A resolve-check on HANDOFF's own rows. Q8 read "SE audio is undecodable
from the disc -- no XACT container exists anywhere". menu-audio-cues.md
retracted exactly that ("### Retracting 'cannot be extracted'") and
locates three cues in Static.slb that decode to PCM: d-pad move 0x1ec0
(4 packets), (B) back 0x0ec0 (2), (A) confirm 0x5d6c0 (6), all mono
48 kHz. The retraction landed in docs/re/ and the page the port reads
kept the superseded text -- the fourth time in this corpus.
Writing the rule down has not worked, so there is a tool now.
handoff_lint.py flags every HANDOFF line making a strong negative claim
that links a doc containing retraction language. First run: found the Q8
row, plus one benign false positive (Q3 links a doc whose retraction is
about a sprite count, not about the tie-break -- checked, and HANDOFF
repeats none of the retracted figures). The lint also caught its own bug
first: it reported existing docs as missing because it joined a guessed
repo root, so it now resolves links relative to the file as markdown does.
Separately, EXTRAS's paint-order risk narrows twice more. Of its 15 tied
pairs only 2 overlap, and of those, ptloop01 x ptloop02 are loop*
animations compose skips by default -- so exactly ONE tie can be drawn:
ptframe3 x ptframe4, overlapping 102x132 px. Against live-extras.png that
contested region correlates +0.9622, better than the whole frame (+0.9440)
and inside the range of regions where order cannot matter (+0.8502 /
+0.9903). Consistent with our order, not proof: correlation cannot see a
swap between locally similar art.
15 -> 2 -> 1 -> consistent is now the whole paint-order risk on the five
screens, and HANDOFF says so.
`compose` claimed the derived paint order "reproduces both measured
orders up to ties". That sentence was never measured and was stale by
one: there are three measured orders, not two. examples/paint_order_audit.rs
checks it.
main menu (entries 5, 8) derived == measured 0 inverted pairs
developer splash (11, 14) derived == measured 0 inverted pairs
title (entry 4) DIFFERS 8, all same-key ties
So the claim holds and the exception is entirely ties -- but two of those
ties are total occlusions, not near-misses. The tied family is the five
ptlogo_back2eff glows (key 32899); back2eff5 is 1133x280 and FULLY
CONTAINS back2eff3 (82,824 px^2 = 100% of the smaller) and back2eff4
(152,047 px^2 = 100%). Derived paints it on top of two glows it entirely
covers; the game paints it underneath. A tie-break by declaration index
can therefore be wrong by a whole layer. The title itself is unaffected --
it has a measured order.
The port's actual exposure, per screen: title, main menu and developer
splash all use MEASURED orders; the publisher splash is derived but has
ZERO ties, so it is fully determined; EXTRAS is derived with 15 tied
pairs of which only 2 OVERLAP. Two element pairs on one screen is the
whole risk, and that is what HANDOFF now says -- not the raw 15, which
would have overstated it 7x.
Reach stated: this compares the derived order against orders measured
from the game, not an independent derivation, so where no measured order
exists only the tie exposure can be checked. Overlap uses pivot*2 as the
element size at its resting placement.
Stale comment in compose corrected. METHOD: a hedge in a code comment is
an unmeasured claim; and count the cases that can bite, not the ones that
match the pattern.
blit() and fill_quad() both opened with
let sx_pct = if kf.scale_x == 0 { 100 } else { kf.scale_x };
so an element whose pose is collapsed to nothing rendered at FULL SIZE.
I first described this as "a 1-pixel sliver", reading the .max(1) in the
size arithmetic two lines below -- the guard above it meant .max(1) never
saw a zero. Read the whole function, not the lines you went looking for.
Control run before the change: if 0 meant "unset", something on the disc
would use it throughout. Nothing does.
elements with a keyframe group 15 493
at least one keyframe at scale 0 2 166
EVERY keyframe at scale 0 0
zero is a transient the element grows out of 1 762
ptlogo_eff3.t32 runs 0% -> 200%. An "unset" marker no element ever uses
throughout is not a marker.
Both functions now return without drawing when either scale is 0.
Reach: 24 of 24 renders byte-identical across GP_TITLE (all 16 builds),
GP_PAUSE_MENU and GP_OPTIONS -- additive on every screen the port needs.
195 elements have a guessed rest pose at scale 0 and 126 of those have a
non-zero alpha, so the old code painted them; all 126 are in
GP_READY_ROOM.pak, which S1 already declared a no-go. They are
tactical-map sprites at sx=0/sy=100, mid-horizontal-wipe, that the
coercion drew full-width.
So: a real correctness fix with no visible effect on the menu port.
Suite green, 122 passed / 0 failed across 3 suites at commit time.
Follow-up on last iteration's unadopted candidate (+36 holds the NEXT
pose's time, not its own). Two new results, pointing opposite ways, and
both are reported.
FOR, and calibration-free: the observed full-alpha hold : fade-out ratio
on palogo_gamearts is 83 : 13 frames = 6.38. The shifted reading predicts
8.00. The current reading predicts 0.25 -- off by 26x. With the glow's 2
units/frame fixed and nothing else free, the current reading says the
logo holds full alpha for 2.0 frames; the capture holds it for 83. This
is no longer the shape argument the candidate rested on.
Also for: rest()'s plain max-dwell fallback picks a=0 -- a transparent
pose, for a publisher logo -- under the current reading, and the visible
a=255 hold under the shift. Only the rest_plateau special case rescues
the render today, and that is the case the port agent reported a bug in.
AGAINST: rendering every build of six UI paks under both readings, 10 of
11 compared are byte-identical and one changes -- GP_TITLE build 7, the
Japanese twin of build 4, by 13.1% of pixels. Build 4, the one verified
against a live capture, is unchanged either way, so the single build the
shift moves is the one with no capture to adjudicate it. The proxy goes
against the shift: language twins are the same artwork, and build 7 reads
70.94 mean luminance as decoded against build 4's 71.41, but 76.32
shifted. Correlation does not separate them (0.6206 vs 0.6201).
These constrain different things -- timing versus pose selection -- and
rest() is a heuristic layered on the times, so moving the times moves its
tie-breaks. Adopting the shift means revisiting that heuristic in the
same change, with no build-7 capture to verify against.
Default UNCHANGED. Experiment reachable via SYLPHEED_KF_TIME_SHIFT=1.
Shifted-mode suite: 122 passed, 0 failed across 3 suites at commit time.
The rotated quads on the title screen come from the keyframe block after
all. The earlier negative -- "every GP_TITLE build 4 element has all three
angle words at zero" -- read the right bytes over too small a region: it
walked the top-level declaration table, and the rotated elements are the
nested leaf records ptloop01.rat / ptloop02.rat.
Confirmed against the framebuffer rather than against our own renderer.
The two records declare +12 = 30 and -45; the GPU capture submits their
quads at +30.26 and -45.28 degrees -- magnitude and sign, two different
values. Corroborated by shape in GP_BUNK 117ca14f, where +12 ramps
0 -> 360 with position, scale and alpha constant: a spin in place.
Identifying which draw it was needed edge lengths, not bounding boxes:
400x1076 and 400x1444 against pteff03/pteff03a 399x180 at the elements'
two different declared scales, 600% (1080) and 800% (1440). The same test
names three known-positives in the capture (ptlogo1, ptcopyright,
ptbtn00), so it passes its own control.
Keyframe gains rotation_deg plus unknown_4/unknown_8, carried rather than
dropped. NOT rendered -- ui_layout::blit is axis-aligned only, so the
reference renderer and the port will both draw these upright until a
rotating blit exists.
The census tool ships with the trap that broke its first version: nested
RATC blobs are not 4-byte aligned, so an aligned scan found 0/3 of its
own control blocks and missed 16 341 blocks. Disc-wide +12 is non-zero in
14.50 % of 83 862 blocks.
sylpheed-formats tests, SYLPHEED_DISC set: 131 passed, 0 failed across the
6 suites finished at commit time; the run had not yet completed.
title's rotation
Looking for where the rotated quads come from, the obvious candidates
were the three keyframe words at +4, +8 and +12 that ui_layout.rs
documents as zero.
They are not zero. Across 72287 keyframe blocks disc-wide they are
non-zero in 4.81, 4.56 and 15.82 percent of blocks, and read as signed
values clustering on 180, -180, 90, -90, 120 and 22 -- degrees. Three of
them, so plausibly rotation about three axes. I have marked that amber
because it is the shape of the numbers and nothing more; no observed
rotation has been tied to a value. The doc comment is corrected
regardless: "0 on every frame seen" was an artefact of the sample.
And they do not explain the screen I was chasing. Every element of
GP_TITLE build 4 has all three at zero, element by element, while the
game demonstrably submits rotated parallelograms there. So the title's
rotation comes from outside the keyframe data and is still unidentified.
