ClassicAPI carries notInterruptible on C_Spell.UnitCastingInfo (arg 8) and
UnitChannelInfo (arg 7); both call sites already stepped over it positionally.
The target, focus and nameplate bars now paint it.
There is no shield art to hang on the icon -- this client ships three
CastingBar textures (Border, Flash, Spark), and Blizzard's shield arrived in
3.x -- so the state is carried by color instead: the bar and the icon's
border both take a new appearance.castbar.shieldcolor, which keeps it legible
on a bar configured without an icon. One api helper paints both so the two
modules cannot drift apart on it.
Dusty rose (.8,.45,.45) rather than a steel grey. Grey is the retail
convention, but pfUI's casting color is already a near-grey pale blue, so the
two separated on brightness alone and washed out against the bar texture. The
rose sits above failcolor in luminance and well below it in saturation, and
the two are further told apart by motion: a failed cast flashes the bar full
and fades, a shielded one fills normally and tints the icon border, which
failcolor never touches.
Not on the player's own bar. The flag is measured against the reader's own
interrupts and silences, so on your own cast it answers a question nobody
asked.
Stamped once at cast start, with the rest of the cast data, rather than
re-polled. ClassicAPI evaluates lazily from the caster's live auras but does
not yet fire the 3.3.5 UNIT_SPELLCAST_INTERRUPTIBLE / _NOT_INTERRUPTIBLE
change events, and polling per frame would undo the event-driven rework these
bars just got -- so a caster who bubbles mid-cast keeps the bar it started
with until the next one.
Two limits worth knowing when it stays dark. The value is relative to your own
kit -- with no interrupt and no silence known it is never reported, whatever
the cast -- and a creature's built-in interrupt immunity lives in
creature_template, which 1.12's SMSG_SPELL_START does not carry, so a boss
with that immunity reads interruptible while your kick still fails on it.
Eight commits off classicapi_next.
ClassicAPI's RegisterUnitEvent registers for an event but only delivers it
when arg1 is one of the given units, so a handler for one unit stops waking
for every other one in the world. The 26 registrations whose unit set is
fixed at registration time now name it. The rule throughout is register the
superset and keep the handler's own check -- the filter narrows what arrives,
it does not decide what to act on. Guards that look unreachable stay put: the
filter applies only when arg1 is a string, so an event that fires with a
number or no argument is delivered as if plainly registered.
Frames whose unit changes at runtime own their subscriptions instead of
sorting events out per event. unitframes points them at the unitstr
UpdateVisibility already computes -- replacing a string concat, and on a miss
a second concat plus a UnitGUID call, for every frame on every unit event in
the world -- and a frame that is not in use drops its unit events entirely.
nameplates registers per plate against the plate's own token, which is also
the only workable shape: slots have no cap, so any nameplate1..N list would
have been a guess that fails in exactly the crowded scenes where plates
matter. marktracking names mark1 through mark8. A registration keeps its
kind, so none of these can be plain-registered first.
Both teardown paths PLAYER_LOGOUT guards -- the crash 132 -- now cover the
per-frame subscriptions: plates tear down rather than dispatching through
logout, and a unit frame takes itself off the visibility scan so it cannot
re-register what it just dropped.
marktracking also drops its once-a-second full rebuild, which ran for the
whole session whether or not a marker existed anywhere. The ticker is created
and cancelled with group membership. It is deliberately not keyed on a mark
being visible -- a marker on an out-of-range unit shows no row, and that is
the case the poll exists to catch.
nameplates gates the per-plate update against the floor across all four
throttle categories before classifying it, instead of running a GetAlpha, a
castbar IsShown, a cast lookup and up to two libthrottle:Get resolutions on
plates throttled to 10fps that were going to return anyway. Nothing that
would have updated can be turned away by a floor. The four throttles resolve
in CacheConfig, where config changes already land.
energytick sweeps the clock the server actually runs. There is one regen
timer for every power, re-armed every 2s by Player::RegenerateAll and never
touched by casting; the five-second rule changes what a tick pays, not when
it lands. The sweep is a free-running phase lock on that clock, so
Illumination refunds, potions and a Mana Spring totem on its own phase no
longer snap the spark mid-cycle, and an 80ms band keeps a correct tick from
hitching it at the wrap. The FSR window shades rather than predicting a share
of spirit the client cannot compute -- the Casting Regen item ladder is equip
auras absent from the buff list. The energy period is summed from
SPELL_AURA_MOD_ENERGY_REGEN_TIME across the spellbook and buffs, so Blade
Rush is found without GetTalentInfo(2, 16), an ordinal that does not fail
when the tree changes but reads another talent's rank.
macrotweak is gone -- ClassicAPI 1.15 covers it -- with its config entry, its
GUI block, its translations in all eight locales, and actionbar's
ButtonMacroScan, the #showtooltip scanner that fed it.
* nameplates: source totem icons from UnitCreatedBySpell
Read the totem's icon from the totem-drop spell (UnitCreatedBySpell +
GetSpellTexture) instead of the self-aura at index 1 plus a
UNIT_SPELLCAST_SUCCEEDED capture for active totems. The drop spell is a
broadcast descriptor field present for every summoned unit in range, so
it resolves immediately for passive and active totems alike and gives the
totem's own icon rather than the attack-spell proxy. Drops the
UNIT_SPELLCAST_SUCCEEDED registration and handler.
Re-read the spell each render and key the cached texture on the spell id
so an in-place totem swap (same unit, new drop spell -- no plate re-add)
refreshes the icon without needing the plate to leave and re-enter view.
* CAPI min bumped to 1.13.1
* Gracefully disable pfUI when ClassicAPI is missing
Add API_Check.lua as the first TOC entry. When the ClassicAPI DLL is
absent or below the minimum version it sets pfUI_disabled and stands up
an inert stub so the rest of the addon no-ops instead of flooding load
errors: modules and skins register their bodies into no-ops, and the
setfenv'd api/lib files run inside an environment where CreateFrame and
any missing global resolve to a null object -- so no real frames or live
handlers are created and missing API calls just return null. pfUI.lua
bails early on pfUI_disabled.
* auras: scan through GetAuraSlots instead of by-index loops
Every aura scan loop (unit frame buffs/debuffs, dispel indicators, buff and
custom indicators, player buff frame, buffwatch bars, tooltip buff row) now
enumerates a unit's auras once with C_UnitAuras.GetAuraSlots and reads each
aura by slot id via the positional C_UnitAuras.UnitAuraBySlot. The by-index
getters re-walk the aura array from slot 0 on every call, so a per-button
loop over them was quadratic in the aura count; one enumeration plus a
by-slot read per aura is linear.
pfUI.api.ScanAuraSlots(unit, filter, buf, max) wraps GetAuraSlots' fill-a-
table form (table as the 5th argument) so no vararg Lua frame is involved:
Lua 5.0 allocates an `arg` table for every vararg call, which showed up as
nameplate OnUpdate/OnEvent memory growth in the first cut of this change.
Single by-index reads in tooltip and click handlers are unchanged (one call
each, and SetUnitAura takes the same index).
