Stock 1.12.1 already defines DAYS_ABBR/HOURS_ABBR/MINUTES_ABBR/
SECONDS_ABBR (bare labels: "Day"/"Hr"/"Min"/"Sec") and the four
*_ONELETTER_ABBR. Our enUS.lua loads after FrameXML, so redefining
them with the modern "%d |4Day:Days;" format silently overwrote the
native values and broke any native/other-addon code reading the bare
labels.
Remove those redefinitions from enUS.lua (keep only the strings stock
genuinely lacks), and drop SecondsToTime / SecondsToTimeAbbrev /
FormatUnit from TimeUtil — they were the only consumers of the
modern-format _ABBR strings. SecondsToClock stays (uses the
stock-missing HOURS_MINUTES_SECONDS / MINUTES_SECONDS). Remaining
enUS entries sorted, all still have live consumers.
`GetPhysicalScreenSize()` → (widthPixels, heightPixels). Modern WoW
added this native in 7.0; FrameXML's PixelUtil builds its entire
pixel↔UI-unit conversion on it (768/physicalHeight). 1.12 lacks it and
has no pre-parsed pixel-dimension global — GetScreenWidth/Height are
UI-space (UIParent aspect at [UIParent+0x7c]), and the engine only
holds the current mode as the gxResolution CVar string, sscanf'd on
demand by SetScreenResolution / display-init. So the native reads the
gxResolution CVar directly (FUN_FIND_CVAR, value at +0x20) and parses
"WIDTHxHEIGHT" — the same source pfUI's GetPerfectPixel uses. Verified
in-game: 2560x1080.
7.0 was the only binary exposing the function, but retail .text is
encrypted on disk so static decompilation yields garbage; derived the
implementation from the 1.12 binary instead.
Also backports FrameXML/PixelUtil.lua (SetPoint/SetSize/SetWidth/
SetHeight/SetStatusBarValue/Get*). Its math deps (Round/Lerp/
ClampedPercentageBetween) already live in MathUtil; it consumes the
native GetPhysicalScreenSize (no Lua polyfill).
`C_Spell.CastAtUnit(spellID, unit)` and `C_Item.UseAtUnit(itemInfo,
unit)` are the unit-position analogs of the existing `…AtCursor` pair:
they commit the engine's in-flight ground-target placement at the
resolved unit's world position instead of the cursor raycast.
ClassicAPI's `/cast [@unit]` / `[@unit]` for AoE reticle spells and
on-use items. Verified in-game against "player", "target", and
"nameplate1".
Implementation reuses the cursor path's placement primitives:
- Extracted `Spell::AtCursor::DispatchSpellCast` + `CommitAtCoords`
(explicit-coords commit, sharing the same placement-ready gate as
the cursor `Resolve`).
- New header-only `Unit::Position` (vtable slot-5 `GetPosition` read);
migrated `UnitInRange` onto it to drop the duplicate copy.
The unit is resolved before the cast/use is dispatched, so an absent
unit fails cleanly without arming placement mode or consuming the
item. Unknown token strings raise the engine's standard "Unknown unit"
error, matching `UnitHealth("garbage")`.
Source the strings FormattingUtil/TimeUtil consume (LARGE_NUMBER_SEPERATOR,
PERCENTAGE_STRING, GOLD/SILVER/COPPER_AMOUNT_SYMBOL, the time
abbreviations + D_* + *_ONELETTER_ABBR, clock formats) from Classic Era
enUS GlobalStrings and define them in locales/enUS.lua, which loads
before the Util files. `or`-guarded so any value the 1.12 client already
defines is preserved (the time abbrevs use |4 plural tokens, which 1.12
supports — our existing ITEMS_* strings already rely on |4).
The Util files no longer define these (single source of truth in the
locale layer); they retain only the non-locale numeric constants
(COPPER_PER_*, SECONDS_PER_*) and the formatting logic. FormatUnit still
inserts the count whether the unit string carries a %d or is a bare label.
1.12's native unit-abbreviation globals (HOURS_ABBR, MINUTES_ABBR, …)
are bare labels ("Hr", "Min") with no %d placeholder, unlike modern's
"%d Hr". So string.format(HOURS_ABBR, n) returned just "Hr" — the count
was lost (SecondsToTime(3661) -> "Hr Min" instead of "1 Hr 1 Min").
