Skip to content

This is the multi-page printable view of this section. .

Return to the regular view of this page.

Design research

Dated experiments and consumer evidence that inform OINK design decisions.
Dated research records

Research records capture measurements, inputs, and tool versions from specific studies. Current behavior is defined by the Design contracts and validated on Hugo Extended 0.168.0 per the toolchain contract.

Publish research in the Design tree only when another maintainer can inspect its method, understand its scope, and repeat the applicable check. Keep raw agent transcripts, temporary build logs, and local absolute paths out of research records.

Research map

Record Evidence
Goldmark block attributes Render-hook visibility and CommonMark container limits on the historical Hugo versions recorded in the study
Consumer and migration evidence A dated corpus survey plus deterministic Book migration results
Comprehensive review, 2026-08-26 Implementation, configuration, output, security, test, performance, and doc audit
Community issue and PR review, 2026-09-19 Reproductions, PR acceptance advice, and remedies for sidebar, focus, and search feedback
OINK 1.1 release review, 2026-09-20 Five runtime repairs, documentation readiness, validation evidence and publication boundaries
CLI acceptance snapshot, 2026-09-29 Executed Starter, real-site, offline, upgrade, and reproducible-archive checks for the local CLI candidate
Visual preset acceptance, 2026-10-05 Paper/Slate local implementation, actual output and bounded browser evidence
Ink and Terminal experiment, 2026-10-05 Explicit experimental presets, design tradeoffs and real-site verification
OINK 1.2 pre-release review, 2026-10-05 Final local candidate checks, cleanup, local resources, compatibility and publication boundaries
CLI maintenance acceptance, 2026-10-03 Historical R1–R8 and A18 source-bound runtime and archive evidence
Theme review, 2026-10-13 Reproduced output and interaction defects, repairs, and local validation scope
Component refinement, 2026-10-25 Final component acceptance, 409 Chromium tests, targeted three-engine visual review, CSS/JS size comparison, and separate gallery baseline failures
Document comparison, 2026-10-29 Dated source review behind the translation comparison decision; historical, not the released contract
Review repairs, 2026-11-06 Reproduced v1.5.0 defects, bounded development repairs, validation and remaining work
OINK 1.5.1 readiness, 2026-11-06 Three repaired candidate regressions, owning checks, browser acceptance, publication evidence and release gates

Publication rules

A research record states its date, inputs, applicable versions, method, result, and known limits, with changing counts labeled as snapshots. External framework comparisons should be refreshed from primary sources before publication and focus on conclusions that affect OINK.

When a result becomes an accepted product choice, link it from an accepted decision. If it proposes behavior that does not exist, move the design question to Proposals.

1 - Goldmark block-attribute evidence

Repeatable probes of lists, images, tables, passthrough blocks, fences, callouts, and nested containers on Hugo 0.160.1 and 0.164.0.
Verified snapshot

These probes produced identical relevant output bytes on Hugo Extended 0.160.1 and 0.164.0. They explain OINK’s native component forms. The current component contract defines required behavior.

Method

The probe used a minimal Hugo site without OINK templates, with render hooks printing their context fields and .Attributes as visible markers. The site enabled Goldmark block attributes and passthrough delimiters for inline and block math, allowed unsafe rendering to inspect raw HTML, and set wrapStandAloneImageWithinParagraph: false.

Each source shape was rendered with the minimum supported Hugo and the then current Hugo version, comparing relevant output byte for byte to record Goldmark’s structural behavior.

Findings

Source shape Hook result Design consequence
Ordered list with paragraphs, fences, callouts, nested lists, and {.steps} The class attaches to the outer <ol> and rich list-item blocks survive A Markdown list is the native Steps form
Heading inside a list item The heading remains inside <li> and enters .TableOfContents Native Steps can carry navigable headings
Nested list with {.filetree} The class attaches to the outer <ul> FileTree needs no outer wrapper to preserve hierarchy
Standalone image plus {#id num= caption= .class} render-image receives IsBlock=true and all attributes A Book figure can have a native image form
Inline image inside a paragraph IsBlock=false; the image receives no block attributes Inline images cannot use the block-figure contract
Block math plus {#id num=} render-passthrough receives block type and attributes A numbered equation can use the native passthrough form
Table plus {.fields #id num= caption=} render-table receives the class and named attributes Field tables, matrix markers, captions, and Book numbering can share one hook
Fenced code plus {#id num= caption=} The code-block hook receives the attributes A numbered example can be the fence itself
Callout plus {icon= tab=} The blockquote hook receives callout metadata and attributes Folding, inline title markup, icon, and tab metadata can coexist
Attribute line separated from its block by a blank line The attribute silently disappears Source checks must reject orphan attribute lines
Adjacent tables with tab= Each table hook receives its own tab label Adjacent-block tabs can extend beyond code fences

Container boundary

Hugo’s % shortcode delimiter renders .Inner as Markdown. Its template must put a blank line before and after the inner Markdown. Without both blank lines, Hugo may treat a following list as literal HTML-block content instead of Markdown.

A multi-line % container inside a CommonMark list item has another limitation: its generated HTML lacks list-content indentation, so the list closes before the container and restarts after it. OINK therefore keeps a full Steps form for steps that must contain another full container. Ordinary rich blocks, fences, and < shortcodes do not have this limitation.

In the applicable collector shape, nested % shortcodes receive inner HTML that is already rendered. If a collector requires the child’s original Markdown, it uses < delimiters and renders the captured body through the shared scoped block renderer.

Attribute ownership

Each hook enforces an explicit attribute allowlist, rejects style and inline on* handlers, and validates URL-bearing values through the shared URL policy. A hook retains a site class only where downstream CSS is an established extension mechanism.

The experiment showed that gallery images inside list items can be block images, but the image hook still does not receive their parent list’s marker. A runtime therefore relies on a theme-emitted marker or a structural fallback.

Limits and verification

These results cover Hugo 0.160.1 and 0.164.0 with the stated Goldmark settings. Before updating this snapshot for a minimum-version change, rerun the focused component, Book, table, gallery, and Markdown-output checks.

2 - OINK 1.3.2 and Hugo 0.168 consumer inventory

Local inventory of 37 maintained OINK sites using v1.3.2 and Hugo Extended 0.168.0, with visible warnings and separate publication evidence.

This dated inventory records local checkouts under ~/pgsty and ~/www. The current build policy is owned by the architecture contract and diagnostics decision.

Baseline

OINK v1.3.2 was published on 2026-10-18 at 13:37 UTC. Its tag resolves to bd2e856e6e2f7027a9189af009701f5b8cf562cf. All maintained sites use this release and Hugo Extended 0.168.0. CI, local checks, preview, and deployment commands omit the warning-fatal flag; HUGO_PANICONWARNING must also remain unset. Warnings stay visible for review. The upstream LibSass deprecation notice is accepted. No external Sass compiler is required, and older Hugo versions are no longer development targets.

The observed local tools were Hugo 0.168.0+extended+withdeploy on macOS arm64 and Go 1.27.2. Module verification disabled environment replacements and both Go and Hugo workspaces. Classic consumers were compared with the release’s runtime sources. All 37 builds preserved Paper and the explicit four-preset reader menu.

Sites

Paths are relative to the home directory. Every row uses OINK v1.3.2 and Hugo Extended 0.168.0. “Module” means the resolved public module; “copy” and “submodule” identify classic themes/oink consumers.

Checkout Theme source Local production build
pgsty/barn.pgsty.com module Passed
pgsty/boar.pgsty.com module Passed
pgsty/caps.vonng.com module Passed
pgsty/cloudexit.org module Passed
pgsty/exp.pgsty.com module Passed
pgsty/oink-starter module Passed
pgsty/oink.pgsty.com module Passed
pgsty/pgetcd.pgsty.com module Passed
pgsty/pgext module Passed
pgsty/pgredis.pgsty.com module Passed
pgsty/pgsilo.pgsty.com module Passed
pgsty/pgsty.com module Passed with local report data; see below
pgsty/pgsty.pro module Passed
pgsty/pig.pgsty.com module Passed
pgsty/pigsty.cc module Passed
pgsty/pigsty.io module Passed
pgsty/repo.pgsty.com module Passed
pgsty/silo.pgsty.com module Passed
pgsty/sow.pgsty.com module Passed
pgsty/wok.pgsty.com module Passed
www/aiempire module Passed
www/ca.vonng.com submodule Passed
www/ddia module Passed
www/dharma module Passed
www/doc.pg.center module Passed
www/fzb/site copy Passed
www/lzj module Passed
www/muse.vonng.com copy Passed
www/pg-internal module Passed
www/pg36g module Passed
www/pgweekly module Passed
www/piglet.run module Passed
www/sg.vonng.com submodule Passed
www/tpme module Passed
www/us.vonng.com submodule Passed
www/vonng.com module Passed
www/zj module Passed

Boundaries

  • pgsty.com could not fetch its separately published report manifest: the remote request returned an unexpected EOF or timed out. Its build passed with the existing local report dataset and HUGO_REPOORT_USE_LOCAL=1.
  • sow.pgsty.com was checked on its existing publication-review branch. Non-Git and source-copy consumers were validated in place.
  • pgsilo/work/silo-docs is a frozen delivery baseline, as its owning repository documents. It is excluded from maintained-site upgrades.
  • This inventory records local source, module-resolution, build, and rendered-menu verification against the published v1.3.2 tag.

3 - Ink and Terminal experiment, 2026-10-05

Explicit experimental presets in real theme output, their design tradeoffs, checks and remaining work.
Local experiment record

Ink and Terminal compile into the theme stylesheet and use the Appearance control. They remain explicitly enabled experiments, pending design acceptance.

Inputs and method

This extends the Paper/Slate implementation on the October 5 working trees. Tools: Hugo Extended 0.166.0, Go 1.27.1, Node 26.9.0, and Playwright 1.62.1 on macOS ARM64. The documentation site’s local sibling-theme build is the integration surface; its published pin remains v1.1.0.

The same Home, configuration, callout, and tab content is compared across four presets, EN/ZH, 390/1440 px, and light/dark mode. Checks inspect real rendered fonts, overflow, appearance controls, and local font requests. Further component checks cover code, parameter fields, Blog, Book, API, Mermaid, ECharts, search, and print. API vendor DOM is excluded from axe under the site’s existing policy.

Design choices

Choice Improvement Cost / limit
Ink: black/white canvas, Inter, red markers, underlined prose links, strong heading rules Clear hierarchy and link affordance with little decoration Heavier headings and repeated rules need long-page editorial review
Terminal: mono controls/headings, sans prose/tables, teal links and amber emphasis A recognizable technical interface while retaining paragraph readability Long Latin navigation labels wrap sooner; CJK uses platform fallback faces
Square Ink geometry, 2 px Terminal geometry, no component shadows Visibly different surfaces using the same content and layout Scoped component rules add CSS; no global spacing/radius token layer
Compact Terminal desktop navigation only More useful navigation rows without shrinking article text Density is fixed per preset
Existing local fonts and state handling No new font files, external font service, framework or persistence mechanism All preset CSS remains in one stylesheet
Explicit experimental menu entries Reviewers can switch immediately without changing ordinary menu choices Four cards make the enabled menu taller

Ink uses #ffffff / #0b0b0b canvases, #141414 / #ededed text, and #c8102e / #ff5c4d accent. Terminal uses #f4f5f2 / #0c0f0e canvases, #1d211f / #d3dbd6 text, #0a6560 / #4cc9bd links, and #935400 / #f0a73a accent. Site/section accent overrides still win. Terminal’s heading markers use empty accessible alternatives; unsupported engines omit them. Its hero cursor is a static shape, with no typing, blinking, scanlines, or glow.

Try it

params:
  ui:
    preset: paper
    preset_menu: [paper, slate, ink, terminal]
    dark_mode: true

The local docs site enables this list. Select Ink or Terminal in Appearance, then choose light, dark, or system independently. Following the October 5 menu revision, all four options use icon-and-name buttons without experiment badges. A site may set either as preset without enabling reader choice. preset_menu: true remains Paper/Slate plus the site default. Selecting the site’s default preset clears the saved preset. Font overrides, system typography, and pre-CSS initialization use the same contracts as Paper/Slate.

Verification

Executed check Result and scope
check-presets.py AA text/link/accent contrast on three surfaces, light/dark token parity, advisory canvas luminance, frozen Slate v1.1.0 palette; seven strict builds and 28 document roots
Theme checks Parameters, font roles, 32 catalogs with 205 keys, generated schemas, component/output contracts, runtime isolation and namespace passed; 52 existing output goldens unchanged
Runtime tests 49 Node tests passed
make check 57 non-browser tests passed; EN/ZH coverage 142/142, Markdown, rendered content and internal links checked
Standard browser suites Eight suites passed 170 tests; the appearance suite passed all 49 after fixing the default-Terminal build issue below (219 checks total across the initial run and focused rerun)
Appearance coverage Four presets × EN/ZH × 390/1440 px × light/dark on Home, configuration, callouts and tabs; local font requests and menu axe checks; state, keyboard, print and Giscus asset checks; four experiment/mode checks over 11 page types plus search, and shared Mermaid contrast
Font/configuration builds Actual docs site rebuilt with Terminal as default: system fonts with/without explicit overrides, plus explicit technical-font overrides; all three passed
Browser engines Six checks passed on Chromium, Firefox and WebKit at 390/1440 px, including pre-CSS state, keyboard selection through Ink/Terminal, persistence and focus return; desktop Chromium also used 4× CPU throttling
Visual review 96 actual-output viewport captures; representative Home, Docs and mobile menu images inspected. The local comparison gallery selects content, language, size and mode without injecting styles

The standard sitemap axe pass used 15 routes: EN/ZH Home, configuration, callouts, tabs, and OpenAPI; English search, Mermaid, ECharts, Blog, and /book/04-design/. The existing responsive axe matrix also ran. Standard browser checks used the 4173 fixture server; the engine gate used the sibling-theme dev server at port 1313. No external font request was observed in the appearance matrix.

The first default-Terminal font build found an omitted entry in the advisory canvas-luminance map, causing false accent-contrast warnings. Both experimental canvases now participate, with checker assertions and authored-accent fixtures. Only the affected appearance suite was rerun after this correction. The added research index entry and updated proposal description were reviewed before refreshing the corresponding two changes in the LLMS golden.

The experiment exposed two new styling faults: the global underline suppression hid Ink’s links, and Terminal’s selected search row retained dim summary text. Both received scoped fixes. A separate inherited Mermaid dark-label pair (#cccccc on #585858, 4.43:1) reproduced in Paper and Slate. The shared mode-only default label background is now #404040; authored Mermaid values retain priority.

Remaining work

Before stable promotion, review the look on real Windows and Android devices, including CJK fallback faces, underlines, mono heading wraps, and long parameter tables. Manual screen-reader speech and first-paint filmstrips remain unverified.

Mermaid/ECharts keep mode-only palettes, API widgets keep vendor styling, and Giscus coverage checks generated palette assets. The decision to promote Ink/Terminal or redesign chart palettes remains open.

4 - OINK 1.2 pre-release review, 2026-10-05

Local 1.2.0 candidate review, cleanup, compatibility, resource provenance and publication checks, with release boundaries.
Local candidate review

This review covers the October 5 working trees prior to tagging v1.2.0. The public theme tag and the documentation site’s consumer pin were still v1.1.0 at review time.

Scope and inputs

The theme starts at a1979a4 and the documentation site at ed2d0e3. The reviewed working trees also contain the CJK keyword-summary, literal-percent outline, and repository-source-path fixes, the site’s existing English editorial changes, and the cleanup below. The separate optional CLI is outside this theme release.

Most checks used Hugo Extended 0.166.0, Go 1.27.1, Node 26.9.0, and Playwright 1.62.1 on macOS ARM64. Official, checksum-verified Hugo Extended 0.160.1 and 0.165.0 binaries were used for selected compatibility checks.

Review findings and cleanup

Finding Correction Impact
The 1.2 release draft omitted the new default and appearance controls Update both release drafts and upgrade guidance with Paper, the Slate compatibility setting, independent persistence, current-state icons and experimental preset selection Readers can identify the visible upgrade change before adoption
Current proposal, decision, experiment and source comments still described preview letters, experimental badges, a separate Default card or an undecided release target Align current descriptions with compact icon/name buttons, site-default reset and 1.2 release preparation; preserve dated test evidence Guidance agrees with the accepted interface while keeping historical results intact
A working-tree ignore rule hid all site tests/ except two files Remove that broad rule; retain existing generated-output exclusions New regression tests remain visible to Git
Development previews do not expose production-only analytics Inspect strict production output and distinguish core local resources from explicitly configured services The local-first boundary is verified against production output

These cleanup findings required no runtime change. Earlier working-tree runtime fixes are covered by their owning checks and the final integration run.

Executed verification

The local candidate passed the following technical pre-release checks. No release-blocking theme defect was found within this scope. Counts are dated snapshots of this working-tree review.

Check Result and scope
Preset checker Passed: seven warning-strict configuration builds, 28 document roots, light/dark token parity, AA text/link/accent checks on three surfaces and the frozen Slate v1.1.0 base palette
Runtime tests 49 Node tests passed, including current-state icons, search summaries, outline tracking, clipboard and dialog focus
Theme regression and tooling checks 40 checker/tool commands passed, including 90 migration tests, snapshot/consumer safety and current output goldens; the PDF browser case was then rerun with an explicit browser, with all three isolation tests passing
Documentation checks Final make check: 57 tests passed; 143/143 bilingual files, 1,197 source headings, 228 rendered content pages and internal links checked; only the new research index entry and changed release title/description required reviewed Markdown-golden updates
Standard browser suites One complete make browser run passed all 219 checks across nine suites, including the full 372-route sitemap axe scan; no failed, flaky or skipped cases
Documentation follow-up A fresh build passed a separate axe scan of 14 updated EN/ZH routes, including this new report pair
Browser engines Six appearance checks passed on Chromium, Firefox and WebKit at 390/1440 px; pre-CSS restoration, keyboard selection, persistence and focus return; desktop Chromium also used 4× CPU throttling
Visual spot checks Current-output captures inspected for the Chinese mobile Paper menu, English desktop dark Paper menu and Chinese Terminal reading on mobile/desktop; two-column icon/name options, current-state icons and reading layout confirmed
Hugo compatibility floor 0.160.1 passed preset and reading/math checkers plus a strict minified production build of the actual documentation site
CI Hugo version 0.165.0 passed a strict minified production build of the actual site, Hugo Module/include/static/print checks, system typography, legacy Sass font overrides and expected rejection of invalid typography
Production resources 28 page visits across four presets and seven routes; core fonts and scripts served from the site’s origin; configured external services recorded separately
Production output security Passed on 921 files in the final strict production build with the documented third-party integration policy
Book publication Root and subpath EPUBs passed the theme checker and EPUBCheck 5.3.0 with zero errors/warnings; both PDFs passed the 23-page, five-chapter structure checks; the root PDF script-isolation probe passed
Published consumer pin Existing v1.1.0 resolved and passed the site’s release-pin check with environment replacements and both workspaces disabled

The full sitemap scan follows the site’s existing axe policy: OINK-maintained surfaces are checked, Giscus requests are blocked, and Swagger UI/Redoc vendor DOM is excluded. Responsive checks cover 360, 768, 820, 1024, 1200, and 1440 px in English/Chinese and light/dark mode.

Appearance checks use the same Home, configuration, callout, and tab content in four presets, both languages, 390/1440 px, and light/dark mode. They also cover search, code, tables, input/focus states, Blog, Book, API, diagrams, and print. Ink and Terminal remain explicitly selected experiments.

Publication used local Pandoc 3.11, Java 26, and Chrome headless-shell 151.0.7922.34. The first attempt with the full Chrome for Testing application timed out on this Mac. Selecting headless-shell explicitly, as CI does, produced the verified PDFs. CI pins Pandoc 3.10 and Java 21 on Linux.

Local-first resource boundary

IBM Plex Sans, Inter, IBM Plex Mono, Chakra Petch, icons, KaTeX fonts, and core browser libraries are bundled locally. Preset switching introduces no runtime font-service or CDN-script dependency. System typography and explicit font-role overrides retain their documented precedence.

The production audit loaded Home, Chinese configuration, math, Mermaid, Markmap, ECharts, and OpenAPI under each preset, then switched dark/light mode. The production base origin was preserved while built files were served locally. Request tracing recorded and blocked off-origin requests; local fonts and diagrams still loaded without uncaught JavaScript or local HTTP errors.

Two configured services requested external scripts: Giscus and Google Analytics. Giscus is an optional comments integration; its OINK palette files are local. This documentation site configures an analytics ID, so production output includes Google Tag Manager’s script. Neither service is required by the new presets. Authored remote media and explicitly selected diagram services retain their opt-in boundaries.

Repeating the checks

Run owning theme checks before the real-site checks. These commands use the sibling theme through the documented Make targets:

python3 bin/check-presets.py
node --test 'tests/js/**/*.test.js'
make -C ../oink.pgsty.com check
env -u A11Y_PATHS -u PLAYWRIGHT_BASE_URL make -C ../oink.pgsty.com browser

The full theme checker set and publication commands are defined in .github/workflows/ci.yml. Run all owning checkers with fresh fixtures, including parameters/schema, vendored assets/fonts, navigation/search/actions, components, output/namespace/goldens, migrations, snapshot protection, consumer tooling, and PDF isolation. The engine suite is npm run test:appearance:engines in the site repository.

For compatibility checks, put the selected Hugo binary on PATH, disable inherited Go/Hugo workspaces, and identify the sibling replacement explicitly. Build the real site with --environment production --minify --printPathWarnings --panicOnWarning into a separate output directory. For published-pin checks, disable the replacement as well.

Remaining release steps and limits

Local technical pre-release acceptance passed. Completing the release requires assembling reviewed changes into commits, running CI on those commits, publishing the tag and module archive, updating downstream pins, and verifying the hosted deployment.

Real Windows/Android font rendering, manual screen-reader speech, and first-paint filmstrips remain unverified. Windows source-path behavior was checked through deterministic fixtures. Ink/Terminal design follow-up remains in the experiment record.

5 - Visual preset acceptance, 2026-10-05

Local Paper and Slate verification, real theme output, and bounded browser evidence.
Local verification record

This record covers sibling-checkout theme output for the Paper and Slate presets prior to tagging v1.2.0.

Inputs

Theme and documentation working trees on 2026-10-05; Hugo Extended 0.166.0, Go 1.27.1, Node 26.9.0, and Playwright 1.62.1 on macOS ARM64. The ordinary browser suite uses Chromium; the additional engine gate uses Chromium, Firefox, and WebKit. The site pinned v1.1.0 at the time of this run; make check, make browser, and make dev select the local sibling theme.

Executed checks

Evidence Result and scope
check-presets.py Paper light/dark token parity and AA text/link/code/copper contrast; frozen v1.1.0 Slate base palette; four strict configuration builds and 16 HTML roots including 404 and print
Existing theme checkers Parameters, font roles, vendor inventory, 32 locale catalogs, actions, shell, output, namespace, Landing and runtime isolation passed; generated schemas match their sources
check-goldens.py 52 surfaces passed after reviewing and updating the 34 HTML/print expectations affected by root attributes, prepaint colors, the menu and its action/runtime; other output formats were unchanged
Strict site build Real sibling-theme EN/ZH site built with --panicOnWarning; translations, rendered Markdown and internal links passed
node --test 'tests/js/**/*.test.js' 49 runtime tests passed
appearance.spec.mjs 25 tests passed: Paper/Slate × EN/ZH × 390/1440 px × light/dark on Home, configuration, callouts and tabs; menu axe checks; keyboard, persistence, default reset, language navigation, cross-tab sync, blocked storage, invalid values, no JS, print, command palette, reading-anchor/breakpoint handling and generated comment stylesheets
appearance-engines.spec.mjs Six checks passed: Chromium, Firefox and WebKit at 390/1440 px; stored state and browser chrome color restored before CSS, native keyboard selection, focus return and language navigation. Desktop Chromium also used 4× CPU throttling
Font requests All observed fonts were local. Paper requested no Inter; Slate requested no Plex Sans. Two real-site overlay builds proved system typography requests no bundled text face and explicit font roles override both presets
make check Complete non-browser suite passed: 57 tests, 141/141 translated pages and the existing Markdown/rendered-content/internal-link checks
make browser 197 tests passed across all nine standard suites; sitemap axe scan scoped to the 15 routes below
Visual inspection Actual Paper desktop Home, mobile long-form Docs, English/Chinese light/dark Appearance panels and Slate dark Home reviewed from browser test captures

The browser suite’s sitemap axe pass is scoped with A11Y_PATHS to 15 representative routes: EN/ZH Home, configuration, callouts, tabs, and OpenAPI; English search, Mermaid, ECharts, Blog, and a Book chapter. The existing responsive axe matrix runs in addition. Cross-origin Giscus and vendor API widget DOM retain the suite’s established exclusions.

The integration run found and fixed Paper dark highlighted-line gutter contrast and smooth-scroll interference with reading-anchor restoration. Theme-color checks now assert both presets: Paper’s opaque warm selection surface and Slate’s existing translucent selection surface.

Limits and next checks

Manual screen-reader speech and visual filmstrip/paint traces remain unverified. The blocked-stylesheet and throttled-CPU assertions verify initialization order. Slate comparison freezes base palette values and verifies rendered font behavior.

Mermaid/ECharts retain mode-only palettes. Ink and Terminal remain research; serif display headings, full geometry/density tokens, and preset-colored charts remain future work.

6 - Consumer and migration evidence

A dated consumer corpus survey that informed OINK shells, authoring primitives, and deterministic Book migration rules.
Dated corpus snapshot

These counts describe repositories inspected in August 2026 to inform OINK’s shell, component, and Book migration design.

Corpus

The authoring survey scanned the content/ trees of eleven OINK consumer sites and found 5,325 Markdown files, of which 5,293 had YAML front matter. The survey included single-language English and Chinese references, bilingual product sites, release archives, custom landing pages, and separate Book consumers.

The survey measured source Markdown, counting shortcode calls, fenced-code attributes, callouts, table markers, raw HTML, front-matter keys, content types, and site-local layouts. A later Book-focused survey added five long-form consumers.

Findings that changed the design

Evidence Resulting choice
Content ranged from nearly plain Markdown to pages with many nested components Native Markdown is the default form; a full form survives only for a named capability gap
Documentation, Blog, Landing, releases, and books repeatedly reimplemented navigation or cards locally Extend the shared shell, registry, and primitive rather than adding a parallel system
Site-specific table classes were common, while canonical Field-table headings were rare Hook attributes use an allowlist but preserve documented site-class extension points; Fields cannot be inferred from arbitrary two-column tables
Book sites carried private figure, table, equation, example, and cross-reference conventions Numbered primitives and migration profiles need deterministic classification, stable IDs, and rendered-target verification
Sites mixed single-language, peer-file bilingual, and generated-language content Language authority and generation boundaries must be explicit; a migration never treats an untracked generated tree as source
Rich HTML pages still needed print, Markdown, feeds, and agent output Every component declares its output degradation before its interactive HTML is accepted

The evidence also ruled out several additions: documentation sites did not require a second Landing system, Book sites did not need a new cover component, and a serial archive did not require a new shell type. Remote API collection belongs in site-side CI, outside a Hugo theme built for local rendering.

Block and table evidence

A focused survey of eleven sites plus Book consumers found 11,484 pipe tables, but only eleven matched the strict Field-table heading vocabulary. Roughly 874 were reference-style tables and about 1,300 were compatibility matrices. OINK therefore uses explicit .fields and .matrix markers rather than inferring these components from table shape.

The same survey found eighteen Steps blocks in the eleven-site corpus, all using the full form with headings and rich content. Platform probes showed that a native ordered list could contain most of that content, but not another full % container inside a list item. OINK keeps both forms because their technical capabilities differ.

Deterministic Book migration

Three dated dry-run profiles checked whether migration rules accounted for every recognized source without inventing meaning:

Profile snapshot Classified result Manual boundary
DDIA v2 106 figures, 3 tables, 22 code examples, and all 304 relevant links accounted for One caption link flattened to visible text; no unaccounted skip
DDIA v1 90 numbered figures and 203 matching references 14 decorative or unnumbered images deliberately left alone
TPME 31 figures, 10 tables, 44 numbered references, and 1,018 generic stable references No skipped recognized item
Private Book profile 119 figures, 5 tables, and 136 numbered references 3 ambiguous images retained for manual review

Each profile first ran without writes, and writes began only after reviewers understood its ambiguous inputs. A second run produced zero changes. Each profile built with warnings fatal and passed rendered kind/number/anchor checks. The public migration toolkit and current profile boundaries are documented in Writing a book and the migration contract.

Publication adoption snapshot

An isolated 2026-08-24 check tested the released generic Book publication path with two consumers:

Consumer Generic publication evidence Downstream status
DDIA 23 ordered pages, 131 typed targets, and 292 resolved cross-references; EPUBCheck, internal, and PDF checks passed At this snapshot it still retained a semantic preprocessor pending independent acceptance of the new gate
TPME 18 ordered pages, 41 typed targets, and 1,062 resolved cross-references; the same generic checks passed A second consumer confirmed portability; it created no upstream migration gate

These results record downstream adoption of the generic publication pipeline.

Limits

Repeating this survey requires a fresh repository list and a new dated snapshot. This public record omits local filesystem paths, uncommitted content, private repository names, and generated build artifacts.

7 - OINK comprehensive review, 2026-08-26

Evidence-based review of post-v0.7.0 implementation, configuration, outputs, security, tests, performance, bilingual contracts, and the real integration site.
Historical review snapshot (2026-08-26)

This page records the 2026-08-26 audit of github.com/pgsty/oink and its integration site. OINK 0.7.1 subsequently fixed code findings F01–F06 (see the 0.7.1 release notes). Current behavior is defined by the Design contracts.

Review verdict

OINK’s main-line quality was substantially higher than that of a typical Hugo theme: the default path built successfully, bilingual coverage was strong, and component tests were broad. The project treated output and trust boundaries seriously. The real site showed no general breakage on desktop, mobile, light/dark modes, or primary accessibility paths. Theme and site worktrees were clean. Their latest remote checks succeeded, and every locally rerun first-party suite passed.

While those checks passed, this review found 4 P1, 9 P2, and 5 P3 findings, where some early or peripheral features fell short of OINK’s modern contract. The gates preserved selected positive scenarios very well, but did not systematically cover configuration combinations, static-output behavior, or the semantic accuracy of public documentation.

Before the next release tag, at minimum:

  1. Disable Swagger UI’s default online validator. Verify no implicit outbound requests with a non-localhost browser test.
  2. Apply common type, range, URL, and CSS-value validation to all public configuration and Landing data.
  3. Redesign Swagger, Redoc, and Asciinema behavior in static outputs and their runtime gates.
  4. Repair generated schemas. Update public configuration and front matter references to match behavior.

Baseline and method

Review baseline

Item Snapshot
Theme repository clean main at fe439fdb1d7c2df745088c9bfcbb8c350403ee63, equal to origin/main
Current stable tag v0.7.0 at cbb6f4e0bfe47e17ba7aa41d04b8651c943cf858
Documentation site clean main at fd5fcde, publicly pinned to github.com/pgsty/oink v0.7.0
Local tools Hugo Extended 0.164.0, Python 3.14.6, Node 26.4.0, npm 11.17.0
Remote CI theme HEAD GitHub Actions run 32792753866 succeeded

Validation executed

  • All 31 theme checkers passed.
  • All 85 migration unit tests passed.
  • All 38 theme browser-runtime unit tests passed.
  • All 40 HTML/Print/Markdown/RSS/LLMS golden surfaces passed.
  • The strict tests/site Hugo build passed.
  • The real bilingual site’s npm test passed: 121/121 page pairs, 886 heading IDs, 24,860 internal links, and 3,172 fragments.
  • The full real-site Playwright suite passed: sitemap-wide axe, 29 accessibility cases, and 45 responsive/navigation cases. It also passed 16 keyboard, 10 content-component, 18 code-block, 4 PRD5, and 5 theme-color cases.
  • Extra visual review at 320 CSS px covered EN Home, ZH configuration, ZH Book, and OpenAPI/Redoc, with no page-level horizontal overflow.
  • npm audit found no advisory among the site’s 79 npm dependencies. An OSV Query API batch for 26 exact VENDOR.json versions found no known advisory.
  • measure-baseline.py assets --fixture-site passed its isolated strict build.

