Files
benilla-octo/crates/benilla-ui/src/script/tick.rs
T
SamandClaude Fable 5.1 c3555b4eea the cooldown goes stock, early-Z comes back, and lists reach the end
- a size-less `<Model>` pane's rect is its file's box in layout units: the reference's geometry overrides (`0x76d080`/`0x76d0d0`) answer the bbox extent whenever no size is authored, `bboxExtent · 768·√(a²+1) · layoutScale` FrameXML units (`1280 · extent` at 4:3), written into the pane's layout input when the file's facts land, re-derived on an aspect change, and ended for good by any authored size; `SetPosition` is the same unit, which is how `0x4a7b20` centres the arrow at `½·GetWidth, ½·GetHeight` through the override — so the map arrow's sprite arm and its hand-sized 33.6 px footprint go (its rect is `0.0262 × 0.0263` layout units, 33.5 × 33.7 FrameXML units at 4:3, 41.0 × 41.2 at 16:9), and the stock `ItemAnim` pane plays the item push unadapted (`ReplaceIconTexture` on the type-14 batch, the 1000 ms clamp, the stock `OnAnimFinished`), `ContainerFrameAdapters.xml` losing 0887's Lua card and its hand-measured 42.41 × 124.72 rect — the implicit one at 16:9 — and keeping the keyring fork and the three bag verbs (2015); and the cooldown is the reference's own model: the texture-transform law `uv' = R_q((uv + t − p) ⊙ s) + p`, `p = (½, ½)`, the raw quaternion component-lerped and never normalised, is one function (`uv_transform`) with the shader's UV fold as its twin — rotation and scale tracks bake per file sequence slot beside the translation, the mat-anim table gains an affine row `[cos − 1, sin, sx − 1, sy − 1]` on `anim_slots.z` whose identity is the zero row so every static batch reads exactly what it did, and a UI tile clones the twin material per animated batch and writes both rows every frame off the pane's play head, never the world clock; the stock `Cooldown.xml` runs off the chain on 2007's two handlers, the file's four quadrant quads turned by their own rotation tracks (`0 → −90°` over each quadrant's 250 ms window, clockwise from 12 o'clock) and its alpha tracks revealing the star for the flash; retired: the native `Cooldown` widget and `SetCooldown` (an Era API 1.12 never had — `CreateFrame("Cooldown")` is the reference's unknown-type refusal now), the loader's `Model → Cooldown` substitution, the app's hand-built pie and star, and our `Cooldown.xml`; `assets/ui` is eight files (2019)
- `UpdateMapHighlight` gets its zone-level exits (B360): at zone zoom ours returned the nil tail for every cursor position, so hovering Deadwind Pass from Duskwood showed no label where 1.12 shows "Deadwind Pass"; the reference's `0x4a7fa0` runs two lookups at zone level and answers a NAME only — first the current zone's exploration-gated overlay list (the builder at `0x4a6ad9` gating on the explored byte), each row's HitRect scaled by 1/1002 and 1/668 against the normalized cursor, edges inclusive, the first hit naming its Area1, then the continent grid re-windowed through the displayed zone's own rect, naming the neighbouring zone or city and nothing for the displayed zone itself — and only the name-only tail is reachable from there, fileName nil and the six coords zero, so the stock frame hides the highlight quad, as the director's own 1.12 shot shows; and `SetVerticalScroll` stores the offset as given (B370, B363): the reference's `0x786db0` compares the new offset against the old and writes it to `[+0x328]` verbatim, `0x787100` re-anchors the scroll child off it, and the range at `[+0x320]` is never read on that path — every clamp the reference exhibits is FrameXML's, through the scroll bar's `[min, max]`; ours clamped into `[0, GetVerticalScrollRange()]`, the child's overflow past the FRAME, which is 40 px short of where `FauxScrollFrame_Update` sends the bar on the 220-tall skills list (three rows lost) and 15 px short on the 287-tall who list (one); the setter fires `OnVerticalScroll` only on a change, as the bytes do, `BenillaScroll_Step` bounds its wheel in Lua where the reference's handler does, and the ScrollFrame kind's prose stops calling its offset unpinned (2017)
