All three MUST verified mechanically: the one-pair-per-action capture
claim PROVEN against RetailDefaults() (9 multi-binding actions, no
counterexample); the identity map now injective (all 122 entries
checked); the confirm dialog real, N-way, accept-only, with the lazy
read's null unreachable in production. The OK/Cancel gesture divergence
is honestly retired (no distinguishing retail affordance = no Risk
symptom to write).
Coordinator third-round residuals: R1 — S2's deferral is commit-message-
only and M2 worsened it (0x6's DAT defaults are the arrow keys, so
Defaults + the flat conflict universe makes 'Move Forward -> Up' trip a
false N-way confirm; the gate script carries no warning). R2 — AP-203's
row body doesn't cover the ten same-verb-sibling-live rows. SHOULD: the
ten Camera Alternate rows render BLANK (retail shows the arrow keys);
identity-map injectivity is load-bearing for M1+M2 with nothing
enforcing it. MERGE NOTE: textually clean vs the campaign tip, but
OP8's six ListBoxes sit on the #372 viewport path — blank on the branch,
auto-healed by merging onto 057d8cd7+; the OP8 gate runs POST-merge.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The 13 ID_ChatOption_TextFilter_* labels (and their _Desc tooltips) live
in string table 0x2300000D, not the 0x23000003 options table the section
headers and slider labels use. Dat-verified: the targeted sweep missed
(0x23000001-0A), the control key resolved ('Auto Target' — machinery
fine), and the exhaustive all-tables sweep (ProbeFilterLabelHome, now a
permanent env-gated probe) hit exactly once: 0x2300000D -> 'Combat'. The
initializer decomp confirms the hash KEYS are the symbol names verbatim
(the vftable-member operands at 0x006f04cd are the known pooled-string
artifact); only the research doc's table attribution was inferred rather
than dat-verified — corrected in §8.
All 13 rows now render their captions instead of the honest-fallback
blanks the first connected gate saw. Full Release suite green (one
Core.Net loss-simulation timing flake on the first run, green targeted
and on rerun).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
M1: SetForAction rebuilds bindings as new Binding(chord, action),
defaulting Press/InputScope.Game and dropping the read chord's
ActivationType/InputScope — MovementWalkMode (Hold), the three melee
heights (Hold+MeleeCombat), Missile/Magic-scoped rows, and
CameraInstantMouseLook (Hold, Device 1) all collapse; RestoreDefaultValue
invokes _apply so ONE Defaults click flattens ~140 actions and OK
persists it (gate step 12 tells the user to click Defaults). M2: the ten
camera 0x5/0x6 rows alias one InputAction each — 20 rows over 10 targets,
twins seed identically, clobber each other, self-conflict; the
round-trip test justifies the dual map on independence the mechanism
makes impossible. M3: silent auto-reassign (NotifyReassigned -> empty,
DisplaySystemMessage skips empty) destroys a binding with zero feedback
while the doc claims it reports; retail confirms first
(OpenOverwriteBindingDialog), the plan gate says 'conflict prompt', the
tested SettingsVM prompt precedent exists, AP-204 concedes
RetailDialogFactory is wired, and a fake asserts a message production
contradicts. SHOULD: non-bindable check ordered after first-match
conflict (inverts retail's any-non-bindable-refuses); ActionMap.
ConflictingMaps never read (false conflicts on legitimately-shared
combat keys); Save lacks the writer's try/catch; sparse-row Mapping-3
lands on Mapping-1; ~330 uncached layout imports at startup.
Clean: the 0x26000000 DID (better-anchored than cited — .actionmap
MasterDBMap range, 0x27 is the type tag); no-Clear-button confirmed
against the fixture; six-box grouping matches lane D; identity
byte-comparison is a permanent test; modal capture real;
AP-202/203/204 accurate; OP3 INERT retired. Brief-premise correction:
ff577641 is the CAMPAIGN tip (claude/latest-commits-cb0c8f), not main's
(852cdda7); OP8 additive, only OptionPageModel.cs is a triple-append
merge watch-point.
Rework round 1 to the isolated op8-keyboard worktree.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The re-reviewer byte-decoded ALL six caption sites independently (no
transposition; a third evidence line from the fixture's authored
left/right label geometry), re-verified the audio chain through the one
place an inversion could still hide (the checkbox pass-throughs), and
hand-traced the 97-key conformance table (exactly 97, none invented or
dropped). Residuals applied here: SF-1 the AudioSettings doc comment's
wrong function attribution (the SetDefaultValue literals live in
gmConfigUI::InitOptions @0x0049E435/E457/E479, not InitUIPreferences);
SF-2 the gate script no longer asserts an unread DAT caption — it gates
on behaviour and asks the tester to report the authored English
verbatim; lane A's stale '24 option rows' corrected to 27 (the 39-item
pin is the authoritative tally). AudioSettings.cs change is
comment-only (no executable-code delta; suite state carries from
67b0815c, re-verified at the next code commit).
Campaign state: OP1-OP7 ALL CODE-COMPLETE; OP3/OP4/OP5/OP6 gates ready;
OP8 (Configure Keyboard) is the sole remaining implementation slice.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
R1: UiRoot now delivers WM_CAPTURECHANGED (0x215 — retail's own Win32
event-id space) to the element losing pointer capture on BOTH release
and re-target; UiScrollbar terminates a mid-drag gesture there,
completing it (one DragCompleted flush persisting the user's last-seen
value) and unlatching IsDragging — a panel-close keybind mid-drag or a
second-button re-target can no longer latch the drag flag forever and
silently suppress every later settings flush. Normal MouseUp paths
no-op (the latch is already clear when capture releases).
R2: the scalar latch arms BEFORE the track-click jump applies, so the
jump's own ScalarChanged tick defers its flush to the MouseUp's single
DragCompleted — one flush per press gesture, never the
inline-then-completed double; the DragCompleted doc now states the real
contract (fires once per value-capable gesture incl. capture loss)
instead of the refuted never-on-jump claim.
Tests: capture-loss mid-drag (ends + completes once + stray-MouseUp
no-double), no-drag capture-change no-op, bare-track-click
single-completion with the latch observed armed during the jump tick.
Also reconciles the research doc's U4 row to its closure (the six
caption pairs, the BN zero-fold post-mortem) per the OP6 rework's flag.
Full Release suite: 13,128 passed / 4 skipped / 0 failed (one
documented #250-class allocation flake on first run, green in
isolation and on full-suite rerun).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes all three MUST-FIX findings from the OP6 REJECT review
(docs/research/2026-08-11-op6-review.md) plus its SHOULD-FIXes and NOTEs.
M1 — the "retail ships zero range captions" claim was a Binary Ninja
constant-folding artifact (the same class the header-string globals a few
lines above already worked around). The six SetSliderLabel call sites
byte-decode to reads of runtime-filled ID_Graphics_Value_* globals, not
immediate zeros (PE-byte-verified against the PDB-paired acclient.exe,
independently re-derived in this session, not just re-asserted from the
review). ConfigOptionsPageController.BuildSliderRow gained optional
rangeLowKey/rangeHighKey parameters wired for all six idx6 sliders (Camera
Stiffness Soft/Hard, Adjustment Speed Slow/Fast, FOV Narrow/Wide, Screen
Brightness Dark/Bright, Graphics Performance Speed/Detail, Degrade Distance
Close/Far) via the same SetRangeLabel mechanism OP5's Chat opacity sliders
already established. Mouse Look Sensitivity (idx3) correctly stays
uncaptioned — the one genuine SetSliderLabel omission. Class doc corrected;
gate-script lines 535/653-equivalent corrected in place.
M2 — the three Sound "Disabled" toggles were semantically inverted:
SoundManager::effect_sounds_enabled/ambient_sounds_enabled/
interface_sounds_enabled are all compiled = 1 in .data, and
UserPreferences::RegisterPreference binds the checkbox's boolean value
DIRECTLY onto those enabled-sense statics — checked-by-default means
enabled-by-default, not disabled. AudioSettings.SfxDisabled/AmbientDisabled/
InterfaceDisabled renamed to SfxEnabled/AmbientEnabled/InterfaceEnabled
(fresh JSON keys — the rejected slice's keys never shipped in an accepted
build); RuntimeSettingsStartupTargets.ApplyAudio now computes effective
volume through the extracted, independently-unit-tested pure function
ComputeEffectiveCategoryVolumes (enabled ? slider : 0f). This closes the
blast radius the review flagged: a missing key in an EXISTING settings.json
now falls back to AudioSettings.Default, which is enabled=true, so a fresh
launch is audible, not muted. AP-199's wording and gate-script step 6
corrected; the enshrined-inversion test rewritten to assert the correct
default and a new SettingsStore test pins the legacy-file fallback path.
M3 — UI_ChatFontFace now ships all five of retail's authored choices
(Arial, CourierNew, PalatinoLinotype, Tahoma, TimesNewRoman — a fixed
compile-time array at gmClient::InitUIPreferences, PE-byte-verified
present verbatim in .rdata, not a per-machine runtime enumeration as the
rejected slice's comment claimed). Default index 2 (PalatinoLinotype) now
indexes a real entry.
S1 — Bind() now emits the sixth trailing AddSeperator retail's own
InitOptions ends with (0x0049E80D), matching retail's 39-item ListBox (6
headers + 6 separators + 27 option-widget-rows) instead of 38.
S2 — Screen Brightness gets its own DisplaySettings.ScreenBrightness field
([-1,1], default 0) instead of overloading Gamma, which has a different
unit system (default 1.0, legacy [0.5,2.0] slider) and its own live
Settings-panel consumer.
S3 — UiScrollbar and UiMenu gained a settable TooltipText surfaced through
GetTooltipText (UiButton's existing pattern). Every slider and menu row's
own interactive widget (not just toggle/trio rows) now carries retail's
"<label>_Help" tooltip, verified as a universal suffix convention across
every AttachPreference site touched by this tab.
