move-cursor border band
- Empty-state text (round 3): the literal-\n split was correct but each
authored LINE rendered as one clipped run. Retail word-wraps each
authored line within the element extent (its GlyphList draw - the
same wrap RetailConfirmationDialogView already uses). Multiline
authored text now wraps through UiText.WrapWords against the widget's
LIVE width/font/color (cached per width+font+color, re-read per call).
Single-line authored labels keep their one-run shape - re-wrapping
every label is a client-wide change no gate asked for.
- Move cursor (round 3): "the frame won the hit-test" is not a border
test - windows whose interior is not fully covered by children (the
inventory panel's empty regions) resolve those pixels to the frame
too. The border is now a geometric 8 px band along the window's outer
edge, AND the frame must win the hit-test so border-adjacent content
keeps its own cursor. Resize-edge claim still takes precedence;
whole-surface dragging unchanged.
App suite 4,984/3 skips (new BuildText_MultilineAuthored wrap test).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
refused-drop yellow notice
Item 4 (confirmation dialogs missing text + names): the missing retail
mechanism was StringTable template substitution - an entry is N+1 literal
fragments interleaved with N named variables, composed by
StringTable::GetString @0x004300D0 (no-metalanguage branch @0x004303B7).
ACE sends the bare player name for types 1/4; retail's OWN CLIENT wraps
it. Ported as DatStringResolver.ResolveTemplate (PLAYER hash 0x05506DA2,
the exact compute_str_hash space; Chorizite stores the variable hashes
directly):
- Server-driven type 4 -> ID_Fellowship_FellowshipRequest, type 1 ->
ID_Allegiance_AcceptSwearConfirmation, injected into
GameplayConfirmationController; null resolve falls back to the bare
wire message, never invented English. The 2/3/5/6 " Continue?" family
never consults the composer.
- Local Swear/Break/Kick: the bind-time fragment-0 latch (which showed
the dangling "Do you wish to swear to ") is replaced by click-time
ResolveTemplate with the target's name.
All five templates verified token-free in the installed DAT - this is
NOT a StringTableMetaLanguage port (AD-81's engine caveat stands).
Item 5 (refused drop shows nothing; retail shows yellow top-center
text): the prevRequest latch was ALREADY ported (InventoryTransactionState);
what was missing was the consumer. InventoryTransactionState now raises
RequestFailed(request, weenieError) when a 0x00A0 clears the latch;
ItemInteractionController composes ServerSaysAttemptFailed @0x0058EAE0's
"The <item> can't be <verb>" (verb table + suffix map ported verbatim in
Core's InventoryFailureMessages, NAME_PLURAL for merge/split) and routes
it as LogTextType 0x1A ClientLocal -> the SpewBox, retail's yellow
top-center line. The dispatcher's second leg (@0x0055B342) also runs:
outside the 7-code exclusion set, WeenieErrorMessages resolves per-code
text/destination; 0x426 AttunedItem has no row in either place beyond
the verb line - faithful single-line output.
Register: AD-85 narrowed to its numeric-field item, AD-81 amended (the
token-free interleave is now ported; meta-token engine + FormatName
remain), AD-93 filed (wire-guid-match vs retail's latched-guid
preference; no Move/Wield latch kinds).
Tests: +2 InventoryTransactionState failure-latch, +5 ResolveTemplate
(constructed StringTable fixtures), +1 composer injection, +1 end-to-end
refused-drop line. Core 4,697/1 skip, App 4,983/3 skips.
Research: docs/research/2026-08-13-confirm-and-weenie-error-display.md
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
empty state, retail amber row selection
- Move cursor (user-directed, ALL windows): HoverWindowMove now
advertises only where the window frame element itself wins the
hit-test - its border pixels; interior points resolve to content
children. Matches retail's Dragbar-chrome-only move cursor.
Whole-surface dragging still works, it just does not advertise.
- Empty-state text (round 2): the DAT stores the LITERAL two-character
escape backslash-n (probe-verified - the dump printed the escape, not
line breaks), so the round-1 newline split never matched. Escapes are
normalized before splitting in DatWidgetFactory authored text.
- Selected fellow amber (user: "check retail"): probe-verified - the
row name band 0x10000282 AUTHORS the retail selected-row art
(DirectState 0x06001450 + Highlight 0x06001451, the amber). Selection
flips the band's ActiveState to Highlight; no invented tint.
App suite 4,976/3 skips. Confirmation-text + refused-drop-notification
research (the round's items 4-5) lands as part 2.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
One user-ordered batch across the FA social panel + world selection.
Every root cause was probe-proven before the fix (new
ProbeSocialClickRouting in SocialPanelLiveMountProbeTests - production
window mount + real UiRoot hit-tests + a synthetic click):
1. STUCK CHECKBOXES (fellowship x4, allegiance x1, "always checked /
can't change any options"): the authored checkboxes carry DAT
ToggleBehavior, so UiButton SELF-FLIPS Selected at MouseUp - the old
handlers read the flipped value and wrote the ORIGINAL back, snapping
every click to where it started (the probe recorded (id, oldValue)).
Fix: SuppressSelfToggle (the CH6a/b mirror discipline) + derive the
next value from the STORE; the per-tick seeding mirrors it back.
2. UNCLICKABLE ROSTER ROWS ("only get the move window cursor"): the row
name text is display-text ClickThrough=true, which the hit-test walk
skips regardless of HandlesClick - the wired OnClick was unreachable.
Fix: UiText.OnClick assignment now clears ClickThrough (central,
documented); the stats text gains the same select handler so most of
the row's width selects the fellow.
3. TRUNCATED EMPTY-STATE ("You do not belong... To create MISSING"):
the authored string resolves COMPLETE (three sentences) but embedded
'\n's rendered as one clipped line. DatWidgetFactory now splits
authored strings into one Line per newline, with the provider still
re-reading DefaultColor live (the state-color contract - caught by
BuildText_AuthoredLineTracksStateFontColor).
4. FELLOW NAMES WHITE (user-directed): the AD-82 invented leader-gold +
selection-blue tints are deleted; names always white (register row
narrowed).
5. ALLEGIANCE HEADER LABELS: bare "0"/"0" -> "Followers: N" / "Rank: [N]"
(user-specified format; the full retail StringInfo composition stays
AD-85's gap), monarch block matching.
6. FRIENDS/SQUELCH LIVE (AD-79 mostly retired): Add friend (name box ->
0x0018, retail clears the box - Request_AddFriend @0x0048D240),
Remove (row-click selection -> 0x0017), Appear Offline (CharacterOption
0x27 via the immediate 0x0005 auto-save, ACE pushes FriendStatusChanged
to your friend-of list), Squelch Character/Account add-by-name
(0x0058 guid0/type AllChannels + 0x0059) and Remove for the selected
row. The wire beneath (builders, WorldSession sends, Runtime commands,
parsers) existed end-to-end since J4.1/FA1 - this is panel wiring only
(docs/research/2026-08-13-social-wire-completion.md, committed here).
Send Tell stays inert (not in the order; AD-79's remainder).
7. WORLD SELF-SELECTION ("clicking my own char should select myself"):
retail has NO self-exclusion (CPhysicsPart::Draw @0x0050D823 arms
every physobj; RecvNotice_SmartBoxObjectFound @0x004E5BAE selects
unconditionally) - the includeSelf gate was an unregistered
divergence, now removed on both the left-click and right-click paths.
Element roles were probe-measured, never guessed (Add 0x10000514 /
Remove 0x10000515 / Send Tell 0x10000516 / Appear Offline 0x1000052C /
name field 0x1000051B; Squelch: field 0x10000540, Remove 0x10000547,
Squelch Character 0x1000054B, Squelch Account 0x1000054C).
Register: AD-79 mostly retired, AD-82 narrowed. Known remainder, filed
not hidden: the fellowship page's authored 600px content vs the 362px
viewport leaves Dismiss/Assign-Leader below the fold until the window is
resized taller (probe-measured; candidate follow-up).