One correction to my own last write-up, flagged rather than left: I
stated that the skewed draw IS the swoosh. It is the only skewed geometry
in the capture and the swoosh is the only diagonal element on the screen,
so the inference is reasonable -- but I never confirmed it by matching
the draw's texture or screen position to that element, and I should have
said so the first time.
residual is smaller than I said
Testing the candidate I raised last iteration rather than carrying it.
Blending bit-0x02 sprites additively moves every measure the wrong way --
whole-frame mean diff +0.55 to +1.04, swoosh-band mean +1.83 to +3.98,
band edge-correlation 0.6971 down to 0.5578. So the bit is real and
independent but does not select an additive blend. I reverted the
experiment and kept the word as T8adImage::flags, documented and not
acted on; the render is byte-identical to before.
Second refutation: the swoosh is not displaced. Shifting the band over
plus or minus 80 by 8 pixels peaks sharply at zero, 0.7342, falling to
0.22 at 24 px. So the pivot story is dead twice over -- inert at scale
100, and no displacement to explain anyway.
And I have restated the residual, because earlier sections overstated it.
The +16 to +34 band tiles I quoted were measured WITHOUT --primitives.
With the dim drawn the band's average is nearly right at +1.83; what is
wrong is its structure, tiles running -38.6 then +33.8 and cancelling.
Six candidates eliminated now and none confirmed.
One caveat I owe the port agent about the capture I gave them: it is at
t=4.0s, roughly 174 keyframe units into a screen whose elements have
keyframes out to t=600. I judged "settled" from mean luminance, which
cannot see a thin sprite still moving. It is settled for the bulk of the
screen and not proven settled for every element -- which is a live
alternative explanation for a structural difference in exactly the band
the sweeps cross.
METHOD: cargo build passing does not mean cargo test compiles. Adding the
field built the library in 1.48s and broke two test-only struct literals;
cargo test failed with exit 101.
VISIBLE, not when its pose repeats
The port agent's report was right about the defect and about which
elements it hits, and its proposed condition does not survive the case
the exclusion was written for.
Their test -- the final untimed keyframe has the same pose as the last
timed one -- is also true of pgptitle.rat, whose trailing run is two
identical transparent frames. Adopting it would erase the word PAUSE
again, which a committed capture disproves. What separates the two is
alpha: an exit fades the element out so its last keyframe is transparent,
while an element with no exit ends on the pose you can see. So
rest_plateau now accepts a trailing run exactly when it is visible.
Verified against a capture rather than against another renderer, which is
the point their own report made: ptframe1's rest moves from (620,108)
t=16 to (440,108) t=62; the changed pixels are 10082 in a bounding box of
x 440-839, y 108-577, which is exactly the 400x470 at (440,108) they
predicted; and correlation against main-menu-oracle.png over that region
improves from 0.9596 to 0.9748. The PAUSE wordmark is unchanged across
all three pause builds.
This also closes a question ui-paint-order-key.md has carried for a
while, that ptframe1 and ptframe2 rest at alpha 0 while the capture shows
the menu frame plainly.
One trap cost most of this iteration and goes in METHOD: CARGO_TARGET_DIR
is redirected in this container, so ./target/debug/sylpheed-cli is hours
stale and every render I made against it was old code. Byte-identical
before and after reads as "no effect" when it actually means "you ran the
old binary".
The full disc-gated test suite was still executing when this was
committed; the verification above is artifact-based.
The trickiest reading on the disc lived in the Bevy viewer: resolving a
cutscene's voice to a continuous byte REGION of the sound stream, because the
movie voices are one XMA stream chunked into VOICE_*.slb entries whose
boundaries do not match the cues -- a cue routinely spans two chunks, so a .slb
need not hold the track its name claims.
That put the logic most likely to be re-derived incorrectly in the crate least
likely to be reused. The Godot port's exporter needs the same answers, and there
must be one implementation of them.
New `sylpheed_formats::media` owns every case where the bytes of one playable
thing are not one archive entry: segment-spanning reads, multi-sub-wave banks,
and the voice-region resolution. Callers supply bytes through a `DiscSource`
trait, so the viewer keeps its ISO/directory abstraction and a headless consumer
gets `DirectorySource` for free.
The seam is deliberate: this module returns XMA RIFFs, not PCM. Decoding means
shelling out to FFmpeg, which is native-only and a policy decision for the
consumer -- everything up to "here are the bytes that belong together" is disc
knowledge, everything after it is a codec choice.
The four moved functions were previously untested; `tests/media_disc.rs` now
pins them, including the negative the corpus paid for -- an unbound movie must
stay unvoiced rather than borrow a neighbour's clip, which was tried and played
the WRONG recording.
The algorithm is unchanged, moved verbatim (same window sizes, same fallbacks).
The new disc tests pass; the broader audio suite was not re-run in this pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`movie_manifest` has been parsed since the movie-voice work and rendered
nowhere: it resolved a voice bank and that was all. So the only way to find a
cutscene was to hunt `.wmv` files in the ISO tree, where nothing tells you which
mission a file belongs to, whether it has subtitles, or what is said in it.
View ▸ Cutscenes lists all 104 manifest slots with mission/phase, kind, movie,
subtitle track, voice token and telop, and -- the part that needed no new
parsing, only a route -- resolves the captions to a readable TRANSCRIPT with a
language selector. Subtitles were previously burned into the video during
playback and reachable no other way.
Three negatives are shown rather than smoothed over:
* 5 manifest-bound movies have no `.wmv` (logo1-4 and an encoder test clip).
They are marked and get no Play button instead of one that would fail.
* 9 of 101 movies resolve no English transcript.
* the `.prt` telop overlay is named by the manifest and we have no parser, so
the reference is shown labelled "not decoded" rather than omitted.
`cutscene_catalog_binds_movies_and_transcripts` pins all of it against the disc
-- 104/101/99/99/22, the exact absent-movie list, 92 transcripts -- because a
browser that quietly dropped these would look complete and be wrong. The counts
independently reproduce docs/re/movie-subtitle-link.md.
Play routes through the normal FileSelected path, so the existing video player
handles it exactly as it would from the tree.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The library enumerator kept only names containing VOICE or \Briefing\, and read
eng\sounds.tbl unconditionally. So the Explorer could reach 4382 of the 9519
banks in sound.pak: no music, no jingles, no sound effects, and no Japanese
voice at all -- roughly half the disc's audio had no route to the UI.
`slb::list_audio_entries` now returns every named bank with the category its
path implies (Music / Jingles / Sound effects / Radio / Dialogue / Movie voice /
Briefing). `list_voice_clips` is that, restricted to the spoken categories, so
its existing test still guards the old behaviour. The 36 root banks carry no
language component and appear whichever table is read; the window gets an
English/Japanese switch that re-reads the other sounds.tbl, since the table name
IS the selector.
Two defects the decode found, both recorded in
docs/re/structures/sound-pak-contents.md:
* `Static.slb` -- the SFX bank -- declares 616768 bytes more than sound.p04
holds. Not our extraction: p04 matches the ISO's own directory record, and a
sweep of every pak on the disc finds this one entry over-running and no other.
It is the highest-offset entry, so its comp_size is an allocation size. A
short read is now allowed for the tail entry ONLY; any other overrun stays an
error, because clamping it would hide real damage behind a half-decoded asset.
The bank went from unreadable to 514 s of audio.
* the left-channel downmix was applied to everything. Right for voice (mono
content however stored), wrong for music (a real stereo mix, half of it
discarded). The caller now decides from the category.
35 of the 36 shared banks decode; JNGL_001 does not, and says so in the player
instead of the panel silently closing. Its payload is not a whole number of XMA1
packets from any known data offset, so it is likely not a plain headerless
stream -- written up rather than papered over.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The UI Screens window enumerated with `is_build` while the PAK browser's
inline preview composes anything `parse_build` accepts. So the browser drew
screens this window flatly refused to list -- most visibly `palogo`, the
publisher splash, which declares its sprites directly and has no `.rat`
layout child.
`compose_screen` now gates on `is_composable`, the documented superset (every
`is_build` bundle passes it), so anything the list offers is drawable and the
two presenters share one rule. A "Fragments" toggle widens the enumeration to
it as well, off by default: the extra ~1786 bundles are mostly two-element
fragments (a button beside its glow) that would bury the real screens. It
re-scans, which renumbers the display ordinal -- harmless, because the pak
ENTRY index is the locator and the stale catalog is cleared.
The toggle sets a `rescan` flag rather than self-sending RequestScreenCatalog:
a system that both reads and writes one event type is a B0002 panic at
startup, which is how the Save browser broke.
