Book-theme parity cutover — tracking

[mmcky]

Repository: QuantEcon/quantecon-theme.mystmd
Revised: 2026-09-11 (eighteenth revision) — main @ d7ec6cd (v2.7.0 plus #200 and #202, no behaviour change); latest release v2.7.0 (2026-09-11); open PRs #195 (#114) off the gate; gate = the gaps found by the 2026-09-11 parity validation study, listed as native sub-issues; maintainer review complete: work the open sub-issues in list order; #183 and #181 wait on the designer
Scope: everything that must land before the all-at-once migration of the QuantEcon lecture series to MyST / Jupyter Book ≥2 on this theme.
Sources: PLAN.md on main; the book-theme parity milestone; the #93 maintainer resolutions (2026-08-24, recorded in #145); the 2026-09-04 scope change on #143; the 2026-09-11 parity validation study (quantecon-book-theme @ 649cc376, this theme @ 5260864, 103 features assessed and adversarially verified, headline claims re-checked against live sites).

This is a long-lived tracker, not a period plan: it is the single source of truth for cutover-gate state, revised in place against live state rather than accumulating narrative, and it is resumed and updated but never session-closed. Revision history is in the comments.

Next session — resume here

Every item on the original gate is merged and released. v2.7.0 (2026-09-11) carried #182 (pinned, scroll-tracking outline, #196), Phase 6 (#92, #197), the declared site options (#173, #194), the 1280–1327px layout correction (#198) and the bundle-manifest change (#191). Both consumers moved the same day: QuantEcon/lecture-wasm#77 is merged and live on v2.7.0, and lecture-python-programming#363 pins it on jb2.

Before closing, a validation study checked parity end to end (section below). It found 8 theme gaps that would visibly break or degrade lecture sites at cutover, or leave the parity record wrong, and they are now this issue's open sub-issues; content-migration dependencies in the QuantEcon mystmd fork are referenced under Sequencing. Work them in sub-issue list order, skipping #183 and #181 until the designer answers: the list is the plan.

Maintainer review of the gate items (complete on 2026-09-11, apart from the designer's two).

• How it ran: each open sub-issue was reviewed one at a time, in list order, before any implementation.
• Where decisions go: each is stamped on its issue as a comment. An issue whose plan changes is revised in place.
• Now: implement. Work the open sub-issues in list order, one issue per PR, following each issue's decision comment. Skip The page-header byline renders in a sky blue nothing else uses and fails AA contrast #183 and The in-content table of contents drops the period after every enumerator, and renders part titles as bare list items #181 until @DrDrij answers the follow-up questions on them.

Item	Review state	Outcome
#203 licence footer	✅ decided (decision)	A theme default: when a site sets no site.parts.footer, the theme renders the CC BY-SA 4.0 notice and "A theme by QuantEcon", hard-coded, with the badge as an inline SVG. A site's footer.md replaces the whole default. lecture-python-programming's jb2 footer.md is to be removed
#204 notebook header {raw} blocks	✅ decided; body revised in place (revision)	The fix leaves the theme, and the sources become mystmd-native. At cutover a rewrite script removes every {raw} block: the 549 notebook headers are deleted, 2 iframes become {iframe}, and 3 tables become HTML in the Markdown. lecture-wasm and the jb2 branch take it now, and this repo's PR preview runs it. The notebook header moves to a project setting (QuantEcon/mystmd#108). Open: where the rewrite script lives
#209 migration checklist	✏️ revised in place to carry #203 and #204 (revision)	Nine checklist items. The footer needs no step; the notebook header setting and the {raw} source rewrite are added
#205 Launch control	✅ decided, one part deferred; body revised in place (decision)	Launch becomes opt-in and explicit: it renders only when launch_notebook_repo is set and launch_colab is true, and no repository name is guessed. The options are flat for now (launch_notebook_repo, launch_notebook_branch, launch_notebook_dir, launch_notebook_source_dir, launch_colab), named as the nested paths they become once mystmd supports structured template options (QuantEcon/mystmd#112). Deferred: whether and where to check that a notebook exists before showing Launch. Current thinking is an in-page failsafe that shows it only when the notebook is found
#206 output plot alignment	✅ decided; plan unchanged (decision)	One CSS rule centres images in cell outputs, both static and under live compute, with a fixture image test. Tables and text outputs stay left-aligned, as on the Sphinx sites. No upstream-registry entry: this is a QuantEcon style, not an upstream feature
#207 canonical link and favicon	✅ decided; body revised in place (decision)	Canonical links come from site_url (the home page points at the site root, the trailing-slash form is used, and nothing is emitted without site_url), and og:url is built by the same function. The scope is widened to every head link that assumes the domain root: the favicon, /myst-theme.css and a full-URL og:image. #209 gains a site_url row. Redirects from the old URLs at cutover are filed as QuantEcon/mystmd#113
#208 outline h4 entries	✅ decided; plan unchanged (decision)	List h2 to h4 in a tree that supports any depth. Expand the current entry's sub-list and its ancestors, as the Sphinx panel does, marking only the current entry. contents_autoexpand: false stays out of scope. The h4 indent needs no up-front design sign-off. If this lands first, #215's criteria widen to h2 to h4
#183, #181	⏸ waiting on @DrDrij	Follow-up questions posted 2026-09-11: #183 (byline colour, size, element, translator line, underline split) and #181 (period separator, <ol>, part titles, self-listing)
#210 parity record	✅ closed 2026-09-11 and split (closing comment)	PLAN.md gets a closing note instead of a rewrite (#218). The docs lines no issue owns go to a final docs review (#217), the no-counterpart notes to #209, the drawer overlap and font-size reset to #219 and #220, and the dark palette to a design question on #187
#217 docs review	🆕 final task	Review README, docs/ and CONTRIBUTING.md against what shipped, before the PLAN.md close-out
#218 PLAN.md close-out	🆕 last task	Move the architecture, upstream-first and PR-preview guidance into CONTRIBUTING.md, then add a closing note to PLAN.md