Tests: Checkbox_Click fact rewritten to the mirror contract (both
directions), monarch-followers label updated, includeSelf expectation
updated, probe extended (click routing, synthetic click, action-widget
role dump). App suite 4,976/3 skips.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
position memory, unified monitor, maximized restore; AD-92
Dual-lens Opus review of e56aa511 (reports committed under
docs/research/). The consolidated corrections:
- Mechanism M1 (load-bearing): on Windows, Silk's GLFW error callback
QUEUES exceptions on a static list instead of throwing - they detonate
later at window close, which is exactly #388's original two-stage
crash shape. catch(GlfwException) was dead code here and a failed
SetWindowMonitor "succeeded". Success is now judged by the NATIVE
POST-CONDITION (GetWindowMonitor after the call) on both enter and
exit; the catches remain only for the throwing platforms.
- M2 (both lenses): same-mode fullscreen re-apply is a no-op BEFORE any
native work (new IDisplayModeSwitcher.CurrentFullscreenMode). Every
Display-backed Config row applies per change - sliders per DRAG TICK -
so without this every tick while fullscreen re-issued a real
display-mode change.
- M3/M5 (both): the remembered windowed placement is process state (two
target instances exist - startup and live-save); a fullscreen boot now
exits through either instance to the real placement, not the (60,60)
literal.
- M4 (both): the switcher resolves the WINDOW'S monitor (attached
monitor when fullscreen, else IWindow.Monitor's index into the GLFW
array - the same monitor DisplayModeCatalog enumerated), primary only
as a last resort; the offered-list/switch-target mismatch is gone.
- Blast M2b: the offered-mode validator falls back to the SAME static
ladder the dropdown falls back to - Full Screen is no longer a
permanent silent no-op on catalog-less hosts (the switcher's own
monitor-mode-list check remains the hard guard).
- Blast M3: a windowed pick on a MAXIMIZED window restores it first
(Size writes are silently ignored while maximized; the deleted
WindowState=Normal write used to do this incidentally). New
IWindowedSizeSurface.IsMaximized/Restore.
- Mechanism M5: no silent bail-outs - the unparseable-resolution
fullscreen path logs, and the failure line no longer claims "staying
windowed" when the state is unchanged (#392 noted inline).
- Q1 nit: one cached Glfw wrapper (per-call GetApi allocated + took a
native refcount); IsFullscreen/CurrentFullscreenMode guarded.
- AD-92: highest-refresh-for-WxH + refuse-and-log versus retail's
pass-through-and-error ForceDisplayResolution.
Known-open tail, filed not hidden: #392 (persisted-flag divergence on a
refused enter - needs an apply-result seam); the mechanism report's
pacing-refresh WATCH rides the same seam.
Tests: +3 (same-mode no-op, unparseable-while-fullscreen refusal,
maximized restore-before-write). App suite 4,975/3 skips.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Slice 5+6 of the display block, one coherent unit (they share the state
machine the goal's dual review covers).
GlfwDisplayModeSwitcher (#376) ports retail's fullscreen semantics -
Device::ForceDisplayResolution @gmClient::Init 0x004047af is a REAL
video-mode change - through native glfwSetWindowMonitor on the same
IWindow.Native.Glfw handle path #348's cursor cache proved. Primary
monitor (retail's primary display device); refresh = the monitor's
highest for the picked WxH; the windowed placement is remembered for the
exit path; every failure is a no-throw (bool, reason) result.
SilkRuntimeDisplayWindowTarget.Apply (#388) becomes the state-aware
machine: fullscreen target = validated native mode switch (mode must be
in #391's DisplayModeCatalog - an offered mode is supported by
construction, making the "Graphics mode not supported" crash class
unreachable from the dropdown); windowed target while fullscreen = the
native exit (which sets the client size itself); plain windowed pick =
the proven #387 size write. A raw Size write NEVER happens against a
fullscreen window - on GLFW that is a video-mode request, and an
unsupported one was the exact unhandled-GlfwException that killed the
user's 2026-08-13 session. The old Silk borderless WindowState path is
deleted from the apply. New IWindowedSizeSurface narrows the window
dependency so the machine is unit-testable (FakePacingSurface idiom).
Live-verified on this machine (goal-sanctioned automated run):
display: fullscreen mode switch 1920x1080@300 -> framebuffer resize
event 1920x1080 -> vulkan: swapchain recreated 1920x1080 ok=True ->
graceful close, desktop mode restored.
Tests: 5 state-machine facts (validated switch/never-size-write,
unoffered refusal, failed-switch usability, native exit, plain windowed
write). App suite 4,972/3 skips. Gate script sections D4-D6 written
(black-screen-risk steps flagged). Dual Opus review of the pair follows
as its own round.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Decomp-first per the block's rule: the research doc
(docs/research/2026-08-13-retail-ui-display-change.md, committed here)
pulled retail's actual mechanism before any code. A display change runs
UIElementManager::RefreshEvent @0x0045C530 ->
UIElement::UpdateForParentSizeChange @0x00462640, which unconditionally
re-applies every floating window's own clamping MoveTo override
(x = max(0, min(x, parentW - selfW)) - top-left priority, oversized
windows pin to 0), then broadcasts global message 0xE whose sole
listener reloads the per-resolution auto layout. No proportional moves,
no resets; retail saves layouts only via @saveui.
Port: RetailWindowLayoutPersistence.ClampAllToScreen() is the cascade
clamp (no store I/O; _restoring suppresses the per-move save so a live
drag-resize cannot write settings.json per frame), and
RetailUiRuntime.Draw carries a two-step screen-size edge detector:
change frame -> clamp; first stable frame -> one
RestoreAll(saveBack:false) per-resolution reload (the 0xE analog; no
lazy save-back, matching retail's save-only-on-command). The login
restore path already used retail's exact clamp math (Apply) - the live
trigger was the missing half, which is precisely the stranding the user
reported.
Deliberate deviation, register AD-91: retail's gmFloatyChatUI windows
have NO clamp and can strand; the block's requirement ("UI windows must
stay reachable") clamps every registered window uniformly.
Tests: 5 new persistence facts (clamp/top-left-pin/no-move/no-save-on-
clamp/no-save-on-live-reload). App suite 4,967/3 skips. Gate script
section D3 filled in.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Dual-lens Opus review of 7e0c1303 (reports committed under
docs/research/). The law, gate, and vertical application are CONFIRMED
at instruction-byte level against the PDB-paired acclient.exe (the BN
text FPU-elides this whole area); the fix round addresses the findings:
- Blast MUST-FIX 1: real schema migration instead of a hand-edited dev
file. SettingsStore v2->v3: a pre-v3 display.fieldOfView was the
applied vertical FOV in degrees; v3 means retail's m_fGameFOV.
LoadDisplay migrates on read - the untouched old default 60 maps to
the retail default 90; a deliberate other value preserves its visible
16:9 framing (x (16/9 - 0.1)), clamped to the registered [10,160];
the next save stamps v3 and migration never reruns. The dev
settings.json hand-edit was reverted so the migration owns it.
- Blast MUST-FIX 2 / mechanism M2: the Field of View now applies LIVE on
Save (retail: Render::GRPCallback_OnRenderPreferenceChanged @0x0054d999
-> SmartBox::SetDefaultFov). RuntimeSettingsTargets gains the camera
graph and applies through ApplyDisplayWindowState - the update-phase
seam, deliberately NOT the render-phase preview path (the review's
WATCH-3 cull-vs-raster landmine).
- Mechanism M1 -> register row AD-90: retail's divisor aspect runs
through the Render.AspectRatio preference (ComputeAspectForViewport
@0x0054f150, (w/h) x pref x 0.75) - exactly raw w/h at the registered
default, which is what acdream assumes; retail's NaN-through-the-gate
quirk (M3) is folded into the same row as deliberately not reproduced.