Also plumbs the last two ComposeOptions the CLI had and the UI did not --
`black backdrop` (what a framebuffer capture must be compared against) and
`primitives` (decoded, but paint order unsolved; hover text says so).
Verified: `screen render --all --build 13 GP_TITLE.pak` draws the SQUARE ENIX
splash, 2/3 elements, the third being the .prm quad. Workspace builds, viewer
reaches steady state, disc-gated suite 20/20 green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three compounding causes, found by tracing every place a `loading` flag is set
against every place it is cleared.
THE COST. `compose_screen` inflated EVERY entry in the pak and held every build
in memory at once, purely to index into the result -- then did it again on every
checkbox click, re-opening the ISO from scratch each time. But the catalog
already records each build's pak ENTRY INDEX, so the locator was there all
along: compose now reads exactly one entry. `RequestScreenCompose` carries
`entry` (the locator) beside `build` (the display ordinal).
`build_screen_catalog` had no budget at all, while the pak browser has capped
exactly this work since it was written. It now skips oversized entries and stops
at a ceiling.
That ceiling is 384 MB, not the pak browser's 64 MB, and the difference is the
point: GP_HANGAR_ARSENAL inflates past 160 MB and holds ~390 builds, so a 64 MB
cap would have quietly hidden most of them -- trading a hang for a wrong answer.
When the ceiling IS hit the pak is marked `truncated` and the UI says so.
THE LATCH. `poll_loader_channel` treated a disconnected channel exactly like an
empty one, so if a worker died every in-flight spinner stayed up for the life of
the process -- and the `if loading { return }` guard at the top of each handler
then refused every retry. Disconnect now clears the flags and reports it.
Verified: the workspace builds, the viewer reaches steady state, and one
composite still draws 11/11 elements of the tutorial pause menu.
A wave runs to data_at + declared_size, and there is seek magic exactly there
whose little-endian packet count at +12 times 2048 equals the declared size --
7620/7620 disc-wide. That is the decoder-independent boundary and the thing that
proves the declared sizes honest, so it should not be able to regress silently.
The boundary routinely lies outside the entry's own comp_size window, so reading
it needs the flat segment stream rather than the entry slice; PakArchive gains a
small data_at(offset, len) accessor for that.
Test walks a bounded slice to stay fast -- 792 banks in this run, all holding.
8 disc tests pass.
I read the seek chunk's packet count big-endian; it is little-endian at seek+12,
with size == 8 + 4*count. And a seek sits immediately AFTER its own data, so an
entry's first seek usually belongs to the PREVIOUS bank (implied start -25232
for D_452, -145988 for TCAF_608). I was comparing an entry's first seek against
its first data -- different waves by construction, which is why no reading lined
up.
With that fixed, the declared sizes are honest: every RIFF-bearing entry on the
disc has seek magic at exactly data_at + declared_size with count*2048 ==
declared. 7620/7620, zero failures.
VOICE_TCAF_608 is not truncated. Its Channels is 2 and I decoded it as mono;
read as stereo it gives 6520176 bytes = 33.96 s, agreeing with both length
signals in the bank (33.88 s from cumulative samples, 33.97 s from
PsuedoBytesPerSec). 170 of 8021 banks (2.12%) are stereo -- exactly the rate of
my 1-in-60 outlier.
This is the mono/stereo trap already documented on this very page, met from the
other direction: I had written 'at two channels every bank yields one frame' and
then spent several passes blaming missing data for a one-frame decode.
Code fix: to_xma_riffs built the leading segment with a hard-wired mono fmt. It
now reads Channels from the bank's first RIFF. 7 disc tests pass.
The ties needed a different signal, not a longer scan. Banks carry one: a seek
chunk sitting on a packet boundary, so seek_pos % 2048 IS the data offset. On
the 6033 labelled banks with a seek before their first RIFF, 6031 agree
(99.97%) -- better than the packet scan and structural rather than statistical,
so scan_data_offset now tries it first.
On the scan's 28 ties it resolves 26 correctly and 0 wrongly (2 have no usable
seek). Combined rule scores 7354/7358 = 99.95%, up from 99.62%. 762 of the 1495
RIFF-less banks carry a seek, so the signal exists where it is needed.
Also ruled out, since a wrong offset was this page's whole subject: the header
is not audio being discarded. Adding 0 to the candidate set, it wins 6 of 7358.
7 disc tests pass.
Two comments claimed it is 'honest per sub-wave'. Measured: 5296 of 7586 banks
declare more than the entry holds and none declares exactly what it holds, so
the existing .min(slb.len()) clamp is load-bearing rather than defensive.
Comment-only change; 7 disc tests still pass.
1495 banks carry no RIFF and take a separate path that hardcoded both the
offset and stereo. Across a random 48-bank sample there was NOT ONE where the
old stereo-at-1392 pair beat the best mono offset; median gain 184x, individual
banks going from 0-4816 decoded bytes to 180000-380000. Stereo shows the same
stop-after-one-frame signature already recorded for the leading segment.
With no RIFF the offset cannot be derived, so scan_data_offset picks among the
four disc offsets by XMA1 packet-header plausibility. Validated on the LABELLED
set -- all 7358 banks that do have a RIFF, where the answer is forced: 7330
correct (99.62%), and all 28 misses are ties on the top score, never a wrong
unique winner. Ties fall back to 1392.
The winning offsets also reproduce, by directory, the distribution measured
independently from the RIFF-bearing banks. jpn\etc splits 1468/1600, so path
alone is not sufficient -- which is why this is a scan and not a lookup table.
7 disc tests pass (build-reborn test -p sylpheed-formats --test
slb_leading_segment_disc, SYLPHEED_DISC wired up).
HEADERLESS_DATA_OFFSET is the value the offset takes in <lang>\etc\, not a
property of the format. The leading stream is a whole number of 2048-byte XMA1
packets ending at the first RIFF, so its start is first_riff % XMA1_PACKET.
Disc-wide that takes four values -- 1392, 1468, 1600, 1728 -- varying by
language and subdirectory.
Verified by decoding, not by arithmetic: on a random 140-bank sample with a
non-empty leading region, the derived offset yields more audio in 85, identical
in 54 (the eng\etc controls, where it must and does reproduce the old
behaviour) and less in 1. Median gain among the improved is 70x --
eng\Voice\VOICE_TCAF_592 goes 1506 -> 97152 bytes, jpn 2910 -> 127178.
This withdraws my own claim from earlier today that the Japanese banks were a
different undecoded layout. They are the same format with a different offset;
I had treated a constant derived from one subdirectory as a property of the
format. The same error was hiding the identical defect in 1873 eng\Voice banks.
ixud.rs now has an IdxdObject-shaped reader, IxudObject, and build_caption_text
reads captions as FIELDS instead of pairing them with whatever token follows in
the pool.
build_demo_text token adjacency 134 ids 537 lines
build_caption_text token adjacency 3721 8074
build_caption_text record fields 4085 8800 = all of them
Verified over the whole disc by tests/ixud_records_disc.rs: 1104/1104 objects
parse, 1476/1476 records and 628165/628165 named fields reproduce their
ixud_hash, 48 positional, zero failures. The header word at 0x08 is record 0's
hash, asserted per object -- there is no schema field, exactly as for IDXD. The
module doc described a 12-byte record directory and a "schema/type hash"; both
were wrong and are corrected.
I also have to correct my own number from the previous commit. "1.3% of the
game's text" counted OCCURRENCES: each family lives in 24-45 IXUD blocks and
the same key repeats across them. Distinct text-bearing MSG_* keys number 8800,
not 44579, and every one has the <id>_<page>_<line> shape. So the real coverage
was 537/8800 = 6.1%, and I overstated the gap about fivefold. Direction right,
magnitude wrong.
The DEMO control is the sharpest evidence for the change: token adjacency finds
537 lines there, the field reader 541. It was dropping lines even in the one
family it was written for -- which is why the test now asserts "must not lose
lines" rather than "must be identical".
Same lesson twice in one session: pool adjacency is a consequence of how
records are written, not a rule of the format.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
build_caption_text generalises the key parser from MSG_DEMO_* to all eight
families. The shapes are uniform and each family is 100% consistent with its
own: seven use MSG_<FAM>_<id>_<page>_<line>, and VOICE alone inserts a family
letter before the id.
ids lines
build_demo_text 134 537
build_caption_text 3721 8074
The DEMO family comes out identical through both readers -- 537 lines either
way -- which is the control that generalising changed nothing that already
worked. Pinned by tests/caption_families_disc.rs, along with VOICE ids keeping
their family letter.
But this does NOT close the gap, and the write-up says so: 8074 against the
44579 text-bearing fields the record-level scan counts is about 18%.
