feat(website): browse Nightly, RC and Stable release tabs #638

Open

bibiburl wants to merge 5 commits from bibiburl/Longshot:docs/release-summary into main

pull from: bibiburl/Longshot:docs/release-summary

merge into: MagicBeanDev:main

MagicBeanDev:main

Conversation 14 Commits 5 Files changed 16 +1191 -3

bibiburl commented 2026-11-15 01:55:10 +00:00

Member

• Copy link
• Copy Markdown

Type: feat · Risk: low · Tested on: Linux, Chromium 153, bounded live collection, fixture and game-free tests · Needs: renewed review; existing publication blockers resolved before deployment

Why

Players can switch between Nightly, Release candidate and Stable to see the latest published game build, relevant changes and release notes. Release tabs update after publication and on the hourly fallback. This revision addresses Jordan’s review.

What changed

Replayed the release-tab commits directly onto main aa95d35f; this PR no longer includes or depends on #634’s hero, or #611’s download changes. The tabs compose with #611 and Jordan’s hero #768.

• Bound each source summary to the newest 50 commits, with the full change count and Compare source link. Forgejo’s paginated commits endpoint excludes the baseline’s ancestry using not; no unbounded compare request is made. The RC panel currently displays 50 of 109 changes, explicitly labelled.
• A failed channel shows an unavailable panel while other channels remain usable. A complete release collection outage produces a valid unavailable snapshot instead of stopping website publication. Malformed output is still rejected; unrelated publication guards remain in place.
• Without an earlier stable, RC compares with the previous RC and Nightly with the previous Nightly. Nightly pruning cannot change an RC baseline. Tag resolution and baseline-ancestry validation remain required.
• Highlight selection prefers features, fixes and performance changes. Exact-SHA editorial keys are checked against real repository commits; unknown or out-of-window keys produce explicit warnings. Retired the three inactive entries. Remaining branded immutable subjects get neutral display text with their source link preserved.
• Refresh incidents must match the authenticated Actions user ID and the marker. A forged marker cannot suppress failures or receive recovery edits; an unverifiable identity causes no writes.
• Disable Website’s repeated failure emails in favor of one incident per outage. Unrelated label events use their own concurrency group, preserving queued previews. Main-only deployment and the fork-preview approval gate are unchanged. Release’s existing failure-email setting remains.
• Tab selections update the URL hash; deep links and keyboard navigation work. Removed duplicate mobile CSS.

Screenshots

After: this PR @ 0ebd90c7, Release candidate at 1440 px, light. The current RC shows the newest 50 of 109 source changes.

Before, mobile Nightly, and empty Stable channel

Before: main @ aa95d35f, 1440 px, light; no release-tab section.

After: this PR @ 0ebd90c7, Nightly at 390 px, light.

After: this PR @ 0ebd90c7, Stable at 1440 px, light; explicitly unavailable until published.

Integration with #611, #709 and Jordan’s #768 hero (local only)

Browser checks: clicking RC updates the hash to #release-rc, deep links restore the panel, and keyboard navigation remains usable. Standalone homepage accessibility audits at 1440/390 report 0 violations, with one incomplete contrast rule for gradient backgrounds/decorative glyphs; screenshots were inspected manually. The old main hero has an inherited halo overflow, which the #768 combination removes; the release panels fit.

Measurements

Verdict: summary collection stays bounded as release history grows, failures degrade channel panels, and incident ownership is verified. These are website/source-change descriptions, not gameplay performance claims.

Hardware: Linux x86_64, Chromium 153 and the pinned Node/npm environment; no GPU or gameplay benchmark.
Scene: built website on loopback, desktop 1440 px and mobile 390 px, light and dark themes.
Method: functional test counts on the named untouched baseline and PR branch; one live release-collection observation where listed, not an interleaved performance measurement.
How to reproduce: in website/, run npm ci, npm run build, npm run check and node --test tests/*.test.ts; at the repository root run python -I -B tests/run_all.py. Use local Astro preview for static behavior and Wrangler for download routes.

Check	Before: main @ aa95d35f	After: this PR @ 0ebd90c7
Website tests	75 pass, 1 existing failure	111 pass, same 1 failure
Game-free suite	509 pass, 14 fail, 70 skip	510 pass, same 14 fail, 70 skip
Live release collection	No release-tab collector	15 API calls, 2.699 seconds in one observation; Nightly 1/1, RC 50/109, Stable empty
Large-range fixture	No release-tab collector	100,000 counted changes, only 50 commit records displayed; two bounded range requests

The live observation is a functional check, not a latency guarantee or repeated benchmark. Category counts cover only the displayed commits; the full range is linked.

Flags and settings

Only Website’s failure-email behavior changes; there are no new game switches.

Flag	What it is	Before this PR	After this PR	Effect	Helps / doesn't help / untested	Recommendation
Website enable-email-notifications	Forgejo workflow failure emails	true	false	One issue records an outage instead of repeat hourly emails	Incident ownership/lifecycle tested with fake APIs; real Actions rollout pending	Disabled; use the authenticated incident and Actions log

Testing

Tested: pinned dependencies; static site build (16 pages); website tests; publication checks; complete game-free suite; six refresh-workflow tests, including execution of the real dispatch shell command and actual concurrency expression with fake contexts; incident spoofing/identity fixtures; source-range cap, partial outage, pruning, branding and tab hash regressions; git diff --check. Browser evidence accompanies the screenshots. #611, #638, #709 and #768 merge without conflict in a temporary local combination.

The website failure is the same HDR Native condition assertion as untouched main. Publication checks still reject existing progress-feed branding in #603, #615 and #715; #731 handles that separately. Release-summary validation passes. The same 14 failing files reproduce on untouched main aa95d35f: tests/test_appimage_player_build.py, tests/test_bake_portability.py, tests/test_cpm_discovery.py, tests/test_crash_dialog.py, tests/test_env_on.py, tests/test_lift_cleanup_error.py, tests/test_linux_preflight.py, tests/test_make_game_cache.py, tests/test_mods_size_table.py, tests/test_options_budget_once.py, tests/test_pipeline_tour.py, tests/test_readback_late.py, tests/test_vk_pgo.py, tests/test_walk_key.py. No new failing files.

Combined integration: #611 + #638 + #709 + #768 merges cleanly, builds 18 pages and runs 125 website tests successfully with the same existing HDR-condition failure. Sixteen checks (home, Install, Transfer a save and Launcher × desktop/mobile × light/dark) have no horizontal overflow. A separate #768 mobile-light tagline contrast finding remains: #ffe100 on #146cea, 20 px normal text, 3.67:1, below 4.5:1. This PR does not modify that hero; resolve it in the hero review before deployment.

Not tested: real Actions job-token identity, runner dispatch/queue execution, email delivery, Cloudflare production deployment, Firefox/Safari or game execution. Local workflow tests use fake endpoints; they do not claim a deployed workflow.

Must be tested:

• Forgejo Actions and Cloudflare production, by hand after approved rollout: publish a test-approved release, inspect the main refresh, exercise one controlled refresh failure and recovery, expecting a single incident owned by the authenticated Actions account, an updated timestamp, and closure only after successful deployment. Resolve existing publication blockers before deployment.

Risks

The latest 50 commits are deliberately a bounded window, not the entire release history; the total and Compare link make that limit visible. An unavailable panel may temporarily replace known release details after an upstream outage. Curated highlights need editorial maintenance and only appear in a displayed range. Incidents require a readable authenticated /user identity; failure to verify it is reported without writing to arbitrary issues. Revert the release-tab and refresh-workflow changes to disable them.

