SamandClaude Fable 5
6e3a67e039
- a corpse is a body you can hover, cursor and click: it enters the same ray pass as units with the winner switched on type at publish (the reference's own shape — one trace over every object), the cursor classifier is transcribed whole (lootable-by-me → the pickup cursor, grayed by range; else the insignia leg behind the learned-skill latch; your own body falls through to the clear), and right-click is "not mounted AND lootable" → CMSG_LOOT; CORPSE_FIELD_FLAGS bit 5 is the PvP insignia flag, not "reclaimable" as an old note had it; the widening also caught that a GameObject anywhere on screen could steal the click from a nearer corpse, and that a left-click on a body cleared your target as if you had clicked ground
- the cinematic sweep closes eleven holes: a shot plays its sequence band's WIDTH, not its end — two shipped fly-bys band away from zero and ran long with their opening frames frozen (the Night Elf intro measured 102333 → 102000 ms); CMSG_NEXT_CINEMATIC_CAMERA announces every shot, the first included — the old "between shots" reading meant it was sent NEVER, since every 1.12 sequence row carries exactly one camera; the ack invariant is stated once (one CMSG_COMPLETE_CINEMATIC per accepted trigger, over-acking tolerated, under-acking never — a dropped ack despawns the world around you until relog); the hardware cursor hides for the shot; and the music cut lands in the frame that starts it — it was one frame late, exactly long enough for a first login's zone fanfare to start under the narration and be stamped as played
- RunBinding exists, and a fly-by stops handing the world its mouse: CinematicFrame's OnKeyDown hands the SCREENSHOT chord back to its binding through RunBinding, a global we never had, so every press during a fly-by raised and took no picture; and the mouse was fully live under a shot — right-drag turned the character, clicks picked targets through the flying camera, both buttons walked the body — so the pointer now reads as over the UI while a shot plays
- AdvanceTime and ReplaceIconTexture are carved off the bytes: AdvanceTime really is a no-op in the reference (its one reachable call is mov eax,1; ret), ReplaceIconTexture is a material swap over a model's type-14 texture slots that stores nothing on the widget, and SetModel(nil) RAISES — the "documented clear" our binding claimed was invented, not read; and the addon roster now says WHOSE package is short when a .toc names files the folder doesn't hold — nine of the corpus's twenty-nine load failures are the addon's own manifest, which the real client also shrugs off ("Couldn't open %s", carry on), so they stop reading as our defects
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
719 lines
45 KiB
XML
719 lines
45 KiB
XML
<Ui>
|
|
<!--
|
|
UIParent itself (ref UIParent.xml l.5, quoted): the real, full-screen, invisible root
|
|
frame the reference UI hangs everything on. Until now benilla shipped only this file's
|
|
Lua (the manage pass below) and no frame - so the Lua global `UIParent` was nil, and
|
|
every ref-quoted `SetOwner(UIParent, ...)` / `GameTooltip_SetDefaultAnchor(this,
|
|
UIParent)` died before it could place anything (the world tooltip stayed parked on its
|
|
load-time anchor - the "tooltip on my character" bug). String anchors ("UIParent" in
|
|
SetPoint) had silently fallen back to the screen root, which has the same rect, so only
|
|
the *global* was ever missing. No enableMouse: the hit test gates on it (pointer.rs), so
|
|
the full-screen plate swallows nothing.
|
|
|
|
The ref's UIParent_OnLoad registers a long list of globally-routed events and its OnEvent
|
|
is a chain of `if event == ...` arms; ours ships exactly the ONE arm that has a consumer
|
|
(the extra bars' login seed, below) and grows an arm at a time as consumers appear. A
|
|
transcription of the whole routing table would be a list of no-ops that reads as coverage.
|
|
|
|
Known seam, stated: the real client PARENTS every FrameXML frame under UIParent (scale/
|
|
visibility cascade). benilla is PART WAY there - 41 of our top-level frames declare
|
|
parent="UIParent" where the reference declares 104, and 32 frames the reference parents
|
|
(MainMenuBar, PlayerFrame, TargetFrame, the four MultiBars, UIErrorsFrame, the panels)
|
|
still attach to the screen root because the transcription dropped the attribute. Same
|
|
rect, so anchors agree.
|
|
|
|
The cascade consumer this note was waiting for HAS appeared (decision 1724): a cinematic
|
|
hides the HUD by doing `UIParent:Hide()` in the reference, and benilla substitutes an
|
|
engine-side draw-kill that also switches the mouse feed off - which is how the world came
|
|
to take every click of a race intro. Restoring the dropped parents is the fix; it is a
|
|
whole-UI arc (draw order, strata, ALT-Z, the world map's own full-screen row) and so is
|
|
proposed rather than taken.
|
|
-->
|
|
<Frame name="UIParent" setAllPoints="true" frameStrata="MEDIUM">
|
|
<Scripts>
|
|
<OnLoad>
|
|
this:RegisterEvent("PLAYER_ENTERING_WORLD")
|
|
this:RegisterEvent("READY_CHECK")
|
|
</OnLoad>
|
|
<OnEvent>UIParent_OnEvent(event)</OnEvent>
|
|
</Scripts>
|
|
</Frame>
|
|
|
|
<!--
|
|
WorldFrame — the reference's full-screen, bottom-of-the-strata frame that the 3D world
|
|
renders into. **Ours renders nothing**: benilla draws the world through Bevy's render
|
|
graph, not through a FrameScript frame, so this is a NAMED HANDLE and nothing more. That
|
|
is a disclosed divergence, not a pretence — and it is what makes the two things addons
|
|
actually do with it work: `WorldFrame:GetWidth()`/`GetHeight()` (the screen rect, which
|
|
this frame really has) and `frame:SetParent(WorldFrame)` / `SetPoint(..., WorldFrame,
|
|
...)` (parking a tooltip or an overlay under everything). Measured demand: 29 of the 218
|
|
vanilla-corpus addons die on their first load with `attempt to index global 'WorldFrame'`
|
|
without it (the corpus harness, `benilla_app::addon_harness`).