The reason is the same lesson this session already learned once.
build_caption_text pairs a value with the key that happens to follow it in the
raw UTF-16 token stream -- the adjacency heuristic that was wrong for IDXD and
is wrong here for the same reason. ixud.rs has no record/field reader at all.
The IXUD record table IS decoded and verified disc-wide (1104/1104 objects,
628165/628165 fields reproducing their key) and was simply never wired into
the crate.
Next step recorded: give ixud.rs an IdxdObject-shaped reader and read captions
as fields rather than adjacent tokens. The decode exists; only the plumbing is
missing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
A disc-gated test that fails if the loaders regress to one flat answer per
object. It checks the three structural claims that came with the migration
request, and one of them is overstated:
* ✅ MainMissionBonus IS per difficulty — 18 of the 24 missions pay three
different values, in a strict 1:2:4 easy:normal:hard ratio (all 24); the other
6 pay 0 at every difficulty. The flat reader returned the Easy value.
* ✅ RankScore_S/A/B/C/D repeat identically across the three Score_* records in
24 of 24 missions, so the single flat answer happened to be right. The
difficulty scaling lives in the earnings (CraftScore_Adjustment,
KillBonus_Maximum, …), not in the thresholds.
* 🟡 SpaceSize is *stored* per phase (three Phase_N records), but only **1 of the
24** missions varies it — the one that reads 250000/100000/100000. The other 23
repeat a single value (500000 ×1, 100000 ×18, 50000 ×4). "Phase_1 = 250000,
Phase_2/3 = 100000" is one mission's numbers, not a rule.
Also pinned: the Delta Saber's 63 Turret_NNN mounts at HP 100 each against a
TurretCount of 4; the SD-Battleship's per-component HP (bridge 10000, thruster
20000, shield generator 5000 at PowerRatio 0.25, hatch 100 launching
Squadron_Test2, anti-ship gun 1500) against a hull of 100000, and its 37 records
that define an HP; 418 hardpoints across the 23 capital ships, none missing HP;
131 weapons whose shell id always differs from the launcher id, 40 with a
ShellWake, and Weapon_NULL as the only one without a Shell.Power; and
fcs_range/shield_ratio/cruising_velocity/maximum_velocity/acceleration/
deceleration now set on 89 of 89 units, all of which the pool reader could miss.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
Measured first, over GP_MAIN_GAME_E.pak, comparing every named-field read the
six struct loaders performed against the record table: 4435 reads, 2872 agreed,
**966 returned None for a field that has a value**, **596 flattened a field that
several records carry**, 1 was wrong (a weapon whose TargetType is the empty
string read back as the neighbouring token "Skip"). The prior report of
4453/2887/974/591/1 is the same picture; the small differences are definitional
(I count a read as flattened only when the records disagree).
Every read now goes through IdxdObject::record, and the types say where a value
comes from:
* Weapon = the `Weapon` record (launcher) + the `Shell` record (projectile).
Both carry an ID and a Name and — with `ShellWake` — an `Interval`, which the
flat reader merged; they are separate fields now. Power/Velocity/ranges/
LifeTime are Shell fields, which is why 427 weapon reads used to miss.
* CraftUnit/Vessel = `Generic` (hull) + `Maneuver` (flight model) +
`StructureCount` (counts) + `Shield`, plus a new `hardpoints: Vec<Hardpoint>`
— one entry per Turret_/Bridge_/Thruster_/Hatch_/ShieldGenerator_ record, each
with its own HP. A flat HP could only ever be one of them.
* PlayerConfig = `Player`, plus `phases: Vec<PlayerPhase>` (SpaceSize/SupplyRange
are per Phase_N) and `score: ByDifficulty<ScoreRules>` (MainMissionBonus is per
Score_<difficulty>; the flat answer was the Easy one).
* Character faces come from the `Faces` record's field names (identical output to
the old token scrape, 0 of 68 objects differ — now by construction).
* Stage = `StageResource` + `phases: Vec<StagePhase>`, and the packages it names.
* The `fields: BTreeMap` on every struct became `records: RecordSet`, which keeps
the record boundary; `RecordSet::everywhere(field)` answers "which record".
The token-scraping loaders move too, and this is where the old reader was worst:
* Arsenal: options are the positional fields of the STANDARD_<slot> records. The
scrape returned 16 nose options of which 8 were field keys and pilot names, and
47 for arm3 of which 38 were junk, while missing Mine_B2A and No_Equipment.
Now 8/12/9/9, all real weapons.
* Squadron: one record per squadron, members are Count*4 positional slots
(unit, message set, n, pilot) — 1160 squadrons with ids and 2295 member tuples,
against 28 idless squadrons and 47 members before. Agrees exactly with the
independent Python decode in docs/re/structures/unit-group-table.md.
* DemoMessage: 11775 lines against 10263, every one with a speaker, a portrait,
a delivery mode and a voice token, from fixed positional slots.
* PilotRoster: assignments are the records the `UNITS` record names, so each one
now carries its unit id, its loadout and the player marker.
* UnitRoster: the roster is the field *names* of the single `EnumUnit` record.
* load_weapons selects on the records (Weapon + Shell) rather than on token[0],
whose first byte is often a stray pool byte ("#Weapon", "%Weapon"). Same 131
objects, no heuristic. GP_HANGAR_ARSENAL.pak holds none of them — the module
doc's claim that player weapons live there was wrong.
schema:: constants keep their names and values but are documented for what they
are: record 0's name hash (PLAYER = Difficulty_Easy, UNIT = Maneuver, VESSEL =
Bridge_000, MESSAGE = Message_000), not a schema id.
Two things the migration exposes and does not fix, flagged in the docs instead:
load_units' bucket is 43 Type=Craft + 46 Type=Vessel objects (new `unit_type`
field lets a caller separate them), and StructureCount.TurretCount is not the
number of Turret_* records (the player's craft says 4 and has 63).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
It is a throwaway measurement harness I wrote in this session ("TEMPORARY
measurement scratch — not for commit") that a concurrent commit picked up
along with its own files. It asserts nothing and duplicates work that is now
covered by tests/game_data_disc.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
to_xma_riffs now emits the leading headerless segment when it sits at a whole
number of XMA1 packets and carries a non-zero byte. VOICE_D_453 goes from a
0.14 s trailing fragment to a 45116-byte leading sub-wave that dominates it.
I withdrew this exact change earlier for two reasons. Both are now answered
rather than argued away:
* "It recovers no audio" -- it used the STEREO format. At two channels every
bank yields exactly 1792 bytes, one frame, whatever its size. Mono yields up
to 113x more.
* "It matches 1524 of 8021 RIFF-bearing entries" -- the byte-level reach is
still 1524, but the audible reach is not. Across the 84 movie-bound banks
the segment adds >1 s to exactly 7, the hokyu_*_H tankers on D_453/D_454 --
precisely the broken ones -- and <=0.25 s to 66 of the rest. The largest
non-resupply addition is S04A at +0.66 s on a 256 s movie.
The safety oracle is recorded with its limits: 8 of the 84 banks ALREADY
exceed their movie's duration before the change, by hundredths of a second,
so it cannot resolve differences at that scale. It establishes scoping, not
correctness. Callers clamp to the movie length regardless.
VOICE_D_451's all-zero leading region is skipped by the non-zero guard, so
the rule cannot prepend silence to a bank that does not need it. Pinned, as
is the packet arithmetic (n = 8, 1, 7, 22, 29) which has no tunable.
slb_disc, movie_subtitle_disc and movie_manifest_disc all still pass.
NOT verified by ear -- that needs a human.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The load-bearing error of the whole voice-bank thread, and it is mine. It
stood for three iterations across two write-ups that each called the result
proven.
I treated a subtitle cue as a timestamp that must fall INSIDE the voice clip,
and concluded a 0.07 s clip could not host a cue at 4.70 s. A cue is when the
line STARTS. The voice plays from the cue, so the clip only has to fit the
window between the cue and the end of the movie. Under that reading every
bank fits at plain 48 kHz:
bank samples @48kHz cue window
D_450 158967 3.31 4.00 5.30
D_451 76084 1.59 3.70 5.60
D_452 119562 2.49 0.00 8.34
D_453 108608 2.26 4.70 4.60
D_454 167828 3.50 0.00 9.50
2-3.5 s is also the right length for the lines. Nothing is missing, and the
17091-20563 Hz window from the previous commit is void with it -- its lower
bound came from the same misreading.
What survives, because it was measured rather than interpreted: the leading
region is XMA1 mono, the decode runs to the final frame, and cue values are
seconds.
Separately settled, and it is what exposed the error: each shared bank holds
ONE generic line. The 3-5 movies bound to a bank have IDENTICAL subtitle
text, 5 banks out of 5 -- "Rhino 3 has landed. Commencing resupply.",
"Resupply complete. You are cleared for take-off!", and so on.