Compatibility: no game files, save formats, launcher behavior, renderer ABI or dependencies change. Main remains the only production source. No dependency on #634; the tabs fit #768 and the download callout from #611.

Issues

See #635, #580, #611, #634, #768, #731 and the linked review above.

Checklist

Everyone

• One topic.
• The title is a Conventional Commit (type(scope): description) of the status line's type.
• Why gives the reason plainly and links the issues that fit (the tracker searched).
• Measurements has a verdict, or one line saying why there is nothing to measure.
• Testing names what was tested, what wasn't, and what must still be tested, where and how.
• Rebased on current main.
• python -I -B tests/run_all.py passes (any failure that also fails on main is named).
• No game files, dumps, build output or generated files.
• third_party/ untouched (a new folder with a PROVENANCE.md instead).
• The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled.

When it applies

• AI assistance recorded with Assisted-by: trailers.
• A check under tests/ that reads changed source updated in this PR, not weakened.
• Before/after screenshots attached with named commits and viewport/theme information.
• Docs updated.

**Type:** feat · **Risk:** low · **Tested on:** Linux, Chromium 153, bounded live collection, fixture and game-free tests · **Needs:** renewed review; existing publication blockers resolved before deployment ## Why Players can switch between **Nightly, Release candidate and Stable** to see the latest published game build, relevant changes and release notes. Release tabs update after publication and on the hourly fallback. This revision addresses [Jordan’s review](https://git.magicbean.dev/MagicBeanDev/Longshot/pulls/638#issuecomment-24095). ## What changed Replayed the release-tab commits directly onto main `aa95d35f`; this PR no longer includes or depends on #634’s hero, or #611’s download changes. The tabs compose with #611 and Jordan’s hero #768. - Bound each source summary to the newest **50 commits**, with the **full change count** and Compare source link. Forgejo’s paginated commits endpoint excludes the baseline’s ancestry using `not`; no unbounded compare request is made. The RC panel currently displays 50 of 109 changes, explicitly labelled. - A failed channel shows an unavailable panel while other channels remain usable. A complete release collection outage produces a valid unavailable snapshot instead of stopping website publication. Malformed output is still rejected; unrelated publication guards remain in place. - Without an earlier stable, RC compares with the previous RC and Nightly with the previous Nightly. Nightly pruning cannot change an RC baseline. Tag resolution and baseline-ancestry validation remain required. - Highlight selection prefers features, fixes and performance changes. Exact-SHA editorial keys are checked against real repository commits; unknown or out-of-window keys produce explicit warnings. Retired the three inactive entries. Remaining branded immutable subjects get neutral display text with their source link preserved. - Refresh incidents must match the authenticated Actions user ID **and** the marker. A forged marker cannot suppress failures or receive recovery edits; an unverifiable identity causes no writes. - Disable Website’s repeated failure emails in favor of one incident per outage. Unrelated label events use their own concurrency group, preserving queued previews. Main-only deployment and the fork-preview approval gate are unchanged. Release’s existing failure-email setting remains. - Tab selections update the URL hash; deep links and keyboard navigation work. Removed duplicate mobile CSS. ## Screenshots After: this PR @ `0ebd90c7`, Release candidate at 1440 px, light. The current RC shows the newest 50 of 109 source changes.  <details><summary>Before, mobile Nightly, and empty Stable channel</summary> Before: main @ `aa95d35f`, 1440 px, light; no release-tab section.  After: this PR @ `0ebd90c7`, Nightly at 390 px, light.  After: this PR @ `0ebd90c7`, Stable at 1440 px, light; explicitly unavailable until published.  </details> <details><summary>Integration with #611, #709 and Jordan’s #768 hero (local only)</summary>   </details> Browser checks: clicking RC updates the hash to `#release-rc`, deep links restore the panel, and keyboard navigation remains usable. Standalone homepage accessibility audits at 1440/390 report **0 violations**, with one incomplete contrast rule for gradient backgrounds/decorative glyphs; screenshots were inspected manually. The old main hero has an inherited halo overflow, which the #768 combination removes; the release panels fit. ## Measurements **Verdict:** summary collection stays bounded as release history grows, failures degrade channel panels, and incident ownership is verified. These are website/source-change descriptions, not gameplay performance claims. Hardware: Linux x86_64, Chromium 153 and the pinned Node/npm environment; no GPU or gameplay benchmark. Scene: built website on loopback, desktop 1440 px and mobile 390 px, light and dark themes. Method: functional test counts on the named untouched baseline and PR branch; one live release-collection observation where listed, not an interleaved performance measurement. How to reproduce: in website/, run npm ci, npm run build, npm run check and node --test tests/*.test.ts; at the repository root run python -I -B tests/run_all.py. Use local Astro preview for static behavior and Wrangler for download routes. | Check | Before: main @ aa95d35f | After: this PR @ 0ebd90c7 | |---|---|---| | Website tests | 75 pass, 1 existing failure | 111 pass, same 1 failure | | Game-free suite | 509 pass, 14 fail, 70 skip | 510 pass, same 14 fail, 70 skip | | Live release collection | No release-tab collector | 15 API calls, 2.699 seconds in one observation; Nightly 1/1, RC 50/109, Stable empty | | Large-range fixture | No release-tab collector | 100,000 counted changes, only 50 commit records displayed; two bounded range requests | The live observation is a functional check, not a latency guarantee or repeated benchmark. Category counts cover only the displayed commits; the full range is linked. ### Flags and settings Only Website’s failure-email behavior changes; there are no new game switches. | Flag | What it is | Before this PR | After this PR | Effect | Helps / doesn't help / untested | Recommendation | |---|---|---|---|---|---|---| | Website `enable-email-notifications` | Forgejo workflow failure emails | true | false | One issue records an outage instead of repeat hourly emails | Incident ownership/lifecycle tested with fake APIs; real Actions rollout pending | Disabled; use the authenticated incident and Actions log | ## Testing **Tested:** pinned dependencies; static site build (16 pages); website tests; publication checks; complete game-free suite; six refresh-workflow tests, including execution of the real dispatch shell command and actual concurrency expression with fake contexts; incident spoofing/identity fixtures; source-range cap, partial outage, pruning, branding and tab hash regressions; `git diff --check`. Browser evidence accompanies the screenshots. #611, #638, #709 and #768 merge without conflict in a temporary local combination. The website failure is the same HDR Native condition assertion as untouched main. Publication checks still reject existing progress-feed branding in #603, #615 and #715; #731 handles that separately. Release-summary validation passes. The same 14 failing files reproduce on untouched main `aa95d35f`: `tests/test_appimage_player_build.py`, `tests/test_bake_portability.py`, `tests/test_cpm_discovery.py`, `tests/test_crash_dialog.py`, `tests/test_env_on.py`, `tests/test_lift_cleanup_error.py`, `tests/test_linux_preflight.py`, `tests/test_make_game_cache.py`, `tests/test_mods_size_table.py`, `tests/test_options_budget_once.py`, `tests/test_pipeline_tour.py`, `tests/test_readback_late.py`, `tests/test_vk_pgo.py`, `tests/test_walk_key.py`. No new failing files. **Combined integration:** #611 + #638 + #709 + #768 merges cleanly, builds 18 pages and runs 125 website tests successfully with the same existing HDR-condition failure. Sixteen checks (home, Install, Transfer a save and Launcher × desktop/mobile × light/dark) have no horizontal overflow. A separate #768 mobile-light tagline contrast finding remains: `#ffe100` on `#146cea`, 20 px normal text, **3.67:1**, below 4.5:1. This PR does not modify that hero; resolve it in the hero review before deployment. **Not tested:** real Actions job-token identity, runner dispatch/queue execution, email delivery, Cloudflare production deployment, Firefox/Safari or game execution. Local workflow tests use fake endpoints; they do not claim a deployed workflow. **Must be tested:** - Forgejo Actions and Cloudflare production, by hand after approved rollout: publish a test-approved release, inspect the main refresh, exercise one controlled refresh failure and recovery, expecting a single incident owned by the authenticated Actions account, an updated timestamp, and closure only after successful deployment. Resolve existing publication blockers before deployment. ## Risks The latest 50 commits are deliberately a bounded window, not the entire release history; the total and Compare link make that limit visible. An unavailable panel may temporarily replace known release details after an upstream outage. Curated highlights need editorial maintenance and only appear in a displayed range. Incidents require a readable authenticated `/user` identity; failure to verify it is reported without writing to arbitrary issues. Revert the release-tab and refresh-workflow changes to disable them. **Compatibility:** no game files, save formats, launcher behavior, renderer ABI or dependencies change. Main remains the only production source. No dependency on #634; the tabs fit #768 and the download callout from #611. ## Issues See #635, #580, #611, #634, #768, #731 and the linked review above. ## Checklist ### Everyone - [x] One topic. - [x] The title is a Conventional Commit (`type(scope): description`) of the status line's type. - [x] Why gives the reason plainly and links the issues that fit (the tracker searched). - [x] Measurements has a verdict, or one line saying why there is nothing to measure. - [x] Testing names what was tested, what wasn't, and what must still be tested, where and how. - [x] Rebased on current `main`. - [x] `python -I -B tests/run_all.py` passes (any failure that also fails on `main` is named). - [x] No game files, dumps, build output or generated files. - [x] `third_party/` untouched (a new folder with a `PROVENANCE.md` instead). - [x] The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled. ### When it applies - [x] AI assistance recorded with `Assisted-by:` trailers. - [x] A check under `tests/` that reads changed source updated in this PR, not weakened. - [x] Before/after screenshots attached with named commits and viewport/theme information. - [x] Docs updated.

