mame.yml filed BBC Micro, Elektronika BK, Enterprise and Vectrex sets under systems its systems: list lacked, so a --system pick of those never reached the profile. The three Astrocade sets were filed under astrocde while mame.yml and groovymame.yml list bally-astrocade, the id a pick of Batocera's astrocade matches. Full packs keep the same files.
naomi.cpp:10918-10921 declares hod2bios, f355dlx, f355bios and airlbios on the naomi machine config as NAOMI BIOS roots, viper.cpp:3213 declares kviper as the Viper BIOS. Every other MAME profile listing them already files them under sega-naomi and konami-viper.
emulatorLauncher writes a BIOS path into an emulator's configuration
only when the file exists: string x = Path.Combine(bios, "name");
if (File.Exists(x)) ini.WriteValue(section, key, x). The extractor now
records each such write as a fixed_name launcher record, with the
enclosing if and early-return conditions, and a key that receives the
bios folder itself as passes_path. A path reassigned before the check,
or a write under a loop or an else, is not read.
Generators read at 48cdc616: KegaFusion (SCD and 32X keys, GenesisBIOS
given the folder), Jgenesis, MelonDS (DS), Mupen64, Project64, mGBA.
The RetroBat paragraph of the Kega Fusion profile now lives in the
launcher layer.
Seven CommonPaths.h refs were written in March against a 2025 layout (bfef65a9..8cbfee21, IPL.bin at 138), and SYSCONF was later set to 118 from the current file, so no revision held all eight. They now cite d8558142, the merge base of shiiion/dolphin@53f53e0f with dolphin-emu master, where each define sits on the cited line, and upstream_commit names it.
EightyOne/Source/ROMs/... answers 404 at charlierobson/EightyOne@4de3cdb8. Source/ROMs/{FDCs/interface1.edition1,spectrum48.arabic,tk90,tk95}.rom hash to the md5 each entry declares. The notes refs drop the clone-directory prefix too and name their repository.
These refs were labelled upstream but land in the port at source_commit: flycast sb_mem.cpp:55-59,196-231, reios.cpp:737 and naomi_roms.h:87-94 in libretro/flycast@bde196bf (filestream_open, get_rom_names), lrps2 BiosTools.cpp:189-229,339-340 and CDVD.cpp:196-245 in libretro/ps2@6eabf224 (RETRO_VFS reads absent from PCSX2), mesence bios.col at FirmwareTypeExtensions.cs:70 in MesenCE@20f497c9 (line 62 in Mesen2).
source_commit 953db38f is the tip of branch groovymame0289, its makefile:1513 writes BARE_BUILD_VERSION "0.289", and the compare with mamedev tag mame0289 is 83 ahead, 0 behind, with no file under src/mame or src/devices. The notes described the earlier pin e3fa1305 (branch groovymame0288).
detect_pinned_tag matched core_version against every view, so an upstream_commit set to the release tag a port was cut from (triforce on Dolphin 2407, dosbox_pure on 1.0-preview7, GroovyMAME on mame0289) froze the whole profile: the port was no longer compared with its HEAD and bump_commit refused. Only the documented build's repository freezes the profile now. An upstream on its tag is self-checked at that tag, is not reported orphaned, and a bump leaves upstream_commit on it.
flycast now reads naomi2.zip, which the RetroArch list RetroPie inherits
already declares. The render added it to the flycast help sentences but
landed_changes counted additions only for entries the platform layer
lacks, so the report missed a name the diff carries. names_anew lets an
exporter whose unit is not the entry say so.
The download page had no Android line and no Windows form with
options, and platform and system pages offered only ZIP parts. One
generator now writes the Linux, Windows and Android commands for the
download page and the platform, emulator and system pages. A system
section lists each platform's command under the system id its install
manifest records, the only names --system accepts.
dosbox_pure takes the upstream tag its core_version names (1.0-preview7), triforce the Dolphin release its tree reports (2407). The others take the upstream commit at profiled_date, the revision profile_sync already compared them against: the standalone build for standalone + libretro profiles, and for cannonball and jl-mod the cited lines land there and not at the fork's merge base. Every upstream-role ref lands on its declared value at the pin.
The which-pack page called RetroPie the largest community with the most
guides, Batocera an easy setup and Recalbox good for a first build,
claims nothing in the repository measures. Each row now says what the
setup is, with board names taken from the platform target lists.
The page opened on a slogan and four tiles that read Verified (100%)
for 5,437 of 5,438 files and Missing 0, above a table of red cells. It
now opens on what it covers and one sentence that says every count of
the platforms' own lists and points to the emulator files below.
An emulator page linked its YAML and api/v1 JSON, platform and system
pages nothing. Each platform page now links its YAML, its own
api/v1/platforms/<id>.json and its install manifest, and each system
page its api/v1/systems/<slug>.json, both validated by the envelope
schema.
np2pi reads 0x4000 bytes of scsi.rom as the SCSI BIOS (cbus/scsiio.c:
208-233). The only 16 KB copy of that name is Clock Signal's Apple II
card ROM, which the name step served and the RetroPie manifest installed.
The entry now carries unsourceable: and np2pi falls back on its stub.
NanoBoyAdvance opens general.bios_path, default bios.bin (src/platform/core/src/config.cpp:36 at 3bb6f478). gba_bios.bin is the path ROCKNIX's config.toml sets, recorded under launchers: in rocknix.yml.
Without the alias the entry met only PCem's and QEMU's bios.bin, 64 and 128 KB, which BIOSLoader refuses (bios.cpp:17-48). verify counted it held by name, so the launcher layer left it alone and the ROCKNIX pack gained PCem's image. A hashless entry whose every same-named copy fails its size is now not held, and the gba_bios.bin ROCKNIX declares settles it.
