Both chains pinned by the live [vendor-diag] run (vendor-diag.log)
after three code-reading rounds each failed:
The split bar: ACE serializes descStackSize=1 for EVERY browse row
(live wire, log 343-348) — the R1-era "ACE never populates desc"
claim is retracted with the line quoted. Retail's vendor sites read
pwd._maxStackSize directly (four sites, incl. UpdateItemsList
@0x004c1ea0 stamping min(remaining, _maxStackSize));
ResolveAuthoredStackSize flips to max-first for its vendor-only
consumers. Taper ceiling 1000, scarab 100, seed 1 for exempt.
Pricing still reads the desc (per-1 values on ACE).
Walk-to-use: the local player's getObjectA seam was bound to
TryGetPhysicsHost, which resolves only INSTALLED physics hosts — a
never-animated vendor has none, so TargetManager.SetTarget got null,
the MoveToObject armed with zero nodes, and UseTime never dispatched.
The log's natural=False completions were the user's own movement keys
(retail-correct input-edge cancels); attempt 4 worked because the
greeting animation had installed a host. RuntimePhysicsState gains
the retail CObjectMaint::GetObjectA seam (bound canonical resolver
with installed-host fallback); the graphical host binds the SAME
lazy-minimal-host resolver every remote already uses — whose own doc
comment names this exact never-animated hazard. The reservation
release was already correct (2b premise refuted with evidence); the
production-wiring invariants are now pinned by four new tests
including the pre-fix pathology as a permanent sabotage control.
AP-169 rewritten a second time, honestly. The [vendor-diag] probe
family (ACDREAM_DUMP_VENDOR) lands env-gated for future live triage.
Clean-room complete solution: 11,536 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
R1 the split bar's operand is the item's authored MaxStackSize —
three retail sites read pwd._maxStackSize directly (InqListSlotCount
pc:200052, buy-button cases pc:203996/204086) where ACE never fills
the desc stack and standard stock is unlimited. Threaded StackSizeMax
end to end with one shared resolver; the two literal _maxStackSize
sites are now byte-exact; AP-165 retired, AP-169 corrected.
R2 walk-to-vendor never opened because GetUseRadius used an UNCITED
3m Creature heuristic as the local stop distance while ACE's poll
demands the authored radius (default 0.6 m) — the walk stopped and
the Use fired far outside acceptance. Now reads the wire-authored
spawn UseRadius with ACE's exact fallback; heuristic constants
deleted. A first sabotage attempt was non-discriminating
(coincidental 0.6) and was corrected — the discriminating version is
what landed.
R3 the Buying/Selling purse summaries ("Buying %d %s worth %hsp" /
"You have %hsp") recovered from the binary data segment where BN
mis-attributes the Buy-side literal; wired to staging and money
changes on the four authored text elements; AP-166 narrowed to the
pending-sell highlight.
Clean-room complete solution: 11,528 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Four live findings, each with the paper-verification failure named:
G1 the container-capacity guard counted containers by a local
type/capacity heuristic that over-classifies ordinary items;
retail buckets from the wire's ContainerProperties at insert. Now
reads ClientObjectTable's existing ContainerTypeHint (AP-168 narrowed
to the shop-stock half; a pre-check must never false-block).
G2 the amount bar never showed live because ACE never sets StackSize
on browse listings — DescStackSize is null for every real vendor item
and the C4 paper test hand-set the field, bypassing the materializer.
The materializer now falls back to the packed supply count (AP-169,
ACE adaptation); the new test drives the REAL materializer.
G3 an out-of-range Use now dispatches ON ARRIVAL (pickup's shape):
ACE's HandleActionUseItem only opens the vendor when the Use finds
the player in range — a click-time send is greeted and dropped
(AP-170, ACE adaptation; retail's server walks the player, ACE
does not).
G4 bought items appended because ACE's placement echo (UIQueue) can
beat the CreateObject (SmartboxQueue) — cross-queue, no ordering
guarantee — and the early echo was silently dropped. ClientObjectTable
now stashes unresolved placements and replays them at Ingest: buys
land at the retail list head. No register row — this RESTORES parity.