bibiburl added 24 commits 2026-11-15 01:55:10 +00:00

docs(guides): add shared launcher screenshot and provenance ... b6694a6385

Assisted-by: Codex (OpenAI)

docs(saves): illustrate launcher transfer controls ... 58ad729b91

Assisted-by: Codex (OpenAI)

docs(guides): flag UI drift and document screenshot ownership ... 68fc4224cb

Assisted-by: Codex (OpenAI)

test(guides): require both menu capture source fingerprints ... d92da61e89

Assisted-by: Codex (OpenAI)

docs(guides): pin the planned launcher review job to its release commit ... 08383c6c9a

Assisted-by: Codex (OpenAI)

docs(guides): capture the real settings menu without game assets ... 861fd7ab39

Render maintained menu code with bundled fonts and deterministic UI-only state; record capture and reviewed sources separately. Share the image with the settings guide and validate site paths containing spaces.

Assisted-by: Codex (OpenAI)

fix(guides): distinguish complete and skipped source reviews ... ea3fab1e26

Assisted-by: Codex (OpenAI)

docs(guides): track launcher fonts and save behavior dependencies ... 71997407f3

Assisted-by: Codex (OpenAI)

docs(guide): illustrate Launcher decisions beside their steps ... 24a92c0fa5

Add real-view captures for Manage saves, game file selection and import confirmation with isolated synthetic state. Keep capture dependencies and source review in the screenshot manifest, and verify image links and transfer-step placement.

Assisted-by: Codex (OpenAI)

fix(docs): preserve Launcher capture build policy ... f7f4ca5741

Stage and fingerprint the Launcher statistics opt-out and explicit NuGet configuration. Recapture neutral example paths, guard fixture ownership, document contributor review, restore Frizti, and retain main's complete HDR wording. Review the settings image against rebased menu sources.

Assisted-by: Codex (OpenAI)

docs(settings): refresh the development menu example ... f035bbd54f

Recapture the real menu after main changed the output-resolution label to reduced. Record current capture provenance and label the guide image as an illustrative development build.

Assisted-by: Codex (OpenAI)

docs(guides): simplify instructions and align save transfers ... d7c62a39d4

Shorten launcher guide copy, make screenshots the full-size links, and align transfer steps and captions with the article. Repair the stale Audio output restart test without weakening its assertion.

Assisted-by: Codex (OpenAI)

docs(guides): review screenshots against current main and simplify mobile navigation ... b905befe78

Integrate main cf605c85 and record why all nine changed menu dependencies leave the retained Graphics capture accurate. Preserve capture provenance. Keep one mobile hamburger per page and remove the duplicate Menu row.

Assisted-by: Codex (OpenAI)

fix(website): unify mobile hamburger styling ... 31a18758e3

Use Starlight icons and shared button styles for home and guide navigation, including light/dark and expanded states.

Assisted-by: Codex (OpenAI)

docs(install): restore Linux launcher permission step ... b792d80516

Restore the AppImage executable-permission instruction before Build and link the platform setup guides.

Assisted-by: OpenAI Codex

docs(guides): refresh captures from current menu and launcher ... e4627be1ee

Integrate main b2911d33, regenerate seven UI-only screenshots, record capture provenance and refresh version-specific guidance. Support registered menu text fields in the website catalogue and neutralize the new item name reported in issue 584.

Assisted-by: OpenAI Codex

docs(guides): reconcile current main and portable save backups ... 397551f680

Integrate main e9bed614. Record the Graphics source review and new threaded-rendering dependency while retaining original capture provenance. Update the restart-list regression and distinguish installed and portable Windows backup locations.

Assisted-by: GPT-6 Astra

Co-Authored-By: GPT-6 Astra <[email protected]>

docs(guides): integrate final main snapshot for review ... a0a93f11fe

Integrate main 2b1d38c6 and record that all 59 menu inputs and the capture helpers still match the reviewed sources. Keep original image provenance unchanged.

Assisted-by: GPT-6 Astra

Co-Authored-By: GPT-6 Astra <[email protected]>

fix(website): improve mobile navigation and launcher downloads ... 00a62f062f

Add responsive artwork, clearer release metadata and a shared download shell. Keep publication and source guards intact.

Assisted-by: OpenAI Codex

Co-Authored-By: OpenAI Codex <[email protected]>

docs(website): include guide additions from PR 264 ... 93f3c0e093

Merge bibiburl/docs/guide-screenshots at a0a93f11, preserving its seven original UI images, provenance, source guards, save instructions and matching navigation controls.

Assisted-by: OpenAI Codex

Co-Authored-By: OpenAI Codex <[email protected]>

fix(website): refresh progress filters after native reset ... 18e1efcd1a

Handle change-only selection events and defer reset rendering until the browser has restored the form values.

Assisted-by: OpenAI Codex

Co-Authored-By: OpenAI Codex <[email protected]>

fix(website): make guide tables and code keyboard accessible ... 02f0299c73

Name the guide table columns and keep horizontal scroll surfaces keyboard reachable without client JavaScript. Preserve table semantics and show focus clearly.

Assisted-by: Codex (OpenAI)

docs(website): add a theme-aware trails hero ... a826af4c0e

Introduce the agreed full-width Longshot hero with original SVG trails, shared primary button styling and light/dark palettes. Preserve the prerequisite launcher guidance, responsive branding and accessible fairy interaction.