Severity

Level Meaning
P1 Breaks a core product, security/privacy, or ordinary-editing invariant; fix before the next tag
P2 Material behavior, contract, or compatibility defect; fix soon with a behavior gate
P3 Maintainability, performance, process, or documentation-governance debt

Finding summary

ID Level Finding Default-site impact
F01 P1 Swagger UI enables its online validator on production URLs Pages using swagger only
F02 P1 Invalid configuration can crash ordinary Hugo or silently emit bad output Depends on authored configuration
F03 P1 Swagger/Redoc/Asciinema violate static-output and runtime-isolation contracts Pages using those shortcodes
F04 P1 Landing sends unvalidated data to safeCSS and lets other bad values pass silently Related Landing fields
F05 P2 Custom page-action and archived-version URLs bypass the shared URL policy Sites configuring those options
F06 P2 Generated JSON Schemas contain wrong defaults, types, descriptions, and candidate keys Authors using editor schemas
F07 P2 The supposedly complete configuration/front-matter references lag v0.7 behavior All maintainers and consumers
F08 P2 Design contracts and proposal lifecycle present conflicting authorities Maintainers
F09 P2 OpenAPI accessibility defects are excluded while Redoc is presented as an alternative OpenAPI readers
F10 P2 Strict-CSP guidance omits theme-owned inline script and style Strict-CSP consumers
F11 P2 No browser support baseline; automation is Chromium-only Firefox, Safari, RTL, forced-color users
F12 P2 Output-security and rendered-Markdown gates have systematic blind spots Consumers relying on those verdicts
F13 P2 Real cross-repository candidate integration is manual and non-atomic Every public behavior change
F14 P3 Checker duplication and source-string coupling are high Maintainers and isolated worktrees
F15 P3 Baseline CSS and fonts remain the main first-visit payload Every HTML page
F16 P3 Vendor integrity is strong, but vulnerability/SBOM and CI supply-chain gates are manual Release maintainers
F17 P3 Changelog, implemented proposals, and behaviorless metadata reduce signal Maintainers and upgraders
F18 P3 The Print isHTML FIXME no longer explains the real dependency Print-template maintainers

Detailed findings

F01 — Swagger UI implicitly contacts the online validator (P1)

layouts/_shortcodes/swagger.html initializes SwaggerUIBundle without validatorUrl: null. Vendored swagger-ui-bundle.js defaults to https://validator.swagger.io/validator. It suppresses the badge only when specification URLs contain localhost or 127.0.0.1. On deployed hosts, it creates an online-validator badge whose request includes the specification URL.

This violates the promises of disabled-by-default theme network features, local same-origin specifications, and local-first operation, and can disclose internal hostnames or specification URLs on intranet deployments. The localhost exemption explains why every local browser test missed the request.

Set validatorUrl: null explicitly. Any future online validator should require an explicit opt-in URL, use the shared URL policy, and have privacy/CSP integration documentation. Test a production-like, non-localhost origin while intercepting every request. Require same-origin specifications to fetch only first-party resources.

F02 — Invalid configuration does not consistently warn and fall back (P1)

ui-param.html says callers validate types. Some callers do not. Minimal builds produced these results:

Input Actual result
ui.blog_index_size: nope ordinary build fails because .Paginate requires a positive integer
ui.sidebar_expand_levels: nope ordinary build fails in add
ui.sidebar_menu_truncate: nope ordinary build fails while first casts the value
offline_search_summary_length: nope ordinary build fails while truncate casts the value
ui.sidebar_width_min: "1; color: red" warning-free build emits --td-shell-sidebar-min: ZgotmplZpx
ui.sidebar_width_min: -50 warning-free build emits -50px
blog_index_columns: 2.5 / section_index_columns: 2.5 warning-free build feeds 2.5 to CSS repeat()
ui.sidebar_item_overflow: clip warning-free build silently behaves as ellipsis
ui.sidebar_menu_foldable: definitely the non-boolean string is truthy and enables folding
ui.blog_index_size: 0 Hugo default silently converts it back to 12

Landing marquee.rows/capabilities.columns and Asciinema numeric parameters also call int/float directly. Other invalid types, including print.toc and offline_search_max_results, silently change behavior.

This directly contradicts the Diagnostics decision: ordinary hugo server can become unusable, while some invalid input passes strict publishing gates without warnings. Add shared validators for integers, positive integers, ranges, paired ranges, and grid counts. Normalize values before arithmetic or output. For every public key, test valid site and page cases, invalid ordinary cases with warnings/fallbacks, and invalid strict cases with failures. Domain resolvers should enforce cross-field rules such as min <= max, pager size >= 1, and integer grid counts.

F03 — OpenAPI and Asciinema remain HTML-only islands (P1)

Architecture and Components require Markdown/LLMS without theme markup, static Print, and safe static RSS or explicit omission. The observed behavior conflicts with these requirements:

  • Redoc emits <style>, <div class="td-redoc">, and <redoc spec-url=...> into generated .md.
  • Swagger puts an executable inline initializer in its shortcode.
  • Asciinema .md contains the complete td-asciinema tree and JSON script.
  • Asciinema Print loads about 185 KB of player JS/CSS and prints only an incidental frame.
  • Swagger/Redoc leave empty Print containers and can select 1–2 MB runtimes.
  • These shortcodes are absent from the Markdown/RSS/Print golden matrix.

The result is theme HTML in agent output, empty shells for paper/EPUB readers, useless runtime in Print, and a strict-CSP violation from Swagger. Reader-facing guides describe these defects as output behavior, conflicting with normative contracts.

Make all three branch on tdOutputFormat. Provide full behavior in interactive HTML. For Print/Markdown/RSS, provide a titled static link with the specification or cast address, or explicitly omit the component. Only interactive HTML should set capability flags. Move Swagger initialization to a stable chunk and Redoc styles to a stylesheet. Add four-output goldens and assertions that unwanted runtimes are absent.

F04 — Landing CSS, URL, and numeric inputs do not share one trust boundary (P1)

hero.html validates title_size, but concatenates media.ratio and media.max_width unchanged before marking the full string safeCSS. This input:

sections:
  - type: hero
    data:
      title: Probe
      image: /icons/logo.svg
      media:
        ratio: "1fr; background-image: url(https://example.invalid/x)"
        max_width: "240px; color: red"

passes a strict build without warnings and emits:

style="--td-hero-columns: 1fr; background-image: url(https://example.invalid/x);
       --td-hero-media-max: 240px; color: red;"

Landing accepts inline sections in page front matter, so authors can supply these values directly. Other sections handle columns, rules, dimensions, styles, icons, and URLs inconsistently. A javascript: URL often becomes #ZgotmplZ without warnings. Invalid columns become ZgotmplZ. Some direct integer casts abort builds.

Add section normalization with common handling for classes, icons, URLs, CSS lengths, grid counts, booleans, and enums. hero.media.ratio should accept two constrained track values, not arbitrary CSS. max_width should use the length validator. All Landing actions should use content/url.html. Each built-in section needs negative tests.

F05 — Two configuration URL surfaces bypass shared policy (P2)

params.ui.page_context_menu.links passes through url-template.html directly to safeURL. url_latest_version is also treated as trusted configuration and marked safeURL. Neither path validates schemes, hosts, whitespace, or protocol-relative URLs. Warning-free builds can produce:

<a class="td-page-actions__item"
   href="javascript:document.body.dataset.pwned=1;undefined">

Clicking executes JavaScript, which violates the published safe-URL model and gives copied configuration fragments unnecessary execution capability.

Use the shared resolver to allow only HTTP(S) and explicitly supported first-party relative URLs. Also validate archived-version URLs. The browser action registry’s second check provides useful defense, but progressive-enhancement anchors must not bypass it.

F06 — Generated schemas disagree with actual YAML (P2)

The small parser in generate-config-schema.py does not remove inline comments. At least 11 defaults become strings. These include print.toc, which becomes "true # ..." instead of a boolean, and print.section_break_wordcount. Both index column counts and enum defaults such as footer_style, blog_index, and typography are also affected.

Comment association is also incorrect. Breadcrumb commentary attaches to section_index; quick-link commentary attaches to sidebar_icon_policy. Taxonomy-icon commentary attaches to pager_types; local-interface commentary attaches to image_zoom.

The front matter schema offers removed detector keys: release, upstream_attribution, and downstream_modified. It incorrectly treats navbar-menu Params.columns as page front matter. The drift check compares the same defective generator with committed output, preserving the error.

Use a real comment-preserving parser or explicit machine metadata markers. Do not extend the ad hoc parser. The scanner must distinguish page, menu, shortcode, and legacy-detector contexts. Tests should compare schema defaults with Hugo’s actual parsed values and exclude removed keys from completion.

F07 — Public configuration and front-matter references are not current (P2)

Both reference pages claim to list every key read by the theme. Material differences include:

  • Long English date defaults, although hugo.yaml uses ISO 2006-01-02.
  • Blog docs omit hero, table, toggle, size, toc_style, and toc_taxonomies.
  • Removed release maps and filters are presented as current, while release_url is absent.
  • images: [] is described as disabling featured images, although bundle discovery continues.
  • upstream_modified is described as adding a line, although it changes the attribution verb.
  • Invalid-input claims conflict: immediate failure versus warnings in ordinary previews and failure only at strict gates.
  • The Book guide says OINK stops at Print HTML, although v0.7 shipped BookManifest/EPUB/PDF tools.
  • Asciinema/OpenAPI guides present static-output defects as product contracts.
  • Features lists 28 vendor dependencies, although the authoritative manifest has 26.

English and Chinese usually share the same outdated answer. Translation parity cannot detect that error. Treat both references as a focused contract migration. Derive comparable key inventories from implementation/schema. Review semantics manually. Check current-key coverage, placement of removed keys, enums, and defaults.

F08 — The Design tree contains conflicting authorities and unretired proposals (P2)

Shell retires navbar columns/mega panels and promises a warning with one column. Landing still says navbar mega-menu columns accept 1–4. Implementation and tests follow Shell.

The lifecycle is also incomplete. config-schema is marked implemented but remains an Active proposal. Book publication has shipped the manifest, EPUB, PDF, and most CI work, but a Draft proposal duplicates the Architecture contract. Media convergence keeps implemented milestones and open M4 in the original design record.

Correct Landing. Move stable config-schema facts to Architecture/Decision and retire the proposal. Reduce Book publication to the remaining consumer-migration question, or replace it with a focused follow-up. Active proposals should not define a second current API.

F09 — OpenAPI accessibility claims conflict with test exclusions (P2)

The axe suite excludes .td-swagger-ui and .td-redoc. Its comments identify Swagger’s unnamed server selector and version stamp that cannot scroll by keyboard. They also identify Redoc operation-description contrast. The guide discloses only Swagger defects and presents rendered Redoc as the alternative. This implies that Redoc meets the site’s zero-violation gate.

Publish the actual boundary in both languages. Fix Redoc contrast in theme CSS where possible. Use a narrow post-render adapter for fixable Swagger DOM. For remaining upstream defects, provide versioned waivers, upstream issue links, and a separate axe report. Do not exclude the entire supported feature while claiming zero site-wide violations.

F10 — The current theme does not directly support a strict CSP (P2)

Deployment guidance says strict CSP is workable. It lists only author scripts, ECharts callbacks, analytics, remote specifications or diagram services, and Giscus. Ordinary Docs pages already emit two theme-owned inline scripts for initial theme display and shell prepaint, plus inline styles. Markmap, Swagger, Algolia, and Google CSE add more theme-owned inline initializers. There is no nonce API, hash manifest, or complete sample policy.

script-src 'self' blocks initial theme display and shell-state restoration. style-src 'self' blocks theme colors, font roles, Landing, and several inline custom properties. Consumers must add 'unsafe-inline', maintain hashes, or override templates. The guide states none of these requirements.

Move stable initializers to same-origin chunks with data/JSON configuration. For unavoidable inline content, provide a generated hash manifest or one nonce hook. Publish policies for minimal core, Markmap/OpenAPI, and third-party integrations. State the style-src requirements.

F11 — Browser compatibility has no baseline or cross-engine proof (P2)

CI installs only Chromium. Product documentation gives no minimum Chrome, Firefox, or Safari version. The implementation uses or enhances content with :has(), dialog, inert, color-mix(), @property, logical properties, and discrete display transitions. Some paths have fallbacks, but there is no engine matrix.

RTL checks mostly use source markers, small JS tests, and one element-level geometry change. They do not use a complete RTL-language site. Most forced-color checks verify SCSS strings instead of computed behavior.

Publish a small support matrix. Test core shell, navigation, content, and dialogs on Chromium, Firefox, and WebKit. Add a real languageDirection: rtl integration configuration. Also test forced colors, reduced motion, 320 px, and 200% zoom.

F12 — Output-security and Markdown gates do not inspect every claimed surface (P2)

For .md, check-output-security.py checks only Markdown links. It does not send raw HTML through the HTML scanner. Redoc/Asciinema scripts, spec-url, and raw href are therefore invisible. It also ignores CSS and JSON URLs. The fixture uses a broad --third-party allowance.

check-rendered-markdown.mjs has a misleading name. It scans generated HTML text for leftover Markdown syntax, not generated .md. The actual Markdown golden set covers 15 pages and omits OpenAPI/Asciinema.

Use clearly named gates for HTML trust, machine-output purity, and rendered-text residue. Give generated Markdown a narrow raw-HTML allowlist. Deliberately parse CSS URLs, form actions, JSON URLs, and non-executable JSON scripts. Every public shortcode should have at least one Markdown/Print/RSS behavior case.

F13 — Candidate integration across the two repositories is manual (P2)

Theme CI tests only synthetic tests/site. Documentation-site CI tests only the public tag in go.mod. Real EN/ZH and Playwright validation for theme PRs depends on local maintainer HUGO_MODULE_REPLACEMENTS. Changes in both repositories cannot be committed atomically.

Both repositories can pass checks while public references differ from implementation, as this review shows. The written separation of release states is correct, but automation does not enforce delivery of implementation, owning checker, and paired contract together.

Add a read-only candidate workflow. Check out a theme PR SHA and a declared documentation-site SHA. Apply a temporary module replacement, then run npm test and critical browser suites. Allow Design-contract PRs to identify candidate theme SHAs. Tag, pin, and deployment remain distinct. The candidate pair gains one traceable joint result.

F14 — Checker maintenance cost and source coupling are high (P3)

Coverage is valuable, but 34 check-*.py files contain 546 read_text() calls. Many repeat require, temporary-site creation, file writes, Hugo invocation, and error aggregation. Many assertions preserve template/SCSS spelling, nearby comments, or complete file equality instead of observable behavior.

Some helpers hard-code theme: oink with --themesDir <repo-parent>, making the checkout or worktree basename an implicit requirement. Without a common Python lint/type gate, checkers are quick to add but can retain shared testing omissions.

Create a common fixture builder and assertion library. Put negative cases in table-driven data. Keep source checks only for actual topology rules. Move other checks to parsed output or computed styles. Load themes through explicit symlinks or module replacements instead of repository basenames.

F15 — Runtime splitting succeeded, but baseline CSS/fonts dominate first visit (P3)

The isolated strict fixture baseline was:

Metric Value
Cold/warm build 1.256 s / 1.273 s
Pages 249
Stable JS chunks 18
Main + Font Awesome CSS 549.8 KB raw / 91.1 KB gzip
Fonts total (FA portion) 999.7 KB raw / 248.5 KB gzip
Median Docs-page JS 176.9 KB raw / 55.3 KB gzip
Generated public 26.2 MB
v0.7.0 Go module zip 7.8 MB (about 20.5 MB and 1,140 files expanded)

Stable first-party capability chunks removed bundles for feature combinations correctly. Large third-party runtimes remain page-local. Bootstrap/theme/Landing CSS and the complete Font Awesome distribution are the remaining common cost.

Do not prune Font Awesome based on theme template usage. That would violate the authoring contract. Measure whether Landing, Book, or Swagger CSS can be independently cached and limited to relevant pages. Inspect fonts requested on first visits. Maintain a trend report instead of an arbitrary hard threshold.

F16 — Vendor builds are reproducible, but advisory and CI supply-chain gates remain manual (P3)

Positive evidence: VENDOR.json pins 26 packages, 56 artifacts, 31 license files, and tree hashes. check-vendor.py passed. OSV and npm audit found no known advisory in this snapshot.

The custom manifest is outside common SBOM/advisory gates. npm audit cannot inspect vendored browser packages. Two documentation-site workflows download Hugo .deb files and immediately run sudo dpkg -i without checking a checksum. Actions use movable major tags. Theme CI uses floating Python 3.x.

Generate CycloneDX/SPDX from VENDOR.json. Schedule OSV scans. Pin Hugo archive/deb SHA-256 values and trusted release actions to commit SHAs. Select a specific Python version or matrix.

F17 — Design and release records have lost signal (P3)

CHANGELOG.md has 1,768 lines. Its v0.7.0 section has about 300 lines, and Unreleased uses about 20 lines for one checker retry. These narratives provide useful engineering history, but make breaking changes, migrations, and observable behavior harder for upgraders to find.

Contracts recognize book_kind and book_part, and content front matter repeats them, but templates explicitly do not read them. They impose API-like authoring work without behavior. Implemented proposals that remain active provide another duplicate answer.

Keep changelog entries focused on observable changes, breaking/migration notes, and concise fixes. Move long design accounts to Blog/Research and link them. Give metadata without behavior a consumer/schema, or make it site-owned.

F18 — The Print isHTML FIXME is no longer accurate (P3)

hugo.yaml says to leave isHTML unset until Hugo fixes issue #14381. Hugo closed it on 2026-01-17, and the fix shipped before OINK’s 0.160.1 floor. Enabling isHTML: true still produces missing page/section/landing Print-layout warnings in this theme, which fail strict builds.

The dependency changed from “waiting for an alias fix” to “the current Print template names rely on non-HTML lookup rules.” Do not simply remove the workaround. First complete the HTML-classified Print lookup matrix and alias/subpath tests. If false remains intentional, explain the actual reason in the comment. Add a test to prevent cleanup based only on the closed issue.

Strengths

  • Source, local validation, commit, tag, public module, consumer pin, and deployment are explicitly separated.
  • The Hugo 0.160.1 floor plus 0.164/0.165 theme matrix is strong.
  • Most newer components follow warning/fallback, four-output, shared URL/attribute, and capability-flag contracts.
  • The 32 locale schemas match, with strong real EN/ZH page, heading-ID, link, and narrow-navigation gates.
  • Search, keyboard behavior, surface coordination, page actions, and theme color have both unit and browser behavior tests.
  • Vendor license/hash checks and EPUB/PDF path, loopback, CSP, and overwrite boundaries are thoughtfully designed.
  • Manual 320 px review found no page-level overflow; current core visual quality is good.
  • Builds are fast, and first-party JS now uses stable capability chunks.

Recommended remediation roadmap

Phase 0: before the next tag

  1. Set Swagger validatorUrl: null and add a production-origin no-network test.
  2. Build the public-parameter inventory and validate every F02/F04 field with negative cases.
  3. Redesign four-output behavior and runtime gates for Swagger, Redoc, and Asciinema.
  4. Validate custom action and archived-version URLs.
  5. Repair the schema parser/scanner and regenerate both schemas.
  6. Synchronize paired Config, Front matter, OpenAPI, Asciinema, Book, Features, and Landing-contract pages.

Phase 1: contract gates

  1. Create a minimum HTML/Print/Markdown/RSS coverage map for all 29 shortcodes.
  2. Split and strengthen output-trust and machine-output-purity gates.
  3. Normalize all Landing section input centrally.
  4. Externalize theme-owned inline initializers and publish CSP guidance.
  5. Add a cross-repository candidate workflow.

Phase 2: compatibility and structure

  1. Add Firefox/WebKit, real RTL, forced colors, and 200% zoom.
  2. Consolidate the Python checker harness and source-string assertions.
  3. Evaluate surface-specific CSS and actual font requests.
  4. Generate an SBOM, schedule OSV, and pin CI download digests.
  5. Retire implemented proposals and reduce changelog volume.

Acceptance criteria

  • A same-origin Swagger specification on a production-like origin makes no third-party request.
  • Every invalid public configuration warns and falls back/omits in ordinary builds, fails strictly, and emits no ZgotmplZ.
  • Generated .md contains no td-*, theme script/style, or empty interactive container.
  • Print loads no Swagger/Redoc/Asciinema runtime and provides an understandable static alternative.
  • Schema default types exactly match Hugo parsing, and removed keys are absent from completion.
  • EN/ZH configuration and front-matter key/enum/default inventories match implementation.
  • Core Playwright passes on Chromium, Firefox, and WebKit, with real RTL and forced-color behavior assertions.
  • Every candidate theme SHA has a traceable joint validation against the real documentation site.

Review limits

This review did not individually audit consumer repositories, production response headers/CDN caches, real Firefox/Safari, or screen readers. It did not manually reverse-engineer 13 MB of minified third-party source. Advisory checks are a 2026-08-26 snapshot and may change. Existing CI/contract evidence covered DDIA/TPME EPUB/PDF consumers.

8 - Community issue and PR review, 2026-09-19

Evidence, acceptance advice, and focused remedies for community issues 40, 41, 42, 44 and pull request 43.
Historical review snapshot (2026-09-19)

This page records source inspection, GitHub status, local builds, and targeted browser observations from 2026-09-19, followed by the same-day implementation and acceptance follow-up.

The initial review remains below; the maintainer later authorized merging PR #43 and implementing the other items directly on main. See the same-day implementation and acceptance follow-up.

Verdict

The reports identify useful problems, but they are not all defects of the same kind. Fix hidden navigation focus first. Accept PR #43’s direction as a small correctness fix after clarifying its boundary and adding coverage. Treat search-tail registration and public sidebar state as additional APIs requiring their own acceptance work.

Item Finding Recommendation
PR #43, MagicFollower Self-root collection ignores an explicit sidebar_root_menu: false. Reproduced. Conditional acceptance: the patch correctly filters the global candidate list. Document the current-root exception, add regression tests, and obtain successful CI.
#41, imbajin Hidden whole-sidebar content remains focusable. Several controllers write disclosure state, and no public API exists. Split a correctness repair from an optional API. The former has higher priority.
#44, lloydsun Pointer focus followed by a key produces the reported outlines. Reproduced on another platform. Improve the main-content focus treatment. Retain useful keyboard cues for scrollable content. The browser heuristic itself is expected.
#42, aucru Existing hiding and divider options do not provide a complete non-link section group with its children. Explain the options separately and obtain the author’s exact minimal example. If the missing group behavior is confirmed, implement it.
#40, imbajin There is no supported way to append a query-dependent action to local search. This is a reasonable small extension proposal. It does not identify a failure of existing local search. Correctness repairs have higher priority.

Baseline and method

GitHub API reads found four open external issues and one open external PR. The other open issue, #37, was the maintainer’s release tracker. The external items had no discussion comments or submitted PR reviews at the snapshot time.

Input Verified snapshot
Remote theme main 93ac292014a3cd81f7c41caec4df98ed9d2dc45a
Local theme 75ddc95. Only the release text in CHANGELOG.md differs from remote main.
PR #43 head 8eeb8ecaf525097cc56572fe22234db381bfc16a, one changed template line
Local documentation ff0ba39. go.mod still requires OINK v1.0.0.
Published release GitHub’s latest release is v1.0.0. The remote query returned no v1.1.0 tag.
Build tools Hugo Extended 0.166.0, Node 26.9.0, npm 11.19.1
Browser observation macOS, Chromium 153.0.0.0, light theme, real sibling documentation site using a command-scoped module replacement

The method combined complete issue/PR bodies and comments, the exact diff, and bilingual Design contracts with inspection of owning templates and JavaScript. It also used a temporary bilingual site under the /sub/ base path and targeted documentation-site browser interactions. No theme implementation changed in the shared checkout.

PR 43: accept the small fix with a precise boundary

The first pass in root-menu-roots.html filters top-level sections by sidebar_root_menu. The second pass collects sections with sidebar_root_for set to self, but omits that filter. Sections excluded in the first pass can therefore return in the second.

The PR adds the same explicit-false predicate to the second pass:

{{- if and .IsSection (ne .Params.sidebar_root_menu false) -}}

The patch keeps the default for absent or true values, the section constraint, and URL deduplication, while preserving navigation-tree order, pager order, and language-specific caching. These properties make it a focused correctness fix with no need for a broader navigation refactor.

However, root-menu-entries.html subsequently appends the current resolved root when absent. This existing behavior is documented in the navigation guide, so it is not a new PR regression; it does prevent the broader claim that false hides the root on every page.

The local probe used a top-level Blog root and nested Docs root, both with sidebar_root_for: self and sidebar_root_menu: false. It also included a visible Docs root and a self-root without a visibility override. English and Chinese results were identical and retained language-aware /sub/ URLs:

Viewed page Before the PR With the PR
An unrelated Docs page Both hidden self-roots appear Both disappear
A page inside the hidden Blog root Blog appears Blog still appears through current-root fallback
A page inside the hidden nested root Nested root appears Nested root still appears through current-root fallback
A visible self-root Appears Still appears without duplication

Recommended contract: false excludes roots from site-wide selectable candidates, but the current root may remain for orientation. Keeping this exception is the smallest compatible interpretation. State it explicitly in both languages. If the intended contract requires absolute exclusion, change the current-root fallback and switcher trigger together. Another predicate alone is insufficient. Check zero and one-entry states.

Before merging:

  1. Add output-based bin/check-shell.py cases for hidden top-level and nested self-roots, absent/true values, deduplication, and current-root behavior. Cover one-entry degradation and EN/ZH subpaths.
  2. Update the Shell contract and navigation guide together. Change the PR description’s YAML comment from // to # so readers can paste its example.
  3. Resolve the workflow’s action_required result and run required checks on the final head. At this snapshot, the PR head had no completed check runs (MERGEABLE, UNSTABLE).

The maintainer can add these small changes while preserving the contribution. Do not make acceptance depend on #40 or all of #41.

Issue 41: repair isolation, then expose state

There are two separate findings.

Whole-sidebar hiding uses transforms and, on desktop, opacity. Drawer and collapse controllers leave hidden controls in keyboard navigation. In the browser probe, Collapse sidebar left focus on the transparent collapse button. Tab then moved focus to the hidden root switcher. The panel had zero opacity and no effective inert or aria-hidden ancestor. This is a reproducible keyboard usability defect. Closed mobile panels use similar off-screen positioning without explicit isolation.

Disclosure writes are duplicated across the click controller and responsive relocation and cached active-path hydration. There is no public setter, getter, or committed-state event. Existing storage for sidebar collapse, width, and scroll position does not persist per-branch disclosure state across page loads. The authoring guide’s local expansion-state claim needs this distinction.

Recommended repair:

  1. Centralize whole-sidebar isolation at initialization and every open, close, collapse, hover-overlay, restore, and breakpoint transition. Remove isolation before moving focus inside. Return focus to a visible external control before making content inert.
  2. Isolate content while keeping the external restore button and desktop edge hover target active. Applying inert to the pointer sensor would break hover interaction. aria-hidden alone does not prevent keyboard focus. The HTML inert contract controls interaction and accessibility exposure.
  3. Route disclosure changes through one commit function separately. Update aria-expanded, the open class, and localized label before emitting one event. Repeated writes of the current value should have no effect.
  4. Expose a small setter/getter and event only after defining stable IDs, invalid-ID behavior, initialization readiness, and restoration order. Keep version/locale storage policy downstream-owned. Let the active path take precedence after restoration.

The proposal needs a scope correction. Responsive TOC/backlink/taxonomy groups can move from the sidebar to the right rail. A controller searching only current-sidebar descendants cannot control wide-layout writes. Register OINK-owned targets independently of their current DOM parent. Restrict the public sidebar API to its intended registered subset.

Acceptance must check real Tab order and the accessibility tree in hidden states. It must also check restored desktop collapse on first load, hover entry/exit, focus return, Escape, backdrop close, and scroll unlock. Cover the 768/1200 breakpoints and preserve #24’s visible no-JavaScript fallback.

Issue 44: real symptom, partly expected behavior

On the documentation configuration page, clicking the article heading focused main#td-main-content. Clicking table headers focused div.td-table-scroll; clicking code focused pre.chroma. In each case, :focus-visible was false after clicking and true after pressing unbound z. Main/code outlines changed from none to browser auto outlines. Tables used the theme’s solid outline. This reproduced the mechanism without the reporter’s Linux compositor or Super key.

The Selectors specification explicitly describes keyboard activity changing focus indication while focus stays on the same element. The report therefore provides useful UX feedback. Browsers are not required to keep mouse-focused elements free of focus rings after keyboard activity.

Recommended treatment:

  • Keep the main skip-link target and focusability. Replace its oversized outline with a local, visible content-entry cue, such as a title-area indicator. Verify actual skip-link activation.
  • Keep visible keyboard focus on scrollable tables and code. Normalize its appearance if needed. Focusability enables keyboard scrolling.
  • Do not apply global outline: none, remove all tabindex attributes, or blur focus on arbitrary keypresses.
  • If OINK suppresses only pointer-origin reading focus, define that behavior explicitly for these non-editable containers. Test pointer, Tab, skip-link, and programmatic focus, including dark and forced-colors modes. A global input-modality framework is disproportionate to this report.

This review recommends a focused presentation improvement while keeping table and code keyboard focus cues.

Issue 42: distinguish hiding from grouping

The question names _index.json and supplies screenshots instead of a source fixture. This review could not visually inspect the original screenshots. The author’s intended first change remains unresolved. When replying, request a small directory tree and actual index/front matter. Without that evidence, do not assume _index.json is a supported page source or a typo.

The existing options have different meanings:

Option Current behavior and limitation
no_list: true Removes the child list from the section’s content body. It does not hide the section’s sidebar row.
hide_summary: true Removes an item from a parent section’s body list. It does not change sidebar grouping.
toc_hide: true The content-tree walker filters out the node before recursion. This also removes its subtree from that tree.
sidebar_root_menu: false Controls root-switcher candidates. It does not control the node’s row in the reading tree. See PR #43.
sidebar_root_link_self: false Redirects a self-root’s row to its parent. It does not make the row a non-link group.
sidebar_divider: true Emits a non-link heading. The shared renderer does not emit the supplied children in that branch.
build.render: link Suppresses the section HTML and retains its permalink. The current sidebar still emits a link to it. This option alone is insufficient.

The temporary site confirmed that divider sections keep child HTML while child sidebar links disappear. Sections with only build.render: link have no section HTML, but keep clickable sidebar rows and children. This matches Hugo’s build-option semantics and the theme’s shared node renderer.

For a real “group label with child links, but no directory-page navigation” requirement, first consider completing sidebar_divider for sections. Keep leaf dividers and section children. Use real disclosure buttons when folding is enabled. Check existing consumers before adopting this interpretation. Add a separate node switch only if the divider contract cannot express it compatibly. Keep section publication separate from whether its navigation label is linked.

The change must preserve hierarchy and active-path expansion in both walkers, and keep children in the pager. If a section is intentionally unpublished, the change must avoid dead targets in breadcrumbs, search, root switching, Print, and machine-readable navigation. Hiding whole nodes with CSS is not a solution.

Issue 40: a narrow extension is reasonable

Source inspection confirms the gap. groupsFor creates only built-in page/action groups. The public Palette object has no provider registration. registerExecutor accepts only built-in action IDs. Static URL commands cannot replace rows carrying the current query.

The proposed search-tail slot is a useful upstream boundary if its scope stays small. It uses synchronous data-only row creation, asynchronous activation, and local results first. OINK controls rendering, selection, keyboard handling, and ARIA. It needs no bundled AI provider, credentials, remote search, or generic plugin system.

Before adopting the proposed API, settle and test:

  1. Define which settled text-search states call providers. Preserve empty, command, choice, loading, and default no-extension behavior.
  2. Snapshot the query/locale used for each rendered row. Preserve native empty/index-error messages, retry behavior, and the difference between local page counts and all selectable rows.
  3. Validate and copy descriptors. Render titles and descriptions as text. Isolate provider exceptions, duplicate IDs, and invalid descriptors.
  4. Handle synchronous throws and rejected promises. Clear pending state and reject duplicate activation. Cancel stale sessions. Ensure unregister handles remain safe when IDs are reused.
  5. Test handoff to another dialog. Palette close already avoids returning focus when focus moved outside. Preserve this guard. Define how successful handoff differs from cancellation. An “every close aborts activation” rule can cancel the assistant being opened.
  6. Preserve default local-only network behavior and conditional bundles. Document that registered extensions run as trusted first-party scripts in the page context.

Before implementation, record the accepted API in a bilingual Design proposal. Downstream Ask AI wrappers can continue until a tagged release provides the hook. This is not required for the small correctness fixes.

Delivery order and ownership

Order Delivery Owning checks and documentation
First PR #43 completion and hidden-sidebar isolation as separate small changes check-shell.py; site responsive/keyboard/accessibility cases; EN/ZH Shell contract and navigation guide
Next Main-content focus styling and clarified grouping behavior Content/reading and navigation checkers as appropriate; browser focus/scroll/skip tests; EN/ZH architecture, shell, and authoring guidance
Later Public disclosure controller, then search-tail API Theme JS tests and check-navigation-contract.py / check-palette.py; real site fixtures; accepted bilingual API contracts

For each behavior change, run its owning checker first. Then use the sibling site’s make check, make browser, and make dev for relevant integration and visual review. Do not combine unrelated proposals into one large sidebar/search rewrite. Do not delay small fixes until all features exist.

Suggested response content, not posted: acknowledge #43’s filter defect and explain its current-root exception. Accept #41’s isolation defect separately from the API request. Acknowledge #44’s reproduction and explain standard focus behavior. Explain #42’s options and request minimal input. Describe #40 as a scoped enhancement, not a local-search failure.

Historical external issues are closed. #22 was resolved by enabling Goldmark passthrough, with reporter confirmation. #21 received the Mermaid viewer and corrected centered presentation. Arbitrary right alignment was explicitly excluded. Do not silently count either issue as a new open defect.

Validation and limits

Executed for this review:

  • python3 bin/check-shell.py passed on the local baseline and in an isolated checkout at PR #43’s exact head.
  • Palette controller and model tests passed: two test files, no failures.
  • Strict temporary Hugo builds before/after the actual PR diff reproduced EN/ZH self-root filtering and fallback under /sub/. The same fixture demonstrated grouping limitations.
  • The real bilingual site built with --panicOnWarning using the local theme. Targeted Chromium interactions reproduced all three focus outlines and the desktop hidden-focus defect.
  • Bilingual, rendered-content, and link checks passed, along with all 57 non-browser tests. Initial make check stopped at the llms.txt snapshot because this report added an index entry. After verifying that one-line addition and updating the golden, affected and remaining groups passed on rerun.