Requires the ClassicAPI build that adds GetAuraSlots' fill form; on an older
DLL the 5th argument is ignored and ScanAuraSlots would read the first slot
id as the count.
* Show Faction/Race icons in chat
* Load pfUI through ClassicAPI's flavor TOCs
ClassicAPI redirects the read of pfUI\pfUI.toc to a flavored file whenever
the DLL is installed, so which TOC the client opens already answers whether
ClassicAPI is there. Split the manifest three ways and let that do the work:
pfUI.toc fallback, reached only when ClassicAPI is missing;
loads API_Check.lua and nothing else
pfUI_ClassicAPI.toc full addon, every non-Turtle client
pfUI_Turtle.toc full addon plus init\turtle.xml, on Turtle
The fallback TOC declares no SavedVariables. It used to, while API_Check.lua
reset pfUI_profiles to an empty table on the disabled path -- which truncated
the player's profiles on logout.
With the missing-DLL case handled by TOC selection, API_Check.lua drops the
null object stub that kept the other ~140 files quiet, along with
pfUI_disabled and the now unreachable early return in pfUI.lua. It keeps the
version gate, which still matters: the flavor redirect landed in ClassicAPI
v1.11.0, below the v1.13.1 pfUI needs, so an old DLL still gets served a
flavor TOC. pfUI.lua also loses a verbatim duplicate of the whole check.
Turtle-only files move to init\turtle.xml: modules\turtle-wow.lua (its
TURTLE_WOW_VERSION guard is now redundant) and the lft, turtle_shop,
barbershop, transmog and ebc skins. turtle-wow registers last instead of
75th of 84; the only ordering it relies on is pfUI.chat, registered 8th.
pfSellData moves to env\selldata.lua, listed only in pfUI_ClassicAPI.toc,
since turtle-wow.lua replaces the table wholesale on Turtle. env\tables.lua
keeps an empty declaration so sellvalue.lua has something to index when the
turtle-wow module is disabled.
The release workflow pinned PFUI_CLASSIC_API_LATEST in pfUI.lua, which has
not held that constant since it moved to API_Check.lua, so the pin was
silently doing nothing. It also switches to brues-code/packager@vCAPI, which
recognizes the _ClassicAPI and _Turtle suffixes and applies the TOC build
type filters to them.
* Split the vendor price tables into their own manifests
Turtle's pfSellData moves out of modules\turtle-wow.lua into
env\selldata_turtle.lua, matching env\selldata.lua for the stock list, and
each is pulled in by the manifest for its client: init\stock.xml from
pfUI_ClassicAPI.toc, init\turtle.xml from pfUI_Turtle.toc. Either way it
loads after init\env.xml and replaces the empty pfSellData declared there.
Turtle's prices used to be assigned inside the turtle-wow module body, which
put them on pfUI.env and skipped them entirely when that module was
disabled. At file scope they land on _G and apply either way.
* Drop the vanilla compat layer
compat\vanilla.lua named the handful of things that differed between clients
back when pfUI targeted several. Only one client remains, so every constant
had exactly one value. Inline each at its use site and delete the file,
init\compat.xml, and both TOC entries.
COOLDOWN_FRAME_TYPE -> "Model"
LOOT_BUTTON_FRAME_TYPE -> "LootButton"
MINIMAP_TRACKING_FRAME -> _G.MiniMapTrackingFrame
FRIENDS_NAME_LOCATION -> "ButtonTextNameLocation"
EVENTS_MINIMAP_ZONE_UPDATE -> the event list, in panel.lua
MICRO_BUTTONS -> a local in panel.lua
NAMEPLATE_OBJECTORDER -> a local in nameplates.lua
ACTIONBAR_SECURE_TEMPLATE_BAR/_BUTTON -> nil, so the argument goes away
NAMEPLATE_FRAMETYPE and PLAYER_BUFF_START_ID had no readers left.
RunMacroText moves to pfUI.lua. compat\vanilla.lua was setfenv'd into the
pfUI environment, so the function only ever existed on pfUI.env; at file
scope it lands on _G as a real export instead. Nothing in pfUI calls it, and
ClassicAPI neither defines nor looks for a RunMacroText global -- it does
the same throwaway edit box natively in src/macro/Execute.cpp and only
defers to a global RunMacro.
* bump CAPI min to 11303
* auras: uncap the self-debuff tooltip lookup
With selfdebuff on, the displayed debuff list is PLAYER-filtered while
GameTooltip:SetUnitAura indexes the unfiltered HARMFUL list, so both
handlers map one to the other by matching name + sourceGUID. That mapping
scanned slots 1..16 only.
The unfiltered harmful list is not capped at 16. Once a unit's 16 debuff
slots are full the server parks further debuffs in buff slots, and
C_UnitAuras classifies by the aura's polarity flag rather than its slot
range, so it reports those as harmful too -- verified live at 18 harmful on
a 20-aura target. Past the sixteenth the lookup found nothing and fell
through to the raw filtered index, opening the wrong tooltip or none.
Both now enumerate however many harmful auras the unit actually has, via
ScanAuraSlots, which also drops the per-index rescan the by-index accessor
was doing. Each handler gets its own slot buffer: OnEnter can fire while a
refresh is showing/hiding frames under the cursor, so sharing the refresh
buffer could clobber a scan mid-walk.
The nameplate module still collects at most 16 debuffs per plate. That one
is a display cap matching its 16 configured icon frames, not an aura-count
assumption, so it is left alone.
* bump CAPI min to 11304
cache.player and cache.minion were filled from unitstr before the two guards
that validate it, so a stale identifier poisoned the cache -- and because the
fill is gated on `== nil`, the wrong answer was never recomputed. Players read
back as cache.player == false until the plate was hidden and shown again, which
only appeared to fix it because pool reuse tripped the name/guid wipe.
Two paths produce a stale unitstr: OnUpdate dispatches a targetUpdate to
OnDataChanged before it refreshes plate.istarget, and frameState.mouseoverGuid
is only updated on gaining mouseover, never on losing it.
Hoist the PLAYER_TARGET_CHANGED distrust and the UnitName mismatch check above
the cache fill so an unverified unitstr leaves the cache nil for the next tick
instead of locking in a wrong answer.
Swap the per-frame aura-refresh loops from table-allocating
GetAuraDataByIndex/GetBuffDataByIndex/GetDebuffDataByIndex/GetUnitAuras
to the zero-allocation positional C_UnitAuras.UnitAura/UnitBuff/UnitDebuff.
unitframes: buff/debuff icon loops, the dispellable-indicator 16-slot
scan, the two GetUnitAuras indicator passes, and the custom-debuff scan.
nameplates: the per-plate debuff scan now writes positional results
straight into the reusable display buffer, dropping the per-aura tables
and the result array.
Filters change from HARMFUL|PLAYER/HARMFUL to PLAYER/nil since UnitBuff/
UnitDebuff lock the range and still honor the PLAYER predicate. Cold
single-lookups (hover/click, tooltip, totem icon, libdebuff) keep the
readable table API.
