{"entries":[{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-08-27T11:20:10.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-08-27T11:20:10.000000-04:00"},"hash":"31fc1aeface2fb9bd986f7afbcfd66b8ff063a56","message":"fix(breaking): bordered elements should reserve layout space for border sides (#116)\n\n* fix: bordered boxes now reserve layout space for their border sides\n\nClay doesn't account for border widths in layout — borders were drawn as\nvisual overlays, so any element with border width > padding had its content\ncollapsed behind the border glyphs (or, for fit-height boxes, collapsed to a\nsingle row with top and bottom glyphs overlapping and children invisible).\n\nFix: at pack time, compute effective padding per side as\nmax(userPadding, borderWidth). Border glyphs draw in the same cells as before;\nonly the Clay layout values change so the engine reserves those cells.\n\nSemantics of the max rule:\n- No explicit padding: border width becomes the effective padding; content\n lands inside the border, not behind it.\n- padding == borderWidth (prior workaround): max evaluates to the same value,\n no double-reservation; these elements render identically.\n- padding > borderWidth: extra padding provides breathing room inside the\n border, measured from the border edge inward.\n\nCallers who set padding == borderWidth as a workaround are unaffected.\nDownstream compensators (e.g. lgtm.shop Panel) will render identically until\nthey drop the manual compensation on their next pin bump.\n\nResolves Open Decision #4 in specs/renderer-spec.md.\n\n* fix!: border padding is now additive, not max\n\nBorder presence implies padding on that side equal to the border width.\nCallers who compensated by setting `padding == borderWidth` now receive\ndouble-reservation and must remove the workaround padding.\n\n`padding: 1` with `border: 1` → effective 2; `border: 1` alone → effective 1.\n\n* refactor: move border padding reservation into the wasm renderer\n\nThe additive effective-padding rule (userPadding + borderWidth per side)\nwas applied in pack() on the TypeScript side, so the packed layout word\ncarried a pre-computed value. It now happens in clayterm.c when the\nPROP_BORDER block decodes border widths: decl is zero-initialized and\nPROP_LAYOUT decodes first, so border-without-layout and\nlayout-without-border both fall out naturally.\n\nThe wire format's padding field now carries raw user padding; the\nrenderer owns the reservation. Behavior is unchanged — all existing\nborder tests pass as-is.","tree":"46bf4e2b7a0cfdb3b35710d6f1deb2257eae5d4b"},"mode":"0100644","name":"border.test.ts","size":20761},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-10-09T10:18:05.000000-05:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-10-09T10:18:05.000000-05:00"},"hash":"d8f2a247b2607f783d8d00790725ff954b3d7410","message":"feat: input parser probe-response recognition → CapabilityEvent (#132)\n\nCo-authored-by: bombshell-cooper[bot] <300334852+bombshell-cooper[bot]@users.noreply.github.com>","tree":"170beaa5910628c0cbd0fd600d8da7bef7163816"},"mode":"0100644","name":"caps.ts","size":361},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-08-27T11:20:10.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-08-27T11:20:10.000000-04:00"},"hash":"31fc1aeface2fb9bd986f7afbcfd66b8ff063a56","message":"fix(breaking): bordered elements should reserve layout space for border sides (#116)\n\n* fix: bordered boxes now reserve layout space for their border sides\n\nClay doesn't account for border widths in layout — borders were drawn as\nvisual overlays, so any element with border width > padding had its content\ncollapsed behind the border glyphs (or, for fit-height boxes, collapsed to a\nsingle row with top and bottom glyphs overlapping and children invisible).