generate_pack split --platform on commas to validate it, then used
the whole string as one platform and crashed. Every script takes a
single platform; a list is refused with a message pointing at --all.
Only a .variants/ copy matched IPLROM.X1 exactly, so x1 got
Recalbox's second md5 while iplrom.x1, the dump every platform
pins, sat beside it. When every exact-case match is a variant, a
primary in the same directory under its own name joins the
candidates. Another machine's font.rom or an alias pointing at a
different file does not.
slots dropped the last extension of any .variants/ name, so a
variant stored under its plain name (63 of them, aes.zip among
them) was not the family of its primary and a platform declaring
both at one path would read as a collision. The database already
stripped an eight-hex suffix only; both now read one function.
The name step only ranked archive copies by members declared
through archive sub-entries, and the contents an archive entry lists
never reached the resolver, so the first copy indexed won. contents
now ranks copies by crc32, or by member name where the code opens
by name, and the report and both extra passes carry it. With no
layout declared, a primary holding every ROM of the others wins;
two revisions keep the index order.
A data directory copy found by name was returned without the size and
Adler-32 checks the name steps over bios/ apply. The verdict now runs
both, for the cache walk, the foreign-tree fallback and release assets.
Dolphin and its forks hash dsp_rom.bin and dsp_coef.bin together and
warn on any combination outside known_roms. The profiles now carry the
accepted pairs, and the resolver keeps both files inside the first pair
the collection holds whole: MMJR and Ishiiruka were served the free ROM
beside the official coefficients.
The declared-names test now builds its own platform: Batocera files the BK under elektronika-bk since its slug map was completed, so the real data no longer shows two ids for one name.
Both run on Linux but inherited RetroArch's case folding, so a core
file differing from a platform name only by case left their packs:
MesenCE's bios.col and np2kai's lowercase 2608_*.wav, which those
cores open with that exact case. They are now case-sensitive, the
shared groups fold only where the platform does, and Lakka gets its
own pack instead of sharing RetroArch's.
lrps2 names no hash for its GameIndex.yaml, and the only full-tail
copy was ARMSX2's: it counted as path proof and replaced the file
RetroArch pins on desktop targets, which have no ARMSX2. Upstream
PCSX2 took ARMSX2's patches.zip by name the same way. A copy in
another emulator's tree now proves nothing at any tail length, and
LRPS2's own GameIndex comes from the tree the buildbot publishes.
Recalbox compiles libretro/mame at 0.258 and 0.278 itself; typed
libretro, both profiles were resolved for every all_libretro platform
although the buildbot RetroArch installs from has neither. A profile
now names the platforms that build it (built_by), and is reached only
when a platform lists it by name.
Abdessamad Derraz