- overdraw's cheap half, the sky first: since 0588 every sky element wrote `frag_depth = 0.0`, which costs a pipeline its early-Z, so on an outdoor frame every full-screen dome fragment under the ground and the buildings was shaded in full (the gradient's `asin`, `atan2` and eight table lookups at dawn and dusk) and thrown away; one shared `sky_vertex.wgsl` — bevy's mesh vertex verbatim plus `out.position.z = 0.0` — is now the vertex stage of the gradient dome, the cloud dome, the star patches and the celestial discs and glares, and the model lane's `WOW_SKY_DEPTH` branch moves from its fragment to its vertex, so no sky fragment declares `frag_depth` and the rasterizer rejects the covered screen before the shader runs; every sky `specialize` zeroes the rasterizer bias constant the `sky_order` rung had been leaking into `DepthBiasState::constant`, harmless only while the fragment overwrote its depth, and the prose beside the pin follows it; then the retained `static_gx` pass draws BEFORE bevy's opaque pass (`StartMainPass → static_gx → MainOpaquePass`) — buildings and doodads are the frame's best occluders and a terrain fragment its dearest, six samples a pixel — so walls reject the terrain behind them instead of the reverse, the pass taking the first-use clear with it and the dev depth probe ordered after it; measured on 16 capture scenarios: 14 bit-identical, `northshire-dusk` 1 px and `water-night` 11 px, attributed to exact depth ties on static/terrain silhouettes and last-bit interpolation noise; a depth prepass and near-to-far bins are declined until a journal says where the time goes (2016)
- the grass cut goes per pixel because of a quarter-mip: `0x6813d0` sets `D3DSAMP_MIPMAPLODBIAS = +0.25` on stage 0 (`0x6813f4`) once per frame before the detail-doodad pass — capture-confirmed on all 29 doodad batches, and outside the `[0x6b1000, 0x6b3200)` scan 2004 had called complete; it decides the whole look because the `clutter_cut` probe finds 6 of 31 shipped detail atlases carry a BINARY alpha pyramid (mip 0 has DXT3's 16 levels, every level below exactly two, both starting zones among them), so `a × ramp` degenerates to a global on/off and every texel of a leaf crosses at the same depth — Elwynn's survivor coverage is flat at 47.6 % across the whole 52.5–61.1 yd band, then 0.0 %, while the probe's box-filtered control erodes 47.7 → 20.9 over the same band, so the art has the gradient the shipped pyramid threw away, and `clutter_size` puts the median tuft at 0.92 yd tall, 1.03 yd deep, a full yard for the boundary plane to sweep; the escape is bilinear: full alpha needs a 2×2 all-opaque texel neighbourhood, and those run out between mip 2 and 3 (179/56/10/0/0/0 on the dominant Elwynn cell), so the reference's λ ≈ 2.9–3.3 sits past the cliff and the tuft erodes continuously — 2.2 % of a tuft's pixels leave in the last half-yard with the bias, 11.6 % without, 24.2 % one mip sharper; wgpu has no sampler LOD bias, so the clutter lane re-samples base colour at `view.mip_bias + 0.25`, keeping the sampler every other model batch shares at the reference's 0; ruled out on the way: our BLP decoder (all nine uploaded levels byte-identical against the archive's 52 793 BLP2s), mip regeneration, the `has_mipmaps` byte's value 2 (a plain boolean, an authoring marker), anisotropy and geometry; what it does not fix is resolution-dependent, and so is the reference's own look — λ with the bias is 3.22 at the capture's 648p but 2.48 at 1080p and 1.48 at 2160p, so at a modern resolution the reference itself pops leaf by leaf, and scaling the bias with the viewport is one reversible line left for a look call (2021)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-06 19:35:31 +07:00

373 lines
19 KiB
Rust

//! The host runtime loop — an `impl UiScript` block beside its concern (the `layout.rs` pattern):
//! the event fan-out to registered frames ([`UiScript::fire_event`]), the per-frame advance
//! ([`UiScript::tick`]: OnUpdate + the engine-side fades and the model panes' clocks), and the FrameXML session clock
//! ([`UiScript::now`]). The low-level handler-firing these drive lives in [`super::event`].
use mlua::Lua;
use crate::widget::FrameHandle;
use super::{editbox, event, tooltip, Model, ScriptValue, UiScript};
impl UiScript {
/// Queue an event to fire at the **start of the next tick**, into the same
/// [`super::Model::pending_events`] list the engine's own bindings use — so it lands *after*
/// everything queued earlier in the frame and before that tick's `OnUpdate` pass.