- Docs: RetailFieldOfView now cites the decisive vertical proof
(D3DXMatrixPerspectiveFovLH fovy slot @0x0059ab71), the unconditional
SmartBox::RenderNormalMode site, and M4's exact horizontal numbers
(89.0/83.9/80.6 deg); the Config FOV row comment updated to LIVE.
- Blast WATCH 4 disposition: the 15 replay-harness PI/3 constants stay -
they are CAPTURE-TIME camera parameters for recorded fixtures, not
production framing; changing them would invalidate the replays.
Tests: +6 SettingsStore migration facts, +1 live-apply fact.
App suite 4,962/3 skips; UI.Abstractions 922.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
User-directed (2026-08-13): "we should only support modern resolutions.
Not any old format." New DisplayModeCatalog enumerates the window's
monitor (Silk IMonitor.GetAllVideoModes) once at GameWindow load and
curates via a pure, tested rule: modern widescreen families only
(16:9/16:10/21:9/32:9 within 2.5%), at least 1280 wide, must fit the
desktop (an impossible windowed pick is not offered - the measured
3840x2160-on-2560x1440 silent clamp class), desktop mode always
included, refresh-rate duplicates collapsed, ascending order.
The Config Resolution row consumes the catalog through two new optional
Bind parameters; its Defaults value becomes the desktop's own mode.
Fixture/headless callers keep the static preset ladder, which now drops
800x600 and is pinned by test to pass the same curation rule (the OP6 S4
"default must be re-selectable" invariant holds on both paths).
Deliberate retail deviation, register row IA-22: retail listed the
adapter's complete enumeration including 4:3 legacy modes and authored
800x600 as the Config default (gmConfigUI::InitOptions
SetDefaultValue(0x03200258); gmClient::Init @0x004047af). The catalog is
also the designated fullscreen mode-switch validation source for
#376/#388 - an offered mode is supported by construction.
Tests: DisplayModeCatalogTests (8 - filter/clamp/dedupe/sort/ultrawide/
desktop-inclusion/fallback-consistency); ConfigOptionsPageControllerTests
row-12 default updated. App suite 4,961/3 skips; UI.Abstractions 916.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Retail's world-camera FOV is not a constant: the applied vertical FOV is
m_fGameFOV / (viewportAspect - 0.1), recomputed on every aspect or
game-FOV change (CreatureMode smartbox sites 0x00452b2f/0x00453b14),
gated by Render::SetFOVRad's open (0, pi) acceptance (0x0054b2d0 -
rejected results keep the previous FOV). m_fGameFOV defaults to pi/2 =
90 degrees (0x00454649) and is what the Field of View option sets in
degrees (0x00451e6a; registered range [10,160] default 90 -
gmClient::InitUIPreferences @0x004035b0). Net effect: the horizontal
view stays ~85-90 degrees across aspect ratios; wide screens trim the
vertical slice instead of ballooning the sides.
acdream hardcoded FovY = pi/3 = 60 degrees on all four world cameras,
aspect-independent, and the Config slider wrote raw vertical-FOV
degrees. New: RetailFieldOfView (the law + gate, decomp-cited),
CameraController.GameFovRadians + SetGameFov + one ApplyProjection
chokepoint recomputing every camera on SetAspect/SetGameFov/
EnterChaseMode/RestoreState; ApplyFieldOfView now feeds the law;
DisplaySettings.Default.FieldOfView 60 -> 90 (the retail registered
default; the stored number changed MEANING with this commit).
The same seam closes a second latent bug the 2026-08-13 "squished" gate
report exposed: SetAspect only ever updated Orbit/Fly - the CHASE
cameras (the ones the player looks through) kept their creation-time
aspect across every mid-session resize, drawing the world at the old
shape stretched onto the new viewport.
The paperdoll camera stays outside the law by design (retail portrait
mode is UseSharpMode, not smartbox - DollCamera's own doc).
Tests: RetailFieldOfViewTests (golden law values at 4:3/16:9/21:9, the
constant-horizontal property, the rejection gate, controller propagation
incl. chase attach/restore + rejected-law aspect-still-propagates);
DisplaySettingsTests + RuntimeSettingsControllerTests updated to the new
semantics. App suite 4,953/3 skips; UI.Abstractions 916/0. AD-89 retired
in this commit; user settings.json migrated 60->90 by hand (stale
pre-port default).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Both from the 2026-08-13 display gate session. #389 carries the full
decomp-verified retail FOV law; #390 requires the retail reposition
mechanism from the decomp before any clamp is implemented.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
GameWindowSlice8BoundaryTests.FramebufferResize_IsOneTypedOwnerHandoff
correctly rejected the log line added to GameWindow.OnFramebufferResize
— the window callback is contractually a one-line handoff. The line now
lives in FramebufferResizeController.Resize after its zero-size gate,
which is also the better home (one owner, all callers covered). Full
Debug App suite 4,941/3 skips.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Three rare-event log lines so any future resize report is diagnosable
from the launch log alone: 'window: framebuffer resize event WxH',
'vulkan: swapchain recreated WxH ok=', and 'display: resolution pick
WxH (window was WxH)'. An instrumented live run on the Windows AMD box
shows the full chain firing for both the programmatic resolution apply
and external window resizes, and screen captures at 784x561 vs 1584x861
confirm fixed-pixel UI with a true pixel-count re-render.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
User report: resolution picks (and window drags) stretched the image
instead of changing the pixel count. Root cause: Campaign V slice V11
deleted the GL viewport target and left a null target, assuming the
driver's OUT_OF_DATE/SUBOPTIMAL acquire/present results would drive
swapchain recreation on resize. That is driver-dependent and
spec-insufficient — this machine's Windows AMD driver keeps presenting
the stale-extent swapchain scaled to the new window indefinitely, so
OnFramebufferResize only ever updated the camera aspect while every
pass (UI included) kept rendering at the old extent.
Fix: SwapchainRecreateViewportTarget implements the existing
IFramebufferViewportTarget seam for Vulkan and arms
VulkanGraphicsContext.RequestRecreate() on every resize event; the next
PrepareFrame rebuilds the swapchain at the live FramebufferSize (bursts
collapse to one recreation, stale events cannot install a stale extent,
minimised sizes stay gated by FramebufferResizeController).
Tests: SwapchainRecreateViewportTargetTests (target contract, size-
agnostic arming, null hook, controller-to-target end-to-end with the
minimised gate). Full Debug App suite 4,941/3 skips.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
LandblockStreamer.HandleJob's near-payload check is "fail loud in Debug
builds and strip in Release" (its own comment, with a Debug.Assert at the
check). FarLoad_StripsEnvCellsAndPhysicsEvenWhenEntityListIsAlreadyEmpty
feeds a deliberately-buggy far factory to verify the Release strip — so
under Debug the assert fires, the test host's listener turns it into an
exception, and the job publishes Failed BY DESIGN. The test asserted the
Release outcome unconditionally and therefore failed on every full Debug
App run (found 2026-08-13 during the #385 session; every campaign gate
runs Release, which is why it never surfaced). It now asserts the Failed
result + assert message under DEBUG and the strip under Release. Verified
green in both configs; full Debug App suite 4,937/3 skips.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
User gate report (Campaign OP happy-testing round, 2026-08-13): every
Config-tab dropdown drew its text gold + left-aligned and its popup a
fixed 6 rows regardless of item count. All three were unmeasured styling
divergences — the authored data (new probe menuprobe3, live DAT) says:
- button label child 0x10000355: fontColor white, hJustify=Center
- row template 0x1000035A: fontColor white, hJustify=Center
- popup ListBox 0x10000358: edge-docked L=T=R=B=1, the authored condition
arming retail UIElement_Menu::RecalculatePopupSize @0x0046caf0 —
popup resizes to the ListBox's summed content height, uncapped
(0x0046e5f4..0046e66c via ResizeScrollableArea's 0x32 broadcast)
UiMenu gains three opt-in properties (ButtonTextCentered,
ItemTextCentered, PopupSizeToContent) plus retail Open @0x0046cc42's
empty-list gate; chat + vendor keep the class defaults, so their shipped
behavior is untouched. ConfigOptionsPageController.ApplyMenuChrome wires
all four corrections for the 8 Config menus with the probe citation.
