go
/
svelte-kit-vice
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '9272c54278'
${ noResults }
svelte-kit-vice/docs/process/CONTINUE-docs-corpus.md

74 lines
14 KiB
Raw Normal View History Unescape

docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
# CONTINUE — reorganización del corpus de documentación
Hand-off para retomar en otra sesión. Trabajo iniciado 2026-06-13/14, rama `active-uix`.
## Objetivo
Convertir la documentación del **framework** (dispersa, ~32k líneas) en un corpus
**estratificado, claro y completo**, en inglés, que sirva de base para (a) un libro
sobre cómo construir frameworks / cómo está construido `active`, y (b) la web de
docs. **Componentes y `web/routes` quedan FUERA** (fase futura).
Diagnóstico de partida: NO es falta de contenido (204 READMEs + 38 docs ad-hoc, sin
stubs). Es **estructura** — sin entrada única, sin canon único, mezcla de
permanente/efímero, drift, dos idiomas.
## Reglas de trabajo (no negociables)
- **PROHIBIDO lanzar agentes / workflows.** Hacer el trabajo a mano. (Memoria: `feedback-no-subagents-do-it-yourself`.)
- **Responder en castellano.** Comentarios de código en inglés.
- **NUNCA tocar words/palabras** (`soma/components/words`, `eidos/components/palabras`).
- **Verificar staged antes de commit**: `git reset -q` → `git add` solo los míos → confirmar `git diff --cached --name-only` sin foráneos (launch/package/animations/palabras/words/settings.local). Commit por bloque, push.
- El libro `docs/Disenando_lo_que_ocurre_v2_3.md` (untracked, del usuario) es la **fuente editorial del canon**. NO tocarlo. Leído entero esta sesión.
## Decisiones acordadas
- **Idioma objetivo: inglés** (migración gradual; los docs de capa siguen en español por ahora).
- **Estructura híbrida**: `docs/` raíz para entrada/canon/RFCs/proceso; referencia de capa in-place.
- **`docs/process/`** = artefactos efímeros (hand-offs, snapshots, auditorías). NUNCA fuente de verdad.
- **Estratos**: E0 orientación · E1 arquitectura por capa · E2 canon · E3 decisiones/RFC · E4 guías · E5 referencia de módulo · process (efímero).
- **Canon**: destila doctrina (del libro) + **enlaza el código para los valores** (no copia tablas que driftan). Cada doc que necesite el vocabulario enlaza `docs/CANON.md`, no lo copia.
- **soma**: dos docs con roles separados (README=onboarding, SOMA_ARCHITECTURE=referencia profunda), sin retirar ninguno.
- **active_architecture**: ya hecho (§0→contratos, §10→process snapshot).
## HECHO (pusheado en `active-uix`)
- **Fase 1 — saneo**: drift de canon corregido contra el código (8 familias con `delegate`, `SEMA_FAMILY_POLICY` dos ejes, verbos reales, DATA_ATTRS "Terra"→LEGACY); hand-offs embebidos de 6 READMEs de capa → `docs/process/handoffs-2026-05.md`; §10 de active_arch. Commits `5c18a892`, `19ce90b0`, `e74a54ba`, `41c3b0f9`, `b7121054`. (+ `67cec940` timer auto-key, no-corpus.)
- **Fase 2 — canon E2**: `docs/CANON.md` (anclado al libro por capítulo + código por `file:símbolo`); 5 docs apuntan a él como autoridad + entrada lectura 0. Commits `c9032a7f`, `462d86c0`.
- **Fase 3 (2/3)**: `active-uix/README` reescrito como referencia atemporal (`633820ef`); `active_architecture` §0→"Contratos mínimos" / §10→snapshot en process (`ae49f737`).
docs(corpus): extract arts layer hand-offs to docs/process/ Close the Phase 1b debt: the "## Handoff 2026-05-14" blocks embedded at the top of arts/README, arts/adom/README and arts/format/README move to docs/process/handoffs-2026-05.md, so the layer references read as timeless. Same mechanical pattern already applied to the UIX READMEs in 41c3b0f9. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 1b cierre — arts hand-offs**: los "## Handoff 2026-05-14" de `arts/README` + `arts/adom/README` + `arts/format/README` extraídos a `docs/process/handoffs-2026-05.md` y quitados de los READMEs (mismo patrón que 1b).