_holding_members kept every same-named copy when the best held nothing,
so galgames.zip, a lone assembler listing, resolved by name for mame,
groovymame and mame4droid-2024, and awbios.zip for mame4droid. Such a
copy now resolves as hash_mismatch.
mame2003_plus declared its contents by name only, while fileio.c:1031-1040
loads a member by CRC when the name is absent. The 61 members now carry
the CRC their driver declares, so stvbios, megatech and the six ZN sets
keep matching the archives the collection holds.
Dolphin and its forks map the Brazilian disc country code B, and the
SYSCONF countries of the Americas, to NTSC-U (DiscIO/Enums.cpp), whose
boot ROM is read from GC/USA. Tagged north-america alone, the IPL a
Brazilian disc boots left a --region brazil,europe pack. GC/USA/IPL.bin
in six profiles and triforce's IPL_TRI.bin now carry latin-america.
validate_pr.py built its folder advice from platform lists alone, so a
file only an emulator profile declares got the generic
bios/Manufacturer/Console/ advice. Profiles now feed the same table
through each entry's system, and the contributing guide says the
command answers where a file goes.
An emulator card's anchor changed when a second entry with the same
path was added, so cited links landed at the page top. A card whose
entry declares a hash is now addressed by it, and the first card of a
path keeps the plain file-<path> id as an alias.
upstream_commit is the merge base the compare API gives between each port at its source_commit and the upstream default branch. holani and melonds_ds take the revision their build fetches (tag 0.9.9, and the melonDS branch whose merge base is 906e9ebb). The upstream-role refs land on their declared values there, except mesence bios.col, whose line 70 is in MesenCE's own FirmwareTypeExtensions.cs.
jp, us, kr and gb were aliases but au was unknown, and the first
spelling offered for it was auto, which means world. Every territory
of the region tree now answers to its ISO 3166-1 code, in the build
scripts and in the installer, whose SHA-256 is pinned again.
verify, generate_pack and cross_reference took -p and -t, slots, generate_truth, diff_truth, export_native and refresh_data_dirs refused them. A test reads every add_argument call.
A fork's clone has origin alone. Pushed to the fork's main, the commit
left origin/main...HEAD empty and --changed reported nothing with rc 0.
When no remote names the project, its main is fetched and compared, and
a fetch that fails is an error.
RetroArch defaults system_directory to :\system, the folder of retroarch.exe (platform_win32.c), for the installer as for the portable archive. %APPDATA%\RetroArch holds only a fallback retroarch.cfg.
The pack README's one-line install was written per platform: ROCKNIX and MiSTer had none, EmuDeck's Windows edition was told to open Konsole, and RomM users were sent to library/bios/ whatever their firmware template. The automatic section now follows each platform's install.detect systems, the EmuDeck and RomM manual steps name every layout, and the download page lists EmuDeck under Windows and macOS.
_foreign_tree compared owner tokens by equality, so cxbx-reloaded-2023
read bios/Other/cxbx-reloaded/ as another emulator's tree and the
case-folded fallback served xemu's encrypted EEPROM. The leading whole
words of a profile key now name its own tree too. Across every profile
entry this moves cxbx-reloaded-2023's EEPROM.bin to the decrypted dump,
turns eight name matches on the same files into path matches, and gives
solarus-2.0.0 the Solarus tree's gamecontrollerdb.txt.
Without jsonschema the step printed only the missing module, and the run then read COMPLETE while PR validation runs the same check. The skip now names the install command and the PR check.
A core that loads no file no longer promises paths and hashes in its
snippet, a description is cut at a word and keeps the underscores of
its code spans, and the JSON-LD url is encoded as mkdocs encodes the
canonical link and the sitemap.
The Batocera example is dated and carries the Not installable line and the closing list of files the collection does not hold, which every Batocera run prints. Disk sizes are no longer hand-kept: the run prints its size. The no-platform case describes the menu install.sh makes possible by handing over the terminal.
The pack integrity check rebuilt the core extras with no taken paths and
tested presence alone, so it credited core files at paths the platform's
own file holds and printed a core count the builder contradicted. It now
replays the builder's placement through packextras.hidden_extras and
compares each placed member with the copy the builder resolves.
archive_members scanned the whole file list on every call, and callers ask once per entry of the same profile: 38 million entries walked in one verify --all over mame.yml. Lists of 64 entries or more are now indexed by archive once (held so the id is not reused, indexed again when the length changes), members still read from the entries at each call. archive_members drops from 2.30 s to 0.14 s tottime under the profiler. verify --all (text and --json), slots --json and the retroarch manifest are byte-identical.
Run from another directory, validate_pr.py read no database, platform or
profile and reported every file as referenced by no platform and outside
bios/. It now moves to the root as verify.py does, reading the files
named from where they were typed.
The platform's systems are known once its config is loaded. The filter ran inside the build, after the output lock and a refresh of every data directory, and ended on a WARNING with exit 1. It is now a usage error before any work.
Without Python, install.ps1 at a console now offers to run winget install Python.Python.3.12, rereads PATH from the registry and carries on. The Store's python.exe placeholder fails the version check like a missing Python.
A browser upload into bios/ lands on a fork and opens the pull request, and CI already hashes the file and names its folder. CONTRIBUTING asked for a fork, a clone, five packages and the whole suite, and promised a PR comment that a fork's read-only token cannot post. The issue chooser links the upload page.
Abdessamad Derraz