Add a FormatUnit helper that string.formats when the unit string has a
%d and otherwise prepends "<count> ", so it's correct whether the global
is modern ("%d Hr") or a bare 1.12 label ("Hr"). Verified expectation:
SecondsToTime(3661) -> "1 Hr 1 Min".
Proactive Tier-1 backports of the most widely-assumed FrameXML utility
globals, ported to Lua 5.0 (no `#`/`%`/string-method syntax; `%` ->
math.mod). Feasible subsets only:
- FrameUtil: RegisterFrameForEvents / UnregisterFrameForEvents,
GetRootParent, SetParentMaintainRenderLayering, DoesAncestryInclude(Any).
Omits the fade/flash helpers (native in 1.12) and the scale/layout/
secure/RegisterUnitEvent helpers (APIs absent here).
- FormattingUtil: FormatLargeNumber, FormatPercentage, FormatFraction,
SplitTextIntoLines, SplitTextIntoHeaderAndNonHeader, GetMoneyString.
GetMoneyString uses text symbols (g/s/c) since coin texture markup
doesn't render in 1.12; currency/cost helpers omitted.
- TimeUtil: SecondsToTime, SecondsToClock, SecondsToTimeAbbrev,
ConvertSecondsToUnits, SecondsToMinutes/MinutesToSeconds + unit
constants. Omits the modern SecondsFormatterMixin.
FrameXML format-string globals (LARGE_NUMBER_SEPERATOR, PERCENTAGE_STRING,
DAYS_ABBR, GOLD_AMOUNT_SYMBOL, …) are defined with `or` fallbacks so a
native/localized value wins when present and we degrade sanely otherwise.
Loaded after MathUtil so FormatPercentage's Round() is available.
Reverts the relocate-to-ColorUtil approach (b2db474): PLAYER_FACTION_COLORS
and BAG_ITEM_QUALITY_COLORS belong in Constants.lua, matching Blizzard's
own layout (its FrameXMLBase/Constants.lua references the named *_COLOR
objects directly).
Blizzard gets away with it via load order: the color objects are
materialized before its Constants loads. We now do the same — move
Mixin.lua + Color.lua above Constants.lua in the TOC, so Color.lua's
C_UIColor loop creates COMMON_GRAY_COLOR / PLAYER_FACTION_COLOR_* before
Constants.lua references them. Safe because Color.lua only depends on
Mixin.lua (+ the native ITEM_QUALITY_COLORS and the DLL's C_UIColor),
nothing else in the load chain; Constants.lua doesn't depend on the
utils that now follow it.
PLAYER_FACTION_COLORS was built in Constants.lua (loads early), but it
references the PLAYER_FACTION_COLOR_HORDE/ALLIANCE color objects that
Color.lua only materializes from C_UIColor at a later load step. So the
table captured nils -> ended up {}, and GetFactionColor always returned
nil. (Can't just reorder the TOC: Color.lua depends on ITEM_QUALITY_COLORS
from Constants.lua, so it's circular.)
Move the PLAYER_FACTION_COLORS construction to ColorUtil.lua, which loads
after Color.lua (and is where its only consumer, GetFactionColor, lives),
so the color objects exist by then. PLAYER_FACTION_GROUP stays in
Constants.lua (no such dependency).
GetFactionColor needs the FrameXML PLAYER_FACTION_GROUP global (faction
tag -> 0/1 index) to pair with our PLAYER_FACTION_COLORS, but we only
shipped the latter, so the lookup always returned nil. Add
PLAYER_FACTION_GROUP next to it, matching Blizzard SharedConstants
({[0]="Horde",[1]="Alliance",Horde=0,Alliance=1}); `or`-guarded so a
native definition wins if present.
Fills the gaps found diffing against Blizzard's reference (same class of
missing-global that broke tContains), porting to Lua 5.0:
Color.lua:
- WrapTextInColor(text, color)
- ColorMixin:IsRGBEqualTo (IsEqualTo now routes through it, per ref)
- ColorMixin:GetHSL
- ColorMixin:GenerateHexColorNoAlpha
ColorUtil.lua:
- CreateColorFromRGBHexString / CreateColorFromRGBAHexString
- IsRGBAEqualToColor
- RGBToColorCode / RGBTableToColorCode
- GetFactionColor (guarded: PLAYER_FACTION_GROUP/COLORS may be absent)
Color.lua and ColorUtil.lua now match the reference's function set. The
deliberately-skipped Secure* variants in Mixin/Pools (modern secure-
template machinery with no 1.12 equivalent) are left out.
tDeleteItem (added in 11191e8) used the 5.1 length operator `#tbl`, which
doesn't exist in 1.12's Lua 5.0. Swap both occurrences to tGetN(tbl)
(table.getn) to match the rest of the file.