docs(corpus): split soma README (onboarding) from SOMA_ARCHITECTURE (reference) The two docs overlapped ~80% (purpose, layers, runtime parts, Soma class, data-* contracts, naming, checklist). Give each a single role: - README (1057→400 L): onboarding + authoring guide — purpose, membership criterion, morfo contract (slim), component + composition pattern, new- component checklist, and a "deep reference" map pointing into the architecture doc. Inventory now points at the live directory tree instead of a hardcoded (drifting) list. - SOMA_ARCHITECTURE (969→1043 L): the complete deep reference. Absorbed the few reference bits that only lived in the README — mergeProps, KEYS / directional keys, focus / roving / scroll-lock, boolean attr helpers, the internal context() helper and the reactive type-aliases — into a new §8.bis. Trimmed its purpose section (now owned by the README). README sections renumbered contiguously; cross-references verified, no other doc links soma/README by section number. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 3 (3/3) — soma de-dup**: `soma/README` 1057→400 L (ahora onboarding/guía: propósito, pertenencia, morfo slim, patrón de componente + composición, checklist, inventory por puntero) y `SOMA_ARCHITECTURE` 969→1043 L (referencia profunda; absorbió mergeProps · KEYS · focus/roving · boolean-helpers · `context()` · type-aliases reactivos en el nuevo §8.bis). Renumerado limpio del README, cero solape, cross-refs verificadas. **Fase 3 cerrada.**
docs(corpus): give the two component checklists distinct roles + cross-links The "3 duplicate checklists" diagnosis was wrong. There are two legitimate, complementary artifacts plus one redundant summary: - COMPONENT_GUIDE.md — the build checklist (ordered authoring steps 1-40 + rationale rules A1-A37). How to build a component. - COMPONENT_COMPLETION_CHECKLIST.md — the acceptance matrix (machine-audited rules with severity/applicability, bound to scripts/component-audit.ts). When a component counts as done. Merging it into the guide would break the script binding. Consolidation = one source per concern, not one file. Both docs now carry a header that states their role and cross-links the other as the complement. soma/README §9 (the only real copy) is reduced to a pointer at both; it keeps no checklist of its own. Flagged out of scope: COMPONENT_COMPLETION_CHECKLIST M-3.3 still says "7 families" (canon is 8 with delegate). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 4 (1/3) — checklists**: el diagnóstico "3 copias" era erróneo. `COMPONENT_GUIDE` = checklist de **autoría** (pasos 1–40 + reglas A1–A37); `COMPONENT_COMPLETION_CHECKLIST` = **matriz de aceptación** machine-auditada (atada a `scripts/component-audit.ts`, `npm run component:audit`). Son complementarios, NO duplicados — fusionarlos a un archivo rompería el binding del script. Consolidación real: una fuente **por concern** + cross-links bidireccionales con roles nítidos; `soma/README §9` (la única copia-resumen real) reducido a puntero a ambos. Pendiente menor (flagged, fuera de scope): drift `7 familias`→8 en `COMPONENT_COMPLETION_CHECKLIST` M-3.3.
docs(corpus): add docs/decisions.md — the RFC/design decision index A single E3 entry point cataloguing the design rationale: the 7 eidos RFCs (color model/engine, typography/depth/shape/structure engines, scaling), the 3 arts subsystem design docs (connection/timer/session), and the cross-cutting decision logs (LIBRO_VARIACIONES, GESTURES). Each entry gives status + the one decision it records, linking the document for the full argument. This delivers the "naming único" goal at the index level. The physical file rename (*_RFC -> rfc-*, DESIGN_* -> design-*) is deferred: those names are cited as provenance anchors in ~30 source files (eidos/lib/*.ts, arts/timer/*, arts/color/*, tests), so a rename only pays off if every citation is swept in the same pass. The index gives consistent naming without that churn. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 4 (2/3) — índice de decisiones**: nuevo `docs/decisions.md` — entrada única E3 que cataloga los 7 RFC de eidos (color-model/engine · typography/depth/shape/structure engines · scaling) + 3 design docs de arts (connection/timer/session) + decision-logs (LIBRO_VARIACIONES, GESTURES), cada uno con estado + la decisión que registra. El **rename físico** de los ficheros se DIFIRIÓ (ver Diferido): los nombres `*_RFC`/`DESIGN_*` están citados como provenance en ~30 archivos de código; el índice da el naming consistente sin tocar las citas.