\n\nFix: at pack time, compute effective padding per side as\nmax(userPadding, borderWidth). Border glyphs draw in the same cells as before;\nonly the Clay layout values change so the engine reserves those cells.\n\nSemantics of the max rule:\n- No explicit padding: border width becomes the effective padding; content\n lands inside the border, not behind it.\n- padding == borderWidth (prior workaround): max evaluates to the same value,\n no double-reservation; these elements render identically.\n- padding > borderWidth: extra padding provides breathing room inside the\n border, measured from the border edge inward.\n\nCallers who set padding == borderWidth as a workaround are unaffected.\nDownstream compensators (e.g. lgtm.shop Panel) will render identically until\nthey drop the manual compensation on their next pin bump.\n\nResolves Open Decision #4 in specs/renderer-spec.md.\n\n* fix!: border padding is now additive, not max\n\nBorder presence implies padding on that side equal to the border width.\nCallers who compensated by setting `padding == borderWidth` now receive\ndouble-reservation and must remove the workaround padding.\n\n`padding: 1` with `border: 1` → effective 2; `border: 1` alone → effective 1.\n\n* refactor: move border padding reservation into the wasm renderer\n\nThe additive effective-padding rule (userPadding + borderWidth per side)\nwas applied in pack() on the TypeScript side, so the packed layout word\ncarried a pre-computed value. It now happens in clayterm.c when the\nPROP_BORDER block decodes border widths: decl is zero-initialized and\nPROP_LAYOUT decodes first, so border-without-layout and\nlayout-without-border both fall out naturally.\n\nThe wire format's padding field now carries raw user padding; the\nrenderer owns the reservation. Behavior is unchanged — all existing\nborder tests pass as-is.","tree":"46bf4e2b7a0cfdb3b35710d6f1deb2257eae5d4b"},"mode":"0100644","name":"clip.test.ts","size":5402},{"lastCommit":{"author":{"email":"rauhryan@users.noreply.github.com","name":"Ryan Rauh","when":"2026-06-14T10:15:47.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-06-14T10:15:47.000000-04:00"},"hash":"be7e746d5921dbc9ee41bb0f0b74b121ea6b6d5d","message":"✨ Add border-cell backgrounds (#94)\n\n* ✨ Add border-cell backgrounds\n\n* format the spec file\n\n* fix merge conflicts\n\n* Adds code comment and validation for bg\n\n* Updated the tests as suggests by paul","tree":"879636da3a6ad198ff8e1586cdd77bfe4a1b224e"},"mode":"0100644","name":"color.test.ts","size":5775},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-10-05T21:04:40.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-10-05T20:04:40.000000-05:00"},"hash":"00aee53c792aab21d1e197c20cc7d5024474fb8b","message":"feat: terminfo capability core (parser, detectTerminal) (#131)","tree":"42f6a1166b2724d037defd20bd235d3df06ecdce"},"mode":"0100644","name":"fixtures.ts","size":6019},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-06-17T11:45:18.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-06-17T11:45:18.000000-04:00"},"hash":"25fb6a091674df8a1c13d109b3f0560f067a64f0","message":"benchmark: replace simulation with walltime macro benches + e2e correctness (#91)\n\n* test: add input event-loop integration test\n\nDrives createInput as a consumer would (read chunk, scan, flush a pending ESC, dispatch) and asserts the same event stream whether input is fed whole, byte-by-byte, or split mid-sequence. Covers the lone-ESC flush path.\n\n* bench: add macro throughput WallTime bench\n\nIn-process bench over a large mixed corpus, measured in WallTime where real work dominates placement/JIT/alloc noise (the Simulation micro-benches sit on codegen cliffs). Feeds the corpus in small reads: a single scan() drains at most 256 events (the wasm event-buffer cap), so small reads keep every call under the cap and process the whole corpus.