|
|
|
|
**No `enableMouse`, deliberately** — the same reason UIParent above ships without it, and
|
|
here it is load-bearing rather than incidental. benilla's `PointerOverUi` is "is the
|
|
topmost mouse-enabled frame under the cursor a UI frame", and it gates camera look, world
|
|
clicks and hover targeting (`player/camera.rs`, `target/hover.rs`); a full-screen
|
|
mouse-enabled plate would answer yes for every world pixel and kill all three. The
|
|
reference gets away with it because *its* WorldFrame IS the world, so a click landing
|
|
there is a world click by construction. Reproducing that here means teaching the arbiter
|
|
that a WorldFrame hit is not a UI hit — engine work with its own tests, not a rename — so
|
|
until then the addon-facing `OnMouseDown`/`OnMouseUp` hooks on this frame never fire.
|
|
-->
|
|
<Frame name="WorldFrame" setAllPoints="true" frameStrata="BACKGROUND"/>
|
|
|
|
<!--
|
|
The UIParent bottom-stack position manager - UIPARENT_MANAGED_FRAME_POSITIONS and
|
|
UIParent_ManageFramePositions(), transcribed from the extracted 1.12 UIParent.lua
|
|
l.1560-1697 (patch-2.MPQ via mpqx, byte-diffed against this install's own chain;
|
|
decision 0272). The reference never leaves a bottom-anchored HUD frame on its static
|
|
XML anchor: every layout pass re-anchors it to a base offset plus per-bar deltas -
|
|
either bottom multibar, the pet/stance bar, the reputation watch bar - fired from those
|
|
bars' OnShow/OnHide (ref BonusActionBarFrame.xml:350-355 et al.) and applied once at
|
|
startup. The loader's post-load call (ui_script/mod.rs) is our startup seam; the stance
|
|
bar's OnShow/OnHide (StanceBar.xml) is the one live re-fire today.
|
|
|
|
Departures, each named:
|
|
- Lua 5.1 dialect (the 1.12 client ran 5.0): pairs() iteration, table.insert for
|
|
tinsert, _G[name] = v for setglobal, and the ref's undeclared `hasBottomRight`
|
|
declared local (it is never read).
|
|
- The ref gates the bottom flags on its interface-option globals
|
|
SHOW_MULTI_ACTIONBAR_1/2; benilla's bottom bars are always-on with no options state
|
|
(decision 0270), so the flags read the bars' own visibility - identical values, no
|
|
parallel global to drift.
|
|
- Frame names: our stance bar, pet bar and bottom multibars wear the reference's own
|
|
names (ShapeshiftBarFrame, PetActionBarFrame, MultiBarBottomLeft/Right), so their
|
|
rows and flags in the table below are LIVE, not the no-ops they were while those
|
|
frames carried a `Benilla` prefix. That is the payoff the bare names buy — the same
|
|
one `group_loot_tests::managed_positions_engage_for_the_bare_frame_name` pins for
|
|
GroupLootFrame1 — and it costs nothing on screen: each frame's static XML anchor was
|
|
already written to the value the row computes (the stance bar's XML (30,45) is the
|
|
row's xOffset 30 + baseY 0 + bottomLeft 45, and MultiBarBottomLeft's XML (0,17) is
|
|
its row's baseY 17), so the pass now re-derives what the XML asserted. TutorialFrameParent
|
|
and FramerateLabel still have no frame here and no-op through the ref's `if ( frame )`.
|
|
- Every frame these flags read now exists, so every flag is a real read — the nil guards
|
|
stay because the ref reads them bare only because it always creates them, and our
|
|
per-window test harnesses load file SUBSETS. `reputation` fires the moment a watched
|
|
faction raises ReputationWatchBar over a shown MainMenuExpBar (decision 1258; it could
|
|
not fire at all until that page landed, and `reputation_frame_tests` pins the 9px it
|
|
moves the bottom stack by). `maxLevel` fires exactly when the level-60 rail replaces the
|
|
XP bar — which is also why the two never fire together: at 60 the watch bar REPLACES
|
|
MainMenuExpBar rather than stacking above it, so `MainMenuExpBar:IsShown()` is false and
|
|
the stack stays put. Both are the ref's own arithmetic.
|
|
- No IsUserPlaced/IsObjectType/SIMPLE_CHAT legs on the SetPoint: benilla has no
|
|
user-placed frames and no simple-chat option, so the ref's guarded SetPoint
|
|
collapses to the plain one.
|
|
- Of the ref function's trailing custom positioning, the pet-bar half IS ours (the
|
|
sliding shelf art + the PETACTIONBAR_XPOS/YPOS re-seat, at the end of the pass,
|
|
decision 0988). BattlefieldMinimapTab and the shapeshift-bar art swap stay out of
|
|
scope - neither frame exists here, our stance bar shipping no end-caps at all (0270).
|
|
- The isVar globals (CONTAINER_OFFSET_*, PETACTIONBAR_YPOS, BATTLEFIELD_TAB_OFFSET_Y)
|
|
are pre-declared to their base values below - the ref declares them in
|
|
ContainerFrame.lua/PetActionBarFrame.lua, which we don't ship - so the ref's
|
|
`if ( frame )` guard (getglobal on a var row yields the number) behaves identically.
|
|
Nothing consumes them yet; they are set for the day the bag anchors do.
|
|
-->
|
|
<Script><![CDATA[
|
|
-- Declared by the ref's ContainerFrame.lua / PetActionBarFrame.lua / WorldStateFrame.lua
|
|
-- (files we don't ship); the manage pass below overwrites them every run.
|
|
CONTAINER_OFFSET_X = 0
|
|
CONTAINER_OFFSET_Y = 70
|
|
BATTLEFIELD_TAB_OFFSET_Y = 210
|
|
PETACTIONBAR_YPOS = 97
|
|
-- ref PetActionBarFrame.lua:3. Not a managed row (the manage pass computes only the Y);
|
|
-- the custom block at the end of the pass recomputes it, for the reason recorded there.
|
|
PETACTIONBAR_XPOS = 36
|
|
|
|
-- ref UIParent.lua:2 — the model panes' held-button spin rate (half a turn per second),
|
|
-- read by every `Model_OnUpdate` transcription (the character window's + inspect's).
|
|
ROTATIONS_PER_SECOND = .5
|
|
|
|
-- ref SecondsToTime (UIParent.lua:1004-1031) and its GlobalStrings, in the ref's own home
|
|
-- file — two consumers now (the quest-timer window and the quest log's "Time Remaining:"
|
|
-- row, decision 1150), so it lives here rather than being redefined per file the way
|
|
-- BuffFrame.xml had to for SecondsToTimeAbbrev when it was the only caller.
