docs(blog): add bilingual TypeScript 2.0 history and core statistics

[loopx-agent]

Goal And Result

Add a complete English/Chinese engineering retrospective for the LoopX2.0 TypeScript control-plane refactor. It follows the August15 RFC, Effect Program cutover, complete-transaction pivot, overlapping authority/store and SQLite work, Song's Semantics contributions, real caller counterexamples and October retirement. The final cleanup PR is presented as one closing slice.

The article includes whole-period statistics restricted to loopx/control_plane/**:1,144 mainline landings,1,121 associated merged PRs, source inventory/churn, weekly cadence and collaboration measures. Nine shared SVGs and downloadable public-safe CSV/JSON inventories make the evidence inspectable. Two locale indexes link the article after the existing first entry. Both articles remain fully static: a desktop contents list and native mobile details need no initialization script. Source pins and limitations distinguish engineering activity from migration completion, productivity or benchmark gains.

Author And Accepted Scope

Written by: model_agent — OpenAI Codex, GPT-6 family.

Specification: maintainer-requested bilingual2.0 engineering retrospective, using the TypeScript migration and Semantics RFCs and merged history pinned to 488e4438e0e51eca023ac3f5e690603120851b0c.

Criterion	Disposition	Evidence
Explain the full migration history and design changes	implemented	Both twelve-section articles link original RFC/PR evidence.
Attribute new maintainer Song's concrete Semantics work	implemented	Introduction, technical section and contributor table link his contributions.
Include reproducible core-directory statistics	implemented	Fixed range, methods, caveats, weekly charts and deduplicated public inventories.
Integrate both languages into the existing editorial site	implemented	Canonicals, hreflang, language navigation, shared article-local assets and locale index entries.

Complete within this publication-proposal scope. The copy is written for the2.0 release; actual merge, deployment and release timing remain maintainer decisions. No runtime, schema, scoring, permissions, scheduler or provider behavior changes. A related refactor was unnecessary; existing editorial styles are reused and additional layout is local to this article.

Validation

Tested revision: 1613d9e848d771f3f1709335edb4b0f38d79f4e2. Local run state: finished. Remote Pages remains blocked as documented below. Inputs: public repository history and public-fixture editorial data.

Check	Result	Evidence / limitation
Full exported frontstage bundle	passed	npm run smoke:frontstage-share-bundle --prefix apps/presentation/dashboard; actual site/dashboard export, static bilingual pages, hosting-base links, public boundary and case restoration.
Production build	passed	npm run build --prefix apps/presentation/site (TypeScript and Vite client/SSR build).
SEO publication	passed	npm run smoke:seo --prefix apps/presentation/site;24 canonical sitemap targets, metadata and assets.
Static article/link checks	passed	One H1,12 sections and6 figures per language; local article/CSS/image/download/language targets exist in the build; approved prose unchanged except asset relocation. Existing MkDocs navigation is assembled separately by Pages.
Statistics reconciliation	passed	1,121 unique PR rows,1,144 mainline rows; review/comment sums and previously validated endpoint/gross-net inventories agree.
Public/private and whitespace	passed	All20 candidate paths checked; no credentials, local paths, private state or raw runtime traces; git diff --check passes. CSV records and SVG geometry/text preserved after whitespace normalization.
Responsive content preview	passed before integration	Both languages were inspected at1440px and390px; charts loaded, no page-wide overflow.
Final integrated browser readback	passed	English/Chinese at1440px and390px render with JavaScript disabled and zero article scripts; all6 figures load, no page-wide overflow; native mobile details toggle and language switching work.
Remote Pages	blocked by baseline documentation links	Run38931020440 passes the bilingual catalog, public-safe bundle and artifact export. Strict MkDocs then fails on the same two unchanged recovery-RFC links reproduced at both base and head; see attribution below.
Runtime / benchmark suites	not applicable	Static editorial additions only; no capability or performance claim is newly qualified.