Local verification used Hugo 0.166.0 on macOS. Linux Super-key behavior, mobile accessibility-tree isolation, dark/forced-colors cases, and the exact #42 screenshots are covered in the follow-up acceptance below.

Implementation and acceptance follow-up

The maintainer chose to merge the contributor patch first, then complete repairs and extensions directly on main without another PR. PR #43 merged as 6e814089. The merge was pulled while preserving the local release-note commit. The implementation follow-up is 56bfe37.

Item Implemented behavior Owning acceptance
#43 Both root collectors honor explicit false. Current-root orientation remains compatible. Dividers and unpublished sections do not become switcher links. Strict EN/ZH subpath fixtures cover hidden top-level/nested roots, absent/true values, deduplication, current-root fallback, zero and one entry.
#41 One disclosure controller commits ARIA, classes, labels and inert state. A late-safe API supports downstream persistence. Hidden whole-sidebar content is isolated. Hover and drawer restoration remain usable. Runtime tests plus browser checks for atomic events, no-op writes, scope, active paths, blocked storage, responsive relocation, focus return, real Tab traversal, Escape, backdrop and breakpoints.
#44 A pointer-origin mark suppresses later incidental container outlines. Tab and fresh programmatic focus retain visible cues. The skip destination outlines the title. Browser checks for article/table/code in light, dark and forced-colors modes, plus keyboard and skip-link regression coverage.
#42 Divider sections keep their children under a non-link label. build.render: never suppresses their own page. Breadcrumb, search, pager, navigation JSON, Book TOC/Markdown and Print agree. Explicit navigation also works under bilingual subpaths. Strict generic/data-tree fixtures, Book depth-three heading checks, EN/ZH browser fixtures and no-JavaScript traversal.
#40 Trusted site scripts can register synchronous data-only search-tail rows and asynchronous activation. OINK continues to own native ordering, ARIA and validation. It also owns error isolation, cancellation, unregistering and focus handoff. Runtime lifecycle tests and a Chinese browser scenario cover pointer/keyboard selection and literal display text. They also cover context snapshots, external-dialog focus and unregistering.

Acceptance used the sibling theme checkout on macOS with Hugo Extended 0.166.0, Node 26.9.0, and Chromium:

  • All 44 theme JavaScript tests passed.
  • check-shell.py, check-reading.py, check-palette.py, and check-keyboard.py passed. A broader run passed 29 of the other 31 theme CI commands. Two local failures involved the media checker’s version-specific processed-image hashes and four goldens with Hugo 0.166’s changed KaTeX output. Ordinary navigation markup matched after preserving its existing whitespace. Fixed-toolchain verification is recorded below. Unrelated expected outputs were not rewritten.
  • make -C ../oink.pgsty.com check passed bilingual source/rendered/link checks and all 57 non-browser tests.
  • make -C ../oink.pgsty.com browser passed all 141 tests: 30 accessibility, 45 responsive/blog/palette, 16 keyboard, 10 content, 18 code-block, 4 scenario, 5 theme-color, and 13 community regressions. Accessibility included the complete multilingual sitemap scan.
  • Strict theme fixture output and namespace checks passed. Local Book packaging produced an EPUB with five chapters and zero checker errors. It also produced a 23-page PDF with all five expected Book pages and zero checker errors.
  • make dev served the real site for desktop-light, Chinese-dark, collapsed-sidebar restore, and 375px mobile-drawer review. Escape returned focus to the visible drawer opener. The temporary viewport and development server were cleaned up afterward.

Accepted contracts are in Shell and Architecture, with matching Chinese sources. Navigation, organization, palette, and Print guides were also updated. The documentation repository owns regression suites. The theme keeps focused checkers and synthetic inputs.

The merged PR’s fixed-toolchain CI passed all three jobs. The final implementation’s CI run also passed at exact revision 56bfe37092a43fc12c0e16f865d3d3407c55cbde. Hugo 0.165.0, browser runtime tests, and Book publication succeeded. This included media and four-state golden checks that differed locally on Hugo 0.166.0, plus root/subpath publication. The supported 0.160.1 floor was not separately retested. The declared continuous-test toolchain remains 0.165.0.

At this checkpoint, the documentation repository stayed pinned to v1.0.0 while awaiting the v1.1.0 tag, and issues #40, #41, #42, and #44 remained open pending release publication.

Image-copy follow-up

The maintainer also reported preview instructions below images after copying blog articles to rich-text editors. The blog pins OINK v1.0.0 and enables params.ui.image_zoom. That release and the reviewed main revision inserted a visually hidden text span after every eligible image. Native Chromium copying reproduced Open image preview and its Chinese equivalent in clipboard content. This is a theme defect independent of the destination editor.

Theme commit 75052f8 puts image descriptions and localized actions in the button’s aria-label, keeping helper text nodes out of the article. Image alt text, authored captions, native button behavior, and dialog focus return remain intact. The component contract and image guide document this copying behavior.

check-image-zoom.py and all 57 non-browser site tests passed. The focused browser run passed 16 tests: 14 content-component cases and two desktop-light/mobile-dark dialog accessibility cases. Component cases included four new EN/ZH image/gallery clipboard regressions. The regressions read plain-text and HTML clipboard data and check text without its original styles. They verify retained image URLs, alt text, captions, and accessible names.

Downstream blogs pick up the fix once they upgrade their theme dependency past v1.0.0.

9 - OINK 1.1 release review, 2026-09-20

Five reproduced runtime defects, documentation corrections, validation evidence, and the OINK 1.1 publication follow-up.
Historical release review (2026-09-20)

This record captures the pre-release validation and five runtime fixes for OINK 1.1, followed by the v1.1.0 publication record.

Scope and baseline

The review starts at theme commit 75052f8a3106d13ef313644836a5ad545135f484, after the community fixes and image-copy repair. It examines the v1.0.0..main changes and behavior requested by issues #40, #41, #42, #44, and merged PR #43. It also examines bilingual documentation and the release boundary. The previous review records those original reports and their implementation.

The method inspects owning JavaScript, templates and contracts, then exercises transition boundaries with focused regressions that compare failing assertions before each fix with the repaired implementation. Theme checkers and the real sibling documentation site’s integration and browser suites provide the subsequent acceptance evidence.

At this snapshot, the public release, documentation go.mod pin and configured public version are still v1.0.0. The blog consumer’s theme pin is unchanged.

Findings and repairs

Five P2 correctness defects were reproduced and repaired in theme commit 08f6563, pushed to main. They concern ordering in the new APIs and existing focus/navigation behavior. They require no new configuration format or content migration.

Finding Trigger and observed failure Minimal repair
P2: sidebar readiness fires before hydration A consumer awaits OinkSidebar.ready or handles oink:sidebar-ready. The readiness microtask can run between DOMContentLoaded listeners. The cached sidebar’s active path is not yet hydrated, so consumers see incomplete initial state. After all initialization listeners and aside placement finish, resolve readiness in the next task. Preserve the existing ready Promise and event contract.
P2: a pending action can enter a native choice menu Start an asynchronous search-tail action, then activate a native choice such as theme selection. The pending guard ran after the choice branch. This allowed that menu to replace the pending action’s rows. Check pending activation before any row-type branch. After completion, ordinary choice activation is available again.
P2: a collapsed right TOC rail remains focusable Collapse the desktop right rail. Its hidden control and links remain keyboard targets. Focus can stay inside the hidden panel. Apply inert and aria-hidden to the rail panel. Move focus to the visible restore control. On restoration, return focus to the column control. When relocated, keep the movable aside outside that isolation.
P2: arrow navigation skips non-link groups From a child of a divider-only group, Left/a cannot consistently return to the parent disclosure and fold it. The group button is absent from the tree’s focus sequence. Include group disclosure buttons in tree focus navigation and direct-parent traversal. Right/d opens or enters the group. Previous/next page navigation still uses links only.
P2: drawer focus wrapping counts inert descendants In the mobile drawer, collapse an aside group and wrap with Shift+Tab. Hidden descendants still count as focusable. They can make the wrap fail or leave focus stuck. Exclude controls under inert or hidden from the drawer’s focusable set. Also exclude controls with hidden/collapsed visibility.

Implementation and regression ownership:

Finding Theme implementation Owning regression
Readiness assets/js/sidebar-state.js tests/js/sidebar-state.test.js; site tests/browser/community-feedback.spec.mjs snapshots the active path from both readiness signals in EN/ZH
Pending choice assets/js/command-palette.js tests/js/command-palette.test.js exercises pending extension → native choice → completion → available choice
Right rail assets/js/docs-shell.js Site tests/browser/community-feedback.spec.mjs covers EN/ZH collapse, Tab traversal, restoration, reload and aside relocation across desktop/tablet/mobile
Group keys assets/js/keyboard-nav.js tests/js/keyboard-nav.test.js covers LTR/RTL, arrows/WASD and link-only paging. Site community tests exercise real EN/ZH groups.
Drawer trap assets/js/docs-shell.js Site community tests collapse relocated groups and wrap Shift+Tab and Tab. They assert that focus never enters an inert or hidden subtree.

The readiness and pending-choice regressions failed against the previous implementation before their fixes. The right-rail browser assertions also failed in both English and Chinese before isolation was added. The repairs remain small: they change timing, move one pending guard earlier, add explicit rail isolation, and adjust tree focus targets and the drawer’s visibility filter.

These changes preserve the already accepted behavior: both root collectors honor sidebar_root_menu: false, and non-link groups retain their children. Pointer focus avoids incidental article outlines while keyboard cues remain visible. Search-tail callbacks retain their cancellation and handoff contract. The image preview fix continues to use an accessible name without inserting helper text into copied article content. Final regression results are recorded below.

Documentation readiness

The current documentation update covers 28 files in 14 EN/ZH pairs:

  • The six Design contract pairs use candidate-v1.1.0. They describe implemented main behavior without announcing a published release.
  • Navigation, layout and front-matter guides match current root filtering, divider groups and bilingual deployment paths. They also match the centered navbar, narrow-screen drawer and in-place navbar reveal behavior.
  • Organization and palette guides explain runtime load order, readiness and feature detection. They also explain site-owned persistence/integrations. Keyboard and image guides explain the fixes and the older-version boundary.
  • Installation guidance distinguishes the validation toolchain from the public version. Existing heading IDs remain stable. The new sidebar API heading has a matching Chinese ID.

The two 1.1.0 release-note files and two upgrade-guide files are also prepared. The release note remains a draft/candidate. Both home-page release entries point to the published 1.0 version.

Validation snapshot

Counts below describe the 2026-09-20 snapshot for theme 08f6563. Site acceptance uses the sibling checkout through a command-scoped module replacement. Local site checks used Hugo Extended 0.166.0, Node 26.9.0 and Playwright 1.62.1. Candidate CI used the pinned Hugo 0.165.0 toolchain. The 0.160.1 floor was checked separately with the official binary.

Check Result and scope
Baseline 75052f8 CI Passed all three jobs: pinned Hugo toolchain, browser runtime tests and Book publication. Exact baseline run.
JavaScript unit suite after the fixes Passed: 44 tests.
Official Hugo Extended 0.160.1 Passed the focused i18n and shell checkers. i18n covers 32 catalogs × 194 messages. The real documentation site also passed a production build against the local candidate with --panicOnWarning: 376 pages per language. The draft release is absent, and both home-page links point to 1.0.0.
Bilingual source and style checks Passed including this report: 129/129 page pairs, 988 source headings, Markdown style and git diff --check.
Final focused theme checkers Passed: shell, palette, keyboard and image zoom.
Final real-site non-browser suite make check passed all 57 tests. It checked 200 rendered content pages, 331 linked HTML pages, 39,803 internal links and 3,863 fragment links. Only the two expected Markdown goldens changed, for the release summary and documentation index. The floor production build also passed link checks: 327 pages, 39,059 internal links and 3,833 fragments.
Final browser suite make browser passed all 149 Chromium tests in eight suites. The suites contain 30 accessibility, 45 responsive/blog/palette, 16 keyboard, 14 content components and 18 code blocks. They also contain 4 scenarios, 5 theme-color and 17 community regressions. Coverage includes the full multilingual sitemap and six viewport widths. It also includes light/dark, forced colors, clipboard and no-script cases.
Agent documentation sample 93/100 (A) across 50 same-origin sampled pages. Sampled links resolve. 49 pages provide Markdown, and all 270 sampled code fences close correctly. The checker warns that the HTML llms.txt discovery hint is missing or too deep.
Rendered review Reviewed the Chinese release note in a desktop dark view and English in a narrow light view. Right-rail collapse removes its descendants from the accessibility tree. It also moves focus to the restore button. Restoration returns focus to the visible rail button.
Final theme revision and its CI 08f6563 passed all three jobs: Hugo 0.165.0, browser runtime tests and Book publication. Exact candidate run.
Public v1.1.0 tag, consumer upgrade and deployment Not performed.

Within this review’s scope, no unresolved implementation blocker remains. The candidate is ready for the publication steps below.

Repeat the checks from sibling checkouts. During development, keep the published dependency pin intact:

# From the theme repository
node --test tests/js/*.test.js
python3 bin/check-shell.py
python3 bin/check-palette.py
python3 bin/check-keyboard.py
python3 bin/check-image-zoom.py

# The site Make targets apply a command-scoped sibling module replacement.
make -C ../oink.pgsty.com check
make -C ../oink.pgsty.com browser

Remaining publication steps

Publication has not been executed. After the final revision passes acceptance:

  1. Finalize CHANGELOG.md, the release date and the release record. Publish the v1.1.0 tag and GitHub Release from the verified theme revision.
  2. Verify that the module proxy resolves that tag to the intended revision.
  3. Update the documentation consumer pin and version configuration together with its home-page release entry and contract status. Set the release-note state to draft: false in the same update.
  4. Rebuild and accept the documentation site using the published dependency, without HUGO_MODULE_REPLACEMENTS. Deploy it and verify the public routes.

A future improvement is an earlier, consistent llms.txt discovery hint for agents entering through HTML.

Limits

Browser verification ran in Chromium. The automated accessibility gate covers theme-owned surfaces while retaining its existing exclusions for vendored Redoc and Swagger UI, and Hugo 0.160.1 floor checks ran on the focused i18n, shell, and production build paths.

Publication follow-up

After this review, the v1.1.0 release was published on 2026-09-20 from 3a18234. Only the changelog differs from the accepted 08f6563 implementation. All three release-commit CI jobs passed before the annotated tag and stable GitHub Release were published.

A download with a fresh cache, using only the official Go module proxy, resolved the tag to that exact commit. Its .info, .mod, .zip, version-list entry and signed checksum record were verified. The module checksum is h1:121L5g57ChRCPyidzEBBcln2Co+0zYRQ+XDDXjymd0Q=. The go.mod checksum is h1:pHvbUhJCfseB41n5RGwsF7abT3i32VSTpofLQoq4b7Y=. The public records are the proxy version and checksum entry.

The documentation publication update pins v1.1.0 in go.mod and go.sum. It aligns the advertised version and both home-page release entries, and publishes both release notes. It also promotes the six contract pairs to released-v1.1.0. The historical acceptance tables above continue to describe the earlier sibling-checkout run. The site’s Site checks and Browser quality workflows track published-dependency validation separately, with both Go and Hugo module workspaces disabled.

Local validation of this published module passed all 57 non-browser tests and 26 focused Palette/community browser tests. It also passed the strict production build with 378 pages per language. The checks ran with GOWORK=off, HUGO_MODULE_WORKSPACE=off, and no HUGO_MODULE_REPLACEMENTS. The publication commit’s Browser quality workflow also runs the complete 149-test browser suite.

10 - CLI maintenance acceptance on 2026-10-03

Dated source and binary evidence for the local R1–R8/A18 program, preserving the initial audit, failed trials, and final acceptance scope.
Historical CLI maintenance acceptance ledger (2026-10-03 to 2026-10-04)

This record preserves the initial audit, intermediate trials, and local R1–R8 and A18 runtime/archive acceptance for the CLI maintenance roadmap. On 2026-10-04, the CLI surface was simplified to the Cobra/text/JSON/YAML command set defined in the current CLI contract and CLI introduction. These results apply only to the recorded historical source and binaries; later changes to the reduced CLI require separate validation.

Scope and evidence rules

The maintenance roadmap defines the authorized R1–R8 scope. The current CLI contract defines its compatibility baseline; the original roadmap does not add Docsy migration, version lifecycle, OpenAPI, theme publication, or the conditional E1–E4 extensions to this program. Hugo remains an external renderer. Generated sites remain ordinary Hugo projects.

Stages are accepted in dependency order. Each stage needs a complete usable flow supported by owning tests and relevant actual Hugo integration, together with known limits, a reviewable diff, and accepted EN/ZH contract and guide updates. New public behavior moves from the proposal into the owning contract only after its implementation and acceptance evidence exist.

In the tables below, existing, not rerun means code or a named test was inspected but its current runtime outcome was not established. Partial means the first candidate provides a reusable part of the required behavior. Open means new implementation or decisive acceptance evidence is missing. Passed, failed, unverified, and unsupported describe a specific executed input and scope when later runs are recorded.

Inspected inputs and tools

The initial 2026-10-03 audit read both repositories’ instructions and the documentation README and translation rules, both maintenance PRD languages, the original proposal, and the current CLI contract. It also inspected existing Go packages and test names. Execution was limited to version and Git inspection commands, without running the owning suites or writing consumer sources.

Input Observed initial state
Host and Go darwin/arm64; go version go1.27.1 darwin/arm64
Hugo hugo v0.166.0+extended+withdeploy darwin/arm64, Homebrew build dated 2026-09-09
Node and npm v26.9.0; 11.19.1; contributor/documentation tools, not CLI consumer requirements
Git 2.54.0 (Apple Git-157)
CLI source e623d93d589c49e5c58b8fae1bd5db720fc904cb, main; clean initial tracked/untracked status; generated bin/, dist/, tmp/ ignored
Documentation source 907d873eb05cfc2e194f492462dfa94849e93474, main; 184 initial porcelain entries, including existing proposals, contracts, guides, and unrelated content changes
Embedded Starter 137843b25bacd76ddd1f7ce71330bf2e3155b954; provenance and license already recorded by internal/starter
Declared theme baseline github.com/pgsty/oink v1.1.0 in Starter and the three selected sites; effective resolved bytes still require each acceptance run

The 2026-09-29 acceptance record provides baseline reproduction inputs from the first candidate.

Stage requirements and implementation evidence

Stage Required complete flow and invariants Initial implementation evidence Acceptance evidence still needed
R1 Shared page identity, languages, publication state, source provenance, actual outputs, translations and observed references from Hugo; oink.yaml owns check policy only; links/translations/style share analysis; severity and exclusions cannot hide required incompletion; trustworthy locations Partial: internal/site isolated snapshots and Page.OutputFormats probe, internal/outputcheck, internal/report; no shared translation/page facts or policy commands at initial audit Real Hugo routes, aliases, mounts, unlisted/generated-source cases and language relationships; public focused-check/policy cases; required unknown/tool/build/input failures remain 2; source locations only when reliable
R2 Three language layouts; strict/manual and localized policies; duplicate, missing and draft states; explicit versioned review records bind source language and source/translation hashes; bounded native syntax rules; effective-theme coverage; visible versioned baseline; reviewed fixes validate before narrow apply Open: no translation/review/native-rule/baseline public command at initial audit; rendered-reference checks remain reusable A04–A07; valid/invalid reviewed content corpus; no mtime review inference; disabled/localized languages handled; acknowledged findings stay visible; missing required checks stay incomplete; fix preservation
R3 Preserve thin default build/dev; build --check checks and manifests one strict Hugo output, exports only to new/empty target; digest/provenance manifest and optional minimal public identity; both local CI templates upload the same tree; release diagnosis; explicit-network public verification Partial: direct wrappers, strict isolated checks and licensed workflow inputs exist; managed build/export, digest verification, CI plans and public verify are absent at initial audit A08–A10; exactly one Hugo build; stale-byte rejection; revision/dirty/input/theme/tool/settings/coverage provenance without secrets or machine paths; workflow customization/conflicts/provenance and immutable source input; example address policy; fallback/language/resource/canonical/timeout/auth/rate-limit HTTP fixtures
R4 new, snippets and editor setup create ordinary inputs without overwrite; docs/blog/book/project profiles compose one licensed Starter; upgrades provide readable diff and old/new routes, aliases and enabled outputs; unsupported migrations give manual action; existing protections survive Partial: fixed archive language profiles and hash-bound single-site module upgrade with candidate validation, backups, dirty/workspace/replacement/vendor protection A11–A12; all new profile/language combinations build with ordinary Hugo; unknown editor settings retained; upgrade route/capability regression and readable diff; source provenance and licenses retained
R5 inspect, impact --since, bounded context, preview move; shared plans include touched files, diff, base hashes, translations, attachments, output/route changes and alias advice; candidate validation and stale/concurrent-safe recovery; ambiguous references require review Partial: module-specific upgrade plan/apply primitives; no shared content plans or inspect/impact/context/move flow at initial audit A13–A15; deleting B includes unchanged inbound A; translation/attachment/derived-output impact; uncertain/global changes force full checks; no content execution; candidate/stale/failed-write preservation and ambiguous-link handling
R6 Explicit versioned site registry reuses single-site engine; per-site and aggregate completion; writes only to selected sites; configured preinstalled markdownlint/Vale/lychee adapters normalize findings and declare syntax/network coverage Open: no workspace/adapter public command at initial audit A07/A15/A18; direct/per-site parity; no sibling discovery, implicit installation or default formatting writes; required missing tool 2, optional omission visible, external network uncertainty distinct
R7 Read-only loopback Studio with overview, issues, translation comparison, page relationships and publication views; filters, known sources, actual Hugo preview, comparisons and copied actions; CLI parity; prebuilt assets; explicit allowlist, separate preview origin, Host/Origin/session protection Open: no Studio server or assets at initial audit A16; browser/keyboard/screen-reader/mobile/light/dark/long-list flows; same underlying results as CLI; unauthorized hosts/origins/sessions and preview-to-management requests rejected; Node unnecessary for consumer runtime
R8 Markdown/text and front matter forms, selected components and collision-safe attachments reuse plans; authorized allowed writes with visible diff, hashes and candidate validation; no-op bytes and unknown fields/comments/order/encoding/whitespace retained; unsupported form syntax stays text Open: editing follows accepted read-only R7; no editor API at initial audit A17/A14; byte-identical no-op, surgical YAML field updates and text fallbacks; stale external-editor saves, traversal/symlink escapes and preview requests fail safely; attachments never overwrite; no management API in static publication

Evidence pages

The stage-by-stage receipts are kept on two companion pages, moved there unchanged on 2026-11-06 so each page stays small enough for people and agents to read whole:

  • R1–R5 evidence: local validation for R1 to R3, authoring and upgrade acceptance (R4), and implementation and documentation acceptance (R5).
  • R6–R8 and A18 evidence: workspace and adapter acceptance (R6), the read-only Studio candidate (R7), accepted reviewed editing (R8), and the 2026-10-04 runtime completion supplement (A18).

Acceptance case ledger

This ledger combines the initial audit with accepted R1–R7 evidence and the qualified R8 candidate gates.

Case Required outcome Code or checker evidence Status and missing decisive evidence
A01 One oink.result/v1 JSON result. Stderr logs. Policy 1, required incompletion 2 Protocol/public R1–R8 commands, frozen owning tests and exact-binary CLI/API reports. Unchanged result schema. Current eight owning gates/Linux qualification plus unchanged-runtime carry-forward of the preceding corpus in #a18 Passed supported current command scope
A02 Hugo resolves slug/url/permalinks/aliases, mounts, unlisted pages and language roots Real PageFacts/manifest/custom-mount/translationKey fixtures. Preserved ordinary artifacts. Final consumer facts R1 scope passed
A03 Definite local missing routes fail. Outside origin/path and declared external scope classified honestly Actual rendered-reference fixture plus subpath/policy regressions and final real sites Passed the required A03 scope. External availability remains unchecked
A04 Filename, directory and translationKey. Duplicate/missing/draft cases. Strict/localized policy R2 translation engine, actual Hugo/public commands, final reports and numeric supplement Passed R2 required scope
A05 Absent record unknown. Changed source/translation hash visible. No mtime inference R2 hash/status/diff, public preview/apply and final reports Passed R2 required scope
A06 Real fences, inline code, shortcodes, HTML, attributes, unknown fields and protected text boundaries R2 actual syntax/provenance fixtures, reviewed corpus and final reports Passed R2 required scope. Catalog and unsupported-source limits remain explicit
A07 Acknowledged findings visible. New findings block per policy. Required unavailable tools cannot pass R2 baseline/public plans. R6 fake/actual protocol, missing/unsafe/offline/network-uncertainty and required-precedence fixtures passed Supported scope passed. Required unavailable/uncertain tools remain 2
A08 Post-check bytes invalidate manifest. Provider uploads verified tree without another build R3 manifest/export/tampering/public one-build tests. Final ordinary-Hugo comparison and provider rehearsal Passed R3 required local scope
A09 Both CI templates. Custom workflows preserved. Permissions/variables/provenance and stale plan protection R3 offline generation/bootstrap, public preview/apply/stale-input tests, custom workflow supplement and local rehearsal Passed R3 required local scope. Custom workflows remain unknown and unchanged
A10 Reject HTTP 200 fallback, wrong language/build, missing resource/canonical mismatch. Incomplete timeout/auth/rate-limit R3 explicit-network local HTTP and public result fixtures, including required identity absence Passed R3 required fixture scope
A11 All declared profiles/languages. Target protection. Ordinary Hugo. Unknown editor settings retained R4 24 ordinary Hugo/public profiles, full Starter authoring/editor flow, snippets, actual mounts, source identities, JSONC preservation and external schema reproof Required supported R4 local implementation/corpus scope passed. Documented unsupported editor inputs remain explicit
A12 Readable diff and route comparison. Dirty/workspaces/replacement/vendor. Recovery/concurrency Frozen actual-Hugo seven synthetic pinned cases, public upgrade, source/external guards, observed alias retarget and guarded partial rollback Required bounded R4 local implementation/corpus scope passed. Unknown redirects/multihost remain incomplete
A13 Deleted B finds unchanged inbound A. Translations/attachments/derived outputs. Global full scope R5 committed Git/actual Hugo deletion, alias-inbound, global/uncertain input and unavailable-baseline fixtures. Exact-binary consumer reports Passed required supported R5 scope. Unavailable or unproven historical inputs stay explicit 2
A14 Candidate before apply. Stale/hash/write failures preserve later edits. Ambiguous references unchanged R2/R4 shared safety, R5 full-mode/inventory moves and R8 regenerated intent/fresh-input candidate validation, guarded writer and stale/late-editor/attachment tests. Current eight owning gates/Linux qualification plus unchanged-runtime carry-forward of the preceding corpus in #a18 Passed supported R5 CLI and R8 CLI/Studio editing scope. Ambiguous or unavailable required inputs still block
A15 Workspace/direct parity. Selected writes only. Bounded context with paths/versions/reasons. No content execution R5 bounded captured-source/context fixtures and four-site queries. R6 registry/direct/aggregate parity and explicit-name saved apply with other sites preserved Supported context/workspace scope passed. No implicit batch writes
A16 Five useful CLI-parity views. Keyboard/mobile/light/dark. Source/preview isolation Accepted R7 evidence retained. Historical source-bound R8 read-only 14 axe/screenshots and Editor 12 axe/screenshots with unchanged UI bytes. Current backend gates, ARM and corpus separately verified, native/API parity, preview isolation, four consumers and canonical render passed. Current eight owning gates/Linux qualification plus unchanged-runtime carry-forward of the preceding corpus in #a18 Passed supported local views. Required partial-preview incompletion/native findings remain visible
A17 No-op bytes. YAML unknown/comment/order preservation. Stale-save and attachment collisions rejected Corrected frozen core/public/guarded-writer race, actual Hugo/tools, source-bound Editor/read-only browsers, exact-binary four-consumer proposal parity/preservation and guarded canonical source/render passed in #r8. Current eight owning gates/Linux qualification plus unchanged-runtime carry-forward of the preceding corpus in #a18 Passed supported local editing scope. Required partial preview/native findings still block Apply. Final A18 separate
A18 Actual declared macOS/Linux runtimes. Child signals. Provisioned offline runs. Honest unsupported inputs Current freeze/source and repeated five-archive reproduction. Darwin arm64, native Linux arm64 and emulated Linux amd64 nonroot ext4/full offline unit-vet/selected Hugo/native archive/signal smokes passed in #a18 Passed current declared runtime/archive scope. Optional guest prerequisites remain explicit skips. Darwin amd64 is experimental/unverified and Windows outside scope

Candidate sites and source preservation

The selected acceptance inputs include the embedded Starter and three distinct maintained consumer sites, reusing the historical corpus without writing consumer sources. The Starter source checkout provides provenance, while generated profile trials use disposable directories.

Input in the sibling checkout layout Initial observed identity and purpose Current candidate acceptance
oink-starter / generated Starter Source 137843b, two initial status entries. Licensed fixed archive, language/profile/root/subpath trials R1 bilingual init/check and R4 all-profile ordinary/public authoring flows passed. Archive/license unchanged
oink.pgsty.com Source 907d873 with existing changes. Bilingual documentation/regression and explicit local-theme trial Final R1 check and source preservation passed. Local-theme evidence remains distinct from public-pin evidence
pig.pgsty.com Source 75050c0, five initial status entries. Root Docs/Blog rewrites and nonrendering sidebar entries. Declared v1.1.0 Final R1 check and source preservation passed
repo.pgsty.com Unborn main, no HEAD revision. Materialized untracked sources, generated catalog and declared v1.1.0 Final R1 check and source preservation passed. Revision remains unknown

For each run, record effective module source and versions, flags/network policy, exit/result/coverage, raw evidence location, and preservation outcome. Before/after inventories must include all tracked and non-ignored untracked source bytes and modes, Git status/index state, workspace/replacement files, and effective vendor inputs. Compare exact inventories. Keep reports, isolated candidates, output and caches outside consumer sources and outside Git. Full-build timing comparisons must use the same current input baseline.

Owning checks and documentation gate

Start with the smallest affected Go packages and public behavior tests. The existing repository gates are make test (offline tests and vet) and make test-hugo (actual Hugo Starter, snapshot, manifest and public command fixtures). The owning Hugo gate now runs all owning packages without the old narrow test-name filter. New fixtures must remain in that gate. Use a race run for concurrent plan/server changes when the focused tests justify it. Unit fixtures remain offline. Networking requires explicit invocation.

For documentation, preserve EN/ZH heading number, order and stable explicit IDs. Use these narrow source checks:

node scripts/check-markdown-style.mjs content/docs/design/research
node scripts/check-doc-translations.mjs

Both source checks passed after this record and its Chinese peer were added. Eight Chinese research files passed the style checker. Translation source coverage was 137/137 pairs with 1,082 source headings.

After building the relevant site, complete the rendered documentation gate:

npm run _check:markdown-style
npm run _check:translations
npm run _check:rendered-markdown
npm run _check:rendered-links

make build validates the declared published pin. make check selects the sibling theme for the complete non-browser regression suite. Studio uses its own actual browser and accessibility checks.

Delivery state and remaining limits

State Current completion evidence. History retained above
Local implementation Finite R1–R8 supported implementation completed locally, including Studio/read-only and opt-in reviewed editing. Current A18 runtime/archive scope passed. Canonical lifecycle rendering is bound separately to these exact bytes
Local validation Historical R1–R8 owning/browser/corpus/render records retained. Current 2026-10-04 backend correction and eight current owning gates, actual three-target runtime/archive checks and unchanged-runtime carry-forward of the preceding four-consumer preservation/parity passed in #a18. Required repository findings/partial preview remain visible. Rendered navigation/URL checks have a separate exact-byte receipt boundary
Commits Baseline CLI commit identified
Archive and runtime qualification Current corrected source: Darwin arm64, native Linux arm64 and QEMU-TCG-emulated Linux amd64 passed installed archive/offline/signal/filesystem flows. Two fresh builds reproduce all five archives. Darwin amd64 remains experimental/unverified after actual failed execution
Public distribution and consumer adoption Not performed in this local validation run
Deployment and public-content verification Verified against local HTTP fixtures

The finite R1–R8 implementation and required current A01–A18 runtime/archive scope have local acceptance evidence. Known repository findings and required preview incompletion remain visible diagnostic outputs.

11 - CLI maintenance acceptance on 2026-10-03: R1–R5

R1–R5 local validation and acceptance evidence from the dated CLI maintenance record.

These R1–R5 sections of the dated record were moved here unchanged on 2026-11-06 so each page stays small enough to read whole. The scope, evidence rules, case ledger and delivery state remain on the main page.

R1 local validation

R1 now provides shared Hugo page/translation/source facts, rendered reference and anchor evidence, strict oink.policy/v1 input, check links, --format json, visible reviewed exclusions/external scopes and required-work precedence. Translation and style selections explicitly return required unsupported coverage while those engines remain unimplemented. Default build/dev remain direct Hugo operations.