Spun out of the review so far:

• Notebook header: ipynb export has no project-level notebook header, so the QuantEcon logo is copied into 549 lecture source files as raw blocks mystmd#108, the header setting for the ipynb export.
• Export gaps: ipynb export writes tables with colspan or rowspan as pipe tables, which cannot express merged cells, so the columns misalign mystmd#109 (span tables), ipynb export with markdown: commonmark writes {iframe} as MyST directive syntax, which Jupyter and Colab show as a literal code block mystmd#110 ({iframe} under markdown: commonmark) and ipynb export drops every raw node, including {raw} text and {raw} latex content, and logs an error that --strict does not count mystmd#111 (raw nodes).
• Source migration: QuantEcon/workspace-lectures#64, the jb1 → jb2 source-migration project. It inventories every construct mystmd does not render across 20 lecture repos, and plans the cutover rewrite as one PR per repo; its tooling is QuantEcon/workspace-lectures#65.
• New constructs: that inventory found four no issue had covered: capitalised {prf:Theorem} (live Unknown Directive callouts on lecture-wasm), {any}, {todo} and {fas}.
• Structured theme options: Template options can only be scalars, so site templates pack lists and nested settings into YAML strings mystmd#112 asks mystmd for object and list template options, so Launch control cannot be turned off and links to a nonexistent notebooks repo on sites without one #205's flat launch options, and languages and translators, can become nested settings.
• Dead Colab links on the Sphinx sites: Drop nb_repository_url, which points at a notebooks repo that does not exist lecture-dp#60, Drop nb_repository_url, which points at a notebooks repo that does not exist lecture-python.zh-cn#279 and Drop nb_repository_url, which points at a notebooks repo that does not exist lecture-intro.zh-cn#305 remove nb_repository_url values that name missing notebooks repositories.
• DataFrame outputs: HTML notebook outputs such as pandas DataFrames are missing from the served page and appear only after JavaScript runs #216. HTML outputs such as pandas DataFrames are missing from the served page and render only after JavaScript runs. It is not on the gate.
• Old URLs at cutover: The site build cannot redirect old page URLs, so every link to a Sphinx lecture page breaks at the move to mystmd mystmd#113. Every Sphinx page URL changes (about_py.html becomes about-py/) and the fork cannot redirect, so redirect pages are to be written at build time.
• From PLAN.md and the user docs claim book-theme parity and behaviour that the code does not deliver #210's close: The open contents drawer covers the start of the lecture text between 1024px and 1339px wide #219 (the open contents drawer covers text at 1024–1339px) and The font-size setting resets on every page load and is not offered on mobile #220 (the font size resets on every load and is not offered on mobile), both unparented, plus a dark-palette design question on Design review of the v2.5.0 lecture build #187.