|
|
--
|
|
-- Departures, both named:
|
|
-- - `GetText(TOKEN, nil, n)` is the ref's plural-aware GlobalString lookup, a C binding we
|
|
-- don't ship. Its enUS rule is the singular token for exactly 1 and the `_P1` twin
|
|
-- otherwise — the same rule already codified in `tooltip_item/render.rs` and MailFrame.xml
|
|
-- — so it collapses to the local `plural()` below and the four shipped string pairs
|
|
-- (GlobalStrings.lua:895-896, 2296-2297, 2693-2694, 3404-3405).
|
|
-- - Lua 5.1 dialect: `math.mod`/`math.floor` for the 5.0 bare `mod`/`floor`.
|
|
--
|
|
-- Everything else is verbatim, INCLUDING the two behaviours that look like bugs and are
|
|
-- not: the thresholds are strict `>` (86400 s reads as hours, not "1 Day"), and `count < 2`
|
|
-- stops after two units, so a 3-day quest reads "3 Days 4 Hrs" and never mentions minutes.
|
|
DAYS_ABBR = DAYS_ABBR or "Day"
|
|
DAYS_ABBR_P1 = DAYS_ABBR_P1 or "Days"
|
|
HOURS_ABBR = HOURS_ABBR or "Hr"
|
|
HOURS_ABBR_P1 = HOURS_ABBR_P1 or "Hrs"
|
|
MINUTES_ABBR = MINUTES_ABBR or "Min"
|
|
MINUTES_ABBR_P1 = MINUTES_ABBR_P1 or "Mins"
|
|
SECONDS_ABBR = SECONDS_ABBR or "Sec"
|
|
SECONDS_ABBR_P1 = SECONDS_ABBR_P1 or "Secs"
|
|
|
|
if not SecondsToTime then
|
|
local function plural(singular, pluralForm, n)
|
|
if n == 1 then return singular end
|
|
return pluralForm
|
|
end
|
|
|
|
function SecondsToTime(seconds, noSeconds)
|
|
local time = ""
|
|
local count = 0
|
|
local tempTime
|
|
seconds = math.floor(seconds)
|
|
if seconds > 86400 then
|
|
tempTime = math.floor(seconds / 86400)
|
|
time = tempTime .. " " .. plural(DAYS_ABBR, DAYS_ABBR_P1, tempTime) .. " "
|
|
seconds = math.mod(seconds, 86400)
|
|
count = count + 1
|
|
end
|
|
if seconds > 3600 then
|
|
tempTime = math.floor(seconds / 3600)
|
|
time = time .. tempTime .. " " .. plural(HOURS_ABBR, HOURS_ABBR_P1, tempTime) .. " "
|
|
seconds = math.mod(seconds, 3600)
|
|
count = count + 1
|
|
end
|
|
if count < 2 and seconds >= 60 then
|
|
tempTime = math.floor(seconds / 60)
|
|
time = time .. tempTime .. " " .. plural(MINUTES_ABBR, MINUTES_ABBR_P1, tempTime) .. " "
|
|
seconds = math.mod(seconds, 60)
|
|
count = count + 1
|
|
end
|
|
if count < 2 and seconds > 0 and not noSeconds then
|
|
time = time .. seconds .. " " .. plural(SECONDS_ABBR, SECONDS_ABBR_P1, seconds) .. " "
|
|
end
|
|
return time
|
|
end
|
|
end
|
|
|
|
-- ── The login seed for the extra action bars ──────────────────────────────────────────
|
|
--
|
|
-- The four extra bars' visibility is stored SERVER-SIDE, in `PLAYER_FIELD_BYTES` byte 2,
|
|
-- and the client never writes that byte itself — `SetActionBarToggles` only posts
|
|
-- `CMSG_SET_ACTIONBAR_TOGGLES` and waits for the field to come back (wow-re
|
|
-- `system/ui/scratch/action-bar-toggles.md` §5: the setter's every store is into its own
|
|
-- stack frame, and nothing in the binary fires on a change to that byte). So the live UI
|
|
-- state is the four Lua globals `MultiBars.xml` owns, and `GetActionBarToggles()` is read
|
|
-- exactly ONCE, here, to seed them. Reading it again later would fight the optimistic
|
|
-- value the options rows just set and lose a round trip's worth of clicks.
|
|
--
|
|
-- This is the reference's own seam, arm for arm (its UIParent_OnEvent does the same two
|
|
-- statements on the same event). `MultiActionBar_Update` is existence-guarded because the
|
|
-- ~25 per-window harnesses that load this file load no bars; the binding itself is engine
|
|
-- level and always installed. Decision 1500.
|
|
function UIParent_OnEvent(event)
|
|
if ( event == "PLAYER_ENTERING_WORLD" ) then
|
|
SHOW_MULTI_ACTIONBAR_1, SHOW_MULTI_ACTIONBAR_2, SHOW_MULTI_ACTIONBAR_3,
|
|
SHOW_MULTI_ACTIONBAR_4 = GetActionBarToggles()
|
|
if ( MultiActionBar_Update ) then
|
|
MultiActionBar_Update()
|
|
end
|
|
end
|
|
-- The ready-check popup's opener (decision 1549), ref UIParent.lua l.570-572. The
|
|
-- event is the engine's own `0x218`, fired by the `MSG_RAID_READY_CHECK` open
|
|
-- handler; the frame and `ShowReadyCheck` live in RaidFrame.xml, so this is the one
|
|
-- place the two meet. Existence-guarded for `MultiActionBar_Update`'s reason: the
|
|
-- per-window harnesses load this file without the raid pane.