S4 — "800x600" added to DisplaySettings.AvailableResolutions: a genuine
retail display mode (Device::ForceDisplayResolution(1,0x320,0x258) at
startup) and the Config tab's own byte-verified Resolution row default, not
an invented preset. Defaults now lands on a highlighted, re-selectable
dropdown entry instead of an orphaned value.
S5 — four new/extended tests: ComputeEffectiveCategoryVolumes gets a
dedicated pure-function value assertion (Theory + a default-profile-is-
audible Fact) in RuntimeSettingsControllerTests, closing the "only event
order was asserted" gap that let M2 ship; a label/choice-key conformance
table in ConfigOptionsPageControllerTests enumerates every key this tab
queries (traced directly from the fixed code paths, not guessed) and fails
on an invented OR a dropped key; a per-row DefaultValue pin asserts every
row's default against the retail literal directly, independent of the
underlying settings-record defaults; and the S1 separator fix gets its own
39-item stacked-ListBox count pin.
NOTEs — AP-198's row count was always ten (its own enumeration never said
nine); the commit-message inconsistency N1 flagged is reconciled in both
the row and the section-summary line, and its Screen Brightness sub-clause
now matches S2. N2: Bind() now reads the scrollbar id from
UiTemplateListBox.ScrollbarElementId (dat property 0x72) instead of a
hardcoded constant. N3 (batch Defaults writes) and N4 (AfterApply on
Config-tab entry, needs no action) are left as recorded — out of this
rework's scope per the review's own disposition.
Full Release suite: 13,125 passed / 4 skipped / 0 failed (baseline
13,117/4/0 — net +8 tests added, 0 regressions, 0 removed).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
All findings genuinely closed: the own-slider push verified against the
never-clamp invariant (mid-drag no-op by construction); the DragCompleted
single-fire proven at all three MouseUp sites with a STRONGER capture
argument than the fixer's (pointer capture + Reset needing its own
click); the S3 review-correction verified from git (the round-trip test
existed at e71e5a96 — the review was wrong, the commit right).
Coordinator residuals (third-round, mine): R1 — _draggingThumb clears
only at MouseUp, but UiRoot can drop capture without one (panel-close
keybind mid-drag; second-button re-target), latching IsDragging true
forever so Reset/Defaults apply live but never persist; the
UiRoot.PointerCaptureChanged seam exists for the root fix. R2 — a bare
track click sets _draggingThumb unconditionally in OnScalarEvent, so it
double-flushes and refutes the DragCompleted doc's never-fires-on-jump
claim. Five NOTEs recorded incl. the Chat-side de-scoping blind spot.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the OP5 (Chat tab) dual-lens review findings against e71e5a96:
- M1 (MUST-FIX): each opacity row's own apply closure now pushes its OWN
slider's thumb from the post-link truth (bindings.Current*Opacity()),
mirroring the OP4 binding pattern. Before this, a single-slider drag
followed by Reset reverted the live value/link but left that slider's
own thumb stuck at the dragged position.
- S1 (SHOULD-FIX): the Chat tab's two opacity sliders no longer round-trip
the whole settings.json on every drag MouseMove tick. UiScrollbar gains
IsDragging + a DragCompleted callback (fires once, at the MouseUp that
ends an actual thumb drag); the opacity apply closures flush immediately
when not mid-drag (Reset/Defaults/discrete edits, same as before) and
defer to DragCompleted otherwise, collapsing dozens of per-tick writes
into exactly one per drag gesture. Live opacity still applies every tick.
- S2 (SHOULD-FIX): filed register row AP-201 and issue #371 for the
UiScrollablePanel whole-row-cull-vs-clip divergence the review found
(predates OP5, made user-visible by OP5's 240-260px filter blocks). Not
fixed in this round (a renderer-level scissor stack is out of scope
here) — corrected the OP5 connected-gate script instead so a straddling
block's disappear-then-reappear-whole is no longer reported as a
self-sizing regression.
- S3 (SHOULD-FIX): the chatWindowMainFilter round-trip test already
existed in e71e5a96 (the review missed it scrolling past line 330);
added the genuinely missing coverage instead — a composed test pinning
RetailUiRuntime.MountChat's window-0 SettingsStore -> ChatWindowState
seed (MountChat itself needs live DAT access and isn't unit-testable
directly).
- N11: ScrollbarLinkage_ModelPointsAtTheChatListBoxScroll now asserts
through the scoped page-slot lookup (UiElement.FindDescendant) instead
of the flat layout.FindElement, which passed for the wrong reason given
the shared scrollbar id 0x10000201 — matches OP6's own scrollbar-linkage
test pattern.
Also updated ConfigOptionsPageControllerTests' local ChatOptionsPageController
Bindings fake for the new FlushOpacity parameter.
Full Release suite: 13,117 passed / 4 skipped / 0 failed (baseline 13,107/4/0
post-OP6 — 10 tests added, zero skips added, zero failures).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
M1: the 'U4 resolved: zero range captions' claim is the SAME BN zero-fold
artifact lane A caught on AddHeader — the bytes at 0x0049E4C6 load two
string-id globals (ID_Graphics_Value_Hard/_Soft) and all six labelled
sliders decode cleanly (Stiffness Soft/Hard, Adjustment Slow/Fast, FOV
Narrow/Wide, Brightness Dark/Bright, Performance Speed/Detail, Degrade
Close/Far) — U4 closes the OPPOSITE way; OP5's Chat sliders already
ported the same mechanism. M2: the three Sound 'Disabled' toggles are
sense-INVERTED (the INI key names the string; the registered static is
effect_sounds_enabled = 1) — SFX/ambient ship muted for every user
including existing settings.json installs, undoing Campaign A; a test,
AP-199's wording, and gate step 6 all enshrine it. M3: UI_ChatFontFace
ships 1 of retail's 5 authored choices (all five font strings verbatim
in the binary) with default index 2 out of range — a self-inflicted
copy of the genuine LandscapeDrawDistance oddity. S1: the sixth
trailing separator missing (retail builds 39 items); lane A's '24 rows'
is itself stale (27 is right). S2: Screen Brightness overloads Gamma
([-1,1] default 0 into a [0.5,2.0] multiplier defaulting 1.0). S3:
tooltips reach 12/30 rows (TooltipText exists only on UiButton). S4:
the 800x600 default is absent from the hardcoded resolution list. S5:
the tests pin safe seams and miss every risky one (no ApplyAudio value
assertion, no label/choice conformance, no per-row default pin, no
count pin). A large verified-clean catalog is recorded so the rework
does not re-litigate.
Rework round 1 queues behind the OP5 fixer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Binds the retail Options panel's Config tab (LayoutDesc 0x21000029, 27
authored rows across 6 sections) through OP2's template mechanism and
OP3's per-page OptionPage model, matching the Character/Chat tab
controllers' established pattern.
The row table is transcribed directly from two decompiled sources —
gmConfigUI::InitOptions @0x0049E400 (row order, widget shape, defaults)
and gmClient::InitUIPreferences @0x004035b0 (the complete
UIPreferences::AttachPreference registration: every label/tooltip key,
every slider's real-unit range, every menu's enum choices) — which
resolves the research docs' own "U4" unverified slider-caption pairing:
retail ships ZERO range captions on this tab (every SetSliderLabel call
passes literal string id 0).
Consumer disposition: LIVE — Sound/Ambient volume-trio sliders and their
toggle halves (AudioSettings.SfxDisabled/AmbientDisabled now gate the
already-live engine write; RuntimeSettingsController.SaveAudio newly
pushes into OpenAlAudioEngine on every change, not just at startup),
Resolution/Full Screen (immediate window resize on save). NEXT-LAUNCH
(pre-existing precedent): Sync To Refresh, Field of View. STORE-ONLY
(register rows AP-198/199/200, TS-74 extended): Sound Features/Interface
trio/Play-Only-When-Active, the nine Graphics/Rendering-Quality rows
(Vulkan has no per-feature render knobs), Camera/Input's six rows and
Use Mouse Turning (no persistent mouse-turning camera mode), Chat Font
Face/Size (distinct new fields from the existing live ChatSettings.FontSize).
AudioSettings/DisplaySettings/CameraTurningSettings/ChatSettings each
gain new fields for their slice of the 27 rows, backed by SettingsStore
round-trips. A real bug caught by testing: the scrollbar scope lookup
used the standalone-layout root id (0x100001FF), which does not survive
base-merge into the host-mounted tree — fixed to scope from the tab
host's own page-slot id (0x10000213), matching Chat's established
pattern for the same shared-scrollbar-id hazard (0x10000201, authored by
both the Chat and Config ListBoxes).
30 new tests (27 authored rows register as 30 IOptionRow instances — the
three toggle+slider trios each register two). Full Release suite:
13,107 passed / 4 skipped / 0 failed (was 13,083/4/0 — net +24, the one
existing RuntimeSettingsControllerTests case updated for SaveAudio's new
live-apply call, not a regression).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The retail port held under independent re-derivation: AddChild takes ONE
64-bit mask (@0x004862A0, the Society-row split-direction confirmation),
both spot-checked filter blocks byte-exact (12/13 rows, Gameplay
fall-through cases), the (0x16,2) enum-map order proven from
GetDIDFromEnum's body, AP-195's LED swap lands on state 6 of the exact
13x13 face child, self-sizing real, fixtures strictly additive (33
files, zero deletions).
M1: the opacity apply closures never refresh their OWN slider — drag
Default only, Reset: the value reverts (windows change) but the thumb
stays; Defaults self-heals only by both-rows luck. S1: SaveChatOpacity
does a full load+WriteAllText per drag TICK (dozens-to-hundreds of
synchronous JSON round-trips per drag; retail batches via the dirty
timer) — mark-dirty + flush on drag-end/Apply/hide. S2: the
whole-row-culling viewport makes a straddling 240/260px block VANISH
(no scissor stack), and the gate script asks the user to report exactly
that as a regression — false-failure generator + missing register row.