\n\n* bench: remove input micro-suite (codegen-cliff artifacts)\n\nThe Simulation input micro-benches (long input burst, printable ASCII single char) move by 50-90% on unrelated changes, even a test-only rename, because their simulated cost snaps to a different value when the combined wasm shifts. Input perf is now gated by the throughput WallTime bench; correctness by the event-loop integration test.\n\n* ci(bench): pin build runner and drop wasm cache\n\nPin ubuntu-24.04 so the wasm toolchain is stable and drop the wasm cache so main and PRs always rebuild identically. The cache froze main's baseline on a stale build, so every PR compared a fresh build against a stale baseline and produced phantom regressions.\n\n* fix(ci): return a promise from throughput bench for CodSpeed walltime\n\nCodSpeed's walltime tinybench plugin only populates result.latency on its async path; a sync task fn leaves it undefined and crashes (Cannot destructure 'min' of result.latency). startup.bench works because its tasks return a promise (spawnFixture). Return Promise.resolve() from the throughput task so the plugin takes the async path — a bare async fn with no await would trip deno's require-await lint. The walltime job runs startup and throughput as separate node processes.\n\n* bench: move render/ops to WallTime macro benches, drop the simulation job\n\nCodSpeed Simulation (Valgrind) is unviable for CI here: flaky measurements (dashboard layout swung 20x, diff render 17% on changes that touch no render code) and unpredictable runtime — the same commit's simulation job finished in ~2 min one run and hung past 30 min the next. Convert render/ops to ms-scale WallTime macro benches (looped, promise-returning, ~7-11ms at <1% variance) run as separate node processes in the walltime job, and drop the simulation job entirely. mod.ts is now a local aggregator for deno task bench.\n\n* ref(test): use built-in\n\n* ref(bench): centralize withCodSpeed workaround\n\n* ref(test): use semantic util functions","tree":"754d2317241a192a71a2eebf8118a65fbd02cac5"},"mode":"0100644","name":"input-loop.test.ts","size":4308},{"lastCommit":{"author":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-08T18:16:46.000000-05:00"},"committer":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-10T12:25:51.000000-05:00"},"hash":"60908220a7433a8c69ee917350263d373ce81ca6","message":"test(input): pin OSC 22 reply edge cases\n\n#101's bespoke parse_osc() drew review for unbounded digit accumulation\non `ESC ] 9999...`. The rebuild drops that parser and relies on #132's\nparse_osc_response(), which already bails once the OSC number passes\n22 and caps payloads at MAX_RESPONSE. These tests pin that behavior for\nOSC 22 so it cannot regress: BEL and ST terminators, the \"0\"\nempty-stack reply, byte-at-a-time delivery, a broken ST, a 64-digit\nOSC number, OSC 220, and an over-long payload (no event, parser\nrecovers for the next reply).\n","tree":"ad8dfb8cb8833254662cf2d0e2ef4ba222df5132"},"mode":"0100644","name":"input.test.ts","size":45599},{"lastCommit":{"author":{"email":"rauhryan@users.noreply.github.com","name":"Ryan Rauh","when":"2026-06-14T10:04:03.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-06-14T10:04:03.000000-04:00"},"hash":"b0ded4ff011737d98581c16c00c3c3b55905c282","message":"feat: expand floating parameters (#29)\n\n* feat: expand floating parameters\n\n* ♻️ use string literals for floating options\n\n* Remove any casts from validation tests","tree":"212ea62592fdd78f475ce108041226dbe1b8f99d"},"mode":"0100644","name":"pack.test.ts","size":2053},{"lastCommit":{"author":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-08T20:10:57.000000-05:00"},"committer":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-10T12:25:51.000000-05:00"},"hash":"72cc4b24db8db0269253dc6839193375476bcf75","message":"feat: emit OSC 22 pointer shapes in output when the terminal supports them (renderer-spec §7.9)\n\nElements declare open(id, { pointerShape }). While\nterm.capabilities.pointerShape is true (raised by #132/#133 from the\nprobe's OSC 22 query reply), each render resolves the shape under the\npointer and appends `OSC 22 ; ST` to result.output when it\nchanges. The host keeps writing one buffer (cowboyd on #101). The\ncapability is the only gate; there is no createTerm option.