Channel end time read SpellDuration.dbc base directly; route it through
the engine's own duration helper FUN_006ea000 (base + level scaling +
duration SpellMod op 1, unit arg 0 = local player) — mirroring how the
cast time is sourced, so the channel bar matches the engine and any
per-level / talent duration is honored. Stock vanilla channels are
fixed-duration, so no change there (verified: Blizzard still 8s).
Completes the SpellMod op map derived this session: duration 1, range 5,
radius 6, cast time 10, cost 14.
GetSpellPowerCost had a duplicate power-type -> token map. Promote the
existing UnitPowerType mapping out of Power.cpp's anonymous namespace
into a shared unit/Power.h (Unit::Power::PowerTypeToken) and have both
consumers use it. Out-of-range PowerType now reports "UNKNOWN" in
GetSpellPowerCost (was ""), matching UnitPowerType.
The mana check read the base ManaCost (+0x80) directly, so a spell with
a cost-reduction talent was wrongly reported as unusable (noMana) when
the player's power sat between the reduced and base cost. Route it
through the engine's effective-cost helper FUN_006e31b0 (op-14 SpellMod +
descriptor power-cost mods + ManaCostPercent), with a base-cost fallback
if the engine can't resolve a cost.
Example: Blizzard base 1400, 3/3 Frost Channeling -> effective 1190; with
~1200 power the spell is now correctly usable instead of noMana.
Returns a SpellPowerCostInfo[] (one entry in vanilla — spells have a
single power cost), or nil for an unknown / no-cost spell. The `cost` is
the player-effective value via the engine's own cost helper FUN_006e31b0
(base + level + ManaCostPercent-of-resource + descriptor power-cost mods
+ the cost SpellMod, op 0xE) — so it matches what casting actually
deducts, talents included.
Verified in-game: Blizzard (10187) on a Mage with 3/3 Frost Channeling
returns cost 1190 = 1400 base x 0.85 (15% reduction).
Keeps C_Spell.GetSpellInfo's `cost` field as the base ManaCost (intrinsic
spell data); GetSpellPowerCost is the player-modified counterpart — same
base/modified split as GetSpellRadius. Calling the full engine helper
beats replicating op 14 via Spell::Mod, since the helper also folds in
the descriptor power-cost modifiers.
Fields: type (Enum.PowerType), name (power token), cost, minCost (==cost,
no optional cost in vanilla), costPercent, costPerSec 0, requiredAuraID 0,
hasRequiredAura false.
Docs: API.md section + README Spell row.
Cast/channel info for the local player with full timing, plus channel
spellID/name/texture for any unit. Vanilla stores no cast times anywhere
readable (confirmed via the cross-binary technique: 3.3.5 reads them from
per-unit CGUnit fields that don't exist in 1.12), so the player's
cast/channel start/end are self-tracked on WorldTick:
- cast: VAR_CURRENT_CAST_SPELL 0->nonzero transition stamps startMs = now
(FUN_OS_TICKCOUNT_MS) and endMs = now + effective cast time. The cast
time comes from the engine's own helper FUN_006e3340 (base + level
scaling + cast-time SpellMod op 10 + the caster's haste/cast-speed
multiplier), the same one the engine cast-start uses -- so end time
matches the cast bar, e.g. a talented Frostbolt reads 2.5s not 3.0s.
Instant casts (cast time 0) and channels are filtered out.
- channel: player's broadcast UNIT_FIELD_CHANNEL_SPELL (+0x228),
endMs = now + SpellDuration base ms.
FUN_OS_TICKCOUNT_MS shares GetTime()'s epoch (x0.001), so startMs/endMs
feed addon progress math directly. Other units: channel info from the
broadcast +0x228 field (no times, since per-unit start isn't tracked);
regular casts return nil (no vanilla data -- remote casts are transient
spell-visual objects, not a queryable record).
isTradeskill is real: the spell's SPELL_ATTR_TRADESPELL flag (Attributes
bit 0x20). Remaining placeholders: castID/castBarID nil, notInterruptible
false, delayTimeMs 0.