That also explains the historical in-game rejection of hokyu_DS_s13A ->
VOICE_D_452 that started this whole thread. The line is generic, identical
for s02A/s07A/s08A/s13A. Someone expecting a stage-13-specific line would
hear the generic one and call it wrong -- while the binding is exactly right.
The disc said so; the subtitle text now says so independently.
Still open: whether the leading mono region is additional audio or an
alternate take, since the totals above add it to the RIFF sub-waves.
Artifact: examples/shared_bank_takes.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
Two things settled, one of them a correction of my own claim from the last
iteration.
The cue unit is verified rather than assumed. parse_timing computes mm*60+ss,
but only if the token really is mm:ss.cc, so I checked against an independent
oracle: the movies are on the disc and a cue must land inside its own movie.
66 English movies with subtitle tracks, 0 cues land after the movie ends.
Centiseconds would have overflowed essentially all 66. The seconds reading
stands and the verdicts built on it survive.
"The sample rate does not converge" does not. I reported implied rates of
39742 / 20563 / 23108 Hz as irreconcilable. They are not estimates of the
same quantity -- each is a ONE-SIDED BOUND. The audio must be at least as
long as the last cue, so samples/cue is an UPPER bound; it cannot outlast its
movie, so samples/movie is a LOWER bound. Intersecting:
bank samples cue movie lower Hz upper Hz
VOICE_D_450 158967 4.00 9.30 17091 39742
VOICE_D_451 76084 3.70 9.30 8180 20563
VOICE_D_453 108608 4.70 9.30 11677 23108
=> 17091-20563 Hz, non-empty. A single rate IS consistent.
I had been comparing them as competing point estimates, which is why they
looked contradictory.
What is still open, and stated as such: that window contains no standard XMA
rate. The lower bound assumes a whole bank plays inside one movie, and each of
these banks is bound to 3-5 movie slots -- so if a bank holds several takes
the lower bound is void, leaving rate <= 20563, which 22050 nearly meets.
Next step recorded: establish whether a shared bank is one line or several.
Artifact: examples/cue_unit_check.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
Retried the format probe with the fmt chunk built to synth_xma1_fmt's exact
byte layout, and with the bank's own RIFF sub-wave decoded through the same
pipe as a CONTROL so a broken harness cannot masquerade as a result.
The channel count is the whole story:
bank lead B channels=2 channels=1
VOICE_D_450 16384 1792 46756
VOICE_D_451 2048 1792 896 (all-zero region: control)
VOICE_D_452 14336 1792 30154
VOICE_D_453 45056 1792 203648
VOICE_D_454 59392 1792 294440
channels=2 yields EXACTLY 1792 bytes for every bank regardless of size -- one
frame, then it stops. That constant is the tell. At channels=1 the same data
yields up to 113x more, and the control sub-wave decodes to 13568, so the
pipe works.
Why the previous probe got 0 bytes everywhere is now named: I read
synth_xma1_fmt(2, 2, 48000)'s second argument as a STREAM COUNT when it is a
CHANNEL MASK, and built the WAVEFORMATEX around that misreading.
Also recorded as a refutation, because it was tempting: solving for the
sample rate as decoded-samples / last-subtitle-cue does NOT converge. D_453
implies 21665 Hz -- close enough to 22050 that I nearly wrote it down -- but
D_450 implies 5844 Hz. No single rate explains both, and the decodes are
visibly partial (samples per input byte ranges 2.10-4.96 where a clean decode
would be near-constant).
So the container is identified and the duration is not. Next step recorded:
find why FFmpeg stops early, likely the hardcoded packet/subframe fields.
Artifact: examples/slb_fmt_probe.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The corpus said 0.14 s is "far too short for the spoken line". That is a
judgement about audio, and audio judgements cannot be made in this container.
The subtitle tracks settle it without listening: each carries cue START
times, and a subtitle that appears at t seconds cannot sit inside a clip
shorter than t.
FFmpeg-measured (not estimated from a compression ratio -- the first version
of this example used an 8:1 guess, which is not good enough to hang a
conclusion on):
hokyu_LS_s02A D_450 cue 4.00 s audio 1.41 s MISSING
hokyu_LS_s09A D_451 cue 3.70 s audio 1.81 s MISSING
hokyu_LS_s02H D_453 cue 4.70 s audio 0.07 s MISSING
hokyu_DS_s13A D_452 cue 0.00 s audio 1.21 s no signal
hokyu_DS_s07H D_454 cue 0.00 s audio 0.21 s no signal
Three of five are decisive; the other two have their only cue at 0.0 s and
say nothing either way. So something is genuinely missing from these banks --
established independently of the leading-region work, and measured rather
than felt.
The fmt-variation probe I recorded as the next step is INCONCLUSIVE and is
written up as such: 36 combinations over VOICE_D_453's 22-packet leading
region all produced 0 PCM bytes, including ones that should be equivalent to
the crate's own synth_xma1_fmt, which does parse. So the probe tested my
hand-built fmt chunk, not the hypothesis, and it is NOT evidence that the
region is non-XMA. The retry should use the crate's helper.
Artifact: examples/voice_len_vs_subs.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The corpus recorded that some unit fields the disc leaves defaulted inherit
from a sibling: Size_Y from Size_X, FCSRange from RadarRange, DefencePoint
from AttackVesselPoint. Size_Y was marked the one to trust, on 9/9 support
across 7 independent ships, and it is restated in INDEX.md.
The premise is false. These fields are not defaulted -- they are on disc for
113-114 of 114 unit tables -- and Size_Y DIFFERS from Size_X in 90 of them.
The mechanism, cross-tabulating "legacy reader missed it" against "equal on
disc":
pair seen+differ seen+equal miss+differ miss+equal
Size_Y / Size_X 90 0 0 24
FCSRange / RadarRange 54 0 1 58
DefencePoint / AttackVesselPoint 51 0 1 61
seen+equal is 0 for all three: a value shared with a sibling is ALWAYS
invisible to the string-pool reader, because the pool stores each distinct
string once. And the reader almost never misses a value that differs. So
"the missing value equals the sibling's" was true BY CONSTRUCTION -- the rule
re-derived the very condition that made the field go missing. That is why the
support looked perfect: it could not fail on the cases it was fitted to.
The two miss+differ cells are its real wrong predictions, both named:
UN_e104_ADAN_Carrier DefencePoint is 0.2 (rule says 0.003), and
UN_e011_ADAN_Attacker_B_HF_Wayne FCSRange is 3000.0 (rule says 6000.0).
Retracted in unit-struct-runtime.md (original reasoning kept below the
correction), live-unit-definitions.md and INDEX.md. Pinned by a disc test
that asserts the seen+equal cells stay zero, so the mechanism itself is
guarded, not just the counts. Artifact: examples/sibling_rule_check.rs.
This one was found by my own check after the subagent assigned to it stalled.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The structure is exact. In all five resupply banks the first RIFF sits at
HEADERLESS_DATA_OFFSET + n*2048, where 1392 is a constant this crate already
had and 2048 is the XMA1 packet size: n = 8, 1, 7, 22, 29. No free parameter
to tune, and the raw bytes agree -- high entropy from offset 0, then a zero
run immediately before the RIFF. VOICE_D_451 is the control, its single
packet being all zeros.
So I made the obvious fix, emitting that region as a sub-wave, and then
withdrew it on two measurements:
* It does not recover audio. Coverage went 5.4% -> 89.9% for VOICE_D_453, but
the emitted stream decodes through FFmpeg to 1792 PCM bytes -- silence --
while the RIFF sub-waves from the same banks decode to 150-270 KB. Byte
coverage was the wrong success metric and it looked like progress.
* It is not narrow. The rule matches 1524 of the 8021 RIFF-bearing entries in
sound.pak, including RT* movie banks that decode correctly today. Landing
it would have risked a wide regression in order to not-fix five banks.
to_xma_riffs is back to its previous behaviour, verified by re-measuring:
coverage is 5.4% / 9.7% again. The refuted attempt is recorded in the code
beside the branch it would have changed, so the next person does not
re-derive the arithmetic and re-make the change.
XMA1_PACKET is kept as a named constant because the blast-radius scan uses
it. Artifacts: examples/voice_bank_shape.rs (structure), voice_bank_dump.rs
(sub-waves for decoding), slb_hybrid_scan.rs (the 1524 count).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
movie_manifest::parse now reads BASE_INFO's positional field keys (the game's
own cutscene ids, stage*100 + slot) and follows each to its record, instead of
scraping the string pool. The pool stores each distinct string once, so a
REPEAT reference produced no token and read as "no binding".
That single cause explains every wrong cell: 13 later references to
VOICE_D_450..454, two to SUBTITLE_hokyu_LS_s11A.tbl, and MS01A's share of
pwterop_s01a.prt. All 18 hokyu movies are bound, not five.