docs(corpus): finish THEMING stub-split — guides (E4) + comparison/FAQ (E3) Completes the renumber-safe split of THEMING.md. - §8 "añadir componente" + §9 "definir theme" -> src/uix/eidos/THEMING_GUIDE.md (E4 authoring guides). - §15 "comparación con librerías" + §17 "FAQ — decisiones polémicas" -> src/uix/eidos/THEMING_NOTES.md (E3 comparison + rationale). THEMING.md keeps numbered pointer-stubs at §8/§9/§15/§17; §16 anti-patterns, the ## Referencias section, and the dated §20-34 changelog stay in place. All 34 sections and every "THEMING §N" citation across the corpus + code survive. THEMING 2571 -> 1930 lines across the three commits; it is now a leaner E1 layer reference with TSC (E2), the guides (E4) and the notes (E3) as siblings. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 4 (3/3) — THEMING split (stubs)**: `THEMING.md` 2571→1930 L, partido renumber-safe en 3 docs-estrato: `eidos/TSC.md` (E2, §7+§18), `eidos/THEMING_GUIDE.md` (E4, §8+§9), `eidos/THEMING_NOTES.md` (E3, §15+§17). THEMING conserva stubs-puntero numerados → las 34 secciones y todas las citas `§N` del corpus/código sobreviven; §16 + `## Referencias` + §20-34 quedan in-place. Además §14 motion saneado (contradicción con eidos-motion.md). **Fase 4 cerrada.**
docs(corpus): add docs/README.md — the corpus entry point (E0), wire it from CLAUDE.md A single front door to the whole documentation set, for new sessions and agents to orient before doing anything: the 60-second framework summary, the strata (E0 orientation / E1 architecture / E2 canon / E3 decisions / E4 guides / E5 module reference / process), a reading order, a per-stratum map of every doc, and task-oriented shortcuts ("I want to build a component / theme it / know why a decision was made…"). CLAUDE.md gets a "Start here" pointer at the top of Reference Documents so the entry point is actually reached on session start (this is the small additive pointer, not the deferred CLAUDE.md slimming). All links verified. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (1/n) — entrada E0**: nuevo `docs/README.md` — puerta única del corpus (lo pidió el usuario: "necesario para que los agentes empiecen y para cada sesión nueva"). Contiene: orientación de 60 seg, los 7 estratos, orden de lectura, mapa de docs por estrato (E1-E5 + process) y atajos task-oriented ("I want to…"). **Cableado desde `CLAUDE.md`** ("Start here", arriba de Reference Documents) para que toda sesión/agente lo lea primero. Todos los enlaces verificados. Pendiente del E0: solo adelgazar `CLAUDE.md` (ver Diferido).
docs(corpus): add docs/glossary.md — the invented vocabulary, one line each Companion to the entry point: an agent or new session can look up the framework's coined terms (morfo, soma, sema, eidos, archetype, runtime part, Presence, polymorphic close, hold, cascade, TSC, recipe, variant, role, scaling, …) in one place, each with a pointer to its authoritative doc. The semantic subset (family / intent / verb / channel) points at CANON.md instead of restating the values, so it cannot drift. Wired into docs/README.md (E0 stratum + reading order). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (2/n) — glosario**: nuevo `docs/glossary.md` — el vocabulario inventado (las capas, morfo, soma, sema, eidos) definido en una línea cada término + puntero a la fuente autoritativa; el subset semántico (family/intent/verb/channel) **apunta a CANON, no se recopia** (sin drift). Enganchado en `docs/README.md` (estrato E0 + orden de lectura).
docs(corpus): add docs/comparison.md — honest framework-level positioning Where UIX sits relative to the two families (headless behavior: Radix/Ark/bits/ React Aria; styled systems: Mantine/Chakra/Radix Themes/shadcn), grounded in UIX's own documented design choices rather than claims about competitors' internals: - morfo as a single declarative contract (compile-time drift), - a perception layer (sema) neither family has, - two-moment motion, theme-as-retint, validated token scope, graceful degradation, - and the honest trade-offs (more to learn, smaller ecosystem, sema only pays off if used). The one competitor-specific claim (Chakra collapses presence onto one axis) is sourced in eidos-motion.md. Per-component prop-parity comparisons stay in each component README, by doctrine. Wired into docs/README.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (6/n) — comparativa**: nuevo `docs/comparison.md` — posicionamiento honesto a nivel framework, anclado en los diferenciadores que el corpus YA documenta (morfo = contrato declarativo · sema = capa de percepción · motion de dos momentos · theme = retintar lo fijo · TSC · degradación headless/SSR) + los trade-offs reales (más que aprender · ecosistema menor · sema solo paga si la usas). NO inventa claims sobre competidores; la única concreta (Chakra colapsa presence en un eje) está sourced en eidos-motion.md. La comparativa per-componente sigue en los READMEs (doctrina). Enganchado en `docs/README.md`.