Verified in-game: Blizzard channel 8s, Conjure Food cast 3s, talented
Frostbolt 2.5s, correct icons/spellIDs, GetTime()*1000 between start/end;
isTradeskill true for Rough Blasting Powder (Engineering).
Docs: API.md sections + README Spell row; TODO #68 marked done.
GetSpellRadius now returns (baseRadius, modifiedRadius). modifiedRadius
applies the local player's talent/item radius modifiers; equal to base
when no modifier matches.
The modifier logic is extracted into a reusable Spell::Mod helper
(src/spell/Mod.{h,cpp}) since the same engine SpellMod mechanism drives
cast time, mana cost, duration, range, cooldown, etc. — only the op
differs. Spell::Mod::Apply(spellRecord, op, base) replicates
FUN_006e6b30/FUN_006e6bf0: gate on the spell's SpellFamilyName matching
the player's class family (ChrClasses.dbc[class]+0x3c), sum the player's
flat/percent tables (0x00CEAD60 / 0x00CECB30, stride 0x74) over the
spell's 64-bit SpellFamilyFlags, then (base + flat) * (100 + pct) * 0.01.
Inherently local-player-only — the client tracks spell mods for the
player alone.
Offsets: VAR_SPELLMOD_FLAT/PCT_TABLE, SPELLMOD_SLOT_STRIDE/COUNT,
SPELLMOD_OP_RADIUS, OFF_SPELL_RECORD_FAMILY_NAME/FLAGS/ATTRIBUTES_EX2,
SPELL_ATTR_EX2_NO_SPELL_MODS, OFF_CHRCLASSES_SPELL_FAMILY.
Verified in-game: Frost Nova (10230) -> (10, 10) without Arctic Reach,
(10, 12) with rank 2 (+20%). op 6 = radius confirmed via FUN_006e6350;
percent constant 0.01 confirmed from .rdata.
C_Spell.GetSpellRadius(spellID) plus the vanilla-positional
GetSpellRadius(slot, bookType). Returns the spell's AOE radius in yards,
or nil for a non-AOE spell.
Radius is per-effect: Spell.dbc EffectRadiusIndex[3] at record +0x160
(verified via FUN_006e6350, which reads spellRec[+0x160]/[+0x164] for
effects 0/1), each resolving into SpellRadius.dbc (records 0x00C0D7B0,
count 0x00C0D7B4; base radius f32 at record +0x04). Returns the largest
base radius across the three effects. Base radius only — no caster-level
scaling, matching modern GetSpellRadius.
Dual signature mirrors SpellHasRange/GetSpellInfo: namespaced form takes
a spell identifier, the bare global takes a (slot, bookType) spellbook
position (bookType "pet" -> pet book, else player book).
Docs: API.md section + README Spell row; TODO #38 marked done.
Add the API.md section (return contract, always-English rationale,
common mechanic IDs + the Sap/Gouge=14 vs nonexistent 30 caveat) and
the README Spell-row entry.
Reads the spell-level Mechanic field straight out of Spell.dbc
(record +0x14) and resolves the name from SpellMechanic.dbc. Covers
every spell the client knows (not just the player's spellbook), with no
caching or network round-trip since the DBCs are resident from boot.
Replaces hand-maintained spellID -> mechanic tables addons keep because
vanilla exposes no Lua reader for the field (e.g. SuperCleveRoidMacros'
CCSpellMechanics, used for [cc:stun] conditionals).
Returns: nil for an invalid spellID; (0, nil) for a known spell with no
mechanic; (mechanicID, name) otherwise. The name is always the enUS
column (locale-independent) since it's a stable token consumers
string-match against; nil if the mechanic has no record / no English
name (numeric ID still returned).
Mechanic offset (+0x14) verified two ways: schema anchor off the
verified Attributes@+0x18, and FUN_006e9ca0 (mechanic-immunity check)
reading spellRec[+0x14] as the spell's mechanic. SpellMechanic.dbc name
at record +0x04 confirmed in-game: 118 -> (17,"polymorphed"),
339 -> (7,"rooted"), 133 -> (0,nil). The .dbc has 27 rows; only the
enUS name column is populated.
commit 52064f025188267ab44dae41af93261ae1bcfc03
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Thu Jun 18 23:33:09 2026 -0500
Item cache: fast-fail server rejections; report pending overflow
Two follow-ups to the WarmCache startup-latch fix.