Remote CI failure attribution

At the tested head, the article export checks pass. The later strict MkDocs build aborts on the English and Chinese complete-state-recovery-v0 RFC links to loopx/control_plane/goals/source_session_registry_state.py. The Python source exists; the relative links point outside the MkDocs documentation inputs.

A controlled mkdocs build --strict using the same dependency environment exits1 with the identical two warnings at base 88bac4bde2602f5bd737ef6fb22cd672901b47c2 and head 1613d9e848d771f3f1709335edb4b0f38d79f4e2. Both RFCs, the target source, MkDocs configuration and dependency declarations have identical Git blobs in that range. The main-branch run38943059450 independently fails on those same links. This attributes the observed docs failure to the baseline; it does not mark Pages green or bypass validation. This PR remains a static editorial proposal, with merge/deployment held for the ordinary checks and review.

Visual Evidence

UI impact: changed — new editorial pages; shared chrome unchanged. Before: N/A for the new article. The locale-index heading, routes and first listing remain byte-for-byte unchanged. After: English article, 中文全文.

The owner reviewed the full current content preview before author submission. One title anchors the viewport; source-bound tables and diagrams earn their space by explaining implementation choices. Narrow screens have scrollable tables/figures and a collapsed contents list. Screenshots below contain only public editorial content, with no browser chrome or account state.

Built editorial views

Desktop, English:

Mobile, Chinese:

Core statistics, Chinese:

Boundaries

• Public-safe editorial files and derived public statistics only; no private operational state or raw traces.
• No benchmark execution, scoring or runtime change.
• Scope is limited to this bilingual article, its index entries and shared assets.
• Author/committer identity is LoopX Agent; every commit includes DCO sign-off.
• Merge and publication remain separate from this PR.

[loopx-agent]

Request changes conclusion (author-owned PR; GitHub blocks formal self-review)

REQUEST_CHANGES — 精确头 1613d9e848d771f3f1709335edb4b0f38d79f4e2 的完整 20 路径、+4488/-0 已独立审完。正文、统计和静态交互大体成立，阻塞点是一处同步存在于双语开头的发布事实错误。比较基线 88bac4bde2602f5bd737ef6fb22cd672901b47c2；不要求重写这篇文章。

动机

希望理解 LoopX 控制面重构的读者，需要可复查的完整历史、贡献归属和代码统计。

此前 blog 没有这份双语全周期复盘；本头提供十二节文章、来源链接与公开清单，但页面开头同时把尚未确认发布的 2.0 候选写成已经发布。

正文、统计和静态阅读路径已独立核验；发布状态表述尚需与真实发布事实对齐。

不改变 runtime、评分、权限、调度或迁移验收，也不把账号数、代码量和 review 数当生产率。

先修正两种语言的已发布断言，或在实际 2.0 发布得到核验后同步上线；其余正文与统计不需重写。

接受前的规格是 docs/development/design.md（88bac4bde2602f5bd737ef6fb22cd672901b47c2）。Product Character 要求公开界面不夸大进度，本头 not_met；Accessibility 和 Responsive Behavior 已通过真实静态阅读验证；Source Of Truth 通过原有 chrome/首条 listing、公开边界和不可变 Git 来源核对。另读了 TS migration RFC 的候选/发布界限，未拿这篇新文章作为其自身正确的规格。

改动思路

文章和下载清单复用现有静态 editorial 路径及共享样式，附加 CSS 只作用于本文章，不需要新 runtime 或状态 owner。

保留完整双语复盘、公开统计与原生目录交互；当前最小修复只处理两处发布事实文案，并核对实际导出页面。

历史需要按不同 owner 和阶段解释，不能把最后一次删除当成全部迁移，也不能把 38% 的 TS 物理行占比说成迁移完成度。本文保留了这些限制，说明 Effect Program、完整事务、共享权威、SQLite retained-history 成本、Semantics 与真实 producer/recovery 的关系；贡献表按具体工作链接社区作者。反对引入这个改动的最强理由是大篇幅和大量派生数据增加维护成本；这里完整下载清单、明确起止与来源链接使这一成本可复查，没必要加 runtime 或新的通用图表框架。

