build_port's four archives carry install.py pinned by SHA-256, so they
change with the installer, not with the packs. The site now builds them
from its own commit and serves them under get/handheld/. Each get/ page
of Batocera, KNULLI, ROCKNIX and AmberELEC names its file, size, hash,
folder and menu, and get/linux-handhelds/ does the same for muOS, ArkOS,
dArkOS and JELOS. The installer's handheld section opens on a download
table, both tables written from sitehandheld.LAUNCHERS.
Every archive has a recipe in recipes/archives.json, and no page a user
reads said so. A short reference page says what a recipe is, how to
rebuild one archive and what exact and canonical mean. Its counts and the
README line are counted from the recipe file at each build.
The README gains a row of four badges: PyPI version, Python versions,
the Validation workflow on main and the license. Tests hold the badges,
the counts and the page's command.
- The installer says "Every file on the list is in place." where it
said "Everything is in place.": a check covers the files the
collection holds, and the required ones it lacks follow.
- The FAQ, first install, legal and home pages state only what the
code shows (PlayStation 1 and Saturn run without a BIOS on some
emulators, save files are never touched, records are added).
- MinUI and MyMinUI show their own boot logos, and the Android,
Linux handheld and iPhone tiles show the Android robot, Tux and a
plain device icon, each credited in sources.json.
- Each Markdown copy, llms.txt and llms-full.txt name the commit
they were built from, version.json the current one, and the
answering rules tell assistants to fetch a stale copy again.
- The deploy lists the pages whose words changed (digests.json) and
posts them to IndexNow once the site is live.
The build writes <page>/index.md beside each page, named in its head,
and llms-full.txt. llms.txt and the setup skill say how to answer from
the site, not from memory, and the setup_bios_files WebMCP tool gives a
setup's folder and exact file names from the picker data.
llms.txt and an ARD manifest name the pages and the JSON API, and lead
an agent setting up a device to --for-games --required-only. Every page
registers four read-only WebMCP tools: the setup command, a hash lookup,
an emulator's files and the site search.
files/?sha1= of a file the collection no longer holds answered that no
file the site knows has the hash. generate_db.py now records each sha1 a
rebuild drops in retired.json, with the day and the file in its place,
backfilled from the history since 0d586ef6b (4,139 files). The lookup
answers with it and leads to the successor's card.
An ArkOS or AmberELEC games part is named on the first screen and comes
first at the folder question. Packs that lack files are said so before
any plan. The served bootstrap gives advice that fits it, and the site
build refuses a copy whose pin is not its installer.
Records that back the same field or issue claim share one line. Each
line says what the runs showed: the BIOS the core ran, the file strace
saw opened, or what a maintainer watched in an issue's run
(tests/campaign/observations.yml). A run with nothing to state stays
off the page.
Emulator pages show a run only when a profile's evidence names it or it
answers an issue, as one line naming the BIOS the core ran. The Tested
blocks, region outcomes, coverage, campaign and pack tables and the
benchmark figures are gone from the site; the records stay.
The site serves install.sh at /retrobios/i and install.ps1 at
/retrobios/i.ps1. generate_site.py writes both copies with the
installer address set to install.py at the commit the site is built
from (GITHUB_SHA in CI, HEAD locally), so the pinned SHA-256 and the
downloaded installer always come from one commit, including while a
deploy is still running.
Every command a user copies names the short address. The raw form
stays on the installer page as the fallback. Deploy Site now runs on
changes to install.sh and install.ps1. The install box weight keeps
the address length its ratio was fitted on, so no console page moves.
MAME's card pages, GroovyMAME and yaps2 read 13,000 to 16,700 words of
cards. Each entry is now a table row (name, members, status, short
hash) that opens to its card in a details block, with the site's filter
box and row counter. Pages from 150 entries and every split part use it.
Visible words: mame-files-1 16,680 to 1,612, groovymame 14,396 to 1,876.
A row costs about five elements, so a part holds 180 cards: MAME goes
from 8 parts to 9, the largest at 567 KB and 8,989 elements.
Each wanted list (TOSEC, No-Intro, Redump) opens on its counts, then one
table folded under a filter box that opens it, in place of a block per
DAT. Only the names are searched: indexed descriptions and hashes put
the PlayStation page ahead of the PlayStation 2 one.
A filtered row stayed visible on a phone: a stacked row's display:block
outweighed the hidden attribute on every stacked table.
The new probes now refuse abbreviated options, read the GitHub token through github_headers, type every signature and document every option. The site build watches region_check.py and the content list, and validate.yml's tests/** counts as watching the list. The testing guide names the two new test modules.
The site compiles the index and the SQLite files from measurements/
instead of a pinned clone. The pin, the sync script, the CI clone steps,
the freshness area and its refresh handler are gone. core_probe --record
writes into measurements/ of the checkout by default, and a test fails
when a line or data file HEAD holds changed or vanished.
Each emulator with records gets a Tested block: claim, result, method, computer, core build and the permalink of its record at the pin. A Measurements page filters every test in the browser from the compact index, and the Data page lists measurements.sqlite with its tables. One SQLite file per emulator is linked from its page. wiki/measurements.md says how a test runs, held to the Start here budgets.
Both workflows fetch the pinned commit alone with git before the build, and the deploy fails without it.
The deploy job sets NODE_DISABLE_COMPILE_CACHE for the build and the search gate, and lighthouse.py passes it to its children. The paths list gains generate_event_schema.py, which the site build imports.
generate_site.py writes the Starlight project in website/ and nothing
else: mkdocs.yml, the MkDocs hooks, theme overrides and stylesheet go,
and the wiki is written in Starlight's syntax. Both workflows set up
Node with the official actions/setup-node pinned by commit, run npm ci
from the committed lockfile, build with the strict link validator and
check the result with validate_site.py.
