Six places read an md5 shorter than 32 digits, each with its own bound.
digest_verdict accepted any prefix down to one digit, while Batocera
compares the whole string (batocera-systems:1088). A cut value now names
a dump only when one held md5 starts with it (cut_md5_owner, read by the
scraper, the resolver and --from-md5), and every verdict compares whole
digests.
gitignored_paths read a missing file as an empty registry: nothing to
restore, nothing unsatisfied, exit 0 with every release asset absent.
The run now stops with the path it could not read.
RetroPie was the only platform with status: archived, which kept it out
of verify --all, slots, the pipeline and list_platforms unless
--include-archived was typed. release_pack already decides which packs a
release publishes and the installer serves every platform, so the status
set nothing apart. The field, its schema entry and the flag are gone, and
RetroPie is covered by every full run like any other platform.
Every source link on the site leads to a path the source record saw at
a commit. The Build file row linked pin_build files unchecked, eight of
them at commits the record did not hold. build_source_paths now records
the build files read at a commit, and the row links only those: a tag
can move and stays plain text. The record run also picked up two cited
paths it had not seen before (pom2 CMakeLists.txt, pcem configure.ac).
The manifest guard compared repo_path with database.json only, so a
git mv under bios/ committed without regenerating either left both
naming the old path and every test passed. A new test requires each
database path to be tracked or gitignored, and a manifest entry whose
repo_path git neither holds nor ignores now fails as untracked.
264 TemporaryDirectory() and mkdtemp() calls without dir= wrote to the
/tmp tmpfs, and a full tmpfs failed modules no change touched. The test
package now points tempfile and TMPDIR at tmp/tests, and sets
GIT_CEILING_DIRECTORIES there so git run in a scratch directory does
not find the checkout above it. The package test asserts both.
test_a_targeted_export_reads_a_full_model runs pipeline.py --offline
--with-export, which is now refused when .cache/upstream-native/ is
empty, as in any fresh checkout. The test is about the truth directory
the export reads.
fetch_large_file verified a cached copy and returned its name, which another run replaces when the release serves a re-upload: the caller then packed bytes it never verified. The returned path is now a hard link under .cache/large/.by-sha1/<sha1>/, kept only if it holds the inode that was hashed. Revisions no name holds are dropped a day later. restore_large_files hashes each copy before moving it into place.
pin_build, in 36 profiles, names the platform repository, revision, file
and variable that hold a pin_frozen revision, and the site read none of
it: the Revision held row printed the file as plain text. A Build file
row links each one at the revision it was read at.
A .variants/ copy downloads as advision.zip.1c086064, its storage name:
raw.githubusercontent.com sends no Content-Disposition and the download
attribute is ignored across origins. A card whose download name differs
from the file name now says what to save it as, at a glance.
Every card read "Not collectable" and the gaps page listed 259 entries as
acknowledged gaps verified from source code. unsourceable: only keeps a
same-named file from being served in the entry's place, and each entry
stays an acquisition target (wiki/profiling.md). The section now names
the catalog or romset that lists each file, from the provenance report
built with the profiles' needs: XM7's KANJI2.ROM is in No-Intro and in
MAME's fm7740sx.zip. The old anchor still leads to the section.
libmt32emu scans the ROM folder and identifies every file with identifyROMFile (src/midi/mt32_lasynth_model.cpp:68-91 at fa1856c0), and the Nuked SC-55 plugin loads with RomLoader::Hashing, matching SHA-256 against its table. The 50 MT-32 and Sound Canvas entries now carry found_by: content, so the truth no longer says dosbox-staging opens their labels.
The entry documents the pcsx2x6 component, whose FindBiosImage (pcsx2/ps2/BiosTools.cpp:268-291 at 583f0e90) takes the first 4 to 8 MB romdir image whatever its name. pcsx2x6.yml already said so, and the RetroBat truth gave the file two identities. The guard no longer needs an agnostic flag to read an entry.
FindBiosImage (pcsx2/ps2/BiosTools.cpp:278-302 under app/src/main/cpp/pcsx2 at bda532f3) keeps the first 2 to 8 MB file with a romdir, whatever its name, as upstream PCSX2 does. The guard reads entry-level agnostic flags too, which is how it surfaced.
FindArcadeBiosFor (pcsx2/ps2/BiosTools.cpp:366-419 at 7f914418) scans the BIOS folder for 2 to 8 MB COH-H images and picks by EXTINFO serial, never by name. r27v1602f.8g, r27v1602f.7d and COH-H31000.BIN now carry found_by: content, and the truth no longer says armsx2 opens those names. The guard covers the 2 MB lower bound.