\n\nEverything lives in TypeScript; the wasm module and the packed encoding\nare untouched, so there is no size or startup cost (#101 regressed\ncreateTerm ~20% and grew the bundle 8.3 KB). Without the capability the\nrender path pays one boolean check and never reads pointerShape (a test\npins this with a counting getter). With it, the directive walk runs only\nwhen the pointer is over something, and only a frame whose shape\nchanged copies output to append the sequence.\n\nResolution takes the last declaring id in Clay's pointer-over order:\npre-order within the topmost layer the pointer reaches, so the innermost\nelement wins and capture-mode floats hide what is beneath while\npassthrough floats defer to it. snapshot() records its declared shapes\nbeside the packed bytes so pre-packed subtrees still participate.\nValues outside the kitty/CSS vocabulary are ignored (and rejected by\nvalidate()), so arbitrary strings never reach an OSC payload.\n\nRestore (dreyfus92 on #101): set-only OSC 22 cannot pop, so the emitted\nshape is the one piece of cross-frame state. It resets to \"default\"\nwhen the pointer leaves or is omitted, and update() returns the reset\nwhen the pointer-shape capability is withdrawn. Resize keeps it, since\nthe terminal's pointer did not change.\n","tree":"1d0aac7e49288eb3725e43a0c7d994017b93631a"},"mode":"0100644","name":"pointer-shape.test.ts","size":8882},{"lastCommit":{"author":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-03-30T15:07:09.000000-05:00"},"committer":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-03-30T15:07:34.000000-05:00"},"hash":"d3636339d50112d7d681e320104c2e2c0b051ff2","message":"✨ add pointer events and interactive hover to Clay render pipeline\n\nClay's pointer hit-testing is now wired through render() via RenderOptions,\nreturning pointerenter/pointerleave/pointerclick events by diffing\nClay_GetPointerOverIds() across frames. The demo keyboard shows hover\nhighlights on keys, semantic element IDs, and configurable event log filters.\n","tree":"35011b93cef94aacd6f4f1bb45b30f275bbc788f"},"mode":"0100644","name":"pointer.test.ts","size":5483},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-10-09T10:00:52.000000-05:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-10-09T10:00:52.000000-05:00"},"hash":"666546e88a54885eb18c5fd5d288f3844483c58b","message":"fix: preserve grapheme clusters (combining marks) in cell emission (#114)","tree":"edb58d7dd985ff216ea292086faad6feae5e50c3"},"mode":"0100644","name":"print.ts","size":2493},{"lastCommit":{"author":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-04-17T12:25:35.000000-05:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-04-17T12:25:35.000000-05:00"},"hash":"a6b1ee1e953c15faaf6a79217695a05bdddfe38d","message":"Merge pull request #14 from thefrontside/decouple-altbuffer-clear\n\nDecouple alternate buffer switching from clearing","tree":"9bbc6827cbea762fdc855974defa40a8cb194faf"},"mode":"0100644","name":"settings.test.ts","size":2323},{"lastCommit":{"author":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-03-14T17:32:11.000000-05:00"},"committer":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-03-15T23:40:26.000000-05:00"},"hash":"d070a2592a55f5d46dadb4b48d36bb09931aab09","message":"🎉 initial commit\n\nWASM-based terminal renderer powered by Clay UI layout engine.