelled. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records placement-map at PORT_DRAW; dupe ledger re-blessed (Closes #11104) * widths ledger: bit_diff and lut_treemap carry 256 KiB text arrays (2^21 bits), host-side plumbing of browser pages, the same class as why_card and deps_card (Closes #11104) * widgets: t27c-stages drawn natively -- 29515-byte wasm, 3431 DOM comparisons, 0 differences (Closes #10938) specs/widgets/t27c_stages.t27 = the trinity spec verbatim + the page in .t27. The page asks the compiler (public/t27/t27_compiler.wasm) through a request/answer pair, w_compile_need and w_compile_give, the way the libm and text services are asked; the host runs t27_analyze. 14 tests (t27b: 14 passed); proven against tool.js in two happy-dom windows with the real compiler. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: gatle drawn natively -- 38685-byte wasm, 1632 DOM comparisons, 0 differences (Closes #10938) specs/widgets/gatle.t27 = the trinity spec verbatim + the page in .t27 (tool.js and the URL-hash half of rule.js; the rest of rule.js is gatle_rule.t27). The Verilog is compiled through the same w_compile_need / w_compile_give service as t27c-stages; the digests are sha256_hex.t27. 19 tests (t27b: 19 passed); proven against tool.js in two happy-dom windows: 52 days, practice, hash round trips, the pool of 37 puzzles, 15 pool variants, 11 failure cases. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> * widgets: catalog records gatle and t27c-stages at PORT_DRAW -- every one of the 28 tools has a native page (Closes #11104) --------- Co-authored-by: Claude <claude@anthropic.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #11358. Part of epic #11275 (T27 OS); see also #11326, #11315.
What lands:
tri golden land | push | status(own git, golden_chain.t27 section 6): the first real link (T27 OS: swarm_pulse.t27 -- the hourly watch decides one action from the pulse numbers (first golden-chain link) #11348) already went through it.Why not
tri land: glass_window.t27 holds @extern host calls, which t27c cannot compile (t27b tests it: 47/47). Same path as #11254.Generated with Claude Code (https://claude.com/claude-code)