The same probe found vendor's authored popup ListBox is ALSO docked while
our vendor dropdown ships G5's fixed 6-row window — filed as #386 +
register row AD-88 (UNCLEAR: the G5 retail screenshot and the decomp
mechanism conflict) instead of silently reworking a user-gated surface.
The "resolution change resizes the window" observation from the same
report is #374's designed windowed-mode behavior (display-mode switching
is #376/#377) — no change.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The retail four-tab social panel (Fellowship & Allegiance) is
code-complete: all six slices landed and reviewed (dual-lens Opus review
-> fix round -> narrow re-review each). Fellowship two-session flow proven
live (FA6 bot gate PASSED). Closeout bookkeeping:
- register AD count 66 -> 67 (AD-87, the deferred allegiance bot gate);
- plan status flipped to CODE-COMPLETE with the OWED connected gates +
#384 (allegiance-swear ACE non-response) called out;
- CLAUDE.md Current-state gains the Campaign FA paragraph
(per feedback_claude_md_staleness), pointing at the memory digest.
Owed: the user's connected gates (§FA3-§FA6 of
docs/research/2026-08-12-campaign-fa-test-script.md) and #384's
ACE-console disambiguation. Full suite 13,304/4/0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Plan ledger: fellowship two-session automated gate PASSED live 2026-08-12
(five of six runs reproduced the decisive cross-session assertion); the
allegiance bot gate is DEFERRED behind AllegianceGateEnabled=false pending
docs/ISSUES.md #384, with commit citations for every fix this slice landed
(confirmation relay, name-matched proximity, the fellowship-only
finalization).
Gate script §FA6: the fellowship automated-gate recipe + actual PASSED
result (the two-session config, the six proof points per stage, the
literal decisive-assertion log lines), the allegiance deferral writeup,
and a new [TWO-CLIENT] manual step (25) the user's own connected gate can
run to help disambiguate #384 (ACE-side rule vs wire-builder defect vs
harness-specific drop) using two real graphical clients instead of the
testaccount/testaccount2 pair.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
docs/ISSUES.md #384 records the live-run evidence trail (six connected
runs, the 0.005 m distance diagnostic, the confirmation-arrival diagnostic
that never fires) behind AllegianceGateEnabled=false.
docs/architecture/retail-divergence-register.md AD-87 records the honest
divergence this deferral creates: the allegiance half of the FA6 bot gate
is written and wired but unverified end-to-end over the wire, unlike the
fellowship half which is proven live.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Six live runs against local ACE (testaccount/+Acdream as Leader,
testaccount2/+Horan as Recruit) converged on a clean split:
- FELLOWSHIP two-session gate PASSES live, reproduced in three separate
runs. The decisive cross-session assertion (the Recruit bot's own
RuntimeFellowshipState — a separate process's canonical Runtime owner,
not the Leader's local echo — flipping IsInFellowship=true,
MemberCount=2, LeaderGuid=<Leader>) holds every time. This ships as the
automated gate.
- ALLEGIANCE swear never completes: ACE returns nothing at all to
Event_SwearAllegiance (0x001D) — no 0x0274 confirmation, no 0x0020 tree
update, no WeenieError — even at 0.005 m separation (run6's distance
diagnostic ruled out retail's 2.0 m swear-distance gate). Ambiguous
between an FA1 wire-builder defect, an ACE-side rule this test pair
trips, or a drop; disambiguating needs an ACE server console this
harness doesn't have. Filed as docs/ISSUES.md #384 and
docs/architecture/retail-divergence-register.md AD-87.
AllegianceGateEnabled (static readonly, not const, to avoid a CS0162
unreachable-code build error from branching on a literal) gates every
allegiance-dependent stage in BOTH policy classes off by default:
Leader's WaitForVassal (skipped straight to the reconnect+teardown that
only need fellowship state), Recruit's Swear/WaitSwornSeed/Break/
WaitBrokenSeed (same). All of that code stays fully written and wired —
flipping the flag re-enables it for a follow-up investigation once #384
closes. WaitReconnectReseed on both sides now asserts fellowship-only
re-seeding when the flag is off, preserving the reconnect-idempotence
proof independent of the allegiance blocker.
The two live-run diagnostics added while investigating #384 (the
confirmation-arrival log line in HeadlessSessionHost's
OnConfirmationRequest, and LogDistanceToPatron in the Recruit policy) are
kept as permanent, clearly-labeled evidence for whoever reopens#384 —
neither is "TEMP, strip later."
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The second live gate run exposed a real environmental hazard: this shared
ACE dev instance has a THIRD player character online (+Je, guid
0x50000001), and after @teleallto it ended up nearer to the Leader bot than
the actual Recruit bot (+Horan, 0x5000000B). RuntimeFriendlyTargetQuery.
FindClosestOtherPlayer — "nearest ANY other player" — picked +Je, and the
fellowship recruit sent to it obviously never completed (confirmed live:
WaitRecruited/WaitForRecruit both timed out, both bots quarantined and
gracefully logged out cleanly).
RuntimeFriendlyTargetQuery.FindPlayerByName resolves the nearest player
whose streamed name matches exactly, with 3 new conformance tests
(preferring the named player over a closer stranger, returning null when
absent, and case-sensitivity/hidden/no-draw/self rejection).
FellowshipAllegianceGateCoordinator (AcDream.Headless.Policies) is a small
same-process, no-locking (single update thread) carrier for the Recruit
bot's own discovered character name — set by its own HeadlessSessionHost
the instant CharacterList selection resolves it, which IS D8's "discover it
live" mechanism, not a hard-coded value. Constructed once per
HeadlessProcessHost and threaded through HeadlessBotPolicyFactory.Create
into the Leader policy, which now name-matches instead of taking whichever
player entity happens to be closest.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The first live two-bot run exposed a real gap: retail always confirms an
incoming allegiance swear to the PATRON (0x0274 Character.ConfirmationRequest,
type 1) before ACE sends 0x0020/0x01C8 to either party
(docs/research/2026-08-11-fa-allegiance-wire.md §3.3) — and unlike
fellowship's FellowshipAutoAcceptRequests (which ACE honors server-side,
never even sending a confirmation), there is no auto-accept character option
for allegiance. HeadlessSessionHost wired OnConfirmationRequest to null, so
a headless bot silently dropped every incoming confirmation and the swear
never completed — both bots timed out waiting for TotalVassals/patron to
seed, confirmed live against ACE (both quarantined cleanly with graceful
per-character logout, proving the self-terminating design and existing
graceful-shutdown path both work correctly; this was an FA6 capability gap,
not an FA1-FA5 wire/state defect).
HeadlessSessionHost now latches the single outstanding confirmation
(matching retail's own one-dialog-at-a-time shape) and exposes
PendingConfirmation/RespondToConfirmation, cleared on every reconnect since
a stale context id would be meaningless post-reconnect. The gate's Leader
policy polls and blind-accepts any pending confirmation on every tick before
its own stage switch — the v1 substitute for a human clicking Accept, safe
because the gate's two sessions are its own known bots.
HeadlessBotPolicyFactory.Create takes two new optional delegate parameters
(default null, so cannot break other policy ids); the Leader gate policy
requires them non-null via a defensive constructor check.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
two-bot gate
Adds the infrastructure docs/plans/2026-08-11-fellowship-allegiance-campaign.md
D8 and docs/research/2026-08-11-fa-acdream-seams.md §6.2 call for:
- HeadlessBotPolicyDescriptor gains an optional typed Role
(HeadlessBotPolicyRole.Leader/Recruit) so two sessions selecting the SAME
policy id run different scripts — fellowship leader/allegiance patron vs
fellowship recruit/allegiance vassal.