Failure fast-path: the escalation path used a no-op callback, so a
server "no such item" reply (high-bit itemID — the FUN_0055BDB0 failure
branch does NOT set the entry loaded flag) went undetected and the item
waited out the full kMaxWaitTicks (~20s) before reporting failure.
Replace it with ItemLoadCallback_Escalated, which marks the pending
entry failed when the engine reports !success. A shared TryCompletePending
helper resolves each entry either way (loaded -> success, failed ->
failure) and is now used by both the response-handler sweep and the
tick, so rejections fire in the same response. Genuinely lost queries
(server silent) still fall back to the timeout.
Overflow: Track now returns false when the pending set is full instead
of silently dropping, and the explicit RequestLoadItemData/ByID Lua
functions propagate that as their boolean return. Bump kMaxPending
256 -> 512 since the unsettled window now also holds implicit warmups.
commit 797a73de44e6a4ceb0ac434944a137dcd47c18f3
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Thu Jun 18 23:24:04 2026 -0500
Item cache: gate WarmCache behind a startup-settled latch
The previous fix (ad13440) deferred only the explicit
RequestLoadItemData path out of the engine startup window. The implicit
WarmCache path — the GetItemInfo hook plus every GetItem* getter cache-
miss warmup — still eagerly called CacheFetch(callback) with no gate, so
it orphaned cache entries during the exact PLAYER_ENTERING_WORLD /
addon-load window the original bug was about. pfQuest/Questie scan item
data there, so those items stayed permanently nil for the session.
WarmCache now gates on a one-way g_settled latch:
- Pre-settle (startup): defer through the same pending set as
RequestLoadItemData, tagged implicit so the response-handler sweep
fires GET_ITEM_INFO_RECEIVED (not ITEM_DATA_LOAD_RESULT).
- Post-settle: eager CacheFetch as before, for low-latency mid-session
lookups.
The latch flips true the moment the engine response handler fires (proof
dispatch is live) or after kSettleTicks world ticks, whichever first.
Per-process, not re-armed on relog (cold-WDB race is initial-login only).
Unify both paths through Track(itemID, implicit); sweep and timeout fire
the correct event by tag. Escalation now gates on g_settled instead of a
fixed per-item tick count, draining the startup backlog as soon as it is
safe. Bump kMaxPending 64 -> 256 to absorb the larger startup burst.
The original implementation only consulted the SMSG_NAME_QUERY_RESPONSE-
fed player NameCache, which only populates for GUIDs the engine has
explicitly queried — so currently-synced players (just-targeted, in
view, etc.) typically miss until the engine sends a name query.
Switch to vanilla's own `Script_UnitName` path: resolve the GUID via
`FUN_CLNT_OBJ_MGR_OBJECT_PTR` with TYPEMASK_OBJECT, then call the
engine's polymorphic name getter `FUN_OBJECT_GET_NAME`. That routes
through the player NameCache AND the creature cache, so both player
and NPC GUIDs return real names for any currently-synced unit. Gate
on the "UNKNOWNOBJECT" / "Unknown Being" sentinels so non-unit
objects fall through. Persistent NameCache fallback kept for ex-units
that have left the engine's sync window.
Modern WoW returns the subtitle as UnitName's second return; vanilla
1.12's UnitName returns only one value, leaving no direct route to
the subtitle. UnitSubName reads it straight off the engine's creature
cache (`[unit + 0xB30] → [+0x10]`), no tooltip scrape required.
Returns nil for players, unsynced NPCs, and NPCs with no subtitle.
Modern backport that reuses GetPlayerInfoByGUID's lookup chain
(engine NameCache → persistent on-disk fallback) but pushes only
(name, realm) — narrower return shape for callers that only need
the name. realm is always "" in vanilla (single-realm, engine
doesn't populate per-player realm).
CMakeLists rejects CLASSICAPI_TAG values that aren't empty or
vMAJOR.MINOR.PATCH. Branch pushes were passing `github.ref_name`
(= the branch name, e.g. "master") and failing the regex.