Before working this plan, re-verify it. The facts below were measured at the revision stamp; anything since is unaudited.

Background

The parity project brings this theme to feature parity with quantecon-book-theme (Sphinx, JB<2) ahead of migrating the lecture series. Phases 0–2 (release infrastructure, git-history headers, launch parity with Colab + in-page JupyterLite/Thebe) shipped in v2.3.0 (2026-08-20); lecture-wasm is live on it and has tracked every release since (v2.7.0 as of QuantEcon/lecture-wasm#77, 2026-09-11). The #93 maintainer decisions (closed 2026-08-24) settled the remaining strategy: features are built local-first with upstreaming deferred to a registry (UPSTREAM-CANDIDATES.yml), and — most consequentially — the migration is all-or-nothing: every lecture series moves at once, onto a single common publishing base, when parity is complete. That decision converts the remaining parity backlog from nice-to-have into a hard cutover gate, and this issue is the single tracker for that gate.

One #93 resolution has since been superseded: the git-metadata plugin was to live permanently in this repo, but decision D8 of the report-theme design review (2026-09-02) created QuantEcon/quantecon-plugins.mystmd as the plugin family's home. The move is QuantEcon/quantecon-plugins.mystmd#11, this repo's side of it is #157, and the living-document corrections are #159. It does not gate the cutover.

Where we stand (verified 2026-09-11)

Phases 0–6 are complete and released in v2.7.0. The gate now holds the gaps found by the 2026-09-11 validation study, filed as sub-issues. The maintainer review is complete: every item is decided except #183 and #181, which wait on the designer, and #210 was closed and split into two final tasks (#217, #218). None is implemented yet.

Area	State
Phases 0–2 (release infra, git history, launch/live compute)	✅ shipped v2.3.0
Live compute correctness on shipped surface (#117)	✅ fixed (#137), released in v2.3.1
Static builds of the self-hosted stylesheets (#138)	✅ fixed (#139), released in v2.3.1
Residual static-build defect in route stylesheets (#150)	✅ fixed (#151), released in v2.4.0
Private JupyterHub launch option (#87)	✅ removed (#152), released in v2.5.0
Phase 3 — code highlighting + colour schemes (#89)	✅ merged (#171), released in v2.6.0 — defaults only, switches deferred; contrast decision open in #172
Phases 4+5 — i18n + RTL (#90, #91)	✅ merged (#174), released in v2.6.0 — no translated edition builds on it yet
Translator attribution (#143)	✅ merged (#174), released in v2.6.0, on the book-theme v0.22.0 placement
Back navigation on static builds (#186)	✅ merged (#192), released in v2.6.1
Pinned, scroll-tracking "On this page" outline (#182)	✅ merged (#196), released in v2.7.0
Phase 6 — Meta/SEO + collapsible stderr + parity docs (#92)	✅ merged (#197), released in v2.7.0
Parity validation study (2026-09-11)	🔄 Gap items from the study, plus #183 and #181 moved from #187. Review complete: #203, #204, #205, #206, #207 and #208 decided (#205's notebook-existence check deferred); #209 revised; #210 closed and split into #217 and #218; #183 and #181 wait on the designer; none implemented
Per-lecture live-compute enablement	➖ off the gate (#114); built as PR #195
Live-compute toggle design review	➖ off the gate (#128), in #187

Parity work that landed outside this item list. v2.4.0 (2026-09-04) and v2.5.0 (2026-09-05) carried a run of visual-parity changes that were never phased in PLAN.md: content typography matched to the existing builds (#155), the Sphinx lecture heading typography restored (#166), content links coloured QuantEcon blue (#167), the icon-only search trigger (#165), the contents drawer rebuilt on the Popover API (#144) and the first-paint fixes (#168, #169). v2.7.0 added the 1280–1327px layout correction (#198). They narrow the visible gap to the Sphinx sites but were not gate items; they are recorded here so the tracker does not understate what has shipped.

