The sandbox bound only /usr, /lib and /bin. On the CI runner setup-python puts Python under /opt/hostedtoolcache, so the worker could not start and every core_probe and campaign run crashed (7 tests, plus their replay without jsonschema). A venv, pyenv or uv do the same. The interpreter's prefix and every link on the way to it are now bound read-only, each at its own path. Reproduced with uv's Python 3.13 outside /usr, 92 tests pass under it and under the system Python.
Merging the card and platform branches left two SHORT_HASH definitions, 8 and 12. The later one silently won, so the platform rows showed 12 characters while their tests expected 8. One constant now, 12, the card anchor's length. The copy-button selector skips buttons that carry their own data-copy, and the tests follow the merged markup.
Each file is a details whose summary is the row: name, folder when it is not the bare name, status chip. The short hash sits on one copy button beside it, since a summary is itself a control. Opened, the evidence runs across the list. A list with no folder drops the column. Code spans in the evidence are plain text.
The lists split by the share of a page they render to (elements against 10,000, bytes against 600,000) instead of Markdown length. With its parts, Batocera goes from 12 pages, 18,854 elements and 1.21 MB to 10 pages, 13,671 and 1.11 MB. A part that folds back into its platform page redirects there, from the record in website/src/data/platform-parts.json.
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 install box's select had a rule of its own that dropped the arrow
and read as a text field. It now takes the shared select rule and
--rb-select-arrow in both themes.
System and emulator file cards show the name, size, download, a short
SHA1 with a copy button, the setups and the emulators. The full hashes
and the other facts sit in one closed fold, each hash with a copy button.
A system's install box shows one command under the OS tabs for the
setup picked in a list, remembered across pages. Without the script the
list of every command shows, open. The Markdown the page budgets are
measured on is unchanged, so no card moves.
Each file of a platform list is one row: name, folder, status chip, a short hash on a button that copies it whole, and a details cell. The details hold every declared hash, the size, the served copy and the readers. A reference's role is one word (glue, port, upstream, build) defined once in a legend instead of a phrase after every reference.
Visible words: Batocera 2,433 to 2,332, AmberELEC 1,301 to 1,299, RomM 1,607 to 1,596, RetroArch 2,207 to 2,088, RetroDECK arcade part 2 19,127 to 2,104. Rows weigh more Markdown, so 38 more systems move to pages of their own under the existing budgets.
source_ref_items() gives each part of a source_ref with the role of the code it cites. _source_ref_markdown() joins them as before, so every other page renders the same text.
Pins and launcher index recomputed for the merged install.py. Mouse modes are now written before the frame, so a screen read back already takes clicks.
core_version is a string the code states, null or keyed by build mode,
and the schema refuses Git, SVN, unknown, master, main, latest, dev and
HEAD, alone or annotated. version_basis is defined per kind with the
fields each one needs, and kind none requires a null version.
tests/test_version_basis.py reads a recorded fixture of forge answers
offline and fails on drift, holds every profile to a basis (tag versions
in describe form, aliases equal to their profile, none as null), checks
that a pin advance reads the version again or is refused, and that the
schema refuses placeholders.
Each standalone profile records where its version is read: the source
at its pin (clk 2026-07-23-136-g9020107, shadps4 0.20.1 WIP, eden
v0.2.0-401-g26a03d2), the APK, binary or archive read (kind package),
or, where no reading can be made, kind unverified with the reason: a
host the tools cannot read, a forge gone or not answering, no build
recorded as read, a pin on another branch. Those keep their value,
unconfirmed.
Each libretro profile records version_basis and core_version is read
there at its pin. snes9x states 1.63 in snes9x.h, nestopia 2.0.0, mgba
0.11.0, vice 3.10, atari800's libretro tree 7.0.0. Cores whose build
takes its version from git carry the nearest tag in describe form
(flycast v2.7-90-ge88b0ce, ppsspp, duckstation, gearcoleco), placeholders
copied from .info files (Git, SVN) are replaced by the code's statement,
and a code that states none is null with kind none.
The installer reads the version the system states in its own file,
without running anything: batocera.version, recalbox.version (a release
only), RetroBat's system/version.info, VERSION_ID in os-release on Lakka
and EmuELEC. When it differs from the version the list was built for, a
line says so before the install and again with the result, and the new
version event carries it for the screen. An unknown version says
nothing, and the install never stops on it.
install/<platform>.json wrote the constant "1.0" as the version of the
list it was built from. The manifest now carries the version the platform
YAML records, and generate_pack refuses a YAML without one.
inherits: no longer passes a parent's version to a child: Lakka was
published as RetroArch v1.22.2 in its manifest, pack name and site page.
lakka.yml records v6.1, the Lakka-LibreELEC tag its target list is read
at. The MiSTer scraper refuses a BiosDB without a timestamp instead of
writing an empty version.
A pack's manifest.json records list_version, so an install from the pack
alone knows it too. Manifests regenerated twice and compared: only
version and generated changed.
Modes 1000 and 1006 only, SGR and legacy reports decoded, a hit map built from each drawn frame, action on release. Off on every way out, on the screens whose text is copied, with a pad route or RETROBIOS_NO_MOUSE.
The run header, the menu, its game data split and g, the plan, the
screen's plan and game data row and both results now divide one tally
of the list the run keeps. Case variants were counted on the run and
missing from the screen's split, g counted the whole list under a head
narrowed by c, and the result left out the data tree files. The wiki
examples come from real runs of 2026-10-27.