Requirement Executed evidence Current outcome
Public result/policy and incomplete precedence make test: all packages and vet; public severity/exclusion/unimplemented-group/JSON-alias tests; TestEveryRequiredUncompletedCoverageFails Passed R1 scope; any required uncompleted status, including not_checked, remains 2
One build and shared facts TestPublicCheckSharesOneBuildAndRenderedFacts Passed; one strict Hugo build supplies page and observed target/anchor facts
Hugo authority and source mapping Actual TestPageFacts* fixtures: translationKey, actual routes/aliases, unknown generated nodes, custom mounts, excluded-page analysis and failure preservation Passed; separate analysis preserves production facts/artifact bytes and source bytes/modes
Repeatable real Hugo gate Corrected make test-hugo includes TestPageFacts* and TestHugoRendered*, alongside Starter and manifest fixtures Passed; scoped route/reference, reviewed external-scope and original-output preservation cases
Fresh Starter Bilingual init, check links, ordinary strict Hugo using isolated provisioned v1.1.0 module archives Exit 0; 223 files, 4,461 references, 66 page facts; dependency preparation remains explicit
R1 documentation source and schema Markdown style under content/docs; bilingual source checker; JSON parse and equality of CLI/docs result schemas; scoped diff whitespace check Passed: 88 Chinese docs, 137/137 source pairs and 1,085 headings; schemas remain additive oink.result/v1 with exits 0/1/2
Final candidate reports and rendered EN/ZH Refreshed current-binary consumer reports; actual rendered source/Markdown/link owning checks below R1 scoped gate passed; existing draft-release omission in production is recorded separately

Earlier offline R1 trials returned 0 without diagnostics on three consumers:

Earlier trial Source files Built files HTML files References Page facts Bytes/modes/Git inventory
OINK documentation 421 1,139 512 74,689 341 Exact before/after equality
PIG project site 858 1,392 424 64,440 248 Exact before/after equality
Repository catalog 2,294 3,287 1,635 851,535 1,572 Exact before/after equality

These earlier reports spell optional unselected coverage not_selected, outside the existing result-schema enum. The final code corrects it to not_checked and includes project.pages coverage. Their measured counts and exact inventories remain valid observations of the earlier binary. The final reports below establish final JSON conformance. Raw evidence stays in task-named local acceptance directories outside consumer sources.

The final R1 binary was rebuilt from the dirty CLI working tree based on e623d93d589c49e5c58b8fae1bd5db720fc904cb. The recorded input inventory includes file hashes, modes and Git-status identity. Its SHA-256, computed over sorted JSON serialization, is 518260f07f3c916468ee3d56c4eeca03c131514155aa82539564ccd2f3c1f664. The exercised binary SHA-256 is 3deb7e357fc86f6907df60da0769d93f2d41ba5e01949b641548a67d7f459d12.

Final current-binary trial Source files Built files HTML files References Page facts Acceptance
OINK documentation 421 1,139 512 74,755 341 Exit 0, valid result, complete page facts, exact source bytes/modes/Git preservation
PIG project site 858 1,392 424 64,440 248 Exit 0, valid result, complete page facts, exact source bytes/modes/Git preservation
Repository catalog 2,294 3,287 1,635 851,535 1,572 Exit 0, valid result, complete page facts, exact source bytes/modes/Git preservation

Each final result has required check.links: complete and project.pages: complete coverage. Unselected translation/style coverage is optional and reports not_checked. The final offline make test and vet passed; the corrected actual-Hugo owning target also passed. The recorded tools remain Go 1.27.1, Hugo Extended 0.166.0, Git 2.54.0 on macOS arm64. The three exact before/after inventories were independently compared while preparing this record.

Production output passed rendered Markdown and link checks. Its global translation checker returned 1 solely because the pre-existing draft content/blog/release/1.2.0.md / .zh.md pair is absent from production. The changed R1 pages rendered in both languages. A separate explicit analysis build set HUGO_BUILDDRAFTS, HUGO_BUILDFUTURE and HUGO_BUILDEXPIRED to true. All three owning checks passed: 137/137 paired sources, 1,085 headings, rendered Markdown and rendered links. That view checks excluded sources in isolation while keeping production output unchanged.

R2 local validation

The local candidate now implements translation policy/status/diff/hash review, syntax-bounded native content rules, visible reviewed baselines and shared oink.plan/v1 preview/validate/apply. Default check requires links, translations and style. Production output and the explicit draft/future/expired analysis are separate. Stable behavior and examples are in the contract. The R2 local gate passed owning checks, final frozen-input consumer reports and rendered bilingual documentation. The exact exercised binary and the subsequent narrow equality fix are recorded below.

Requirement Executed owning evidence Outcome and limit
A04 translation relationships/policy Actual TestHugoFilenameDirectoryAndTranslationKeyLayouts; scope, duplicate/missing/disabled-language, draft, strict/localized and selected-constraint tests Passed owning tests; no universal heading/code/localization parity
A05 explicit review and diff Full byte hash/current/source/translation/both-changed, mtime-independent, unknown/unreadable/ambiguous and malformed-record tests; public status/diff/review preview/apply fixtures Passed owning tests; review state is change evidence, not semantic judgment
A06 source boundary and provenance Actual Hugo enabled/disabled canonical title/block attributes and configured passthrough fixtures; front matter/CRLF/BOM/shortcode/code/HTML tests; every public v1.1.0 source/license SHA verified Passed owning tests; unsupported syntax remains incomplete and custom hooks remain outside catalog attestation
A07 baseline scope Capture/visible acknowledgement/new finding/incomplete precedence and malformed-record tests; public baseline preview/apply fixtures Passed R2 baseline scope; external tool adapter acceptance belongs to R6
A14 shared metadata plans Stale bytes/modes/existence/guards; edits during validation; exclusive commit collision; partial restore; later editor bytes/modes/deletion; old open inode write; new-directory children; confinement/identity/diff tests Passed owning tests and vet; candidate/source overlap refused; later move/reference ambiguity remains R5 scope
Frozen runtime gates macOS arm64 make test/vet, owning actual Hugo and focused race runs Passed; logs /tmp/oink-r2-frozen-go-gate.log, /tmp/oink-r2-frozen-hugo-gate.log, /tmp/oink-r2-frozen-race-gate.log; final all-owning-package Hugo gate /tmp/oink-r2-owning-hugo-final.log explicitly includes configured passthrough
Bilingual documentation Narrow source style/pairing/IDs, equal result schemas, scoped whitespace and actual production/analysis node checks Passed scoped gate: 88 Chinese docs, 137/137 source pairs and 1,092 headings; production draft omission separately recorded below

The frozen parser corpus at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r2-source-corpus-lqx25kwr/summary.json records parser input SHA-256 a601200ec4fe275d2bd4baf11d4db7d46a2cc6f1674900c1fd801769e55d12de. Each scope parsed completely with zero findings. The core uses actual Hugo site-source identities from the recorded configuration. Supplemental Markdown includes disabled/unpublished files. It does not invent routes or relationships. These captures precede the authorized R2 documentation edits.

Corpus Unique actual Hugo source files Supplemental local Markdown Source inventory files Bytes/modes/Git
Starter 52 78 97 Exact before/after equality
OINK documentation 272 274 421 Exact before/after equality
PIG 212 212 858 Exact before/after equality
Repository catalog 1,568 1,572 2,294 Exact before/after equality

Preliminary public-command reports at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r2-final-qu_zprps/summary.json used binary c8d87e6d73d3101fefcb62c5d6845518573c02c400f474dc9f4603afafc774d5 and CLI input inventory d8a75be0e074365a4164b7aaaa27d82a1e844e04406a36c3dd6d39ff2b6e873f. They precede the final parser/doc freeze and are not final acceptance evidence. The initial Starter invocation selected the enclosing evidence folder and returned 2. This was a validation setup error. Selecting its actual site child returned 0, with evidence in /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r2-starter-27vmi0w5.

Preliminary check Exit Page facts Built files References Translation statuses Outcome
Correct Starter child 0 66 223 4,461 28 Complete; 97 source files/inventory unchanged
Documentation 0 341 1,139 74,755 144 Complete; 421 source files/inventory unchanged
PIG 0 248 1,392 64,440 120 Complete; 858 source files/inventory unchanged
Repository catalog 1 1,572 3,287 851,535 788 Completed policy check: 10,462 actual HTML_ID_DUPLICATE findings in existing merged-print output; 2,294 source files/inventory unchanged

The repository catalog run completed with policy findings from existing duplicate HTML IDs in merged-print output; informational review states remain visible.

The final frozen-input reports at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r2-candidate-6xzcs7fk/summary.json exercise binary SHA-256 ff88b407a6cddb9007f94275c65a80ed4c9c4fd13f5e821f9b7a4a8973abaa56 from CLI input inventory bd8c71b55250a89dc15c7534924bb82a5447d6f2628eba23c8cb3d864309ee9f. All four exact source byte/mode/Git inventories were independently compared equal before/after. These reports supersede preliminary public-command trials:

Final check Exit Source files Page facts Built files References Translation statuses
Starter 0 97 66 223 4,461 28
Documentation 0 421 341 1,139 74,825 144
PIG 0 858 248 1,392 64,440 120
Repository catalog 1 2,294 1,572 3,287 851,535 788

No final report has incomplete diagnostics. The repository catalog retains 10,462 actual merged-print HTML_ID_DUPLICATE findings and 788 informational review states; the other sites retain informational unknown review states.

Kept production docs at /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-site-2201601475/public passed rendered Markdown and links. Global translation checking returned 1 only for the existing draft release 1.2.0 pair absent from production. The separate draft/future/expired analysis at /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-site-3698773232/public passed all three node checks: 137 pairs/1,092 headings, 216 content pages/41,586 text nodes, and 347 pages/48,682 internal links/4,171 fragments. Evidence logs are /tmp/oink-r2-docs-production-{translations,markdown,links}.log and /tmp/oink-r2-docs-analysis-{translations,markdown,links}.log.

A final review identified optional equal_fields comparing JSON 7.0 with YAML/TOML numeric 7 by representation. The narrow supplement now normalizes decoded numeric values recursively to an exact rational number tag, preserving strings versus numbers, map keys and array order. Tests cover decimals/exponents, negative zero, integers beyond float64 precision, and nested differences. They also cover source byte preservation and required incompletion for unrepresentable values. Actual-Hugo translation tests passed in /tmp/oink-r2-numeric-translations-gate.log; all public maintenance actual-Hugo cases passed in /tmp/oink-r2-numeric-public-gate.log; owning vet and whitespace checks passed. Supplemental source SHA-256 values are:

Source SHA-256
internal/translations/check.go 24664377e14b4ae2fc554d0d7fde2ec33cc987707250e130fd88d9a25d5e1637
internal/translations/translations_test.go f58f4a305fe9fe3f5500ddfcf85faf3cfa37d72f8c220a1cb16ce4ccfbddb74d

The frozen real-site reports and Linux qualification predate this supplement; because those sites configured no numeric equality constraint, the fix does not affect their recorded outputs.

At this stage, macOS arm64 was exercised; an attempted Darwin amd64 runtime on this host failed with arch -x86_64 / posix_spawn: Bad CPU type in executable (/tmp/oink-r2-darwin-amd64-gate.log) because Rosetta was not installed. Both native Linux arm64 and Docker Desktop Rosetta-emulated Linux amd64 executed the same runtime/schema/license inputs. Their SHA-256 is 0f786df68ef3c4844c983a51595f79242d1cb1d2bf6c5b5eb7f2c6415fb8d861. Evidence is retained at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-a18-linux-ajbbnvki in arm64-results, amd64-results, commands.json, candidate-inputs.json, preparation.json and qualify.sh. Each target passed 270 test/subtest cases with zero failures and two optional external corpus/provenance skips. Checks included full actual-Hugo go test ./..., vet, and built CLI version/bilingual init/doctor/full check/translation status, plus missing-Hugo exit 2 smoke. JSON stdout and source byte/mode inventories were verified. Go 1.27.1 ran on Linux arm64; Hugo Extended 0.166.0 architecture assets were SHA-verified.

R3 local validation

R3 adds managed build --check, oink.artifact/v1 sealing/export/local verification, explicit-network HTTP verification, release diagnosis and guarded local CI generation. The default build/dev path remains ordinary Hugo. The executed runtime and paired contract/guide gates passed.

Requirement Executed owning evidence Outcome and limit
A08 one checked artifact Public fake/actual Hugo one-renderer tests; exact export, manifest/marker, post-check byte/mode/missing/extra/symlink tampering, failure/concurrency and source-preservation tests Passed local scope; a failed/incomplete check cannot seal/export; local artifact verification does not rebuild
A09 both CI providers Offline deterministic generation, pinned source/Hugo archives and action revisions; safe bootstrap archives; guarded public preview/apply/stale-input cases; actual-Hugo original-input binding Passed local configuration scope; every existing generated target is refused and custom workflows remain unchanged
A09 upload identity Both local provider rehearsals and TestProviderUploadRehearsalPreservesActualSealedManifestIdentity Passed: one managed build, separate verification, then the same tree; GitHub tar includes the hidden marker, Cloudflare rehearsal receives that verified directory; no provider upload executed
A09 custom workflow diagnosis Generated-plus-other-custom and standalone-custom/no-metadata public tests, actual-Hugo preview and owning vet Passed supplement in /tmp/oink-r3-ci-custom-owning-gate.log and /tmp/oink-r3-ci-custom-vet-gate.log; each unrepresented workflow stays unknown, informational and optional release.ci: not_checked, including beside valid generated metadata
A10 deployed identity Local HTTP fixtures for all recorded files/routes/languages, marker, canonical/base/inert-template behavior, HTTP 200 fallback, wrong bytes/language/build, missing resources/Markdown/search JSON Passed local fixture scope; definite mismatches return 1; browser JavaScript is explicitly unchecked
A10 unknown network state Explicit network/credential refusal, timeout before headers/during body, authentication/rate-limit/server errors, required marker absence, bounded body/gzip and redirect/no-cookie fixtures Passed local fixture scope; incomplete states return 2 with remaining requests unknown; no public deployment was contacted
Frozen runtime gates Full tests/vet, owning actual Hugo and focused race checks Passed on macOS arm64 in /tmp/oink-r3-frozen-go-gate.log, /tmp/oink-r3-frozen-hugo-gate.log, /tmp/oink-r3-frozen-race-gate.log; the subsequent custom-CI change has the focused supplement above

The latest single-binary corpus is retained at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r3-ci-final-ahc4csjk/summary.json. It compiled an exact captured CLI input copy, with binary SHA-256 425845c1d2db7b1cd3c3cdb5f28475cb06ba6f656054909759e2359a39925dd2, 67 runtime/schema/license inputs SHA-256 6789a3a0235ff8d81453b9bfde37824979eac7d56af3710da390e4d2ef8479dc, and 108 broader CLI inputs SHA-256 4ca473a4cb586d232baeb4cee029b281469c5bb03c831cc199b95451e6832c60. Runtime inputs remained exactly equal after all runs. Live tools were Go 1.27.1 and Hugo Extended 0.166.0 on Darwin arm64. The manifest’s normalized Hugo version excludes vendor build text and private paths.

Each run used offline build --check. It used fresh external destination/manifest paths, the optional marker, and retained isolated work. These existing local consumer inputs were checked without --release. Their configured workspaces were preserved. Every raw report records exactly one strict Hugo renderer, zero incomplete diagnostics and zero required unfinished coverage.

Final managed build Exit Source files Copied source inputs Page facts Built files References Exported files Local artifact verify
Starter 0 97 94 66 223 4,461 224 0
Documentation 0 421 427 341 1,139 74,825 1,140 0
PIG 0 858 861 248 1,392 64,440 1,393 0
Repository catalog 1 2,294 2,299 1,572 3,287 851,535 None Not exported

Both inventories compare exact bytes, modes and file types before/after; the primary inventory also compares logical Git state. Git sites include tracked and non-ignored untracked sources; the non-Git Starter includes its existing generated files and lock. The supplemental copied-source inventory also includes ignored workspace/editor metadata read by snapshots, excluding existing generated output/cache trees. All four comparisons were exactly equal.

The repository catalog retains 10,462 existing merged-print HTML_ID_DUPLICATE findings, so neither destination nor manifest was created. Production review states number 28/143/120/786. Analysis includes unpublished pages, which explains the earlier R2 144/788 counts. Existing custom workflow information remains visible, and Starter’s example address is a warning in this non-release run.

The three fresh exports match the retained independent ordinary-Hugo trees in every original file’s SHA-256, size and mode. The sole extra file is .well-known/oink-build.json. Their original source inventories and complete manifest input hashes match the earlier capture, so reusing those ordinary trees does not substitute different inputs. The helper source SHA-256 is 13e4957a3d7847eb28c8b1eeba3588a4f4a9982c2bfca2ebc729ab2827159607; its binary SHA-256 is b2699fe7a7aa3a34c41f9e4aba4b22d39cf8d0c156a369f3dfc4ca8c8c0fbce5. Raw helper and ordinary evidence are in /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r3-candidate-wb643dhh; the temporary compilation source was removed after building the helper.

Earlier R3 captures preceded the runtime freeze (oink-r3-final-pisrr21h preceded custom-CI diagnosis), and an initial /Users/vonng/pgsty/PIG target selection returned 2 before switching to pig.pgsty.com.

The scoped R3 documentation gate passed at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r3-docs-render-pljj5aqd/summary.json. Ten paired contract/guide/roadmap/index/overview edits were installed only after the original bytes/modes matched. Their recorded hashes are at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r3-doc-drafts-s0b3g48l/applied-files.json. Source style passed 88 Chinese docs; translation coverage passed 137 pairs and 1,099 headings; schemas remained equal and scoped whitespace passed. Actual production Markdown passed 214 pages/41,871 text nodes; links passed 345 pages/48,344 internal links/4,171 fragments. Production translations returned 1 only for the unchanged draft release 1.2.0 absent from production. A separate draft/future/expired analysis passed Hugo and all three owning checks: 137 pairs/1,099 headings, 216 pages/42,177 text nodes and 347 pages/48,720 links/4,199 fragments. All 421 canonical and 488 copied source files retained exact bytes/modes throughout these rendered checks.

R4 authoring and upgrade acceptance

The supported R4 implementation and read-only corpus scope are locally accepted after frozen owning/full gates. This record covers profiles, ordinary authoring/editor/snippets and bounded upgrade views, along with guarded canonical documentation promotion and fresh scoped rendered validation.

Executed profile evidence Outcome and limit
One fixed licensed archive Snapshot verification against commit 137843b25bacd76ddd1f7ce71330bf2e3155b954 passed without --write; archive SHA e55bde279715f6d8d19d3d88671a2cf7561b515be46915b0f12c640d0ce1d958 and MIT license unchanged; projection metadata/script match
Composition and preservation Default/explicit project byte parity; selected archived model/localized home, invalid profile, nonempty target, concurrent validation/publication and cancellation recovery tests passed; unit/vet/race gates passed
Ordinary Hugo All four profiles × three language choices × root/subpath passed 24 actual warning-strict offline builds using provisioned public OINK v1.1.0; complete source byte/mode/no-extra-file and rendered-reference checks passed
Public init workflow Four profiles with en/en,zh, actual subsequent root/subpath Hugo URL facts/checks, workflow/license preservation and default parity passed; unknown/nonempty refusals 1, missing/failed Hugo 2, empty/absent targets and pure JSON/separate logs verified
Public authoring and source identity Actual candidate/apply/ordinary Hugo, review-unknown and source preservation passed in /tmp/oink-r4-authoring-public-gate.log; fresh-directory/site guards and vet passed in /tmp/oink-r4-new-input-race.log and /tmp/oink-r4-public-core-vet.log. Actual ignored input refuses 2 without a saved plan or source writes even when source groups are disabled; selected draft peers still force analysis identity in /tmp/oink-r4-authoring-sourceproof-gate.log. Supported owning scope passed
Bounded upgrade owning gate Seven actual-Hugo synthetic pinned module-fixture cases, observed-stream digest/inventory fidelity, independent cross-page alias-retarget blocking, source/concurrency/exclusive installation and later-edit rollback protection passed under race; vet passed. Final hardening logs /tmp/oink-r4-hardening-owning-gate.log, /tmp/oink-r4-hardening-final-focused.log, /tmp/oink-r4-hardening-vet.log; final public/full frozen gates passed
Integrated authoring/editor hardening Actual-Hugo language-directory plan/apply/ordinary builds, link/never new-source refusal, external schema/license/full-mode/module identity and legacy schema reproof, shared translation/baseline/CI regression and full Starter docs→new draft peer→editor→check→ordinary Hugo flow passed. /tmp/oink-r4-app-authoring-hardening-gate.log (58.241s), focused race and vet passed; post-candidate external mutation proof /tmp/oink-r4-app-external-during-validation.log passed. Opaque saved external-input hashes and canonical workspace-origin guards passed /tmp/oink-r4-external-plan-binding-final.log, /tmp/oink-r4-workspace-origin-gate.log and their vet logs. Frozen full-stage, corpus and scoped canonical rendered documentation gates passed
Actual language mounts Standalone ordinary per-language contentDir and explicit site-matrix fixtures each passed config/mounts/strict-render with source bytes/modes unchanged; Hugo0.166 emits sites.matrix.languages and distinct physical files with reciprocal public translations. /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r4-language-mounts-lgmve1sk/summary.json; public actual language-directory plan/apply/ordinary-Hugo integration passed

Owning evidence is retained at /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r4-starter-owning-0pv41lw5/summary.json. Logs are /tmp/oink-r4-starter-{unit,hugo,vet,snapshot,race}-gate.log and /tmp/oink-r4-public-init-gate.log, /tmp/oink-r4-public-init-vet-gate.log. Generated source counts are project 94, docs 58, blog 40 and book 34.

Final frozen gates all returned 0: make test/vet /tmp/oink-r4-frozen-go-gate.log, make test-hugo /tmp/oink-r4-frozen-hugo-gate.log and actual-Hugo core race /tmp/oink-r4-frozen-core-race-gate.log. The final public flow includes Starter docs → primary/translation draft → editor → check → ordinary Hugo; post-candidate external schema mutation still refuses before source writes. Seventeen owning and three final gate logs are retained verbatim with hashes in the final corpus’s owning-gates.json.

The exact four-site evidence is /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r4-corpus-lw2cjyq6/summary.json with bounded summary.compact.json, raw JSON/logs and per-command inventories. Binary SHA is c169b3d4d046c811dca80867068b86fb66ada5c8ce6910dd5cda7353c406f377; 82 runtime-input files bind SHA fdff7f50d49b44f03fa1db79eec6b6c9b9bd5e52f8967a88aed84b3207a7b3c6, which equals the final root inventory with no runtime changes. The broader 139 CLI inputs bind SHA 562e838d9eccb628eac86ae59b9b9587c1e23ad52991ec50eafb1e604e3924da. Driver SHA is d3ac41dc2e18295bfb26134d1a696935c8174913e2801a5766dbf7a1139d89f8. Actual tools were Go 1.27.1 and Hugo 0.166.0 Extended on macOS arm64.

Frozen consumer Primary/copied source files Pages; output files; references Managed build / artifact verify Read-only upgrade / new / editor
Starter 97 / 94 66; 223; 4,461 0 / 0; 224 exported files including marker 0 / 0 / 0
Documentation 421 / 427 341; 1,139; 74,937 0 / 0; 1,140 exported files including marker 0 / 0 / 0
PIG 858 / 861 248; 1,392; 64,440 0 / 0; 1,393 exported files including marker 0 / 0 / 0
Repository 2,294 / 2,299 1,572; 3,287; 851,535 Completed finding 1; no export/manifest Completed finding 1; new/editor not attempted after blockers

Each managed build used exactly one strict production Hugo render and had no required incompletion or uncompleted required coverage. Primary Git-visible source bytes/full modes/logical Git state, supplemental copied inputs and source directory modes matched exactly before/after every command and each complete site flow. Repo’s 10,462 existing merged_print duplicate HTML IDs remain visible.

Consumer upgrade previews selected the available public v1.1.0 pin and did not apply writes. Cross-version route/alias/output regressions use explicit synthetic fixture pins. New/editor plans were validated previews without saving or applying consumer plans. Multi-host and unknown relative-alias identities stay incomplete, and nondeterministic output requires a fresh v2 plan preview.

Fresh canonical documentation acceptance passed after applying the ten guarded files. Evidence is /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r4-docs-render-v01eima0/summary.json; the promotion manifest is /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r4-doc-drafts-3i8bw994/applied-files.json. The frozen c169b3… CLI performed one strict production render and its focused link check returned 0.

Fresh canonical documentation gate Executed result
Source owners Translations 0: 137/137 pairs and 1,104 headings; complete canonical style 0: 137 Chinese files, 181 strong spans, no emphasis; ten-file whitespace check and public JSON schema equality passed
Production rendered Markdown/links Both 0: 214 content pages / 42,214 text nodes; 345 pages / 48,360 internal links / 4,187 fragments
Production translations 1 solely for the pre-existing draft content/blog/release/1.2.0.md absent from production; no new pairing/heading finding
Separate nonpublishable analysis Fresh ordinary Hugo with the actual original snapshot environment/rebased paths and explicit draft/future/expired flags returned 0; all three owners 0: Markdown 216 pages / 42,520 nodes, links 347 pages / 48,736 links / 4,215 fragments, translations 137/137 pairs / 1,104 headings
Source preservation Canonical 421 Git-inventoried files and 427 copied inputs retained exact bytes, modes, Git state and directory modes; production copied 428 and analysis copied 427 files remained unchanged through their checks; analysis build also retained copied full modes

The temporary helper copied the frozen core without modifying it: helper source SHA 7faea7e726a6c6fb2e0747be1a4428f4c5fb5734fa52b6f981157a5fe37d9989 and helper binary SHA 532638e76f96f8b173c122e512b3bf5fc2c4d4a7130f59c99c2c69e135e87073 are retained with raw logs.

R5 implementation and documentation acceptance

R5’s supported local scope is accepted after focused public/core, corrected frozen full-stage, exact-binary read-only consumer and guarded canonical source/rendered documentation gates.

The actual public Git/Hugo flow at /tmp/oink-r5-public-final-flow.log passed in 53.963 seconds. Its committed synthetic site owns its local theme, bilingual pages and binary attachment. Ordinary modes 0640 and 0600 remain full current facts. Historic Git comparison uses only executable bits. Deleting B selects unchanged inbound A, the remaining translation, removed attachment and actual RSS output. Actual alias-inbound uncertainty, global configuration/template/data and unknown-input changes expand full scope.

Completed inspect/impact/context returns 0 with separate current-check findings 1. Check-since retains current quality 1 and full validation scope. Missing/unborn/foreign history returns 2. It retains every known current page/attachment/reference/output. It does not invent a prior identity or change. Malformed selectors/limits, missing tools and failed renderer logs are tested. Bounded context gives reasons/versions/source and excerpt hashes, visible omission/truncation and no execution of literal document instructions.

Saved move preview/apply and subsequent ordinary Hugo passed, retaining binary bytes, raw full modes, unrelated files and Git index/revision. Actual opaque HTML/shortcode references remain manual. Inline/fenced/opaque spans stay unchanged. Their broken final candidate returns 1, with no saved plan or source writes. Source/config/attachment/mode/fresh-target drift returns 2 and preserves the later edit. A deterministic mutation after the actual candidate renderer also refuses before writes and preserves editor bytes/mode. Focused actual move race passed in 8.286 seconds at /tmp/oink-r5-public-move-race.log; app vet passed at /tmp/oink-r5-public-vet.log.

Before the cached-module supplement, frozen parent make test/vet and make test-hugo both passed at /tmp/oink-r5-frozen-go-gate.log and /tmp/oink-r5-frozen-hugo-gate.log (actual app fixtures 185.709 seconds). Actual move/source race and vet passed /tmp/oink-r5-move-hugo-gate.log, /tmp/oink-r5-source-move-race-gate.log and its vet counterpart; full inventory/mode/selector plan safety passed /tmp/oink-r5-plan-owning-final.log.

The first frozen consumer trial exposed a cached-public-module guard gap. The original graph contained resolved module inputs that were absent from a fresh outer candidate hash. It returned false incomplete 2, without source writes. Content plans now resolve/capture the same module inputs before comparison. Legacy metadata/authoring plan scopes remain unchanged. The separate checksum-verified public OINK v1.1.0 regression passed preview, fresh saved apply and ordinary bilingual Hugo in 27.42 seconds (package 28.220) at /tmp/oink-r5-public-cached-module-move.log, including raw modes, binary bytes, unrelated inputs and Git preservation. Corrected current-binary corpus and supplemental race evidence remain separate from the earlier unaccepted trial.

The corrected current candidate passed full make test/vet at /tmp/oink-r5-corrected-frozen-go-gate.log and actual make test-hugo at /tmp/oink-r5-corrected-frozen-hugo-gate.log (app 278.787 seconds). The cached/public and committed/in-site move safety race passed in 38.578 seconds at /tmp/oink-r5-public-cached-seam-race.log; its app vet also passed.

/tmp/oink-r5-corrected-runtime-freeze.json records 96 runtime inputs with SHA-256 e5b6e0eda972116dbb94a8086668e6ef34bfaa56138cf31f1f71f4832c477842 and 165 broader CLI inputs with SHA-256 4965a0c92cb6126f67a6dabd548c7c25ee5d7c9c57e14cce5e55ebb7a22fca2d. The corrected binary SHA-256 is d7675aecca2f77b1eb37bb4f664c3314cf5207149e6abbb86523686c5c50bff0. The corrected four-consumer capture completed 16 commands in 831.825 seconds at /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r5-corpus-corrected-y2eue81h. Its bounded final-receipt.json has SHA-256 b86e0e6c7dbfbaed62c845c03a55d963068f7d9a16771de5ff3b0974e76fce3b; the receipt retains 12 copied owning gate logs and all 20 observed move routes. It links to the complete raw JSON/logs and per-move classifications.

Site Primary/copied inputs/directories Inspect/context Current check Impact Move preview
Starter 97/94/21 0/0 0 2: no Git baseline 0: validated, unapplied
Documentation 421/427/109 0/0 0 0: complete historical comparison 1: six candidate missing references
PIG 858/861/52 0/0 0 2: historical foreign-input provenance incomplete 1: 32 candidate missing references
Repository 2294/2299/48 0/0 1: 10,462 existing duplicate HTML IDs 2: unborn HEAD baseline unavailable 1: the same existing duplicate IDs

Starter and Repository impact retain known current facts without inventing prior pages or changes. PIG’s actual baseline is complete (theme v1.0.0 versus current v1.1.0). Required foreign-input provenance is incomplete. The comparison therefore expands full scope and returns 2. These are distinct outcomes. Documentation impact completes with 192 captured input changes, 343 affected prior/current pages and full scope. All completed fact queries expose current quality findings separately. Repository inspect/context remain 0.

Documentation move proves 18 rewrites and four routes. Two ordinary literal /docs/admin/comments/ occurrences in content/docs/customize/repository.md at lines 216 and 313 remain manual because repeated source/output occurrences cannot be attributed precisely across ordinary and print outputs. Their six missing candidate references block validation. PIG moves two Markdown files and four binary attachments, with eight proven page/processed-resource routes. Equal-byte paired outputs prove processed featured_hu_* resources, but not new URLs for the four original absolute image references /article/pgext-day/{featured,topic,venue,schedule}.webp. Their 32 candidate missing references block validation. Original-asset rewrites are not guessed. These ordinary Markdown limits are separate from opaque HTML/shortcode limits.

Repository move proves ten rewrites and four routes. Its candidate retains only the same 10,462 existing duplicate-ID findings. It has no new missing reference or required incomplete finding. Starter’s zero-link bilingual move is validated. All four moves remain unapplied. No consumer plans were saved. No consumer source writes occurred. Failed candidates have validated: false. All JSON stdout is pure. Primary/copied inputs, full modes, directory inventories and logical Git/index state are unchanged. These comparisons include ignored copied inputs. The Git metadata inventory excludes immutable object storage. The runtime and broader CLI inventories still match the captured identities.

The first ten-file guarded canonical promotion was qualified in 62.37 seconds with the corrected frozen binary and runtime hashes above. The separate rendered receipt is /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r5-docs-render-ks2tw82c/summary.json, SHA-256 06ae844a4b3f1c01bb5faa8a21091d5461c28aab592d12c4310fb34bc176c5d4.

Canonical documentation gate Executed result
Source owners Translation 0: 137/137 pairs, 1,109 headings; style 0: 137 Chinese files, 181 strong spans, no emphasis; scoped whitespace and public JSON schema equality passed
Frozen CLI Production check links returned 0 using the exact corrected binary
Fresh ordinary production Hugo Build 0; Markdown 0: 214 pages/42,571 nodes; links 0: 345 pages/48,376 links/4,203 fragments
Production translation owner 1 only for the pre-existing draft release-1.2 omission from ordinary production output; no new R5 discrepancy
Separate ordinary analysis Hugo Fresh nonpublishable -DFE build 0, without the CLI probe; Markdown 0: 216 pages/42,877 nodes; links 0: 347 pages/48,752 links/4,231 fragments; translations 0: 137 pairs/1,109 headings
Input preservation All per-command and overall guards passed: 421 primary files, 427 copied inputs, 109 directories and 36 mutable Git files retained bytes/full modes/logical Git state; both isolated source copies remained unchanged

The permanent first-promotion applied-files.json in /var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r5-doc-drafts-t3_klnck retains the ten authorized files and their original modes. Core owning logs /tmp/oink-r5-frozen-core-hugo.log, /tmp/oink-r5-owning-race.log and /tmp/oink-r5-owning-vet.log passed. A13 impact, A14 move safety, and A15 bounded context passed their required supported CLI scope.

12 - CLI maintenance acceptance on 2026-10-03: R6–R8 and A18

R6–R8 and A18 workspace, Studio, editing and runtime evidence from the dated CLI maintenance record.