Parity validation study (2026-09-11)

The study inventoried quantecon-book-theme in eight independent sweeps (config surface, template, scripts, styles, user docs, changelog and tests, the fixtures corpus, and which options the lecture repos actually set across 29 repos), consolidated 100 features, assessed each against this theme with an adversarial reviewer and adjudication of disputes, and added 3 features a completeness critic found. Result: 28 implemented, 50 partial, 13 missing and untracked, 1 missing and tracked, 6 dropped by earlier decision, 5 not applicable. The headline claims were then re-checked against the live lecture-wasm site, the lecture-python-programming preview and the lecture sources; one was corrected and then re-corrected: both forms of the notebook header ({raw} jupyter and {raw} html) render as literal markup under the page title; the study's scale was essentially right (about 456 lecture files across the lecture repos; the #204 review later counted 549 header blocks across 18 lecture repos).

The gate items it produced, by phase. Membership and order are the sub-issue list's; this table carries only what each phase is for and when it is finished.

Phase	Intent	Exit criterion
Decisions	Settle the question a fix depends on	The licence-footer default is decided
Content correctness	Remove defects a reader sees on lecture pages at cutover	No literal notebook-header markup; Launch shows no dead links and can be switched off; plots centred; canonical emitted and the favicon resolves on sub-path sites; byline and translator links pass WCAG AA; landing-page listings match the Sphinx enumerators and part captions
Navigation	Match the Sphinx outline depth	h4 sections appear in "On this page" and expand under an active h3
Migration readiness	Make the theme configuration move cleanly	A migration checklist page names every step the Sphinx build did implicitly: maths macros, git-metadata history depth, notebook repo naming, the notebook header setting, the {raw} source rewrite, and language and analytics settings. Under #203 the licence footer needs no step
Close-out	Close on an accurate record	The docs describe what shipped (#217), and PLAN.md carries a closing note, with its lasting guidance moved to CONTRIBUTING.md (#218)

Gates. The close-out phase waits on the others: the docs review (#217) first, then the PLAN.md close-out (#218). #183 and #181 wait on the designer. Everything else can start now.

Sequencing

Phases 0–6 are done and in consumers' hands (v2.3.1 through v2.7.0; lecture-wasm and lecture-python-programming#363 pin v2.7.0). What remains is the study's gap list, worked in sub-issue order: decisions, content correctness, navigation, migration readiness, then the close-out: the docs review (#217) and the PLAN.md close-out (#218). When the list is empty: cut a release, bump both consumers, and the gate is green.

The cutover itself — repointing every lecture repo and running the migration — belongs to the lecture migration project and is deliberately not a sub-issue here; this gate is an input to its migration-route decision.

Migration dependencies outside the theme. The study found lecture-source constructs the QuantEcon mystmd fork does not yet handle: {nb-exec-table} (QuantEcon/mystmd#76), {only}, {glue:figure} and {youtube} render as error callouts until supported or rewritten, and the scroll-output cell tags are silently ignored, so long outputs render at full length. They are filed as QuantEcon/mystmd#104 ({only}), QuantEcon/mystmd#105 ({glue:figure}), QuantEcon/mystmd#106 (scroll-output tags) and QuantEcon/mystmd#107 (the {youtube} decision). They are filed in QuantEcon/mystmd without a parent: they gate the content migration (they surface when the lecture sources build on the fork), not this theme's parity, so they are referenced here rather than tracked.

Added by the #204 review:

• ipynb export has no project-level notebook header, so the QuantEcon logo is copied into 549 lecture source files as raw blocks mystmd#108: a settings.myst_to_ipynb.header project setting, so the notebook logo header comes from configuration instead of copies in lecture files.
• Three ipynb export gaps: ipynb export writes tables with colspan or rowspan as pipe tables, which cannot express merged cells, so the columns misalign mystmd#109 (span tables), ipynb export with markdown: commonmark writes {iframe} as MyST directive syntax, which Jupyter and Colab show as a literal code block mystmd#110 ({iframe} under markdown: commonmark) and ipynb export drops every raw node, including {raw} text and {raw} latex content, and logs an error that --strict does not count mystmd#111 (raw nodes).

All four are filed without a parent. The lecture-source side is coordinated in QuantEcon/workspace-lectures#64: the inventory, the rewrite rules and the one-PR-per-repo rollout at cutover. That issue also tracks the constructs its inventory found that no other issue covers: {prf:Theorem}, {any}, {todo} and {fas}.