- HeadlessBotPolicyFactory.Create widens from Create(string id) to
Create(HeadlessBotPolicyDescriptor, GameRuntime) — the gate policies need
RuntimeFriendlyTargetQuery, which (like its RuntimeHostileTargetQuery
sibling) takes the concrete GameRuntime rather than the narrower
IGameRuntimeView a policy's own Tick receives (IRuntimeEntityView's
snapshot carries no name/PWD-bitfield). The single call site
(HeadlessSessionHost.cs) already has the constructed runtime in scope, so
no new constructor parameter or cross-session coordinator was needed.
- FellowshipAllegianceLeaderBotPolicy / FellowshipAllegianceRecruitBotPolicy:
a full stage-machine pair covering proximity (retail's admin @teleallto —
"teleport everyone online to me" — needs no cross-session name sharing,
unlike @teleto <name>; D8's proximity requirement is load-bearing, recruit
fails without it), fellowship create+recruit, the D4 0x00A6 panel-open
declaration with a vitals-presence assertion, the decisive two-session
assertions (the RECRUIT bot's own RuntimeFellowshipState/
RuntimeAllegianceState flipping — not the Leader's local echo), a
mid-flow reconnect on both bots proving FA2's reset-and-reseed semantics
over the real wire, and teardown (disband / break) with matching
decisive-clear assertions.
Every pre-FA6 policy (idle, lifecycle-smoke, observer-movement,
portal-route-smoke, jump-probe) is unaffected; the widened factory
signature is the only touch point.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
RuntimeHostileTargetQuery only classifies hostile monsters (via
CombatTargetPolicy.IsHostileMonster) — the FA6 two-bot fellowship/allegiance
headless gate needs the OTHER bot's server guid as a FRIENDLY target instead.
RuntimeFriendlyTargetQuery.FindClosestOtherPlayer mirrors the hostile query's
shape exactly (same hidden/no-draw filtering, same landblock-absolute
distance metric), substituting the retail PWD-bitfield IsPlayer bit (0x8,
via the existing EntityCollisionFlagsExt.FromPwdBitfield decoder) for hostile
classification. TryGetName resolves the streamed WeenieHeader name for
reporting/logging.
4 new conformance tests mirror RuntimeHostileTargetQueryTests's fixture
pattern: cross-landblock distance, hidden/no-draw/self/non-player rejection,
null-without-player-or-target, and unresolved-guid name lookup.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The FA5 mechanism review found the offline-vassal name-grey
(OfflineNameColor) is an invented visual: retail's UpdateVassalsData
@004924c3 writes the vassal name with no colour change, and the offline
cue is EXCLUSIVELY the authored 0x100004AA marker (already wired,
SetVisible per online state). Removed OfflineNameColor; the vassal name
always renders in the normal white. The Allegiance page now carries NO
invented tint (unlike Fellowship's registered leader/selection tints).
Pinned by Allegiance_OfflineCue_IsTheMarkerOnly_NameStaysWhite (marker
visible iff offline, name always white). AD-82's FA5 addendum corrected
(it had described the now-removed grey as 'covered by the marker'); AD-86
count corrected seven -> nine.
Full Release suite: 13,297 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
FA5 mechanism-faithfulness review of 7ed79eaf/bc29a1db/7e394cbf. Verdict
APPROVE-WITH-FIXES. Every high-stakes claim re-derived from the PDB-paired
2013 decomp: CF-1's unconditional 0x001F post-world arm (00490d59 sits
OUTSIDE the busy-count guard), the monarch/patron/self field sources
(UpdatePlayerData/UpdateMonarchData/UpdatePatronData), the SF-7
per-relationship gate, swear=world-selection/no-SetSelectedObject, and the
AD-86 ACE-zeroed-field citations all match retail.
MANDATORY live-mount probe RAN and PASSED against the real installed DATs
(1/1) — the scoped doubled-0x10000492 NotSame assertion and a full
production Bind() with zero "not found" held. FA5 unit suite 36/36 green.
One SHOULD-FIX (LOW): FA5 greys the offline vassal NAME (OfflineNameColor)
— retail's UpdateVassalsData @004924c3 sets the name with no color; the
offline cue is exclusively the authored 0x100004AA marker toggle. Either
drop OfflineNameColor or honestly register it (the AD-82 addendum's
"covered by the marker" framing understates it). Does not block the gate.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Suite-accounting SHOULD-FIX: the FA5 ledger/commit cite FA4's INTERMEDIATE
13,285 figure as the baseline and claim +11 net, but FA4 CLOSED at 13,286
(its 'Final full suite' figure) and the real net is +10 (verified per-file
[Fact] counts: SocialPanelControllerTests 22->31, Confirmation 4->5, probe
1->1) -- the ledger's own itemization already sums to +10, contradicting
its +11 headline. End figure 13,296/4/0 is itself correct; documentation
fix only.
Verified clean: all three Callbacks/Bindings construction sites pass the
widened Allegiance binding; every production accessor fed from a real seam;
the 0x001F and 0x00A6 toggles are independent edge-triggered latches with
no cross-talk (38 Fellowship tests green); ResolveWorldObjectName reuses
the Toolbar's ClientObjectTable read and ShowConfirmation is a pre-existing
shared method with no Fellowship collision; @allegiance info/0x027C path
untouched (52 Core.Net + 16 Runtime allegiance tests green); FA5 makes zero
Runtime changes; register 63->66 rows accurate (AD-84/85/86 + AD-82
addendum); NUL-fix correct and no residual control bytes in any of the 11
touched files.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A tool-layer artifact from the original FA5 commit (7ed79eaf) silently
replaced the leading and trailing ASCII space (0x20) in
`BlankSentinel = " blank "` with NUL (0x00) bytes -- verified byte-for-
byte via PowerShell (two NUL bytes total in the whole file, both
adjacent to the literal's "blank" text). C# tolerates an embedded NUL in
a string literal (it compiles and runs fine, since the constant is only
ever used as an internal dedup sentinel, never rendered), so this never
surfaced as a build or test failure -- caught only by an incidental
`file`/`grep -a` binary-content check while re-reviewing the finished
slice. Replaced the two NUL bytes with the intended spaces at the exact
byte offsets; swept every other file this slice touched (Runtime/App/
tests/docs) for the same corruption and found none.
No behavior change: full solution suite still 13,296 passed / 4 skipped
/ 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Register:
- AD-84 -- Swear button's missing "target is a player" enable-rule gate,
same class as AD-83's Recruit-button gap.
- AD-85 -- the unported StringInfo variable-substitution engine (AD-81's
same root cause) extended to the Allegiance page's numeric-only
followers/rank/experience-passed-up fields and its three local
confirmation dialogs (verbatim-or-bare-name, never invented).
- AD-86 -- ACE's deliberate zeroing of seven AllegianceProfile/
AllegianceData fields (officers, officer titles, MOTD, MOTD-set-by,
name-last-set-time, lock, approved vassal, timeOnline, allegianceAge),
dropped past acdream's own parse layer to match retail's own
gmAllegianceUI, which has no widget for any of them either.
- AD-82 addendum: the vassal-row click-target-only selection shares
point (3)'s limitation, but NOT the invented leader/selection tints
(point 1/2) or the Fellowship-only world-selection sync (point 4) --
Allegiance's list-selection message has no SetSelectedObject call.
Gate script: new docs/research/2026-08-12-campaign-fa-test-script.md
SFA5 section, mirroring SFA4's structure -- the CF-1 subscription steps
(including the reconnect-while-closed MF-3-REOPEN analogue), the SF-7
per-relationship monarch/patron steps, vassal-list steps, swear/break/
kick with their confirmations, the ACE-zeroed-field honesty note, and
full "what to report"/"explicitly not in scope" lists.