S3: no test for chatWindowMainFilter round-trip or the window-0 seed.
N11 (forward, OP6): 0x10000201 is authored under BOTH the Config and
Chat slots and the flat FindElement returns Chat's (last-write-wins);
OP5's linkage test passes through the flat lookup for the wrong reason.
Fix round queues behind the OP6 builder.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Binds LayoutDesc 0x2100005C through OP2's template-list mechanism and
OP3's per-page OptionPage model: the General Options header + two
DualHash-linked opacity sliders (Option_DefaultOpacity_Property
0x10000080 / Option_ActiveOpacity_Property 0x10000081, live-apply on
drag through RetailWindowOpacityController, defaults read from the
installed DAT's DBProperties collection at DID 0x78000001 via
ChatOptionsDatDefaults), and the five per-window text-filter blocks
(main window 12 rows minus Gameplay, four floaties 13 rows each — the
byte-verified authored order cross-checked against the raw
gmChatOptionsUI::InitOptions/AddCheckboxBitfield64Option pseudo-C, not
just the research doc's own table) writing AcDream.Core.Chat.
ChatWindowState directly, the same state CH6's chat windows already
read.
AP-195 retired: ported both halves left open at the OP2 re-review —
the ALL-set LED media swap (new UiButton.FaceFileOverride, driven by
the block-level P0x10000082/P0x10000083 sprites now threaded through
ElementInfo/DatWidgetFactory) and the CreateChildren self-sizing tail
(UiCheckboxBitfield64.Height grows with its stacked row content; the
enclosing ListBox reflows around the block's FINAL height via the new
UiTemplateListBox.AddPrebuiltRow, reusing the ListBox's own stacking
rather than a third stacking path). AP-187 broadened to cover the main
window's own filter (previously only the four floaties) and the new
live-editing write path.
The main chat window's filter (retail window id 8, ChatWindowState id
0) gains its own settings.json persistence (ChatSettings.
ChatWindowMainFilter) alongside the pre-existing floaty 1-4 fields;
opacity persistence is now wired on every live slider change, not only
through the old dev-scaffold Settings panel.
Fixture regeneration (ACDREAM_REGENERATE_UI_FIXTURES=1) picked up the
new ElementInfo.LedCheckedSprite/LedUncheckedSprite fields across all
19 committed layout fixtures — purely additive, confirmed against the
live installed DAT (0x10000520's own 0x82/0x83 properties resolve to
0x06004D17/0x06004D19 exactly as AP-195 documented).
Conformance: FilterRows/FilterBlocks pinned against the byte-verified
authored order and ChatWindowState's own default constants; the AP-195
LED swap and self-sizing behavior; the DAT opacity-default extraction
against the live installed DAT; live filter/opacity writes reaching
ChatWindowState/RetailWindowOpacityController; OnShown re-seed and
Reset/Defaults ghosting per the OP4 binding-pattern discipline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MF-1/M1 CLOSED with the blast scenario walked (Reset now restores live
values; UiLocked converges on every seed). MF-2 CLOSED with a
load-bearing fixture check (all nine verb buttons author Ghosted and
lack DAT 0x0B — TrySetRetailState would otherwise short-circuit). M2
store retirement verified to zero remaining references; M3 walked; S2's
.spv hygiene verified by recomputing the compiler hash scheme.
R1 MUST-FIX: the timestamp now prefixes ChatEntry.Text (the BODY), so
six of ten chat kinds render 'Alice says, "13:05:09 hi"' — retail
composes the line FIRST and prepends the timestamp as a separate string
at display time (AddTextToScroll @0x00563c50 receives composed lines;
fprintf("%ls%ls" ts, text) @0x00563e5b). Fix at ChatVM's display
composition; the two new tests pin the defect and must be rewritten.
R2: the seed-event re-read for OPEN panels (Combat vs Character can
disagree until re-shown; a panel open across reconnect reopens the
wrong-direction Reset for that window) — wire the existing
OnCharacterOptionsChanged hook like LockUI or register it. R3: no test
drives the refresh widget push. R4: headless AutoRepeatAttack default
flip false->true unnamed in AP-196.
Coordinator fixes directly (second round) once OP5 frees the tree.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Both OP4 reviews converged on one headline bug (Character-tab rows never
re-read live server truth after their pre-login constructor-word seed) plus
overlapping MUST-FIXes. All ten converged/consolidated findings land here:
MUST-FIX:
- BoolOptionRow.SaveCurrentValue now re-reads its live binding (retail's
GetValue()-into-SaveCurrentValue) on every OnShown — panel open, tab
switch in, initial activation — instead of trusting the pre-login
constructor word it was built with. Reset/tab-switch can now only
restore values that were actually live at the last show. LockUI's
host.Root.UiLocked one-shot mount seed now also converges on every
PlayerDescription via the existing OnCharacterOptionsChanged hook.
- Apply/Reset are wired to OptionPage.OnOptionChanged in production
(Ghosted when nothing changed, Normal when dirty, run once at bind so
both start disabled per retail's PostInit); Defaults stays ungated.
- The Combat panel's three LEDs (Repeat Attacks/Auto Target/Keep in View)
now read/write the same RuntimeCharacterOptionsState seam the Character
tab uses instead of a disconnected client-local GameplaySettings copy —
closes the "two writable copies" divergence. The three now-orphaned
GameplaySettings fields and RuntimeSettingsController's mirror
properties/SetCombatGameplay are deleted outright; the headless host's
hardcoded AutoRepeatAttack/AutoTarget now read the live option bit.
- RuntimeSettingsController.SetUiLocked's convergence guard now compares
against the last value actually applied to the runtime target instead
of the persisted GameplaySettings.LockUI snapshot, which could already
match a server-derived request without ever having been pushed.
SHOULD-FIX:
- DisplayTimeStamps now prefixes every chat producer (ChatLog.Append is
the one seam all of them funnel through), not just AddText's own
callers — heard speech, emotes, Turbine channels, and combat text were
previously missed. The prefix format escapes its colons and forces
InvariantCulture instead of the culture-dependent TimeSeparator
placeholder.
- sky.frag now honors uFogParams.w (fog mode) like the mesh/terrain
shaders, so Disable Distance Fog stops the sky dome's horizon band from
blending toward fog color too.
- Corrected the "byte-verified" overclaim on the timestamp format string
doc comment (BN-sourced, wire doc U6) and the AP-194 anchor-column
class-name typo; the RunAsDefaultMovement doc comments now cite retail's
actual acclient.h enumerator name.
- Added: DispatcherMovementInputSource's option x modifier truth table
(incl. || AutoRunActive with the option off), the per-page Apply/Reset
enable-gate tests, a real checkbox.OnClick/ToggleBehavior-driven click
test, and hash-pins for the six header string keys.
- Gate script step 8 corrected for the logout-flush false-failure
(closing the panel before relogging is load-bearing); a new step
documents the enable-gate sequence and the Combat-panel/Character-tab
cross-check.
Register: AP-196 (the Group-C default-source change + GameplaySettings
retirement) and AP-197 (the ignored per-character timestamp format
override) filed in this commit.
Full Release suite: 13,044 passed / 4 skipped / 0 failed (was 13,008/4/0;
net +36 tests from new coverage and legitimate assertion updates from the
GameplaySettings retirement).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MF-1 converges with blast M1: rows seed once at startup pre-login and
never re-read; retail's own schedule is the fix (UIOption_Checkbox::
SaveCurrentValue @0x004868E0 = m_current = GetValue() live, m_saved =
m_current, on every OnShown — predicted by OP3 review note N2). MF-2:
the OnOptionChanged Apply/Reset ghosting seam (Ghosted 0x0D / Normal,
PostInit runs it once so both start disabled) is wired by NOTHING in
production. The Defaults correction is independently VERIFIED from the
decomp (SetPlayerOption @0x00486e80 sets m_default from
GetDefaultOptionValue; InqDefaultGameplayOptionProperty's only caller is
the Chat tab's slider path — the plan's U1 directive would have made
Defaults restore nothing). Cross-page button collision confirmed from
the fixture (0x100001FC/FD/FE x3). 18 rows traced end-to-end; six header
hashes + 49 label globals hand-verified. SHOULD: culture-dependent
timestamp separator + overclaimed byte-verification; no
OnClick/ToggleBehavior-driven LED test; gate item 8's logout-flush false
failure. NOTEs incl. sky-dome fog and no mid-session echo repaint.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
M1: the Character tab binds at startup, pre-session — rows seed from
constructor words and nothing re-seeds on PlayerDescription; Reset (and
tab-switch, which is OnHidden->Reset) then publishes the STALE baseline
over server truth, and the one-shot UiLocked assignment can permanently
disagree with the radar's polled read. M2: the Combat panel's three LEDs
still write/read the orphaned GameplaySettings copy — decorative,
wire-less, and visibly divergent from the Character tab's rows for the
same PlayerOptions (the CH3 two-writable-copies mode); headless
hardcodes for the same ids filed alongside. M3: ToggleUiLock's
convergence guard (initialized converged, now comparing values from two
different stores) can early-return past ApplyUiLock — one press flips
the wire and radar but not the windows.
S1: AddText is NOT acdream's transcript chokepoint — speech/tells,
emotes, Turbine channels, combat text and @-replies write ChatLog
directly, so the timestamp prefix reaches only ServerMessage/WeenieError
lines and gate step 13's tell/say half cannot pass. S2: sky.vert never
reads the fog-disable param — crisp terrain against a fogged horizon
with the option on. S3/S4: missing register rows for the seven
default-source changes (ViewCombatTarget's observable default flips) and
the orphaned-but-written GameplaySettings; the timestamp format is
culture-dependent where retail strftime is literal. S6: the two changed
consumer files ship without their test files touched.
Clean: scoped per-page searches structurally sound (fixture-verified),
GetOptionBit thread-safe/allocation-free, the run-XOR byte-correct vs
SetHoldRun @0x006b3370 with wire encoding untouched.
