if C_AddOns.DoesAddOnExist('pfUI') then EventUtil.ContinueOnAddOnLoaded('pfUI', function() -- A pfUI that sets `handlesHookScript` was written against our -- additions and manages them itself, so none of the shims below apply -- to it. if pfUI.handlesHookScript then return end -- pfUI\pfUI.lua redeclares the RAID_CLASS_COLORS object at least 14 times (insane!!) -- This stops it from replacing all our ColorMixin values pfUI.UpdateColors = function() end RAID_CLASS_COLORS = setmetatable(RAID_CLASS_COLORS, { __index = function() local unknownColor = CreateColor(0.6, 0.6, 0.6) unknownColor.colorStr = unknownColor:GenerateHexColor() return unknownColor end}) -- Compatibility with OLDER forks of pfUI. Our DLL adds the modern -- HookScript widget method; older pfUI forks branch on -- `if button.HookScript` and take a modern code path they were never -- written for on their pfActionBar buttons, which breaks them. While -- the actionbar module builds those buttons, temporarily wrap -- CreateFrame to shadow HookScript with a falsy field so those forks -- fall back to their vanilla SetScript path. The wrapper is removed as -- soon as the actionbar module finishes loading so it affects nothing -- else. The maintained "brues" fork, identified by its X-Website tag, -- handles HookScript on its own buttons and is exempt from this part; -- every other fork gets the shim. local website = GetAddOnMetadata("pfUI", "X-Website") local hasBruesWebsite = website and strfind(website, 'brues') if not hasBruesWebsite then local _createFrame = CreateFrame local HookedCreateFrame = function(frameType, name, parent, template) local frame = _createFrame(frameType, name, parent, template) if frameType == "Button" and strfind(name or "", "pfActionBar") then frame.HookScript = false end return frame end hooksecurefunc(pfUI, 'LoadModule', function(frame, moduleName) if moduleName == "updatenotify" then CreateFrame = HookedCreateFrame elseif moduleName == "actionbar" then CreateFrame = _createFrame end end) end end) end -- ShaguTweaks libpredict register's TBC events that ClassicAPI backports. if C_AddOns.DoesAddOnExist("ShaguTweaks") then EventUtil.ContinueOnAddOnLoaded("ShaguTweaks", function() if not ShaguTweaks or (ShaguTweaks.API and ShaguTweaks.API.classicapi_version) then return end local libp = ShaguTweaks.libpredict if libp then libp.sender:UnregisterEvent("COMBAT_LOG_EVENT_UNFILTERED") libp.sender:UnregisterEvent("UNIT_SPELLCAST_START") libp.sender:UnregisterEvent("UNIT_SPELLCAST_STOP") libp.sender:UnregisterEvent("UNIT_SPELLCAST_FAILED") libp.sender:UnregisterEvent("UNIT_SPELLCAST_INTERRUPTED") libp.sender:UnregisterEvent("UNIT_SPELLCAST_SENT") end end) end -- Guard UPDATE_MOUSEOVER_UNIT handlers (issue #12). function CAPI_MouseoverClearedCompat(frame) if not frame then return end local originalOnEvent = frame:GetScript("OnEvent") frame:SetScript("OnEvent", function() if event == "UPDATE_MOUSEOVER_UNIT" and not UnitExists("mouseover") then return end if originalOnEvent then originalOnEvent() end end) end CAPI_MouseoverClearedCompat(GameTooltip) -- Restores vanilla calling for ONE frame's script handler. Our DLL hands script -- handlers modern positional arguments (`self`, then the event's own args) on -- top of the `this` / `arg1` globals vanilla uses. That is additive for a -- handler declared `function()`, but a handler declared with a parameter of its -- own -- one written to be BOTH called directly and used as a script -- now -- receives `self` in that parameter instead of the nil it was written for. -- -- The wrapper below takes no parameters, so those extra arguments stop here, -- and the original runs with the empty argument list vanilla gave it. The -- globals are untouched by the nesting, so the handler still reads `this` and -- `event` normally. function CAPI_VanillaScriptArgsCompat(frame, script) if not frame then return end local original = frame:GetScript(script) if not original then return end frame:SetScript(script, function() original() end) end -- LunaUnitFrames sets `SetupGroupHeader(groupType)` as its group headers' -- OnEvent while also calling it directly with a group name. As a handler it -- expects `groupType` to be nil so it falls back to `this.unitGroup`. Given -- `self` instead, it looks the header up by frame rather than by name, finds -- nothing, and indexes nil (units.lua:358, issue #41). Its raid counterpart -- takes the header frame itself, so `self` is already what it wants -- only the -- group headers need this. if C_AddOns.DoesAddOnExist("LunaUnitFrames") then EventUtil.ContinueOnAddOnLoaded("LunaUnitFrames", function() if not (LunaUF and LunaUF.Units) then return end local function FixGroupHeader(unit) CAPI_VanillaScriptArgsCompat(getglobal("LUFHeader" .. (unit or "")), "OnEvent") end -- Headers are built on demand, so cover the ones already up and every -- one created later. FixGroupHeader("party") FixGroupHeader("partytarget") FixGroupHeader("partypet") hooksecurefunc(LunaUF.Units, "LoadGroupHeader", function(self, unit) FixGroupHeader(unit) end) end) end if C_AddOns.DoesAddOnExist("Puppeteer") then EventUtil.ContinueOnAddOnLoaded("Puppeteer", function() CAPI_MouseoverClearedCompat(PTEnemyUpdater) end) end -- Cartographer_Notes gates a WorldMap release-spirit hook behind an -- `if lua51` probe, treating a 5.1-capable Lua as a modern client that has -- the WorldMapDeathRelease frame. Our 5.1 syntax