Plan ledger: FA5 row filled in against 7ed79eaf with per-item summary,
directly-measured totals (13,296/4/0, +11 net from FA4's 13,285/4/0),
and the two primary-source resolutions this slice needed beyond the
research docs (the self-rank field's live buffed-quality source, and
"your follower count" == _total_vassals, confirmed by a fresh targeted
decompile of UpdatePlayerData rather than inferred).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Campaign FA slice FA5. The Allegiance page (0x10000291) goes from FA3's
empty-state shell to fully live, wired against FA1's parser and FA2's
RuntimeAllegianceState/IRuntimeAllegianceCommands (both already shipped
the full command surface, including SetUpdateSubscription).
CF-1 (the corrected data subscription): 0x001F AllegianceUpdateRequest,
not 0x027B, is the panel's data source (0x027B/0x027C are text-only chat
per FA2 MF-2). Wired at retail's three arming points -- Bind's PostInit
attempt (almost always a pre-world no-op), the post-world EnteredWorld
seam (RedeclareAfterWorldEntry, UNCONDITIONAL -- does not check the
current latch, matching retail's own PlayerDescReceived arm and avoiding
the exact MF-3-REOPEN bug class FA4 hit for 0x00A6), and the visible
branch (SetPageVisible, edge-triggered, folded into
SocialPanelController's existing window-shown+active-tab conjunction
alongside Fellowship's 0x00A6).
Monarch/patron/self blocks: per-relationship empty-state gate (fix-round
SF-7) replacing FA3's coarse HasProfile-only gate -- the monarch block
hides when there is no monarch OR the monarch is the viewer; the patron
block hides when there is no patron OR the patron is the monarch (in
which case the monarch block's 0x10000490 sub-block reveals and its
label swaps to PatronSlashMonarchLabel). Field sources decompiled fresh
from gmAllegianceUI::UpdatePlayerData/UpdateMonarchData/UpdatePatronData:
0x10000251 is the ALLEGIANCE's own name (not the viewer's), follower
counts are TotalVassals/TotalMembers-1 directly off the wire, and the
"experience passed up" text (0x10000492, doubled -- scoped FindDescendant
under each of its two parents) is the viewer's own CpTithed under the
monarch/patron blocks and each vassal's own CpTithed in their row.
Vassal roster: flat list built via UiTemplateListBox.FlushPreservingScroll
in the FA4 roster-diff pattern (guid-set diff, in-place update on an
unchanged set), rendering in the bindings' own already-reversed order.
Swear/break/kick: each opens a local confirmation dialog
(RetailDialogFactory via ShowConfirmation) before sending, mirroring
retail's MakeSwearConfirmationDialog family -- Swear targets the WORLD
selection (via the same ClientObjectTable name resolver
ToolbarRuntimeBindings.ResolveName already uses), Break targets the
current patron, Kick targets the panel-local selected vassal row (no
world-selection sync for Allegiance, unlike Fellowship). The
server-driven "accept incoming swear" (ConfirmationType 1) needed no new
code -- GameplayConfirmationController already handles every type
generically; a new test verifies it explicitly.
Runtime/composition plumbing: DeferredGameRuntimeStateCommands gains
Allegiance{Swear,Break,Kick,SetUpdateSubscription}; SocialRuntimeBindings
gains the Allegiance view/command projections; SocialPanelController.
Callbacks.AllegianceSnapshot widens to a full
SocialAllegiancePageController.Bindings record, mirroring FA4's
Fellowship widening.
Tests: SocialPanelControllerTests.cs gains 10 tests covering the SF-7
gate (4), roster population, swear/break/kick wiring (3), and the CF-1
subscription arming points (2); GameplayConfirmationControllerTests.cs
gains the type-1 verification test.
Also extends SocialPanelLiveMountProbeTests.cs (production-mount
assertions: scoped 0x10000492 resolution, the vassal row template, the
checkbox, confirmation-dialog string resolution, and a full production
Bind() pass) -- not yet run against live DATs in this worktree (no
Documents/Asheron's Call present here).
Release build green; full solution suite 13,296 passed / 4 skipped / 0
failed (13,300 total), up from FA4's 13,285/4/0 baseline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The re-fix moves the 0x00A6 re-declaration off the pre-world reset seam
and onto the post-world EnteredWorld seam, and stops the widget latch from
advancing on a dropped publish. Verified in the diff:
- SetPageVisible advances _pageVisible ONLY on RuntimeCommandStatus.Accepted
(the widget-level root of the REOPEN); a dropped Inactive publish leaves
the latch clear so the in-world attempt is not deduplicated.
- ResetSessionDeclaration (pre-world) now only clears the latch;
RedeclareAfterWorldEntry (new) does the re-evaluation, wired through
RetailUiRuntime.RedeclareSocialPanelAfterWorldEntry into
LiveSessionRuntimeFactory's EnteredWorld RestoreLayout delegate.
Seam ordering traced and confirmed inverse of the pre-world SessionDialogs
stage: StartCore runs ResetHostBeforeStart (pre-world reset, latch clear)
at :555, then ActivateCommands :639, _inWorld=true :642, and
ApplyEnteredWorld :644 -> LiveSessionHost.ApplyEnteredWorld ->
RestoreLayout delegate -> RedeclareAfterWorldEntry. So SetPanelOpen's
requireWorld gate is Accepted and 0x00A6 publishes on the fresh server.
Idempotent and load-bearing (the social panel isn't state-managed
visibility, so RestoreLayout fires no OnShown edge).
Tests model the world gate (fake returns Accepted only when in-world) and
would fail against pre-fix behavior: the widget test's second attempt is
deduplicated if the latch advances unconditionally; the reconnect test's
DoesNotContain-after-reset fails if the pre-world declaration is
reintroduced (the coordinator's RED-verification). Binary confirmed
post-fix (new tests reference RedeclareAfterWorldEntry); 3/3 new + 58/58
touched classes green. 13,286/4/0 reconciles (+1, 0 deletions).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The FA4 fix round's MUST-FIX 3 placed the 0x00A6 reconnect re-arm at the
wrong lifecycle point (re-review 8bbceff5): ResetSessionTransientUi runs
via the SessionDialogs reset stage BEFORE _inWorld=true, so SetPanelOpen
(world-gated, Validate requireWorld:true) returned Inactive and published
nothing — yet _pageVisible was latched true anyway, so no later hook
re-declared and fellow vitals stayed frozen for the whole new session.
The unit test passed only because the fake recorded unconditionally.
Two-part fix, both retail-faithful mechanisms not suppressions:
- SocialFellowshipPageController.SetPageVisible advances the edge-trigger
latch ONLY when the declaration is Accepted (published), so a dropped
pre-world send leaves the latch clear and a later attempt retries.
- ResetSessionDeclaration (pre-world) now ONLY clears the latch; the new
RedeclareAfterWorldEntry fires from the LiveSession EnteredWorld seam
(wired via RestoreLayout, idempotent if a persisted layout already
re-showed the page) so a still-open Fellowship page re-declares 0x00A6
in world and vitals resume.
Regression pins that actually catch it (the prior test could not):
- SetPageVisible_DoesNotLatch_WhenDeclarationDropped_SoItRetriesInWorld
(widget-level root, world-gated fake);
- Reconnect_ReDeclares0x00A6_AfterWorldEntry_NotDuringPreWorldReset +
Reconnect_StaysSilent_WhenFellowshipPageIsNotActuallyOpen (panel-level,
world-gated). RED-verified: reintroducing the pre-world declaration
fails the reconnect test.
Full Release suite: 13,286 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Re-derived each disposition from the actual fix diffs (290f9b58/5499f058/
df000306/300d8189/55b17e15/1d743277/f041b09b), not the commit claims.
CLOSED (4/5 MUST-FIX, all 9 SHOULD-FIX, all 4 NIT, blast SF-1):
- MUST-FIX 1: (int)((double)pct*100.0) truncation + 6->44%/8->34% pinning
cases + gate step corrected.