Gate steps 3 and 13 cannot pass pre-fix — the OP4 gate must wait for
the fix round.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Binds LayoutDesc 0x21000028 (gmCharacterSettingsUI) through OP2's
template-list mechanism and OP3's OptionPage model: 6 authored group
headers + 50 toggle rows (49 from the 2013 build + D3's "Listen to PK
death messages", AP-193) in research doc §2's authored order, each row
resolved by PlayerOption id through CharacterOptionTable, seeded from
live RuntimeCharacterOptionsState, defaulted from CharacterOptionTable.
ClientDefault (byte-verified against UIOption_Checkbox::SetPlayerOption
@0x00486e80's own GetDefaultOptionValue call — AP-194 updated to confirm
the directive was followed), labels/tooltips resolved by name from
string table 0x23000003 (never hard-coded English), and registered with
OptionsPanelController.CharacterPage. Apply/Reset/Defaults
(0x100001FC/FD/FE) are now wired per-page via a scoped subtree search
(UiElement.FindDescendant, promoted from UiTabPanel) since Character/
Chat/Config each author their own physical instance under the SAME
element ids.
Consumers: Group A (29 ids) wire+store only via the existing
SetSingleCharacterOptionRuntimeCmd/TrySetOption seam. Group B: Display
Timestamps prefixes new transcript lines (RuntimeCommunicationState.
DisplayTimestampsSource); Disable Distance Fog forces FogMode.Off
(WeatherSystem.DisableDistanceFogSource, retiring half of TS-73); Run as
Default Movement inverts the walk-mode modifier's default
(RuntimeLocalPlayerMovementState.RunAsDefaultMovementSource). Group C
re-points AutoTarget/AutoRepeatAttack/ViewCombatTarget
(CharacterOptionCombatSettingsSource), VividTargetingIndicator/
CoordinatesOnRadar/LockUI/AcceptLootPermits from the client-local
GameplaySettings record to the canonical server bit — closing two
previously-unfiled divergences where AutoRepeatAttack and
AcceptCorpseLootingPermissions never reached the wire despite being
retail auto-save ids. TS-73 narrowed to its two still-open cases;
TS-75..TS-80 file the genuine gaps (no day/night force, no weather-
particle/profanity-filter/salvage/housing/pickup-preference subsystem,
fellowship-create's unaudited client-sourced field) rather than
inventing stand-ins.
Conformance: CharacterOptionsPageControllerTests pins all 50 rows
against CharacterOptionTable in both directions (an invented or dropped
row fails the build), the authored group/order row-by-row, and the
build/seed/Apply/Reset/Defaults/wire-publish behavior end-to-end against
the committed fixture. 52 new tests; full solution suite 13,008 passed /
4 skipped / 0 failed (was 12,956/4/0).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Evidence appended to the test script: run-1 diff-and-send exact to
contract; mid-run reconnect idempotence; run-3 cross-process persistence
proof (the fresh seed echoed the blob-only SalvageMultiple value);
graceful converged exits; no refusals, no pre-LoginComplete sends.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
R1: the gate script no longer promises a timestamp prefix on the Magic
macro lines — acdream renders no chat timestamps yet (the Display
Timestamps consumer is OP4 scope; no chat-log file exists, TS-69). A
bare light-blue transcript line is the CORRECT gate outcome.
R2: IsGrounded yields null (silent) for a NULL controller in player
mode — the prior pattern returned false and fired the mid-air refusal
retail cannot produce in that state; comments now match the code.
R3: the dormant-ActivePageChanged pin now applies the real stimulus —
every authored tab button on a dormant host must carry NO click handler
(RetailTabBinding.SetClick never ran), which is AD-73's actual dormancy
mechanism; SwitchTo deliberately has no guard.
OP3 is CLOSED: dual APPROVE-WITH-FIXES -> fix round 386076af ->
re-review REOPEN(narrow) -> this pass. Connected gate now READY.
Full Release suite: 12,956 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
TWO work products share this commit (a staged-index collision between the
coordinator's docs commit and the OP7 fixer's staged files — content
verified complete and coherent; only this message was wrong before the
amend):
1. OP7 review fixes (all nine findings from
docs/research/2026-08-11-op7-review.md):
- M1: HeadlessSessionDescriptor is a record; WithAccount uses 'with' non-destructive record copy,
so a future property cannot be silently dropped; direct-CLI
regression test proves CharacterOptions survives --user/--password.
- M2 root fix: LiveSessionEventRouter skips BOTH Replace and the
options notification on a trailer-truncated PlayerDescription — a
truncated re-seed can no longer install zeroed words under an armed
latch for OP7's automation to flush into 0x01A1.
- SF1: schema keys validate as ordinal strings against the allowed
names (numeric / comma-combined aliases rejected). SF2: both-true
fellowship exclusion rejected at load, naming both keys. SF3: the
onLoginCompleteSent observer moved after transit.EndTeleport().
SF4: production-hook coverage for all three LoginComplete sites.
SF5: test-script OP7 wire expectation corrected (batched ids ride
only the 0x01A1).
2. docs/research/2026-08-11-op3-rereview.md — OP3 re-review verdict
REOPEN (narrow): M1 byte-decode independently re-verified (6a 07 at
all six sites); residuals R1 (gate script promises a timestamp prefix
acdream doesn't render), R2 (null-controller player-mode still
refuses), R3 (dormancy pin lacks stimulus) — coordinator fixes follow.
Full Release suite at this tree: 12,956 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Consolidated fix round for the two OP3 dual-lens reviews
(docs/research/2026-08-11-op3-review-{mechanism,blast}.md), both
APPROVE-WITH-FIXES.
MUST-FIX:
- The six "Use Mouse Turning Settings" chat lines were typed
RetailLogTextType.ClientLocal (0x1A); retail types them 0x07 (Magic).
BYTE-VERIFIED against the PDB-paired binary at all six
gmConfigUI::SetMouseTurningDefaults call sites (0x0049E972/E9E2/EA52/
EAA4/EAF6/EB48): every site pushes `6a 07` (type=7) immediately before
the text-pointer push and the AddTextToScroll call. Added a dedicated
OptionsRuntimeBindings.DisplayMouseTurningMacroLine seam routed at
Magic (scrolling chat transcript, light blue, timestamped) instead of
the 4-slot SpewBox ClientLocal uses; the mid-air refusal and UA/RA
keep ClientLocal (both independently confirmed correct).
- Filed AD-77: the client-wide floating-only gmPanelUI host divergence
(retail also exposes a docked 0x21000017 host) the plan §5 delegated
to this review, scoped to every main panel, not just Options.
SHOULD-FIX:
- gmGameplayOptionsUI is not an OptionPage in retail (acclient.h:55857,
UIElement_Field). OptionsPanelController now constructs the Gameplay
slot's OptionPage with AfterApply deliberately null, so entering/
leaving that tab never publishes SaveCharacterOptionsRuntimeCmd.
Corrected OptionPageModel's doc comment and rewrote the two tests
that pinned the wrong (Gameplay-flushes) shape.
- Added the OptionPage.OnOptionChanged seam (PlayerOptionPage::
OnOptionChanged @0x004F27D0) — fires as the last step of Apply/
Reset/Defaults, plus once per live LED edit via a new
IOptionRow.AttachPageNotify hook (BoolOptionRow wires it into
SetCurrentValue only, matching retail's Apply(1)-only
HandleDialogAndNotices path). OP4-6 will bind Apply/Reset enable
state to this.
- Exit to Character Selection's mid-air refusal is now tri-state
(Func<bool?> IsGrounded): retail's UseTime only reaches the airborne
test inside `else if (smartbox->player)`, so outside player mode (or
with no live controller) the button is a SILENT no-op, not a
refusal. Fixed the inverted comment at both call sites.
- Options panel geometry now matches its nine gmPanelUI siblings
sharing RetailPanelUiController's one main-panel rectangle
(ResizeX=false, bottom-edge-only resize, no invented Min/MaxWidth/
Height) instead of being the only all-four-edge/horizontal-resize
outlier whose width silently reverted whenever a sibling was shown.
- Added the three missing test pins: Options/Character mutual
exclusion through a REAL RetailPanelUiController registration,
RetailDialogFactory.MakeConfirmation's omitted-queueKey overload
sharing DefaultQueueKey, and UiTabPanel.ActivePageChanged never
firing on a dormant (non-activated) host.
- TS-74's What/Where now names the five store-only CameraTurning
preferences explicitly instead of only mentioning them in Risk.
- Test script gains the toolbar-button ghosted->enabled+highlight
check, UseMouseTurning-survives-relogin and the five prefs-survive-
relaunch steps, a UA/RA legibility eye-item, and the corrected
bottom-edge-only geometry description for step 5.
One-liners fixed in files already touched: symmetric close-button
resolve-failure logging in OptionsPanelController.Bind (blast NOTE 8).
Full Release suite: 12,947 passed / 4 skipped / 0 failed (baseline
12,935/4/0 post-OP7 — 12 net new tests; the two OptionPageModelTests
"wrong-shape" tests were renamed/rewritten in place, not removed).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The 26-name allow-list cross-checks clean against research §5.2,
CharacterOptionTable, and retail's PlayerOption enum — all three risky
renames verified right (ToggleRun IS PlayerOption 0x0A run-as-default,
not the keybind latch). #368 honored; existing configs byte-identical;
per-route seeder leak-free; the Runtime observation hook never fires in
the graphical host.