///
/// **Ordering, not laziness, is the whole reason this is public** (decision 1750). A benilla
/// drain runs a step behind the input pass that fed it, so an event it fires *immediately*
/// arrives BEFORE events the same input already queued — the reverse of the reference, where
/// the deferral happens inside the very call that queued them. That inversion is not
/// hypothetical: the soulbind confirm raised by placing an item into a worn slot was cancelled
/// one tick later by that same place's own `CURSOR_UPDATE`, whose `StaticPopup_Hide` is
/// `UIParent.lua:356-360` transcribed correctly. Queueing restores the reference's relative
/// order — cursor-change first, question second — while a *later* cursor change still retires
/// the question, which is exactly what that arm is for.
///
/// Use [`Self::fire_event`] for anything whose ordering against in-flight queued events does
/// not matter, which is nearly everything.
pub fn queue_event(&mut self, event: &str, args: Vec<ScriptValue>) {
self.model_mut()
.pending_events
.push((event.to_string(), args));
}
/// Fire an event to every frame registered for it, invoking their `OnEvent` handlers (RF-0025):
/// both the legacy `this`/`event`/`arg1..argN` globals *and* the modern `(self, event, ...)`
/// arguments. Events reach registered frames regardless of visibility, **in registration
/// order** — the client's `SignalEvent 0x703e50` walks the per-event listener list head-first
/// (tail-append insert `0x7052d0`, wow-re `event-dispatch-order.md`: FIFO, first-registered
/// fires first). Cross-frame order is a LAW consumers depend on: both ZoneText frames write
/// `PVPInfoTextString` on one event — the last writer decides. Handler errors are collected
/// into [`UiScript::errors`], never panicking.
///
/// **The walk steps by a next-listener SAVED BEFORE the handler runs** (`0x703ee8: mov
/// edi,[eax+0x4]` — "save the next node before firing"; decision 1324), never by list index.
/// That is what makes a handler that UNREGISTERS ITSELF mid-dispatch unable to rob its
/// successor — and it is load-bearing in the wild: AceEvent-2.0 treats `PLAYER_LOGIN` /
/// `VARIABLES_LOADED` as fire-once events and unregisters its frame *inside* the handler, so
/// an index walk skipped whichever addon registered right after AceEvent (Bagnon_Forever's
/// saved-bag DB never initialized — the director's `SaveBagData:76` error dialogs). A frame
/// registered mid-dispatch tail-appends and is still visited; a handler that unregisters the
/// walk's SAVED next stops the dispatch there (the reference frees the node and walks into
/// its zeroed links — memory-accident territory we render as a deterministic stop, a knowing
/// divergence from an accident, not from a mechanism).
pub fn fire_event(&mut self, event: &str, args: Vec<ScriptValue>) {
fire_event_into(&self.lua, event, args);
}
}
/// [`UiScript::fire_event`] against the VM directly, for a caller that holds `&Lua` rather than
/// `&UiScript` — which is every Lua binding, and therefore `UpdateSpells` (decision 1924).
///
/// Identical behaviour; `fire_event` is the same call with the script's own VM. The `&mut self` on
/// the method was never load-bearing — the body reaches everything through `lua.app_data_mut()` —
/// and this is the shape [`crate::loader::load_into`] already uses for the same reason (1188 §2).
pub(crate) fn fire_event_into(lua: &Lua, event: &str, args: Vec<ScriptValue>) {
let model_mut = || lua.app_data_mut::<Model>().expect("model app_data set");
{
let mut at = {
let model = model_mut();
model
.event_to_frames
.get(event)
.and_then(|l| l.first().copied())
};
while let Some(h) = at {
let mut model = model_mut();
// The saved handle must still be registered — its removal (by the previous handler)
// ends the walk, per the doc above.