|
|
if ( event == "READY_CHECK" ) then
|
|
if ( ShowReadyCheck ) then
|
|
ShowReadyCheck()
|
|
end
|
|
end
|
|
end
|
|
|
|
-- ── Everything that has to move when the bottom of the screen fills up ──────────────────
|
|
--
|
|
-- One row per thing whose seat depends on which bars are raised. A row reads:
|
|
--
|
|
-- baseY where it sits on the y axis with nothing in the way
|
|
-- baseX / xOffset the same for the x axis, under two names: the rows that write a var
|
|
-- spell it `baseX`, the rows that seat a frame spell it `xOffset`, and
|
|
-- baseX wins if a row ever names both
|
|
-- bottomLeft, bottomRight, bottomEither, pet, reputation, maxLevel (y axis)
|
|
-- rightLeft, rightRight (x axis)
|
|
-- what this row owes each raised bar, ADDED for every flag standing
|
|
-- anchorTo / point / rpoint who to hang off; defaulted to UIParent BOTTOM-to-BOTTOM
|
|
-- isVar the row is not a frame at all: it names a GLOBAL, and the pass writes
|
|
-- the computed offset into it for other files to read
|
|
--
|
|
-- The KEY is the global name — a frame to seat, or the var to write — and other windows'
|
|
-- OnShow/OnHide drive the pass by it, so both the keys and the fields are a published
|
|
-- shape. Written as one constructor rather than a dozen index assignments: it is a
|
|
-- constant, and a constant reads better as a table than as a program that builds one.
|
|
UIPARENT_MANAGED_FRAME_POSITIONS = {
|
|
-- Frames to seat
|
|
MultiBarBottomLeft = {baseY = 17, reputation = 9, maxLevel = -5, anchorTo = "ActionButton1", point = "BOTTOMLEFT", rpoint = "TOPLEFT"},
|
|
GroupLootFrame1 = {baseY = 60, bottomEither = 42, pet = 42, reputation = 9},
|
|
TutorialFrameParent = {baseY = 55, bottomEither = 47, pet = 42, reputation = 9},
|
|
FramerateLabel = {baseY = 64, bottomEither = 42, pet = 42, reputation = 9},
|
|
CastingBarFrame = {baseY = 60, bottomEither = 40, pet = 40, reputation = 9},
|
|
ChatFrame1 = {baseY = 85, bottomLeft = 17, pet = 17, reputation = 9, maxLevel = -5, anchorTo = "UIParent", point = "BOTTOMLEFT", rpoint = "BOTTOMLEFT", xOffset = 32},
|
|
-- ChatFrame2 is NOT managed here, and its absence is the fix for the latent
|
|
-- decision 1499 flagged in the ClearAllPoints comment below. The reference's row
|
|
-- seats an UNDOCKED ChatFrame2 against the screen's bottom-right; ours is DOCKED to
|
|
-- ChatFrame1, and a docked frame takes its rect from the dock, never from this pass.
|
|
-- With the row present the pass cleared ChatFrame2's authored anchors and re-seated it
|
|
-- with a single BOTTOMRIGHT point and no size — an unresolvable rect, so the window
|
|
-- rendered nothing at all. That was invisible while ChatFrame2 had no tab; the moment
|
|
-- one existed (1575) it became "the Combat Log tab shows an empty screen".
|
|
-- Re-add this row the day windows can be undocked, which is chat *settings* (0288 §2).
|
|
ShapeshiftBarFrame = {baseY = 0, bottomLeft = 45, reputation = 9, maxLevel = -5, anchorTo = "MainMenuBar", point = "BOTTOMLEFT", rpoint = "TOPLEFT", xOffset = 30},
|
|
|
|
-- Globals to write
|
|
CONTAINER_OFFSET_X = {baseX = 0, rightLeft = 90, rightRight = 45, isVar = "xAxis"},
|
|
CONTAINER_OFFSET_Y = {baseY = 70, bottomEither = 27, bottomRight = 0, reputation = 9, isVar = "yAxis", pet = 23},
|
|
BATTLEFIELD_TAB_OFFSET_Y = {baseY = 210, bottomRight = 40, reputation = 9, isVar = "yAxis"},
|
|
PETACTIONBAR_YPOS = {baseY = 97, bottomLeft = 43, reputation = 9, maxLevel = -5, isVar = "yAxis"},
|
|
}
|
|
|
|
-- ── Consumers of the VAR rows ──────────────────────────────────────────────────────────
|
|
-- A row with `isVar` leaves a NUMBER in a global and anchors nothing, so whoever seats
|
|
-- itself from that number has to be told when the number changes — otherwise it keeps the
|
|
-- seat it computed the first time and the bars grow underneath it. That is not a
|
|
-- hypothetical: the open-bag stack read a file-local copy of CONTAINER_OFFSET_X/Y and sat
|
|
-- at the no-bars corner forever, so a bottom multibar drew straight through the lowest bag
|
|
-- window (director-caught, decision 1499).
|
|
--
|
|
-- The reference closes the same loop by hard-calling `updateContainerFrameAnchors()` from
|
|
-- the end of its own pass (UIParent.lua:1774). This is that call generalized: a consumer
|
|
-- registers itself and the pass calls it after every recompute, so the NEXT var consumer
|
|
-- does not have to edit this file to be noticed. Order of registration is the order of
|
|
-- call; listeners are independent (nothing here reads what another writes).
|
|
--
|
|
-- The law, stated once so it can be pointed at: **nothing outside this file may keep its
|
|
-- own copy of a managed offset.** Read the global fresh at use, and if you cache a SEAT
|
|
-- derived from it, register here. `bottom_hud_tests.rs` enforces both halves.
|
|
UIPARENT_MANAGED_POSITION_LISTENERS = {}
|
|
|
|
function UIParent_RegisterManagedPositionListener(fn)
|
|
table.insert(UIPARENT_MANAGED_POSITION_LISTENERS, fn)
|
|
end
|
|
|
|
function UIParent_ManageFramePositions()
|
|
-- Frames that affect offsets in y axis
|
|
local yOffsetFrames = {}
|
|
-- Frames that affect offsets in x axis
|
|
local xOffsetFrames = {}
|
|
|
|
-- Set up flags. Departure: the bottom flags read the always-on bars' visibility,
|
|
-- not the ref's SHOW_MULTI_ACTIONBAR_1/2 option globals (0270 ships no such state).