MUST-FIX 1: HeadlessProcessHost.WithAccount hand-copies six of seven
descriptor fields — CharacterOptions is dropped, so the K3 direct-CLI
launch mode silently no-ops the whole feature. MUST-FIX 2: a truncated
PlayerDescription RE-seed re-opens OP1's wipe class — Replace installs
zeroed words while the latch stays armed from the earlier complete seed,
and OP7's automation then flushes those zeros into 0x01A1; fix at the
seam (a truncated parse installs nothing, notifies no one). SHOULD:
validate schema names as strings (Enum.TryParse accepts numeric and
comma-combined keys that alias into the allow-list); reject the
contradictory fellowship pair at load (declared both-true oscillates
against retail's mutual exclusion); move the mid-teleport hook to the
sequence tail; cover the three production LoginComplete hooks with the
real controller argument; correct the OP7 test-script line expecting a
0x0005 for batched SalvageMultiple (only the 0x01A1 carries it).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Adds an optional, strict `characterOptions` block to the headless bot config
(D8): keys are exactly the lane-B tier-1 (22) + tier-2 (4) bot-declarable
CharacterOptionId enum-member spellings; an unknown/out-of-tier name fails
config load naming the offending key, before it can ever reach the wire.
HeadlessCharacterOptionsSeeder diffs declared-vs-actual once both of ACE's
real preconditions are known true — GameActionLoginComplete sent (the
FirstEnterWorldDone gate SetCharacterOptions 0x01A1 needs) and a real
PlayerDescription has seeded RuntimeCharacterOptionsState
(HasServerSeed) — learned from whichever of two hooks lands second. Every
differing id routes through OP1's shared IRuntimeCharacterCommands seam:
auto-save ids send SetSingleOption (0x0005) immediately; batched ids also
call SetSingleOption (which only dirties the module) followed by exactly
one SaveOptions flush after the whole declared set has been walked.
Idempotent on reconnect by construction — no dedupe latch, the diff simply
finds nothing once the server agrees.
RuntimeLiveEntitySessionController gains a passive onLoginCompleteSent
observation hook (additive only, never changes when/whether it sends) so
the headless host can learn ACE's gate opened from any of its own two
internal send sites; the third site (direct first-entry completion) is
already owned by HeadlessSessionHost itself. All wiring is synchronous
delegate calls on Runtime's one dedicated update thread — no new
async/Task continuation, honoring #368.
Tests: schema (valid parse, unknown/tier-3 name rejected naming the key,
non-bool rejected, empty/absent no-op), the diff engine against a fake
IRuntimeCharacterCommands (nothing-to-send, auto-save-only, batched-with-
flush, mixed ordering, reconnect idempotence), and two wiring integration
tests — one dispatching a real PlayerDescription game event end-to-end to
a captured wire action, one proving the send lands on the same dedicated
thread every Tick runs on. Full solution suite: 12,935 passed / 4 skipped
/ 0 failed (+17 over baseline 12,918/4/0).
No register row: the characterOptions bot-config surface is acdream-
native tooling over retail's own wire mechanisms (both already ported by
OP1), not a retail UI port with a divergence to record.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Converged MUST-FIX: the six mouse-turning macro lines are typed
ClientLocal 0x1A where retail's AddTextToScroll sites pass type 7 (Magic,
light blue — mechanism lens decoded the stray [4]=7 writes at all six BN
call sites); 0x1A additionally routes to the SpewBox where the 4-slot cap
discards two of the six lines before they draw (blast lens). Fix: a typed
message seam, Magic for the macro, ClientLocal retained for the mid-air
refusal and AD-75; byte-verify the type argument during the fix.
Mechanism: the Gameplay page is NOT an OptionPage in retail
(gmGameplayOptionsUI : UIElement_Field, acclient.h:55857) — the
auto-flush-on-visibility acdream gave it flushes the blob at moments
retail would not, and two green tests pin the wrong shape; the
OnOptionChanged enable-gating seam (@0x004F27D0 — Apply/Reset gated,
Defaults never) is missing from the model; the mid-air refusal fires
where retail is silent (non-player-mode/null controller).
Blast: the plan-assigned docked-host (0x21000017) register ruling was
not filed — deemed a divergence, row owed client-wide; Options is the
only shared-geometry panel with ResizeX/four-edge/min-size, which the
shared _mainPanelGeometry silently reverts when siblings show; three
targeted seams lack pins (real-sibling mutual exclusion, DefaultQueueKey,
dormant ActivePageChanged silence); UA/RA legibility flagged for the
gate. Clean: exactly one toolbar button changed (previously ghosted),
F11 same-action since K.1c, no queue key invented, settings additive.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mounts retail's Options panel (LayoutDesc 0x2100002B resolved through host
0x2100006E slot 0x1000018D, gmPanelUI key 10) via the same catalog-import
pattern CharacterController already validates, registered through
RetailPanelUiController so it shares retail's "one active gmPanelUI child"
mutual exclusion with every other sibling panel for free. F11 and the
toolbar's options button (0x1000019B, already authoring panel id 10) both
now open it; the close button fires the same ToggleOptionsPanel action.
OptionPageModel (OptionPage/BoolOptionRow) ports retail's exact
Apply/Reset/Defaults/visibility semantics from
UIOption_Checkbox/PlayerOptionPage — LED clicks apply live immediately,
Apply commits every row unconditionally + flushes the batched blob, Reset
reverts only Changed rows, Defaults restores without committing, and
tab-switch/window-hide revert uncommitted edits. Wired for all four tabs;
this slice registers real rows on none of them (Gameplay authentically has
none — a pure button list). UiTabPanel gains an ActivePageChanged event so
the page model can hook every tab transition, including the initial
default-tab activation.
The seven Gameplay-tab buttons: Exit Game reuses the existing graceful
window-close path; Exit to Character Selection gets retail's confirmation
dialog and byte-verified mid-air refusal but still behaves as Exit Game
(AD-74 — no pre-world character-select flow exists); Configure Keyboard
and In-Game Help Files are inert this slice (AD-76 for Help — the
plugin retail depends on doesn't exist); Urgent Assistance/Report Abuse
short-circuit to their own byte-verified failure text through the
interface-text seam instead of ShellExecute against a dead URL (AD-75);
Use Mouse Turning Settings runs the pure MouseTurningSettingsMacro port,
persisting five new CameraTurningSettings preferences and sending
PlayerOption.UseMouseTurning — TS-74 records that acdream has no
persistent mouse-turning camera mode for the bit to drive yet.
Full Release suite: 12,918 passed / 4 skipped / 0 failed (baseline
12,871/4/0 — only new tests added).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Closes the mechanism lens's REOPEN (one MUST-FIX) and both lenses' small
residuals on the OP2 rework (b236a442); the blast lens re-review was
CLOSED outright. Fable-direct per the two-failure escalation rule.
- AP-195 filed: UIOption_CheckboxBitfield64 ports HALF of Refresh
@0x004859C0 — the ANY-set checkbox predicate is exact, but the ALL-set
LED media swap (P0x10000082=0x06004D17 / P0x10000083=0x06004D19) and
the ListBox self-sizing tail (ResizeTo/CalculatePaperSize — the block
IS a UIElement_ListBox in retail) are unported, and the block's row
stacking is a second divergent implementation beside UiTemplateListBox.
All due at OP5 before the Chat tab's connected gate; the IsSet doc
comment now names both halves instead of quoting only the ported one.
- Row tooltips: UiButton gains settable TooltipText surfaced through the
shared GetTooltipText hover pipeline (UiCatalogSlot's pattern);
UiCheckboxBitfield64.AddChild applies the row tooltip retail stamps in
CreateChildren @0x00485DF0, and documents that the 0x10000084 row-index
attribute stamp is deliberately replaced by the typed mask closure.
- AD-73 addendum: UiTemplateListBox.ConsumesDatChildren=true is inert
only while no authored Type-5 element carries children — that premise
is now conformance-PINNED across all 32 fixtures (a future DAT
regeneration surfacing an authored child fails the build instead of
silently dropping it).
- Plan doc: OP2's contract names UiTabPanel.cs (retail UIElement_Panel),
not the fictional-class-named UiTabControl.cs; ledger records OP1 and
OP2 both CLOSED.
Full Release suite: 12,871 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Focused re-review verifies all ten fix dispositions genuinely close their
findings (fellowship order re-derived from OnChanged @0x0059A8E0; the
seed latch confirmed upstream of every one of the three blob call sites;
TS-71 retirement and TS-73 accuracy checked). Residuals owed to a small
coordinator pass, none reopening: R1 closure hoisting defeats the
per-tick allocation guard (move the IsDirty check into the ctor
lambdas); R2 concurrent MarkDirty during a flush is silently erased by
the trailing clear (needs the dirty-generation token); R3 NOTE — a
truncated-trailer PlayerDescription (options words zero) arms the seed
latch, newly wire-reaching via the timer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Blast lens: the unguarded Type-8 arm re-classes five pre-existing
elements in four shipped panels (vendor backdrop sprite stops drawing;
character sheet + spellbook swallow clicks; IUiDatStateful propagation
severed), the Type-5 guard's premise is factually wrong (ten shipped
elements author non-empty 0x64), and the 27 stale fixtures make the
harness structurally blind to both — the green suite was true but
meaningless. Merge semantics, fixture provenance, hermeticity, and the
children-attached hook all verified clean.
Mechanism lens: readers decode correctly through the canonical
effective-state path; Type 8's real retail class is UIElement_Panel
(Update @0x0046BD00 — switching behavior confirmed faithful); both
claimed structural identities (UIOption_Slider = horizontal scrollbar,
UIOption_Menu = UiMenu shape) CONFIRMED from fixture fingerprints;
U10 closed (0x10000521 is the bitfield row template, consumed by
CreateChildren @0x00485DF0 via AddItemFromTemplateList(0)); TS-72 is
backwards — the toggle math is decomp-confirmed right, IsSet's all-bits
predicate is confirmed wrong (retail Refresh checks ANY mask bit).
Missing: GetChildRecursive resolution, no-0x32 means NO default switch,
SetupTabPageHash's malformed-entry filters.