Assisted-by: Codex (OpenAI)

feat(website): show latest build and release changes ... 6f72040e19

Add a theme-aware release summary with exact published commit ranges, reviewed highlights and a complete source-change disclosure. Keep launcher versions separate and reject incomplete release metadata before publication.

Assisted-by: Codex (OpenAI)

bibiburl added a new dependency 2026-11-15 01:55:43 +00:00

#264 docs(guides): illustrate installation, settings and save transfer

bibiburl added a new dependency 2026-11-15 01:55:44 +00:00

#611 fix(website): improve mobile browsing and launcher downloads

bibiburl added a new dependency 2026-11-15 01:55:44 +00:00

#634 docs(website): add a theme-aware trails hero

bibiburl added the

status

blocked

label 2026-11-15 01:55:44 +00:00

bibiburl commented 2026-11-15 01:55:45 +00:00

Author

Member

• Copy link
• Copy Markdown

Superseded screenshots — earlier version without tabs. The current Nightly / Release candidate / Stable design is shown directly in the main PR description. These images are retained only as historical comparison.

First attempt for #635 — actual website output, not a mockup. After: this PR @ 6f72040e, built with the pinned dependencies and served by the local Wrangler Worker. Before: #634 @ a826af4c. Desktop captures use a 1440px viewport; mobile uses 390px. The hero's existing punchline varies on reload.

The new section shows Nightly 170, its publication date, three reviewed highlights, and all six source commits since Nightly 164. No stable release exists at capture time. Once one does, prereleases compare with the latest stable; a new stable compares with the previous stable. Launcher versions stay separate.

After: this PR @ 6f72040e — dark section:

After: this PR @ 6f72040e — light section:

Full-resolution homepage comparisons:

• Before: #634 @ a826af4c — dark
• After: this PR @ 6f72040e — dark
• Before: #634 @ a826af4c — light
• After: this PR @ 6f72040e — light

Additional after views:

• Dark mobile, 390px
• Light mobile, 390px
• All six source changes expanded with the keyboard

Validation: 107/107 website tests pass; 16 pages build; no horizontal page overflow at 320–1920px in either theme; native disclosure works with Enter and visible focus. Section-scoped axe reports 0 violations, 22 passes and one incomplete gradient-contrast category per theme. The affected palette's gradient endpoints were checked separately (minimum 6.15:1 text contrast). No browser errors. These checks are not a full accessibility or real-device certification.

The inherited publication guard still rejects public wording from progress issue #615. The repository game-free suite reports 500 passed, 13 known baseline failures and 69 skipped; details are in the PR description. No game or runtime experiment was performed.

Still WIP, with native dependencies and status/blocked: #264 → #611 → #634 → #638. Do not approve/merge out of order. No hosted preview or deployment was requested.

> **Superseded screenshots — earlier version without tabs.** The current Nightly / Release candidate / Stable design is shown directly in [the main PR description](https://git.magicbean.dev/MagicBeanDev/Longshot/pulls/638). These images are retained only as historical comparison. First attempt for #635 — actual website output, not a mockup. **After: this PR @ 6f72040e**, built with the pinned dependencies and served by the local Wrangler Worker. **Before: #634 @ a826af4c**. Desktop captures use a 1440px viewport; mobile uses 390px. The hero's existing punchline varies on reload. The new section shows **Nightly 170**, its publication date, three reviewed highlights, and all **six source commits since Nightly 164**. No stable release exists at capture time. Once one does, prereleases compare with the latest stable; a new stable compares with the previous stable. Launcher versions stay separate. After: this PR @ 6f72040e — dark section:  After: this PR @ 6f72040e — light section:  Full-resolution homepage comparisons: - [Before: #634 @ a826af4c — dark](https://git.magicbean.dev/attachments/083700bc-3d33-4f56-bacf-10a94c2c0181) - [After: this PR @ 6f72040e — dark](https://git.magicbean.dev/attachments/9a3394fa-cd0f-445f-bf4d-ca72a3179ea8) - [Before: #634 @ a826af4c — light](https://git.magicbean.dev/attachments/31a4bec6-2e29-4feb-a8cf-d9508c78800c) - [After: this PR @ 6f72040e — light](https://git.magicbean.dev/attachments/6cc86438-1d23-46ba-b7c9-f74fe5cd697f) Additional after views: - [Dark mobile, 390px](https://git.magicbean.dev/attachments/0481cdf8-3bc7-4964-aae6-b028144d4bf0) - [Light mobile, 390px](https://git.magicbean.dev/attachments/2b76756b-68df-4929-b4ca-7e2acaa2a140) - [All six source changes expanded with the keyboard](https://git.magicbean.dev/attachments/67db90a8-727f-4ce9-bf97-00fb8273b42f) Validation: 107/107 website tests pass; 16 pages build; no horizontal page overflow at 320–1920px in either theme; native disclosure works with Enter and visible focus. Section-scoped axe reports 0 violations, 22 passes and one incomplete gradient-contrast category per theme. The affected palette's gradient endpoints were checked separately (minimum 6.15:1 text contrast). No browser errors. These checks are not a full accessibility or real-device certification. The inherited publication guard still rejects public wording from progress issue #615. The repository game-free suite reports 500 passed, 13 known baseline failures and 69 skipped; details are in the PR description. No game or runtime experiment was performed. **Still WIP, with native dependencies and status/blocked: #264 → #611 → #634 → #638. Do not approve/merge out of order. No hosted preview or deployment was requested.**

after-dark-section.png

174 KiB

before-dark-full.png

1.2 MiB

after-dark-full.png

1.4 MiB

before-light-full.png

1.1 MiB

after-light-section.png

165 KiB

after-light-full.png

1.3 MiB

after-dark-mobile.png

148 KiB

after-light-mobile.png

142 KiB

after-dark-expanded.png

226 KiB

bibiburl commented 2026-11-15 01:56:33 +00:00

Author

Member

• Copy link
• Copy Markdown

Attached nine before/after desktop, mobile and expanded-change screenshots and linked them from the description. Native dependencies are #264, #611 and #634; status/blocked and the ordered approval/merge hold remain. Final website result: 107 tests pass, 16 pages build, and the inherited #615 wording guard still blocks publication. Kept WIP; no hosted preview or deployment requested.

Attached nine before/after desktop, mobile and expanded-change screenshots and linked them from the description. Native dependencies are #264, #611 and #634; status/blocked and the ordered approval/merge hold remain. Final website result: 107 tests pass, 16 pages build, and the inherited #615 wording guard still blocks publication. Kept WIP; no hosted preview or deployment requested.

bibiburl added 1 commit 2026-11-15 02:08:56 +00:00

feat(website): add nightly, release candidate and stable tabs ... 6fc80620db

Collect an independent release comparison for each channel, identify RCs explicitly, and retain useful empty states. Add keyboard-accessible tabs with a readable no-script fallback and show the snapshot time in UTC.

Assisted-by: Codex (OpenAI)

bibiburl commented 2026-11-15 02:09:46 +00:00

Author

Member

• Copy link
• Copy Markdown

Updated first attempt: Nightly / Release candidate / Stable tabs, with an explicit Release candidate badge. After: this PR @ 6fc80620; before: first attempt @ 6f72040e. These are rendered screenshots from the local Worker, not mockups.

The tabs currently show Nightly 170 versus Nightly 164 (six source changes), RC 3 versus RC 2 (twelve), and “No stable release yet.” Each available channel has its own highlights and complete source links. Older RCs never compare against a stable published after them. The checked timestamp now includes UTC time.