\nIncludes command buffer protocol for TS→C communication,\ndouble-buffered ANSI output, and Deno test harness.\n","tree":"c4243fb11dc4cfa6b51a0db5e5aa025f0f3baa0a"},"mode":"0100644","name":"suite.ts","size":99},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-10-09T10:24:17.000000-05:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-10-09T10:24:17.000000-05:00"},"hash":"2c6215cf2d255ccf06b609327c0d89c0e2bfab09","message":"feat: term.update() folds CapabilityEvents into runtime capabilities (#133)","tree":"c76527accbf43c62cbff41b848a43d9067210a77"},"mode":"0100644","name":"term.test.ts","size":28171},{"lastCommit":{"author":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-09T13:35:43.000000-05:00"},"committer":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-10T12:25:51.000000-05:00"},"hash":"770d205673a064ad44e04a7f09e71b88f86b66f0","message":"feat: seed pointerShape from a known-support table in detectTerminal()\n\nThe OSC 22 query is answered by kitty alone; every other set-supporting\nterminal (ghostty, foot, xterm ≥ 367) never replies, so detection could\nnever see them and the demos silently degraded in the most popular\nmodern terminal.\n\nterminfo-spec §6.1 now carries a known-support table of terminals with\nprimary evidence for OSC 22 set support — ghostty (its own docs), kitty\n(its own spec), foot (foot-ctlseqs(7)), and xterm (ctlseqs, version\nfloor 367 via XTERM_VERSION). detectTerminal() seeds pointerShape from\nthe environment; the probe reply remains authoritative evidence, and the\nmanual CapabilityEvent opt-in stays for anything the table misses.\n\nmintty is deliberately excluded: it implements OSC 22 but its accepted\nnames are Windows cursor names (hand, ibeam), not the CSS names §7.9\nemits. Rio and Konsole lack primary evidence (Konsole is contradicted\nby ucs-detect, which — like our probe — tests the query form).\n\nThe spec records the accepted false-positive trade: nested contexts\nthat inherit the environment unchanged (Emacs term-mode) are granted by\nassociation; muxers and IDE terminals rewrite their identity and are\nnaturally excluded.\n","tree":"686f097e42f13919e79d154ac06ed16ac4a769ae"},"mode":"0100644","name":"terminfo.test.ts","size":9002},{"lastCommit":{"author":{"email":"cowboyd@frontside.com","name":"Charles Lowell","when":"2026-06-26T01:10:28.000000-05:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-06-26T09:10:28.000000+03:00"},"hash":"5e253a722d6a2a4072797dbc76729a8a0c990ece","message":"✨Transitions (#27)\n\n* 📝 add transitions design specification\n\nDesign spec for clayterm transitions: frame-snapshot-compatible interpolation\nof element position, size, and color properties. Defines the deltaTime\nconvention, the animating signal on RenderResult, declarative enter/exit\nsemantics that replace Clay's function-pointer callbacks, and cancellation\nas a structural consequence of re-describing state. Implementation is\ngated on bumping the Clay submodule past the upstream transition commit.\n\n* ⬆️ bump Clay submodule to latest main (transitions API)\n\n* 🔧 adapt clayterm to new Clay signatures (OpenTextElement, EndLayout)\n\n* ⬆️ pin Clay to 938967a (work around upstream CLAY_WASM_EXPORT typo)\n\n* ✨ add deltaTime parameter to reduce()\n\n* 🔧 add deltaTime to Native.reduce signature\n\n* ✨ track deltaTime on Term, accept deltaTime override\n\n* ✨ add animating_count to Clayterm context\n\n* 🔧 expose animating() via Native binding\n\n* ✨ surface animating: boolean on RenderResult\n\n* 📝 rewrite transitions spec for v1 (Clay-supported subset)\n\nScope v1 to what Clay currently supports without userData on transition\ncallbacks: one duration and one easing per element, applied to all listed\nproperties. Drop per-property longhand, enter/exit deltas, cubicBezier,\nand corner radius — each with an explicit \"Deferred Until Upstream Clay\"\nentry in §13 referencing nicbarker/clay#603 and the forthcoming exit-flag\nwork. Easings are plain string literals (\"linear\" | \"easeIn\" | \"easeOut\"\n| \"easeInOut\") since v1 has no parametric easings.