Added by the #207 review: QuantEcon/mystmd#113, redirect pages written at build time for the old Sphinx page URLs and the existing redirect maps. Without it every inbound link to the lecture sites breaks, so unlike the other dependencies it blocks the cutover. Each repository's redirect map is recorded on QuantEcon/workspace-lectures#64.

Not in scope (and why)

• Feature: Add BinderHub support to Launch Notebook #26 (BinderHub), Feature: JupyterHub launch support — config-driven links (replacing the option removed in #87) #153 (JupyterHub launch links), Feature: connect in-page compute to a running Jupyter server (Thebe server: backend, incl. localhost) #154 (Thebe server/localhost compute) — demand-driven launch/compute enhancements beyond the book-theme feature set as the lecture repos actually configure it; none gates the cutover. Deliberately kept off the milestone. The study confirmed no lecture repo sets jupyterhub_url.
• React hydration fails on every page load (#418/#423), which makes the FOUC control test racy #126/Sidebar toggle icons still animate their first-paint correction (same mechanism as the contents panel) #127 (hydration, first-paint animations) — closed. Sidebar toggle icons still animate their first-paint correction (same mechanism as the contents panel) #127 was closed by refactor: rebuild the contents drawer on the Popover API #144's rebuild; React hydration fails on every page load (#418/#423), which makes the FOUC control test racy #126 was re-measured on 2026-09-07 and closed as not a defect. Migrate off Remix v1 — target React Router 7, upstream-first (tracks jupyter-book/myst-theme) #28 (React Router 7) remains upstream-gated and unrelated to parity.
• Continuous line numbering for prf:algorithm nested blocks (algorithm2e parity) via CSS counters #119 (algorithm2e line numbering) — a content-rendering enhancement beyond the book-theme feature set.
• Lecture-level live compute: per-lecture WASM/JupyterLite enablement (page metadata + compatibility testing) #114 (per-lecture live compute) — removed from the gate 2026-09-10. The book theme's Thebe support is one project-wide flag and the Sphinx lecture sites set thebe: false; built as PR feat(compute): per-lecture live compute via an enable_live_compute site option #195 on its own merits.
• Review: live-compute (JupyterLite) toggle — placement and integration #128 (live-compute toggle placement) — removed from the gate 2026-09-10 and moved to the design-review project Design review of the v2.5.0 lecture build #187.
• The remaining design-review decisions in Design review of the v2.5.0 lecture build #187 (Whether content links keep a resting underline or underline only on hover #176–Adopt the Sphinx theme's unused boxed inline literal, or keep the bare literal both builds render #180, Non-executable code blocks get no frame, and every code block renders 13px against 18px prose #184, Admonition, exercise and proof bodies render at the full 18px content size, with no step down from the surrounding copy #185, Light-mode site footer falls short of WCAG AA once the 70% opacity is composited #201) stay there. The page-header byline renders in a sky blue nothing else uses and fails AA contrast #183 (byline and translator link contrast) and The in-content table of contents drops the period after every enumerator, and renders part titles as bare list items #181 (landing-page listing) moved from Design review of the v2.5.0 lecture build #187 to this gate on 2026-09-11, by maintainer decision, because the study found both differ from the live Sphinx sites; DrDrij remains their designer.
• Study findings filed without a parent because they do not block the cutover (A reader's light or dark choice is overwritten when the operating system switches appearance #211, Downloads toolbar button opens an empty menu on pages with no downloads #212, The hide_toc and hide_search options are declared and documented but have no effect #213, The contents drawer loads closed on every page, with no counterpart to the book theme's persistent_sidebar #214, Readers cannot copy a section link from the "On this page" outline, which Sphinx lecture sites allow #215, and from the review HTML notebook outputs such as pandas DataFrames are missing from the served page and appear only after JavaScript runs #216, The open contents drawer covers the start of the lecture text between 1024px and 1339px wide #219 and The font-size setting resets on every page load and is not offered on mobile #220): A reader's light or dark choice is overwritten when the operating system switches appearance; Downloads toolbar button opens an empty menu on pages with no downloads; The hide_toc and hide_search options are declared and documented but have no effect; The contents drawer loads closed on every page, with no counterpart to the book theme's persistent_sidebar; Readers cannot copy a section link from the "On this page" outline, which Sphinx lecture sites allow.
• The study's remaining partial rows (cosmetic and behavioural differences such as font-size persistence, the dark-mode palette and the header site title) are recorded in the study report and are not gate items.
• Accessibility of TOC when reading a lecture #129, Visual snapshots: 1% diff budget hides structural changes (sidebar TOC drifted stale on main) #113, Track KaTeX upgrades: own the stylesheet href, bump to 0.16.x now, 0.18.x in lockstep with QuantEcon/mystmd #118, Contents sidebar carries two unprefixed width utilities, so its base width depends on Tailwind's emission order #130 — design-review, QA-infrastructure and dependency-tracking items with their own carriers. Retire the in-repo git-metadata plugin once the standalone release exists in quantecon-plugins.mystmd #157, PLAN.md, #147 and the README still record this repo as git-metadata's permanent home; D8 reversed that #159, Stale and mis-qualified issue references in #143 and #28 #160 — plugin retirement and living-document hygiene.