Counts, verified independently by me against the disc before recording:
104 cutscene SLOTS binding 101 distinct MOVIES; 99 slots / 96 movies with a
voice track, 99 / 96 with a subtitle, 22 / 22 with a telop. The docs' old
94 / 83 / 21 are exactly the counts of DISTINCT POOL STRINGS -- not wrong
measurements, measurements of the wrong thing. Three denominators were being
conflated; the new test pins all three.
Two assertions in movie_manifest_disc.rs were false and are corrected:
hokyu_DS_s13A binds VOICE_D_452 and resolves to eng\etc\VOICE_D_452.slb. The
in-game verdict that rejected that value tested an INFERENCE from a shared
demo id, on a decoder that discards 85-87% of banks in this class -- see
voice-bank-leading-region.md, committed earlier today.
The ~104 script ids are no longer open: they are literal positional keys,
each naming its record, and all 104 resolve. The old "counts differ by three,
positional pairing does not work" has a concrete cause -- three resupply
movies are bound by TWO slots each.
Also corrected: the naming convention has 3 subtitle exceptions (s24A/s27A
borrow s11A's track) and 18 voice exceptions, not one and five.
The legacy scraper is kept as a fallback for blobs with no record table, so
the synthetic unit fixtures still exercise it.
Artifacts: examples/movie_map_csv.rs regenerates the CSV, now slot-keyed
(104 rows; the movie-keyed version silently dropped one slot of each
duplicate). Disc tests green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The record table gives a DIRECT binding hokyu_DS_s13A -> VOICE_D_452, where
the corpus records the movie as unbound and movie_manifest_disc.rs asserts
None, citing an in-game verdict that this exact value was "the wrong
recording". That is the only place on the disc where a runtime observation
disagrees with the record table, so it was worth settling.
First, shape: these banks are SHARED. Five slots bind VOICE_D_452, five bind
451, four 450, four 453, three 454 -- 21 hokyu slots over five banks, and the
movies repeat too. Generic resupply cutscenes, not per-stage recordings.
The recorded explanation for 453 decoding to 0.14 s and 454 to 0.43 s was
that the banks are "likely multi-subwave / not cleanly sliced". Refuted: the
count of RIFF magics EQUALS the number of sub-waves recovered in all five
banks, and the last data chunk ends exactly at EOF in four of them. Nothing
between or after sub-waves is being missed.
The real defect: slb::to_xma_riffs finds audio by searching for the RIFF
magic, and a large region PRECEDES it. 87% of VOICE_D_453 and 85% of
VOICE_D_454 sit in front of the first RIFF -- 21-27% zero over 256 distinct
byte values, i.e. content, not padding. VOICE_D_451 is the control: its
leading region is 100% zero, 1 distinct value, real padding.
So the in-game verdict listened to a decode that had discarded most of the
bank, for exactly this bank class. It is evidence about the decoder, not
about the mapping. Note also that what was rejected was a value INFERRED
from a shared demo id; the record table supplies the same value as a stored
field, and only the inference was ever tested.
This does NOT establish the binding is right -- it removes the only recorded
evidence against it. What the leading region actually holds is undecoded, and
confirming the binding needs a human listening.
Artifact: examples/voice_bank_shape.rs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
With the record table decoded there is finally a ground truth to check the
old string-pool reader against. It infers `key -> value` from pool adjacency,
which is a consequence of how records are written, not a rule of the format.
Verified by hand against the disc, with an independent parser:
* `FCSRange` = 500000.0 — the module docs' own canonical example of a field
"left at its default" that "omits the value string".
* `ShieldRatio` = 1.0, where `tests/pak_idxd_disc.rs` asserted None with the
comment "a defaulted/omitted field must be None". That test encoded the
false belief; it now keeps the None as a deliberate characterisation of the
legacy reader, with the true value asserted beside it.
* `get_raw("Model")` on GP_HANGAR_ARSENAL returns the first record's model for
every record — silent corruption, not an absent value. New test pins four
records that disagree with it.
The cause is the flat API having no way to name a record: only 548 of 6325
objects have one. `HP` on the DeltaSaber answers 1000.0, the hull, while 63
Turret_* records each carry their own 100.0 (measured — a first draft said 34,
taken from a report rather than from the disc).
Disc-wide rates are recorded as single-source and labelled as such: get_raw
52% wrong, typed getters 38% miss, but 100% correct on single-record objects.
Also records a negative result: the 504 unnamed field keys were NOT recovered.
A 572464-string dictionary and 73191 variants gave 0/42. The key deltas do
prove the preimage ends with the two decimal digits.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The binary region in front of the string pool was the parser's oldest open
note ("Not yet decoded"). It is a uniform 16-byte record array sorted by
name hash, a field count, a 12-byte field array sorted by key, a pool size,
and the pool. The trailing `pool_size == file_len - pool_base` identity makes
the layout self-checking, which is what caught the first wrong version.
Verified over the WHOLE disc with zero failures: 7750/7750 IDXD objects,
190782/190782 records reproducing their stored tag_hash, 1271462/1271462
named fields reproducing their key. IXUD is the same container with
ixud_hash, UTF-16BE and every offset in chars — 1104/1104 objects,
628165/628165 fields, checked with an independent parser.
Field names are stored on disc, so no preimage search is needed: a field's
middle word points at its own name. Only 504 fields disc-wide are hash-keyed
with no name; the other 1485073 nameless fields are positional, keyed by a
literal integer (line slots, movie ids).
Two long-held beliefs are WITHDRAWN:
* The word at 0x08 is not a schema hash. It is record 0's name_hash — the
format has no type field at all, and an object's kind is known only from
the caller that loads it. It survived as "schema" because tables of one
kind share their lowest-hashed record name. Caught by a test asserting
every movie id names a real record: 1005 -> STAGE10_PHASE01 failed because
tag_hash("STAGE10_PHASE01") IS 0x067025B9, that table's supposed schema id.
* The field's middle word is not an always-0xFFFFFFFF flags word. It is
0xFFFFFFFF for 54% of fields, enough to look constant in a small sample;
the tell was that it is constant per key ACROSS records, which a per-record
flag cannot be but a per-name pointer must.
`schema_hash` keeps its name rather than churn 33 call sites, with corrected
docs. The first sweep globbed dat/** and missed hidden/DefTables.pak (1425
objects); the test now walks the whole disc root.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
Found the routines in the disassembly DB rather than guessing from data:
sub_82447DF0 IDXD tag hash (lbz+extsb, modulus 0x00FFFFDF, magic 0x2101)
sub_82447E70 IXUD tag hash (lhz, 64-bit, modulus 0xFFFFFF67 then 0x00FFFFDF)
Both transcribed instruction-for-instruction into Python and Rust.
IXUD SOLVED. It defeated every single-modulus search because it chains TWO
exact moduli -- the loop reduces mod 2^32-153 in 64-bit arithmetic and only the
result is folded mod 2^24-33. A polynomial mod M1 folded through M2 is not a
polynomial mod anything, which is exactly why the gcd test returned 1. Verified
independently: 86/86 record keys and 108,261/108,261 field tags in
GP_MAIN_GAME_E.pak, and NoRecord -> 0x1c6d9c96.
CORRECTION 1: tag_hash must SIGN-EXTEND each byte (extsb). My reconstruction
used unsigned bytes and matched all 1.27M disc names -- every one is ASCII --
while disagreeing on ~90% of random inputs with a byte >= 0x80 (verified:
18096/20000). The disc could never have caught this; only the disassembly did.
CORRECTION 2: name_hash's reduction is EXACT, not lossy. The module doc claimed
the missing conditional subtract made it something other than %. rlwinm r6,r6,
9,23,31 is just hi>>23, and with RECIP = floor(2^55/M)+1 that is Granlund-
Montgomery magic division -- 0 wrong at every quotient boundary across the full
32-bit domain. Retracted.
cargo test -p sylpheed-formats --lib hash: 10/10.
Closes the 4-byte record key. tag_hash is name_hash's shape -- byte-sum
checksum in the top byte over a 24-bit modular polynomial -- with two different
constants: modulus 0x00FFFFDF (2^24-33, prime) instead of 0x00FFF9D7, and no
lowercasing, so tags are case-sensitive. name_hash explains 0 of 8643.
Recovered from the tables rather than the executable: every inline field name
is a known (name -> tag) pair, and comparing names differing in one character
gives the per-position weights 1, 0x100, 0x10000, 0x21, 0x2100, ... -- a byte
leaving bit 24 re-enters as 33, i.e. reduction mod 2^24-33. Holds where it is
easy to get wrong (distance 8 and 9 carry correctly).