This change updates several text setters to use `SetFormattedText` instead of passing `string.format(...)` into `SetText`. It also switches a few regex captures from `string.find` to `string.match` for cleaner capture handling, and replaces explicit width/height setters with `SetSize` in the first-run dialog for consistency.
The class-colour write sourced r, g, b from the health bar rather than
from PFUI_CLASS_COLORS[class], which the condition tested for but never
read. So "class colours on friendly names" only produced a class colour
when friendclassc - a bar option - happened to be on as well; otherwise
the name took the generic friendly colour. In every case it also
inherited the tapped-grey and barcombatstate overrides meant for the bar,
neither of which belongs on the name.
The name now reads the class table directly.
The discarded "and PFUI_CLASS_COLORS[class]" term was dead weight: the
table carries an __index metamethod returning a grey ColorMixin for any
missing key, so it was always truthy and merely allocated a throwaway
table per evaluation. "and class" is the real guard. One behaviour change
falls out - a friendly player of an unrecognised class now takes that
grey fallback rather than the bar colour, which is what the option means.
(cherry picked from commit f880dc3528db2ba0d0936b00a4dff8deec48e6d2)
(cherry picked from commit 111d70004d4fb172f168aa48a4cb6177a7f0d393)
cache.namecolor was written by two writers holding unrelated quantities:
the computed health-bar colour in OnDataChanged, which class-colours
friendly player names when friendclassnamec is on, and Blizzard's own
name FontString in the OnUpdate sync. Sharing one key let either suppress
the other, so the option silently lost - names came out white with
namefightcolor off, or the plain reaction colour with it on.
It also forced a full data pass every throttled tick, because the sync
writer sets update = true whenever it writes and the two kept flipping
the key. That defeats the half-second tick gate entirely.
Compounded by nameplate.cache surviving pool reuse: OnShow re-runs the
data pass but never clears the table, which is built once at plate
creation, so a recycled plate could inherit the previous unit's name
colour.
Ownership is now explicit through cache.ownname, each writer keeps its
own key, and both are invalidated when ownership flips.
(cherry picked from commit 0a4b3a7d6ae523cf70584adbb694a8740024b7cd)
(cherry picked from commit 0df9bb9fa907349182fb03217cab6aa1777aa7d9)
The level string was brightened by +0.3 on all three channels before
display, in the levelFromDB path and again in the OnUpdate colour sync.
A flat offset desaturates toward white and costs the high tiers most,
because their green/blue channels start near zero: verydifficult
1.00/0.50/0.25 becomes 1.00/0.80/0.55 and reads as yellow, impossible
1.00/0.10/0.10 becomes 1.00/0.40/0.40 and reads as orange. Orange and
yellow ended up 0.2 apart on a single channel, so a mob 3-4 levels above
the player showed yellow and a skull-range mob showed orange.
Both offsets dropped. Ownership of the colour is now explicit through
cache.levelfromdb so the two writers no longer race: the sync block
stands down on ?? plates, and the DB path clears cache.levelcolor so the
sync re-asserts cleanly once the level resolves. That also fixes a stale
cache - nameplate.cache is built once at plate creation and survives pool
reuse, so a recycled plate could keep the previous unit's colour.
Deleting the sync block instead would be wrong: it is the only thing that
colours the level on non-?? plates, and plate.level is created with no
colour at all.
Reported by Iden via Discord.
(cherry picked from commit eaa9beac51953d2a959c67d5ff5bc50273a919c3)
(cherry picked from commit 3fdb3b23136b260d111509e24e6d7b0d0b9b1b33)
- combopoints: only ever show on the target's plate, so re-issue Show/Hide
only when the target's combo count changes, and skip non-target plates once
cleared (they start hidden) -- instead of hiding all five every tick on every
plate. plate.cpShown tracks the shown count and stays in sync with the frames.
- player class color: merge the identical ENEMY_PLAYER / FRIENDLY_PLAYER
branches into one condition.
Use C_UnitAuras.GetBuffDataByIndex(guid, 1) instead of GetUnitAuras(guid)[1]:
a passive totem's provided buff is its single helpful aura, so the buff-
filtered single-slot fetch is both more precise and cheaper than building the
full aura list to take the first entry.
Replace the localized L["totems"] / L["critters"] name-substring matching
with ClassicAPI's UnitCreatureTypeID (CreatureType.dbc: 8 = Critter,
11 = Totem), cached per plate and reset on reuse. Locale-independent,
covers custom/Turtle units, rejects mobs merely named "...Totem...", and
cheaper than the per-plate name loop.
Totem icons now come straight from the game instead of a hand-picked table:
passive totems self-cast their provided buff, so their single aura's icon
is read via C_UnitAuras.GetUnitAuras; active totems (Searing/Magma/Fire Nova)
carry no self-aura, so their icon is captured from UNIT_SPELLCAST_SUCCEEDED
(C_Spell.GetSpellTexture) and cached. Detection and icons read through the
plate's ClassicAPI-provided GUID (cachedGuid).
Compare each plate's guild name against a cached myGuild value instead
of calling UnitIsInMyGuild for every plate. Refresh the cache on
PLAYER_GUILD_UPDATE and PLAYER_ENTERING_WORLD. This replaces a per-plate
C call with a Lua string compare.
Replace raw strsplit(",", ...) color parsing across action bars, chat,
nameplates, third-party skins, player, and roll with the cached
GetStringColor / GetStringColorObject helpers, dropping the per-build
string-table allocations they created.
* Migrate player castbar to ClassicAPI UNIT_SPELLCAST_* events
Drive the player cast path off ClassicAPI's synthesized, player-only
UNIT_SPELLCAST_* events instead of the vanilla SPELLCAST_* plus nampower
SPELL_{START,GO,DELAYED}_SELF mix. START/CHANNEL_START poll UnitCastingInfo/
UnitChannelInfo and stamp; STOP/CHANNEL_STOP/FAILED/INTERRUPTED clear;
SUCCEEDED drives tradeskill craft counting; DELAYED/CHANNEL_UPDATE re-poll.
Since UNIT_SPELLCAST_START fires for chained same-spell recasts, the nampower
SELF deps and their RunNextFrame co-hook-ordering workarounds are gone.
Remote target/focus bars keep SPELL_*_OTHER (the new events are player-only).
Fix the cast-start flash: StampBar now primes the fill on the stamp frame,
and ClearBar no longer snaps the bar to full (that snap lingered in the fade
tail and flashed when the next cast stamped).
Add a red flash on a cancelled cast: new appearance.castbar.failcolor,
flashed by ClearBar only when a cast was actually in progress.
Handle pushback Quartz-style: DELAYED/CHANNEL_UPDATE re-poll just the times
and accumulate the endMs shift into this.delay for the +/- indicator (cast
+X, channel -X), rather than a full restamp.