The FAQ table of arcade romsets was typed by hand and told readers to
collect MAME 0.160 and FinalBurn Neo 1.0.0.2 sets for cores whose source
states 0.159 and 1.0.0.03. generate_site now writes it between generated
markers from the profiles, and a test holds the page to the generator.
An emulator page names where its version is read (the file, the tag and
the commits after it, the package), and says when the code states none
or the version could not be confirmed. HBMAME no longer claims a MAME
romset. tools.md and profiling.md document version_basis and the two
modes.
An alias carries the version of the profile it aliases (version_basis
kind alias): mednafen_saturn and mednafen_vb read v1.32.1 and v1.31.0,
the mupen64plus_next builds 2.6.0, flycast_gles2 the flycast tag form.
The libretro-samples test cores without a pin take the commit their
profiled_date names and read library_version there.
core_version is read at each pin where version_basis says. MAME 2014,
2015, MESS 2015 and CDi 2015 state 0.159 in src/version.c (the libretro
wrapper reports 0.160), MAME/MESS sits on the 0.289 makefile of its pin,
FinalBurn Neo states 1.0.0.03 in src/burn/version.h, FB Alpha 2012 and
its CPS-3 build 0.2.97.30. mame_version of MAME 2014 and 2015 follows,
as MESS 2015 and UME 2015 already said.
FBNeo bumps src/burn/version.h without tagging: the scraper wrote the
last tag, v1.0.0.02, while the source said 1.0.0.03. It now reads the
header the way the libretro core prints it. The merge no longer writes
core_version: the hashes come from finalburnneo/FBNeo's tip, which the
libretro/FBNeo pin does not name, so the version stays the one
version_basis reads at that pin. A MAME release profile is written with
its version_basis.
version_basis says where a profile's version is read: a pattern in a
file, the nearest tag and its distance, an alias, a package, none, or
unverified with the reason. profile_sync --verify-version reads it again
at the pin and exits 1 on a difference, --set-version writes what the
code states. A pin advance reads the version at the new pin in the same
write and is refused when it cannot.
upstream.describe counts the commits after the nearest tag in a treeless
clone kept under the cache, and graph_ancestry answers a pair GitHub
refuses to compare.
Ubuntu 24.04 blocks the unprivileged user namespaces bubblewrap needs, so the campaign refused to run the fake cores and four FakeCorePasses tests failed, also inside the no-jsonschema replay. CI now installs bubblewrap, lifts the AppArmor restriction and proves a sandbox starts. Locally the tests name the missing sandbox instead of failing on a KeyError.
One table of roles from the 16 basic colours (accent, success,
warning, failure, muted, strong, selected) read by the full screen and
by the end of a line-mode run. Status words take their role from one
pattern table, the text never changes, and colour is off with NO_COLOR,
TERM dumb or unset, a terminfo entry naming fewer than 8 colours, or an
output that is not a terminal.
Colour goldens keep the escape codes, and a test strips them back to
the text goldens.
--offline given with the variable set is named as the option, and the
runs the screen starts get it only when the user gave it. The first
install page stays under its 400 words, the 3.8 runs skip only the
pack test that needs generate_pack.
The options and environment tables say what --no-input finds and when
it stops, and how the words name RETROBIOS_INSTALL_OFFLINE. A test holds
the options screen to leave out parallel downloads offline.
Long messages scroll with their position, a small window says why the
questions come line by line, one bar rule per width, the folder a
config names sits under it and wraps whole, the help while typing a
filter lists what does not type, the plan names --core and --region,
pc finds the computer setups, rows keep sizes near their names, a
partial result counts once, a quit is said as one, the WSL browser
names drives by letter and No connection names the pack.
A full or read-only card is freed or unlocked, checked again first,
never left for another drive. Line mode names --dest where it said
choose another one, numbers out of range, the folder copied from, the
frontend's save directory and the command after Enter. The words name
RETROBIOS_INSTALL_OFFLINE when it, not --offline, made a run offline.
The installer knew BIOS files from game data and no count showed it.
The plan, the line menu and the result count them apart for a list that
marks game data, the plan gets a game data row and the menu a g key,
each toggling it in one press, and one sentence says what game data is.
A Termux run named /sdcard and /storage/emulated/0 for one storage, read
saves/<core> as a folder, split a backup path and the restore command
inside their words and named no folder for missing game data. Paths are
spelled as the BIOS folder was named, folders per core are said in
words, the result keeps paths and commands whole, and the game data line
names the folder it goes in.
Started with no job-control shell, a Ctrl-Z stopped the screen and its
launcher for good. The key now stops it only when its process group is
a shell's job, and the bar says why otherwise.
The MinUI family's lists record cores and regions and no system: every
narrowing was refused with a remedy that fetched the same list. Each
option is now refused by the axis missing, with what works, and a pack
names its platform's list only when that list records the axis. Line
mode names what c chooses and skips a question with one answer, and
Options leaves out parallel downloads when nothing is downloaded.
RetroArch's folder question numbered a Batocera SHARE drive and filled
its root. A drive is now offered by its own system's BIOS folder, one
nothing names by its bios folder, and another system's is left out and
named with the options that reach it. A list or pack card setups read
too opens on those setups, in both modes.
Abdessamad Derraz