Was tag-only. Now builds on every push to any branch, uploading the
DLL as a workflow artifact (ClassicAPI-<sha>, ~90-day retention) so
pre-release DLLs are downloadable from the run page without cutting
a tag. Tag pushes still build + stamp the embedded TOC version +
publish the GitHub release.
Renamed release.yml -> build.yml to match the broader role.
The engine stores expiration timestamps as GetTickCount-derived uint32
milliseconds. We were reading as signed int, then bailing on `<= 0` to
catch the "no timing" sentinel of 0. Once system uptime crossed 2^31 ms
(~24.86 days) the high bit flipped, signed reads went negative, and
every active buff started reporting expirationTime = 0 while
GetTime() kept working through a different code path.
Switch to uint32_t with `== 0` sentinel check. The unsigned read flows
correctly through GetTickCount's 2^32-ms (~49.7-day) wrap because
addon math (`expirationTime - GetTime()`) operates on values that wrap
together — same as the engine's own `expirationTable[slot] - currentMs`
subtraction.
Vanilla's CMSG_SPLIT_ITEM rejects splits where count == srcStack
(EQUIP_ERR_COULDNT_SPLIT_ITEMS — would leave source empty). The
drag-and-drop UI's "drop whole stack on matching stack" goes through
CMSG_SWAP_ITEM, which the server merges for same-item destinations.
In MoveCount, read ITEM_FIELD_STACK_COUNT off the resolved source
CGItem and route count == srcStack to Item::Swap::Containers. Also
return false locally for count > srcStack (server would reject anyway,
fail fast). count < srcStack keeps the existing SPLIT path.
3.3.5+ fires UNIT_FACTION("player") on rep-state changes that vanilla
silently drops. Hook each path's inner setter / SMSG handler and fire
the event post-original via the engine's printf-style dispatcher:
- FactionToggleAtWar — fire if rep slot's flag byte changed
- SetFactionInactive / SetFactionActive — same shape
- SMSG_SET_FACTION_STANDINGS — fire if any of 64 flag bytes changed
(catches AT_WAR side-effect flips on standing threshold crossings)
- SMSG_SET_FACTION_ATWAR — unconditional after byte write
- SMSG_INITIALIZE_FACTIONS — fire after initial sync at login
Hoist the event-table strcmp walk into Event::Custom::LookupByName
(sibling to Lookup, which only consults our reservation cache). Replaces
duplicated walks in event/Util.cpp::IsKnownEventName and the polyfill.
Re-registers the standard five entries (create / resume / yield /
status / wrap) on the global `coroutine` table. The C-level
machinery (lua_resume, lua_yield, lua_newthread, threads as tag 8)
was always linked into the engine; only the script-side library
was stripped from the registration tables at build time.
Two WoW-specific quirks discovered along the way and worked
around:
- `lua_resume` returns 1 for BOTH yield and caught runtime error.
Disambiguated by checking `co->ci > co->base_ci` — a yielded
coroutine still has its body frame on the call stack; an
errored one has been unwound to base_ci.
- The byte at L+0x31 is `allowhook` (saved/restored around
protected calls), not `L->status` as it initially looked.
Status reading is replaced by the ci comparison above.
Single-call kitchen-sink reader for the cached `ItemStats_C` record.
Returns a table with ~50 fields covering everything the cache holds:
identity (name, description, icon, class/subclass/equipLoc strings),
quality + bind type + decoded flag bits (isConjured, isOpenable,
isLootable, isWrapper), stacking, economy, requirements, equipment
stats (sparse map keyed by statType), resistances, weapon damage
arrays, spell slots, and the misc fields (lockID, itemSet, pageText,
startQuest, area/map, etc.).
Returns nil for uncached / invalid input — passive reader, no cache
warming. Caller wanting to ensure load should `RequestLoadItemData`
and listen for `ITEM_DATA_LOAD_RESULT`.
Implements via existing offsets where available; adds the remaining
`ItemStats_C` fields to `Offsets.h` based on VanillaHelpers's full
struct layout, anchored against the offsets we'd already verified
(`m_stackable@+0x60`, `m_lockID@+0x1AC`, `m_bagFamily@+0x1D0`, etc.).
Bursts of `WarmCache` calls for the same uncached itemID (e.g. Bagnon
re-enumerating a slot while the SMSG_ITEM_QUERY_SINGLE_RESPONSE is in
flight) were each appending a fresh `Implicit` callback to the
entry's pending list. When the response landed the engine walked the
list and fired our callback N times, surfacing as duplicate
`GET_ITEM_INFO_RECEIVED` events for the same itemID.