* Route target/focus castbars through remote UNIT_SPELLCAST_*
ClassicAPI now fires UNIT_SPELLCAST_* for remote tokens (target/focus/...)
via PollRemote, so drop the nampower SPELL_*_OTHER path and drive all three
bars off one event model. The gate is now token-based (arg1 == this.unitstr,
plus the arg1=="player" + UnitIsUnit case for target=self); the RunNextFrame
defer is gone since the remote poll fires after UnitCastingInfo is readable.
Cast detection is now nampower-free across every bar.
Use the event's rank payload (arg5): thread it through RefreshBar into
StampBar, which now only calls C_Spell.GetSpellSubtext as a fallback on the
retarget re-poll (PLAYER_TARGET/FOCUS_CHANGED) where no event is in hand.
* Drive nameplate castbars off UNIT_SPELLCAST_*
ClassicAPI now fires UNIT_SPELLCAST_* for nameplate tokens, so populate the
per-GUID castState cache from UNIT_SPELLCAST_{START,CHANNEL_START} (poll
PollCastInfo for the timing the payload omits) and clear it on
{STOP,CHANNEL_STOP} -- the remote poll fires STOP for natural end, interrupt,
and cancel alike. Replaces the nampower SPELL_{START,FAILED}_OTHER path; the
^nameplate arg1 gate ignores the target/focus/party fires of the same events,
and the NAME_PLATE_UNIT_ADDED seed still catches a plate spawning mid-cast.
Nameplate cast detection is now nampower-free.
* We no longer use these features from SuperWoW/Nampower
* Use secure unit attributes; remove mouseover scripts
Switch unitframes to attribute-driven clicks and remove mouseover tooltip handlers. Adds SetAttribute("unit") in UpdateVisibility and sets frame attributes (unit and type1='target') when creating frames. Removes OnEnter/OnLeave functions and their SetScript registrations and eliminates direct TargetUnit calls in ClickAction
* Use SetShown and simplify component default
Replace explicit Show/Hide conditionals in api/unitframes.lua with SetShown(...) for ressIcon, leaderIcon, lootIcon, pvpIcon, and restIcon to reduce branching and improve readability. Also remove the local shadowing of the parameter 'component' by using assignment (component = component or ""). No intended behavior changes.
* Drive unit-frame clicks via secure attributes
Replace the Lua OnClick dispatch (OnClick/ClickAction/RightClickAction) with
secure frame attributes: default type1=target / type2=menu, and EnableClickCast
now writes type/spell/macrotext/menu/target/focus attributes per button+modifier
instead of caching a clickactions table read at click time. Drops the now-dead
clickactions table and buttons list.
Also set a "unit" attribute on each frame (at creation and re-synced in
UpdateVisibility to the live token, so a party shown on the raid grid reports
partyN, not raidN) so the hovered unit resolves from the frame.
Match the target/focus/menu click keywords exactly rather than by prefix, so a
spell whose name starts with "focus"/"target"/"menu" (e.g. Focus Magic) casts
instead of being swallowed as the keyword action.
* Simplify marktracking colors and refresh per mark
Replace the hand-rolled ParseColor plus its default-color table with
GetStringColor -- the raidmarkercolor_* config keys already carry those same
defaults, so the fallback was dead code.
Restructure the refresh off observed mark tokens: UNIT_HEALTH/UNIT_MAXHEALTH
now arrive as arg1 == "markN", so refresh just that one row (UpdateRow) instead
of rescanning all eight on every nearby unit's health tick. A visibility flip
(into range / death / hp crossing 0) re-packs the rows, so UpdateRow hands off
to a full UpdateDisplay; RAID_TARGET_UPDATE / PLAYER_ENTERING_WORLD stay full
refreshes, with the 1s poll as the range-change backstop.
* not true anymore
* Poll marktracking fallback via C_Timer instead of OnUpdate
The 1s range-change safety net ran a per-frame OnUpdate that no-oped ~59 of
every 60 frames. Replace it with C_Timer.NewTicker(FALLBACK_INTERVAL,
UpdateDisplay) -- one wakeup per second off the shared timer driver -- and drop
the elapsed accumulator. The scanner frame is now purely event-driven.
* SuperWoW is now optional
* Update README feature list
Add 'Mouseover Unit Frames' and 'Click-casting' to the main features list. Update the SuperWoW entry to reflect that it tracks party/raid units on the minimap (replacing the prior SetMouseoverUnit note).
* Replace fixed-interval OnUpdate polls with C_Timer/RunNextFrame
Swap hand-rolled per-frame throttles for the modern timer primitives:
- turtle-wow: one-shot next-frame defer (self-hiding OnUpdate frame) -> RunNextFrame
- panel: clock, combat, and fps widgets -> NewTicker(1); guild roster -> NewTicker(60)
- minimap: coordinates text -> NewTicker(0.1)
- addonbuttons: one-shot init -> RunNextFrame; 5s button rescan -> NewTicker(5),
guarded by IsShown() to preserve the old "paused while hidden" behavior
Ticker callbacks reference frames via upvalue since 'this' is unbound outside
OnUpdate. Genuine per-frame work (bar fills, fades, drag) and polls already
coordinated through pfUI.throttle are left as-is.
* update pfQuest link
* Use SetShown/SetSize in panel module
Replace manual Show/Hide toggles with SetShown(not ... ) and replace SetWidth/SetHeight with SetSize in modules/panel.lua. Changes simplify toggle logic (timer, WorldMap, chat hide buttons) and unify sizing calls for timer, frames, textures, and microbutton. No behavior changes intended; purely refactor for conciseness and consistency.
* Bump ClassicAPI min version to 1.9.0
* Drive reagent counter from events instead of a polling OnUpdate
The pfReagentCounter frame ran an OnUpdate that (1) progressively rescanned
all 120 action slots one-per-100ms on any slot change and (2) recounted
reagent inventory on a 1s throttle. Replace both with direct event handling:
- ACTIONBAR_SLOT_CHANGED updates just the changed slot (arg1), full-scanning
only when arg1 is 0/nil, instead of restarting a ~12s rescan on every edit
- BAG_UPDATE_DELAYED recounts tracked reagents directly (it is already
Blizzard's coalesced bag event, so the extra 1s throttle was redundant)
- PLAYER_ENTERING_WORLD seeds the full reagent map once
UpdateSlot now seeds a new reagent's real count via GetItemCount so a freshly
placed reagent spell shows the correct number immediately. The updatecache ->
BarsUpdate render path and the IsReagentAction/GetReagentCount accessors are
unchanged.
hasTarget was always snapshotted alongside targetGuid (UnitExists("target")
next to UnitGUID("target")), so it was just targetGuid ~= nil. Drop the
field and derive it from targetGuid at both read sites; remove the now-unused
UnitExists perf-cache local.
frameState.hasTarget/targetGuid were polled via UnitExists("target") every
central-loop tick, while targetPlateGuid held the same guid but was already
event-driven -- redundant. Set frameState target fields in PLAYER_TARGET_CHANGED
(seeded on PLAYER_ENTERING_WORLD for a target held across reload), remove the
poll, and collapse targetPlateGuid into frameState.targetGuid (the castbar
frame now reads that). The central loop's only remaining per-tick poll is
GetTime().