These R6–R8 and A18 sections of the dated record were moved here unchanged on 2026-11-06 so each page stays small enough to read whole. The scope, evidence rules, case ledger and delivery state remain on the main page.

R6 workspace and adapter acceptance evidence

R6 supported scope is accepted locally after frozen owning/runtime, exact-binary consumer parity/preservation and guarded canonical source/render gates. The supported registry/tool fields are documented in the contract and command reference.

The registry has an independent version, oink.workspace/v1. It requires strict one-document regular YAML, 1–64 entries, and at most 256 KiB. It requires exact ASCII names, literal relative/absolute directories, and proven canonical identities. It refuses overlap. Missing-site results stay per-site incomplete while later selected sites run. Selection preserves registry order. There is no default named site, sibling discovery, Hugo settings duplication or automatic multi-site apply. Optional tools extend oink.policy/v1, with pinned protocol versions, configuration/full-mode provenance and typed omissions/coverage.

Workspace owning receipts

Focused gate Executed local evidence
Registry core Strict fields/document/bounds/names/literal paths, existing aliases/case-inode ancestry, duplicate/overlap refusal, missing-directory listing and exact subset order; go test -race ./internal/workspace -count=1 passed in 1.414 seconds, /tmp/oink-r6-workspace-core-race.log
Public actual Hugo OINK_TEST_HUGO=1 go test ./internal/app -run '^TestPublicR6Workspace' -count=1 -v passed in 10.498 seconds, /tmp/oink-r6-workspace-public-hugo.log
Public race The same workspace public suite with -race passed in 12.426 seconds, /tmp/oink-r6-workspace-public-race.log; excluded commands specifically reject registry selection
Vet go vet ./internal/workspace ./internal/app completed with exit 0, /tmp/oink-r6-workspace-vet.log
Public outcomes Actual bilingual committed fixture sites retain direct diagnostics/coverage/exit parity for links and full checks. A missing first site yields 2 while later clean/finding sites yield 0/1; explicit subsets preserve registry order, invalid unregistered siblings remain untouched and human output retains findings
Selected application Saved translation-review preview is validated but unapplied; a different registered name is refused before writes and preserves plan/source bytes/full modes/Git. Explicit matching-name apply writes only its planned review file; other registered and unregistered sites remain unchanged

The inspected core workspace.go SHA-256 is cf2cbc9509e8c83eedf6d8833c9eb0ea6492de9a85c959798112fa3f105213f4; its owning test is 9070a8e2c3e58680f6567f2394160ec682bf0457c068c2addf354921e7612d6b; the public test is 3e57a6417ae2e7604f7cb06933759bb06a2f40758ff7059848593cedbaa6570a. All three inspected files retain mode 0600.

Corrected protocols and stage gates

Protocol or gate Recorded status
Actual markdownlint-cli 0.49.1 and Vale 3.24.0 Corrected public trial passed: exactly one finding mapped to the original UTF-8/BOM/CRLF line; excluded front matter/shortcode/math/raw HTML/enabled attributes/code produced no false original attribution
Actual lychee 0.24.2 Corrected trial actually reached the local HTTP fixture: 200 → 0, 404 → 1, 401/403/429/503/timeout → required 2. Optional offline → 0, required offline → 2, both with zero HTTP requests
Final focused actual-tool receipt /tmp/oink-r6-public-actual-tools-final.log passed in 15.192 seconds; the earlier corrected 14.686-second run is retained as prior evidence. Node preload and discovered JS configuration did not execute; source full modes/Git were preserved
Fake/protocol failure receipt /tmp/oink-r6-public-fake-tools-final.log passed in 13.089 seconds: malformed output, version mismatch, timeout, unsafe configs, required missing/optional/group omissions and raw stderr normalization
Focused public race/vet /tmp/oink-r6-public-tools-race.log passed in 30.273 seconds across fake and actual cases; /tmp/oink-r6-public-tools-vet.log completed with exit 0
Frozen runtime inputs Parent freeze at 2026-10-03T10:58:01.807947Z, /tmp/oink-r6-runtime-freeze.json: 103 runtime inputs bind b85affd96378b45bfc56a996b0c5672d02ee4c6cc9bc95335fa5072f6c42a03b; 179 broader CLI inputs bind fbb8176ebc58f1aa26336f4e6036cf9bd5f7a0d62b142f16532b50a8071e9fbe. The exercised 0.3.0-r6-local binary SHA-256 is aa8b347fbe01071f9da729f4d98aa2f50d7264456be6c5f05771bcfadadc371f
Frozen owning suites Full Go/vet completed with exit 0, /tmp/oink-r6-frozen-go-gate.log; full actual Hugo plus pinned tools completed with exit 0, /tmp/oink-r6-frozen-hugo-gate.log (app 382.832 seconds). Workspace/core/protocol/source-mask/policy/report race and vet receipts passed and are copied into the final receipt
Four consumer sites Qualified: exact-binary direct/aggregate diagnostics, coverage, exit, identity and registry-order parity for all four sites; per-command/overall source byte/full-mode/type/logical and mutable Git/ignored-input/directory guards passed. Aggregate completed 4, finding 1, incomplete 0, exit 1
Canonical EN/ZH Passed: guarded first ten-file promotion, source owners and fresh ordinary production/nonpublishable rendered evidence; only the known production draft-release omission remains
Stage decision Supported R6/A07/A15 scope accepted locally after the required receipts

The durable focused-tool receipt is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r6-tools-6bf67ltv/r6-public-tools-acceptance.json, SHA-256 16b6e47fc0618c76d2f9e3680a4112b6e47b478af8aabd3f2fc84821f840cc8a. It binds the provision record, executable/configuration evidence and 1,422 resolved Node package files. Markdownlint reports original content/tools.md line 7, bytes[80:92] (ppears here.); Vale reports the same line, bytes[71:78] (BADTERM). Each of the seven network cases actually makes one HTTP request.

The exact-binary consumer receipt is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r6-corpus-59_asiyr/final-receipt.json, 42,212 bytes, SHA-256 0ad86afaf235bdcff0c474e76b08e0591591a7b22c7992b02e20fb17975d029e; the completed summary binds 467b66eb6d178829508115050d4243909313acf1b4d59317ecda37ab7383ca55. It retains 14 copied owning/completion gate logs. The six original operations sum to 314.912887 seconds, excluding candidate compilation and receipt-only correction. workspace list returns 0; four direct full checks return 0/0/0/1; the aggregate returns 1 with all four sites completed.

Site Preserved source files Direct/aggregate child exit Diagnostics/coverage Recorded finding boundary
Starter 97 0/0 28/29 Translation review information only
Documentation 421 0/0 144/34 Translation review information only
PIG 858 0/0 120/41 Translation review information only
Repository 2,294 1/1 11,250/29 Existing 10,462 duplicate-ID findings and 788 translation review information items

Direct and aggregate child identities, order, every diagnostic and coverage record match. After every operation and overall, all four consumer sources retain full modes/types, logical and mutable Git metadata, ignored copied inputs and directory inventories. The complete root CLI inventory still equals its frozen capture. The 478,603,149-byte direct repository JSON and 635,470,795-byte aggregate JSON were validated streamingly rather than truncated. Separate pinned-tool fixtures qualify optional-tool protocols. The consumer registry is task-local and writes no consumer policy.

The initial acceptance driver overwrote a summarized result’s command string with invocation argv, producing a false parity exception after all six CLI operations and their per-operation guards had completed. The failed driver and summary remain preserved as pre-correction.r6_qualify.py and pre-correction.summary.json. Receipt completion corrected only invocation metadata. It verified unchanged raw-result SHA-256 values and header commands. It retained all original full-stream diagnostic/coverage digests and rechecked overall consumer/root guards. No CLI runtime correction or Hugo/CLI rerun was needed. The receipt-only completion took 2.002 seconds and exited 0 in /tmp/oink-r6-corpus-receipt-completion.log.

The executed driver SHA-256 is 4d2a360c6f7f6f96c38698bd189bc4d4b2cb02a7509858920d752897fdd85988; the corrected driver is 4870f5c0374fcc11ad1a6b2e3aefe36f493b6f9f4666c293993dc6a59df8a11b; the receipt-completion driver is cd50d3fe704370f73fa4e7d94ce8e4bc925d11ec8ef04d463aacec37c2053daf. The streaming helper binds 144f778cdb7907372797b47b97f817f340e70423701a2a958dee589281a9a11c, and the inventory helper binds d3ac41dc2e18295bfb26134d1a696935c8174913e2801a5766dbf7a1139d89f8. This receipt qualifies local darwin/arm64 with Go 1.27.1, Hugo Extended 0.166.0, Node 26.9.0 and Apple Git 2.54.0.

The initial actual-tool trial exposed Darwin /var versus /private/var staging identity, actual loopback proxy routing, and an invalid inline-block-attribute/line assertion in the Vale fixture. Staging is now canonical. The Vale fixture uses a real standalone block attribute. The source-mask boundary is unchanged. The qualified child environment forwards literal NO_PROXY/no_proxy host-list data while omitting proxy URLs/credentials and Node preload settings.

Supported source diagnostics require proven original ranges. Rendered lychee locations remain output file/DOM pointers without inferring a Markdown line. Offline lychee is not invoked. Authentication/rate-limit/server/transport uncertainty cannot become required success through severity, exclusions or baseline acknowledgement.

The first guarded ten-file promotion and its fresh rendered qualification are complete. The receipt is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r6-docs-render-dcwtcmyl/summary.json, 555,297 bytes, SHA-256 ee153932900dc6f1ec62beef1a75927fc60b857efccfbcc558bf0e2c2b12cc04. The 64.17-second run used the exact qualified aa8b347f…371f binary and unchanged 103-input b85affd9…a03b runtime inventory recorded above.

First-promotion documentation owner Actual result
CLI production links 0; one strict production Hugo renderer, no analysis build
Ordinary production Hugo / Markdown / links 0 / 0 / 0; 214 content pages and 43,376 text nodes; 345 HTML pages, 48,438 internal references and 4,259 fragments
Ordinary production translations 1 only for the existing draft content/blog/release/1.2.0.md absent from production; not a new R6 failure
Independent ordinary nonpublishable Hugo / Markdown / links / translations All 0; 216 content pages and 43,682 text nodes; 347 HTML pages, 48,814 internal references and 4,287 fragments; 137/137 pairs and 1,118 headings
Source owners / schema Translation, style and whitespace all 0; 137/137 pairs, 1,118 headings; 137 Chinese files, 181 strong marks, zero emphasis marks; CLI/documented result schema both bind 7468c2d04cde8a368ce0ba44a1f27125b5fca364b6d4672353519b9545b3bdda
Preservation All 12 owner commands, CLI/schema checks and overall comparison preserve 421 primary files, 427 copied inputs, 109 directories and 36 mutable Git files with full modes/types/bytes and logical Git; both private ordinary source copies and all 103 runtime inputs unchanged

The first-promotion installer receipt, oink-r6-doc-drafts-ymjop499/applied-files.json, binds 13c965592d64056d8365aed1927d2d422fadec8adc54ee7050b22e2ea0ad6270. It retains captured actual original inodes in private temporary storage, preserving later old-open-handle writes. Recovery also preserves later target edits or deletion.

R7 read-only Studio candidate evidence

R7 implements the embedded five-view browser and authenticated loopback API candidate described in the contract. Frozen core/browser, exact-binary four-consumer qualification, and guarded canonical rendered gates passed within the declared scope.

Focused native and browser evidence

Owning boundary Evidence status
Native/public parity Actual shared check reports preserve diagnostic/coverage/exit identity for 0/1/2; explicit selected workspace startup, cleanup/signal and no-source-write proofs are recorded separately by the owning test receipts
HTTP management/source/preview Literal-loopback selection; exact Host/origin/Bearer checks; no arbitrary request paths/writes; captured source/diff bounds and source-mode/output inventory guards; focused core/new browser and current corpus receipts below bind this supported scope
Initial held browser 14 axe checks with zero violations and 14 screenshots; five desktop light views, captured BOM/CRLF source/diff, desktop dark, mobile dark and all five 320-pixel light views plus capture changes. Synthetic actual-Hugo fixture retains 228 native diagnostics and coverage parity; copied suggestions use a private test clipboard, leaving the host clipboard unchanged
Browser preview attack Actual attack script executes in the isolated preview, but parent access, management fetch and popup are blocked; token query refused. Actual draft-only page remains production 404
Snapshot preservation Captured source instructions/HTML stay literal data; the initial capture retains its old bytes before refresh after an explicit task-fixture external edit. Source full modes/Git/directories preserved except that declared fixture edit; changes show the actual modified captured input
Preceding held browser Passed refreshed held UI/backend receipt: all 14 axe checks zero violations and 14 screenshots, including declared/captured theme rows
New partial-preview browser Passed new frozen partial-preview runtime: 14 axe checks with zero violations and 14 screenshots; actual Hugo normal HTML 200 and 67,108,865-byte output 413; native 1/228 diagnostics and coverage retained, required partial coverage visible and Studio/refresh 2

Initial browser receipt: /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-studio-browser-AfwaYi/summary.json, SHA-256 930fbd1ab86806069b963bff2e3e95aaa07e3634400cec65e2b8c7922e2a1707. Its exact binary binds 92e5b962e40fbe828a0b006f3ae76a2bddf4ad7e8b1c5b6967e365b8f1827879. The qualified local versions are Node 26.9.0, Playwright 1.62.1, @axe-core/playwright 4.13.0 and Chromium 151.0.7922.34.

The refreshed held UI/backend browser receipt is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-studio-browser-vn0ofb/summary.json, SHA-256 520603779f712539865c6e9ef7a9ad3ad21ec1906adfb067ec607cc69d071b7e, with exact binary 73bf90c69dce84849ee20ddfbfe825b9f2dd46f0cd37a228ce1b041a83afa33f. All 14 axe runs and 14 screenshots passed, including declared/captured theme metadata and all five views at 320 pixels.

The initial parallel whole-suite trials in /tmp/oink-r7-frozen-go-gate.log and /tmp/oink-r7-frozen-hugo-gate.log failed due to ten-second CI-test deadlines under parallel package load and a graph test observer refreshing its own Git index. The isolated CI target sets then passed in 12.149 and 2.291 seconds. The controlled Git-observer graph run passed in 1.354 seconds. Only internal/projectgraph/hugo_test.go changed: its read-only observer disables Git optional locks, filesystem monitoring and the untracked cache. The ordinary actual-Hugo graph run passed in 1.562 seconds after that test-only correction. No runtime or embedded UI bytes changed.

The corrected freeze is recorded in /tmp/oink-r7-corrected-runtime-freeze.json: the 113 runtime inputs retain 15a7de85a1ae9e6a73d8ea6570aa4f97bdd0ad5677ad7ca996fdd081ad43f7b5; the 193 broader inputs now bind 67c6d36cf91d175f208f79cdd4d337aab6d2ef71e43453b20394b677678725e8, with only the test-observer file changed from the preceding 85ad60d24c93e899020fbdcd34f8252c578253ce5afaf8652e561a432ecc8067 freeze. Corrected serial Go tests and vet passed in /tmp/oink-r7-corrected-go-gate.log. The corrected serial actual-Hugo/pinned-tool suite also passed in /tmp/oink-r7-corrected-hugo-gate.log, SHA-256 2aed822ff6fc8be04919aa74ca6ada721789232c14c1d77f1d44113bc0d235a7; the application package took 220.782 seconds. The independent post-Go source-preservation audit is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r7-postgo-audit-qygh7bcc/receipt.json, SHA-256 5ee45fe041536243bc1229054516835c61b27909a4f296ac226b1a138e4bc8dc.

The durable corrected owning-gate receipt is /tmp/oink-r7-corrected-owning-gates.json, SHA-256 c7f94a740e33a7349886b3f3689419719f857f39031375e80ca384a3dac36e67. The prepared documentation renderer requires the completed consumer receipt to prove a private captured-source rebuild byte-identical to the final browser binary. Its source-only independent audit, oink-r7-docdriver-audit-ig_r8ud1/receipt.json, binds SHA-256 858cc48bf602fbdb26fcbda03c78ca485338b1295d64257d32cfb14b24f1ade3 and prepared driver f25d9ac4bf1a7ef43d5526b7b3cbadf84dd64ad8a57fd82d1c82e76fdb2b3435.

The first consumer driver trial, oink-r7-corpus-h1lOFl, stopped with KeyError('preview_base_path') because it indexed a field omitted when the actual preview base path is empty. Its private rebuild was byte-identical to the final browser binary 73bf90c69dce84849ee20ddfbfe825b9f2dd46f0cd37a228ce1b041a83afa33f, and all four consumer source inventories and the root inventory were preserved. The fresh oink-r7-corpus-corrected-cByXTa driver changes only those two accesses to get(..., ''), with SHA-256 4c409acacbb9b82e658e6705eddefd9a3541def5c63c340b65678a6c9a8354e4. That fresh run subsequently failed when the repository produced an inventoried file larger than 64 MiB: the preceding runtime refused all production preview. Its native check retained outcome 1, while required unavailable preview made Studio outcome 2. Starter, docs and PIG completed that run with outcome 0. All four source inventories and the root inventory were preserved.

A narrow runtime/test correction for partial preview keeps the 64 MiB limit, exposes guarded production files within the limit, returns 413 for the exact skipped oversized paths and keeps required studio.preview coverage incomplete. Native check outcome remains unchanged. Studio still returns 2 for required incomplete preview. The embedded UI is held unchanged. The bounded coverage detail records actual omitted relative paths, sizes and count. A prepared real-Hugo browser fixture adds static/oversized.bin at 64 MiB plus one byte to exercise an available guarded HTML preview, an exact skipped-file 413, native outcome 1 and required partial-view outcome 2.

The new partial-preview freeze is /tmp/oink-r7-partial-preview-runtime-freeze.json, SHA-256 c426ce3e641ed7b39bb711a26006306cab22e761c5062f2164f10deb4bea8765. Its 113 runtime inputs bind 4900ae05abbdf4409b0be54f276fb4135269cf0a49e9071013ccf42544d35c84; 193 broader inputs bind 8b172cef2b228e2642f0139d6cc569136e86843f818e52e412fa4a2d56add25d. Only internal/studio/preview.go changed among runtime inputs. The broader changes also include its test and scripts/test-studio.mjs. All three UI files retain their exact bytes and modes. Focused core final race passed in 1.748 seconds, with vet and scoped whitespace checks also passing. Its receipt, oink-r7-partial-preview-owning-a56dunn4/receipt.json, binds SHA-256 7b6ecb491f283d04fe54347e564dba426b1a84d152040a1d945af54bc67756ac. In the initial sparse-fixture mode trial, host umask 0077 made a requested 0640 file actually 0600; explicit fixture chmod to 0640 corrected the setup. Focused proof covers normal 200, oversized GET/HEAD 413, changed identity 409, private path 404 and refusal for other unknown output errors.

Whole serial Go tests for the new partial-preview freeze then passed in 61.481 seconds, and vet passed in 0.571 seconds. Completed logs are /tmp/oink-r7-partial-preview-go-gate.log, SHA-256 be7d6eccf99a6f4c1b8f09d1fb782455c7cbd3bad4a2f37e2f0e9da916bcb313, and /tmp/oink-r7-partial-preview-vet-gate.log, the empty SHA-256 e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855. The independent held-input audit oink-r7-partial-held-audit-o6p98i1l/receipt.json, SHA-256 e567efc5f580db9395afab8ad36c4db842c3db95db442eb1cb1c740dbd43ec31, verified all 113/193 inputs and physical/logical identities during that parent Go/vet run. The new whole actual-Hugo/pinned-tool invocation subsequently completed with exit 1 after 285.649 seconds. Its sole failure was the parent’s unavailable Markdownlint preparation path /md/node_modules. All other actual cases passed. The log is /tmp/oink-r7-partial-preview-hugo-gate.log, SHA-256 1ab6b8cfd399d484e08a1d1f05d25475754caa731991dd1eec1cca03cf6ce970. The sole owning case was rerun with the exact provisioned /markdownlint/node_modules executable. Source/runtime did not change. The case passed: application package 2.317 seconds, wall 3.265 seconds. Its receipt is /tmp/oink-r7-partial-preview-corrected-tools-gate.json, SHA-256 62b75e563e8074995ed9dd354434e653b2f5f2c6d20767226286d0c08d4c667c; log SHA-256 is e9bddac210654d219d9c5d6ebabaa3b91a0f5f4de4daf228b3ae21aeaac7673a. The executable comes from provision receipt 268e601e81bc03a263296d57257b85635371bda642d0632532a7d9318c981461. The independent case-matrix/held-source audit verifies cumulative executed actual-owning-case coverage 0 from the whole invocation plus that corrected case. Its receipt, oink-r7-partial-case-matrix-audit-16_bl9hr/receipt.json, binds SHA-256 0a9e4a1a1e1a08f597becb2f27e743c9f23df672c713c2757241704edb16b51e. All 113/193 physical/logical inputs remain frozen. Optional TestArtifactCorpus and TestPublishedRuleSourceProvenance cases were explicitly skipped.

The new post-Go input audit, oink-r7-partial-postgo-audit-g1hcqdtl/receipt.json, SHA-256 b369737ec48456f673c850ea702cb3cb7efffb8ecc129d87e00dc03af82b2e3b, then confirmed the complete held 113/193 physical/logical inputs after Go/vet.

The new partial-preview browser passed against exact binary f39d6754f7ad13599e4e849394e0f470b2c6f26edf96ce40f199d27b65a8030e, version 0.4.0-r7-local, 16,000,578 bytes and mode 0700. Its summary is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-studio-browser-cE9be3/summary.json, SHA-256 9099e6c407fd0f9de3c29ce80e03f034a4223d7d7a7c1f1378052e8b9084e0ae; provenance SHA-256 is 85b9f537fb09eecbb09d133b53a297c78184c11200ab0938734c0d10f3449095. The private oink-r7-browser-partial-ZZIqzZ/source-binding.build.json, SHA-256 7a89ab8318c3a38455ab6ce12bcdbc53ae5ce0674fb5aaaa1df0fcf68a093399, binds all 113 runtime and 193 broader source inputs plus physical identities before capture/build/after to the new freeze; root inputs stayed unchanged. All 14 axe checks had zero violations and all 14 screenshots passed, retaining the keyboard/mobile/light-dark/source/clipboard/security checks described above. Actual Hugo emitted oversized.bin at 67,108,865 bytes. A request through its rendered /sub/oversized.bin link returned 413. Ordinary actual HTML returned 200. Required partial preview coverage stayed visible. The CLI/API/UI matched 228 typed native diagnostics and native coverage/outcome 1. Studio and the subsequent refresh returned 2. Source preservation still excludes only the declared task-fixture external edit.

Another prepared corpus driver had assumed that an available normal preview always appends a studio.preview coverage row. Actual normal Starter/docs/PIG Overviews do not emit that row, so the driver assumption was corrected without changing native coverage. The repaired fresh oink-r7-corpus-partial-pZLwY0 driver, SHA-256 47778df62505beeb7432985be927f1b001e03824e9dee3a6dbed9d9b2dbe049c, was reviewed against those three retained actual Overviews and the current partial browser capture. Normal availability still requires its actual preview URL and independent HTML 200. A partial capture retains its actual required row, omitted count/identities and 413. The preparation audit is oink-r7-partial-driver-correction-audit-sa98cm6k/receipt.json, SHA-256 a519a5bc6ae83438146ff4710d53f5edb0e656a05d0532c02123e5771416f07e.

The fresh four-consumer qualification completed with driver outcome 0 in 234.6425 seconds. Its current summary is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r7-corpus-partial-pZLwY0/summary.json, SHA-256 d7b5a4f1607b6f75ae6a596c19cbab28685fb67dac750096173060ed097c8bf5; log /tmp/oink-r7-partial-corpus-gate.log binds SHA-256 a7b724500569bd594d8e01502ec1956eb089cc9153b04b693992a9322013c811. The durable current corpus qualification.receipt.json binds SHA-256 4e5df7c3fda9f0b763091af3e6cb85c68c736c319a1de68030b81d5cd5b384bc; primary source inventories contain 97/421/858/2,294 files respectively. The private captured-source rebuild is byte-identical to the new browser binary f39d6754f7ad13599e4e849394e0f470b2c6f26edf96ce40f199d27b65a8030e. Runtime 113/broader 193 inputs and root physical identities stayed frozen. Every operation and the overall boundary preserve all four consumer bytes, full modes/types, logical/mutable Git, ignored copied inputs and directories.

Consumer Native outcome Typed diagnostics Studio outcome Actual captured pages
Starter 0 28 review-info records 0 66
docs 0 144 review-info records 0 343
PIG 0 120 review-info records 0 248
repo 1 10,462 existing duplicate-ID findings plus 788 review-info records 2 1,576

Nested native headers, typed diagnostics, coverage and exit match direct CLI checks exactly for all four sites. All issues were paginated. Other views used bounded samples. Captured physical source and translation diff were available on all four. The first three actual production previews returned HTML 200 and emit no studio.preview omission row. The repository normal HTML returned 200 with 60,100 bytes. Its current capture exposes four oversized print paths, where each actual HEAD returned 413 with zero response-body bytes:

Captured omitted relative path Captured byte size
_print/pkg/index.html 73,976,221
_print/pkg/pgsql/index.html 69,903,999
zh/_print/pkg/index.html 73,086,240
zh/_print/pkg/pgsql/index.html 69,052,754

Required studio.preview remains incomplete, so repository Studio 2 retains native 1. Actual analysis-only draft routes returned production 404 in docs and repo (Starter/PIG had no unique captured draft route). Workspace subset/full/healthy-subset sessions selected only registered sites and closed listeners without captures. Unknown or selected missing sites returned 2 before startup. Independent final corpus audit oink-r7-final-corpus-audit-xr3_u5dt/receipt.json, SHA-256 b8a8eedf5c899fe5830bdde959783c46b3f191ab144c0d3798077555d55238fc, verifies raw typed native/API/shutdown parity, all 132 operation preservation comparisons and four overall guards.

For temporary disk capacity, the parent retired three private Go build caches totaling 366,184,826 bytes, as recorded in /tmp/oink-r7-private-cache-retirement.json.

Remaining stage gates and promotion boundary

Required gate Current status
Frozen runtime input/binary identity New partial-preview freeze binds 113 runtime inputs 4900ae05abbdf4409b0be54f276fb4135269cf0a49e9071013ccf42544d35c84 and 193 broader inputs 8b172cef2b228e2642f0139d6cc569136e86843f818e52e412fa4a2d56add25d; all UI bytes/modes unchanged. Source-bound browser binary f39d6754f7ad13599e4e849394e0f470b2c6f26edf96ce40f199d27b65a8030e passed; fresh private consumer rebuild is byte-identical
Full Go/vet and actual Hugo New whole serial Go/vet and browser passed. New actual-Hugo/pinned-tool whole invocation remains exit 1 for a preparation path; sole corrected owning case passed 0, yielding independently verified cumulative executed actual-case coverage 0; two optional cases explicitly skipped
Four consumers Completed exact-binary CLI/API qualification; native 0/0/0/1, Studio 0/0/0/2, exact nested native parity and source byte/full-mode/type/Git/ignored-input/directory guards; repository partial preview remains required incomplete
Canonical paired sources/render First guarded TEN promotion and scoped actual render passed
Stage decision R7/A16 supported local scope accepted

The documentation installer retains the actual captured old inode outside canonical source storage on both success and recovery, without unlinking its last name after an earlier target identity check.

The completed first-promotion rendered gate is /private/var/folders/df/bfm8q07d7bv3kpjf1fjchq4m0000gn/T/oink-r7-docs-render-n8tw2tbw/summary.json. Its SHA-256 is 35e79f51d39803b3e4cdf134ed277957dd627acba42e0e0dc785e4745ec3c481. The log SHA-256 is de3a07eac661c15805070e0ed2e364a71ebbd38e15d8907aab3bdf716e95131d. The run took 63.33 seconds. Exact qualified binary f39d6754f7ad13599e4e849394e0f470b2c6f26edf96ce40f199d27b65a8030e passed production CLI links with one strict Hugo build. Ordinary production Hugo without a probe passed rendered Markdown (214 pages/44,075 text nodes) and links (345 pages/48,482 internal links/4,303 fragments). Its translation owner retained exit 1 only for the existing nonpublished release 1.2.0 draft. Independent draft/future/expired analysis passed Markdown (216 pages/44,381 nodes), links (347 pages/48,858 internal links/4,331 fragments) and translations (137 pairs/1,129 headings). Source translation/style/whitespace checks passed and CLI/docs schema 7468c2d04cde8a368ce0ba44a1f27125b5fca364b6d4672353519b9545b3bdda remained identical. All twelve operations, schema and overall guards preserved 421 primary files, 427 copied inputs, 109 directories, 36 mutable Git files and 113 runtime inputs.

The first guarded promotion receipt is oink-r7-root-promotion-p9g1u7pz/summary.json, SHA-256 323a5ce267e39aaf8f97dc4a12cccbdde83730155a199efb5f3815e29b334e4a, verifying that actual original inodes are retained outside canonical source.

R8 accepted reviewed editing evidence

R8/A17 supported editing scope is accepted locally after the corrected frozen owning/browser/corpus and guarded canonical rendered gates recorded below. CLI edit text, field, snippet and attachment preview the same bound oink.edit/v1 intent. Explicit studio --edit uses this intent. Saved-plan apply or explicit acknowledged Editor Apply owns selected source writes. The default Studio session remains read-only.

Candidate scope and preservation

Known site-owned UTF-8 Markdown is bounded to 1 MiB. Full text and supported ordinary top-level YAML scalar forms retain the declared BOM/line-ending and source-span preservation boundaries. Unsupported form shapes remain text. Exact value_json numeric tokens avoid browser Number rounding. Scalar forms limit numeric literals to 4,096 bytes and absolute decimal exponent 10,000. Larger/nonfinite constructs remain manual text. Field JSON is at most 1 MiB. Escaped lone surrogates are refused. Valid Unicode pairs are supported. Catalog components use original UTF-8 body byte offsets. Attachments require actual leaf-bundle identity, at most 4 MiB and an exclusive new clean basename. Source hashes, full modes, all site/external inputs, regenerated intent and fresh actual Hugo validation bind the same shared guarded application path.

The Editor displays the complete UTF-8 review, selected-file base/after identity and native candidate result. The review is capped at 2 MiB. Its literal bytes are hash-checked before acknowledgement. The actual selected candidate HTML is draft/future/expired analysis, visibly nonpublishable and separate from the original production preview. Required candidate-view incompletion may raise the proposal/session to 2 without changing native findings. For page-file edits, the selected candidate source hash/full mode matches its reviewed After state. Attachment/no-op proposals retain the selected page’s reviewed Base state. Stale/replayed plans, attachment collisions and untrusted preview requests are refused. Applied-with-refresh-error remains explicitly applied.

Focused preparation receipts

Candidate evidence Current observation and boundary
Pure editing core Owner’s focused exact-numeric tests passed for 18446744073709551615 and 7.12345678901234567890123456789, exact no-op raw bytes and changed final decimal digit; broader frozen receipt pending
Actual DFE output ownership Live ordinary output is copied through a confined os.Root, exclusive target files and guarded streaming reads; files over 64 MiB can be captured while serving limits remain unchanged
Copy cancellation/race Context-aware helper focused 0 in 0.703 s, race 0 in 1.856 s, vet/whitespace 0; actual first-chunk cancellation retains partial output, source bytes/modes/identity unchanged; source FIFO replacement cannot block before descriptor proof
Helper log identities Focused c227a88210ab0dc46b24eaff50a347d5c494e9ce23f5bdef5d5b822efab4976f; race 13f0616d55fd4df791ecded0712a18096392c88cb9b849383414c305e50b6779; vet is empty SHA-256 e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855
Read-only candidate integration review Selected actual HTML URI/base prefix and inventory, retained private DFE lifetime, source SHA/full-mode equality, native versus view coverage and cancellation reviewed; no new material defect found within this code review scope
First Editor browser trial Harness stopped on ambiguous global Open editor selector; retained as failed trial
Corrected-selector Editor trial Desktop field/component/binary apply checks and axe checks passed before 320 px draft-review horizontal overflow failed; retained original trial
Narrow layout correction Editor review hashes/receipt text wrap and intrinsic widths are bounded; prior CSS and failure evidence retained separately. Development rerun passed 12 axe checks/screenshots, including real 320 px dark review and light receipts/refusals; final frozen-source/binary rerun pending

The preceding preparation rows were captured before the first complete R8 freeze. The first whole frozen gates later passed for binary 84b804d3246a5be581e44884ed910fa3f45d8be29734b8babdeeb763a11fa882 (0.5.0-r8-local), runtime 123/58517b8e98b80df6642be4ee6275a0074ec187768b20e009f41da2607b635d46 and broader 212/d81335c78413acc60e27adee0ac794862285e41cb2a3a7687ec820c1065ba003. The whole-gate summary is b49f4a3272af3e3dcc92e7e9b38d4289bb49cb19aa3e9354ea148d2d8cb2cea8: Go tests 0/56.523 s, vet 0/0.904 s, whole actual Hugo plus all three pinned tools 0/320.044 s, core race 0/16.610 s and public R8/helper race 0/48.319 s. Its source guards passed.

The first frozen browser receipt eb6977235ef9ae6cec28651b5654eb101685af67abe0cbed0acb0f8375458d9c binds that same binary and source freeze. Editor had 12 axe runs with zero violations and 12 screenshots. Retained read-only Studio had 14/zero/14. The actual form preserved the literal 1e400, its planned after hash and diff. It then discarded the proposal. Exponent ±10,001 and a 4,097-byte numeric literal were refused locally without an API request. Four acknowledged field/component/binary/draft applications occurred only in disposable fixtures. Default read-only refusal, stale preservation, no-op source bytes, preview isolation and native-result independence passed.