Rework round 1 follows: UiTabControl/UiTemplateListBox become
UiDatElement subclasses with dormant behavior (controller-activated),
bitfield rows build from the authored template, IsSet goes any-bit,
all 32 fixtures regenerate as the acceptance gate.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mechanism lens: 53/53 table rows verified four independent ways; golden
vector recomputed byte-for-byte; 15/15 citations resolve. MUST-FIX: TS-71's
deferral rationale asserts a nonexistent obstacle — both hosts already
funnel one Runtime tick seam (LiveSessionController.Tick), so the 480 s
timer + logout flush wire with zero host edits. SHOULD-FIX: port
CPlayerModule::OnChanged cases 2/0x12 (fellowship mutual-exclusion emits a
second 0x0005); add the id-by-id 53-row word/mask pin.
Blast lens: CH3/CH4 seams bit-identical; routes single-write; reset clean;
the blob echo reads canonical J4.3/J4.5 owners (the important negative).
MUST-FIX: SaveOptions before the PlayerDescription seed would flush CLIENT
DEFAULTS over server options — needs a HasServerSeed latch (set by
Replace, required by TryFlush, cleared by ResetSession). SHOULD-FIX: router
test substitutes a fake binding for the production seam; flush callback
runs under _dirtyGate (deadlock with the router gate once the timer
wires); ledger blind to IsDirty; SaveOptions result encoding differs
between adapters; Replace leaves stale dirty state; no cross-check between
PlayerDescriptionParser enums and CharacterOptionTable.
Fix round follows as one consolidated pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Final of four Opus research lanes for the retail Options panel campaign:
- One LayoutDesc 0x2100002B owns the tab control (0x10000208, Type 8),
the four mounted tab pages, and the seven option-row templates. Pages:
gmGameplayOptionsUI 0x2100002A/0x10000202 (default tab),
gmCharacterSettingsUI 0x21000028/0x100001F9 (InitOptions @0x004A02F0),
gmChatOptionsUI 0x2100005C/0x1000050A (@0x0049FC60),
gmConfigUI 0x21000029/0x100001FF (@0x0049E400). Tab declarations are a
data table (property 0x2E structs {0x30 button, 0x31 page, 0x32
default}); rows build via UIElement_ListBox::AddItemFromTemplateList
against authored template lists in ListBox property 0x64 (all three
template arrays decoded).
- Open path: input action 0x1000001A ToggleOptionsPanel, retail-default
F11 (VERIFIED in retail-default.keymap.txt:148); toolbar button
0x1000019B authors P0x12=0x1000001A (VERIFIED in the committed
toolbar fixture); host is gmFloatyPanelUI 0x2100006E slot 0x1000018D —
the same floating shell CH6 ports.
- Apply/Reset/Defaults are PER TAB, and clicking an LED APPLIES
IMMEDIATELY (SetCurrentValue -> Apply(1)); Apply commits the undo
baseline + CPlayerModule::SaveToServer (flushes 0x01A1 if dirty);
Reset reverts to baseline; Defaults applies live without committing.
Hiding a page auto-reverts uncommitted edits; showing auto-applies.
- Chat tab fully enumerated (2 linked opacity sliders + 5 per-window
filter blocks, 13 checkbox masks byte-decoded; main window omits the
Gameplay checkbox - 12 rows vs floaties' 13). Config tab = 6 sections
/ 27 rows, ALL UserPreferences.ini-backed, nothing on the wire.
- Two new widgets needed: Type 8 tab control, Type 5 template-list
ListBox (both also needed by Configure Keyboard). U1: the Character
tab's Defaults behavior is genuinely unestablished (never calls
SetDefaultValue). U3: the 50th row (PK deaths) is a later-build
addition (DAT string exists, hash-verified). U7: Urgent Assistance /
Report Abuse ShellExecute dead support URLs — register-row candidates.
Research phase complete: all four lanes landed and seam-verified.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Third of four Opus research lanes for the retail Options panel campaign:
- Gameplay Options tab (gmGameplayOptionsUI::ListenToElementMessage
@0x0049E110): five of seven buttons have code handlers. Exit to Char
Selection -> local EndCharacterSession notice -> confirm dialog ->
airborne refusal (byte-verified) -> 0xF653 logoff; Exit Game -> the
epilogue path, no confirmation; Urgent Assistance / Report Abuse are
WEB LINKS in the EoR build (same support-site URL, byte-verified) —
the legacy wire paths still exist (0x0140 AbuseLogRequest: ACE names
it but has NO handler; Help channel 0x400 broadcast: ACE handles it);
Use Mouse Turning Settings is a 6-option macro (SetMouseTurningDefaults
@0x0049E8F0), not a screen — five client-local prefs + one server bit
(PlayerOption 0x31); Configure Keyboard / In-Game Help have no element
handler in the class (help = external ACHelpPlugin.dll via keystone).
- Configure Keyboard (gmKeyboardUI): six ActionClass list boxes, rows
from DAT ActionMaps (DBO 0x27), all 19 ID_InputMap_* strings
byte-verified; N-way cross-map conflict handling; storage is a LOCAL
.keymap file named in UserPreferences.ini, never wire-synced; Reset
reloads the DAT master maps (DIDs 0x14000000/0x14000002, which
tools/dump-keymap already extracts). retail-default.keymap.txt is a
user-saved keymap, not the DAT default.
- LOAD-BEARING CORRECTION (verified at the seams): the F11 SettingsPanel
is NOT rendered anywhere post-V11 — ToggleSettingsPanel() is an empty
no-op, the only IPanelRenderer implementation is the test fake, and
SettingsDevToolsComposition documents the keybinds.json fallback. The
retail Options panel is therefore acdream's FIRST shipping in-client
settings surface, and Configure Keyboard is the ONLY rebind screen —
it also clears Campaign V's carried panel debt (#258 adjacent).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two of the four Opus research lanes for the retail Options panel campaign
(docs/research/2026-08-10-settings-track-handoff.md):
- Lane B: the complete Character-tab option map. 50 rows / 6 groups
decomp-authored from gmCharacterSettingsUI::InitOptions @0x004a02f0
(the screenshots' PK-death row is 2015-client-only; the 2013 enum caps
at 0x33). Wire routing is retail's byte-verified lookup table
CPlayerModule::IsAutoSaveOption @0x0059a600 — ~21 ids send 0x0005
immediately, the rest dirty the module for the batched 0x01A1. Retail's
Defaults-button table reconstructs Options1 = 0x50C4A54A exactly
(independent confirmation of ACE's constant) and Options2 = 0x00008700
vs ACE creation's 0x00948700 (a real client-vs-server distinction, not
a bug). Per-option ACE handling + acdream consumer inventory included.
- Lane C: the real 0x01A1 body is PlayerModule::Pack @0x005D45C0
(builder CM_Character::Event_CharacterOptionsEvent @0x006A10C0), flag
enum PlayerModulePackHeader verbatim at acclient.h:7835;
SetPackHeader @0x005D44A0 always sets 0x460 and never 0x02/0x04/0x10/
0x80, so ACE's extra reader branches are dead legacy. Flush triggers:
Apply, logout, 480 s autosave. ACE stores options words raw, discards
the rest, refuses only pre-LoginComplete; unknown option ids THROW.
CH3 post-mortem: the deleted 16-byte builder put a CharacterOptions1
word in the section-flag slot.
Both lanes independently converged on the same latent defect: the
headless DirectGameRuntimeCommandAdapter.SetSingleOption sends the wire
but skips the local Options.SetOptionBit write the graphical path does
(LiveSessionRuntimeFactory.cs:348) — the CH4 stale-membership-gate bug
class reproduced on the bot side. Flagged for the campaign plan, not
fixed here (research-only lanes).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Collects the post-gate polish left uncommitted by the killed round-5 agent
(S1/S3 + review fixes N1/N3/N4) and completes the missing S2 half:
- S1: UiText multi-line transcript + colored-run label now submit EVERY
line/run's outline pass before ANY fill pass, matching retail's
UIElement_Text::DrawSelf @0x00467aa0 whole-block walk. DrawStringDatPass
is exposed for block-level batching; single lines keep DrawStringDat.
- S2 (completed this commit): authored outline 0x21/0x22 now reaches every
text-bearing widget — UiButton, UiDatElement, UiField, UiMeter, UiMenu,
UiCatalogSlot — seeded from the element's effective-default state exactly
like UiText (BuildButton lifts the label-bearing Text child's authored
value first, same chain as the label color). Per-STATE outline switching
(dialog/character/combat buttons author 0x21 in state 0x3 only) is NOT
ported — filed as register row AP-192 in this commit.
- S3: ChatWindowController reconciliation comment corrects the misread
indicator action ids 0x10000514-17 -> 0x10000114-17 and re-attributes the
id-coincidence to the pagination widget's m_prevButton/m_nextButton, not
gmFriendsUI; register + window-shell research doc corrected to match.
- N1: LayoutImporter's duplicate per-state any-state-first-wins 0x21 read is
deleted — ElementReader.ApplyCanonicalLegacyProjection's DirectState-then-
effective-default resolution is the single source (the duplicate would have
lit state-0x3-only outlines permanently once S2 widened consumption).
- N3: the outline pass tints with the outline color's OWN alpha, not the
fill's (retail tints m_curOutlineColor and m_curTextColor independently).
- N4: the outline-inflated glyph SOURCE rect is clamped to the atlas bounds
with matching dest shrink, porting CreateCharRectPair @0x00441480's edge
behavior — edge glyphs crop instead of sampling a neighbour's texels.
Full Release suite: 12,610 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Transcribes the user's four retail screenshots, inventories the existing
plumbing (CH3 option wire, CH6 UI machinery, the F11 surface), the
research questions, and the binding process lessons from Campaign CH.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Campaign CH round 4, user-gate items 1+2. Root cause: retail ships a
second (background) glyph atlas per font, dilated 2px on every side,
plus two border-pixel scalars (Font.NumHorizontalBorderPixels/
NumVerticalBorderPixels) that acdream's font reader never read — so
even the pre-existing outline parameter drew almost nothing once
enabled. Landed together (either half alone is a no-op or a
regression):
- UiDatFont carries BorderX/BorderY from the DAT font resource.