具体改动

两份静态文章均有一条 H1、十二节和六幅实际加载的图。目录使用桌面原生 anchors、移动端 details/summary，不需要初始化脚本。独立生产 build、SEO（24 canonical targets）、完整 frontstage-share-bundle smoke 全通过；实际构建产物在 EN/ZH、1440/390px 下 scripts=0，没有整页横向溢出。已用键盘展开目录、进入统计章节并切换语言；表格580px内容限定在350px移动容器中。

统计没有只检查总数：我从 d1fe05932ac28690d393a38c78b4738ac12c78c0 到 488e4438e0e51eca023ac3f5e690603120851b0c 独立重建了全部1,144条first-parent提交及逐条Python/TS增删，核对1,121个唯一PR及其merge commit位于主干、周分桶、作者账号和review/comment快照求和。端点源码为Python80,473→142,371行，TS0→87,268行，累计Python+120,899/-59,001、TS+104,769/-17,501；这些与JSON/CSV/正文一致。API快照求和不是当前review状态或生产率证明。关键历史的原PR（包括3226、3416、3464、4433、6396）也已直接读回；旧checkpoint成本描述没有冒称当前性能。

[P2，需修复] 发布状态必须与公开事实同步。 English opening 写着 “LoopX 2.0 is out.”；中文开头 同样宣称已经发布。当前 正式版本仍是1.4.2，本头包版本是2.0.0a1；当前100条公开release记录没有2.0 release。build.mjs 直接复制并收录这些HTML，既有Pages workflow在main相关改动上部署；PR正文“面向将来发布”的说明不会替换网页中的断言。最小修复就是改为候选版/重构阶段表述，或与核验后的实际发布同步上线。保持其余文章、统计和归属。

对主干的风险

全20路径是静态文章、局部CSS、SVG/PNG和公开派生清单；没有运行时状态、schema、权限、评分、调度或provider变更。既有blog头部、原第一条listing及导出机制保持不变；新第二条入口与文章是有意新增。全部候选路径的凭据、机器绝对路径、私人链接扫描无命中，SVG无脚本/foreignObject/onload。图片与大量CSV行是可复查证据，不计为产品能力。

严格MkDocs没有被称为通过：在同一实际依赖环境运行base/head，均exit1，逐项相同的两个complete-state-recovery双语RFC跨docs源码链接warning。两RFC、确实存在的Python目标、MkDocs配置/依赖和Pages workflow均逐blob相同；这归于原基线路径，而不是只凭失败数相等。独立文章生产build/导出/浏览器证据通过，不能代替整个Pages部署资格。当前 scoped wait_for_ci=false，本轮没有查询、等待或重跑CI；作者及其他线程转述的CI信息没有被当作本轮核验。

我的整体评价

文章和下载清单复用现有静态 editorial 路径及共享样式，附加 CSS 只作用于本文章，不需要新 runtime 或状态 owner。

保留完整双语复盘、公开统计与原生目录交互；当前最小修复只处理两处发布事实文案，并核对实际导出页面。

代码与展示层面的完整性足以保留，future-facing pass 在这个局部静态边界无需新增抽象。当前结论仅因双语发布状态而请求修改；它不是对原迁移实现、社区贡献或统计方法的否定，也不要求先补齐所有RFC持续运行资格。修改后应重新核对精确头、两份导出开头与实际发布事实；没有执行合并、发布或旧review处置。

English verdict: REQUEST_CHANGES - The bilingual retrospective, fixed-source statistics and script-free reading journey are substantiated, but both exported openings prematurely claim that2.0 is released. Correct the candidate/release wording or synchronize publication with a verified release; the unrelated baseline MkDocs failures remain separate.

Reviewer: model_agent; model=gpt-6.1-sol; provider=OpenAI; reasoning_effort=max; declaration_source=runtime_reported.