The first exact-binary consumer trial then stopped at Starter after 37.533 s. Its immutable failed-trial receipt is 853a8397ba4c527c03aa3cc9ac7cacc41c0ab0379549d145b874f1d1ecc3901c. Two failures are recorded separately. A driver event hash depended on JSON object-key order. Recursive comparison nevertheless proved API/CLI arrays equal: 28 diagnostics, 29 coverage rows and native exits 0/0. Separately, repeated resolved cache capture produced a genuine duplicate module input, .gitattributes. Required candidate graph capture became incomplete. Mutation outcome was 2, and native check remained 0. There was no actual selected DFE HTML or applicable lease. The public published-cache fixture reproduced that defect; its retained failing log is 50ab704e715665096e0f36391bb1364841c3a2b2ac88dd262915ce53fb66afa6. All 48 recorded per-operation source proofs and four overall consumer guards preserved bytes, full modes, types, Git, ignored copied inputs and directories; root inputs stayed exact. No Apply or saved plan was performed.

The narrow runtime correction gathers complete resolved-module rows before committing additions. Identical repeated or reordered captures retain the original inventory. Changed hashes/full modes and added or removed paths within an existing scope are refused. Missing scopes, conflicting identities and capture errors are also refused. Refusal does not refresh prior evidence or append partial additions. Non-module rows remain exact. Site owning receipt e7b284b9dece1c5f2b696cd76166d2286fcbec69e6c48912ab9a78204bdb980b records focused 0/0.746 s, site 0/2.123 s, focused race 0/1.958 s and vet 0/0.167 s.

The corrected published-cache receipt 6358dc81f06e34b789cb47a0f4442d6d036d2e33423a06f5dbe85b3064766da6 records actual github.com/pgsty/[email protected] from a task-local copied checksummed archive, with no downloads or replacement. Original and candidate graphs each have 1,256 unique inputs, including 1,198 module inputs. Full API/CLI typed diagnostics, coverage, exits and plan identity match. Actual selected DFE HTML returns 200. Source bytes/full modes/Git are unchanged. No Apply or saved plan occurred. Owning race passed in 30.255 s; vet passed. Its corrected race log is c7d21a30b8af141d9d9604a80ddf9cf3f608320b97a441a06a742376e2119551.

The current complete corrected freeze is fb276500a3d2643bd0aa220f8bebb380fce2c98493d62b0502b6497b2f02949f, runtime 123/cdf629eeb4bbef6d4d88ee27fe3fb0a73b07b6bf6438336e033a18fb7feb1c17 and broader 212/f6e305e792733a550814eb841615d12fa14a9a6bb2a97c4ada85f7275183e579. Only source_inputs.go, its owning test and the public published-cache test differ from the first freeze. Held UI/helper bytes and all full modes remain unchanged. The rebuilt candidate is bd25f9e0b35ec10e227aabf9582ae40b0b367390f64b93668de6ae85222c3d71 (0.5.0-r8-local). Its observed corrected source-bound browser receipt 33a698985a55c14c3e64e981da1f8e74c686497083dc0edb241406f185e3eeb8 again reports Editor 12/zero/12 and read-only 14/zero/14 with complete 123/212 pre/post source preservation.

At the next evidence observation, the corrected whole gates completed for that held bd25f9e0…22c3d71 binary and fb276500…02949f freeze. Summary 208f156c0954e803eccbada678a4689683dc1576c543c379e5cc04b3497ef772 records Go tests 0/57.826 s, vet 0/0.521 s, whole actual Hugo plus all three pinned tools 0/378.847 s, core/site/Studio race 0/17.937 s and public R8/attachment/output-helper race 0/85.159 s. Every gate’s source pre/post guard passed. The actual-Hugo log has 434 top-level passes and zero failures. The two optional external fixtures TestArtifactCorpus and TestPublishedRuleSourceProvenance remained explicitly skipped.

The corrected whole actual-Hugo log is 4eb1a2afdce2adbe570b10922fd53b6d8954f7c95747370c3c661e94d2f71a05; Go log ea59463e9649ffe2f8aff9da66c91cf6895c86fde96a524db23c89cd4eb35925, core race cdd3d761b5ca7b5e986b25aee3129d65663e3e5ebb83eecb6fbb080387298a58 and public race bb610bdcd7a299cb9b66f4c69e30e246c20546efae47653c01d350b1026ea2de.

The 434 passes and two optional skips above are top-level counts. The same whole invocation also skipped the nested Unix-socket refusal fixture because the Darwin temporary pathname exceeded the socket limit. A first shorter private-path trial still skipped. Receipt 93106854ca890b497d3c74522b895f597ac60cec55ced42cad7b187d334da200 retains process exit 0 and records no actual socket execution.

A subsequent nonresolved short private TMPDIR executed the same frozen socket fixture under race detection without a skip: 0/2.954 s. Receipt 531a503b3b91e1b423c2be61b92ed806d3a813738c38d57e5ec122577b4337f9 and log 236c84f1842ffce76174c834f3888718a109cec6377ada6f3242b02f551f00b3 bind fb276500…02949f and all 123/212 logical/physical/Git inputs unchanged before/after.

The corrected four-consumer trial subsequently completed in 845.705 s. Summary af4fc53326163c4a03aa2982c1f01363fbbdd5a447c9baed3639bd8599d46370 and qualification receipt 4d6fd02543c1920497e1a1bb0a68fcf89b0546130fd9cc12c1df391b7e673e75 bind a byte-identical private rebuild of bd25f9e0…22c3d71, the complete held 123/cdf629ee…feb1c17 runtime and 212/f6e305e7…5183e579 inputs. All 2,383 entries of the first failed trial retained their exact bytes/full modes/types.

Corrected consumer Native/current and native candidate exit API candidate outcome Full diagnostics/coverage Actual selected analysis HTML
Starter 0 / 0 0 28 / 29 200, 48,149 bytes, /blog/design/content-model/
Documentation 0 / 0 0 144 / 34 200, 61,738 bytes, /blog/oink/immersive-reading/
PIG 0 / 0 0 120 / 41 200, 55,641 bytes, /404/
Repository 1 / 1 2 11,250 / 29 200, 92,352 bytes, /blog/infra/2020-12/

These are actual selected Hugo HTML routes in the separate, visibly nonpublishable draft/future/expired candidate view. Each expected candidate marker was present. API and CLI matched full typed diagnostics/coverage, native exits, and plan ID. They also matched Base/After hashes and full modes, unified diff, and the selected page’s proposed source. The complete literal API review and its hash were independently validated. No consumer Apply or saved plan occurred, and no listeners remained. The repository retained 10,462 existing duplicate-ID findings and 788 information records. Its four actual PRINT outputs remained required partial-preview incompletion: _print/pkg/index.html 73,976,221 bytes, _print/pkg/pgsql/index.html 69,903,999 bytes, zh/_print/pkg/index.html 73,086,240 bytes and zh/_print/pkg/pgsql/index.html 69,052,754 bytes. Each bounded HEAD request returned 413 with zero body. Selected in-limit HTML remained 200. Native 1 remained unchanged. Proposal/session 2 and Apply refusal stayed visible. The other three complete previews had no invented explicit complete-coverage row. Actual guarded HTML 200 supplied the evidence.

Exactly 53 protected operations each checked all four sources: 212 per-operation source proofs plus four overall proofs, with source bytes/full modes/types, copied ignored inputs, directories and logical/mutable Git unchanged. Each source proof compared four inventory categories, yielding 864 raw inventory pairs including the overall comparisons. All 53 root guards and the final complete 123/212 logical/physical inputs also matched. All issues and pages were paginated. The other five view endpoints were sampled to their first 50 records. Each site included one known source and one bounded diff. Full native/CLI record parity used the declared bounded complete-record codec. Object order was canonicalized. Array order, types, null and field presence remained significant.

Independent audit receipt 6b72ca06d8392a5271fc40757f176f26e1144c21eec93dbd035e0a1bd645657b verified 69 artifact hashes, complete bounded API/spool/shutdown typed records and the large native/CLI raw-file digest bindings. Its additive receipt 335f137663d4ec0b2a0d3e49c8d70b9918f86078ab6264855171a2644d8aa6c6 also verified the fresh current root’s complete logical Git inventory against the freeze. The original audit stayed immutable.

The reviewed first TEN promotion subsequently passed through the guarded retained-inode installer, root receipt 7cd9b4604d2340b9e46965a26281c921b967060909d518b8b4b31e5f42d0120c. Its actual original source inodes were retained outside the documentation site. Unselected source/copied inputs, directories and Git, and complete CLI 123/212 logical/physical inputs stayed unchanged.

The canonical render then completed in 67.21 s, summary bb0d0710294f810fb14284f7b5b0329befbd290b66c21397b9a45e8382287fb6, qualification receipt 32d3ffeca43bc9ad4615edcca0d3cc47cc932bbc93c6576724c800e0a62405b1. It used the exact qualified bd25f9e0…22c3d71 binary and corrected 123/212 freeze. Actual CLI production links passed 0 with one strict Hugo build. Independent ordinary probe-free production Hugo/Markdown/links passed: 214 Markdown pages/44,691 nodes and 345 link pages/48,532 internal references/4,351 fragments. Its translation owner retained 1 solely for the existing release/1.2.0 draft absent from production. Separate explicitly nonpublishable draft/future/expired Hugo analysis passed Markdown (216 pages/44,997 nodes), links (347 pages/48,908 references/4,379 fragments) and all translations 0.

Source translations passed 137/137 pairs and 1,143 headings. Chinese style passed 137 files/181 strong spans/zero emphasis. Whitespace passed. Schema SHA-256 7468c2d04cde8a368ce0ba44a1f27125b5fca364b6d4672353519b9545b3bdda matched exactly. All 12 commands, schema and overall guards preserved 421 primary source files, 427 copied inputs, 109 directories and 36 mutable Git files, plus all 123 runtime/212 broader CLI logical/physical inputs. The qualification receipt binds 60 canonical inventory pairs, 15 CLI guard pairs and six private-copy source pairs.

R8/A17 supported local editing scope is accepted after corrected owning, socket, source-bound browser, exact-binary four-consumer preservation and these guarded canonical gates. R1–R8 are accepted locally. Native repository findings and required partial-preview 2/Apply refusal remain visible. The separate platform-authority audit 762571dab9a07651ac8e4c71764bfef292f8d5eba729a089e72d9d755b7e2d7c confirms that the initial contract qualifies actually exercised architectures: macOS arm64, native Linux arm64 and emulated Linux amd64. Darwin amd64 remains an experimental archive with failed actual execution/unverified runtime.

Required gate matrix

Required gate Current status
Final immutable runtime/source freeze and exact CLI binary Corrected complete 123/212 freeze and bd25f9e0…22c3d71 bind completed owning/browser/corpus/canonical scope; first-freeze trials separate
Public CLI/JSON/exit, stale source/config/external-input and guarded writer tests Corrected public R8/attachment/output-helper race, whole Go/vet/actual Hugo and canonical stage gates passed
Frozen whole Go/race/vet and actual Hugo/ordinary Hugo after selected application Corrected whole Go/vet/actual Hugo and core/public race passed; 434 top-level passes, zero failures, two optional external fixture skips explicit; first trials remain separate
Editor browser five-view parity, text/forms/components/binary attachments, exact numeric/no-op, stale rejection and preview isolation Corrected source-bound Editor 12 zero-violation axe runs/12 screenshots and read-only 14/zero/14 passed; bound completed corpus and canonical acceptance
Exact-binary four-consumer read-only qualification Corrected all-four completed in 845.705 s; full typed native/API/CLI candidate parity, 212 per-operation source proofs plus four overall, no Apply/save/source writes; first failed trial preserved
Protected canonical TEN promotion, EN/ZH source/schema/style/whitespace and actual production/analysis render Guarded first promotion and independent exact-binary 67.21 s render passed; only known draft translation omission in production; rendered and status bytes separately bound
R8/A17 stage decision Supported local scope accepted after corrected whole owning/browser/corpus/canonical gates; R1–R8 accepted locally
Final A18/platform/archive delivery Completed in the 2026-10-04 supplement below

Prior R1–R7 sections, whole-invocation failures and scoped acceptance receipts remain unchanged. Temporary qualification files stay outside canonical content and Git.

Current runtime completion supplement on 2026-10-04

This supplement records the current candidate on 2026-10-04 (Asia/Shanghai). The dated page URL and all initial 2026-10-03/R1–R7 records remain unchanged. The three UI files retain the browser-qualified bytes and full modes. Current Go, Hugo, platform, archive and four-consumer checks refresh the changed backend.

The first current ARM offline unit run exposed an output-copy integrity gap. On ext4, adding a directory entry could preserve the parent’s allocation size and observed timestamp. That failed run stopped before later qualification steps. The correction captures and rechecks actual sorted directory membership and entry identity. It also checks regular-file bytes and full modes. Only internal/app/studio_output.go and its owning test changed. The failed receipt and independent audit are retained.

The preceding integrity-correction qualification freeze is 683daca0e522193c7ff1b0de6ac2fee5d2fca080811bf184a8dfd5b90a33f224: 123 runtime inputs hash to d346ad15cd4239004e32e1b9f30d727eaf156be0187dc165ca874032a7cf962a, and 212 complete CLI inputs hash to 2abd1a044d8192b07f9bbc06b55dc8b4544d66ca17b8867971cec702ba3af088. The 0.5.0-r8-local Darwin arm64 candidate is 74ad94e73557f6538cd64edd1766d6df92c596d98411031159d94af072c186ec. The integrity-correction receipts below bind that source scope. The later one-test fixture amendment has a separate complete source identity and completed formal qualification boundary. Both are recorded below.

The current complete source freeze is now 196245a3ba09305e34b86539c8eb79f1473e4373ee47aa1f56f8933b04a42d43. The 123 runtime inputs remain d346ad15cd4239004e32e1b9f30d727eaf156be0187dc165ca874032a7cf962a. The 212 complete CLI inputs are 2c487bfb4c65ed40ff78356b2860de627e6ac1afa0da2df433b09345dab7f5b0. Only the owning published-cache test changed, to source 54c10ef89310256b5f4c165c7de5de9668e1d4d2991b71751076680141dbe779. The fixture amendment is separately guarded. Production bytes and all semantic assertions remain unchanged. Formal qualification of this complete source passed. It includes the current eight owning gates, reproduced archives, and full plain-Go AMD/ARM runs. Root A18 proof 2c018cb2afa3f26699a9e6b5a0971096246b12405fde5a27e43a9e213e46da60 binds all three declared supported targets and five reproduced archives.

Current proof and preserved earlier input boundary Observed result and bound receipt
New complete-source formal qualification Freeze 196245a3ba09305e34b86539c8eb79f1473e4373ee47aa1f56f8933b04a42d43, owning test 54c10ef89310256b5f4c165c7de5de9668e1d4d2991b71751076680141dbe779, unchanged runtime123. Current eight gates, host/archive and both full plain-Go Linux flows passed, bound by root A18 proof 2c018cb2afa3f26699a9e6b5a0971096246b12405fde5a27e43a9e213e46da60
Narrow integrity correction Owning receipt 6966d768025497b45958073d4c53a2a2981065c8a95857834dcf6a4faa4f0201. Independent audit 6cb5d2eabf57b41079026a38a674f46def9f56a15df17ad27e671e4f765798df
Prior-source six frozen owning gates Build, full offline Go unit/vet, whole actual Hugo/pinned tools, core race and public R8/output-helper race all 0. Elapsed 3.501/68.257/3.909/333.801/37.070/79.670 seconds. Summary b40b7787b3da8dc1e0763812b6dde529b4b5b69fe479d79940f1161223124e1d. Independent audit 20780662b7ff35019b2c8c84e6dc763f9351ae0816f6ef7a7789f7a15be99167
Eight current frozen owning gates Selected published-cache actual Hugo and race, build, full offline unit/vet, whole actual Hugo/pinned tools, core race and public R8/output-helper race all 0. Current summary d6272fcc4dfab114aecfcdf19a7e2b78f1e931817331b460f43ff2056bf754a4. Raw whole Hugo has 435 top-level passes, no failures, two optional top-level skips and the explicit long-path socket child skip. Runtime/binary bytes remain identical
Prior-source Darwin arm64 and archives Current extracted candidate runs outside the checkout with no consumer Node requirement. Seventeen commands and eight actual process tests, including child signals, passed without process-test skips. Two fresh release directories contain byte-identical five archives and checksums. Source/license/provenance/canonical tar checks pass. Summary bcb4d7599e965c1b3cfe7fe698ca14061ad53d45e7a194337aeebb8d37aa77c1. Independent audit 60a04771365d8be15ac91fbbd8d485b019aae081598e468018861ca5734e387c
Current Darwin arm64 and deterministic archives Seventeen extracted-archive/ordinary-Hugo/process commands passed expected exits, with missing Hugo explicitly 2. Eight actual signal/process cases ran without skips. Two independent fresh builds reproduced five byte-identical archives from current complete source. Summary 3890fd8468b6bce5271bb32ffa1a18bd5daf99c19c43becac0be8e3b908a5d57. Source, tools, module-cache and smoke-source guards remained equal
Prior-source Linux arm64 Actual nonroot Linux arm64 on ext4 with Go 1.27.1, Hugo Extended 0.166.0 and Git 2.47.3: full offline Go unit/vet, all 13 required pure top-level pass records and the membership case plus its four children without skips, 10 selected actual-Hugo cases without skips, native rebuilt archive identity, installed bilingual/offline/ordinary-Hugo/missing-Hugo 2 JSON and signal/source-mode checks passed. Guest summary 409990bc1425f4bf219f8911a71581af6e68729865580121dbeb6d85a06d2ea7. Outer receipt a022e40f068703cd59ce6d6a7fb6530cce6907681baa26eb1dfc77c09f0c8898. Exported-record audit 24afc50f6f860394d1ebfa7a8b754ddd9cb97f9e88a0dcfcbcb659193ecbfe5f
Current Linux arm64 Current nonroot Linux arm64 ran on ext4, with native ARM through QEMU HVF and Go1.27.1/HugoExtended0.166.0/Git2.47.3. Full offline unit/vet passed with 370 top-level passes. The following cases passed without skips: all13 decisive pure cases and4 membership children, and 10 selected actual-Hugo cases. Exact native/installed current archive identity and bilingual/offline/ordinary-Hugo/signal flows passed. 24 commands reached expected exits, including missingHugo2. Guest 268102f69c0950f9d2994d22cd2fd290fc11e24bd6d6f916fd70a93ca4946c74. Outer f3c066fdc9b97feff92160346185a1af978a5172eed5c81904ac7c0e5fc6c982. Source/SDK/borrowed/old-task guards were equal. The owned VM was reaped. Default optional unit skips retain their named gating reasons
Prior-source Linux amd64 failed trial Preserved current TCG trial failed. Outer receipt 3543664ba5590f2ba5a8f676b196bb636b72bc819913289f415d0a8a841c1bdb, guest summary 16075204d287713c7f7650c0a65dd289dd4bd83db07c9ba4b85b3f21244d5240. Full offline units (370 top-level passes), vet and the first three selected Hugo cases passed. The published-cache candidate request reached the test HTTP client’s 90-second deadline. Candidate parity, the remaining six selected Hugo cases, native rebuilt archive and installed archive smokes were not reached. Deadline review 12917b9eb89e3abc5893e08da3b6b6e20743dcb4c14e6f7ba8561628e3566934
Current Linux amd64 Current nonroot Linux amd64 ran on ext4 under QEMU TCG emulation, with Go1.27.1/HugoExtended0.166.0/Git2.47.3. Full offline unit/vet passed with 370 top-level passes. The following cases passed without skips: all13 decisive pure cases and4 membership children, and 10 selected actual-Hugo cases. Exact native/installed current archive identity and bilingual/offline/ordinary-Hugo/signal flows passed. 49 commands reached expected exits, including missingHugo2. Guest 3a1a32979efc843de8b95b7c13824026e17f71c06d4c458b738c0b9583fb4723. Outer 30cf4950cc83fa0732047d9a0f89bb59e68779ee2e8f5c755724c9679be265e3. Source/SDK/borrowed/old-task guards equal and owned VM reaped. Default optional unit skips retain their named gating reasons
Runtime-equivalent preceding four-consumer candidate corpus Source epoch 683daca0…33f224. Runtime123/binary74ad is byte-identical to current 196245a3…42d43. The corpus was not rerun after the test-only amendment. 853.249 seconds. Native/candidate-native 0/0/0/1, API/view 0/0/0/2. Diagnostics 28/144/120/11250, coverage 29/34/41/29. Summary a1e98ca3e10095a1134381666bacf256f8e8827cd3900c9e811b7120de4c2974, receipt 05c4562a50d9f83ba2c99879ec841870c5e753199e41792bd5bc718cf8046e7b, independent audit 3a1b0b6e3a8c6b1a0d82c5f82b46c84b1e44d6c30bab655610cb9e86e6a30b47
Final canonical lifecycle and rendered checks The promoted TEN bytes use independent canonical rendering and navigation/URL receipts

The preceding six-gate b40b7787b3da8dc1e0763812b6dde529b4b5b69fe479d79940f1161223124e1d, host/archive bcb4d7599e965c1b3cfe7fe698ca14061ad53d45e7a194337aeebb8d37aa77c1, and ARM outer a022e40f068703cd59ce6d6a7fb6530cce6907681baa26eb1dfc77c09f0c8898 / guest 409990bc1425f4bf219f8911a71581af6e68729865580121dbeb6d85a06d2ea7 / audit 24afc50f6f860394d1ebfa7a8b754ddd9cb97f9e88a0dcfcbcb659193ecbfe5f remain retained alongside the new exact-source proof. Historical 26 axe checks/screenshots and 22 codec cases are carried with unchanged UI/codec/runtime inputs.

The initial max-CPU AMD trial remains failed: receipt 3543664ba5590f2ba5a8f676b196bb636b72bc819913289f415d0a8a841c1bdb, guest summary 16075204d287713c7f7650c0a65dd289dd4bd83db07c9ba4b85b3f21244d5240. The test client timed out after 90 seconds while awaiting candidate headers. Candidate parity and the remaining six selected Hugo cases/native rebuild/installed smokes were not reached. Guest inputs stayed exact. The host guard recorded only a .git directory timestamp change. The separate qemu64 one-test preflight also failed at the unchanged 90-second HTTP client deadline: outer receipt fc68173ccdfd8ce263ecdf082a533d9da666a4cc2e1e5e29880ee827286132ac, guest summary e350ff65feeee166ffac1d337db9bbd70d3895b1fb6d93df0a30ca4de09019fc. The named case took 177.71 seconds. The first trial took 176.64 seconds. Its inputs remained exact and its VM was reaped.

A later Go-overlay diagnostic preserved the same production source and every original semantic assertion. Outer receipt 0bc6d563b7cd9ca862717c2123ee0836d83b6b927d00204a0b031049c38e93f0 and raw-bound classification 66a1422cdb79ab9f1cf683f441ade0ce4adb4a7a666d524c4b9ed98ebee28708 record a passed named case in 352.40 seconds. Original capture took 26.254 seconds, Studio capture 26.211, candidate HTTP 94.312, direct preview 94.318 and CLI preview 81.962. Both graphs retained 1,256 unique inputs, including 1,198 module inputs. The original-capture context had expired when the HTTP result arrived. Fresh independent direct/CLI contexts completed normally. All 20,564 host guards and five guest command guard pairs stayed exact. The owned VM was cleanly reaped. The scoped owning-fixture amendment now uses a 300-second budget for that candidate request and fresh direct/CLI operations, about 3.18 times the slowest observed operation. General/original-capture 90-second limits, restoration of the shared client, shutdown 15 seconds and Go’s default ten-minute cap remain unchanged.

The consumer driver compared complete typed diagnostics, coverage, native exit, PlanID, selected Base/After/full modes, unified diff and selected source between API and CLI. It separately verified the complete literal API review and its hash. For attachments and no-ops, the selected page retains reviewed Base. Page-file edits match reviewed After. Nonissue views are bounded samples. All issues and pages are paginated. The 53 protected operations have 212 all-four per-operation source proofs plus four overall proofs (864 raw inventory pairs across four categories) and 53 root pairs. Seventy-one retained artifacts are bound. No Apply, saved plan or consumer write occurred. The completed corpus’s file-only collector needed two preserved metadata corrections for absent historical trial/self-test files. No CLI/Hugo operation was rerun.

The repository retains its 10,462 pre-existing duplicate-ID findings and 788 review-info records. Its selected actual DFE HTML is available. Four oversized actual PRINT files remain unserved (413, zero response body): _print/pkg/index.html 73,976,221 bytes, _print/pkg/pgsql/index.html 69,903,999, zh/_print/pkg/index.html 73,086,240 and zh/_print/pkg/pgsql/index.html 69,052,754. The 64 MiB per-file preview bound is unchanged. Required partial-preview incompletion remains 2. Native findings remain 1. Apply is refused.

Linux prerequisites were prepared in exclusively owned guests from signed Debian metadata. Exactly ten new packages and three approved existing-package updates were verified before and after installation. SDK/Hugo/module caches were provisioned separately and reused offline. Qualification runs as an ordinary user on ext4. Cached inputs and all 212 source files’ bytes/full modes stay guarded. Default unit skips retain their actual gating/not-applicable reasons. All required pure top-level cases, no-skip membership children, selected Hugo cases, and signal cases must execute. Linux amd64 uses QEMU TCG on the ARM host and is explicitly emulated. Darwin amd64 remains an experimental archive. Actual execution returned Bad CPU type (errno 86). Windows is outside the declared scope.

The current Linux archive digests are ac883e54a1df0b820696279c63881ba75a00d279f507330128fe8d5aff59c52e (arm64, 4,552,687 bytes) and 2dde43bf94ef35aac2111b07dcb9b2766fbf9f883fe39ccd646d14a98b94d734 (amd64, 5,034,668 bytes). The preceding 683daca0…33f224 archive digests c191383af21913be6940ec41be11755b3d985344bbc0f65cc3f5de16424a96a4 and 6531b27d889260afe804c1f49f37541fbae46e57b5d17a20178c28cb51968794 remain historical. SDK/guest preparation failures, the first ext4 membership failure, and earlier private host metadata/resources trials remain immutable evidence. Uninvoked E1–E4 extensions are separate inactive scope.

13 - Document comparison research, 2026-10-29

Dated evidence for a generic OINK side-by-side reading mode across languages and revisions.
Historical research; see the 1.5.0 decision

Research date: 2026-10-29. This bounded source review of OINK v1.4.0 informed the translation reader in OINK 1.5.0. The accepted decision and the shell contract describe the released behavior and supersede the recommendations below.

Scope and evidence

The question was how OINK could support a generic comparison reading mode for roughly 30–40 local sites: translated documents, eight-language sets, proofreading books, arbitrary languages, and old/new revisions in one language. The review covered Docusaurus, VitePress, Material for MkDocs with mike, Read the Docs, Hugo, Sefaria, MediaWiki ContentTranslation, OmegaT, Weblate, Joplin, VS Code, and CodeMirror. Evidence came from official documentation and pinned source revisions; no upstream application was started and no browser behavior was claimed from source inspection alone. Unpinned official documentation records what was read on 2026-10-29; it is not an immutable source snapshot. OmegaT evidence is limited to the 6.1.0 manual, not an inspected implementation.

Project Fixed documentation/source revision Verified relevance
Hugo v0.168.0 source commit e27a6760c5547237f87d353aebe0580feadb20f4 Page relationships and native version rotation; no paragraph alignment.
Docusaurus, VitePress, Material/mike Docusaurus v3.10.2, VitePress v1.6.4, Material 9.8.0, mike v2.2.0 Stable identity, explicit anchors, version metadata, and existence-aware navigation.
Read the Docs and VS Code RTD platform commit 9b012b1df15fc153590b4db5f4da7bd3eb8ae74a; Addons d8f99069d9a8b9647f5bd68fdedb95ffa11e63ce; VS Code d7658021df8289f549bf2af8c8dec131f2580d69 Real rendered HTML/Markdown comparison and mapped previews; neither proves a generic OINK reader.
Sefaria and ContentTranslation Sefaria af96a1a1695a6307a3d616493455a6802455be1b; ContentTranslation 8d5dde493528330efcb373dc448f352e81815115 Independent panes and strict segment rows are separate products.
Joplin, OmegaT, Weblate, CodeMirror Joplin b48e75856d81152c32e71f21075e1c043d9ea922; OmegaT Standard Manual 6.1.0; Weblate 503129cadb056e2ea0db629fe8e37907b41ff054; CodeMirror 53c8f7b213332743a1f4f7d26bf40cfe2b512b53 Logical scroll maps, manual alignment, stable message context, and geometry spacers.
DDIA consumer sample Vonng/ddia a6b2e6f85e1249720c8a349273455190d2fc9515 Four Hugo members with repeated Chinese locales and distinct editions.

Identity and relationship lessons

For the original general-purpose scope, comparison would need to keep candidate-set identity, member identity, Hugo site key, actual locale and direction, edition/version, canonical URL, and anchor correspondence separate. A language code is not a document identity. The pinned DDIA configuration provides a concrete case: zh, tw, v1, and v1_tw are four configured members that include repeated Chinese locales and different editions. Deduplicating by locale would remove valid choices and mislabel editions as translations.

Hugo 0.168.0 provides a sound starting point. .AllTranslations includes the current page and .Translations excludes it; an explicit translationKey can relate pages whose paths differ. .Rotate "version" can adapt native version dimensions while holding other dimensions, including language and role, fixed. That adapter is optional for comparison and does not require a consumer to reorganize content. It must not be confused with an OINK params.versions menu, which is a navigation configuration rather than a semantic partner resolver. Custom version or role dimensions still need an explicit adapter so ambiguous native matches are not silently accepted.

Automatic translation candidates should therefore begin with the actual Hugo page relationship, then pass an eligibility filter: published, built, renderable HTML siblings; the selected page is excluded; home pages used as language-navigation fallbacks are excluded; foreign role or version members are excluded when the set requires that boundary. A navigation link proves that a destination exists, not that it is a translation or a comparable document. Missing translations remain missing; OINK must not copy source text behind the reader’s back.

The original generic design used two selectors for an eight-language set. It does not require 28 prebuilt pair pages or eight simultaneously loaded documents. Same-language v1/v2 comparison uses two member IDs and explicit labels. The same identity model also supports a proofreading book whose chapter structures differ, without pretending that equal filenames or title text establish correspondence.

What existing tools actually compare

Docusaurus document identity, localization, and versioning separate ID, slug, locale, and version tree; its heading guidance advises stable IDs for localized pages. VitePress constructs locale links from the current relative path in the inspected language helper and supports custom anchors, but that helper does not establish a semantic partner registry. Material uses mike version metadata and target sitemaps to find a corresponding URL; that confirms addressability, not semantic equivalence. These are useful identity and validation patterns, not evidence of native two-pane scrolling.

Read the Docs supplies direct fixed comparison evidence: its visual-diff documentation and Addons docdiff.js compare a rendered HTML body, mark changed DOM regions, and preserve a URL mode flag. This is a single-page rendered diff for review workflows. VS Code’s Markdown diff documentation and diffScrollSync.ts show two rendered previews with logical mappings and feedback suppression. They demonstrate that rendered comparison and mapped scrolling are feasible, while leaving OINK’s content identity and static-host constraints unresolved.

Sefaria explicitly distinguishes an independent Compare Text view from a shared-segment bilingual view: TextRange.jsx renders both texts inside a shared segment, while TextColumn.jsx keeps comparison columns independently scrollable. MediaWiki ContentTranslation’s alignSectionPair uses stable section IDs and height/margin compensation for paired translation regions in a single editing flow. OmegaT’s Align Files manual combines automatic matching with manual split, merge, move, and acceptance operations. Weblate’s Unit model keeps message IDs and context separate from text hashes. CodeMirror’s updateSpacers shows the geometry cost of aligned rows. Together they show why automatic similarity can suggest a relationship but cannot silently publish a paragraph map for arbitrary multilingual HTML.

Three reading semantics

The design must name three different behaviors. Independent scrolling gives each pane its own position and is the fallback for every valid pair. Anchor-linked scrolling keeps two independent panes and uses shared semantic anchors to estimate corresponding progress between anchors. It is a nearby section reference, not permanent paragraph or line alignment. Strict row alignment puts actual blocks into paired rows, uses the maximum row height and spacers for the shorter side, and needs one-to-many mapping, missing-block placeholders, and an explicit order policy. It is a separate later feature.

For OINK, the proposed default depends on mapping quality: enable anchor-linked scrolling only when the conservative conditions below pass, otherwise use independent scrolling. The reader can always turn synchronization off. A pair with zero or one valid internal anchor, duplicate explicit targets, missing explicit targets, or non-monotonic moved anchors stays independently readable. Ordinary unmatched sections are content gaps, not invalid targets. A direct “jump to counterpart” action can still work where continuous synchronization cannot.

Rendering and isolation finding

OINK’s v1.4.0 Docs shell, content rendering, Book ID handling, tabs, table reader, diagrams, and capability-selected scripts make arbitrary single-DOM body concatenation a large refactor. Two same-origin document frames preserve each page’s lang, dir, resource base, renderer context, IDs, and reader ownership. A frame is a document boundary, not an independent process or a security sandbox; same-origin content remains trusted content. See the iframe reference and same-origin policy.