docs(corpus): add docs/testing-and-tooling.md — testing, validation, codegen, SSR Consolidates the cross-cutting verification story that was scattered across package.json, COMPONENT_GUIDE and SOMA_ARCHITECTURE into one reference: - The two-project vitest suite (browser client / node server) and how to run one. - Each validator and the class of bug it catches: morfo:check (DOM vs contract), morfo:vocabulary (verb drift), component:audit (acceptance), perm:check (state transitions), smoke (hydration), translations:check, eidos-lint. - Codegen vs authored: generate:eidos-css, generate:contracts-docs, and the compileMorfo primitive. - The SSR posture: dom:false -> disabledDom, ActiveDom owner-document resolution, ornamental sema, and why smoke (not HTTP 200) is what catches hydration bugs. All grounded in the real package.json scripts. Wired into docs/README.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (5/n) — testing/tooling/codegen**: nuevo `docs/testing-and-tooling.md` — consolida la historia transversal de verificación (estaba dispersa en package.json / COMPONENT_GUIDE / SOMA_ARCHITECTURE): suite vitest de **2 proyectos** (client browser / server node), **validadores** (`morfo:check` · `morfo:vocabulary` · `component:audit` · `perm:check` · `smoke` · `translations:check` · eidos-lint — qué caza cada uno), **codegen** (`generate:eidos-css` · `generate:contracts-docs` + `compileMorfo`: generado vs autorado), y **postura SSR** (`dom:false`→disabledDom, ActiveDom owner-doc/iframe-safe, sema ornamental, hidratación = lo que `smoke` caza). Anclado en scripts reales de package.json. Enganchado en `docs/README.md` ("I want to… test").
docs(corpus): add docs/getting-started.md — clone to first change The narrative onboarding path that complements the map (docs/README.md) and the build guide (COMPONENT_GUIDE): run it, the mental model, and your first change, in order. Grounded in the real routes (/uix/components/*, /active, /temas) and the real scripts (check / test / morfo:check / smoke / component:audit / perm:check), it points at COMPONENT_GUIDE / THEMING_GUIDE rather than duplicating them. The centerpiece is "see the whole architecture in one element": inspect a toggle and watch morfo's data-toggle, soma's data-state, sema's data-event-* and eidos's CSS all on one node — the doctrine made visible. Wired into docs/README.md (top callout + first "I want to…" row). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (4/n) — getting-started**: nuevo `docs/getting-started.md` — el camino narrativo de cero a primer cambio: run it → modelo mental con la cadena de transcripción → **ver las 4 capas en un solo elemento del DOM** (toggle: `data-toggle`/`data-state`/`data-event-*` + el CSS de eidos) → loop de verificación con scripts reales → 3 caminos de primer cambio. Anclado en rutas reales (`/uix/components/*`, `/active`, `/temas`) y scripts reales (`check`/`test`/`morfo:check`/`smoke`/`component:audit`/`perm:check`); apunta a COMPONENT_GUIDE/THEMING_GUIDE sin duplicar. Enganchado en `docs/README.md` (callout arriba + fila "I want to"). Hay además un get-started de la app en `/active/get-started/*` (incl. ai-agents) — este es el doc-side.
