tests/__init__.py keeps a run offline, its scratch under tmp/ and its
installer cache its own, but only discover -s tests -t . imported it.
discover tests ran with RETROBIOS_OFFLINE unset, scratch on the /tmp
tmpfs and the user's real installer cache, the setting in which e2e
fixtures had written into the shared .cache/large. Every test module now
imports the package first, and a test fails on a module that does not.
test_verify_listing.py took its line with the previous commit.
Geargrafx drops a 512-byte header before its crc32 (src/media.cpp:754-773
at 7c447b0b), and its profile lists both sizes. slots compared the crc32
of the whole file, so EmuELEC's headered syscard3.pce read as contested.
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.
Walking the systems sorted reordered the declarations doubled.py
reports from, so the export report named the CoCo reader first. The
declarations keep the platform's order again, and the packed file is
picked from the sorted walk the builder makes.
A collision line named two files and neither the system declaring each
nor the one the pack ships, so the reader went back to the platform
file. find_collisions walks the systems in the builder's sorted order,
takes the builder's preferred declaration, and the line and the JSON
now name each system and the file packed. RetroDECK's bios/disk.rom
reads as the CoCo ROM packed, the PC-88 one left out, as its manifest
holds.
A core with found_by: content reads no file by the destination's name. Any dump its profile accepts for the system settles the slot: Batocera, KNULLI and ROCKNIX ship scph1001.bin v2.0, which DuckStation and rustation accept under scph1001_v20.bin.
architecture.md said the pack and verify decide through slots.arbitrate(),
while a test forbids them to call it: they read slot_choices() through
pack_overrides(). build_claim_index, the name-keyed index the module
docstring described, had no caller outside its own tests.
A keep line gave every emulator the file of the first profile claim:
Recalbox's neocdz.zip line said geolith reads the archive fbneo reads.
The JSON 'expected' lists each file with its emulators (finding 4).
Each content-checked line ended with a verdict on the platform's list,
printed at pipeline step 3b. The line now says which emulators read the
other file, and the closing count says the rest follow the platform's
list. The module source carries no such verdict, a test reads it.
A destination no platform declares went to the first claimant in profile
order whatever the proof: fbalpha2012 asks for hiscore.dat by name and
took the root of six packs over cores that pin it by sha1. Claims on a
contested path are now ranked by what the collection holds for each (a
held declared hash, then a path or every required archive member, then a
name, then a contradicted hash), and profile order only breaks ties.
Measured: 14 slots change file across the platforms, among them
hiscore.dat, kick13.rom on RetroPie, cgb_boot.bin and dmg_boot.bin on
EmuDeck, and ROCKNIX neogeo.zip and neocdz.zip.
Whether the frontend compares content was decided per platform. EmuDeck
(md5) declares neocdz.zip with no hash and checkBIOS.sh never names it,
yet the proven claims of geolith and FBNeo were dropped and the pack
shipped a neocdz.zip without 000-lo.lo. checked_by_path() decides per
entry: no hash of the kind the frontend compares (any of md5, sha1,
crc32 under RomM's rule) and no zipped_file means a path. Measured: the
EmuDeck neocdz.zip slot and three RetroBat DS slots (dsi_bios7.bin,
dsi_bios9.bin, firmware.bin) now take the file the cores prove.
--emulator accepts a buildbot core name from a profile's cores:, and a
mistyped target gets the closest names instead of every target. diff_truth
shows destinations and reports unpaired entries as evidence, not removals.
The verify summary counts discrepancies, a shared destination names both
systems, a truth hash conflict names both cores, and pipeline steps carry
their place in the real run instead of N/8. Every GitHub API caller takes
the token from github_token(), which falls back to gh auth token.
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.
System.dat pins the 2 KiB X1 font as xmil/IPLROM.X1T, where xmil reads
a 32 KiB Turbo IPL. A held copy of the exact size a profile declares now
proves its claim against a file of another size, so the existence packs
ship the IPL. The font's duplicate under the IPL name is removed.
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.
A filename-agnostic profile ships its scan candidates, never its named
entries, and the gap report skips it whole. Its entries no longer
arbitrate a destination: RetroArch keeps the GameIndex.yaml it declares
instead of the ARMSX2 copy.
Abdessamad Derraz