|
|
local bottomLeftShown = MultiBarBottomLeft and MultiBarBottomLeft:IsShown()
|
|
local bottomRightShown = MultiBarBottomRight and MultiBarBottomRight:IsShown()
|
|
if ( bottomLeftShown or bottomRightShown ) then
|
|
table.insert(yOffsetFrames, "bottomEither")
|
|
end
|
|
if ( bottomRightShown ) then
|
|
table.insert(yOffsetFrames, "bottomRight")
|
|
end
|
|
if ( bottomLeftShown ) then
|
|
table.insert(yOffsetFrames, "bottomLeft")
|
|
end
|
|
|
|
if ( MultiBarLeft and MultiBarLeft:IsShown() ) then
|
|
table.insert(xOffsetFrames, "rightLeft")
|
|
elseif ( MultiBarRight and MultiBarRight:IsShown() ) then
|
|
table.insert(xOffsetFrames, "rightRight")
|
|
end
|
|
|
|
if ( ( PetActionBarFrame and PetActionBarFrame:IsShown() ) or ( ShapeshiftBarFrame and ShapeshiftBarFrame:IsShown() ) ) then
|
|
table.insert(yOffsetFrames, "pet")
|
|
end
|
|
if ( ReputationWatchBar and MainMenuExpBar and ReputationWatchBar:IsShown() and MainMenuExpBar:IsShown() ) then
|
|
table.insert(yOffsetFrames, "reputation")
|
|
end
|
|
if ( MainMenuBarMaxLevelBar and MainMenuBarMaxLevelBar:IsShown() ) then
|
|
table.insert(yOffsetFrames, "maxLevel")
|
|
end
|
|
|
|
-- Walk the rows and settle each one against the flags now standing. Rows are
|
|
-- independent — nothing here reads a var another row writes — so the arbitrary order
|
|
-- `pairs` hands back is not a hazard.
|
|
for index, row in pairs(UIPARENT_MANAGED_FRAME_POSITIONS) do
|
|
local frame = getglobal(index)
|
|
if ( frame ) then
|
|
-- Start from where the row sits with nothing in its way. A zero base is a real
|
|
-- base, not an absent one, which is why `or` chains do the right thing here.
|
|
local xOffset = row.baseX or row.xOffset or 0
|
|
local yOffset = row.baseY or 0
|
|
|
|
-- Then add what the row owes each raised bar, keeping note of which ones it
|
|
-- actually paid — two of them interact below.
|
|
local paid = {}
|
|
for _, flag in pairs(yOffsetFrames) do
|
|
if ( row[flag] ) then
|
|
yOffset = yOffset + row[flag]
|
|
paid[flag] = true
|
|
end
|
|
end
|
|
for _, flag in pairs(xOffsetFrames) do
|
|
if ( row[flag] ) then
|
|
xOffset = xOffset + row[flag]
|
|
end
|
|
end
|
|
|
|
-- The bottom-left bar and the pet/stance row do not simply stack: clearing both
|
|
-- costs 23px more than the two figures added together. The reference carries the
|
|
-- same correction (l.1657-1658) with no derivation, and it is a fact about the
|
|
-- bars' art rather than arithmetic we can recompute.
|
|
if ( paid.bottomLeft and paid.pet ) then
|
|
yOffset = yOffset + 23
|
|
end
|
|
|
|
-- A row either NAMES A GLOBAL to leave the number in, or is a frame to seat.
|
|
if ( row.isVar == "xAxis" ) then
|
|
_G[index] = xOffset
|
|
elseif ( row.isVar ) then
|
|
_G[index] = yOffset
|
|
elseif ( frame.IsUserPlaced and frame:IsUserPlaced() ) then
|
|
-- **A frame the PLAYER placed is not this pass's to seat.** benilla's rule,
|
|
-- not the reference's: 1.12 has no *movable* managed frame — its ChatFrame1
|
|
-- is docked and its tab click refuses to drag a docked window — so the case
|
|
-- cannot arise there. Ours does, since an unlocked chat window drags and
|
|
-- resizes, and without this the next pass would `ClearAllPoints` the window
|
|
-- and drop it back on the bottom-stack seat. That is not a rare path: the
|
|
-- stance bar fires this on every show and hide.
|
|
--
|
|
-- `IsUserPlaced` is exactly the right question rather than a chat-specific
|
|
-- one: the engine sets that bit at the drag that moved the frame, and the
|
|
-- layout cache (`benilla_app::ui_layout`) sets it again for every frame it
|
|
-- seats out of the player's file — so "the player owns this frame's
|
|
-- position" and "do not manage it" are the same fact.
|
|
--
|
|
-- The `isVar` rows above are untouched: they write a NUMBER and seat no
|
|
-- frame, so there is no position for a player to own.
|
|
else
|
|
-- ClearAllPoints first: the row states the frame's WHOLE seat, so the pass
|
|
-- must own it outright. `SetPoint` only replaces the same-NAMED point, so
|
|
-- any other anchor the frame was authored with survives and fights this
|
|
-- one — ChatFrame2 is declared in XML with both a TOPLEFT and a
|
|
-- BOTTOMRIGHT onto ChatFrame1, and the row's BOTTOMRIGHT replaced only
|
|
-- the second, leaving the pane stretched from ChatFrame1's top-left
|
|
-- corner to the screen's bottom-right. Masked today only because
|
|
-- ChatFrame2 ships hidden, which is the worst kind of latent (decision
|
|
-- 1499). Every row here supplies a single complete anchor, so this is a
|
|
-- no-op for the other eight and the reference's plain SetPoint is
|
|
-- behaviour-identical on its own frames.
|
|
frame:ClearAllPoints()
|
|
frame:SetPoint(row.point or "BOTTOM", row.anchorTo or "UIParent",
|
|
row.rpoint or "BOTTOM", xOffset, yOffset)
|
|
end
|
|
end
|
|
end
|
|
|
|
-- Custom positioning the row table cannot express (ref l.1705-1738). Both rules
|
|
-- belong to the pet/stance ROW — the one row two different bars share — and without
|
|
-- them the pet bar lands on top of the bottom-left multibar's buttons
|
|
-- (director-caught: the bar drew across a row of spells, border and all).
|
|
--
|
|
-- (a) The shelf art IS the pet bar's border, and it only makes sense when the bar
|
|
-- sits directly on the main bar. Raised a row it would draw across whatever is
|
|
-- underneath, so the ref hides both strips (l.1707-1708) — the same move it makes
|
|
-- for the stance bar's own end-caps, which is why benilla's stance bar has none
|
|
-- (0270 anchors it statically at the raised position).