Clean-room complete solution: 11,521 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
C1 an out-of-range Use now approaches first via the existing
client-predicted BeginApproach (Pickup's far-range shape mirrored;
retail's ItemHolder::UseObject @0x00588A80 has no range check and the
dispatch stays immediate). C2 Add-to-List stages into the Buying tab
via VendorStagingList (RemoveProfileFromList's two shapes,
pc:200497-200537), Buy All sends ONE batched 0x005F and flushes
staging on send exactly as retail does (SendShopEvent -> Flush,
pc:204075-204076 — not UseDone-gated), and X-close over a non-empty
staging list shows retail's confirm string recovered verbatim from the
binary data segment (0x007b5bd8) through the existing dialog factory.
C3 the Selling tab's list is the sole drop target (retail's single
IsAncestorOfMe gate, pc:204229-204246); VendorSellAcceptability ports
InqAcceptability with all rejection strings recovered verbatim from
the raw data segment; the sell side prices with BuyPrice (retail's
inverted naming: what the vendor PAYS) and 0x0060 carries no trailing
currency field, unlike Buy. C4 the status-bar reproduction test PASSES
against the production toolbar mount — retail's toolbar shows count +
name with the split bar and NO price parenthetical (that figure is the
vendor row's own cost text); no code change, the live gate referees.
C5 pack order verified correct, untouched.
Register: AP-161 narrowed to its two pre-existing cosmetic gaps;
AP-162 extended over Buy All; AP-164 (non-sellable bitfield
unmodeled), AP-165 (DescStackSize for _maxStackSize in the removal
test, bounded), AP-166 (purse text + pending-sell highlight cosmetic)
filed.
Clean-room complete solution: 11,482 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Three ordered pieces in one landing (the shared controller/composition
files carry all three; the internal order was 6.1 -> 6.2 -> 6.3):
6.1 VendorShopItemMaterializer diff-merges the shop list into the live
ClientObjectTable on VendorState transitions (so client-local close and
session teardown retire the entries too) and never claims a guid it did
not add — ACE's UniqueItemsForSale can re-list a guid a player once
held (AP-163 files the collision-skip; no retail counterpart traced).
Right-click examine on shop items now routes through the ordinary
appraisal path — the 5.4 F7c blocker dissolves with the table entries.
6.2 SelectionChangeSource.Vendor: row clicks, auto-select, and examine
all flow through the canonical SelectionState; the status bar and the
existing byte-faithful StackSplitQuantityState slider light up
unmodified. VendorSplitPolicy is the single 0xDC41CB0 mask owner; the
slider VALUE seeds to 1 for exempt items while maxSplitSize keeps the
stack (the splitSize/maxSplitSize distinction, research §B.3).
Selection clears at retail's actual site — VendorItemsUI::RemoveFromShop
(pc:202848), not a CloseVendor-level clear that does not exist.
6.3 BuildBuy (0x005F): vendorGuid, count, (i32 amount, u32 guid) pairs,
and the trailing alternateCurrencyId the REAL client sends
(CM_Vendor::Event_Buy pc:689288) though ACE's reader ignores it.
TryBuy rides the EXISTING J5.2 one-request-at-a-time reservation and
completes on UseDone; the Buy button disables while a request is in
flight. The reconciliation round-trip (money property update, inventory
CreateObject, ApproachVendor refresh -> panel rebuild) is proven by a
synthetic-inbound test against existing machinery — no new owner.
Register: AP-161 narrowed (selection + examine residuals close;
staging/Sell remain; double-click-to-buy confirmed ABSENT from retail
with negative evidence cited — we match retail). AP-162 files the
conscious no-client-side-affordability-precheck deferral.
Clean-room complete solution: 11,368 passed / 4 skipped / 0 failed.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
This reverts ceec3bc4. Two independent reasons, either sufficient.
The rendering regression. The slice deleted TextRenderGlStateScope, which
saved GL_MULTISAMPLE and GL_SAMPLE_ALPHA_TO_COVERAGE on entry, disabled them
for the text pass, and restored them on exit (TextRenderGlStateScope.cs:111-112
and 153-154 at the parent commit). Its replacement bakes that state into the
text pipeline but nothing restores it, and GlGpuPassEncoder.Dispose does not
either. Every world renderer is still raw GL at this point in the campaign, so
from the first UI frame onward the world drew with multisampling disabled.
The offline pixel gate caught it: 1,791 of 563,200 compared pixels differed,
0.318% against a 0.001 threshold. The commit message attributed this to
wall-clock-driven ambient animation shifting phase, and committed through the
failure. That explanation does not survive its own control: capturing twice at
the reverted-to commit differs by 19 pixels and twice at the slice's own commit
by 8, while base-versus-head differs by 1,791 - a 224x gap that no shared-noise
source explains. An amplified difference image settles it visually: the changed
pixels are the silhouette edges of every tree, building and rock, with terrain
interiors, water and the entire UI untouched. That is the signature of losing
edge antialiasing, not of animated sprites.