- UiRenderContext.DrawStringDat inflates the background blit's source
and destination rect by that margin and restructures into retail's
exact two-pass whole-string outline-then-fill model
(UIElement_Text::DrawSelf), plus the 8-neighbour +-1px fallback for
fonts with no background atlas. Corrects the stale "property 0xd"
comment to the real ids, 0x21 (Outline) / 0x22 (OutlineColor).
- LayoutDesc property 0x21/0x22 import (ElementInfo.Outline/
OutlineColor, LayoutImporter.ReadState, ElementReader.Merge/
ApplyCanonicalLegacyProjection, DatWidgetFactory.BuildText) so every
authored-outline element across the DAT set is correct at once.
- SpewBox: RetailFontId corrected from a round-3 heuristic
(0x40000025) to the actually-authored 0x40000001 (18px bold serif),
Outline=true set on the controller's UiText. Fill colour stays the
user-gate-round-1-pinned yellow — font atlases are alpha-only
(PFID_A8), so there is no baked shading that could explain the
screenshot's gold as anything other than the outline itself.
- Chat transcript: default fill now seeds from its authored
ARGB(255,204,204,204) instead of an unrelated color-table slot
(ChatTranscriptRenderer.BuildLines takes the transcript's own
DefaultColor as a parameter); the 34-entry LogTextType table is
untouched, and every existing CH1 conformance test stays green
unmodified.
Regenerated the committed chat_2100006f.json fixture from the real
installed DAT, confirming end to end (not by missing-field default)
that the transcript carries no outline.
Tests: font-reader border fields + inflation math pinned against the
real DAT font, two-pass draw ordering/tint/inflation via a new
TextRenderer.DebugSpriteSegmentVerts test seam, property 0x21/0x22
import at both the ElementReader.Merge and StateDesc-property layers,
SpewBox font/outline, and the chat default-shade seed with the color
table proven untouched.
Full Release suite: 12,610 passed / 4 skipped / 0 failed
(AcDream.slnx, complete solution).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Item 3 (#364): every honesty marker is now gone from user-visible /help
text. AllegianceOverview/HouseOverview's "[IMPLEMENTED]" tags and trailing
"Subcommands NOT marked..." sentences, and Day/Log/Render/Motd's appended
"NOT YET IMPLEMENTED in acdream" tails, are removed; the underlying retail
text is corrected/completed against the pseudo-C's own pristine
consolidated data dumps (Log and Motd had been silently truncated; Render
was entirely acdream-authored and is replaced with the real retail usage
string). The three PARTIAL /help group topics (channels/chatting/commands)
are now COMPLETE verbatim listings: HelpStupidChannelHack's three
"vtable slot" operands, previously believed undecodable, are the same
pooled/mislabeled-data artifact this campaign has hit before (AP-113's
precedent) — reading the function's own disassembly for the push imm32
preceding each constructor call resolves all three directly. messagetypes
is now a real ported construction (IsLegalChannel's 14-id whitelist +
LogTextTypeToString's name table + the exact join/wrap format) instead of
an acdream summary. Register row AP-184 retired.
Item 5: the main window's 1/2/3/4 indicator buttons now toggle their
floating chat window on click, per the user's retail memory overruling
the earlier decomp-only reading. UIElement_Button::HandleButtonClick has
its own generic click-driven action dispatch (property 0x12) reaching the
same DoVisibilityToggleAction the Alt+1..4 keybinds use; the button
fixture confirms this half is genuinely armed, but the floating-window
fixture authors no matching listener-registration property, so the
generic mechanism has no proven target in the data on hand. Per
CLAUDE.md, the user's retail memory is the axiom regardless:
ChatWindowController.BindIndicatorClicks wires each indicator's click
through the same ToggleFloatingChatWindow chokepoint the keybinds use,
as explicit user-directed retail behavior. SetIndicatorOpen stays the
sole writer of the Selected mirror so the visual stays consistent
through the click round trip.
Full reconciliation in docs/research/2026-08-09-chat-retail-window-shell.md
§1.4. Campaign plan gets the round-4 findings section; items 1+2
(text-style) are under parallel research, item 4 passed, item 6 deferred
to the settings track.
Suite: 12,579 passed / 4 skipped / 0 failed (Release, complete solution),
up from baseline 12,553/4/0 — net +26 tests, zero regressions.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
#368's entry now records the fix mechanism (dedicated
acdream-headless-update thread owning Start + every scheduler turn;
synchronous TimeProvider-timer scheduler loop; guard untouched, zero
shared Runtime changes) and the 3/3 live-ACE verification vs the 3/3
pre-fix quarantines. The #365 entry and diagnosis doc get dated
pointers: their open question is answered — the airborne residual
persists with threads provably single, refuting the
unsynchronized-thread hypothesis — and is split off as #370 with the
evidence and starting points.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ChatVM gains an OnInterfaceText hook + ShowInterfaceText(text), the
App-layer composition wires it to RuntimeCommunicationState.AddText,
and ChatCommandRouter routes every retail-0x1A command refusal through
it instead of the chat log's 0x00 sink. UI.Abstractions still never
references Runtime directly; unwired hosts (headless, tests) fall back
to the chat log tagged ClientLocal so no text is ever silently lost.
Reclassified per register row AP-183 (DoChannelList/On/Off, DoAllegiance,
DoHouseAvailableList — the last also corrected to retail's own bad-house-
type string instead of a synthesized "Usage:" line) and newly wired two
sites that previously showed nothing at all (DoStupidChannelHack's bare
legacy-channel-verb refusal, DoReply's message-but-no-last-teller
refusal). The generic bad-args fallback now resolves WeenieErrorMessages
0x026 ("That is not a valid command.", retail's HandleFailureEvent(0x26))
instead of synthesizing "Usage: {Usage}". DoSpeaker/DoEndurance/DoTitle
are untouched — already correct at 0x00.
Also closes#367 (DoHelp's "Unknown command" fallback and the degenerate-
prefix refusal now reach the SpewBox too) and retires register row
AP-186, whose own filing proposed exactly this seam shape.
Full Release suite: 12,542 passed / 4 skipped / 0 failed (baseline
12,466/4/0 at ff2784ea).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
BLOCKER: ChatSettings.DefaultOpacity shipped retail's base ChatInterface
value (0.5) as ONE shared global default applied to every
RetailWindowManager-registered window, not just the four floating chat
windows retail itself fades. That faded the whole out-of-box registered
UI (radar, vitals, toolbar, main chat, ...) to 50% opacity, including
several windows that can never take keyboard focus and so were stuck at
0.5 permanently. Fixed to gmMainChatUI's 1.0/1.0 override
(0x004CD0F0) instead — retail-identical opaque presentation for the 11
non-chat windows and the main chat window; only the four floating chat
windows now diverge from retail's 0.5-while-idle default, and the
Settings -> Chat transparency slider remains fully user-settable.
AP-190 reworded and gains two new decomp-verified clauses: (3) retail
eases opacity toward its target by 5% of the delta per tick
(ChatInterface::ListenToGlobalMessage @0x004F3840, armed from the focus
element-messages at @0x004F5275) where acdream snaps -- deferred, needs
a UI frame-tick hook the opacity controller doesn't have; (4) retail's
focus predicate is the chat ENTRY FIELD specifically
(ChatInterface::IsTextEntryFocused @0x004F30A0) where acdream uses
any-focusable-descendant. Both findings + the pre-existing UiMenu.cs
PushAlphaAbsolute(1f) popup bypass are folded into the window-shell
research doc's opacity section.
NITs: fixed the stale "text bypasses the alpha" comment in
UiElement.DrawSelfAndChildren (CH6c already routed DrawStringDat/
DrawString through the same ApplyAlpha chokepoint as sprites/rects);
added RetailWindowManager.WindowUnregistered + wired
RetailWindowOpacityController to detach and forget a window unregistered
while it held focus (previously only Dispose detached, leaking any
window unregistered mid-focus for the rest of the session); added
post-Dispose no-op guards to the three Set* opacity mutators; added a
DrawString (BitmapFont path) alpha regression test and a DrawStringDat
outline/background-pass alpha test (the existing tests only ever
exercised the foreground/fill pass).
Also fixes RuntimeSettingsControllerTests.SettingsViewModelSavePreserves
SectionAndTargetOrder's now-stale "target-chat-opacity:0.5:1" expectation
(caught by the full-suite run this fix requires) to match the new 1.0
default.
Campaign ledger CH6c row updated to APPROVE-WITH-FIXES.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Retail's ChatInterface::SetOpacity (0x004F3120) fades the WHOLE composited
window surface with one alpha; UiRenderContext.ApplyAlpha already gated
DrawSprite/DrawRect/DrawFill (since 1da697ec, pre-CH6) but DrawStringDat and
DrawString still passed applyAlpha:false, so text stayed sharp over a
translucent window. Both now route through the same chokepoint.
RetailWindowOpacityController (new) subscribes to a new
RetailWindowManager.WindowRegistered event and drives every registered
window's live Opacity from keyboard-focus state, applied to EVERY window
(chat, floaties, vitals, toolbar, ...) rather than retail's ChatInterface-only
scope — register row AP-190, retiring the stale AP-40 "fixed 0.75, no focus
transition" row in the same commit.
Verified retail's shipped opacity defaults from the decomp (constructor
literals, no cdb needed): the base ChatInterface ctor sets
DefaultOpacity=0.5/ActiveOpacity=1.0, kept unmodified by the four floating
windows; gmMainChatUI's own ctor overrides the main window to 1.0/1.0
(always fully opaque). acdream ships one shared global default (0.5/1.0)
rather than replicating the per-class override — also AP-190. The linking
invariant (raising default above active drags active UP; lowering active
below default drags default DOWN — never a clamp) is ported verbatim as
ChatOpacityLink in AcDream.UI.Abstractions, shared by the live controller
and the new Settings -> Chat tab's two linked opacity sliders.