|
|
-- (b) The pet bar is the only managed entry whose row is a VAR: the loop above writes
|
|
-- the NUMBER PETACTIONBAR_YPOS (97, +43 with the bottom-left bar up) and anchors
|
|
-- nothing, so the ref re-seats the frame with it here (l.1735-1736).
|
|
--
|
|
-- Departure: the ref computes PETACTIONBAR_XPOS in ShowPetActionBar() and reads it
|
|
-- here, which only works because its bar slides in over several frames. Ours appears
|
|
-- at once, so the value is computed where it is used. Same rule
|
|
-- (PetActionBarFrame.lua:170-174): the two bars share this row, so a pet bar shown
|
|
-- while the stance bar is up starts 20px past the last stance button. That case is
|
|
-- rare but real — a possessed or charmed unit gives a pet bar to a class that also
|
|
-- has forms.
|
|
local bottomLeftUp = MultiBarBottomLeft and MultiBarBottomLeft:IsShown()
|
|
if ( SlidingActionBarTexture0 and SlidingActionBarTexture1 ) then
|
|
if ( bottomLeftUp ) then
|
|
SlidingActionBarTexture0:Hide()
|
|
SlidingActionBarTexture1:Hide()
|
|
else
|
|
SlidingActionBarTexture0:Show()
|
|
SlidingActionBarTexture1:Show()
|
|
end
|
|
end
|
|
if ( PetActionBarFrame ) then
|
|
PETACTIONBAR_XPOS = 36
|
|
local forms = GetNumShapeshiftForms and GetNumShapeshiftForms() or 0
|
|
if ( forms > 0 and ShapeshiftBarFrame and ShapeshiftBarFrame:IsShown() ) then
|
|
local last = getglobal("ShapeshiftButton"..forms)
|
|
local right = last and last:GetRight()
|
|
if ( right ) then
|
|
PETACTIONBAR_XPOS = right + 20
|
|
end
|
|
end
|
|
-- The ref guards this on IsShown() because a hidden bar of its is mid-slide and
|
|
-- owns its own anchor; ours has no slide, so re-seating a hidden frame is a
|
|
-- no-op and the guard would only add an ordering trap (OnShow fires the pass).
|
|
PetActionBarFrame:SetPoint("TOPLEFT", "MainMenuBar", "BOTTOMLEFT", PETACTIONBAR_XPOS, PETACTIONBAR_YPOS)
|
|
end
|
|
|
|
-- The under-minimap right stack (ref l.1740-1770): anchorY walks down capture bars →
|
|
-- quest timer → durability → quest watch, each shown frame pushing the next one down
|
|
-- by its own height. benilla ships the quest-timer, durability and quest-watch arms
|
|
-- (below); the capture bars above them don't exist yet, so anchorY starts at the ref's
|
|
-- initial 0. The reference never leaves DurabilityFrame on its static XML anchor
|
|
-- (whose +40 hangs the side glyphs off-screen — director-caught): every pass seats it
|
|
-- at -CONTAINER_OFFSET_X, minus 20 more when a showSeparate glyph
|
|
-- (shield/off-weapon/ranged) extends the art to the right — then drops the quest
|
|
-- tracker to the running anchorY so the two never collide at the same corner.
|
|
local anchorY = 0
|
|
-- ref l.1753-1757 verbatim (decision 1150). Seated whether shown or not — the ref
|
|
-- SetPoints unconditionally and only the height accrual is gated — so the frame is
|
|
-- already in place the frame it appears. Nil-guarded like every row here: the
|
|
-- durability-only test harness ships no quest-timer frame.
|
|
if ( QuestTimerFrame ) then
|
|
QuestTimerFrame:SetPoint("TOPRIGHT", "MinimapCluster", "BOTTOMRIGHT", -CONTAINER_OFFSET_X, anchorY)
|
|
if ( QuestTimerFrame:IsShown() ) then
|
|
anchorY = anchorY - QuestTimerFrame:GetHeight()
|
|
end
|
|
end
|
|
if ( DurabilityFrame ) then
|
|
local durabilityOffset = 0
|
|
if ( DurabilityShield:IsShown() or DurabilityOffWeapon:IsShown() or DurabilityRanged:IsShown() ) then
|
|
durabilityOffset = 20
|
|
end
|
|
DurabilityFrame:SetPoint("TOPRIGHT", "MinimapCluster", "BOTTOMRIGHT", -CONTAINER_OFFSET_X-durabilityOffset, anchorY)
|
|
if ( DurabilityFrame:IsShown() ) then
|
|
anchorY = anchorY - DurabilityFrame:GetHeight()
|
|
end
|
|
end
|
|
-- QuestWatchFrame is last in the ref's walk (l.1770), seated at the running anchorY so
|
|
-- it drops below the durability guy when he shows instead of overlapping him (both
|
|
-- anchor this same MinimapCluster corner — director-caught). benilla's tracker is
|
|
-- QuestWatchFrame; nil-guarded because the durability-only test harness (and
|
|
-- any config without a quest log) ships no such frame.
|
|
if ( QuestWatchFrame ) then
|
|
QuestWatchFrame:SetPoint("TOPRIGHT", "MinimapCluster", "BOTTOMRIGHT", -CONTAINER_OFFSET_X, anchorY)
|
|
end
|
|
|
|
-- Last: everything that seats itself from a var row this pass just rewrote. After the
|
|
-- frames, so a listener may measure one (the bag stack reads GetScreenHeight and its
|
|
-- own windows' heights, not another managed frame — but the ordering costs nothing and
|
|
-- removes a class of question).
|
|
for _, fn in ipairs(UIPARENT_MANAGED_POSITION_LISTENERS) do
|
|
fn()
|
|
end
|
|
end
|
|
|
|
-- The published binding-label API — benilla's implementation of the behaviour ref
|
|
-- UIParent.lua:1819-1878 specifies (decision 1008, which closed 0997's "abbreviates
|
|
-- through KEY_ABBR" residue with the REAL mechanism: there never was a KEY_ABBR set, the
|
|
-- whole abbreviation is this one function). Addons call it by name, so the contract is:
|
|
--
|
|
-- * a nil name reads as the empty string;
|
|
-- * "CTRL-SHIFT-2" splits at its LAST dash — up to and including that dash is the
|
|
-- modifier run ("CTRL-SHIFT-"), what follows is the key ("2"). A name with no dash is
|
|
-- all key;
|
|
-- * the key renders through the `prefix..key` global (the KEY_* family), preferring a
|
|
-- `_MAC` twin on a Mac client, and falls back to the raw key name;
|
|
-- * returnAbbr — what the action bar's HotKey corners pass — shrinks ONE modifier to
|
|
-- its initial, so SHIFT-2 reads "s-2" (the director's ask), and collapses TWO OR MORE
|
|
-- to "·". The full form (tooltips, the Keybindings page) keeps the whole prefix.