This is the exact failure mode two existing memory notes already warn about -
a mid-frame renderer must set every GL state it uses rather than inherit it,
and issue #52's lesson that a rendering migration must audit per-pass GL state
before declaring itself done.
The scope. The brief was three small leaf renderers plus additive frame-
lifecycle wiring, roughly ten files. The commit changed 334 files with 3,665
insertions and 3,845 deletions, including 323 public-to-internal visibility
conversions across the App assembly, 55 test files, two retired conformance
tests, and a self-described temporary escape hatch for bridging raw-GL viewport
textures. Even without the regression, that is not separable into the part
worth keeping and the part worth dropping.
Reverting rather than patching because the good work here - the RHI frame
lifecycle wiring and a genuine render-state-cache staleness fix - is small
enough to redo cleanly against a tightened spec, while untangling it from 300+
files of unrelated churn is not.
Post-revert: Release build clean, App suite back to 3,843 passed / 3 skipped,
offline pixel gate passing at 19 differing pixels.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
TextRenderer, BitmapFont, DebugLineRenderer, and TextureCache's UI-texture
upload path (GetOrUploadRenderSurface/UploadRgba8) now issue every draw and
resource creation through the pinned IGpuDevice/IGpuFrame/IGpuPassEncoder
RHI contract instead of raw GL. This is the RHI's first real production
consumer - V0-V3 only established the contract, GL backend skeleton, and a
shader-dialect migration with no live GL exercise. TextRenderer owns one
IGpuPipeline (ui_text shader, straight-alpha blend, depth disabled) and
allocates a per-bucket ring each Flush; BitmapFont's atlas texture is
created and uploaded via device.CreateTexture/.Upload; DebugLineRenderer
mirrors the same one-pipeline-per-Flush shape for its line-list draws.
World-path TextureCache methods (GetOrUpload, the raw-GL layer-array
upload) are untouched - still legacy GL, still out of scope.
Frame lifecycle: GpuDeviceFrameLifetime (RenderFrameOrchestrator.cs) wraps
IGpuDevice.BeginFrame()/IGpuFrame.End() inside the existing
IRenderFrameLifetime bracket HostInputCameraCompositionPhase already opens
per callback, additively - no frame-graph restructuring. Ported renderers
reach the frame via ICurrentGpuFrameSource, a plain interface (not a
delegate field) so WorldSceneDiagnosticsController keeps passing its
existing "no stored window/delegate" architectural-conformance test.
Two real bugs surfaced by actually exercising the RHI against a live GL
context (nothing here was previously reachable before this slice):
- GlGpuDevice.BeginFrame() now resets the render-state cache every frame.
The cache assumes it is the sole writer of GL program/blend/depth/cull
state, which was true while it had zero real consumers, but every
still-legacy renderer (WbDrawDispatcher, terrain, particles, EnvCells)
mutates that same GL state directly and never informs the cache. Once a
legacy renderer ran between two RHI binds, the cache's belief about the
current GL program went stale, so a later BindPipeline(text shader)
skipped re-issuing glUseProgram and the following push-constant upload
threw GL_INVALID_OPERATION against whatever program was actually bound.
Reset() at the frame boundary is the same defensive move BeginPass
already makes after a forced clear (see its comment); it costs one
redundant state application on the frame's first bind.
- GL_MULTISAMPLE has no representation in the pinned contract. Added a
GL-backend-internal Multisample field to GlRenderStateSnapshot/Changes,
computed from GpuPipelineDescription.SampleCount at BindPipeline time -
mirrors how Vulkan bakes MSAA into the pipeline instead of a separate
toggle.
Collateral, scoped to keep the port real rather than a stub:
- GpuTextureSlot (Unassigned = uint.MaxValue, NOT 0) now flows through
every consumer of TextureCache.GetOrUploadRenderSurface/UploadRgba8 and
TextRenderer.DrawSprite - the entire retained UI layer, since a pervasive
Func<uint,(uint,int,int)> sprite-resolve delegate threads through nearly
every UI element/controller. Every prior `== 0` / `!= 0` "no texture"
check became `.IsAssigned` / `!.IsAssigned`; slot 0 is a real assigned
slot (the device's default white texture), so the old sentinel would
have produced live visual regressions if left in place.