Other views:

• Nightly, light
• Release candidate, dark
• Stable, light
• Release candidate on mobile, dark
• Stable on mobile, light
• Full homepage, dark
• Full homepage, light

Validation: 112/112 website tests pass; 16 pages build. All three tabs in both themes at 320, 390, 768, 1024, 1440 and 1920px fit without horizontal overflow. Arrow wrapping, Home/End, Enter/Space, Tab into panels, the RC disclosure and deep links pass. Blocking all page scripts leaves every channel readable and anchor navigation working. Axe reports zero violations for all six channel/theme states; Nightly and RC retain one incomplete gradient-contrast category, Stable has none. No normal-preview browser errors.

The repository game-free suite remains 500 passed, the same 13 baseline failures and 69 skipped (100s). Publication is still blocked by the inherited #615 wording check; no guard was bypassed. Full details are in the updated description.

Freshness: the existing workflow is already configured for an hourly rebuild at :07, website-source pushes on main and manual rebuilds. Successful builds refresh release data; handwritten highlights remain editorial. No workflow change, hosted preview request or deployment was made.

Keep WIP and the ordered hold: #264 → #611 → #634 → #638.

Updated first attempt: **Nightly / Release candidate / Stable tabs**, with an explicit Release candidate badge. After: this PR @ `6fc80620`; [before: first attempt @ 6f72040e](https://git.magicbean.dev/MagicBeanDev/Longshot/pulls/638#issuecomment-21684). These are rendered screenshots from the local Worker, not mockups. The tabs currently show Nightly 170 versus Nightly 164 (six source changes), RC 3 versus RC 2 (twelve), and “No stable release yet.” Each available channel has its own highlights and complete source links. Older RCs never compare against a stable published after them. The checked timestamp now includes UTC time.    Other views: - [Nightly, light](https://git.magicbean.dev/attachments/954c434b-7c9d-4867-8664-d0b8695a8462) - [Release candidate, dark](https://git.magicbean.dev/attachments/9b2417a2-f53a-4704-9432-8669a3ff308b) - [Stable, light](https://git.magicbean.dev/attachments/c31d6e84-47fa-47f7-9e20-58e909588a1e) - [Release candidate on mobile, dark](https://git.magicbean.dev/attachments/cc592b25-6249-475e-b9d8-85534146c8ac) - [Stable on mobile, light](https://git.magicbean.dev/attachments/2686a465-45e5-4645-a71d-dfcab4c4309f) - [Full homepage, dark](https://git.magicbean.dev/attachments/0c4a709c-97c3-4e8c-92f8-c4b5f759cd69) - [Full homepage, light](https://git.magicbean.dev/attachments/5cb93709-ffa9-4cf9-9f5b-ce642d8b6fca) Validation: 112/112 website tests pass; 16 pages build. All three tabs in both themes at 320, 390, 768, 1024, 1440 and 1920px fit without horizontal overflow. Arrow wrapping, Home/End, Enter/Space, Tab into panels, the RC disclosure and deep links pass. Blocking all page scripts leaves every channel readable and anchor navigation working. Axe reports zero violations for all six channel/theme states; Nightly and RC retain one incomplete gradient-contrast category, Stable has none. No normal-preview browser errors. The repository game-free suite remains 500 passed, the same 13 baseline failures and 69 skipped (100s). Publication is still blocked by the inherited #615 wording check; no guard was bypassed. Full details are in the updated description. Freshness: the existing workflow is already configured for an hourly rebuild at :07, website-source pushes on main and manual rebuilds. Successful builds refresh release data; handwritten highlights remain editorial. No workflow change, hosted preview request or deployment was made. **Keep WIP and the ordered hold: #264 → #611 → #634 → #638.**

stable-dark-desktop.png

41 KiB

nightly-dark-desktop.png

178 KiB

rc-light-desktop.png

171 KiB

rc-dark-mobile.png

154 KiB

nightly-light-desktop.png

169 KiB

stable-light-mobile.png

37 KiB

full-dark.png

1.4 MiB

rc-dark-desktop.png

180 KiB

stable-light-desktop.png

41 KiB

full-light.png

1.3 MiB

bibiburl commented 2026-11-15 02:10:29 +00:00

Author

Member

• Copy link
• Copy Markdown

Added Nightly, Release candidate and Stable tabs at 6fc80620, with independent release ranges, explicit RC badges, accessible keyboard controls and a readable no-script fallback. UTC timestamps now include the time. Attached ten current screenshots; 112 website tests and the 16-page build pass. The inherited #615 publication blocker and #264 → #611 → #634 → #638 hold remain. Existing hourly refresh behavior is documented; deployment automation is unchanged.

Added Nightly, Release candidate and Stable tabs at 6fc80620, with independent release ranges, explicit RC badges, accessible keyboard controls and a readable no-script fallback. UTC timestamps now include the time. Attached ten current screenshots; 112 website tests and the 16-page build pass. The inherited #615 publication blocker and #264 → #611 → #634 → #638 hold remain. Existing hourly refresh behavior is documented; deployment automation is unchanged.

bibiburl added 1 commit 2026-11-15 02:19:08 +00:00

feat(website): refresh release tabs after publication ... a134608098

Dispatch the website from completed releases, retain the hourly fallback, and report production failures with a single incident issue that closes on recovery.

Assisted-by: Codex (OpenAI)

bibiburl changed title from WIP: feat(website): show latest build and release changes to WIP: feat(website): browse Nightly, RC and Stable release tabs 2026-11-15 02:19:22 +00:00

bibiburl commented 2026-11-15 02:19:24 +00:00

Author

Member

• Copy link
• Copy Markdown

Reframed this PR around the Nightly, Release candidate and Stable tabs, with the current tab screenshots leading the visual evidence. Added supporting automatic refresh after successful release publication, the existing hourly fallback, a shared production queue and one failure issue per outage that closes after recovery. Failure emails are enabled as a secondary route. Head a1346080: 119 website tests and 5 workflow tests pass; the full game-free suite has 501 passes, the same 13 recorded baseline failures and 69 skips. The #615 publication-wording blocker remains. No live dispatch, alert, preview or deployment was performed. Keep WIP and do not approve, merge or deploy before #264 → #611 → #634 are approved in order and merged.

Reframed this PR around the Nightly, Release candidate and Stable tabs, with the current tab screenshots leading the visual evidence. Added supporting automatic refresh after successful release publication, the existing hourly fallback, a shared production queue and one failure issue per outage that closes after recovery. Failure emails are enabled as a secondary route. Head a1346080: 119 website tests and 5 workflow tests pass; the full game-free suite has 501 passes, the same 13 recorded baseline failures and 69 skips. The #615 publication-wording blocker remains. No live dispatch, alert, preview or deployment was performed. Keep WIP and do not approve, merge or deploy before #264 → #611 → #634 are approved in order and merged.

bibiburl commented 2026-11-15 02:19:44 +00:00

Author

Member

• Copy link
• Copy Markdown

Checklist finalized: the new source-reading workflow test is included, and rollout verification is explicitly a manual maintainer check. WIP and the ordered dependency hold are unchanged.

Checklist finalized: the new source-reading workflow test is included, and rollout verification is explicitly a manual maintainer check. WIP and the ordered dependency hold are unchanged.

bibiburl commented 2026-11-15 02:25:53 +00:00

Author

Member

• Copy link
• Copy Markdown

Added the current tabbed screenshots directly to the main PR description: all three channel states, mobile views, both themes and full-size homepage links. The earlier version without tabs is explicitly marked superseded so reviewers know which design this PR proposes. WIP and prerequisite order are unchanged.