docs(corpus): add docs/authoring.md — the rules for writing docs here So a session or agent that creates or edits documentation does it consistently with the corpus, not against it. Codifies what the migration learned, each rule tied to the failure it fixes: 1. Link the canon, never copy it (the anti-drift law — "7 families" survived in three docs because they re-transcribed instead of linking). 2. Every doc belongs to one stratum (E0-E5 / process); where each kind lives. 3. Reference docs are timeless — hand-offs, dated status and hardcoded catalogs go to process/ or become pointers. 4. One source per concern; two docs on a subject get distinct stated roles. 5. Frontmatter convention. 6. Sections cited by §N are load-bearing — stub-split, never silently renumber; verify moved links resolve. 7. English target; code comments always English. 8. Naming (rfc-* / design-*), markdown relative links, pre-commit checklist. Wired into docs/README.md ("I want to… write a doc" + a note after the strata). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Fase 5 (3/n) — reglas de autoría de doc**: nuevo `docs/authoring.md` — cómo se crea/edita documentación en el corpus, codificando lo que aprendió la migración: (1) linkear el canon, nunca copiarlo (la ley anti-drift); (2) cada doc tiene un estrato + dónde va; (3) docs de referencia atemporales (hand-offs/fechas/catálogos hardcoded → process o puntero); (4) una fuente por concern (roles distintos si hay 2 docs del mismo tema); (5) frontmatter; (6) ediciones renumber-safe (stub pattern para `§N` citadas, verificar links); (7) idioma EN; (8) naming `rfc-*`/`design-*`; + checklist pre-commit. Cada regla cita el fallo real que arregla. Enganchado en `docs/README.md` ("I want to… write a doc" + nota tras los estratos).
docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
## PENDIENTE
docs(corpus): finish THEMING stub-split — guides (E4) + comparison/FAQ (E3) Completes the renumber-safe split of THEMING.md. - §8 "añadir componente" + §9 "definir theme" -> src/uix/eidos/THEMING_GUIDE.md (E4 authoring guides). - §15 "comparación con librerías" + §17 "FAQ — decisiones polémicas" -> src/uix/eidos/THEMING_NOTES.md (E3 comparison + rationale). THEMING.md keeps numbered pointer-stubs at §8/§9/§15/§17; §16 anti-patterns, the ## Referencias section, and the dated §20-34 changelog stay in place. All 34 sections and every "THEMING §N" citation across the corpus + code survive. THEMING 2571 -> 1930 lines across the three commits; it is now a leaner E1 layer reference with TSC (E2), the guides (E4) and the notes (E3) as siblings. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
### Fase 4 — COMPLETA (resumen en HECHO; detalle de THEMING ↓ para el registro)
docs(corpus): THEMING §14 — resolve the motion contradiction, point at eidos-motion.md THEMING §14 described motion as "deferred / data-motion-ref does not exist / superseded by the TSC event:* scope". eidos-motion.md says the opposite: the two-moment model is implemented (F1-F7, 2026-06-04), the engine was relocated to arts/motion as a service, and it is the event:* scope that is now obsolete. §14 was a stale snapshot contradicting the canonical motion doc. Rewrite §14 as a lean pointer to eidos-motion.md (canonical, current) with a status note flagging the obsolete data-motion-ref / event:* framing, and keep the theming-specific bit (the --motion-scale-lift "pickup" token). Renumber-safe — §14 stays §14, so the §N citations across the corpus are untouched. The full stratum-split of THEMING (TSC -> E2, guides -> E4, comparison/FAQ -> E3) is deferred: it renumbers sections that are cited by number across the corpus + code, so it needs a dedicated pass with a citation sweep. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Partir `THEMING.md`** (2571 L, **34 secciones**, multi-estrato): TSC (§7, §18) → canon visual E2; "añadir componente" (§8)/"definir theme" (§9) → E4 guía; comparación (§15)/FAQ (§17) → E3. El resto (§1-6 mental model/capas/roles/sizes/naming, §10-13 runtime/bundle/validación/sema) = E1 referencia de capa.
- **YA hecho** (commit de §14): §14 motion saneado — tenía una **contradicción** con `eidos-motion.md` (THEMING decía "motion deferred / data-motion-ref no existe / superseded por TSC event:*"; eidos-motion.md dice F1-F7 implementado + motor en `arts/motion` + el event:* scope es el obsoleto). Reescrito como puntero a `eidos-motion.md` (canónico) + conservado el token theming `--motion-scale-lift`.
docs(corpus): extract TSC from THEMING into eidos/TSC.md (stub-split) First piece of the THEMING split, using the renumber-safe stub approach so the many "THEMING §N" citations across the corpus + code keep resolving. §7 (Token Scope Contract) and §18 (universal coverage, v2.2) move to a new src/uix/eidos/TSC.md — the eidos visual canon (E2), with CANON.md-style frontmatter. THEMING.md keeps numbered pointer-stubs at §7/§18, so section numbers (and therefore §23/§25/§26/§27/§28 citations) are untouched. 2550 -> 2271 lines; 34 headers intact, TOC anchors still resolve. Remaining split pieces (own commits): guides §8/§9 -> E4, comparison §15 + FAQ §17 -> E3. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **RIESGO** (= el del rename): las secciones están citadas por `§N` a través del corpus (`CLAUDE.md` §23/25/26/27/28, RFCs §25, `arts/color/README` §26, `THEMING_AUDIT` §23) + las §25/29/30/31 solapan los RFCs de color/depth/shape/structure (ya en `docs/decisions.md`). Renumerar rompe esas citas.