let Some(pos) = model
.event_to_frames
.get(event)
.and_then(|l| l.iter().position(|&x| x == h))
else {
break;
};
let next = model
.event_to_frames
.get(event)
.and_then(|l| l.get(pos + 1).copied());
let id = model.frame_id(h);
drop(model);
if let Err(e) = event::fire_event_handler(lua, id, event, &args) {
model_mut().record_script_error(e.to_string());
}
at = next;
}
}
}
impl super::UiScript {
/// Advance time: run `OnUpdate(self, elapsed)` on every *effectively-visible* frame that has one
/// (RF-0025: OnUpdate → `this` + `arg1 = elapsed`). Errors collected, never panicking.
/// Also advances the `GetTime()` clock — the FrameXML session clock (seconds, monotonic,
/// arbitrary epoch like the real client's), kept as the `__benilla_now` global so the stdlib
/// binding reads it without a host round-trip. Reference FrameXML (CastingBarFrame & co.)
/// anchors durations on GetTime; the clock advancing in the same call that fires OnUpdate
/// keeps the two views of time consistent within a frame.
/// The current `GetTime()` value (`__benilla_now`, seconds) — the FrameXML session clock. The app
/// reads it to stamp an absolute expiry into the clock a Lua countdown reads (the aura feed's
/// `expirationTime`, decision 0257); it is the same value `GetTime()` returns inside the VM.
pub fn now(&self) -> f64 {
self.lua.globals().get("__benilla_now").unwrap_or(0.0)
}
pub fn tick(&mut self, elapsed: f32) {
let clock = {
let g = self.lua.globals();
let now: f64 = g.get("__benilla_now").unwrap_or(0.0);
g.set("__benilla_now", now + f64::from(elapsed))
};
if let Err(e) = clock {
self.push_error(e);
}
// The focused edit box's caret blink (`0x77a790` runs on the client's frame tick).
editbox::tick_blink(&self.lua, elapsed);
// …and, from inside that same `0x77a790` (`0x77a7a1`), the dirty-word drain that fires the
// `OnTextChanged`s an edit only marked (decision 1831). It runs BEFORE this tick's
// OnUpdate sweep, matching the reference's order: the box's own OnUpdate override drains
// before the FrameXML handlers that read the box get their turn.
editbox::drain_text_changed(&self.lua);
// Events queued by Lua bindings last tick (`Model::pending_events` — e.g. `SetMapZoom` →
// `WORLD_MAP_UPDATE`; the cursor arc's `CURSOR_UPDATE`/`ITEM_LOCK_CHANGED`/
// `DELETE_ITEM_CONFIRM`, decision 0216) fire first, so handlers see them before this
// frame's OnUpdate runs.
let pending = std::mem::take(&mut self.model_mut().pending_events);
for (event, args) in pending {
self.fire_event(&event, args);
}
let ids: Vec<u32> = {
let mut model = self.model_mut();
// The OnUpdate population is its own list (decision 1446) — maintained by
// `SetScript`, `scripts`' one writer — so this reads a few hundred handles instead
// of re-filtering the whole scripts map. A destroyed frame's handle stays until its
// liveness check fails once, then compacts here.
let frames: Vec<FrameHandle> = model
.on_update_frames
.iter()
.copied()
.filter(|&h| model.arena.frame(h).is_some_and(|f| f.effective_visible))
.collect();
if model
.on_update_frames
.iter()
.any(|&h| model.arena.frame(h).is_none())
{
let arena = &model.arena;
let live: Vec<FrameHandle> = model
.on_update_frames
.iter()
.copied()
.filter(|&h| arena.frame(h).is_some())
.collect();
model.on_update_frames = live;
}
let mut ids: Vec<u32> = frames.into_iter().map(|h| model.frame_id(h)).collect();
// **Deterministic order, by frame id — i.e. creation order.**
//
// `model.scripts` is a hash map, so this sweep used to fire OnUpdate in an order that
// varied per process. That was invisible for as long as every geometry getter answered
// from the PREVIOUS frame's resolve: each handler saw the same stale world regardless
// of when it ran. The moment the getters began settling on demand (so a handler can
// observe work an earlier handler did this same sweep), the order became load-bearing
// and the macro window's tab width started flipping 164/167 run to run.