A record's key is the tag of its own name: FormationSet rosters 362/362,
UnitGroup rosters 281/281, S02 squadron names 111/111 -- so records can be
addressed by name without reading the roster first.
Implemented in Python (unitgroup.tag_hash) and Rust
(sylpheed_formats::hash::tag_hash) with 3 new unit tests carrying disc-derived
vectors; cargo test -p sylpheed-formats --lib hash is 8/8 green.
Not settled: the guest routine is unlocated, so this uses exact modular
arithmetic where the game may use a Barrett step without final fixup.
TextIndex in sylpheed-formats::localization already indexes per-phase objective
text under S<NN>_P<n>_Objective_<i>, with hints and lose conditions alongside.
Querying it settles what the wave work spent many iterations circling, and a
disc test now pins it.
Stage 02 phase 1 is "Shoot down all invading enemy fighters while watching out
for attacks on the ACROPOLIS", losing if the ACROPOLIS is sunk. Phase 2 is
"Protect the Admiral's ship the CALIBAN until it has entered the safe zone",
losing if the CALIBAN is sunk. Phase 3 is "Destroy the interstellar cruise
missiles headed for Schlos Base", matching phase 3's roster of exactly nine
ISCMissile and SUBOBJ_013.
That explains the phase advances and shows they are not all the same kind. Phase
1 is a kill-all-marked-fighters gate -- fighters, not the turrets every run
killed almost exclusively. Phase 2 is positional and ends when the CALIBAN
reaches a safe zone, so "what advances a phase" never had a single answer.
It also closes a loop to the first session. The hints say red mission markers
[OB] indicate your targets, so REMAINING OB -- the HUD counter hunted in the
earliest iterations and located at 0xbdb59668 -- is the count of remaining
marked objective targets, which is phase-1 progress itself. The old reframing
that the counter is not a roster was right, and this names what it is. It is
also the correct signal to watch for an advance: not deployed, which only
changes once the next phase deploys, but REMAINING OB reaching zero, which is
the cause.
Method lesson recorded because the reflex will recur: localization.rs has had
objectives, hints and lose_conditions for some time, and several iterations were
spent reconstructing the same information by hand from SUBObjective strings and
the guide script. The reconstruction was not wasted -- it produced the IXUD
UTF-16BE and language\ prefix findings independently -- but the corpus should
have been searched before the disc was, and grep -rl TextIndex crates/ would
have saved the detour.
Minor discrepancy noted, not investigated: ixud.rs documents the string pool as
UTF-16BE and localization.rs as UTF-16LE.
The 16.16 reading of +0x04 rested on twelve bundles at 30.0. They are not twelve
witnesses: they are TWO bundles - pghud_range_main_em/emeff and
pghud_range_nose_em/emeff, both dur 30 - repeated across six language PAKs. A
30-frame flash is equally consistent with 0.5 s at 60 and 1 s at 30, so the
observation that "30.0 bundles are shorter" says almost nothing.
Worse for the clean fixed-point reading: the four bundles at 0x3C0001 would be
60.0000152 fps, which nobody authors. They are py_ranking_jump/py_ranking_next
dialogs, all dur 60. So the better reading is <rate:16>.<flag:16> - a rate-like
number in the high half and a small low field that is 0 on 2843 bundles and 1 on
four, meaning unknown.
Also stated: nothing in this container can settle it by timing, because the
emulator runs on software Vulkan far from real time - a stopwatch would measure
lavapipe rather than the game.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The rate/duration reading of the bundle header came from the values alone, so it
was checked against something the file states independently: the keyframe `time`
fields in the placement region.
Over the 2313 bundles that have both, the largest keyframe time is <= +0x08 in
EVERY one, none exceeds it, and 444 reach it exactly. The refutation attempt is
in the same data: a large unrelated constant would bound every time too, but the
max_time/+0x08 ratio would then pile up near zero - instead it peaks at exactly
1.0 with 520 bundles. Asserted, including the shape of that histogram, so the
bound cannot later be waved through as vacuous.
Stated precisely because the units are a separate claim: what is proven is that
+0x08 is the animation length in the SAME UNIT as a keyframe's time. That the
unit is frames still rests on the values (30/60/120/1200) and on the 16.16
reading of +0x04, which stays amber - the only new evidence for it is that the
twelve 30.0 bundles cap at +0x08 = 30 while the 2843 60.0 ones reach 1440, and
twelve bundles is not a demonstration.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The backlog asked what makes a bundle a screen rather than a fragment, and the
obvious suspect was the 32-byte header. Swept over all 2859 composable bundles
with a real declaration table.
The answer to the question is NO, and it is asserted rather than argued: no bit
of the flags word at +0x10 labels a screen. The best any bit manages is bit 13 -
403 bundles, 179 of them carrying a full-screen element, a 44% hit rate against a
12.8% base - and the commonest bit is set on 91% of everything. Enrichment, not a
marker.
The sweep found more than it was asked for, though. The header is not dead space:
+0x18 is 1280 on 2829 bundles and +0x1c is 720 on 2823 - the design resolution at
bundle level, the same pair the parser already reads out of a .rat record, and
asserted here. And +0x04 takes only three values, 0x3C0000 on 2843 and 0x1E0000
on 12, which are exactly 60.0 and 30.0 in 16.16 fixed point, with +0x08 taking
30/1200/120/60 - a frame rate and a duration in frames would fit a format whose
records are keyframe lists. That reading is marked amber: it comes from the
values alone and is not verified against an animation.
Also recorded, since the file will not say: element counts are min 1, median 2,
p75 5, p95 23, max 56, and only 365 bundles carry a full-screen element. The
population is mostly fragments and the separation is shape.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
Two readings were on record and both were wrong in different directions: the
structure doc called it "normal state -> focused state" from a single example,
and the backlog called it "refuted as focus; unexplained otherwise".
Classified every link reachable from a declaration table: 1467 links, and ALL
1467 resolve to a RATC child of their own bundle, ALL are .rat -> .rat, none
dangle and none self-link. 1076 (73%) match the <stem>f focus pattern; the other
391 are chains between effect records - px_bunk_eff01 -> pjex_eff -> pjex_eff07,
pveff01 -> pjeff02 -> pjeff21 - which also explains why only 227 targets are
themselves declared elements: the middle of a chain is, the end is not.
So `opt ` is a record-to-record reference within the bundle, and focus is its
commonest use rather than its meaning.
Coverage is stated rather than glossed: the bundles hold 18718 raw `opt ` tags
against the 1467 classified, because opt_link reads the first tag of a DECLARED
element's record. Roughly 92% of occurrences sit deeper in the chains (or are
byte coincidences - the scan is unaligned) and are untested. The numbers are
asserted so the answer cannot drift back into an anecdote.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The backlog's cheapest open question about the 60-byte entry: kind is a flags
word (0x10 untextured primitive, 0x4 repeated instance, 0x3002 button record), so
a focus bit would be the obvious answer and the name-pairing in
mark_focused_states would be a convention standing in for a real field.
It is not. Over every screen build on the disc: 54 name-paired focused/base
pairs, all 54 with IDENTICAL kind (all 0x0), NO bit ever set on the focused entry
and clear on its base, and the only words of the entry that ever differ are +48
and +52 - the pivot. The two entries differ in where the sprite sits and in
nothing else.
The test asserts all four numbers rather than just printing them, so the negative
is pinned instead of decaying back into a suspicion. Also noted: these buttons
carry kind = 0x0, so the documented 0x3002 "button record" belongs to the .rat
records and not to the .t32 sprites a menu draws.
Closes the declaration table, not the question - the focused state could still be
marked in the .rat record, the RATC child stream, or only in the game's code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
The three orders the derived rule was built from all live in GP_TITLE.pak, so
they cannot confirm it - the rule was fitted to them. These two are from
GP_SAVE_LOAD.pak, read off the running game now that the Canary threading fix
makes the main menu dependable.
The 9-element slot-list header composites EXACTLY as the sort predicts, on all 6
instances of it, and nothing about this screen was fed into the rule:
measured 7 8 0 1 2 3 4 5 6
derived 7 8 0 1 2 3 4 5 6
including TWO tied groups (0xb102 x2 and 0xb210 x5) that both come out in
declaration order, and the unkeyed pfeff00.prm fade quad last.
The 13-element save/load frame differs in exactly the two open questions and no
new ones: two unkeyed pfbase.tbm backgrounds paint FIRST where the sort puts the
keyless last - the splash's palogo_eff0.prm behaviour in a different file type,
so implied_layer_key now covers it - and the 0xb100 group of four paints
10,11,8,12 where declaration order is 8,10,11,12.