- **Enfoque elegido (usuario): split con stubs** — renumber-safe, sin sweep. Extraer cada bloque a su estrato y dejar un stub-puntero numerado en THEMING manteniendo el número de sección → las citas `§N` sobreviven.
- **HECHO — TSC**: §7 (Token Scope Contract) + §18 (cobertura universal v2.2) → `src/uix/eidos/TSC.md` (canon visual E2, frontmatter como CANON.md). En THEMING quedan stubs §7/§18 apuntando a TSC.md. 2550→2271 L. Verificado: 34 headers intactos, §8/§19 limpios, anchors del TOC OK, §23/25/26/27/28 sin tocar.
docs(corpus): finish THEMING stub-split — guides (E4) + comparison/FAQ (E3) Completes the renumber-safe split of THEMING.md. - §8 "añadir componente" + §9 "definir theme" -> src/uix/eidos/THEMING_GUIDE.md (E4 authoring guides). - §15 "comparación con librerías" + §17 "FAQ — decisiones polémicas" -> src/uix/eidos/THEMING_NOTES.md (E3 comparison + rationale). THEMING.md keeps numbered pointer-stubs at §8/§9/§15/§17; §16 anti-patterns, the ## Referencias section, and the dated §20-34 changelog stay in place. All 34 sections and every "THEMING §N" citation across the corpus + code survive. THEMING 2571 -> 1930 lines across the three commits; it is now a leaner E1 layer reference with TSC (E2), the guides (E4) and the notes (E3) as siblings. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **HECHO — guías + notas**: §8+§9 → `src/uix/eidos/THEMING_GUIDE.md` (E4: añadir componente + definir theme); §15 (comparación) + §17 (FAQ) → `src/uix/eidos/THEMING_NOTES.md` (E3). Stubs-puntero en THEMING; §16 anti-patterns y `## Referencias` conservados in-place; §20-34 (datados, changelog) in-place (citados). **THEMING 2571→1930 L; split completo; 34 secciones + citas `§N` intactas.**
- **NOTA harness**: escribir un archivo EXISTENTE con PowerShell `Set-Content` o con regex (`\d+`, ` / ` sueltos) en el comando lo bloquea un analizador (lo lee como `Remove-Item path`). Funciona: `[System.IO.File]::WriteAllLines(abs, $arr, utf8NoBom)` + límites por `.StartsWith()` (sin regex). Crear archivos NUEVOS con `Set-Content` sí va. Las tools Edit/Write también.
docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
### Fase 5 — huecos de libro (escribir nuevo)
docs(corpus): add docs/comparison.md — honest framework-level positioning Where UIX sits relative to the two families (headless behavior: Radix/Ark/bits/ React Aria; styled systems: Mantine/Chakra/Radix Themes/shadcn), grounded in UIX's own documented design choices rather than claims about competitors' internals: - morfo as a single declarative contract (compile-time drift), - a perception layer (sema) neither family has, - two-moment motion, theme-as-retint, validated token scope, graceful degradation, - and the honest trade-offs (more to learn, smaller ecosystem, sema only pays off if used). The one competitor-specific claim (Chakra collapses presence onto one axis) is sourced in eidos-motion.md. Per-component prop-parity comparisons stay in each component README, by doctrine. Wired into docs/README.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
~~Entrada E0~~ (hecho) · ~~Glosario~~ (hecho) · ~~reglas de autoría~~ (hecho) · ~~arco *getting-started*~~ (hecho) · decision-log consolidado (semilla: `src/docs/LIBRO_VARIACIONES_Y_EXTENSIONES.md`) · walkthrough "construye tu propia capa" · ~~comparativa honesta vs Radix/Ark/Mantine~~ (hecho) · ~~historia transversal SSR/testing/codegen~~ (hecho).