The initially considered path used a prototype with ordinary eligible HTML pages in two same-origin frames and guards for navigation, keyboard shortcuts, analytics, comments, and recursive compare entry, then considered an optional content-only HTML view. The revised first phase keeps ordinary pages for the implementation too; a dedicated output is only a future optimization if measurements justify it. Hugo parameters alone cannot create an output format; any auxiliary output needs explicit output plumbing, Page Store/capability handling, relative-resource rules, and SEO policy.

Scroll model and bounded conclusions

The useful engine pattern is Joplin’s logical map and VS Code’s mapped rendered previews. Joplin’s scrollmap.js and VS Code’s diffScrollSync.ts support collecting layout-valid anchors, interpolating within intervals, and clamping positions. The revised translation MVP derives a strictly monotonic mapping from shared stable heading IDs; it needs no separately authored or reviewed pair map. Crossing or reordered anchors reject that map; the engine must not silently choose an LIS, drop moved anchors, or reorder the document. Geometry is invalidated when fonts, images, diagrams, tabs, details, or width changes alter layout. A state owner and sequence token prevent programmatic scroll events from echoing back. The user should keep focus and pane ownership; ordinary scrolling must not open hidden content or move focus.

An unmatched section is an ordinary content gap when the page and its surrounding anchors are valid. If valid anchors bracket it, the reader may label the interval approximate and interpolate through it. In a future manual-mapping feature, an explicit target that does not exist, is duplicated, or crosses another explicit target is an invalid mapping; it disables continuous synchronization for the pair rather than becoming an ordinal fallback. A direct explicit jump can report the invalid target.

The evidence supports a focused proposal, not an implementation claim. The upstream source investigation did not run the candidate comparison feature, a server, a browser, accessibility checks, or an eight-language consumer. It also did not prove that every framework lacks plugins or experimental comparison modes. Those limits are carried into the proposal’s acceptance matrix and keep the recommendation honest: first prove normal Docs/Book reading with two panes, then decide whether richer output or strict row comparison earns its maintenance cost.

Phase-one scope revision

Following the request to simplify on 2026-10-29, the companion proposal now assumes existing translation siblings with matching heading IDs. Its only site setting is params.ui.compare: true. Hugo supplies the existing translation links; the browser chooses one target and displays two ordinary pages in same-origin frames, with an on/off synchronization switch. The current language is fixed and only the target needs a selector.

The broader identity discussion above remains research for possible version comparison. It does not require phase-one compare sets, member IDs, manual anchor maps, a CLI, new layouts, new outputs, or prebuilt language pairs. Ordinary pages in frames are the recommended first implementation, not a necessary step towards a content-only output. The existing host remains behind the reader: two child documents plus that host means three documents, a cost to measure before adding another rendering path.

14 - CLI acceptance snapshot, 2026-09-29

Executed Starter, real-site, offline, upgrade, and reproducible-archive checks for the local CLI candidate.
Historical CLI acceptance snapshot (2026-09-29)

This page records the 2026-09-29 validation of the local 0.1.0-dev CLI candidate at commit e623d93. Current CLI behavior is defined in the CLI and result contract.

Inputs and method

The CLI belongs to the independent oink-cli Go repository. The CLI and result contract defines its accepted boundary. The CLI introduction gives reproducible user steps. Hugo remains an external renderer. Generated sites contain normal Hugo inputs.

Input Observed baseline
Host macOS, darwin/arm64
Go go1.27.1
Hugo 0.166.0+extended+withdeploy
CLI 0.1.0-dev, local commit e623d93d589c49e5c58b8fae1bd5db720fc904cb
Embedded Starter Commit 137843b25bacd76ddd1f7ce71330bf2e3155b954, complete licensed Git archive
Generated theme pin Public github.com/pgsty/oink v1.1.0, with recorded Go checksums
Documentation-site theme Local theme HEAD b0af631 plus uncommitted changes

The Starter archive hash is e55bde279715f6d8d19d3d88671a2cf7561b515be46915b0f12c640d0ce1d958. Its recorded projections select an existing language profile, pin OINK v1.1.0, and set enableGitInfo: false for a new directory. An observed failure required the last projection. The original enableGitInfo: true caused a warning-strict build to fail before the site’s first Git commit. No Git repository or commit was created to hide that failure.

Checks used disposable source copies, module/render caches, and output directories. Original Starter and consumer source trees were not written by the CLI checks. Existing unrelated theme and documentation edits were retained. The counts below describe those inputs and CLI revisions on 2026-09-29.

Starter and ordinary Hugo

All six ordinary-Hugo cases passed --environment production --panicOnWarning, with provisioned modules and isolated caches:

Language profile Root URL /manual/ subpath Reported Hugo pages
en Passed Passed EN 90
en,zh Passed Passed EN 91, ZH 89
all Passed Passed EN 91, ZH 89, FR 89

The tests checked the expected language roots and representative Docs, Blog, and Book outputs. They compared generated source bytes before and after Hugo. The public CLI’s init command also passed separately for all three profiles, returning zero diagnostics and 94 generated source files per profile. The profiles select different root configurations and disable untranslated sample files while retaining them in the snapshot.

Starter package unit, race, and vet checks passed. Failure cases cover nonempty and symlink targets, validation failure, and target replacement after planning. They also cover cancellation rollback, concurrent modification/deletion, and archive path rejection. The regeneration script reproduced the fixed archive, provenance, and license exactly.

Real-site inspection snapshot

Each run below returned CLI exit 0 with zero recorded diagnostics.

Site shape and theme source Files HTML files References Machine artifacts
Three-language Starter, public v1.1.0, release check at /manual/ 316 142 7,042 6
OINK documentation/regression site, local theme HEAD b0af631 plus dirty changes 1,127 506 72,562 8
PIG project site, root Docs/Blog route rewrites, public v1.1.0 1,392 424 64,440 4
Repository documentation with generated catalog, public v1.1.0 3,287 1,635 851,535 12

The last three are local consumer repositories: PIG and the catalog site validate published-pin resolution, while the OINK documentation run validates the local theme checkout.

The inspection covered HTML links, anchors, resources, and emitted machine artifacts. The completed Hugo manifest enumerated 261, 766, 662, and 3,192 output declarations respectively. Every supported enabled machine output was required by its actual language and URL. Before/after manifests compared all tracked and non-ignored untracked source bytes, modes, and Git status. All four sites were unchanged (94, 415, 858, and 2,294 source files respectively).

Two real regressions were fixed during this work. A NAVJSON template rendered only English. This previously hid the missing Chinese output. The check now returns policy exit 1 with the missing output location. The check initially treated PIG’s intentional build.render: link sidebar entries as missing pages. Hugo’s effective build parameters now exclude them. Direct and cascaded regression cases verify this behavior. A paired build of the bilingual Starter also proved all 223 ordinary artifacts byte-identical before and after adding the isolated probe.

Offline execution and upgrade

On macOS, a full check of an initialized bilingual Starter passed under sandbox-exec with (deny network*), after dependency provisioning. The result was exit 0, zero diagnostics, 223 files, 95 HTML files, 4,461 references, and four machine artifacts. A separate English init also passed under the same OS-level network denial, returning exit 0, zero diagnostics, and the expected 94 generated files.

A cold-cache fixture requiring example.invalid/[email protected] returned CLI exit 2 and retained Hugo’s original module lookup disabled by GOPROXY=off evidence. A missing dependency was therefore reported as incomplete work, without silently enabling resolution.

A separate temporary site exercised a real public-module upgrade from v1.0.0 to v1.1.0. The original consumer was not used as a write target:

Operation Observed result
Preview Exit 0; candidate validated; applied: false; plan named only go.mod and go.sum
--write --expect-plan Exit 0; the matching plan was validated and applied
Repeat the same target version Exit 0; candidate validated; no proposed changes and applied: false

All three operations preserved unrelated dirty README.md content and untracked user-note.txt. The preview and write used the same plan ID and before/after module-file hashes. Replacement, workspace, dirty-file, rollback, and failure-protection cases passed the final focused Go tests and race run. Only go.mod and go.sum changed on write; backup manifests retained original bytes. Preview and repeat executions preserved all source bytes.

Real thin-wrapper smoke tests also passed in a disposable initialized site. build --json returned 0 and produced index.html. dev --json served HTTP 200, forwarded SIGINT to Hugo, and closed the listener. Hugo exited 0. The canceled wrapper reported 2, as specified by the cancellation contract. These runs used provisioned local caches without --network.

Final make test (all packages plus vet), make test-hugo (ordinary Hugo, workspace/config precedence, output manifest, missing-language regressions), and go test -race ./... passed. All three public init profiles were rerun under OS-denied networking; the cold dependency fixture again returned 2.

Archive and installation preparation

One frozen CLI source snapshot produced four binary archives and one source archive, plus SHA256SUMS. Rebuilding independently from the source archive produced the same SHA-256 values for all five archives. The tested packaging input hash was:

b07c5b98ef787dfe9924ce7b50c57d018c6149ec493124bb0103551a01535547

This final snapshot supersedes the intermediate archive experiments. All five archive checksums were verified and reproduced from the extracted source archive. Both source and binary archives include the versioned JSON schema, licenses, dependency pins, and Starter provenance. Local make install into a temporary prefix and the installed binary’s --version succeeded.

Target Evidence
darwin/arm64 Compiled; host binary executed; local installation path exercised
darwin/amd64 Cross compiled only; not executed on that architecture
linux/amd64 Cross compiled only; not executed on Linux
linux/arm64 Cross compiled only; not executed on Linux

The archive builder records the toolchain, flags, source-input hash, and platform targets.

Reproduce the relevant checks

From a CLI checkout with dependencies already provisioned:

make build
make test
make test-hugo
go run scripts/snapshot-starter.go --source ../oink-starter

To repeat rendered-site checks in the sibling layout, keep JSON and logs outside each consumer’s source tree:

oink_acceptance_dir="$(mktemp -d)"
mkdir "$oink_acceptance_dir/reports"
./bin/oink init "$oink_acceptance_dir/my-docs" --languages all
./bin/oink check --site "$oink_acceptance_dir/my-docs" --release \
  --base-url https://example.org/manual/ --json \
  > "$oink_acceptance_dir/reports/starter.json" \
  2> "$oink_acceptance_dir/reports/starter.log"
HUGO_MODULE_REPLACEMENTS="github.com/pgsty/oink -> $(cd ../oink && pwd)" \
  ./bin/oink check --site ../oink.pgsty.com --json \
  > "$oink_acceptance_dir/reports/docs.json" \
  2> "$oink_acceptance_dir/reports/docs.log"
./bin/oink check --site ../pig.pgsty.com --release --json \
  > "$oink_acceptance_dir/reports/pig.json" \
  2> "$oink_acceptance_dir/reports/pig.log"
./bin/oink check --site ../repo.pgsty.com --release --json \
  > "$oink_acceptance_dir/reports/catalog.json" \
  2> "$oink_acceptance_dir/reports/catalog.log"

On a macOS host providing sandbox-exec, after initializing a bilingual site:

./bin/oink init "$oink_acceptance_dir/my-bilingual-docs" --languages en,zh
sandbox-exec -p '(version 1) (allow default) (deny network*)' \
  ./bin/oink check --site "$oink_acceptance_dir/my-bilingual-docs" --json \
  > "$oink_acceptance_dir/reports/offline.json" \
  2> "$oink_acceptance_dir/reports/offline.log"
sandbox-exec -p '(version 1) (allow default) (deny network*)' \
  ./bin/oink init "$oink_acceptance_dir/offline-en" --languages en --json \
  > "$oink_acceptance_dir/reports/offline-init.json" \
  2> "$oink_acceptance_dir/reports/offline-init.log"

The command reference describes preview, plan review, and explicit application. Use a separate review copy for write-path testing. For the archive experiment, keep the same Go toolchain and release version:

make release VERSION=0.1.0-dev DIST=dist/first
mkdir -p dist/rebuild
tar -xzf dist/first/oink_0.1.0-dev_source.tar.gz -C dist/rebuild
make -C dist/rebuild/oink_0.1.0-dev_source release \
  VERSION=0.1.0-dev DIST=dist
cmp dist/first/SHA256SUMS \
  dist/rebuild/oink_0.1.0-dev_source/dist/SHA256SUMS

Limits and delivery state

State At this snapshot
Local implementation Six first-stage commands and versioned result format exist
Executed validation The runs described above passed for their recorded inputs
Owning checks and documentation-site make check Passed after implementation and bilingual-document updates
Commit, tag, push CLI committed locally as e623d93; no tag, remote, or push. Documentation changes remain local alongside existing work
Public CLI release or distribution Not performed
Consumer source adoption or production deployment Not performed by these checks
Independent-user study or adoption No measured 4-of-5 / 15-minute study, retention, or independent-team adoption data

Raw JSON, logs, source-preservation manifests, upgrade recovery evidence, and archive-verification results are stored in the CLI checkout’s ignored tmp/acceptance/ directory, with release archives in dist/first/.

This first-stage candidate focuses on existing OINK sites; Docsy assessment and later migration, theme descriptor, version lifecycle, OpenAPI, MCP, and Studio remain tracked in the CLI roadmap.

15 - Component refinement, 2026-10-25

Local four-preset component validation, asset measurements, and separately reproduced gallery baseline failures.
Local acceptance record

Phases 0–3 and final acceptance in phase 5 are complete for this component change. The moving selection trial remains deferred.

Inputs and method

The final review used isolated sibling checkouts based on theme 7f9589f and documentation 855d890, with only this task’s source changes applied. Concurrent PlantUML, toolchain, and workflow work was excluded from the candidate and preserved in the shared worktrees. The accepted implementation is theme commit 15c39ef; guides, contracts, examples, and browser coverage are in documentation commit 81e2be9.

Validation used macOS ARM64, Hugo Extended 0.168.0 with built-in LibSass, Go 1.27.3, Node 26.12.0, and npm 11.20.0. Browser previews used the isolated sibling theme and hugo.yml,tests/fixtures/browser.yml, with both workspace mechanisms, watching, and live reload disabled. Screenshots came from actual documentation output. The bilingual comparison guide owns the reusable examples.

Executed validation

Check Result and scope
Theme component owners SCSS/compiler hooks, fonts, presets, content primitives, components, code blocks, and Landing passed. Invalid variants and inapplicable command strips warn and preserve content.
Theme shell/output owners Shell, actions, palette, keyboard, runtime isolation, and output checks passed. Output security passed on 501 fixture files with the fixture’s intentional third-party embeds enabled.
Real site make check passed all 64 non-browser tests, including bilingual headings, rendered Markdown, and internal links. Reviewed navigation/backlink additions account for the site’s Markdown golden changes.
Full browser suite make browser passed 409 Chromium tests. After the last correction to hide an empty command action column without JavaScript, all 20 presentation tests and the SCSS checker passed again against a fresh build.
Presentation coverage Four presets × light/dark × EN/ZH; 1440/768/390/320px layouts; literal clipboard text; independently operable native and full card links; native disclosures; no JavaScript; print; coarse pointers; reduced motion; forced colors; and Landing feedback width/contrast.
Additional engine review Chromium 151.0.7922.34, Firefox 153.0, and WebKit 26.5 passed 48 preset/mode/locale combinations covering Tabs, Fields, card links, and 1440/390/320px layouts. The final no-JavaScript command presentation also passed in all three engines.
Visual review Reviewed actual Paper, Slate, Ink, and Terminal cards, disclosures, Tabs, command strips, and field labels in light/dark and desktop/mobile output. Stable viewport captures preserve each preset’s geometry and reading hierarchy.
Production The final candidate passed a minified production build. Resource measurements below use the CSS linked by the generated page and the same documentation source for both builds.

The full suite’s A11Y_PATHS covered EN/ZH Presentation, Cards, Callouts, Tabs, Code, and Fields, plus English Badge and Kbd: 14 routes. The existing responsive accessibility matrix ran as well. Presentation axe scans included the article at 390px; the Landing check included its command box in all eight preset/mode combinations. Giscus requests were blocked as in the existing suites.

Asset cost

The exported theme base and the final candidate used the same documentation source and production build mode:

Asset Before After Change
Minified main CSS 528,665 bytes 539,450 bytes +10,785 bytes
Main CSS, gzip 79,541 bytes 80,887 bytes +1,346 bytes
Built JavaScript 39 files Identical hashes 0 bytes

These final measurements supersede the earlier 539,437-byte / 80,856-byte initial refinement measurement. No framework, font, remote asset, or JavaScript dependency was added.

Tab review follow-up

Maintainer review rejected the raised selected Tab. The accepted treatment keeps individual labels transparent at rest, supplies a subtle fill only on hover-capable devices, and uses link color plus a 2px bottom rule for selection. Labels retain the same medium weight and position. Keyboard focus has its own outline; coarse-pointer targets remain at least 44px high. Final engine review verified these states, arrow-key selection, and narrow layouts.

Field names and final corrections

Terminal field names now use the same unboxed, emphasized monospace treatment as the other presets. Both native Markdown tables and full Fields shortcodes were checked in EN/ZH and light/dark output. Ordinary prose code and keycap borders retain their separate role.

Final review reproduced and fixed two further edge cases:

  • A native card’s stretched title link intercepted a secondary link nested inside emphasis. Secondary controls now remain above that target; pointer checks verify both the nested link and the rest of the card navigate correctly. The title’s keyboard focus outline follows the card’s own shape.
  • A command strip retained a small empty action column when JavaScript was disabled. The wrapper now disappears until its copy control is available. In all three engines, the source column occupies the full 356px inner width of the 390px sample, while the normal enhanced copy control remains usable.

Baseline failures and limits

Two theme checks fail identically in the exported base and the component candidate:

  • check-namespace.py: committed blog gallery output uses unprefixed data-height, data-width, data-mode, and data-size, with corresponding unprefixed dataset reads.
  • check-goldens.py: the committed gallery’s prepaint condition, toolbar icon, and gallery markup are absent from the existing theme goldens. Comparing base and candidate with the checker’s normalizer found no additional difference across all 52 golden surfaces.

The component checker also includes a targeted correction: reject obsolete configuration keys rather than any occurrence of gallery, allowing the valid blog_gallery key.

Local acceptance covers this component change and its two final repairs. Ink and Terminal retain their experimental status. Runtime JavaScript is unchanged.

16 - Review repairs, 2026-11-06

Reproduced build, feed, navigation and interaction defects, bounded repairs, and remaining review work.
Development evidence

This record describes local changes after v1.5.0. It does not establish a new release, consumer upgrade, or deployment. The original review’s roughly 170 entries have not all been accepted or fixed.

Inputs and method

The review started from theme 84b93e4 and documentation a8eeb92. The candidate keeps Hugo Extended 0.168.0 and its built-in LibSass, with Go 1.27.3, Node 26.12.0, npm 11.20.0, and Python 3.14.8 on macOS ARM64. The Hugo 0.169.0 executable on the normal system path is outside this validation baseline. Its golden differences are not evidence of a defect on 0.168.0.

Build and output claims were checked with synthetic inputs in the owning theme checkers. Interaction claims were checked against the sibling documentation site through Playwright. Browser previews used the local theme replacement and hugo.yml,tests/fixtures/browser.yml, with both workspace mechanisms, watching, and live reload disabled. Existing passing snapshots were reviewed before recording intentional changes.

Confirmed defects and repairs

Area Confirmed behavior and repair Repeatable evidence
Invalid input Positional arguments reached map-only shortcode operations and failed the build. Named-parameter guards now warn and safely omit or fall back. Negative checks require exit status zero. check-content-primitives.py, check-book.py
Book numbering Invalid numbers produced empty targets such as fig-. Unnumbered content now retains readable output without target registration. check-book.py: HTML, print, Markdown, BookManifest and index rows, with valid-number controls
Nested components Independent Markdown documents and sibling shortcode containers reused local ordinals. Generated IDs and their references now include the containing scope. Authored code IDs are preserved. check-code-blocks.py: repeated nested code, fields, charts and FileTree in HTML and print
Feeds Weight-based order was limited before date sorting, undated items emitted year-one dates, and several components emitted raw Markdown. Both feeds now sort before limiting and omit unavailable dates; components use static HTML or their documented omission. check-content-primitives.py, check-book.py, RSS goldens
References and versions Missing manual references could fail Hugo; version links could repeat the deployment prefix. Shared resolvers now provide warning fallbacks and join version paths once. check-agent-indexes.py, check-actions.py
Navigation under subpaths Language-neutral navigation IDs and section-index keys retained the deployment prefix. A shared key resolver now removes deployment and language prefixes while published links keep both. check-agent-indexes.py: EN/ZH under root and subpath deployments, with and without a default-language directory
Landing Invalid record shapes and optional fields could break Markdown output. HTML and Markdown now share record and URL validation, preserve fragments and queries, and include standalone CTA links. Unsafe targets do not become empty links. check-landing.py
Blog navigation List pages inherited article-sequence navigation and English pagination. Localized semantic pagination now owns list navigation and head links. check-reading.py, check-shell.py, browser pagination regression
Focus and menus Native disabling removed focused controls; hidden self-links were unreachable; focus-opened menus reopened after Escape. Guarded aria-disabled controls retain focus, self-links remain reachable, and Escape restores focus before closing. JS units and browser code, shell, comparison and Expo suites
Mobile shell Navbar-only articles lacked mobile utilities, and closed sidebar drawers leaked a shadow. Site drawers now expose language, version and repository links; shadows appear only when open. check-shell.py, responsive browser regressions
Layout Book TOCs displayed two counters, prose margins affected component lists, Landing padding was overridden, and global link rules suppressed print underlines. The owning rules are now scoped correctly. Browser review regressions and SCSS checks
Locale and schema Pagination and breadcrumb labels were untranslated; several locale strings and the appearance legend were incorrect; the schema rejected language-code translation notices. Labels now cover all 32 locales and generated schemas match the resolver. check-i18n.py, generate-config-schema.py --check
Preset and alternate rendering Terminal’s current-row chevron selector was wrong; focus rings were translucent or clipped; forced colors hid current-state cues; the RTL artifact overwrote font roles. Targeted rules now retain these states. Repeated marquee rows are hidden from assistive technology. Preset/SCSS checks and browser accessibility coverage

Qualifications to the original review

The original report’s severity labels grouped build failures, accessibility defects, wording, and design preferences together. They should not all be treated as release-blocking failures. Reproducible build and output defects deserve the first repair batch; cosmetic consistency needs its own acceptance criteria.

“No TOC” does not mean “empty right rail.” Blog indexes can still contain author, category, series, and tag clouds. Removing the fallback control made that rail impossible to collapse and failed the existing RSS/browser regression. The final repair omits genuinely empty rails and labels populated non-TOC rail controls as sidebar actions. It preserves collapsing, focus restoration, and RSS access.

RSS fixes do not require every component to appear in feeds. Release assets keep their explicit omission. The Book checker now exercises the ordinary section feed instead of treating raw RenderShortcodes output as the RSS contract.

Different corner radii, duplicated template shells, and native/shortcode layout differences are maintenance or design concerns until a specific contract or user failure is demonstrated. Merging base templates solely because their bytes are equal can change Hugo lookup precedence. No such broad consolidation is included.

Validation

The theme’s focused checks, schema parity, 32-locale parity, 80 JavaScript tests, and 90 migration tests were run on the fixed toolchain. Output changes cover 52 HTML, print, Markdown, RSS, and LLMS golden surfaces. These checks are local unit and output evidence; the separate CI PDF/EPUB publication job is not claimed.

The documentation site’s final make check passed, including all 64 non-browser tests, with a larger Node heap for its rendered-link scan. A local production build also passed and generated 462 pages per language, with only the accepted upstream LibSass deprecation warning. The first browser pass completed all 30 accessibility tests; the responsive group passed 117 tests after retaining the taxonomy rail control. The other canonical browser groups also passed in their respective runs: keyboard 16, content components 70, code blocks and Landing 37, PRD5 8, theme color 10, community feedback 42, appearance 53, Expo 50, gallery 8, PlantUML 8, diagrams 19, table reader 15, standard comparison 19, and eight-language/RTL comparison 6. The canonical browser set totals 508 passed tests across these runs. A later focused accessibility run passed all 30 cases on the new research and blog pages. After the Escape correction, a fresh shell, blog and menu regression run passed 56/56. The final Landing checker passed after its metrics links were routed through the same URL policy.

The browser counts describe this repair candidate before separate comparison interface work appeared in the shared checkout. That concurrent work is outside this record’s browser acceptance. The completion audit additionally verified navigation keys under language/deployment prefixes and invalid Book numbers in HTML, print, Markdown, BookManifest, and the generated indexes.

Remaining work and order

This is a bounded repair batch. The following work remains outside its acceptance:

  1. Reader control over single-character shortcuts, command-palette handoff and localized error messages, and plural-aware contributor/search/code counts.
  2. Non-default-language Google search destinations and the documented meaning of relative content URLs.
  3. Remaining contrast cases and the previously recorded narrow, large-text footer overflow. These require their own rendered reproductions and browser checks.
  4. Geometry, interaction-state and overlay tokens, remaining manual-link owners, shared shell structure, and stylesheet size. These need bounded design decisions and measurements rather than a blanket replacement of literals.

The component, shell, architecture, and Landing contracts describe this batch’s development behavior. Historical release claims remain unchanged.

17 - Ink and Terminal need reading acceptance

Source review, bounded visual evidence, current failures and promotion limits for Ink and Terminal on 2026-10-26.
Pre-release research snapshot (2026-10-26)

This record preserves inputs and measurements from before the OINK 1.4.0 release preparation. At 320px with a 32px root/default font, footer links cause document overflow across all four presets. Ink and Terminal remain experimental; the stabilization proposal defines desired behavior and promotion gates.

Keep Ink neutral sans editorial and Terminal a readable engineering handbook. Both can become coherent official styles through their existing role system and shared shell. A new serif identity would weaken Ink’s distinction from a future Folio; a simulated terminal would weaken long-form engineering reading. The immediate work is shared text resilience, quieter hierarchy, and component consistency.

The evidence identifies a local snapshot

This 2026-10-26 review combines primary-source text/HTML/CSS inspection, an exported source snapshot, real documentation rendering, and a bounded injected CSS comparison. Theme HEAD was 7bc8658bc1be13ef0ad3050b9d9f6862f8f1f65f; documentation HEAD was 99751179eefada3fa7a91318c792ddac46d7e6bf. The snapshot also included uncommitted changes, including concurrent table work; the retained source manifest records that full state. Public tag lookup observed v1.3.2, peeled to bd2e856e6e2f7027a9189af009701f5b8cf562cf; documentation pinned v1.3.2. See the earlier experiment and component review for prior context.

The documentation server used the sibling snapshot with hugo.yml,tests/fixtures/browser.yml, both workspace mechanisms off, memory rendering, no watching/live reload, and enableGitInfo=false because the export lacked Git metadata. Tools were macOS ARM64, Hugo Extended 0.168.0, Go 1.27.3, Node 26.12.0, npm 11.20.0, and Playwright 1.62.1; engines were Chromium 151.0.7922.34, Firefox 153.0, and WebKit 26.5 (Playwright WebKit scope).

The reproducible method loads the same real route, language, mode, and preset at each width, compares baseline with the listed overlay, records full-document dimensions and computed styles, then captures relevant components. Source manifests, review-metrics.json, api-supplement.json, text-resize-findings.json, stress-metrics.json, reference-rendering.json, and screenshots are stored outside Git.

Whole-document overflow outweighs targeted passes

At 320px/32px root font, English Ink had scrollWidth=424 (+104px), Terminal 408 (+88px). Chinese Chromium overflow was +12px/+21px respectively; English Terminal also exceeded 768px by 11px. The overlay produced the same failures. Firefox’s font.size.variable.x-western=32 preference reproduced English overflow across all four presets; Paper reached 413px and Slate approximately 400px. Footer links including “Customization”, “Components”, and “Docsy discussions” exceed two narrow columns. This is a shared baseline blocker across all presets.

These probes exercise CSS root enlargement and native default-font preferences. Combining 320px width with enlarged text is a stronger stress condition than standalone reflow. Full manual 200% text enlargement and 400% browser zoom remain pending (Resize text, reflow).

Owning checks: check-presets.py passed text/link/accent checks over three surfaces, mode parity, frozen Slate, and configuration output; check-font-tokens.py passed.

Existing suites: 28 appearance-engine, text-reflow, and component tests passed. Their page-context/topline assertions missed the footer. The source-snapshot make check passed 64 non-browser tests before these documents were added. With the new documents, the site built and passed Markdown, bilingual heading, and internal-link checks; 63 of the 64 non-browser tests passed, with llms.txt golden differing only by the two new English document entries.

Main measurement matrix: Chromium: 11 valid routes × two presets × two modes × EN/ZH × 320/390/768/1440px × baseline/overlay. Firefox/WebKit: presentation/configuration/Book only at 320/1440px. Normal-size pages showed no document overflow.

Counting correction: The initial ledger’s 960 entries were computed-state observations. A mistaken API route produced 32 harness 404s; the corrected /docs/write/openapi/ supplement contributed 64 valid observations with HTTP 200 and no document overflow.

Additional stress: 64 simultaneous spacing observations used line-height 1.5, paragraph margin 2em, letter spacing .12em, and word spacing .16em; no page/heading overflow was observed. Sixteen 320px long-unbroken heading/Field-label probes preserved text without page overflow.

axe: 32 article scans at 390px covered presentation/configuration, both languages/modes/presets, and baseline/overlay. No violations were reported; all scans retained contrast incompletes on aria-hidden=true pager-arrow spans with non-text characters, which require manual graphic-contrast inspection.

Visual material: The archive contains 352 comparison captures, including Blog/Expo/Mermaid/ECharts scenes, plus eight external references, from which representative desktop/mobile and light/dark pages and controls were reviewed.

Long reading: The real EN/ZH /blog/oink/oink-implementation-diary/ article was checked in four selected combinations: Ink EN/1440/light, Ink ZH/390/dark, Terminal EN/390/light, and Terminal ZH/1440/dark. Baseline/overlay captures at top/middle/footer produced 24 additional images; normal-size document overflow was zero throughout. Eight representative images were inspected, including English before/after prose, Chinese prose, Terminal’s English mobile footer, and Ink’s English desktop opening.

The Chromium routes were home, Presentation, Configuration, Fields, Code, Table, Book structure, immersive-reading Blog, Expo studies, Mermaid, and ECharts, plus the corrected OpenAPI guide, with Chinese peers included. The capture pass also verified the first rendered Mermaid diagram and an ECharts canvas. The comparison gallery spans 11 scenes; seven control scenes cover both presets, modes, languages, and two widths. Offsite requests, including Giscus, were blocked in the main matrix.

The overlay tests a limited editorial change

Ink kept its palette/fonts while changing article h1 from 800/-.035em to 750/-.025em, h2/h3 from 750/-.025em to 700/-.012em, and its 2px emphasis top rule to a 1px existing strong-line rule. It removed enforced uppercase/tracking on minor/table/callout headings, used zero Chinese heading tracking, and compared native underlines at max(1px,.06em)/.22em, with Chinese .25em and skip-ink: none.

Terminal kept teal/mint and amber, used heading weight 600/zero tracking, removed generated heading prefixes and callout brackets, changed descriptive table headers to body 0.875rem/600 and search titles/excerpts to body, and reduced decorative code/tab/table/callout frames to the existing border role. Technical paths/controls stayed mono and Field names stayed unboxed. FileTree’s omitted literal 0.6rem radius became 0px/2px for Ink/Terminal. Flat Tabs, mobile targets, and vendor boundaries remained intact. The overlay tested CSS adjustments only, without adding new theme templates, fonts, runtime, or configuration.

Authored accent colors on actual pages were documentation blue #245f94/#5da2dd, Book orange #9a3412/#fb923c, and Blog violet #6d28d9/#a78bfa. Separately labeled palette probes reset only that authored accent to the preset default on both variants for controlled comparison. The architecture contract preserves overrides and the shared shell.

Palette calculations expose state risks

Opaque sRGB calculations and alpha compositing used source tokens and the WCAG luminance formula; displayed ratios are rounded while pass decisions use unrounded values (Contrast method, palette source).

Pair, light / dark Calculated contrast Implication
Ink accent #c8102e on white / #ff5c4d on #0b0b0b 5.88 / 6.46 Retain mode-specific red; keep red accents focused rather than repeating decorative red rules.
Terminal body on canvas 14.88 / 13.64 Strong neutral reading pair.
Terminal link on canvas / secondary surface 6.31 / 9.54; 5.75 / 8.73 Existing teal/mint supports these exact surfaces.
Terminal tertiary text on secondary surface 4.63 / 5.93 Light muted text has little margin for further tint/opacity.
Terminal amber on canvas 5.47 / 9.44 Amber remains a separate emphasis role.
Terminal 32%-alpha border on canvas 1.97 / 2.36 Suitable for decorative rules; insufficient if it alone identifies a required control or state.
Terminal 55%-alpha strong line on canvas 3.63 / 4.73 Candidate functional boundary on these canvases.
White on dark mint #4cc9bd 2.02 Avoid white labels on light mint fills.

Cross-mode reuse fails ordinary text contrast: mint on the light canvas is 1.85, dark teal on the dark canvas 2.79. Semantic tones, author accents, syntax comments, selected search excerpts, focus, and text-selection composites still require state-specific review. Non-text contrast applies to necessary identifying boundaries (WCAG 1.4.11).

Mature precedents support roles, not transplantation

All external references were accessed 2026-10-26. External Chromium rendering at 1440/390px returned HTTP 200 for Tufte, GOV.UK Headings, Effective Go, and A List Apart. Eight captures were saved, with visual review focused on Tufte desktop, GOV.UK mobile, Go mobile, and A List Apart desktop; the remaining precedents received source review.