Added the current tabbed screenshots directly to the main PR description: all three channel states, mobile views, both themes and full-size homepage links. The earlier version without tabs is explicitly marked superseded so reviewers know which design this PR proposes. WIP and prerequisite order are unchanged.

bibiburl commented 2026-11-15 03:32:20 +00:00

Author

Member

• Copy link
• Copy Markdown

Confirmed #264 merged at d7aeaa26 on 2026-11-15 at 03:24 UTC. The main description now marks that prerequisite complete and shows the remaining order: #611 → #634 → #638. Keep this PR WIP and blocked until the remaining prerequisites are approved and merged in order. The current-main checklist is now pending because the stack still needs reconciliation and revalidation against the merged results; no code, preview or deployment was changed.

Confirmed #264 merged at d7aeaa26 on 2026-11-15 at 03:24 UTC. The main description now marks that prerequisite complete and shows the remaining order: #611 → #634 → #638. Keep this PR WIP and blocked until the remaining prerequisites are approved and merged in order. The current-main checklist is now pending because the stack still needs reconciliation and revalidation against the merged results; no code, preview or deployment was changed.

bibiburl referenced this pull request 2026-11-15 04:41:38 +00:00

fix(website): improve mobile browsing and launcher downloads #611

bibiburl referenced this pull request 2026-11-15 04:41:40 +00:00

fix(website): improve mobile browsing and launcher downloads #611

bibiburl referenced this pull request 2026-11-15 04:41:46 +00:00

docs(website): add a theme-aware trails hero #634

bibiburl changed title from WIP: feat(website): browse Nightly, RC and Stable release tabs to feat(website): browse Nightly, RC and Stable release tabs 2026-11-15 04:41:48 +00:00

bibiburl commented 2026-11-15 04:41:51 +00:00

Author

Member

• Copy link
• Copy Markdown

Marked ready for review at the author’s explicit request (#611 was already non-WIP). #264 is merged; the remaining approval/merge order stays #611 → #634 → #638. The description records the current merge conflicts and outstanding validation; unchecked work remains unchecked. Dependencies and blocked labels are preserved. No code, approval, merge or deployment was performed.

Marked ready for review at the author’s explicit request (#611 was already non-WIP). #264 is merged; the remaining approval/merge order stays #611 → #634 → #638. The description records the current merge conflicts and outstanding validation; unchecked work remains unchecked. Dependencies and blocked labels are preserved. No code, approval, merge or deployment was performed.

madebyjordan requested changes 2026-11-15 11:01:48 +00:00

Dismissed

madebyjordan left a comment

• Copy link
• Copy Markdown

Reviewed at a1346080 (only this PR's three commits on top of #634). The workflow side is sound:

• Untrusted code can't reach the dispatch or the incident reporter. Release dispatches only after a push-triggered publish, and the report step is skipped whenever IS_PR.
• Production now always deploys main.

The tabs follow the ARIA tabs pattern properly and work without JavaScript. 119 website tests pass, and a live collection takes 12 API calls. Changes needed:

Blocking

1. The nightly comparison will stop every production deploy.

   • Once a stable exists, a nightly's baseline is that stable, and the range grows with every merge.
   • Forgejo's compare endpoint slows with it, timed live: 300 commits took 15 s, 819 took 30.4 s, and 2,261 took 50 s and returned 7 MB. The timeout is 30 s × 3 attempts (progress-request.ts:16).
   • After that, buildReleaseSummary returns null and check-site.ts (~61–70) fails, so nothing deploys: not the hourly refresh, not a website fix. Same-repo PR previews fail too.
   • At main's pace, that's one to two weeks after the first stable.
   • Fix: cap or paginate the range (the count plus a Compare link), and let a failed channel degrade its own panel instead of failing the build.

2. Anyone can take over the refresh incident (website/scripts/report-refresh.mjs ~22). It treats any open issue whose body starts with the marker as the incident, whoever posted it. Someone opens an issue starting with <!-- longshot-website-refresh-failure -->, and then:

   • real failures never open an incident;
   • the next successful deploy uses the job token to comment on that stranger's issue and close it.

   Match on the poster (the Actions user) or a label.

Should fix

3. The baseline depends on which nightlies are still unpruned (release-summary.ts:39). With no stable, a prerelease compares against older[0], the previous build of any channel, and Release keeps only 5 nightlies.

   • rc.4 compares against rc.3 today only because the nightlies before it were pruned; when it was new, its baseline was a nightly.
   • If an rc is tagged on a commit older than a nightly published before it, the ancestry check (:172) throws and (1) follows.

   Comparing an rc with the previous rc avoids both.

4. Curated highlights mostly don't show.

   • Three of the six keys in release-highlights.json (a1f0c539, 4b9fad06, fabcd2c0) are in no visible range today, and a nightly highlight is visible for about an hour.
   • Nothing checks that a key is a real commit, so a typo is silently ignored.
   • Without curation the newest three commits of any type show, so the live nightly headline is a chore(agents) commit. Prefer feat/fix/perf.

5. A commit subject can't be edited, and check-site.ts:64 applies the branding check to subjects. publicGameText leaves several single names alone, so one such subject would block deploys for as long as it's in a range. None of main's current subjects trip it, but neutralising or dropping such subjects is safer than failing the build.

6. In website.yml:

   • Line 25: failure emails go out on every failed scheduled run, so (1) means 24 a day.
   • Label events (pull_request_target: [labeled]) also create runs in each PR's preview group. If Forgejo keeps one pending run per group, adding any label during a preview build can cancel the newest commit's preview.

7. Smaller points:

   • ReleaseSummary.astro ~162–174 repeats the base rules inside the 30rem query.
   • tests/test_website_refresh.py only checks the YAML for substrings.
   • Clicking a tab doesn't update the hash, so a chosen channel can't be shared once JavaScript is on.

This is built on #634, which won't merge (#768 replaces the hero), so it needs moving onto main or #611. The tabs sit below the hero and fit #768's design as they are.

Review drafted with Claude Code and checked by Jordan.