Replaced numerous SetWidth/SetHeight calls with SetSize for consistency and brevity across UI code. Touched api/ui-widgets.lua, api/unitframes.lua and multiple modules (actionbar, addonbuttons, addons, afkcam, autovendor, bags, bgscore, buff, buffwatch, nameplates). Also simplified some sizing math in buff module. No functional behavior intended to change — code modernization only.
ClassicAPI polyfills UPDATE_MOUSEOVER_UNIT to fire on mouseover loss as well
as gain (src/unit/Mouseover.cpp), so the per-tick UnitGUID("mouseover") poll
in the central loop is no longer needed. Update frameState.mouseoverGuid on
the event and flag just the plates losing/gaining hover (via plateByGuid,
which holds overlays) for an immediate refresh.
- The central OnUpdate looped the entire plate pool (registry) every tick with
an IsVisible guard. Maintain a visiblePlates set via NAME_PLATE_UNIT_ADDED/
_REMOVED and iterate that instead, so hidden pool slots aren't touched. The
IsVisible guard stays as a safety net.
- Fix the global refresh propagation: it set .eventcache on the base frame,
but OnUpdate reads/clears it on the overlay (nameplate) -- so the propagation
was a no-op and global refreshes (e.g. target-change alpha/strata) fell
through to the 0.5s catch-all timer. Set plate.nameplate.eventcache and scope
to visiblePlates, so those refreshes land on the next tick as intended.
OnConfigChange still iterates the full registry (hidden plates must pick up
config changes before they next show).
Both mouseover checks relied on original.glow:IsShown() (a fragile proxy for
the hovered plate, on a Blizzard texture pfUI hides/restyles) gated by
UnitExists("mouseover"). Cache the engine mouseover unit's GUID once per
central-loop tick (frameState.mouseoverGuid) and have each plate compare its
cachedGuid against it.
More correct for the data path: the "mouseover" unit token is only chosen
for the plate whose GUID actually matches the engine's mouseover unit, so
overlapping plates can't double-match and paint the wrong unit's data. Also
cheaper -- one UnitGUID("mouseover") per tick plus a per-plate table compare,
versus the old per-plate IsShown gated on a per-tick UnitExists.
The unthrottled target castbar frame called C_NamePlate.GetNamePlateForUnit
every frame. Stash the target's GUID on PLAYER_TARGET_CHANGED (seeded on
PLAYER_ENTERING_WORLD for a target held across reload) and resolve the plate
through plateByGuid instead -- a table lookup plus a real idle skip when
there's no target. The plate is looked up per frame (not cached) so a plate
spawning/despawning while the unit stays targeted still resolves correctly.
Previously every visible plate polled C_Spell.UnitCastingInfo each throttled
tick just to detect casts. Now cast state is event-driven:
- SPELL_START_OTHER (nampower) stamps a per-GUID castState cache (spellId,
timing, channel flag from spellType); SPELL_FAILED_OTHER clears it; normal
completion expires at endTime. GetCastInfo just reads the cache, so all
call sites (castbar update, non-target detection, casting name-color) stop
polling.
- plateByGuid routes events to plates in O(1) and bounds the cache to on-screen
casters; NAME_PLATE_UNIT_ADDED seeds an already-casting unit with one poll.
- The vestigial nameplate.castUpdate flag now fires on cast start to bypass the
throttle for an immediate bar.
With the poll gone, per-tick work is a cache read + SetValue, so the dedicated
target castbar frame now runs unthrottled (every frame) for the smoothest
sweep, and the non-target nameplates_castbar default rises 50 -> 100 FPS.
Known limits: other-unit cast pushback isn't reported by nampower, and a
mob-cancelled channel with no fail event lingers until endTime.
BigWigs (SpellRequests) redefines the global string:split to return a table,
and another addon clobbers the strsplit global too. pfUI.api.strsplit and the
bare strsplit callers inherited the broken versions depending on load order,
producing 'attempt to compare number with nil' from GetStringColor.
- Make pfUI.api.strsplit fully self-contained (no delegation to global
strsplit / string.split), so no override can reach it.
- Optimize the hot path: single-char delimiters use plain-text find (no
pattern compilation, no per-call char-class string); localize string.find
and string.sub.
- Route the per-frame castbar/nameplate color splits through GetStringColor,
which caches, instead of re-splitting a constant string every update.
commit 70b1e66c2d
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sun Jul 26 18:30:18 2026 -0500
Use SetSize/SetShown and simplify minimap/map
commit 7b6bfe9975
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 11:27:14 2026 -0500
Include raidpet frames in /pftest test mode
The unit-frame test toggle (showall) only previews frames that exist and
are positioned. Raidpet frames are created when raidpet.visible == "1",
but LayoutPets' collapse mode only positions pets whose raidpet<N> unit
actually exists, so solo/in test mode they stayed hidden.
Add a showall branch to LayoutPets that mirror-lays every pet cell and
shows it (bypassing collapse and the roster gate), still guarded by the
existing showpets check so nothing happens when raidpet is disabled. Call
LayoutPets from the /pftest handler so the grid updates immediately on
toggle-on and restores to the normal layout on toggle-off.
commit 4c65c38647
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 10:34:33 2026 -0500
Remove unused RangeCache local in UnitInRange
commit 6c96bbbe6c
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 10:34:33 2026 -0500
Use Clamp() for the two-sided clamps in ui-widgets
Replace the hand-rolled math.min/math.max and if/elseif clamp idioms in the
status bar and scroll frame with Clamp(). Equivalent for every value seen
(min <= max always holds), just clearer.
commit 830e0a0be1
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 10:34:33 2026 -0500
Drop vanilla GetItemInfo shim for C_Item.GetItemInfo
The compat/vanilla.lua GetItemInfo override reshaped vanilla's tuple into
retail order by inserting nil at slot 4 (itemLevel), but truncated at 8
fields, dropping itemEquipLoc/itemTexture. ClassicAPI's global GetItemInfo
keeps the vanilla shape (its hook only warms the item cache), so the shim
stayed necessary but incomplete.
Remove it and point the three callers at ClassicAPI's C_Item.GetItemInfo,
which returns the full 18-field retail tuple. Positions 1-8 are identical
to the old shim output, so bags/roll are unchanged; character.lua's
itemSlot (equipLoc, field 9) was always nil under the truncating shim and
now resolves correctly for ShaguScore.
commit c0da63657d
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 10:09:34 2026 -0500
Various cleanup
Removes version compatibility checks for TBC/Vanilla, consolidating code to target a single WoW version. Refactors repetitive SetWidth/SetHeight calls to SetSize for cleaner code. Removes duplicate function definitions from unitxp module and simplifies conditional show/hide operations using SetShown.
commit 11a6302c2d
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 09:53:37 2026 -0500
Delegate pfUI.api.strsplit to ClassicAPI's strsplit
Replace the Lua pattern-based implementation with a thin wrapper around
ClassicAPI's C-level strsplit. Keeps the pfUI.api.strsplit entry point for
backwards compatibility with addons that call it, while dropping the
redundant reimplementation.