Practical Typography. Puts body text first; prescribes Latin-oriented size, spacing, and measure. Its underlining advice opposes link underlines.

Review complete paragraphs before display titles, while keeping OINK’s bilingual rhythm and open-source font constraints.

Tufte CSS. Inspected desktop shows a narrow serif reading column and underlines. CSS declares ET Book, light #fffff8/#111, dark #151515/#ddd, and mobile note toggles.

Borrow proximity of explanation/figures and note hierarchy; leave serif/margin geometry to future Folio exploration.

GOV.UK Headings. Inspected mobile view has generous sans headings and underlined links; cookie banner was visible. Current type scale uses 19px/25px body sizing on small and large screens. Links distinguish prose from navigation; tables emphasize structure.

Strongest functional input for Ink’s sans hierarchy and persistent link cues, adapted to OINK’s bilingual scale and dark mode.

A List Apart: A Dao of Web Design. Inspected desktop distinguishes serif body from sans headings/controls. Current CSS uses Georgia 1.125rem/1.7, red #ba3925 h2, and blue links.

Separate reading/UI/code roles and keep adaptable content while keeping Ink’s sans-serif identity and restrained accent frequency.

Rust Book: Ownership. Source exposes numbered chapter structure, explanations, and code; CSS separates Open Sans prose and mono code.

Strong Terminal precedent for explanation/syntax separation and real chapter numbers, while keeping OINK’s mono headings and standard browser shortcuts.

Effective Go. Source/CSS separates platform sans and mono; inspected mobile TOC is dense and two-column.

Borrow straightforward technical hierarchy while keeping OINK’s responsive TOC.

man7 open(2). Stable reference sections, signatures, links, and preformatted text; (2) names a manual section.

Preserve predictable lookup and complete identifiers without forcing bilingual prose into fixed-width terminal cells.

CLI Guidelines. A navigable explanatory web guide recommends intentional, limited color.

Borrow restraint and clear labels while keeping standard web interaction semantics.

Paper’s warm general reading, Slate’s established technical presentation, Ink’s neutral editorial restraint, and Terminal’s technical role contrast remain distinct within the shared shell.

Language and accessibility need independent evidence

Actual Chromium per-glyph inspection on macOS found Inter for Ink Latin, IBM Plex Sans for Terminal prose, IBM Plex Mono for Terminal headings, and PingFang SC for Chinese in both. Windows and Android fallback rendering still require direct device verification (CSS font matching, Windows fonts, Android fallback).

Chinese underline clearance remains open. The CSS Text Decoration Working Draft explains Han collisions and fragmented ink skipping; CLReq’s Group Note Draft discusses interlinear marks without banning hyperlink underlines. Next steps should compare language-scoped text-underline-position: under with a small offset and initial skip-ink: auto, including Chinese inside English pages and mixed-script wrapped links (CSS draft, CLReq).

WCAG A/AA governs color-independent meaning, text/non-text contrast, keyboard operation, resize, reflow, spacing, language, and status announcements. OINK’s proposed 44px routine touch targets, 7:1 main text, and stronger 2px-area focus treatment are tracked as project targets above AA minima (WCAG 2.2, axe-core, focus appearance).

Still pending are actual Windows/Android loaded/blocked/system-font comparisons; branded Safari/iOS rendering; manual NVDA, VoiceOver, and TalkBack tasks; full manual 200%/400% zoom; and comprehensive keyboard, forced-colors, and state-contrast review covering navigation, Book numbers, table relationships, field errors, Chinese search input, status feedback, and dialog return focus.

Promotion should follow resilient reading

The footer overflow highlights the need for whole-document regression checks alongside article-top checks. Shared repair and whole-document regression should therefore precede preset promotion. FileTree’s radius leak and competing Terminal punctuation are bounded owner-level changes.

For 1.4, refine existing owners and promote Ink or Terminal independently once their acceptance gates close; otherwise retain experimental status. Folio (academic serif) and Canvas (creator geometry) remain deferred to avoid font/licensing and QA overhead in this cycle.

18 - OINK 1.5.1 readiness review, 2026-11-06

Three reproduced candidate regressions, complete browser acceptance, and remaining publication gates.
Local repair acceptance complete

The three reproduced regressions below have been repaired in the working trees and the targeted, theme, site, and browser checks pass. This dated review does not announce a release, consumer upgrade, or deployment.

Inputs and method

The review covered theme v1.5.0 (84b93e4) through c55849e: all six post-tag commits and 186 changed files. The documentation site was 43990ac. Theme main was three committed changes ahead of origin/main (d9ad654). Both working trees were clean before the initial review. Its original evidence is retained below; the later repair pass is recorded separately. No tag, push, consumer update, or deployment was performed.

The formal local toolchain was Hugo Extended 0.168.0 with built-in LibSass, Go 1.27.3, Node 26.12.0, npm 11.20.0, and Python 3.14.8 on macOS ARM64. CI uses Node 24. Supplemental Hugo 0.169.0 checks do not replace the pinned baseline. The accepted upstream LibSass deprecation remained visible.

Methods included implementation and contract inspection, the ordered theme CI commands, sibling-site checks, production builds, Chromium tests, and controlled minimal reproductions. Temporary fixtures and generated artifacts stayed outside Git. Local-theme and published-module checks used separate commands with both workspace mechanisms disabled.

Confirmed regressions

P2: RSS format restoration contaminates another render

The final assignment in layouts/_partials/content/rss-description.html restores a captured tdOutputFormat into the page Store, which shortcodes also read. In the reproduction, a summary render restores html while the page’s RSS shortcode render still uses that Store, producing mixed static and interactive output.

The formal bin/check-release-assets.py run failed because the RSS checksum table header disappeared and data-td-asset and <button appeared. A standalone rerun and eight single-page reproductions passed; rerunning until green does not resolve the original failure.

For a more reliable reproduction, expand that checker’s RSS fixture to 24 RSS-only pages, each with release-card and 20 release-assets blocks. Keep its custom layouts/docs/single.rss.xml, which sets tdOutputFormat to rss before .RenderShortcodes, and keep the ordinary home feed enabled. Build fresh output on Hugo 0.168.0 and inspect each page XML for interactive markers.

Three candidate builds all contaminated docs/rss-9/index.xml. One captured output had six static tables, fourteen interactive tables, and 28 buttons. The same fixture against published v1.5.0 passed three times. Overriding only the candidate summary partial to omit the final Store restoration also passed three times.

The confirmed scope is the custom RSS-page render exercised by the required checker, not every ordinary feed. Keep format state stable within its rendering context and add multi-page coverage while preserving HTML, Print, and normal RSS summaries. The temporary removal was a causal probe, not an accepted fix.

P2: Blog pagination duplicates Print aggregates

layouts/_partials/head.html calls shell/blog-index-state.html for every blog section, including Print. That partial calls .Paginate, so Hugo generates Print archive pages that each contain the entire section.

Use three dated blog posts, outputs.section: [HTML, print], and params.ui.blog_index_size: 1. The candidate generates _print/blog/index.html, _print/blog/page/2/index.html, and _print/blog/page/3/index.html; all three contain all three bodies. Overriding only head.html with its v1.5.0 version generates one aggregate. Both builds succeed on Hugo 0.168.0.

The documentation site’s production build also adds four duplicate aggregates across its English and Chinese blog roots and release indexes, totalling 1,120,141 bytes. Large archives amplify this waste. Limit blog-index pagination and its head links to HTML, then verify one Print aggregate and correct HTML pagination links.

P2: A stalled Duo View document never reaches Retry

assets/js/document-compare.js returns immediately for about:blank, including when its 15-second timeout calls ready(true). A pending main-document request therefore consumes the timeout without invoking failure handling. Loading and busy states remain without Retry or a direct-link fallback.

In the sibling site’s browser fixture, open /tests/compare-reader/ and hold /zh/tests/compare-reader/?oink_compare_pane=1 pending with a Playwright route. Open Duo View and wait 17 seconds. The source pane is ready, the target remains about:blank, and there is one loading note, zero failure panels, and zero Retry buttons.

This is not covered by the existing 404 and slow-image tests. Timeout expiry must fail a still-blank document while ignoring its initial blank load event and preserving parsed articles with slow images. Add the stalled-document case to the site’s comparison suite and retain the localized fallback promised by the shell contract.

Initial validation

Check Result and scope
Theme CI commands 48 of 49 runnable steps passed on Hugo 0.168.0. The release-assets RSS step failed; theme acceptance is not all green.
Theme tests 81 JavaScript tests, 90 migration tests, and 22 snapshot/consumer-tool tests passed. Locale parity, schemas, vendor inventory, output goldens, and Sass-boundary checks passed.
Site make check Passed, including 67 Node tests and source, translation, rendered-Markdown, and rendered-link checks.
Site make browser A single complete invocation passed 518 Chromium tests in approximately 23 minutes. A11Y_PATHS was unset: the sitemap scan was unrestricted.
Candidate production build Passed with the local theme and ordinary production configuration. Only the accepted LibSass warning appeared. Print duplication was inspected separately.
Publication isolation All three PDF safety tests passed with real Chromium, including remote CSS blocking and disabled scripts.
Publication smoke test Current-candidate EPUB/PDF generation and owning artifact checks passed: five chapters, 31 EPUB targets, and a 23-page PDF. Local Pandoc was 3.12.1. The CI-pinned Pandoc 3.10, Java 21, EPUBCheck, and subpath matrix were not rerun locally.
Published baseline Go proxy v1.5.0 resolved to 84b93e4 with the site’s recorded checksums. The release-pin check and production build passed with replacements and both workspaces disabled.

Browser groups passed 30 accessibility, 117 responsive-shell, 16 keyboard, 70 content-component, 37 code/Landing, 8 PRD5, 10 theme-color, 42 community-feedback, 53 appearance, 50 Expo, 8 blog-gallery, 8 PlantUML, 19 diagram, 15 table-reader, 26 standard comparison, and 9 multilingual comparison tests. The local sibling theme used hugo.yml,tests/fixtures/browser.yml and the separate eight-language comparison fixture. Firefox and WebKit were not rerun.

The suite at the initial review did not cover the additional stalled-request and Print cardinality cases above. The earlier review-repair record explicitly excludes later Duo View work from its 508-test result; this review supplies fresh evidence for the current candidate.

Repair verification

The repair pass changes the working trees based on theme c55849e and site 43990ac. The changes remain uncommitted and are not included in the remote CI runs linked below.

  • RSS summary rendering no longer restores the stale format value. The owning release-assets checker now renders 24 RSS pages with 20 asset tables each alongside the home feed, asserting every static table and checksum and the absence of interactive controls. A temporary override with the pre-fix partial fails this check; the repaired partial passes.
  • Blog pagination in the shared head is limited to HTML. The reading checker verifies one complete Print aggregate, three distinct HTML archive pages, correct HTML prev/next links, and no Print pagination links. A temporary override with the pre-fix head generates three Print aggregates and fails; the repaired head passes.
  • Duo View distinguishes document timeout from the initial blank load event. The new browser case holds the target document pending, checks the Retry label and direct link, and verifies that Retry restores the pane when the request can proceed. It passes with the repair. The earlier slow-image case continues to cover readiness without waiting for all resources.

The bilingual architecture and shell contracts and the theme’s Unreleased changelog accompany these fixes. The site’s LLMS golden gains only the new research-page entry; the output diff was reviewed before updating it.

Repair acceptance Result
Theme CI commands All 49 locally runnable steps pass on Hugo Extended 0.168.0, including 81 JavaScript, 90 migration, and 22 snapshot/consumer-tool tests.
Site make check Passes all source/rendered checks and 67 Node tests.
Site make browser Final complete invocation passes 519 Chromium tests: 510 in the main suite and 9 in the eight-language comparison suite. A11Y_PATHS was unset, so sitemap accessibility coverage was unrestricted.
Production build Passes with the local theme. Each EN/ZH blog and release index has one Print aggregate and no Print pagination links; all four duplicate aggregates are gone.

Browser acceptance uses the installed Playwright browsers without an executable path override, allowing the bfcache case to select full Chromium. That case and the new stalled-document Retry test also pass individually. These results use the same local tool versions recorded above; the publication-toolchain limits in the initial validation still apply.

All three reproduced regressions are locally resolved. The next release gates remain commit and push of the reviewed changes, green remote CI for that exact revision, then the separate tag, Go-proxy, consumer-pin, deployment, and production-identity states.

Publication and adoption

The latest observed public tag remains v1.5.0. The green theme CI run, including Book publication, belongs to d9ad654, not local c55849e. The green site checks and browser workflow also tested the remote theme available then. They do not certify the three later local commits.

A read-only bin/update-consumers.py v1.5.0 inventory found 50 primary checkouts: 42 current, three module consumers skipped for non-default branches, and five classic themes/oink consumers. All 45 module pins were v1.5.0. Four classic theme repositories were clean at the exact v1.5.0 tag. The remaining source copy matched sampled implementation files; that is not a full identity audit. No consumer was upgraded or deployed. The skipped branches and classic copies need individual handling during eventual adoption.

Release exit criteria

The three repairs, their owning checks, browser coverage, and bilingual contracts are present locally, and the working-tree acceptance recorded above is green. The next gate is green remote CI on the final release commit. Deferred design and accessibility work from the earlier review remains a separate backlog; this record does not claim every historical finding is fixed.

After code acceptance, prepare the 1.5.1 changelog and current version references, push reviewed commits, and verify CI for that revision. Tagging, push and Go-proxy verification, consumer pinning and appearance checks, deployment, and production identity are separate subsequent states. Those future steps are normal release preparation and remain distinct from local repair acceptance.

19 - Theme review, 2026-10-13

Reproduced image escaping and interaction defects, their repairs, and local theme and consumer validation.
Local review snapshot

This record describes local source and tests against the v1.2.0 consumer baseline.

Scope

The review began at theme commit f8f476e, with existing consumer-maintenance and compact-sidebar changes and an ongoing English documentation revision in the shared checkouts. Those edits were preserved. During the review, the existing sidebar repair was committed separately as 8f1f08d; its file contents remained unchanged.

Work covered the theme’s owning checkers, runtime unit tests, synthetic output fixtures, the real bilingual documentation site, and two different consumers: DDIA’s book and SILO’s large documentation tree. Source inspection focused on URL handling, image output, message boundaries, focus transitions, runtime isolation, and navigation. Visual review used real consumer pages and the documentation site.

Reproduced defects and repairs

Defect Reproduction and observed result Repair and regression
Image URL escapes its HTML attribute An ordinary image or Gallery URL containing a quote followed by an attribute-shaped query value creates an extra HTML attribute. An ampersand can also change the browser’s decoded URL. Escape URLs as HTML attributes after validation. check-media-primitives.py parses block and inline image attributes in HTML, Print, and RSS; check-gallery.py checks the quoted Gallery URL. The harmless extra-attribute probe failed before the repair.
Draw.io loses endpoint parameters and mishandles messages An endpoint with an existing query or fragment receives malformed embedding parameters. Null or malformed messages can throw; message handling does not require the configured origin. Build the URL with URL and searchParams; require the active iframe and configured origin; ignore malformed data; send to the exact origin and accept only PNG/SVG image export URLs. New Draw.io unit tests and the site’s editor integration tests exercise these paths.
Page Actions loses the keyboard interaction contract ArrowDown opens the menu before its CSS visibility transition permits focus. Tab can leave the disclosure while it remains open. Make visibility immediate, retain the opacity and movement transitions, and close on focus leaving the entire disclosure without reclaiming focus. Runtime and real-browser regressions cover forward and backward traversal.
Appearance clears a newly reopened dialog’s state Close and immediately reopen the dialog before its queued native close event arrives. The dialog stays visible while aria-expanded becomes false. Ignore an earlier close event while the dialog is open. Real-browser tests cover the race in Chromium, Firefox, and WebKit at desktop and phone widths.

The two taxonomy HTML goldens also needed the already accepted sidebar repair: 33 hidden-leaf markers in each snapshot were removed after reviewing the complete diff. The 52-surface golden gate then passed.

The behavior is recorded in the bilingual architecture and shell contracts. No configuration migration or new dependency is required.

Validation

Local tools were Hugo Extended 0.166.0, Go 1.27.1, Node 26.9.0, Pandoc 3.11, and Chromium 151.0.7922.34 for PDF generation.

Layer Evidence
Theme runtime All 52 JavaScript tests passed after the repairs. Python discovery passed 116 tests; the 90-test migration suite was also run separately.
Theme ownership Every bin/check-*.py checker was exercised, including explicit output-security arguments and packaged EPUB/PDF checks. Configuration schema generation was checked. Changed media, Gallery, output and golden owners were rerun after their repairs. Builds used warning-failure gates where the checker builds valid publishing output.
Documentation integration make check passed with the local theme replacement: source/rendered Markdown, bilingual heading coverage, internal links, alternate-site builds, favicons, output goldens and release-pin contracts.
Browser integration make browser passed 229 tests, including the full sitemap accessibility scan, six responsive widths, EN/ZH, navigation, search, components, code, community regressions and all four visual presets. The additional engine gate passed 6 tests across Chromium, Firefox and WebKit at 390 and 1440 px.
Publication fixtures EPUB: 5 chapters, 31 link targets, no internal-check errors; EPUBCheck 5.3.0 reported zero errors and warnings. PDF: 23 pages, all 5 expected titles, no internal-check errors; rendered pages were inspected.
DDIA sample Local-theme make check-local passed, including 14 translation chapters, 196 figure-layout checks and 38,170 internal references across 57 second-edition pages. Desktop, phone and dark reading views and the four-style menu were sampled.
SILO sample Local-theme make check passed, including 536,561 internal references across 1,552 HTML files. Desktop documentation and phone search results were sampled.

The initial broad run exposed the stale goldens. One standalone security invocation also lacked its required CLI arguments; its corrected invocation passed over 493 fixture output files. A new Print assertion initially omitted Print from the test page’s own output list; correcting that test setup produced a passing HTML/Print/RSS regression.

Performance observations

A strict, minified local production build of the documentation site was inspected. For /docs/customize/config/, the referenced local JavaScript totals 62,192 bytes gzip (about 61 KiB, including Lunr); the two stylesheets total 98,838 bytes gzip (about 97 KiB), excluding fonts, images, search indexes, and third-party service traffic.

The ordinary documentation page omits the large diagram bundles. The Mermaid sample adds a 976,006-byte gzip Mermaid dependency (about 953 KiB) plus its small page runtime. Runtime-isolation and output checkers passed.

Limits and delivery state

The accessibility suite excludes upstream Redoc and Swagger UI DOM under its existing policy. Consumer browser checks sampled key desktop and mobile routes across DDIA and SILO. The repairs were grouped into local theme and documentation commits.

20 - Give large tables room to read

Framework and component evidence, measured reading concepts, and a bounded direction for OINK technical tables.
Historical research; see the 1.4.0 decision

Research date: 2026-10-26. This study informed the table-reading design in OINK 1.4.0. The accepted decision and table guide describe the released behavior and supersede the entrance and scope recommendations below.

An explicit Read table action opening a large native dialog is the strongest candidate for further evaluation. It gives desktop tables more room while preserving text size, original table semantics, and a short return path. Inline widening remains sufficient for many tables; the measured package excerpt already fits that mode on desktop. On a 390px viewport, the dialog adds only 16px of width, so horizontal scrolling remains on narrow screens. Five documentation frameworks use native tables and overflow handling by default, while third-party table libraries add data operations alongside additional integration weight.

Evidence separates layout, content, and release state

The public tag snapshot was v1.3.2, with tag ref 68cfbbdc177064e2c28277c09c385d78e5363ffd. Research began at theme HEAD 7bc8658bc1be13ef0ad3050b9d9f6862f8f1f65f and site HEAD 99751179eefada3fa7a91318c792ddac46d7e6bf, with uncommitted changes in both working trees. The short .wide alias tested in this experiment was subsequently shipped in OINK 1.4.0 as documented in the component contract.

The local owners inspected were oink/layouts/_partials/content/table-render.html, table-body.html, assets/scss/td/_content.scss, _table.scss, the preset/font tokens, and the surface coordinator. .wide/.td-table--wide control intrinsic sizing and token wrapping; .full-width reaches the available article canvas. .matrix supplies first-column row headers and bounded two-axis scrolling, with existing sticky header/first-column CSS. That CSS gives each header top:0, without accumulated multirow offsets or a mobile first-column width guard. .fields emits a definition list. Ordinary captions belong inside the table; numbered Book captions belong to the outer figure (Table guide, output contract).

Selected consumer source Bounded sample and purpose
pigsty.cc/content/docs/infra/param.md, NGINX group Four columns × 14 body rows; linked parameter identifiers and descriptions
pgext/content/list/pkg.md, lines 44–69 Six columns × 24 body rows; package/version/repository/category/RPM/DEB comparison
pigsty.cc/content/docs/ref/linux.md, overview Eight columns × 14 body rows; OS/architecture and PG18–PG14 support cells
piglet.run/content/docs/reference/vscode/_index.md Two columns × seven body rows; small-table counterexample

Consumer snapshots were pigsty.cc at efaf7b19b442ddb284106418c78854552d949e9e, pgext at ab87fc7e5707d57f55129d3c45874abcc16fdef8, and piglet.run at 5a1572e6f9110cd5e8ef6e75cd50d9c4b6429f98; each pinned OINK v1.3.2. The separate 616-row extension catalog was left outside this prototype’s scope.

These examples distinguish many columns, many rows, long cells, and complex headers. Wider viewports primarily help the first; sticky orientation helps the second; wrapping helps the third. Multirow and merged headers need explicit associations and offset handling beyond wider layout. The selected prototype tables all have one header row and no merges. GFM pipe-table syntax itself defines a single header row and inline cell content (GFM tables, WAI multilevel headers).

Five frameworks make overflow the common baseline

The survey inspected official documentation and pinned default-theme source across five documentation frameworks. None of the five provides a built-in fullscreen table reader by default. Documentation links point to pages accessed on 2026-10-26; source links pin implementation commits.

Framework snapshot Default table behavior Optional/custom boundary
Material for MkDocs, 0f044dbe, 2026-10-24 Markdown tables; original table wrapped in local horizontal-scroll containers, with unwrapped fallback Sorting uses an opt-in Tablesort 5.3.0 recipe; no built-in table reader, sticky columns, or table copy (Guide, CSS).
Docusaurus Classic 3.10.2, f37f9035, 2026-07-10 MDX/GFM renders tables; Infima 0.2.0-alpha.45 applies display:block; overflow:auto Custom table views are built through MDX/remark/rehype (MDX mapping, Infima CSS).
VitePress 1.6.4, 1fc537b7, 2025-08-05; 2.0.0-alpha.21 snapshot d6d7a970, 2026-10-26 Both snapshots use block tables and horizontal overflow The 2.0 snapshot adds default tabindex="0"; Vue components handle custom behavior (1.6.4 CSS, 2.0 focus plugin).
Astro Starlight 0.42.7 snapshot af9557ea, 2026-10-24 Markdown table styling uses display:block; overflow:auto .not-content supplies a styling boundary; full-view and code-fullscreen features come from community plugins (CSS, directory).
Nextra 4.6.5 snapshot d6e80e1d, 2026-06-23 Official Table components wrap styled HTML with horizontal overflow Literal MDX <table> requires Table or the documented whitelist (Source, HTML boundary).

The inspected code licenses are MIT: Material, Docusaurus, VitePress, Starlight, and Nextra. Docusaurus documentation separately uses CC BY 4.0.

For static Hugo, the directly applicable patterns are semantic HTML, scoped scrolling, build-time focus attributes, and optional progressive enhancement. Material’s mounting code uses RxJS; Classic uses React/MDX, VitePress Vue/Vite, Starlight Astro, and Nextra Next/React.

Richer libraries add data operations and ownership

Current versions matter. DataTables 3.1.4 runs without jQuery, unlike 1.x releases. TanStack Table 9.3.1 uses constructTable, feature selection, and Store reactivity (DataTables v3 dependency boundary, TanStack v9 announcement).

Component Fit and available operations Integration and maintenance boundary
DataTables 3.1.4, released 2026-10-13; MIT, no required core runtime dependency Enhances HTML tables using an internal data model; core sorting/search/paging; optional Buttons, FixedColumns, FixedHeader, Scroller Closest candidate for later systematic table operations. Scrolling can duplicate or move headers; supports complex thead, not tbody spans (Release, package, span limit).
Tabulator 6.7.1, released 2026-10-22; MIT, no required runtime dependency Grid rendering, sorting/filtering, frozen columns, grouped columns, clipboard and virtual rendering HTML importer reads first-body td cells, dropping .matrix row-header th semantics, and does not preserve general caption/scope/headers (Release, package, importer).
TanStack Table core 9.3.1, released 2026-10-12; MIT, @tanstack/store ^0.11.2 Headless sorting/filtering/pinning/selection/spanning logic; vanilla integration is supported Maintainer owns HTML import, rendering, controls, styles, focus, export, and dialog; virtualizer is separate (Metadata, fixed package, vanilla guide).

DataTables’ FixedHeader/FixedColumns are optional extensions. Scroller requires equal-height rows, which conflicts with wrapped technical descriptions. Tabulator emits ARIA grid structures designed around cell navigation. TanStack’s spanning API requires explicit column definitions rather than importing arbitrary HTML merges. None of these packages includes a built-in fullscreen reading dialog (Tabulator’s printFullscreen targets print layout) (Scroller, Tabulator navigation, Print source, TanStack spanning guide).

Copy defaults also differ. Buttons 4.1.2 exports display data with current order/filter rules and span placeholders, without formula escaping by default. Tabulator 6.7.1 joins cell values with tabs/newlines without general quoting or formula neutralization. Tablesort 5.7.1 is a lightweight MIT, dependency-free sorting library that preserves table rows and handles Enter on headers without adding a reader, filter, or export layer (Buttons source, Tabulator Clipboard, Tablesort source).

Compare delivered bytes with their exclusions

Measurements used 14 fixed npm tarballs verified against registry integrity. Publisher-minified files were measured unchanged. Gzip used Python 3.14.8, zlib 1.2.12, gzip.compress(data, compresslevel=9, mtime=0); totals sum separately compressed files.

Measured delivery Minified bytes Gzip-9 bytes
DataTables 3.1.4 core JS + default CSS 153,523 46,181
Above + Buttons 4.1.2, FixedColumns 6.1.1, FixedHeader 5.1.2 JS/CSS 249,015 72,480
Scroller 3.1.1 JS/CSS, additional 13,300 4,389
Tabulator 6.7.1 full UMD JS + default CSS 483,872 107,512
TanStack 9.3.1 selected reader logic, local bundle 49,505 14,248
TanStack 9.3.1 all core exports plus bindings/render helper, local bundle 115,146 30,983
TanStack virtual-core 3.19.7, separate local bundle 32,872 9,745
Tablesort 5.7.1 core + number comparator 3,756 1,763

DataTables core alone is 123,516/42,338 JS bytes and 30,007/3,843 CSS bytes, minified/gzip respectively. Extension totals use the exact Buttons, FixedColumns, FixedHeader, and Scroller packages plus matching -dt CSS. Optional style-adapter JS adds 2,501/1,279 bytes to the four-package total; Scroller’s adapter adds 650/326. Initialization, dialog code, translations, JSZip/pdfmake, and source maps are excluded. Buttons copy resides in its main file, so the compatibility shim is not counted again. Tabulator’s full package includes all built-in modules; Tablesort supplies only sorting.

TanStack measurements come from local esbuild bundles. Node 26.12.0/npm 11.20.0 and esbuild 0.25.11 bundled core 9.3.1 with Store 0.11.2 using --bundle --minify --format=esm --platform=browser --target=es2020 --legal-comments=inline and production NODE_ENV. The selected entry exports constructTable, tableFeatures, sorting/filtering/pinning/spanning features, sorted/filtered row models, storeReactivityBindings, and FlexRender. The all-core entry exports the core namespace plus those two helpers; the virtual entry exports virtual-core separately. Metafiles found no external imports. The 14,248-byte gzip logic sample covers headless state management only, excluding UI, CSS, HTML import, accessibility controls, dialog, copier, and virtualizer (Core package, Store, Virtual).

The concept widens desktops much more than phones

The external prototype used a real sibling documentation-site Hugo build with an external content mount. PGEXT-specific shortcodes became explicit labels/links; parameter fragments resolved to their source URLs; support icons received “Supported” text alternatives. Added captions and .wide were experimental adaptations. Every reading mode reused the same generated table data. The build used the sibling replacement with both workspaces disabled and external destination/cache/resource directories.

The comparison shell used a 230px rail, 40px desktop main gutters, 68ch normal measure, and 1130px available canvas, styled with generated OINK CSS and fonts. A separate generated table-guide reference measured a 770px desktop / 358px mobile viewport.

Package-table scroll viewport Normal measure Inline widening Native dialog Separate page mode
Desktop Paper/Terminal 652.80px 1130px 1358px 1408px
Desktop Slate/Ink 686.38px 1130px 1358px 1408px
Mobile, all four presets 358px 358px 374px 374px

These are local qa.json measurements at 1440×1000 and 390×844, DPR 1; light/dark geometry matched within each preset. Text remained 16px and code 14px throughout. Mobile intrinsic table widths remained about 969–1009px. Desktop inline widening already fits the six-column excerpt; the dialog adds bounded height, persistent controls, and isolation from prose. Its desktop visible table height was 847.20px, versus 897.20px in page mode; mobile dialog heights ranged 711.02–741.20px as controls wrapped.

Inline widening has the lowest context-switching cost. The dialog gains desktop space while retaining a direct return to the article. The separate page gains only another 50px on desktop and no mobile width; its main advantage is reloadability and URL sharing, at the cost of losing article scroll/focus on return. None of these modes scales text or invokes the Fullscreen API.

Chromium 151.0.7922.34, through Playwright 1.62.1, checked 16 preset × color × device combinations, each with three reading concepts plus baseline. The package sample passed original-node identity, unique IDs, caption/header scopes, loaded styles/fonts, title initial focus, Enter open, Escape/Close return, both Tab boundaries, stable toolbar, separated control/table regions, and page-mode reload. No page overflow was recorded; three additional mobile samples and a 320px Terminal/dark overflow probe extended the scope. A separate inspection preserved link targets and selection and restored inline horizontal scroll to 180px. Screenshot review covered 48 concept captures and contact sheets, and a Paper/light/desktop dialog axe scan reported zero violations.

The parameter and version samples’ first columns occupied 59.3% and 58.3% of the 374px mobile reader, motivating an optional mobile width guard. Preset styling remained distinct: Paper’s Plex Sans and restrained framing, Slate’s Inter and rounded controls, Ink’s flat sharp geometry, and Terminal’s mono chrome with Plex Sans prose.

Reproduction uses the external builder, source manifest/hashes, generated reference output, QA/inspection JSON, and screenshot set with Hugo 0.168.0+extended+withdeploy, Go 1.27.3 on macOS ARM64, Node 26.12.0, npm 11.20.0, and Python 3.14.8.

This prototype scoped evaluation to Chromium desktop and touch emulation with single-row headers; Firefox/WebKit, screen-reader testing, forced-colors, 200/400% zoom, merged headers, coordinator integration, and print-while-open belong to implementation acceptance.

A small reader preserves more than a generic grid

The recommended first stage is an explicit Read table action, a native dialog, and one original-node move/restore lifecycle. Hover-only entry is undiscoverable on touch and keyboard; whole-table click competes with selection, links, and scrolling. A visible title/Close control, named scroll region, initial title focus, contained Tab navigation, Escape, and guarded focus/scroll return are core requirements. Reuse OINK’s surface coordination conventions. A near-full dialog is distinct from browser fullscreen, whose standard excludes <dialog> as the request target (APG dialog, Fullscreen Standard).

Retain caption/th/scope/headers and original IDs. Moving the original table avoids clone collisions and divergent state, while inherited styling, outer numbered captions, external ARIA references, and footnotes still need explicit handling. Native tables can contain links and sorting buttons without becoming ARIA grids. Start with single-row sticky headers; leave complex headers scrollable without stickiness until offsets are validated. Comfortable spacing is default; compact mode changes padding while keeping font size. Optional first-column pinning must leave useful mobile reading width (APG table, APG grid, header semantics).

Keep .wide/.td-table--wide layout-only, .full-width bounded by article canvas, .matrix semantics intact, and .fields ineligible. No-JS content remains readable; Print expands static tables, Markdown retains source, and RSS retains its static/summary boundary without leaking reading controls.

If TSV copy is added in the future, it should copy the whole eligible table, including offscreen rows/columns, in author order: links become labels, code text remains, hidden controls/decorative glyphs are excluded, and meaningful icons use explicit text equivalents. It should reject hidden data rows/columns, nested tables, unequal row lengths, multiple header rows, or any DOM rowSpan/colSpan other than 1 (including rowspan="0"), using TAB between cells, LF between rows, normalized in-cell whitespace, and a leading apostrophe for formula-like cells starting with = + - @ (HTML table model, DOM text, Clipboard 2026-06-24 Working Draft, OWASP spreadsheet-injection limits).

A focused native reader leaves data-grid adoption possible later while avoiding a second data model. See the accepted decision for the shipped implementation boundary.

Accepted direction after review

Maintainer review selected explicit {.wide .expand} composition and a direct corner action, revealed on hover/focus and visible on touch, omitting the above-table toolbar, ellipsis menu, and TSV copy. The accepted decision and current contract define the shipped behavior.