Reviewed at a1346080 (only this PR's three commits on top of #634). The workflow side is sound: - Untrusted code can't reach the dispatch or the incident reporter. Release dispatches only after a push-triggered publish, and the report step is skipped whenever `IS_PR`. - Production now always deploys main. The tabs follow the ARIA tabs pattern properly and work without JavaScript. 119 website tests pass, and a live collection takes 12 API calls. Changes needed: **Blocking** 1. **The nightly comparison will stop every production deploy.** - Once a stable exists, a nightly's baseline is that stable, and the range grows with every merge. - Forgejo's compare endpoint slows with it, timed live: 300 commits took 15 s, 819 took 30.4 s, and 2,261 took 50 s and returned 7 MB. The timeout is 30 s × 3 attempts (`progress-request.ts:16`). - After that, `buildReleaseSummary` returns null and `check-site.ts` (~61–70) fails, so nothing deploys: not the hourly refresh, not a website fix. Same-repo PR previews fail too. - At main's pace, that's one to two weeks after the first stable. - Fix: cap or paginate the range (the count plus a Compare link), and let a failed channel degrade its own panel instead of failing the build. 2. **Anyone can take over the refresh incident** (`website/scripts/report-refresh.mjs` ~22). It treats any open issue whose body starts with the marker as the incident, whoever posted it. Someone opens an issue starting with `<!-- longshot-website-refresh-failure -->`, and then: - real failures never open an incident; - the next successful deploy uses the job token to comment on that stranger's issue and close it. Match on the poster (the Actions user) or a label. **Should fix** 3. **The baseline depends on which nightlies are still unpruned** (`release-summary.ts:39`). With no stable, a prerelease compares against `older[0]`, the previous build of any channel, and Release keeps only 5 nightlies. - rc.4 compares against rc.3 today only because the nightlies before it were pruned; when it was new, its baseline was a nightly. - If an rc is tagged on a commit older than a nightly published before it, the ancestry check (`:172`) throws and (1) follows. Comparing an rc with the previous rc avoids both. 4. **Curated highlights mostly don't show.** - Three of the six keys in `release-highlights.json` (a1f0c539, 4b9fad06, fabcd2c0) are in no visible range today, and a nightly highlight is visible for about an hour. - Nothing checks that a key is a real commit, so a typo is silently ignored. - Without curation the newest three commits of any type show, so the live nightly headline is a `chore(agents)` commit. Prefer feat/fix/perf. 5. A commit subject can't be edited, and `check-site.ts:64` applies the branding check to subjects. `publicGameText` leaves several single names alone, so one such subject would block deploys for as long as it's in a range. None of main's current subjects trip it, but neutralising or dropping such subjects is safer than failing the build. 6. In `website.yml`: - Line 25: failure emails go out on every failed scheduled run, so (1) means 24 a day. - Label events (`pull_request_target: [labeled]`) also create runs in each PR's preview group. If Forgejo keeps one pending run per group, adding any label during a preview build can cancel the newest commit's preview. 7. Smaller points: - `ReleaseSummary.astro` ~162–174 repeats the base rules inside the 30rem query. - `tests/test_website_refresh.py` only checks the YAML for substrings. - Clicking a tab doesn't update the hash, so a chosen channel can't be shared once JavaScript is on. This is built on #634, which won't merge (#768 replaces the hero), so it needs moving onto main or #611. The tabs sit below the hero and fit #768's design as they are. *Review drafted with Claude Code and checked by Jordan.*

bibiburl force-pushed docs/release-summary from a134608098 to 0ebd90c776 2026-11-15 12:24:56 +00:00 Compare

bibiburl referenced this pull request 2026-11-15 12:25:27 +00:00

fix(website): improve mobile browsing and launcher downloads #611

bibiburl commented 2026-11-15 12:25:44 +00:00

Author

Member

• Copy link
• Copy Markdown

Updated review screenshots for 0ebd90c7. The main description will use these current images. Before images are main aa95d35f. Individual after images show only this PR; combined images are a local integration of #611, #638, #709 and #768, not a deployed site.

Combined integration: no merge conflicts, 18 pages built, 125 website tests pass with the same existing HDR-condition failure. All 16 viewport/theme/route layout checks have no horizontal overflow. The mobile light hero from #768 has a separate tagline contrast finding (3.67:1 versus 4.5:1 required); the download and release-tab changes do not edit that hero.

Updated review screenshots for `0ebd90c7`. The main description will use these current images. Before images are main `aa95d35f`. Individual after images show only this PR; combined images are a local integration of #611, #638, #709 and #768, not a deployed site.       Combined integration: no merge conflicts, 18 pages built, 125 website tests pass with the same existing HDR-condition failure. All 16 viewport/theme/route layout checks have no horizontal overflow. The mobile light hero from #768 has a separate tagline contrast finding (3.67:1 versus 4.5:1 required); the download and release-tab changes do not edit that hero.

638-before-main-home.png

819 KiB

638-after-rc.png

214 KiB

638-after-stable.png

293 KiB

combined-home-390-dark.png

558 KiB

638-after-nightly-mobile.png

109 KiB

combined-home-1440-light.png

706 KiB

bibiburl referenced this pull request 2026-11-15 12:26:39 +00:00

fix(website): improve mobile browsing and launcher downloads #611

bibiburl commented 2026-11-15 12:26:41 +00:00

Author

Member

• Copy link
• Copy Markdown

Addressed Jordan’s review in 0ebd90c7, rebased directly onto main aa95d35f with no #634/#611 commits. Release ranges now use bounded paginated commit requests (50 displayed plus full count and Compare link), failed channels degrade independently, RC baselines ignore nightly pruning, and incidents require the authenticated author. Added regression coverage, player-facing highlight selection and shareable tab hashes; isolated unrelated label queues and disabled repeat Website emails. Current screenshots and validation are in the main post. Please re-check this head. Neither this update nor the integration preview requests deployment.

Addressed Jordan’s review in 0ebd90c7, rebased directly onto main aa95d35f with no #634/#611 commits. Release ranges now use bounded paginated commit requests (50 displayed plus full count and Compare link), failed channels degrade independently, RC baselines ignore nightly pruning, and incidents require the authenticated author. Added regression coverage, player-facing highlight selection and shareable tab hashes; isolated unrelated label queues and disabled repeat Website emails. Current screenshots and validation are in the main post. Please re-check this head. Neither this update nor the integration preview requests deployment.

bibiburl removed a dependency 2026-11-15 12:27:05 +00:00

#634 docs(website): add a theme-aware trails hero

bibiburl removed a dependency 2026-11-15 12:27:06 +00:00

#611 fix(website): improve mobile browsing and launcher downloads

bibiburl removed a dependency 2026-11-15 12:27:06 +00:00

#264 docs(guides): illustrate installation, settings and save transfer

bibiburl referenced this pull request 2026-11-15 12:27:07 +00:00

docs(website): add a theme-aware trails hero #634

bibiburl referenced this pull request 2026-11-15 12:27:07 +00:00

docs(website): add a theme-aware trails hero #634

madebyjordan requested changes 2026-11-15 12:48:38 +00:00

madebyjordan left a comment

• Copy link
• Copy Markdown

Re-checked at 0ebd90c7. Points 1–6 are fixed:

• The comparison is bounded: a live build makes 15 API calls in 3.5 s, and a 2,056-commit range takes 0.93 s with an exact count.
• A failed channel shows "unavailable" without failing the build.
• RC baselines no longer depend on pruning.
• Highlights prefer feat/fix/perf and skip unknown commits.
• Branded subjects are neutralised.

It merges cleanly with main, #611 and #768. One blocker is left, and it's about one line:

The refresh reporter fails on every production run (website/scripts/report-refresh.mjs:19). A job token authenticates as the Actions user, whose ID is −2: GET /user answers id: -2 (Forgejo's models/user/user_system.go, services/auth/method/action_task_token.go). The new check identity.id < 1 throws on that. So "Report production refresh health" fails on every main run, and a run that deployed fine still ends red. A real outage never opens an incident, and with Website emails now off, nobody hears about it. The test misses it because its mock returns id: 42 (refresh-report.test.ts:12). Accepting the Actions user (id === -2, login forgejo-actions) keeps the protection, because other accounts still can't post as it. A mock with id: -2 would catch it.

Small notes, not blocking:

• github.run_id is empty when Forgejo evaluates concurrency, so every unrelated-label run shares website-ignored-. That's harmless, since the job is skipped and nothing is cancelled, but the ignored-42 case in test_website_refresh.py:73 checks something that can't happen.
• release.yml:27 adds enable-email-notifications: true, which turns on failure emails for the whole Release workflow. If that's intended, worth saying so in the description.
• If collection fails as a whole (repository or release list), a deploy now succeeds and shows every tab as unavailable until the next run. That's fine once the reporter works; today nothing would say so.