Behavioral note: the old version used ([^delim]+) which silently collapsed
empty fields; delegating to real strsplit now preserves them
("a,,b" -> "a", "", "b"). All in-repo callers split numeric color tuples,
version strings, or build name-keyed lookup tables, none of which are
affected by empty-field preservation.
commit 67c126eac8
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 01:01:02 2026 -0500
Bump ClassicAPI minimum version to 1.8.0
commit a6cf157518
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 00:49:50 2026 -0500
read spell rank via C_Spell.GetSpellSubtext
The rank string comes from the spell subtext ("Rank N"), which ClassicAPI's
C_Spell.GetSpellSubtext returns directly -- so drop nampower's
GetSpellRecField(spellId, "rank") here. The existing "Rank " gsub parse is
unchanged.
commit 4b1ba99b4f
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 00:33:20 2026 -0500
Move player info overlay onto ClassicAPI (drop Nampower)
The haste/spell-power overlay read both values through Nampower and disabled
itself entirely without it. Move both to ClassicAPI so it works on any
ClassicAPI client:
- Spell power: GetSpellPower("net") (nampower) -> GetSpellBonusDamage(i)
(ClassicAPI, same per-school field). Merged the build+scan into one loop;
the default school seeds the tiebreak so equal schools still favor it.
- Haste: GetUnitField("player", "modCastSpeed") -> UnitSpellHaste("player"),
which returns the haste percentage directly (the (1/modCastSpeed-1)*100
conversion is now baked in, off the same UNIT_MOD_CAST_SPEED field, signed).
- Dropped the "if not GetUnitField then return" gate -- UpdateInfoText no
longer touches Nampower, so the overlay runs everywhere.
commit 7df4aa6d50
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sat Jul 25 00:32:44 2026 -0500
Hook the real global _G.UnitHealth for feign death
Inside a RegisterModule body, `function UnitHealth(...)` defines UnitHealth on
the pfUI environment, not the real global -- so the feign-death real-HP fix
only reached callers that resolve UnitHealth through pfUI's env, and missed
_G consumers (Blizzard frames, other addons). Hook _G.UnitHealth explicitly
(and capture oldUnitHealth from _G) so the un-gate applies everywhere.
commit b0bf2fd869
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Fri Jul 24 21:10:58 2026 -0500
Refactor nampower module
Simplify reactive spell storage by using spell IDs instead of texture/name tables. Update to use modern C_Spell APIs (GetSpellTexture, GetSpellName) instead of manual texture paths. Consolidate SetWidth/SetHeight calls to SetSize and use print() instead of DEFAULT_CHAT_FRAME:AddMessage().
commit 0b06961333
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Fri Jul 24 20:54:23 2026 -0500
Use GetNamePlateForUnit for target lookups; drop dead ScanGuid block
- Replace the three GetNamePlateForGUID(UnitGUID("target")) round-trips with
GetNamePlateForUnit("target"), which resolves the token to a GUID internally
-- no UnitGUID string detour, and it returns nil for no-target/no-plate so
the UnitExists guards collapse. GetNamePlateForGUID is left for raw-GUID cases.
- Remove the dead libunitscan.ScanGuid nameplate block (ScanGuid was deleted
from libunitscan long ago, so the guard was never true) -- it carried the
last GetUnitField("npcFlags") read.
commit 3e6210086e
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Fri Jul 24 20:54:23 2026 -0500
Drop Nampower stats system and polling from unit frames
GetUnitStats now reads health/power straight from UnitHealth/UnitPower
(the descriptor fields the server broadcasts) -- for a real unit token
GetUnitField read nothing different, so the whole Nampower-vs-fallback
apparatus was measuring a distinction that no longer exists. Removed:
- The stats system: pfUI.uf.stats, pfUIStatsFrame + UpdateStatsDisplay, the
lastUnitStats change-cache, the /pfuistats slash command, and every counter
increment (event/heartbeat/earlyReturn/throttledSkip/nampower/fallback).
- All GetUnitField health reads (GetUnitStats, heal-prediction, health-gradient
color) -- collapsed to the UnitHealth/UnitHealthMax they already fell back to.
- The heartbeat-polling backstop and its fallbackThrottle/lastEventUpdate deps.
Frames now refresh on events only; eventless frames still use their own tick,
and range/aggro still run in the eventless-actions block.
commit 63001b7b0c
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Fri Jul 24 20:09:42 2026 -0500
Move nameplates onto ClassicAPI stable nameplate tokens
ClassicAPI now assigns retail-exact, per-plate-stable "nameplateN" tokens and
fires vanilla UNIT_* events for them, so the nameplate module can key off the
token instead of Nampower/SuperWoW GUID primitives:
- Events: UNIT_AURA_GUID/UNIT_FLAGS_GUID -> UNIT_AURA/UNIT_FLAGS, matched by
the "nameplateN" token (guarded on the token prefix, since these also fire
for target/party/raid). Registered unconditionally -- no GetUnitField gate.
- Health: GetUnitField(guid, "health"/"maxHealth") -> UnitHealth/UnitHealthMax
on the plate's cached token. Same UNIT_FIELD_HEALTH the server broadcasts
(real HP on Turtle; the ~= 100 scaled-vs-real guard is unchanged).
- Combat/target in GetCombatStateColor: GetUnitField("flags") + HasFlag ->
UnitAffectingCombat(token); GetUnitField("target") + SuperWoW "<guid>target"
-> the "nameplateNtarget" chain. Dropped the now-dead HasFlag,
UNIT_FLAG_IN_COMBAT, and guidTargetTokenCache.
- GetCastInfo now takes the token directly; every caller already holds it, so
the UnitTokenFromGUID(guid) round-trip is gone (unused module-wide now).
NAME_PLATE_UNIT_ADDED caches the token as plate.nameplate.unit alongside the
GUID; REMOVED clears both. GUID is retained only as the stable cache key
(debuffCache/threatMemory/combatColorCache). Only GetUnitField("npcFlags")
remains -- it has no token/vanilla equivalent.
commit ce1c49fcbb
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Fri Jul 24 18:20:06 2026 -0500
Make focus/focustarget event-driven via ClassicAPI unit events
ClassicAPI now fires UNIT_* (health/mana/aura/...) with arg1 == "focus" and
arg1 == "focustarget", observed per-unit like target/party/raid. Both frames
already registered those events (focus/focustarget are in pfValidUnits) and
their OnEvent already matches arg1 == label, so the 0.2s polling ticks were
pure workarounds for the missing events. Drop both ticks; the frames now
refresh on-event like target, with range/glow still on the shared 0.5s state
pass and PLAYER_FOCUS_CHANGED still driving assign/clear.
- Replace all UnitMana/UnitManaMax calls with UnitPower/UnitPowerMax. The
no-arg form returns the unit's primary power from the same field vanilla's
UnitMana read, so these are behavior-preserving. Drops two dead
UnitMana/UnitManaMax local caches in nameplates.lua.