|
|
--
|
|
-- Two knowing departures from the reference, both about its code rather than its
|
|
-- behaviour: its `gsub(tempName, "CAPSLOCK", "Caps")` line is dead — gsub returns a new
|
|
-- string and that result is discarded, so nothing can observe it — and has no counterpart
|
|
-- here; and its GetLocale()/IsMacClient() calls ride behind existence guards because this
|
|
-- VM defines neither yet, with the getglobal fallback covering the gap.
|
|
|
|
-- What a single modifier shrinks to on a button corner. German calls Control "STRG" (the
|
|
-- deDE swap below runs first), which is the one that needs two letters to stay legible.
|
|
-- The order is fixed so the pass is deterministic, but it does not matter: no abbreviation
|
|
-- occurs inside another modifier's name, so no substitution can feed the next.
|
|
local MODIFIER_ABBREVIATIONS = {
|
|
{ "CTRL", "c" },
|
|
{ "SHIFT", "s" },
|
|
{ "ALT", "a" },
|
|
{ "STRG", "st" },
|
|
}
|
|
|
|
function GetBindingText(name, prefix, returnAbbr)
|
|
if not name then
|
|
return ""
|
|
end
|
|
|
|
local modifiers, key = "", name
|
|
local modifierCount = 0
|
|
-- The greedy `.*` walks to the LAST dash, which is the one that divides the modifier
|
|
-- run from the key; gsub's second return then counts the dashes, one per modifier —
|
|
-- plus one more when the key ITSELF is a dash ("CTRL--"), which is why that binding
|
|
-- abbreviates as if it had two. The reference's dash-walking loop reaches both of the
|
|
-- same answers, and both are pinned by keybindings_tests.
|
|
local _, lastDash = strfind(name, "^.*%-")
|
|
if lastDash then
|
|
modifiers = strsub(name, 1, lastDash)
|
|
key = strsub(name, lastDash + 1)
|
|
_, modifierCount = gsub(modifiers, "%-", "-")
|
|
if GetLocale and GetLocale() == "deDE" then
|
|
modifiers = gsub(modifiers, "CTRL", "STRG")
|
|
end
|
|
end
|
|
|
|
if returnAbbr then
|
|
if modifierCount > 1 then
|
|
-- Two modifiers never fit a corner: the whole label becomes the dot.
|
|
return "·"
|
|
end
|
|
for _, abbreviation in ipairs(MODIFIER_ABBREVIATIONS) do
|
|
modifiers = gsub(modifiers, abbreviation[1], abbreviation[2])
|
|
end
|
|
end
|
|
|
|
prefix = prefix or ""
|
|
local label
|
|
if IsMacClient and IsMacClient() then
|
|
label = getglobal(prefix .. key .. "_MAC")
|
|
end
|
|
label = label or getglobal(prefix .. key) or key
|
|
return modifiers .. label
|
|
end
|
|
|
|
-- ref UIParent.lua:1051-1057, transcribed 1:1 (decision 1062). The reference's own
|
|
-- list joiner, and it is exactly as plain as it looks: a bare ", " between every item,
|
|
-- with NO "and" before the last one and NO localization string of any kind — 1.12 has
|
|
-- no LIST_DELIMITER/AND global for it to reach for, and it does not build one. (The
|
|
-- richer `BuildColoredListString` right below it in the reference — the per-item
|
|
-- met/unmet red — already lives in TradeSkillFrame.xml as a guarded global, 0437 phase
|
|
-- 3; folding that one up here beside this is a follow-up, not this slice's churn.)
|
|
--
|
|
-- Callers: the pet paper doll's diet tooltip and PetStable.lua's, both as
|
|
-- `format(PET_DIET_TEMPLATE, BuildListString(GetPetFoodTypes()))`.
|
|
--
|
|
-- ZERO arguments returns NIL, not "" — the ref reads `arg[1]` off an empty vararg and
|
|
-- returns it unchanged. That is load-bearing: `format("%s", nil)` errors, so a caller
|
|
-- that can pass nothing (a warlock minion's empty diet) must be gated by its own UI, and
|
|
-- it is — the diet icon is hunter-only. Quietly returning "" here would hide that.
|
|
--
|
|
-- Lua-5.0 `arg`/`arg.n` written as `select("#", ...)`/`select(i, ...)`, the 5.1-native
|
|
-- spelling — TradeSkillFrame.xml's own substitution.
|
|
function BuildListString(...)
|
|
local n = select("#", ...);
|
|
if ( n == 0 ) then
|
|
return nil;
|
|
end
|
|
local list = select(1, ...);
|
|
for i = 2, n do
|
|
local item = select(i, ...);
|
|
list = list..", "..item;
|
|
end
|
|
return list;
|
|
end
|
|
|
|
-- BenillaChatEdit_InsertLink(link) → true if it went in, else nil (decision 1059).
|
|
--
|
|
-- Every shift-click site in this UI does the same two things before inserting: check the
|
|
-- chat edit box is up, and — ours alone — check the link is not nil. That nil case is
|
|
-- **benilla's, not the reference's**: our item templates resolve asynchronously
|
|
-- (`CMSG_ITEM_QUERY_SINGLE`, answered a frame or many later), so a `Get…Link` getter
|
|
-- legitimately answers nil for a slot whose name and quality have not landed, where the real
|
|
-- client's synchronous item cache always has one. `EditBox:Insert` takes a string (the
|
|
-- byte-modelled `0x77bee0`), so an unguarded nil is a script error — which is exactly the
|
|
-- bug this function exists to make impossible to reintroduce at the next site. A dozen call
|
|
-- sites were growing a dozen copies of the same two-line guard, and three of them shipped
|
|
-- without it.