backport makes the probe -- pass, so Cartographer takes the modern branch and hooks a frame 1.12's -- WorldMap never had (WorldMapDeathRelease:SetScript in OnEnable) -> nil -- index. Provide an inert, hidden stub so the hook succeeds and does nothing -- -- the same outcome vanilla had, where the branch never ran. The embedded -- !!!ClassicAPI addon loads first, so the frame exists before Cartographer's -- OnEnable. if C_AddOns.DoesAddOnExist("Cartographer") and not WorldMapDeathRelease then CreateFrame("Button", "WorldMapDeathRelease", WorldMapButton or UIParent):Hide() end -- ModernSpellBook (Spellbook\MSB_Spellbook.lua) re-hooks its "show all ranks" -- checkbox by calling HookScript as a GLOBAL -- HookScript(frame, "OnClick", -- handler). That global never existed on any client: HookScript has always -- been the widget method frame:HookScript, which our DLL provides. MSB even -- guards the call with `if checkbox.HookScript` (the method IS present), then -- calls the global (nil) -> "attempt to call global 'HookScript'". Define the -- global so the call resolves, forwarding to the method. Gated on the addon -- being present so the global namespace stays clean otherwise. if C_AddOns.DoesAddOnExist("ModernSpellBook") and not HookScript then function HookScript(frame, script, handler) if frame and frame.HookScript then frame:HookScript(script, handler) end end end -- Compost-2.0 (rev 17406+, bundled by BigWigs among others) replaces most of -- its own body with stubs once it believes it is on Lua 5.1, on the reasoning -- that table recycling is pointless when the collector already does it: -- -- local lua51 = loadstring("return function(...) return ... end") and true or false -- ... -- if lua51 then function lib:Erase() return {} end -- else function lib:Erase(t) --[[ clears t in place ]] end -- -- Our 5.1 syntax backport makes that probe pass, so the stubs win. For -- GetTable and Reclaim that is harmless -- both are pure recycling hints. Erase -- is not: callers hand it a live table and expect its keys gone. The stub takes -- no parameter at all, so every `compost:Erase(t)` that clears in place (rather -- than assigning the returned table) silently stops clearing. -- -- Puppeteer is the reported casualty. PTUnit:ClearAuras erases its -- TrackedDebuffTypes, AfflictedDebuffTypes and aura-ID sets in place, so a -- dispel type never leaves the set once applied: a cursed raid member keeps its -- debuff colour after the curse is cleaned, and HasDebuffID answers true for -- auras that are long gone. It takes a second addon shipping the newer Compost -- to show up, because AceLibrary arbitrates on the highest $Revision and -- ignores load order -- so a lone Puppeteer keeps its own working rev 11579. -- -- Restore an in-place Erase matching the pre-stub body. Probed by behavior -- rather than by revision or addon name, so a Compost that already clears is -- left alone. -- -- Re-checked on every ADDON_LOADED rather than once, because AceLibrary can -- upgrade a library at any point during load -- including from an addon loaded -- on demand mid-session -- and an upgrade installs the new copy's functions -- over ours. We cannot simply wait for a late event either, since this addon -- loads first and AceLibrary does not exist yet at that point. -- -- Repeating is cheap because the verdict is cached against the function that -- earned it: a check settles to two identity comparisons once the live Erase -- has been either vetted or replaced, and only an Erase we have never seen -- costs a probe. The probe table is reused across checks (refilled each time, -- so leftover state cannot skew the answer) rather than built per check. local compostProbe = {} local vettedErase local function EraseInPlace(self, t) if type(t) ~= "table" then return end setmetatable(t, nil) for k in pairs(t) do t[k] = nil end table.setn(t, 0) return t end local function EnsureCompostErasesInPlace() if not (AceLibrary and AceLibrary.HasInstance and AceLibrary:HasInstance("Compost-2.0")) then return end local compost = AceLibrary("Compost-2.0") if compost.Erase == EraseInPlace or compost.Erase == vettedErase then return end compostProbe[1] = "array" compostProbe.key = "hash" compost:Erase(compostProbe) if next(compostProbe) == nil then vettedErase = compost.Erase -- clears in place; leave it alone return end compost.Erase = EraseInPlace end EventRegistry:RegisterFrameEventAndCallback("ADDON_LOADED", EnsureCompostErasesInPlace) EventRegistry:RegisterFrameEventAndCallback("PLAYER_LOGIN", EnsureCompostErasesInPlace) EnsureCompostErasesInPlace() -- Some addons carry a copy of Blizzard's own color definitions -- a plain -- `NORMAL_FONT_COLOR = {r=1.0, g=0.82, b=0}` and its friends -- and assign -- them at file scope. That takes ColorMixin off those globals for every addon -- that loads later, which is what breaks pfUI's who list when it calls -- `NORMAL_FONT_COLOR:GetRGB()` (brues-code/pfUI#56). -- -- Re-apply after each addon loads, so a clobbered color is repaired before the -- next addon reads it. This also covers an addon loaded on demand mid-session. -- The pass mixes the methods back into the table that is there, and costs one -- lookup per color when nothing is broken. EventRegistry:RegisterFrameEventAndCallback("ADDON_LOADED", CAPI_ApplyStandardColorGlobals) EventRegistry:RegisterFrameEventAndCallback("PLAYER_LOGIN", CAPI_ApplyStandardColorGlobals)