- Replace magic power-type numbers (0/1/2/3) with Enum.PowerType.Mana/Rage/
Focus/Energy in the GetUnitStats branches, the power-bar color block, the
druid mana bar, GetStatusValue's powerdyn, and energytick's mode checks.
commit d63057474083aa67fda2aa28ee6301334ffc4fdf
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Mon Jul 6 00:59:17 2026 -0500
Use C_Map.GetMapOverlays instead of hardcoded pfMapOverlayData
ClassicAPI's C_Map.GetMapOverlays reads WorldMapOverlay.dbc directly and
returns the full overlay list for a zone (explored + unexplored) — the
data vanilla's GetMapOverlayInfo withholds. mapreveal now iterates it
straight (named fields: textureName/texturePath/width/height/offsets),
dropping unpack_hash and the pfMapOverlayData tables entirely.
Also fixes the explored-check: it compared the full texture path against
GetMapOverlayInfo's bare-name keys, so the magnifying glass never
suppressed on explored overlays. Now matches on the bare name.
Removes ~870 lines of hand-measured overlay data (base + Turtle).
commit ad12c8209a80cd1211fe08e83f7013d24b5efd2d
Author: Brues <5278969+brues-code@users.noreply.github.com>
Date: Sun Jul 5 23:45:09 2026 -0500
utilize HookScript from ClassicAPI
ClassicAPI's remote-cast cache is stamped from SMSG_SPELL_START and only
expires by computed end time — 1.12 keeps no per-unit interrupt record,
so an interrupted cast kept animating on the plate until its would-be
finish (BG flag caps being the loudest repro).
Nampower does surface the missing signal in Lua: SPELL_FAILED_OTHER
(casterGuid, spellId; fired from the SMSG_SPELL_FAILED_OTHER handler)
and UNIT_DIED (guid). Stamp a guid-keyed suppression time on either
event and have GetCastInfo drop any cast that started before the stamp;
a newer cast clears its unit's entry. Both castbar paths (dedicated
target frame + central loop) already funnel through GetCastInfo, so one
check covers them. The handler only stamps when the unit actually has a
tracked cast, and flags the plate via castUpdate for a same-tick hide.
libhealth ships with pfUI and Nampower's GetUnitField (hard-dep) covers
the real-HP read; the MobHealth integration was a chained last-resort
fallback that could never actually win, and on the nameplate path it
also mixed percentage and real-value scales into the bar's SetMinMaxValues
since it overwrote hp/hpmax without hpmin.
GUID is the unique identity; the name check was leftover from the
name-only era and became dead weight once cachedGuid (NAME_PLATE_UNIT_
ADDED) was added alongside it.
The plate-reuse reset only cleared name/guid/player/cdCache, leaving
hp/hpmax/rgb/namecolor/levelcolor/target/mouseover/inCombat behind. The
PERF gates downstream ("only update X when X changed") then skipped bar
fill / color / text updates when the new occupant happened to share a
cached value with the previous one — most easily reproduced on
plate-pool reuse in starting zones (mobs churn fast, percentage values
collide). Nuke the whole cache via table.wipe.
Refs #8.
ClassicAPI's recent C_Spell additions cover remote-unit casts natively
(SMSG_SPELL_START co-hook caching per caster GUID), so the two parallel
cast trackers pfUI was running — libcast.lua and the libdebuff_casts
table inside libdebuff.lua — can both retire.
Migrations:
- modules/castbar.lua: focus/player cast-info gathering reads
C_Spell.UnitCastingInfo / UnitChannelInfo directly. Fallback ladder
(libdebuff_casts → pfGetCastInfo → pfGetChannelInfo) collapses into a
single call. Pushback handlers stop writing back into a non-existent
cache; the local this.endTime is the source of truth.
- modules/nameplates.lua: GetCastInfo(guid) now resolves to a unit token
via UnitTokenFromGUID and queries C_Spell, returning the same compact
struct shape downstream code expected. UpdateCastbar collapses from a
three-branch hierarchy (dead IterDebuffs / libdebuff_casts / libcast)
to one C_Spell read.
- modules/afkcam.lua: pfGetCastInfo+pfGetChannelInfo round-trip becomes
a single C_Spell.UnitCastingInfo("player") or UnitChannelInfo fallback.
- libs/libpredict.lua: HealComm timing uses C_Spell on the sender's unit
token after a small group-roster walk to resolve the sender's name.
Deletions:
- libs/libcast.lua entirely (-571 lines) plus its init/libs.xml entry.
- libdebuff_casts / libdebuff_item_icons tables and their write sites
in libs/libdebuff.lua (the SPELL_START_*, SPELL_GO_*, SPELL_FAILED_*
event handlers stop maintaining them but keep firing the
libdebuff_*_hooks broadcast surface for actionbar / swingtimer /
libtotem). SPELLCAST_CHANNEL_STOP now reads the active channel from
C_Spell.ChannelInfo.
- modules/superwow.lua's supercast block — UNIT_CASTEVENT writes into
libcast.db are redundant now that C_Spell co-hooks the same packet.
- The cast-bar item-icon override that swapped in a potion/trinket
icon for item-triggered casts. Spell icon stays; the item-icon
metadata path (libdebuff_item_icons) went with libdebuff_casts.
Steady Shot synthetic cast bar — Turtle WoW-specific:
- castbar.lua gains a pfUI.synthetic_casts[unit] fallback that fires
only when C_Spell returns nil, so abilities the engine treats as
instant but which have a meaningful wait window can still render a
cast bar.
- modules/turtle-wow.lua replaces the old libcast.customcast block with
a Nampower SPELL_QUEUE_EVENT subscriber. ON_SWING_QUEUED matching
the localized "Steady Shot" name writes a 1.4s synthetic entry;
ON_SWING_QUEUE_POPPED clears it; castbar's endMs guard self-expires
the entry as a safety net. Note: haste scaling (libcast.ApplyShotHaste)
is gone — bar may finish slightly early under +ranged haste buffs.
Net: 152 insertions, 972 deletions.
With every external caller of libdebuff:UnitDebuff / :UnitOwnDebuff
now on C_UnitAuras, the two public per-aura readers and the
nameplate-side cache they were feeding have no consumers.
- libs/libdebuff.lua: removes libdebuff:UnitDebuff (~120 lines),
libdebuff:UnitOwnDebuff (~75 lines), the _ownDebuffSortFunc helper,
and the local cache table. The slotOwnership / ownDebuffs /
allAuraCasts / pendingCasts bookkeeping stays — GetBestAuraCast
(libpredict) and GetEnhancedDebuffs (CleveRoids) still read it, and
the event handlers maintain it. GetSlotCaster / GetDebuffSlotMap
stay too; the DEBUFF_ADDED_OTHER handler and the debug printer use
them. File goes 2010 → 1870 lines.
- modules/nameplates.lua: deletes PlateCacheDebuffs (was already
rewritten on C_UnitAuras and unused once the display loop bypassed
the cache), PlateUnitDebuff, the cachedVerify scaffolding, and the
nameplate.UnitDebuff / nameplate.CacheDebuffs registrations.