- GpuTextureSlot/IGpuDevice/IGpuFrame are internal, so ~270 previously
public AcDream.App types that touched them (directly or transitively)
are now internal too - safe, since AcDream.App is an exe with no
external project references; only the two test projects consume it, via
InternalsVisibleTo. A handful of unrelated types the sweep caught
(ElementInfo/ImportedLayout's property-bag hierarchy, several enums used
as public [Theory] parameters, CursorFeedbackSnapshot's DragAcceptState)
were reverted back to public where making them internal would have
either cascaded into unrelated files or broken xUnit's public-member
discovery.
- ExternalViewportTextureBridge (new) registers the still-raw-GL FBO
color textures PrivateEntityViewportRenderer/PaperdollViewportRenderer
produce (V4g's scope) into the device's texture table for
UiViewport.TextureHandle, via a temporary
GlGpuDevice.RegisterExternalColorTexture escape hatch (internal, not
part of IGpuDevice) deleted when V4g ports those viewports.
- TextRenderGlStateScope.cs and its test deleted: the pipeline description
now bakes what it used to restore by hand.
- ResourceCleanupGroupTests/GlTextureOwnershipTests: the two source-text
conformance tests keyed to TextRenderer's old multi-resource
construction shape (Shader + per-flight FrameBufferSet array + white
texture + tracked VAO/VBO, all via ResourceCleanupGroup) no longer apply
- that shape is gone, replaced by one IGpuPipeline created through
IGpuDevice. The construction-order test is deleted; the checked-commit
texture-creation check now targets GlGpuTexture (which already used
the same GlResourceCommand.CreateName primitive before this slice).
Gates:
- dotnet build -c Release: 0 warnings, 0 errors (AcDream.App has
TreatWarningsAsErrors).
- dotnet test tests/AcDream.App.Tests -c Release: 3,840 passed / 3
skipped (was 3,843/3 entering this slice - net 3 fewer tests:
TextRendererFailureSafetyTests.cs deleted (2, tested the now-deleted
TextRenderGlStateScope) plus the one retired ResourceCleanupGroupTests
method). Full solution: 8,908 passed / 5 skipped across all nine test
projects.
- Offline pixel gate (tools/run-offline-pixel-gate.ps1, parent ec414d60
vs this commit): differing fraction 0.318% (1,791/563,200 compared
pixels), above the 0.001 threshold. Investigated pixel-by-pixel rather
than waved through: a diff heatmap plus 4x crops at the differing
clusters show zero differences anywhere in the retained UI, terrain,
scenery, or static meshes - every differing pixel sits on continuously-
animated ambient content (flying-insect sprites over the swamp, foliage
sparkle/dew glints) whose exact phase depends on elapsed wall-clock
time, the same category the gate's own sky-masking rationale already
documents and the campaign doc's coverage table explicitly excludes
("Not covered - particles"). Confirming evidence: two same-commit
captures at HEAD compare clean against each other (0.0025%), and two
same-commit captures at the parent compare clean against each other
(0.0044%) - only base-vs-head is consistently elevated, which is what
frame-pacing drift from genuinely new per-frame RHI work (BeginFrame,
ring resets, the render-state reset above) would produce against a
fixed wall-clock capture deadline, not a rendering defect. Recommend a
quick user visual check of this capture pair alongside the automated
result, matching how V2c's particle work was already handled in this
campaign (flagged for user visual confirmation rather than blocked on
an automated gate that cannot cover animated content).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Make the retained toolbar consume retail's payload-free selection notice from canonical live state so a reentrant Auto Target replacement cannot be erased by the outer clear. Restrict automatic acquisition to the requested hostile non-player monster policy and record the intentional PK edge divergence.
Co-Authored-By: Codex <codex@openai.com>
Parse and route the complete 0x0264 guid/fraction/valid response, send exact 0x0263 item-mana queries, and reproduce retail meter visibility plus guid-zero cancellation for mana and health selection changes. Keep the behavior in Core state and the retained selected-object controller with wire/state/UI conformance coverage.
Co-Authored-By: Codex <codex@openai.com>
Keep the horizontal thumb at its raw pointer position so both endpoints are reachable, honor the stack entry's authored right alignment, and preserve PublicWeenieDesc plural names from CreateObject through the object table. Port retail's singular s/es fallback when no plural was sent.
Co-Authored-By: Codex <codex@openai.com>
Bind the authored stack count entry and horizontal slider to one Core split-quantity owner, preserve retail count-first naming and exact 1000-step rounding, refresh on stack changes, and consume the selected amount during merges. Conformance covers the production DAT fixture and retained pointer/focus paths.