Tracker hygiene

✅ Applied. The gate items from the study are on the book-theme parity milestone. The milestone and this tracking issue describe the same gate — the milestone for label-level filtering, this issue for native sub-issue progress and sequencing.

[mmcky]

Phase 0 is code-complete — status refresh (2026-08-26)

The gate has moved since the 2026-08-24 snapshot this issue was filed against. Both Phase 0 defects are fixed on main, and the issue body has been updated to match.

What landed: #137 (closes #117 — code cells nested in a directive now register with the kernel, so their run buttons execute), #139 (closes #138 — built stylesheet asset URLs are relative, so all 60 KaTeX font references resolve in myst build --html and under a baseurl), #140 (self-hosts the Source Sans 3 webfont, removing the last third-party origin from the critical path) and #148 (refreshes the linux mobile-chrome baselines after #140's glyph-metric drift).

What is left in Phase 0: cutting the release. Both fixes are on main but unreleased, so lecture-wasm is still pinned to a v2.3.0 that ships dead run buttons and system-glyph maths. PR #149 prepares v2.3.1 — a patch, since it is two bug fixes plus an asset-hosting optimization with no new configuration surface. After it merges the tag needs pushing (git tag v2.3.1 && git push origin v2.3.1) to trigger release.yml.

Tracker hygiene from the original proposal is done: #114, #128 and #138 are on the milestone and #26 has been removed, so the milestone and this issue now describe the same gate.

Remaining gate: #89 (Phase 3), #90 → #91 (Phases 4–5), #92 (Phase 6), #114, #128, #87.

Suggested order. #87 is the cheapest item on the gate — a maintainer call with no implementation and no dependants — so settling it first shrinks the board for free. #89 is the next substantial build item and can start immediately against the existing visual-preview harness. #114 and #128 can run in parallel with it, since #114's implementation channels are already scoped and #128 is a design conversation.

Also worth noting for accuracy: PLAN.md's open-question 4 still named #117 as an open defect gating the cutover. That paragraph is corrected in PR #149 to point at the live gate items and at this issue.

[mmcky]

Session update — Phase 0 released, #87 resolved (2026-08-26)

The gate has moved twice since the last update. The body above is current.

v2.3.1 is released. #149 merged and the tag was pushed, so release.yml published v2.3.1 with quantecon-theme.zip. Both defects that made the shipped v2.3.0 surface incorrect — dead run buttons in directives (#117) and 404'd KaTeX fonts in static builds (#138) — are now in consumers' hands. Phase 0 of this gate is complete.

Verifying the published artifact found one more defect. Unzipping the release showed #139 had fixed only the stylesheets in _assets/; Remix also emits route and shared-chunk CSS elsewhere in the build, and four of those files shipped still carrying absolute /myst_assets_folder/… URLs. Filed as #150 and fixed in #151, now on main. Impact is a missing plotly icon on statically built sites, so it did not warrant re-tagging — it rides the next release.

#87 is resolved: remove. The demand signal came back empty (no lecture repo has ever set jupyterhub_url in the Sphinx theme, so that button never rendered on a live site), and probing a live build showed the MyST version hardcoded a /jupyter/hub/ path segment and discarded the reader's typed URL whenever the popover closed. PR #152 removes it and collapses the launcher to a direct Colab link; it is green and awaiting review. Once it merges, this gate contains only unstarted build work.