- api/config.lua + modules/gui.lua: drops the now-defunct
"guessdebuffs" knob — its only effect was gating the dead cache.
The per-unit cache reset only fired when the plate's name changed, so
pool reuse between same-named units (player "Ironforge Guard" → NPC
"Ironforge Guard") carried the previous unit's cache.player into the
next OnDataChanged call. GetUnitInfo then returned the wrong table's
class, unittype flipped to FRIENDLY_PLAYER, and the bar drew in MAGE
class color. Gate the reset on cachedGuid changes too.
Stores cache.player as true/false/nil instead of "PLAYER"/"NPC" so it
flows straight into GetUnitInfo as the isPlayer hint — both in
OnDataChanged and the wait_for_scan retry, which previously dropped
the hint and ping-ponged the flag every frame.
Hoists the name-change reset above the GetUnitInfo lookup and folds the
two UnitIsPlayer call sites into one. The earlier shape resolved
cache.player upfront from cachedGuid, then redundantly re-resolved from
unitstr after the name-change reset — and the upfront block could read
stale cache.player on plate reassignment, passing a wrong isPlayer hint
to GetUnitInfo.
GetUnitInfo() probed the players table first, so an NPC sharing a name
with a known player (e.g. Chromie) inherited the player's class, level,
and guild. Adds an isPlayer hint to GetUnitInfo and resolves each
plate's actual unit type from its cached GUID via UnitIsPlayer before
the lookup. The plate's cache.player gates further work.
Routes player and nameplate castbar progress through
PixelUtil.SetStatusBarValue so a 1.4s cast on a narrow bar no longer
shimmers across fractional pixels as it ticks. Adds GetMinMaxValues /
GetValue handlers to pfUI's custom StatusBar so PixelUtil can read the
current range.
Six allocation sites firing per visible plate per central tick (or per
throttled OnDataChanged tick) were generating heavy GC pressure visible
in pfDebug's analyzer:
- IterDebuffs callback in PlateCacheDebuffs was a fresh closure per call.
Hoisted to a module-level function with state in _pcdSelf / _pcdNow /
_pcdId, matching the pattern the icon-grid IterDebuffs already used.
- string.format("%.2f", remaining) in UpdateCastbar fired every frame per
active castbar. Factored into SetCastbarText helper that compares a
rounded integer (centi- or deci-seconds) and only formats + SetTexts
when the displayed value would actually change. lastTextTick reset
alongside lastEndTime so back-to-back casts don't collide.
- guid.."target" concat in GetCombatStateColor hit Lua's string intern
pool every call. Cached in guidTargetTokenCache, evicted alongside
the other per-guid caches in NAME_PLATE_UNIT_REMOVED.
- plate.name:SetText was unconditional; moved inside the existing
plate.cache.name change gate so it only fires on real unit changes.
- plate.level:SetText (string.format) was unconditional; gated behind a
new (level, elite) cache so the format only runs on level-up or
elite-state flips.
- C_NamePlate.GetNamePlates() poll every 0.5s allocated a fresh Lua
table just to read its length. Replaced with a visiblePlateCount
counter maintained event-driven from NAME_PLATE_UNIT_ADDED/_REMOVED.
- Switch the four GUID reads in OnDataChanged from plate.parent:GetName(1)
(SuperWoW idiom) to plate.cachedGuid (set by NAME_PLATE_UNIT_ADDED via
UnitGUID(token)). Move the initial OnDataChanged call out of
OnConfigChange's CREATE path so it runs after UNIT_ADDED has populated
cachedGuid; re-add it explicitly in the user-config-change loop.
- Reject GetUnitField's health/maxHealth when maxHealth == 100 — the
engine writes (hp_percent, 100) into UnitFields for non-detailed units
(UPDATE_PARTIAL packets carry percent only). Without this guard,
Nampower's raw field read returns the percent and the nameplate displays
"5 / 100" as if it were real HP. Mirrors libhealth's heuristic so we
fall through to its estimator instead.
- New config: nametextpos (LEFT/CENTER/RIGHT, defaults to CENTER).
Decouple the bar's anchor from the name so the name's JustifyH can
shift left/right without dragging the bar with it.
* nameplates: drive discovery + GUID lookup via C_NamePlate
ClassicAPI now provides the modern C_NamePlate API and the matching
NAME_PLATE_CREATED / NAME_PLATE_UNIT_ADDED / NAME_PLATE_UNIT_REMOVED
events. Replace the WorldFrame:GetChildren polling + manual GUID
dictionary with event-driven plumbing.
- Drop the per-tick WorldFrame:GetNumChildren / GetChildren scan
- Delete IsNamePlate() border-texture filter (events only deliver
nameplate frames)
- Delete the guidRegistry table + its OnShow/OnUpdate maintenance;
every lookup now goes through C_NamePlate.GetNamePlateForGUID
- Move per-GUID cache cleanup (debuffCache / threatMemory /
combatColorCache / libdebuff_casts) from the OnUpdate visibility
scan to a NAME_PLATE_UNIT_REMOVED handler — fires exactly once
per unit going out of range
- NAME_PLATE_CREATED drives nameplates.OnCreate; NAME_PLATE_UNIT_ADDED
sets cachedGuid and drives nameplates.OnShow (so OnCreate no longer
calls OnShow directly and no longer HookScripts the parent)
- visiblePlateCount uses table.getn(C_NamePlate.GetNamePlates())
Net -64 lines, with a meaningful reduction in per-tick CPU (no more
WorldFrame child iteration + region/texture filter every 50ms).
* nameplates: switch UNIT_ADDED/REMOVED handlers to nameplateN tokens
ClassicAPI's NAME_PLATE_UNIT_ADDED / _REMOVED events now ship the
"nameplateN" unit token as arg1 (matching modern WoW) instead of the
GUID string.
- UNIT_ADDED: resolve plate via C_NamePlate.GetNamePlateForUnit(arg1);
derive the cache-key GUID via UnitGUID(arg1)
- UNIT_REMOVED: same lookup pattern; compute UnitGUID(arg1) inside the
handler before the slot recycles (per docs guarantee)
The other six GetNamePlateForGUID call sites (libdebuff aura callbacks,
UNIT_FLAGS_GUID, PLAYER_TARGET_CHANGED, combo-point handler, castbar
frame) stay on the GUID flavor — they receive real GUIDs from other
event paths.
---------
Co-authored-by: Brues <5278969+brues-code@users.noreply.github.com>
Drop the now-vestigial expansion plumbing.
- Delete modules/thirdparty-tbc.lua + its xml Include
- Strip 10 tbc-tagged CreateConfig calls in modules/gui.lua
- Drop the expansion arg from CreateConfig() signature + the disabled-
entry rendering path that depended on it
- Drop the showdisabled GUI toggle + its default
- Simplify pfUI:RegisterModule / pfUI:RegisterSkin to (name, func) only
- Strip the leading version arg ("vanilla:tbc", etc.) from all 114
Register call sites
- Delete the pfUI.expansion variable
Brues