Co-Authored-By: Codex <codex@openai.com>
Visual gate against retail surfaced several fidelity gaps in the selected-object
strip; all fixed and user-confirmed. Faithful to gmToolbarUI::HandleSelectionChanged
(acclient_2013_pseudo_c.txt:198635) + RecvNotice_UpdateObjectHealth (:196213).
- UiMeter.DrawHBar: guard each slice on `id != 0` BEFORE resolve. resolve(0)
returns the 1x1 magenta placeholder with a non-zero GL handle, so the single-
image meter (caps id=0) was drawing 1px magenta caps at the bar's ends. The
3-slice vitals meter (all ids set) was unaffected. (the magenta-lines bug)
- SelectedObjectController: meter visibility is now UpdateHealth-driven (shown when
health is known for the selected guid — HasHealth at select or HealthChanged),
not shown-on-select; brief green selection flash via Tick revert; overlay floated
above the meter so the flash isn't hidden by the bar; name top-aligned into the
bar sprite's black band (NameBandHeight) with the bar below.
- GameWindow.IsHealthBarTarget: gate the health bar on the server PWD bits
BF_ATTACKABLE (0x10) | BF_PLAYER (0x8) — friendly/vendor NPCs and attackable
Doors (Misc type) are name-only; players/monsters get the bar. Replaces the
too-loose IsLiveCreatureTarget. Wired SelectedObjectController.Tick in OnUpdate.
- CombatState.HasHealth(guid): distinguishes a known health value from the 1.0
default, so a re-selected already-assessed target shows its bar immediately.
- TextureCache.GetOrUploadRenderSurface: resolve the surface's DefaultPaletteId
so paletted (P8/INDEX16) UI sprites decode instead of falling to magenta.
- ToolbarController.HiddenIds: also hide 0x100001A3 (stack-entry box) — retail
hides it in HandleSelectionChanged; it was rendering as a stray black box.
Divergence register: AP-47 (meter-visible timing) retired (now faithful); AP-46
rewritten to the BF_ATTACKABLE/BF_PLAYER gate approximation. Full suite green
(2,688 passed / 4 skipped). User-confirmed: name on top, NPC name-only, monster
bar on assess, green flash, no magenta.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Port of gmToolbarUI::HandleSelectionChanged (acclient_2013_pseudo_c.txt:198635).
When the player selects a world object the action bar's bottom strip shows the
object name + (for player/pet/attackable targets) a live Health meter; deselect
clears it. Mana (#140) + stack slider deferred.
- SelectedObjectController (new): clear-then-populate on selection change; sets
name (UiText child, VitalsController pattern), overlay state (ObjectSelected /
StackedItemSelected via UiDatElement.ActiveState), shows the health meter and
sends QueryHealth for health targets. Subscribes via a delegate seam (no
GameWindow coupling).
- GameWindow: _selectedGuid field -> SelectedGuid property + SelectionChanged
event (fires on actual change only); 3 write sites converted, reads untouched.
All selection-write paths (LMB pick, Tab/Q, despawn-clear via Tick()) run on
the render thread, so the event-driven UI mutation is single-threaded.
- WorldSession.SendQueryHealth (0x01BF) — wraps SocialActions.BuildQueryHealth.
- DatWidgetFactory.BuildMeter: handle the single-image toolbar meter shape
(back-track on the element's own DirectState, fill on one Type-3 child). The
sprites go in the TILE slot (DrawMode=Normal tiles to full bar geometry per
UIElement_Meter::DrawChildren) — a left-cap assignment would gap/clamp a
sub-140px sprite. Vitals 3-slice path unchanged.
- ToolbarController.HiddenIds: A1 (health) now owned by SelectedObjectController;
A2 (mana) + A4 (stack) stay hidden (deferred) so their dat back-tracks don't
render as stray empty bars.
Adversarial Opus review found + fixed: the mana-meter orphan (A2 left unhidden)
and the meter tile-vs-cap render bug (C1). Divergence rows AP-46 (health gate
approximation: IsLiveCreatureTarget vs IsPlayer||pet||attackable) + AP-47
(meter shown on select vs on UpdateHealth reply). Spec §5 corrected.
Build + full test suite green (2,684 passed / 4 skipped). Health meter render
fidelity (full-width fill + fraction mapping) pending the user's visual gate.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