//
// Random order was always wrong — the reference sweeps a defined order — the staleness
// was merely hiding it. Creation order is stable, matches FrameXML declaration order,
// and is what a handler reading a sibling's geometry should see.
ids.sort_unstable();
ids
};
for id in ids {
if let Err(e) = event::fire_update_handler(&self.lua, id, elapsed) {
self.push_error(e);
}
}
self.tick_model_panes(elapsed);
// Advance every ScrollingMessageFrame's per-line fade (the client's OnUpdate `0x788460`).
// Independent of the frame's own OnUpdate script — the fade is C++ behavior, not Lua. The
// AtBottom freeze gate lives inside `ScrollingMessageState::tick`.
let mut model = self.model_mut();
// The sibling class's OnUpdate (`0x786200`): the same two-phase fade with no scroll gate,
// plus the capacity law that is this class's stand-in for `maxLines` — the cap is what fits
// vertically, so it needs the frame's resolved rect and is collected first (the arena walk
// below holds a mutable borrow that cannot also read `model.resolved`).
// Both per-kind walks below ride the arena's ticked-kind registry (decision 1446):
// dozens of handles instead of two full-arena sweeps per tick.
let ticked: Vec<FrameHandle> = model.arena.ticked_kinds().to_vec();
let message_frames: Vec<(FrameHandle, usize)> = ticked
.iter()
.copied()
.filter(|&h| {
model
.arena
.frame(h)
.is_some_and(|f| matches!(f.kind_state, crate::widget::KindState::Message(_)))
})
.map(|h| (h, Self::message_viewport_rows(&model, h)))
.collect();
for (h, viewport_rows) in message_frames {
if let Some(crate::widget::KindState::Message(mf)) =
model.arena.frame_mut(h).map(|f| &mut f.kind_state)
{
mf.tick(elapsed);
mf.trim_to_viewport(viewport_rows);
}
}
for &h in &ticked {
let Some(frame) = model.arena.frame_mut(h) else {
continue;
};
if let crate::widget::KindState::ScrollingMessage(smf) = &mut frame.kind_state {
smf.tick(elapsed);
}
}
drop(model);
// Advance fading tooltips (FadeOut's ramp + end-of-ramp hide) — engine behavior like the
// message fade above, decision 0274.
tooltip::tick_fades(&self.lua);
// The hover RE-PICK ([`Model::hover_repick`]): the world under a stationary cursor
// changed this tick (the hovered frame hid, or a frame was shown over the cursor), so
// re-run the hover walk at the SAVED cursor position — the reference's own pump tail
// (`0x765650` → `0x7660d0` with the saved event, didn't-move coalesce bypassed).
// `mouse_move` at the unchanged point is exactly that walk: the hidden frame's OnLeave
// already fired at hide time and `mouseover` is clear, so only the new winner's
// boundary fires — no second OnLeave, `OnEnter` with no physical mouse move.
let repick = {
let mut model = self.model_mut();
let due = model.hover_repick;
model.hover_repick = false;
due.then_some(model.cursor_pos)
};
if let Some((x, y)) = repick {
self.mouse_move(x, y);
}
// `WOW_UI_HANDLERS=<secs>` — who spent the frame (decision 1395). Last, so a report covers
// everything this tick fired; a no-op unless the instrument is armed.
self.report_handler_profile(elapsed);
}
}
impl UiScript {
/// The model panes' per-frame pass (decision 2007; wow-re `modelframe-render-law.md` §4).