Review drafted with Claude Code and checked by Jordan.

Re-checked at 0ebd90c7. Points 1–6 are fixed: - The comparison is bounded: a live build makes 15 API calls in 3.5 s, and a 2,056-commit range takes 0.93 s with an exact count. - A failed channel shows "unavailable" without failing the build. - RC baselines no longer depend on pruning. - Highlights prefer feat/fix/perf and skip unknown commits. - Branded subjects are neutralised. It merges cleanly with main, #611 and #768. One blocker is left, and it's about one line: **The refresh reporter fails on every production run** (`website/scripts/report-refresh.mjs:19`). A job token authenticates as the Actions user, whose ID is −2: `GET /user` answers `id: -2` (Forgejo's `models/user/user_system.go`, `services/auth/method/action_task_token.go`). The new check `identity.id < 1` throws on that. So "Report production refresh health" fails on every main run, and a run that deployed fine still ends red. A real outage never opens an incident, and with Website emails now off, nobody hears about it. The test misses it because its mock returns `id: 42` (`refresh-report.test.ts:12`). Accepting the Actions user (`id === -2`, login `forgejo-actions`) keeps the protection, because other accounts still can't post as it. A mock with `id: -2` would catch it. Small notes, not blocking: - `github.run_id` is empty when Forgejo evaluates `concurrency`, so every unrelated-label run shares `website-ignored-`. That's harmless, since the job is skipped and nothing is cancelled, but the `ignored-42` case in `test_website_refresh.py:73` checks something that can't happen. - `release.yml:27` adds `enable-email-notifications: true`, which turns on failure emails for the whole Release workflow. If that's intended, worth saying so in the description. - If collection fails as a whole (repository or release list), a deploy now succeeds and shows every tab as unavailable until the next run. That's fine once the reporter works; today nothing would say so. *Review drafted with Claude Code and checked by Jordan.*

mooki added the

kind

feature

platform/linux

platform/windows

priority

low

labels 2026-11-15 13:56:47 +00:00

malon referenced this pull request 2026-11-15 15:11:03 +00:00

Game: Keeps on crashing in death mountain #817

HumanBean force-pushed docs/release-summary from 0ebd90c776 to 733f20c345 2026-11-15 18:48:30 +00:00 Compare

HumanBean commented 2026-11-15 18:50:12 +00:00

Owner

• Copy link
• Copy Markdown

Rebased onto main @ 7bacf846. One conflict in website/scripts/check-site.ts (the release-summary.ts import next to main's public-copy.ts import): kept both imports. The PR's own diff is the same files and line counts as before. No tests were run here; the test boxes stay yours.

Rebased onto `main` @ 7bacf846. One conflict in `website/scripts/check-site.ts` (the `release-summary.ts` import next to main's `public-copy.ts` import): kept both imports. The PR's own diff is the same files and line counts as before. No tests were run here; the test boxes stay yours.

Some checks failed

Hide checks

CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped

Details

CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped

Details

CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped

Details

Third-party parts / The same as upstream (pull_request) Has been skipped

Details

Website / Website on Cloudflare (pull_request) Has been skipped

Details

PR description / Description follows the template (pull_request_target) Successful in 3s

Details

CI / Conventional Commit title (pull_request) Successful in 3s

Details

CI / Workflow lint (actionlint) (pull_request) Failing after 4s

Details

Guide screenshots / Guide images and local UI source review (pull_request) Failing after 5s

Details

Third-party parts / As their notes say (pull_request) Successful in 8s

Details

This pull request doesn't have enough approvals yet. 0 of 1 approvals granted.

This branch is out-of-date with the base branch

You are not authorized to merge this pull request.

View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.

git fetch -u 'docs/release-summary:bibiburl-docs/release-summary'

git switch 'bibiburl-docs/release-summary'

Sign in to join this conversation.

Reviewers

No reviewers

madebyjordan

Labels

Clear labels   

area/accessibility

Barriers for people with disabilities

  

area/audio

Sound and music

  

area/input

Controllers, keyboard, mouse

  

area/install

Installing, updating, building from your game

  

area/menu

In-game menu and settings

  

area/renderer

Vulkan renderer, shaders, effects

  

area/saves

Save files

  

area/upscaling

DLSS, FSR, frame generation

  

beanstalk: approved @ 1697603

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 2c5a62e

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 4fd82b7

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 64d6d09

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 6f91fae

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 70a78f9

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 8015c9a

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 8d361dc

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ 9d8356e

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ abf2227

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ d82f202

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ e5f1e40

Set by Beanstalk for show; the farm's database is what counts

  

beanstalk: approved @ ee8ab5f

Set by Beanstalk for show; the farm's database is what counts

  

ci: approved

A maintainer approved CircleCI for this fork PR (sticky; removed after 14 quiet days)

  

good first issue

A good start for a new contributor

  

gpu/amd

AMD GPUs

  

gpu/igpu

Integrated GPUs

  

gpu/intel

Intel GPUs

  

gpu/nvidia

NVIDIA GPUs

  

help wanted

Extra hands or knowledge wanted

  

kind

bug

Something doesn't work as it should

  

kind

chore

Build, CI, dependencies, tidying

  

kind

docs

Documentation

  

kind

feature

A new feature or an improvement

  

kind

performance

Frame rate, frame time, loading, memory

  

kind

question

A question rather than a change

  

needs measurements

Waiting for before/after measurements

  

platform/linux

Affects Linux (a change for every system carries both Windows and Linux)

  

platform/macos

macOS (Apple silicon or Intel Macs)

  

platform/steam-deck

Steam Deck

  

platform/windows

Affects Windows (a change for every system carries both Windows and Linux)

  

preview

Maintainers: deploy a preview of this PR's website (read its changes first). Removed after the run.

  

priority

critical

Crashes, data loss, or blocks most players

  

priority

high

Important: next in line

  

priority

low

Nice to have

  

regression

Worked in an earlier build, broken in a newer one: highest priority, above other bugs

  

status

blocked

Waits on something else first

  

status

confirmed

Reproduced or agreed on

  

status

duplicate

Already reported or proposed

  

status

invalid

Not a problem in this project

  

status

needs-info

Waiting for more information from the reporter

  

status

wontfix

Will not be worked on

  

testing wanted

Wanted: a test on other hardware (posts a test request in #test-requests)

  

ci: approved

No labels

area/accessibility

area/audio

area/input

area/install

area/menu

area/renderer

area/saves

area/upscaling

beanstalk: approved @ 1697603

beanstalk: approved @ 2c5a62e

beanstalk: approved @ 4fd82b7

beanstalk: approved @ 64d6d09

beanstalk: approved @ 6f91fae

beanstalk: approved @ 70a78f9

beanstalk: approved @ 8015c9a

beanstalk: approved @ 8d361dc

beanstalk: approved @ 9d8356e

beanstalk: approved @ abf2227

beanstalk: approved @ d82f202

beanstalk: approved @ e5f1e40

beanstalk: approved @ ee8ab5f

ci: approved

good first issue

gpu/amd

gpu/igpu

gpu/intel

gpu/nvidia

help wanted

kind

bug

kind

chore

kind

docs

kind

feature

kind

performance

kind

question

needs measurements

platform/linux

platform/macos

platform/steam-deck

platform/windows

preview

priority

critical

priority

high

priority

low

regression

status

blocked

status

confirmed

status

duplicate

status

invalid

status

needs-info

status

wontfix

testing wanted

ci: approved

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

3 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

MagicBeanDev/Longshot!638

No description provided.