The search weighs a page by what it answers: Start here pages first,
a system page's cards and file names before a guide citing them, deep
guides, launchers, aliases and frozen copies after. Status badges,
hash blocks and the export change lists leave the index. The ranking
sits at the centre of a range where the 13 queries of search_gate.py
all answer, in the search box itself.
839 pages, every URL and 45,611 ids match the MkDocs build. The tests
read the Starlight content and syntax.
generate_site.py --engine starlight writes the pages, navigation and
data files of an Astro Starlight project in website/. Links resolve to
site addresses, every heading keeps its MkDocs id, and pages split by
size are measured in the MkDocs Markdown so both builds cut the same
files. The 839 pages and their 45,613 ids match the MkDocs build.
The redesign moved the technical questions to technical-faq.md, made the Get your files rows generated from each platform key and reduced getting-started.md to links. Tests now read those places: the pack-row test reads the rendered page. The new site scripts go through run_cli and refuse abbreviated options. The three maintenance image tools are exempt by name from the build dependency rule: their output is committed and the build only reads it. siteguides.py now triggers the site deploy, and the recipe counts match the store.
scripts/sitediagrams.py writes the explainer, install flow, verification modes, MinUI card, pipeline and data layer diagrams as inline SVG from the registry, the platform files, the install manifests and the pipeline step list. A page asks for one with an image comment, which the decoration pass resolves. The two mermaid diagrams of the architecture page are replaced, so no page loads mermaid from unpkg.
Platform logos and system icons are copied once by fetch_site_images.py with their source and owner recorded, and the build rewrites the remote links to the copies. validate_site refuses an image from another host. Three registry logos named a personal avatar instead of the project's (RomM, ROCKNIX, MiSTer FPGA) and MyMinUI's was a GitHub identicon, now dropped.
One 1200x630 sharing card per section, a 404 page with search and the two entry pages, and a list of screenshots that need a real device.
The install manifest marks each game data file "category": "game_data":
a platform entry only game data profile entries read, or a core file its
profile declares game data. Firmware carries no category. Every manifest
says "categorized": true so the installer can refuse an older list.
generate_pack.py --no-game-data cuts those files from the platform list
and the core profiles before the build, names each one with the cores that
read it as game data, and tags the pack _NoGameData through _narrowings.
verify.py and the pack check take the same cut, so --verify-packs and the
report describe the pack the builder wrote. Modes without a platform
(--emulator, --system alone, --from-md5) and --all-variants refuse it.
A names_searched list is one place the launcher fills with the first name it finds, so a core file it would never open once another name of the list is held is not reported missing. A place no declared file and no packed core file fills receives, under the launcher's name, a file the core accepts by content. The layer is applied inside find_undeclared_files, which verify and the pack builder both read, and a test reads their source to keep it there.
export_native now writes into exports/ by default, and beside each
platform's corrected file a report naming every change, where the run
printed five per kind. A new site page lists each platform with its
original file, a download of the corrected one and every corrected,
filled or withheld value. Each platform page links to its row.
Profiles, the wiki, comments, docstrings, command output and schema descriptions carried em-dashes, clause-joining semicolons, unicode arrows, box-drawing banners and check-mark glyphs. Each sentence was rewritten rather than having its character swapped. Hashes, paths, source_ref values and every cited file:line are unchanged, which a comparison of each parsed profile against the previous revision confirms.
validate_pr comments now say PASS and FAIL in words, and source_ref values in a profile's structured blocks render as code. test_prose_style fails if any of these comes back, in tracked files or in the generated site.
The site imported two parsing functions from profile_sync, so every change to the comparison tool redeployed an unchanged site. The grammar now lives in sourceref.py.
The sitemap gave every page the build date. A page now states its own date
and a mkdocs hook passes it on, or leaves lastmod out. The values added to
the site's code spans today are also escaped: a backtick in a profile or
file name could close the span and inject markup.
The Wanted page listed 79 dumps the collection already produces,
such as a Kickstart's byte-swapped words or the halves of a C128
character ROM. The pipeline now records which missing dumps can be
derived, in derivable.json, and the page moves them to a section of
their own that names the held file and the transform.
The site deploy still triggered on seven scripts it stopped importing
when the pack and site modules were split, and on the test suite it
never runs, so a builder-only push rebuilt the site for nothing. The
triggers now match the build's import closure in both directions.
A CI checkout omits every file over 50 MB, so verify and generate_pack
resolve those database entries against a disk that does not hold them and
report them missing. The generated README then stops matching the
committed one for a reason that has nothing to do with staleness.
restore_large_files.py writes them back from the release cache, matched by
SHA1 rather than by name, and only where the path is gitignored and
absent. The site workflow runs it, and refreshes the data directories, before
generating.
PR validation keeps its four jobs, its comment and its labels. The two
inline schema heredocs become validate_schemas.py --source-only, the
changed-file list is NUL-separated so a path with a space survives, and
the whole suite runs instead of one module.
The offline pipeline, manifest regeneration and mkdocs build stay out of
the pull-request path: they belong to the site workflow, which only runs
on main, and stacking them on every PR does not fit the free-tier
budget.
Deploy Site gains schema validation and rendered-site validation, and
checks that the committed README and CONTRIBUTING match what the
generator produces. write_if_changed compares with the timestamp line
stripped, so that check reports staleness rather than the clock.
Abdessamad Derraz