The intent behind #87 is preserved, not dropped — three future launch/compute targets are now tracked with background research: #153 (JupyterHub links, config-driven), #26 (BinderHub, refreshed — it is really two capabilities and the in-page half needs no URL builder), and #154 (connecting in-page compute to a running Jupyter server, including localhost). All three are demand-driven and deliberately excluded from this milestone, since none gates the cutover.

Remaining gate: #89, #90 → #91, #92, #114, #128. #89 is the recommended next build item, and its scope is now materially smaller than PLAN.md implies — no lecture repo sets qetheme_code_style or color_scheme, so real parity is the seoul256 and QE-token defaults rather than the configuration surface. The concrete values and the one genuinely open design question (prism cannot be a literal Pygments port) are in the scoping comment.

[mmcky]

Revision log — 2026-08-26 13:15 AEST (session close)

Body re-stamped and given an explicit resume pointer. No state changed since the previous comment 20 minutes ago — this revision is about making the artifact resumable, not about new facts.

Two things fixed in the body itself:

The stamp was a bare date. This body was revised three times today, every one of them reading 2026-08-26, so the stamps could not be ordered against each other or against the comment thread — precisely the date-rollover failure the plan convention warns about. It now carries time and timezone.

There was no explicit front of the plan. The sequencing section said what could be worked in parallel, which is not the same as saying what to pick up first. A new "Next session — resume here" section names #89, says why it leads, and carries the two things a fresh context would otherwise have to rediscover: that its scope is much smaller than PLAN.md implies (real parity is the seoul256 and QE-token defaults, not the switch surface), and that the one genuinely open question is whether the QE tokens can be expressed as a prism theme — which should be tested on the preview harness before an approach is chosen.

Genre recorded. This issue is a long-lived tracker, not a period plan: it is resumed and updated but never session-closed. Stated in the body so a future session does not close it on completing a phase.

Note for whoever resumes. The live-state facts were measured at the stamp above. Three PRs were open, green and mergeable at that moment — #152 (removes the private JupyterHub option, closes #87), #144 (contents drawer on the Popover API) and #141 (first-paint transitions) — and may well have merged since. Re-verify before working the plan; the gap since the stamp is unaudited.

[mmcky]

Scope change, 2026-09-04: translator attribution is on the gate

Taken as part of the theme-estate review on QuantEcon/workspace-themes#1. #143 moves from Not in scope into the work plan.

The rationale it was excluded on had already expired when it was written. The Not in scope entry read "#143 (author/translator attribution) — content-rendering and attribution enhancements beyond the book-theme feature set". quantecon-book-theme v0.22.0 shipped the translators option on 2026-08-25 (QuantEcon/quantecon-book-theme#426, __init__.py:722); this issue was revised on 2026-08-26. Translator credit has been inside the book-theme feature set for the entire life of the exclusion.

It is live on a translated edition today. lecture-python-programming.ml pins quantecon-book-theme==0.22.0 (environment.yml:14) and configures a translator (lectures/_config.yml:62); the credit renders on its content pages. Verified against the deployed site on 2026-09-04.

And this issue already treats the translated series as in scope. The work plan keeps #90 and #91 on the gate "required for the translated series", and the sequencing note says Phase 2 "matters specifically for the translated lecture series, which the all-or-nothing scope includes". Cutting over as previously scoped would move .ml from a theme that credits its translator to one that cannot, with no step anywhere recording the loss.

The exclusion of #119 is unchanged and stands on its own merits.

Scope of the added item

Only the translators half. The theme has rendered per-page authors since d1ac95aa (2025-02-14) — app/components/PageContent.tsx:52-55 passes authors into app/components/ProjectFrontmatter.tsx, which renders them as linked Author components — and PLAN.md:57 already lists "author header" under Already at parity, ticked on both stacks. #143 is being rescoped accordingly; its opening claim that the theme "has no per-page author attribution" is wrong.

What remains is the translators block, the per-page override, and a localisable "Translators" label. The label mechanism overlaps Phase 4 (#90), which is on the gate for the same translated-series reason — worth building once rather than twice.

The placement question stays genuinely open: the two themes' header structures are inverses, so matching book-theme's reading is a design decision, not a port. Note QuantEcon/quantecon-book-theme#432 while designing it — on the Sphinx side the credit is rendered into the landing page and then hidden by display: none, and page-header.js recovers only the authors, so the front door of a translated edition currently shows no translator at all. Worth not reproducing that shape here.