\n\n* ✨ add transition property names, bitmask helpers, and Easing\n\n* ✨ add transition field type to OpenElement\n\n* ✨ encode transition block in pack()\n\n* ✨ register Clay handlers, interpolate on property change\n\nCo-Authored-By:\n\n* ✨ reset deltaTime to 0 after idle (preserve transitions across long gaps)\n\n* ✅ verify color transitions work in line mode\n\n* 🎨 apply deno fmt and clang-format\n\n* ✨ add transitions demo (collapsing sidebar)\n\n* 📝 reference transitions-spec from renderer-spec\n\n* ✨ rewrite transitions demo as interactive full-screen menu\n\n* ✨ add clay-transitions demo port (v1-compatible subset)\n\nPorts the spirit of the raylib-transitions demo to clayterm: a 4×4 grid\nof colored boxes that animate position, size, and bg color. Shuffle (s)\nanimates positions via Clay's transition system; recolor (c) toggles\nbetween two palettes with animated bg interpolation; hover tints each\nbox by blending its bg toward white (overlay-color field is not yet in\nthe v1 command buffer, so lighten-on-hover substitutes). Full mouse\ntracking is wired via mouseTracking() + pointer state from input events.\n\n* 🎨 let clay-transitions demo rows fill available height\n\n* 🎨 remove modeline from clay-transitions demo\n\n* 📝 note ct_active_context is a workaround for Clay userData PR\n\n* 🎨 use border-only boxes in clay-transitions demo\n\n* 🎨 prevent menu text from wrapping during sidebar transition\n\n* 🔥 drop unused grow import in transitions-run test\n\n* Update transitions branch with upstream changes (#53)\n\n* ✨ add snapshot() for pre-packing directive subtrees\n\nIntroduces a new `snapshot(ops)` constructor that pre-packs a directive\narray into its transfer encoding. The returned opaque `Op` can be\nspliced into any directive array, and during packing its bytes are\ncopied directly into the command buffer without re-encoding.\n\nThis enables higher-level frameworks to implement dirty tracking:\nunchanged component subtrees can reuse a cached snapshot, skipping\nthe per-frame packing cost entirely.\n\n* docs: add maintainer build guide\n\n* 💄 format build docs\n\n* 🐛 improve pack string overflow errors\n\n* os matrix test in ci (#36)\n\n* 🧼 optimize build\n\n* 🧼 compress bundled wasm\n\n* 🧼 optimize wcwidth.c size\n\n* ⚙️ update npm settings\n\n* 🐛 install wasm-opt in ci\n\n* ⚡ use brotli-11 + z85 wasm encoding\n\n* 📌 pin @types/node to v22\n\n* 🧼 apply @ghostdevv review suggestions from PR #35\n\nCo-Authored-By: ghostdevv \n\n* 🔨 add type to bundle-wasm\n\n* 💌 signed, sealed, delivered\n\n* chore: use hashes for versions\n\n* chore: don't save git credentials\n\n* chore: use array syntax\n\nfor some reason the schema for the actions wants it to be an array\n\n* perf: set concurrency limits to reduce cost and improve dx\n\nWithout this it means that, for example, if I push a change to a PR then\nshortly push again this workflow will be running twice. This change will\ncancel the old run before starting the new one, which reduces the\noverall actions cost and DX as you don't have extra runs\n\n* chore: use hashes for versions\n\n* chore: update node version\n\n* perf: set concurrency limits to reduce cost and improve dx\n\nWithout this it means that, for example, if I push a change to a PR then\nshortly push again this workflow will be running twice. This change will\ncancel the old run before starting the new one, which reduces the\noverall actions cost and DX as you don't have extra runs\n\n* chore: don't save git credentials\n\n* chore: mitigate potential template injection\n\nSee https://docs.zizmor.sh/audits/#template-injection\n\n* chore: update ::set-output command to new