Persistence: ChatSettings.DefaultOpacity/ActiveOpacity round-trip through
SettingsStore; Save pushes both through IRuntimeSettingsTargets.SetChatOpacity
into the live controller, no restart required.
Rider (CH6a/b re-review): strengthened the grip-media regression guard past
a bare SpriteFile != 0 check — ChatLayoutConformanceTests now drives each
live grip through a real UiRenderContext/TextRenderer (backed by the
in-memory RecordingGpuDevice test double) and asserts the draw call chain
actually queued sprite geometry, via a new TextRenderer.DebugSpriteSegments
test-only accessor.
Full Release suite 12,459 passed / 4 skipped / 0 failed (baseline
12,420/4/0). No subagents, no client launches (session hard constraints);
pending the next connected user gate for visual confirmation.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Applies docs/research/2026-08-10-ch6ab-review-findings.md in full:
- BLOCKER 1: UiResizeGrip now carries its ElementInfo/resolve pair and
draws its own authored DirectState media (a synthetic parameterless
grip still draws nothing, preserving existing resize-drag tests).
DatWidgetFactory.BuildResizeGrip threads resolve through. All seven
live grips on the main chat window now resolve a non-zero sprite,
restoring the visible borders/corners CH6a silently dropped.
- SHOULD-FIX 2: ChatWindowState gains BroadcastTargetWindow, a sentinel
distinct from every real window id (0-4), fixing the bug where the
main window's explicit-addressing branch coincided with the broadcast
check (both were literal 0). SetFilter's main-window no-op is dropped
— the main window's filter is now genuinely settable. ChatWindowController
.Bind takes a ChatWindowState (the same canonical instance the floating
windows already share) and GetTranscriptLines builds a real accept
predicate instead of accept:null. Verified safe: ClientLocal (0x1A)
never reaches ChatLog (AddText routes it to the SpewBox and returns),
so nothing observable regresses.
- SHOULD-FIX 3: UiButton.SuppressSelfToggle stops the four chat-window
indicator buttons (DAT property 0x0B=true, no retail click handler)
from flipping their own Selected mirror on a stray click.
- SHOULD-FIX 4: generated and committed chat_floaty_2100005b.json from
the real installed dats; added the permanent RetailLayoutFixtureGenerator
entry. All three flagged FloatingChatWindowController assumptions
(input field, title bar, close button) are confirmed correct against
real data — no controller code changes needed. New finding: unlike the
main window, ALL EIGHT floaty border/corner elements are live Type-9
grips (the floaty's own title bar is its move handle), so a floaty
window resizes from every edge and corner.
- SHOULD-FIX 5: register row AP-189 documents the shared-500-entry/
200-line-tail vs retail's per-window 10,000-line scrollback depth gap.
- NITs 1-5: documented the filter-persistence-only-on-/saveautoui
asymmetry and the reconnect-preserves-filters intent; corrected the
research doc's modifier-mask mislabel and the "ONLY function" false
superlative; moved WrapText off ChatWindowController onto
ChatTranscriptRenderer, closing the circular dependency.
Full Release suite: 12,420 passed / 4 skipped / 0 failed (baseline
12,392/4/0 at 22020ef2; net +28 tests, zero regressions).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Mounts retail's four floating chat windows as always-resident, born-hidden
children per gmGamePlayUI::SetupChildren @0x004E9EC0, all sharing LayoutDesc
0x2100005B (window ids 0x10000505/0x1000050E/0x1000050F/0x10000510). New
FloatingChatWindowController (AcDream.App/UI/Layout) binds each window's own
widget tree — built fresh per instance from one shared imported ElementInfo
— reusing ChatWindowController's word-wrap + retail color-carry algorithm via
the extracted ChatTranscriptRenderer instead of duplicating it. A floaty
window has no talk-focus menu (research doc §2.2), so its entry field always
sends on Say; the mismatch against retail's possible shared-channel behavior
is UNVERIFIED and filed as #369/AP-188.
Runtime owns the per-window filter/open state: ChatWindowState (new,
AcDream.Core.Chat) seeds retail's exact PostInit defaults per window
(window 1 0x0000101C Speech/Tell/DirectSend/Emote, window 2 0x00040C00
Social/SocialSend/Allegiance, window 3 0x00080000 Fellowship, window 4
0x78000000 Turbine General/Trade/LFG/Roleplay) and implements the full
ShouldDisplay(windowId, targetWindowId, logTextType) display predicate from
ChatInterface::RecvNotice_DisplayFinalStringInfo @0x004F4640. It lives on
RuntimeCommunicationState.ChatWindows so every host borrows the same
instance. The main window's filter (0xFBFFFFFF, "no user filter") never
actually gates anything because its own explicit-address branch already
covers every broadcast line — that's why UpdateFromPlayerModule early-returns
for window 0 in retail, ported here by construction rather than a special
case.
Keybind wiring: InputAction.ToggleFloatingChatWindow1..4 and their
KeyBindings.RetailDefaults() chords already existed since Phase K.1c
(unwired until now). The MetaKeys table confirms retail's default is Alt+1
through Alt+4 (index 3 = bit 0x00000004, cross-checked against the same
file's Alt+A/D strafe and Alt+Enter/Tab/F4 rows). Routes through
GameplayInputCommandController -> RetainedGameplayWindowCommands ->
RetailUiRuntime.ToggleFloatingChatWindow -> the generic UiHost.ToggleWindow,
whose visibility-change event is the single chokepoint that syncs
ChatWindowState.SetOpen and mirrors the main window's 1-4 indicator button
regardless of what changed a window's visibility (keybind, close button, or
a restored layout).
A direct decomp read of gmMainChatUI::ListenToElementMessage @0x004CDA80 —
the only function in the whole binary that branches on a click message —
settles what the research doc had left as a hedge: it handles exactly
0x1000046f (max/min) and the talk-focus menu's selection message, with NO
case for 0x10000522-0x10000525. The four indicator buttons are PURE
one-directional mirrors in retail; clicking them does nothing.
ChatWindowController.SetIndicatorOpen ports this with no OnClick at all.
Corrected research doc §1.4 accordingly.
Persistence is local-only (register row AP-187; the retail 0x1000008C
GameplayOptions wire remains deferred to CH6f): window geometry and
open/visible state ride the existing generic RetailWindowLayoutPersistence
path for free once each window registers under its own WindowNames entry;
the four filter masks get a dedicated ChatSettings round-trip
(ChatWindow1Filter..ChatWindow4Filter, defaulting to the retail PostInit
constants) loaded at mount and saved alongside SaveLayout().
Tests: ChatWindowStateTests (defaults, TypeIsActive, the full display-rule
matrix, toggle/reset, revision counter), FloatingChatWindowControllerTests
(bind smoke tests against a synthetic 0x2100005B tree, per-window filter
routing, filter-change cache invalidation, fixed-Say submit), new
ChatWindowController.SetIndicatorOpen tests (Highlight/Normal state,
cross-window isolation, range validation), GameplayInputCommandController
routing for the four toggle actions, and a SettingsStore filter round-trip.
Full Release suite: 12,392 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Root cause (measured live via ACDREAM_PROBE_PARK=1): HeadlessSessionWorldProjection
drove the first-entry conductor unconditionally, including while
HeadlessCollisionNeighborhood's own 3x3 publication plan held a genuinely open
RuntimeCollisionAdmission for the local player's landblock. Every
TrySealCollisionEvaluationAuthority attempt during that window failed
(IsCollisionEvaluationPrefixAdmissible false) and retried forever without
recovering — measured verdict: "seal-refused" repeating with no preceding
[rearm] verdict= line (the operation never even reached the AwaitingCell park).
This is the diagnosis doc's "structural half" mechanism; no evidence of the
"circular HasOldPrefixPlacementDebt" hypothesis was observed, so that shape
was not needed.
Step 1 (enabler): HeadlessStaticStateAudit.ValidateProcessIsolation now takes
sessionCount and only refuses process-global physics probes for
sessionCount > 1 — its own multi-root-attribution rationale never applied to
a single session, and it was blocking the exact probe built to diagnose this
class of stall.
Step 3a (root cause): new IHeadlessCollisionNeighborhood.IsQuiescent gates
ProjectSpawn/ProjectPosition/PumpFirstEntry's conductor-drive calls — the
conductor is never driven while the neighborhood's own publication owns
collision authority for that tick.
Step 4 (defense-in-depth): HeadlessLocalPlayerFrameHost.CanAdvancePlayer now
requires Controller.CanExecuteLiveMovement instead of just a non-null
controller — the headless-only gap that turned the (now-fixed) hydration
stall into a hard crash reaching SuspendObjectUpdate on a dormant controller.
RuntimeLocalPlayerFrameController's three shared entry points gained the same
guard, contract-preserving for the graphical host.
Verified end-to-end against live ACE (jump-probe policy, three runs):
hydration succeeds cleanly (136 entities load vs. 0 before), no seal-refused
spam, no crash from the original bug, graceful logout every time. Full
airborne-transition confirmation is blocked by a separate, newly-discovered,
pre-existing defect filed as #368 (the headless scheduler's
Task.Delay(...).ConfigureAwait(false) tick loop can resume on a different
ThreadPool thread mid collision-generation, tripping
EnsureCollisionMutationThread) — explicitly out of scope here, not mentioned
anywhere in the #365 diagnosis, and unsafe to fix without graphical-host
verification this session was constrained not to perform.
New tests: the real-admission hydration test (fails on the pre-Step-3a tree,
verified by temporarily reverting the three gates and confirming failure,
then restoring), the PumpFirstEntry quiescence-gate test, the
CanAdvancePlayer publication-lifecycle test, the dormant-controller
sabotage tests for RuntimeLocalPlayerFrameController, and the audit
single/multi-session tests. RuntimeLocalPlayerPhysicsPublicationStateTests
is untouched.
Full Release suite: 12,343 passed / 4 skipped / 0 failed (baseline ~12,330/4
plus 11 new tests).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