That second point is a SECOND screen with a mis-ordered tie, which is what the
question needed, and it immediately kills a candidate: 10 and 11 are kind=0x2002
while 8 and 12 are 0x0000, so "descending kind then declaration index"
reproduces 10,11,8,12 exactly - and then fails both title groups, where every
element of 0x8083 is kind 0 and where 0x80a0 would predict 2,3,4,5,0,1,7 against
a measured 0,2,4,7,1,3,5. Seven candidates refuted now.
16 disc tests green.
With the layer key and the primitives' implied keys in place, the tie-break -
how the game orders elements sharing a key - is all that is left between the
derived order and ground truth. Three measured screens now constrain it.
On the menu and the splash every tied group comes out in declaration order,
which is what the stable sort already gives. The title is the only screen that
discriminates, and nothing predicts it: 0x8083 x5 paints eff1, eff2, eff5, eff3,
eff4, and 0x80a0 x7 paints logo1 x3, tm, logo2 x3.
Refuted: declaration order; RATC child order; first keyframe time (52, 56, 62,
58, 60 - the measured order is not sorted by them); resting keyframe time;
resting X or Y (938, 938, 64, 788, 447); and T8aD header words +00, +04, +0c and
+10, which are either identical within a group or unsorted.
Child order is worth its own line: a strict improvement over declaration order
(7 misplaced positions on the title instead of 9, and it recovers the logo
grouping) and exactly right on the other two screens. NOT adopted, because on the
only screen that can tell them apart it is still wrong.
Adds a test that measures what the residual costs instead of assuming it. Of the
3 disagreeing pairs of drawn elements across all three screens, all 3 have
overlapping bounding boxes and 2 share opaque pixels: ptlogo_back2eff5 against
eff3 (22568 px) and eff4 (32395 px). The third pair, ptlogo2 vs ptlogo_tm,
overlaps by two columns and shares NO opaque pixel - the wordmark is transparent
there. A bounding-box test called that a defect; reading the alpha says it is
not, which is why the test reads pixels. The set is pinned, so a change that
makes it worse fails.
15 disc tests green.
Two places checked for the key a .prm element sorts by, both empty:
* the declaration entry's four unread words are constant across every element
of all three measured screens (+28=0, +36=0xffffffff, +56=0, and +44 is a
button ordinal 1-5, 0xffffffff elsewhere);
* the bundle carries NO data for a primitive at all - the menu build declares
pteff00.prm, pteff02.prm and pteff05.t32 and has zero RATC children for any
of them, its 34 children being 21 T8aD sprites and 13 .rat records.
So the layer comes from the game code. But it is consistent, which is what makes
a per-name table a measurement rather than a fudge. Bracketing each unkeyed
element between the keys of its measured neighbours: pteff02.prm falls in
(0x8010, 0x8040) on BOTH screens it appears on, pteff00.prm is past the maximum
on both, and palogo_eff0.prm is below the minimum on the splash.
implied_layer_key records exactly those and nothing else; an unlisted primitive
keeps u32::MAX and still sorts last. With it, derived_paint_order produces the
same layer-key sequence as the order read off the running game on all three
measured screens - primitives included - and matches element-for-element on four
of the five bundle instances. The fifth is the title, differing only inside its
tied groups, which is a separate open question.
This does not make include_primitives safe by default: the 36 builds that come
out one colour are wiped by pzeff00.prm and pceff00.prm, never measured, hence
not in the table.
14 disc tests green.
Read off the running game with screen_children.py and identified by pivot
signature as GP_TITLE.pak ratc-index 8, the NEW GAME / LOAD GAME / TUTORIAL /
OPTIONS / EXTRAS screen:
paint order: 1 3 4 2 5 8 9 6 7 15 10 11 12 13 14 0
It is the first measured screen carrying TWO primitives, and they land in
different places, which is the point. pteff02.prm (the 25% dim) paints 4th,
beneath the whole UI; pteff00.prm (the transition fade, resting transparent)
paints last. Both match their positions on the title screen exactly. So a
primitive's place is per-element and stable by role - backdrop first, dim at
slot 4, fade last - and there are now three permutations to test a derivation
against rather than two.
Wired into measured_paint_order, keyed by element names so both language builds
get it. The English build composited with --primitives edge-correlates at 0.9591
at shift (0,0) against a framebuffer capture taken in the same session - a third
screen confirming paint order, resting pose, fade alpha and primitives at once,
against a capture this project had not seen before.
13 disc tests green.
fill_quad composites an untextured primitive as a solid rectangle of the
keyframes fade colour, pivot x 2 in size, placed and scaled exactly as a sprite
is. Behind ComposeOptions::include_primitives and screen render --primitives.
On the title screen, whose paint order is ground truth, it is measurably right:
mean luminance 76.30 -> 63.72 against the captures 64.58, i.e. from +18% to
-1.3%, and mean absolute difference 16.07 -> 13.08. The background was ~40% too
bright; pteff02.prm, a 25% black dim, was what was missing. The wordmark is not
dimmed by it because the measured order paints that quad at slot 4, beneath the
logo. Edge correlation moves 0.9538 -> 0.9480, which is not informative here: a
uniform dim scales gradients uniformly so a normalised edge score barely sees it.
OFF BY DEFAULT, and that is the finding. A primitive has no T8aD header, so no
layer key, and derived_paint_order sorts the keyless to the end. GP_DIALOGs
pzeff00.prm is a single keyframe of opaque black at full screen; painted last it
wipes the build. Of the 125 builds that draw a visible primitive, 36 come out
>99% one colour with the derived order.
No constant default works, because the two screens read off the running game
disagree: the splash paints its primitive FIRST (the black backdrop) while the
title paints one at slot 4 and another LAST (the fade-out). Declaration order
fails the title too. A disc test measures the damage rather than asserting the
feature works, so the number moves when the ordering is solved.
Also records a false alarm worth keeping: a first pass reported 36 GP_DIALOG
builds at "100% black", which was a crude near-black pixel threshold and not a
black screen - those dialogs are dimmed 50% and perfectly legible. The genuinely
wiped builds are a different set.
blit modulated by tint only, which is 0xffffffff on essentially every keyframe,
so the fade word was decoded, stored and then discarded. Applying it as an ARGB
modulate takes the title composite's edge correlation against the framebuffer
capture of the running game from 0.4597 to 0.9538, both at zero shift. The white
wordmark with its blue outline, the trademark, the copyright and the orange
exploding planet all appear -- the planet because a full-screen blue effect that
rests at alpha 0 had been painting over it at full opacity.
ARGB is measured: across a fade-in the high byte walks 0x00 -> 0x80 -> 0xc0 ->
0xe0 -> 0xff while the low three stay ffffff, and the low 24 bits are 0xffffff on
5276 of the disc's 5453 resting keyframes.
A modulate can only remove pixels, so the risk is a blank screen. Measured: no-op
on 4060 of 5200 sprite elements, partial on 453, hides 687 -- transient HUD
indicators that should not be lit at rest -- and NO build is left with nothing
visible. A disc test asserts that, and that the no-op share stays high, which
also guards the resting rule against drifting onto ramp frames.
Both changes are in one commit because the second is invisible without the first.
Applying fade erased the word PAUSE, which the running-game capture plainly
shows: pgptitle.rat has three runs of two identical keyframes -- pre-roll, hold,
exit -- and the 'later run wins' tie-break grabbed the exit. A group carries the
screen's entry animation AND its exit, so a run ending on the last keyframe is
now excluded unless it is the only one. The title correlation is unchanged.
Element::rest() picked the keyframe with the largest gap to the next keyframe's
time. That reads a keyframe as a value held until the next one; it is the start
of a ramp toward it. A long gap after keyframe k means the screen spends that
time arriving at k+1, so the settled pose is at the far end of the gap.
The title wordmark zooms in over five frames and holds at (184,193) at 100% from
t=251 to t=264. The old rule picked the frame before the long gap: (179,186) at
101%, still mid-zoom.
Measured against the framebuffer capture of the running title screen, which is a
1:1 crop so frame coordinates map directly (confirmed: the copyright line lands
on row 669 in the capture and in both composites). Edge-correlated over the
wordmark box:
plateau (landed) best 0.4597 at shift (0,0)
longest dwell (old) best 0.1511 at shift (+3,+8), 0.1268 at (0,0)
The old composite scores 3x lower and only peaks after being moved, by about the
(-5,-7) that picking kf4 instead of kf5 predicts.
It also fixes six title elements the old rule rested at alpha 0x00 where the
capture plainly shows them, and pteff00.prm - the full-screen fade quad painted
last - which rested at opaque black. That was the blocker on .prm compositing.
Adds tools/re-capture/align_to_capture.py, which is how this was scored, and
turns the .prm test that deliberately asserted the old defect into a guard on
the fix.
Not settled and now the next item: compose ignores the keyframe fade alpha
entirely (blit modulates by tint only), which is why choosing the wrong keyframe
was invisible until now.