syntax\n\nhttps://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/\n\n* chore: use hashes for versions\n\n* chore: don't save git credentials\n\n* chore: limit id-token permission to the publishing steps\n\n* chore: explicitly disable npm cache to mitigate cache poisoning attacks\n\n* chore: mitigate potential template injection\n\nSee https://docs.zizmor.sh/audits/#template-injection\n\n* chore: use oidc\n\nThis should be using OIDC for publishing\n\n* chore: update ::set-output command to new syntax\n\nhttps://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/\n\n* 🙅 revert aggressive optimization experiments pending benchmark\n\nReverts three changes that need benchmarks before landing:\n- -Oz / wasm-opt\n- brotli+Z85 wasm compression\n- wcwidth.c rewrite\n\n* Add CodSpeed performance benchmarks\n\n* 🔨 use deno\n\n* 🧼 deno fmt\n\n* ⚙️ vitest -> tinybench\n\n* 🧼 remove codspeed assets\n\n* 🔨 fix ci\n\n* add type module\n\n* fmt\n\n* chore: update github url (#38)\n\n* downgrade to tinybench@5\n\n* move to examples folder with readme\n\n* 🔧 export animating from wasm build\n\n* 🧪 cover transitions in snapshots and validation\n\n* ✅ enforce nonnegative transition duration\n\n---------\n\nCo-authored-by: Charles Lowell \nCo-authored-by: Jacob Bolda \nCo-authored-by: Nate Moore \nCo-authored-by: ghostdevv \nCo-authored-by: Nate Moore \nCo-authored-by: codspeed-hq[bot] <117304815+codspeed-hq[bot]@users.noreply.github.com>\nCo-authored-by: Nate Moore \n\n* remove uncecessary note\n\n* clean up and consolidate\n\n* re-organize transitions examples\n\n* 🔥remove redundant cast\n\n---------\n\nCo-authored-by: Ryan Rauh \nCo-authored-by: Jacob Bolda \nCo-authored-by: Nate Moore \nCo-authored-by: ghostdevv \nCo-authored-by: Nate Moore \nCo-authored-by: codspeed-hq[bot] <117304815+codspeed-hq[bot]@users.noreply.github.com>\nCo-authored-by: Nate Moore \nCo-authored-by: Ryan Rauh ","tree":"ba60a90298902acf4ff586e009bd1b69d3490149"},"mode":"0100644","name":"transitions.test.ts","size":3193},{"lastCommit":{"author":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-08T20:10:57.000000-05:00"},"committer":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-10T12:25:51.000000-05:00"},"hash":"72cc4b24db8db0269253dc6839193375476bcf75","message":"feat: emit OSC 22 pointer shapes in output when the terminal supports them (renderer-spec §7.9)\n\nElements declare open(id, { pointerShape }). While\nterm.capabilities.pointerShape is true (raised by #132/#133 from the\nprobe's OSC 22 query reply), each render resolves the shape under the\npointer and appends `OSC 22 ; ST` to result.output when it\nchanges. The host keeps writing one buffer (cowboyd on #101). The\ncapability is the only gate; there is no createTerm option.\n\nEverything lives in TypeScript; the wasm module and the packed encoding\nare untouched, so there is no size or startup cost (#101 regressed\ncreateTerm ~20% and grew the bundle 8.3 KB). Without the capability the\nrender path pays one boolean check and never reads pointerShape (a test\npins this with a counting getter). With it, the directive walk runs only\nwhen the pointer is over something, and only a frame whose shape\nchanged copies output to append the sequence.\n\nResolution takes the last declaring id in Clay's pointer-over order:\npre-order within the topmost layer the pointer reaches, so the innermost\nelement wins and capture-mode floats hide what is beneath while\npassthrough floats defer to it. snapshot() records its declared shapes\nbeside the packed bytes so pre-packed subtrees still participate.\nValues outside the kitty/CSS vocabulary are ignored (and rejected by\nvalidate()), so arbitrary strings never reach an OSC payload.\n\nRestore (dreyfus92 on #101): set-only OSC 22 cannot pop, so the emitted\nshape is the one piece of cross-frame state. It resets to \"default\"\nwhen the pointer leaves or is omitted, and update() returns the reset\nwhen the pointer-shape capability is withdrawn. Resize keeps it, since\nthe terminal's pointer did not change.