|
|
--
|
|
-- The RETURN is what lets a caller fork on "did chat take it?" — a bag slot's shift-click
|
|
-- posts the link when chat is open and opens the stack splitter when it is not (ref
|
|
-- ContainerFrame.lua:567-577). Blizzard themselves eventually factored this exact helper out
|
|
-- for this exact reason: 2.0's `ChatEdit_InsertLink(text)`, returning true/false, called from
|
|
-- every one of these sites. 1.12 has no such function, so ours carries a benilla name rather
|
|
-- than pretending to be reference API.
|
|
function BenillaChatEdit_InsertLink(link)
|
|
if not link then
|
|
return nil
|
|
end
|
|
if not ChatFrameEditBox or not ChatFrameEditBox:IsVisible() then
|
|
return nil
|
|
end
|
|
ChatFrameEditBox:Insert(link)
|
|
return true
|
|
end
|
|
|
|
-- ── LowerFrameLevel / RaiseFrameLevel (ref UIParent.lua l.1890-1896), verbatim ────────
|
|
--
|
|
-- Two lines each, and the reason to have them is that they are FRAMEXML, not engine: an
|
|
-- addon calling `RaiseFrameLevel(f)` is calling a function the client's own UI defines, and
|
|
-- without it that is `attempt to call global` at load (BetterCharacterStats dies there).
|
|
-- Our own ActionBar.xml already describes this operation in prose at three sites while
|
|
-- doing it by hand; those can move onto the real name whenever that file is next opened.
|
|
function LowerFrameLevel(frame)
|
|
frame:SetFrameLevel(frame:GetFrameLevel() - 1)
|
|
end
|
|
|
|
function RaiseFrameLevel(frame)
|
|
frame:SetFrameLevel(frame:GetFrameLevel() + 1)
|
|
end
|
|
|
|
-- ── The 1024x768 scale ratios (ref UIParent.lua l.1794-1802) ────────────────────────────
|
|
--
|
|
-- Two FrameXML helpers addons call by name (the raid pullouts and a dozen third-party
|
|
-- drag handlers), so they exist for the reason `MouseIsOver` and `RaiseFrameLevel` do:
|
|
-- missing, the caller dies on `attempt to call global`. The reference's constants are the
|
|
-- 1024x768 reference resolution, and its own callers multiply a `GetCursorPosition()`
|
|
-- reading by them to convert into pixel space.
|
|
--
|
|
-- **benilla's own drag code does NOT use them**, and that is deliberate rather than an
|
|
-- oversight: our UI space is 1:1 logical px and `GetCursorPosition` already reports in
|
|
-- the same space `SetPoint` offsets are read in (the same identity `MouseIsOver`'s
|
|
-- comment below records for `GetEffectiveScale`). Multiplying here would place a dragged
|
|
-- frame at 1.5x the cursor on a 1600-wide window. Kept for the addons, unused by us.
|
|
function GetScreenHeightScale()
|
|
return GetScreenHeight() / 768
|
|
end
|
|
|
|
function GetScreenWidthScale()
|
|
return GetScreenWidth() / 1024
|
|
end
|
|
|
|
-- ── MouseIsOver: is the cursor inside this frame, with optional per-edge padding ─────────
|
|
--
|
|
-- The idiom every hover-driven addon writes, and **9 corpus addons call it** by name.
|
|
-- Missing, each is `attempt to call global 'MouseIsOver'`. So the CONTRACT is 1.12's, quirks
|
|
-- and all (ref UIParent.lua l.1388-1417) — three of which look like bugs and are not:
|
|
--
|
|
-- · each offset is ADDED to its own edge, in that axis's positive direction. A positive
|
|
-- `topOffset` raises the ceiling (growing the box upward) while a positive
|
|
-- `bottomOffset` raises the floor (shrinking it from below). The reference's own caller
|
|
-- proves the intent: `MouseIsOver(chatFrame, 45, -10, -5, 5)` is 45 up into the tab
|
|
-- strip, 10 in from the bottom, 5 out each side.
|
|
-- · the padding is ALL OR NOTHING on `topOffset`: pass none and the other three are
|
|
-- dropped with it, so `MouseIsOver(f, nil, 5)` silently ignores the 5. Pass a topOffset
|
|
-- and omit any of the rest and it is an arithmetic error, not a zero — 1.12 raises
|
|
-- there too, and an addon written against it may lean on either half.
|
|
-- · the bounds are strict, so a cursor exactly on an edge is outside.
|
|
--
|
|
-- A frame that has never been laid out answers nil for its rect; nil is the right answer to
|
|
-- hand back for it, and the reference's own comment on that guard ("a symptom not the real
|
|
-- issue") is about where the unresolved frame came from, not about the guard.
|
|
--
|
|
-- ONE DIVERGENCE, and it is a no-op: the reference divides the cursor by
|
|
-- `frame:GetEffectiveScale()` because 1.12 hands Lua screen pixels. benilla's UI space is
|
|
-- 1:1 logical px and `GetEffectiveScale` is the constant 1 (`script/pointer.rs`,
|
|
-- `object.rs` l.271), so the division is written out rather than left in to divide by one
|
|
-- forever. If a real UI scale ever lands, this is one of the sites that needs it back.
|
|
function MouseIsOver(frame, topOffset, bottomOffset, leftOffset, rightOffset)
|
|
local x, y = GetCursorPosition()
|
|
local left, right = frame:GetLeft(), frame:GetRight()
|
|
local top, bottom = frame:GetTop(), frame:GetBottom()
|
|
if not left then
|
|
return
|
|
end
|
|
if not topOffset then
|
|
topOffset, bottomOffset, leftOffset, rightOffset = 0, 0, 0, 0
|
|
end
|
|
-- All four edges are padded before any of them is tested, and deliberately so: folding
|
|
-- the additions into the comparison would let a short circuit swallow the arithmetic
|
|
-- error that a half-supplied offset set is supposed to raise.
|
|
left, right = left + leftOffset, right + rightOffset
|
|
top, bottom = top + topOffset, bottom + bottomOffset
|
|
if x > left and x < right and y > bottom and y < top then
|
|
return 1
|
|
end
|
|
return nil
|
|
end
|
|
]]></Script>
|
|
</Ui>
|