**Restan de Fase 5** (los más opinables/arriesgados): **decision-log consolidado** (riesgo: duplicar `LIBRO_VARIACIONES`, que YA es el decision-log impl-vs-libro; lo no-consolidado son los ~10 "Session hand-off" de CLAUDE.md → encaja mejor con el ítem Diferido "adelgazar CLAUDE.md") · **walkthrough "construye tu propia capa"** (especulativo — requiere validar que el patrón de capa se extiende limpio a una 5ª; mejor hacerlo deliberado, no de relleno).
docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
### Diferido (decisión del usuario)
docs(corpus): add docs/decisions.md — the RFC/design decision index A single E3 entry point cataloguing the design rationale: the 7 eidos RFCs (color model/engine, typography/depth/shape/structure engines, scaling), the 3 arts subsystem design docs (connection/timer/session), and the cross-cutting decision logs (LIBRO_VARIACIONES, GESTURES). Each entry gives status + the one decision it records, linking the document for the full argument. This delivers the "naming único" goal at the index level. The physical file rename (*_RFC -> rfc-*, DESIGN_* -> design-*) is deferred: those names are cited as provenance anchors in ~30 source files (eidos/lib/*.ts, arts/timer/*, arts/color/*, tests), so a rename only pays off if every citation is swept in the same pass. The index gives consistent naming without that churn. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Rename físico de RFCs/design** (`*_ENGINE_RFC`/`*_RFC` → `rfc-*`; `DESIGN_CONN`/`DESIGN`/`DESIGN_TIMR` → `design-*`): los nombres actuales están citados como provenance en ~30 archivos de código (`eidos/lib/*.ts` ×~20, `arts/timer/*`, `arts/color/*`, `arts/session/types.ts`, tests) + ~10 docs + CLAUDE.md. `docs/decisions.md` ya da el naming consistente a nivel índice sin tocar nada. El rename físico solo vale la pena si se barren TODAS las citas en el mismo pass (si no, drift) — decisión del usuario por el coste/beneficio (cosmético vs ~40 archivos + `check`). De paso: ruta mala `soma/layers/GESTURES.md` (real: `layers/gesture/GESTURES.md`) en SOMA_ARCHITECTURE/old-README.
docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
- **Poda física** de las copias de vocabulario (Fase 2): hoy tienen puntero de autoridad pero conservan la lista; reemplazar por el puntero. Opcional.
- **Commitear el libro v2.3** (hoy untracked → el enlace de `CANON.md` no resuelve en remoto). Decisión del usuario.
- **Migración de idioma** coordinada (docs de capa es→en).
- **`DATA_ATTRS.md`** (fósil Terra, 2158 L, hoy LEGACY): mover a process o regenerar desde los morfos.
docs(corpus): add docs/README.md — the corpus entry point (E0), wire it from CLAUDE.md A single front door to the whole documentation set, for new sessions and agents to orient before doing anything: the 60-second framework summary, the strata (E0 orientation / E1 architecture / E2 canon / E3 decisions / E4 guides / E5 module reference / process), a reading order, a per-stratum map of every doc, and task-oriented shortcuts ("I want to build a component / theme it / know why a decision was made…"). CLAUDE.md gets a "Start here" pointer at the top of Reference Documents so the entry point is actually reached on session start (this is the small additive pointer, not the deferred CLAUDE.md slimming). All links verified. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 months ago
- **Adelgazar `CLAUDE.md`**: mover sus ~10 "Session hand-off" embebidos a process/ — su propio paso, archivo sensible. (El índice E0 ya existe: `docs/README.md`, cableado desde CLAUDE.md.)
docs(process): add CONTINUE hand-off for the doc-corpus migration Self-contained hand-off to resume the documentation-corpus reorganization in a fresh session: objective, work rules (no agents, Castilian, never touch words/palabras, verify-staged), agreed decisions (EN target, hybrid layout, docs/process for the ephemeral, the E0–E5 strata, canon links code for values), what's done (Phases 1–2 + Phase 3 2/3, with commit hashes), and what's pending (soma de-dup — heavy, do with fresh context; the 3 arts hand-offs from 1b debt; Phases 4–5; deferred items). Lives in docs/process/ (a hand-off is process, not a CONTINUE-*.md loose at the repo root — which is what we've been removing). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
4 months ago
## Referencias
- Mapa de trabajo (scratch): `G:\tmp\docs-corpus-map.md`.
- Canon: `docs/CANON.md`. Proceso: `docs/process/`. Arquitectura: `src/uix/active_architecture.md`.

Powered by TurnKey Linux.