\n","tree":"1d0aac7e49288eb3725e43a0c7d994017b93631a"},"mode":"0100644","name":"validate.test.ts","size":8769},{"lastCommit":{"author":{"email":"natemoo-re@users.noreply.github.com","name":"Nate Moore","when":"2026-07-29T16:19:32.000000-04:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-29T16:19:32.000000-04:00"},"hash":"d4bc79b55e33e07d9c227626367404e11d4804f0","message":"🧼 generate compact `wcwidth` tables (#49)\n\n* ref(wcwidth): generate efficient wcwidth tables\n\n* ref(wcwidth): deslop\n\n* perf(wcwidth): add BMP filter + fast path, use single table for binary search\n\n* ref(ci): add verify task\n\n* deno fmt\n\n* ref(wcwidth): generate clang-format friendly code\n\n* fix(verify): handle --check flag\n\n* deno lint\n\n* ref(wcwidth): add datahash\n\n* ref(wcwidth): correctness pass\n\n* ref(wcwidth): performance pass\n\n* ref(wcwidth): upgrade to unicode 17\n\n* ref(wcwidth): regen\n\n* fix format\n\n* fix(render): measure non-printable codepoints as one cell\n\nThe draw path coerces wcwidth() == -1 to one cell and emits U+FFFD, but\nthe Clay measure callback skipped those codepoints entirely, so a\nfit-sized box around text containing a control character or\nnoncharacter measured narrower than what the renderer draws. Mirror\ndraw_text in measure(): coerce -1 to 1, and substitute U+FFFD on UTF-8\ndecode failure instead of passing an uninitialized codepoint to\nwcwidth().\n\n* fix(wcwidth): treat surrogates as non-printable\n\nutf8_decode() accepts CESU-8 surrogate encodings, and the Unicode 17\ntables default unlisted codepoints to width 1, so a surrogate reaching\ndraw_text was emitted to the terminal as invalid UTF-8 (the old\ntermbox2 table returned -1 for U+D800..U+DFFF and the emit path\nsanitized it to U+FFFD). Guard the surrogate range in the generated\nwcwidth() so iswprint() rejects it and the emit path substitutes\nU+FFFD again.\n\nAlso derive the generated header's Unicode version from the same\nconstant as the data URL; it was left saying 16.0 after the 17.0\nupgrade.","tree":"b3c282619bd969253734bd9996b6412a5da48f3c"},"mode":"0100644","name":"width.test.ts","size":3000}],"lastCommit":{"author":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-09T13:35:43.000000-05:00"},"committer":{"email":"git@natemoo.re","name":"Nate Moore","when":"2026-10-10T12:25:51.000000-05:00"},"hash":"770d205673a064ad44e04a7f09e71b88f86b66f0","message":"feat: seed pointerShape from a known-support table in detectTerminal()\n\nThe OSC 22 query is answered by kitty alone; every other set-supporting\nterminal (ghostty, foot, xterm ≥ 367) never replies, so detection could\nnever see them and the demos silently degraded in the most popular\nmodern terminal.\n\nterminfo-spec §6.1 now carries a known-support table of terminals with\nprimary evidence for OSC 22 set support — ghostty (its own docs), kitty\n(its own spec), foot (foot-ctlseqs(7)), and xterm (ctlseqs, version\nfloor 367 via XTERM_VERSION). detectTerminal() seeds pointerShape from\nthe environment; the probe reply remains authoritative evidence, and the\nmanual CapabilityEvent opt-in stays for anything the table misses.\n\nmintty is deliberately excluded: it implements OSC 22 but its accepted\nnames are Windows cursor names (hand, ibeam), not the CSS names §7.9\nemits. Rio and Konsole lack primary evidence (Konsole is contradicted\nby ucs-detect, which — like our probe — tests the query form).\n\nThe spec records the accepted false-positive trade: nested contexts\nthat inherit the environment unchanged (Emacs term-mode) are granted by\nassociation; muxers and IDE terminals rewrite their identity and are\nnaturally excluded.\n","tree":"686f097e42f13919e79d154ac06ed16ac4a769ae"}}