- MUST-FIX 2: intercept deleted, every type routes to the generic
controller, type-4 dialog test added; type-1 allegiance path unaffected
(was never intercepted).
- MUST-FIX 4: world->panel selection sync reproduces retail's found/
fallback arms; AD-82 records the deferred generic UiTemplateListBox
selection-model port honestly -- minimal-observable-contract, not a
hidden gap.
- MUST-FIX 5: AD-82/AD-83 well-formed; AD-78 count corrected to 34/16.
- D6/D7/SF-8 dimming (audited from source): 34 dimmed / 16 live is
correct, not split-the-difference. FellowshipShareLoot has NO client
value-reader (only an editor/display surface; 0x00A2 sends shareXP
alone; ACE authors loot server-side) -> dimmed faithful.
FellowshipShareXP is genuinely read by the Create click -> Live right.
REOPEN (MUST-FIX 3): the 0x00A6 reconnect re-arm is placed at a pre-world
reset seam. ResetSessionTransientUi runs via the SessionDialogs reset
stage at ResetHostBeforeStart / retired-scope teardown -- both BEFORE
_inWorld=true and before command activation for the new generation -- and
SetPanelOpen requires world, so the re-declaration returns Inactive and
nothing is published, yet _pageVisible is still set true and no
post-world-entry hook re-evaluates. The new server never receives 0x00A6
and fellow vitals stay frozen -- the exact bug the fix targets. The unit
test passes only because its fake command records unconditionally.
Recommend moving the re-declaration to an in-world seam (EnteredWorld).
Totals/probe: 109/109 touched App test classes green on post-fix
binaries; live-mount probe PASS 1/1; +13/0-deletion delta and 13,285/4/0
corroborated on the touched projects and by arithmetic (not re-run
end-to-end).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
FA4 row now names all six fix-round commits (290f9b58, 5499f058, df000306,
300d8189, 55b17e15, 1d743277) alongside the original landing's three, and
records: build/test green at every commit; the +13/0-deletion test delta
broken down per file; the final directly-measured 13,285 passed / 4
skipped / 0 failed (13,289 total); every MUST-FIX/SHOULD-FIX/NIT applied;
and the corrected dimmed-row arithmetic (35 -> 31 FA4-original -> 34
fix-round final, net one row). Also corrects item (4) of the
"contradictions/deferrals" list, which called the missing Recruit
is-a-player register row an acceptable inline comment -- MUST-FIX 5 named
that the wrong call; it is now register row AD-83.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Register (docs/architecture/retail-divergence-register.md):
- AD-78: the Character-tab dimmed count had drifted stale through two
campaigns (still read "35" after FA4 shipped 31; now 34 after the fix
round's three reversions). Addendum explains the full D6/SF-8 chain.
Blast review's own SHOULD-FIX 1.
- AD-82 (new): the invented leader-tint/selection-tint colors, the
name-text-only row click target, and the page-local (not generic
UiTemplateListBox) world->panel selection sync -- MUST-FIX 4's
disposition plus two items MUST-FIX 5 named as owed rows.
- AD-83 (new): the Recruit button's missing "target is a player" gate,
previously an inline comment, not a register row -- MUST-FIX 5's third
item. Section header bumped 61 -> 63 active rows.
Gate script (docs/research/2026-08-12-campaign-fa-test-script.md):
- SF-7: fixed step 3's self-contradiction ("only Quit" then "Disband and
Open should ALSO be enabled").
- MUST-FIX 3: new reconnect step after the existing close/reopen step.
- MUST-FIX 4: new world-selection step under the recruit/dismiss/quit
section.
- MUST-FIX 1: new HARD-check step for the 6/8-fellow 44%/34% truncation
(distinct from the existing SOFT 9-member ACE-divergence note).
- MUST-FIX 2 correction: the old invite steps tested whether acdream's
CLIENT gates the dialog on the option bits -- a mechanism that never
existed in retail and no longer exists in acdream. Rewritten to test
the corrected behavior (the dialog always shows regardless of the
target's own checkbox state) and to explain what ACE-side filtering
would look like if the local server implements it, so a tester doesn't
misattribute ACE's behavior to a client bug.
- Renumbered steps 9-22 to 9-25 to fit the two new steps; updated the
"what to report" section's step cross-references and rewrote its
invite/dimming bullets to match the corrected mechanism.
Plan (docs/plans/2026-08-11-fellowship-allegiance-campaign.md):
- D7 addendum: SF-8's further correction (FellowshipShareLoot reverts
too; only FellowshipShareXP survives as genuinely live).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
SocialPanelLiveMountProbeTests wrote the four checkbox labels and the two
Open/Close captions to the console with no assertion, yet the FA4 ledger's
live-DAT paragraph cited them as verified -- the same finding FA3's own
mechanism SF-3 raised for a different table ("printed but never asserted
-- deserves a real assertion, not just a hope"). Now asserts each label is
non-null/non-empty and the two captions equal the exact retail strings
"Open"/"Close". Env-gated (ACDREAM_PROBE_LIVE_MOUNT=1, real installed
DATs) -- inert in this session's build/test run.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The corrected plan D6 (docs/plans/2026-08-11-fellowship-allegiance-campaign.md)
established that retail's client reads neither IgnoreFellowshipRequests nor
FellowshipAutoAcceptRequests on the fellowship-invite path -- both are pure
server-side filters with no client consumer, exactly like the two
allegiance bits they were always meant to parallel. Their claimed consumer
(RetailUiRuntime.TryAutoRespondToFellowshipInvite) is deleted in a sibling
commit this fix round. Both rows revert from Live to StoreOnly.
Mechanism review SF-8 additionally found FellowshipShareLoot's claimed
consumer -- "a second live checkbox surface on the fellowship page" -- is
not a consumer at all: nothing in acdream reads the stored value back
(FormatStatsText uses snapshot.ShareXp only; the 0x00A2 Create builder
carries shareXP alone), and the live-DAT dump confirms its checkbox is a
child of the NOT-in-fellowship frame -- invisible whenever you actually
have a fellowship to loot-share within. A second EDITOR of a value is not
a CONSUMER of it under AD-78's own "drives nothing observable client-side"
definition. FellowshipShareLoot reverts too.
Only FellowshipShareXP survives as genuinely live -- the Create-flow click
reads it directly as the sent shareXP wire bit. Net: 35 (pre-FA4) -> FA4
shipped 31 -> fix round reverts three -> 34 of 50 dimmed / 16 live, ONE
net un-dim from the pre-FA4 baseline, not four. Updated the class doc's
derivation table, the conformance test's ExpectedStoreOnlyIds set, and the
31/19 count assertions to 34/16.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MUST-FIX 3 -- SocialFellowshipPageController.SetPageVisible is edge-triggered
on a bool that survives a generation reset unchanged while the panel stays
open, so a reconnect never re-sends 0x00A6 and fellow vitals freeze for the
rest of the new session. SocialPanelController.ResetSessionDeclaration
clears the fellowship page's latch (SocialFellowshipPageController.
ResetPageVisibleLatch, this commit's counterpart) and re-evaluates the
existing "window shown AND Fellowship active" conjunction, wired into
RetailUiRuntime.ResetSessionTransientUi -- a seam that already runs on
every generation reset. A still-open Fellowship page re-declares; a closed
or other-tab page correctly stays silent.
SF-4 -- SocialPanelController's constructor subscribed an anonymous lambda
to UiTabPanel.ActivePageChanged with no way to remove it; Dispose only set
a flag. A tab switch after Dispose still reached
UpdateFellowshipPageVisibility and issued a Runtime command, since Tick's
own _disposed guard doesn't cover this event path. Stored the handler as a
field and unsubscribe it in Dispose.
Also adds the panel-level D4 conjunction test mechanism SF-5 flagged as
missing (the only prior D4 test exercised the PAGE controller's own
edge-trigger directly, never SocialPanelController's "window shown AND
Fellowship active" logic or its ActivePageChanged subscription).