///
/// Three things the reference does each frame for a **visible** model pane, in this order:
/// the widget's own `OnUpdate` (`0x76d7f0`, walked by the UI pump with the Lua OnUpdates the
/// caller just fired) advances its private scene clock by `trunc(elapsed · 1000)` ms — a
/// truncation, no `+0.5` (`76d854 __ftol`); the paint (`vt+0x98`, `0x76d1a0`) fires
/// `OnUpdateModel` before it builds the scene; and the scene's animate runs the completion
/// callback (`0x76cdc0`) when the armed sequence has run its length — `OnAnimFinished`, on
/// natural completion only (mode 0), once per arm, for a looping sequence's first pass as
/// much as for a clamped one's end (the flag test sits after the enqueue — wow-re
/// `modelframe-texanim-and-sequence-law.md`, Q4). A hidden pane gets none of the three: its
/// clock stands still and it completes nothing, which is what makes the minimap ping "resume
/// where the last one stopped".
///
/// The completion is read AFTER `OnUpdateModel` because the handler may re-arm (the cooldown
/// flips to its flash there), and a fresh arm has nothing to complete.
fn tick_model_panes(&mut self, elapsed: f32) {
let dt_ms = (elapsed * 1000.0).trunc().max(0.0) as u64;
let update_ids: Vec<u32> = {
let mut model = self.model_mut();
let ticked: Vec<FrameHandle> = model.arena.ticked_kinds().to_vec();
for h in ticked {
let Some(f) = model.arena.frame_mut(h) else {
continue;
};
if !f.effective_visible {
continue;
}
if let crate::widget::KindState::Model(m) = &mut f.kind_state {
m.clock_ms += dt_ms;
}
}
// The handler population, maintained by `SetScript` like `on_update_frames`; a
// destroyed frame's handle compacts out on its first miss.
if model
.on_update_model_frames
.iter()
.any(|&h| model.arena.frame(h).is_none())
{
let arena = &model.arena;
let live: Vec<FrameHandle> = model
.on_update_model_frames
.iter()
.copied()
.filter(|&h| arena.frame(h).is_some())
.collect();
model.on_update_model_frames = live;
}
// Visible MODEL panes WITH A MODEL INSTALLED: the handler is `CSimpleModel::LoadXML`'s
// (`+0x3cc`) and the paint that fires it (`76d1bc`) is reached only past the
// `[widget+0x318] ≠ 0` gate at `76d24c` — a file set, resident or still streaming;
// a pane with no file paints nothing and fires nothing (wow-re
// `modelframe-texanim-and-sequence-law.md`, Q4). A script of that name on any
// other kind is inert, as it is in the reference.
let frames: Vec<FrameHandle> = model
.on_update_model_frames
.iter()
.copied()
.filter(|&h| {
model.arena.frame(h).is_some_and(|f| {
f.effective_visible
&& matches!(&f.kind_state, crate::widget::KindState::Model(m) if m.path.is_some())
})
})
.collect();
let mut ids: Vec<u32> = frames.into_iter().map(|h| model.frame_id(h)).collect();
ids.sort_unstable(); // creation order — the OnUpdate sweep's own law
ids
};
for id in update_ids {
if let Err(e) = event::fire_widget_handler(&self.lua, id, "OnUpdateModel", Vec::new()) {
self.push_error(e);
}
}
let finished_ids: Vec<u32> = {
let mut model = self.model_mut();
let ticked: Vec<FrameHandle> = model.arena.ticked_kinds().to_vec();
let mut due: Vec<FrameHandle> = Vec::new();
for h in ticked {
let Some(f) = model.arena.frame(h) else {
continue;
};
if !f.effective_visible {
continue;
}
let crate::widget::KindState::Model(m) = &f.kind_state else {
continue;
};
let Some(path) = m.path.as_deref() else {
continue;
};
let Some(facts) = model.model_facts.get(&crate::widget::model_key(path)) else {
continue;
};
if m.completion_due(facts) {
due.push(h);
}
}
let mut ids = Vec::with_capacity(due.len());
for h in due {
if let Some(crate::widget::KindState::Model(m)) =
model.arena.frame_mut(h).map(|f| &mut f.kind_state)
{
if let Some(a) = &mut m.armed {
a.finished = true;
}
}
ids.push(model.frame_id(h));
}
ids.sort_unstable();
ids
};
for id in finished_ids {
if let Err(e) = event::fire_widget_handler(&self.lua, id, "OnAnimFinished", Vec::new())
{
self.push_error(e);
}
}
}
}