Step 5 builds the --required-only, --no-game-data, --region, --target
and --system packs of every platform in memory and compares their paths
with verify run under the same narrowing. verify_platform gains
required_only, generate_pack a record mode. A run samples three regions,
targets and systems per platform (155 packs, about 3 minutes);
--full-narrowings compares all 1,150 (25 minutes). A test fails when
_narrowings() gains a dimension that is neither compared nor exempted.
A platform pack carried only the trees its own systems declared, so for_platforms reached RetroArch, Lakka and RetroPie alone and every other platform running blueMSX, FBNeo or Dolphin got no tree. The pack, the install list and the pipeline's tree members now add the trees of the libretro cores the platform runs, below the system directory and build folder the core reads, unless the platform installs or never opens them. A platform entry the manifest resolved into a carried tree becomes that tree's member, which the installer can fetch: AmberELEC and EmuELEC no longer hold 192 blueMSX machine configs back. stonesoup reads its tree beside the content and pcee2 compiles its resources in, so neither tree reaches a BIOS pack.
The --skip-packs help and the validate_pr fetch error were reworded, and a batocera page with no tab bar is plain Markdown that keeps the underscore entity, read here as a browser shows it.
upstream_commit 25b54e3f was dev's HEAD at the 2026-10-23 bump. dev has
been rewritten since, and the pin is on no branch (diverged 109/114), so
profile_sync reports it orphaned. libretro-pcem shares no history with
sarah-walker-pcem/pcem: it took mainline patches by hand until 2016-04-13
(c9f3593f, 223f434f). Its sound_adlibgold.c holds 10739b96, the last
upstream commit before that date, and the port has none of the 6x86 work
of 2016-04-27. The pin is held there with upstream_frozen.
The pin 87b176a0 and tag v2.0.1 sit on preview (ahead 129, behind 0) and
not on the default xenios branch, which diverged (ahead 1784, behind 24).
The preview-v2.0.1 releases carry the dmg and ipa the notes name. With no
source_branch, profile_sync called the pin orphaned and listed 19 refs to
review that only measured the branch divergence. It now reads 1.
ecwolf.pk3 carried md5 c011b428, which its note said came from the .info.
The code checks no hash (src/wl_iwad.cpp:664-672) and the pk3 is built
from wadsrc/static/ at the pin, so no shipped file is designated by it.
The entry still resolves to the same file by name, and the platform lists
that pin this md5 are untouched.
FS-UAE scans its Kickstarts directory and names a ROM by checksum
(src/fs-uae/config.c:121-140, main.c:431-470), opens kick.rom by name
only as a fallback (memory.cpp:1597-1606) and loads AROS when no ROM is
found (memory.cpp:2374-2379). required: true and the launcher's dialog
came from RetroPie's fs-uae.sh, which retropie.yml launchers: records.
The entry is now optional and found by content. kick13.rom stays as its
label, the name amiberry and uae4all give this 256 KB dump.
find_missing kept undeclared files reached only by bundled entries, so auto_fetch searched for and offered files the emulator installs itself, and a placed copy would fail bundled_only.py --strict. RetroPie drops from 55 to 14 missing files.
--emulator mednafen_psx refused with 'use --emulator beetle_psx' although the profile names its target. The alias now selects that profile and says so on stderr, in verify, generate_pack and cross_reference alike. Launchers are still refused, and pipeline step 5 skips aliases since they compare under the profile they name.
pyproject had no [build-system] and no module list, so setuptools' flat-layout discovery found scripts/ and tests/ and refused to build. Nothing is packaged now and pip install -e ".[dev]" installs the tooling.
With found_by content the openMSX entries read as labels, and two of
them, COLECO.ROM and fmpac.rom, are names other profiles give another
dump. openMSX's own machine XMLs name both (ColecoVision_SGM.xml:56,
Boosted_MSX2_EN.xml:83) and try them beside the XML, so they are code
names, not labels.
Continued if-conditions were cut at their first line, so GLES3 held on every board, and selects guarded by a negation stayed on once fired. Derived flags are read line by line, each pass rebuilds the symbol set from the board, and the architecture comes from the board's BR2_<arch> choice.
Comments, docstrings and test texts ended on tags that name private review notes no clone holds. The tags go, the facts stay. Test methods named after a finding number are named after what they hold. test_prose_style refuses a tag that comes back. install.py changes in comments only: its pin and the launcher index follow.
Abdessamad Derraz