`_GetRecord`'s 6th arg (`param_5` in the decompilation, previously
documented as the unused `requestIfMissing` bool) controls whether
the engine dedups appends. With `0` it always appends; with non-zero
it walks the existing list and skips the append on a matching
`(callback, userData)` node. We always have the same callback ptr and
encode the itemID in userData, so flipping the bit collapses repeat
warmups for the same item to a single registration.
`Item:CreateFromItemID(X):ContinueOnItemLoad(cb)` was leaving items
permanently uncached when pfQuest called it at addon-load /
PLAYER_ENTERING_WORLD against an empty WDB. Three intertwined bugs:
1. `GetItemInfoInstant` returned no values for uncached items, so
`ItemMixin:GetItemID()` returned nil and `AddCallback(nil, cb)`
crashed with "table index is nil".
2. `DoesItemExistByID` returned false for uncached items, which made
`IsItemEmpty()` return true and erroneously gated `ContinueOnItemLoad`
on cache state — modern semantic is "is this a valid ID", independent
of cache.
3. Our `RequestLoadItemData` path eagerly called `CacheFetch` with our
own callback at addon-load. That created a cache entry before the
engine's natural inventory prefetch had a chance to run; the engine
then saw the entry already existed and skipped its own SMSG, leaving
the entry permanently pending. `GetItemInfo` retries didn't help —
subsequent `_GetRecord` calls just appended callbacks to the dead
list, no new query was sent.
Fixes:
- `GetItemInfoInstant` always pushes the itemID (echoed from input) and
the remaining 6 fields when cached / nils when not. No cache touch.
- `DoesItemExistByID` returns `itemID > 0`. No cache touch.
- `RequestLoadItemData` tracks itemIDs in a pending set without touching
the cache. We hook `FUN_ITEMSTATS_CACHE_RESPONSE` (the engine's
`SMSG_ITEM_QUERY_SINGLE_RESPONSE` handler at `0x0055BDB0`) and sweep
pending after the engine fills entries — observing the natural
prefetch instead of racing it. `WorldTick` only handles escalation
(after ~60 ticks, kick a query for items the engine doesn't prefetch
itself, e.g. quest rewards) and the hard timeout.
Also fix the truthiness bug in modern (id, success) events:
`Event::Custom::Fire("%d%d", id, success)` was pushing `success` as a
Lua number — both 0 and 1 are truthy, so `if success then` fired the
success branch on failed loads. New `Event::Custom::FireIdSuccess`
pushes `1` for success and `nil` for failure (via the engine's
`lua_pushstring(NULL) → lua_pushnil` tail-jump). `ITEM_DATA_LOAD_RESULT`,
`GET_ITEM_INFO_RECEIVED`, and `QUEST_DATA_LOAD_RESULT` now all use this.
Hoists the macro-parser hook + long-line guard + per-line walk out
of Spell::CastNoToggle into a new Spell::MacroPrimarySpell module
that dispatches to (prefix, extractor) pairs registered by feature
modules. The body walk now visits each line once and checks every
registered prefix at each position; extractors only parse args
after a matched prefix.
Registers a pattern for C_Spell.CastAtCursor(<spellID>) so macros
like `/run C_Spell.CastAtCursor(10)` get tagged with the spellID
the same way `/cast Blizzard` does, driving action-bar cooldown
overlays and auto-repeat highlight in pfUI etc.
Vanilla 1.12.1 requires a manual click on the AoE reticle to land a
ground-target spell. This pair of functions fires the cast (or
on-use item spell) and immediately commits placement at the
cursor's current world position, matching modern `[@cursor]`
semantics:
C_Spell.CastAtCursor(spellID)
C_Item.UseAtCursor(itemInfo)
Chains the engine's own primitives — name-resolver to spellbook
slot, cast dispatcher, then a screen→world raycast via the existing
mouseover machinery and the placement-commit function. Cancels
cleanly when the spell isn't ground-target or the cursor is over
UI / off-screen.
Also extracts the bag-search pattern (used by `UseAtCursor` and
existing by-name action APIs) into `Item::Location::FindByArgInBags`,
alongside the existing `FindByItemID` / `FindByGUID` family.