Per docs/research/2026-08-12-fa4-review-mechanism.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MUST-FIX 1 -- D5's percentage conversion rounds where retail truncates.
gmFellowshipUI::UpdateFellowStats @0x0048ECC9 forms pct*100.0f on the x87
stack then calls _ftol2 (MSVC's round-to-truncate helper), never
MathF.Round. The stored floats for 6 and 8 fellows are 0.44999998807907104
and 0.3499999940395355 (byte-read from the PDB-paired binary), so retail's
own products truncate to 44/34, not 45/35 -- and (int)(pct*100f) alone does
not fix it, since 0.45f*100f already rounds UP to exactly 45.0f in single
precision. Fixed as (int)((double)pct * 100.0), forming the product the
same wider-than-single-precision way retail's x87 does. Pinned with new
[InlineData] cases for both sizes.
MUST-FIX 4 -- gmFellowshipUI::UpdateFellowSelection @0x0048F0F0 (the
world->panel arm of retail's two-directional selection coupling) was never
ported; only the panel->world arm (SelectFellow) shipped. Selecting a
fellow in the WORLD left Dismiss/Assign-Leader disabled and showed no row
highlight. SyncSelectionFromWorld/SetSelectedFellow reproduce the
observable contract (button-enable + a row tint) against this
controller's own guid-keyed row dictionary instead of porting retail's
generic ListBox SetAttribute_InstanceID/SetSelectedItem primitive (scoped
disposition recorded at register row AD-82).
Also in this pass over the controller:
- SF-1: cache the fellowship-name LinesProvider; only reassign on an
actual name change (was allocating once per Tick, even while hidden).
- SF-2/SF-3: track true membership in _memberGuids, independent of which
rows finished building. Fixes an unbounded DAT-locked rebuild retry
when a row template permanently fails to build, and fixes Recruit's
"already a fellow" check reading render rows instead of membership.
- N-0: the Open/Close caption now flips optimistically on click, matching
retail's pre-toggle-before-server-echo (lane B feature 11).
- N-1/N-2/N-3: doc-only notes on the meter-child-text gap, the max>0
guard, and Tick's two-read non-atomicity.
- ResetPageVisibleLatch: the fellowship-controller half of MUST-FIX 3
(see the SocialPanelController commit for the panel-level half).
Per docs/research/2026-08-12-fa4-review-mechanism.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
RetailUiRuntime.TryAutoRespondToFellowshipInvite auto-declined/auto-accepted
fellowship invites based on IgnoreFellowshipRequests/FellowshipAutoAcceptRequests
before the dialog ever reached GameplayConfirmationController. Byte-verified
across Handle_Character__ConfirmationRequest @0x005640A0,
RecvNotice_FellowshipRequest @0x00490880, and MakeFellowRequestDialog
@0x00490620 (whose only guard is m_fellowRequestContext) plus a whole-file
sweep of both option accessors: retail's client reads neither bit on any
confirmation path. ACE filters both bits server-side, so the interceptor was
dead code against a correct ACE and actively harmful against a drifting one
(IgnoreFellowshipRequests defaults true, so it would silently swallow real
invites with no dialog and no chat line).
HandleConfirmationRequest now routes every confirmation type, including 4,
straight to the generic controller -- exactly like retail. No tests existed
for the deleted interceptor (nothing to remove); added a test proving the
type-4 dialog renders the server message verbatim (not "Continue?"-suffixed)
and sends accept/decline through the generic path.
Per the corrected plan D6 (docs/plans/2026-08-11-fellowship-allegiance-campaign.md).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
D6 asserted the client consumes IgnoreFellowshipRequests/
FellowshipAutoAcceptRequests on the invite path; the FA4 mechanism review
(913e35cd MUST-FIX 2) byte-verified retail reads NEITHER bit client-side
(Handle_Character__ConfirmationRequest @0x005640A0, RecvNotice_
FellowshipRequest @0x00490880, MakeFellowRequestDialog @0x00490620) — ACE
filters both server-side. Same class as the D2 reset-lifetime correction.
D6/D7 corrected in-place with dated addenda: the client-side intercept is
removed, the invite dialog always shows, and the two un-dims revert
(dimmed 31->33). D8 records the user-provided second account
(testaccount2/testpassword2) and the recruit-proximity requirement.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A second mechanism-lens review landed on the same path at 6849b457 while
this one was in progress and was overwritten by 913e35cd. Its text is
recoverable from git and is now cited from a new appendix, its two unique
findings are carried forward, and the two places the reviews disagree are
adjudicated from primary source.
Carried forward:
- SF-9: AD-78's register row still says "35 of 50 rows dimmed" (the D7
addendum landed in the class doc, not the row's Where column).
- N-0: the Open/Close caption does not optimistically pre-toggle; lane B
feature 11 records that retail's handler pre-toggles _open_fellow
locally before sending 0x0291.
Adjudicated:
- _ftol2 vs MathF.Round: 6849b457 filed it a NIT ("round and truncation
agree on every table value"). That holds for the DECIMAL literals, not
the stored floats -- 0x007C91D4 = 0.44999998807907104 and 0x007E72BC =
0.3499999940395355, so retail truncates 44.999998/34.999999 to 44/34
while acdream rounds to 45/35. Stays MUST-FIX 1.
- D6 invite auto-response: 6849b457 passed it as verified-clean after
confirming the code matches the plan. The binary says retail has no
such client-side read on any confirmation path. Stays MUST-FIX 2.
The reviews agree on the reconnect D4 hole, the missing panel-level D4
conjunction test, the leader-tint register omission, and the live-DAT
probe result.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Findings persisted before any fixer dispatch, per the campaign's §7
review protocol.
MUST-FIX, in the order they were derived:
1. D5's percentage conversion rounds where retail truncates. Byte-decoded
0x0048ECC9..0x0048ECD8 (fld pct; fmul [0x007A5170]=100.0f; call
_ftol2 @0x005DE394 -- the fld/fst/fistp/fild truncation dance), so a
6-fellow roster displays 45% where retail shows 44%, and 8 fellows
shows 35% vs 34%. The table itself IS byte-exact; only the ->int
conversion diverges, and it is not fixable by a plain cast because
0.45f*100f already rounds up to 45.0f in single precision.
2. D6's client-side invite intercept has no retail anchor. Read in full:
Handle_Character__ConfirmationRequest @0x005640A0 (bare jump table),
RecvNotice_FellowshipRequest @0x00490880, MakeFellowRequestDialog
@0x00490620 (only guard is m_fellowRequestContext), plus a whole-file
sweep of both option accessors -- zero reads on any confirmation path.
The code comment cites ACE's Fellowship.cs as "retail". ACE filters
both bits server-side, so the intercept is dead against a correct
server and harmful against a drifting one -- and IgnoreFellowshipRequests
defaults to TRUE client-side.
3. D4 never re-declares 0x00A6 after a generation reset: the edge-
triggered _pageVisible latch survives reconnect, so fellow vitals stay
frozen for the whole new session. ResetSessionTransientUi is the seam.
4. gmFellowshipUI::UpdateFellowSelection @0x0048F0F0 is not ported --
selecting a fellow in the WORLD leaves Dismiss/Leader disabled and no
row ever shows selected; the plan's contracted UiTemplateListBox
selection model + 0x1000000D row instance-id were not added.
5. Three shipped deviations have no register row (invite intercept, gold
leader tint, name-text-only row selection); the Recruit is-a-player
gate's "inline comment, not a register row" call is also wrong.
Re-derived rather than trusted: the live-mount probe was re-run against
the installed DATs (every ledger element/string claim CONFIRMED, Bind()
warning-free), the GetEvenSplitXPPctg table was byte-read from the
PDB-paired binary, FlushPreservingScroll's shrink semantics were traced
through UiScrollablePanel/UiScrollable (sound), and the five touched
test classes pass 93/93.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
