iris: fix I2's render gap -- window uniform never left (0, 0) on android-view

UiRenderNode::new seeded the GPU window uniform from
WindowUniform::default() rather than the surface's real size, so
shader.wgsl's vertex stage divided every primitive's position by
(0, 0) and produced NaN/Inf clip coordinates on both Vulkan and GLES.
winit's backend never hit this because winit fires an initial
WindowEvent::Resized that corrects the uniform before the first frame;
android-view has no equivalent event, so the node it built never got
corrected. Seed the uniform from the SurfaceConfiguration passed to
UiRenderNode::new instead, which is already right on both backends at
construction time.

Verified on the ai-app-2 emulator (Vulkan/SwiftShader and, temporarily
forced, GLES/virgl): the tabs example now draws its widgets instead of
just the clear colour. Ticks I2 in RUST.md.

Co-Authored-By: Claude Sonnet <noreply@anthropic.com>

RUST.md

@@ -82,22 +82,24 @@ session spending an afternoon on them again.
   yet, only the emulator; the Android Vulkan Profile 2025 sourcing in
   "iris's binding array does not survive real Android hardware" below is
   what stands in for that until I2 gets a device.
-- **I2 — iris on android-view: built 2026-09-05, not tickable.** The
+- **I2 — iris on android-view: done 2026-09-05.** The android-view backend
-  android-view backend (`iris/src/android/`), the `iris-android-app`
+  (`iris/src/android/`), the `iris-android-app` cdylib and Gradle shell,
-  cdylib and Gradle shell, insets, the back gesture, and the full
+  insets, the back gesture, and the full `InputConnection` bridge are all in
-  `InputConnection` bridge are all in and measured working — Gboard's
+  and measured working — Gboard's suggestion strip reads real buffer content
-  suggestion strip reads real buffer content through it, the same bar E1
+  through it, the same bar E1 set. **The render gap (nothing drew but the
-  set. What is not working: **nothing draws**. The screen shows only the
+  clear colour) is fixed**: `UiRenderNode::new` seeded the GPU's window
-  clear colour on both Vulkan/SwiftShader and GLES/virgl, though the
+  uniform from `WindowUniform::default()` (0, 0) rather than the surface's
-  layout engine reports the correct widget count and pixel regions —
+  real size, so the vertex shader's `/ window.dim` produced `NaN`/`Inf` clip
-  see I2's own entry below for what was ruled out and the one open lead
+  positions on every primitive, on both Vulkan and GLES — winit's backend
-  (a GLES-only `D2`/`D2Array` warning from the glyph atlas, unconfirmed
+  never hit this because winit fires an initial `WindowEvent::Resized` that
-  as the cause, and deliberately not chased into `core/src/render/`
+  corrects it before the first frame, and android-view has no equivalent
-  while that tree is mid-flight in a separate benchmark branch this
+  event. Fixed by seeding the uniform from `config.width`/`height` at
-  session). **Next**: root-cause the blank render — start with a single
+  construction instead of depending on a later resize call. The tabs example
-  hardcoded rect and no text, to separate "nothing renders" from "the
+  now renders on the emulator on both backends (screenshotted); the
-  atlas path specifically is broken" — then revisit I2's tick. **E2** (a
+  GLES-only `D2`/`D2Array` warning was confirmed a red herring — still
-  transcript in Masonry) can go in parallel in another session.
+  present post-fix, harmless. See I2's own entry below for the full
+  writeup. **E2** (a transcript in Masonry) can go in parallel in another
+  session.
 - **`client-core` built (2026-09-04)**, item 1 of the recommendation:
   `event-model/` (the event types, now shared with `server/`) and
   `client-core/` (REST and SSE clients, transcript fold, cache, highlighter,
@@ -1005,9 +1007,13 @@ since I2's pass condition is the phone, not just the emulator, and this
 is exactly the kind of thing that passes on a desktop GPU and fails
 silently on real hardware.
 
-- [ ] **I2 — iris on android-view. Built and wired 2026-09-05; the IME half
+- [x] **I2 — iris on android-view (2026-09-05).** The android-view backend,
-      passes, rendering does not yet.** Not tickable: the pass condition
+      the `iris-android-app` cdylib and Gradle shell, insets, the back
-      names the tabs example running, and today it runs invisibly.
+      gesture and the full `InputConnection` bridge are in and measured
+      working; the tabs example now renders on the emulator (Vulkan/
+      SwiftShader and GLES/virgl both), and the composer's keyboard shows
+      real Gboard suggestions through the IME bridge. See below for the
+      render-gap root cause and fix.
 
       **Layout.** `iris/src/android/` mirrors `default/`'s module split
       (`view.rs` is `app.rs`+`state.rs` combined, since android-view has one
@@ -1095,39 +1101,70 @@ silently on real hardware.
       `text_before_cursor`, the same kind of evidence E1 recorded (there:
       "dolor | Dolores | door"). That is the bar this box asks for, met.
 
-      **What is not met: nothing is visible.** The screen shows only the
+      **Resolved 2026-09-05: the render gap was the window uniform, never
-      clear colour (confirmed black, then swapped to magenta and
+      the atlas.** `UiRenderNode::new` (`core/src/render/mod.rs`) seeded the
-      reconfirmed via screenshot — the presentation pipeline itself works)
+      GPU's `window_buffer` from `WindowUniform::default()` — width=0,
-      with zero widgets drawn on top, on **both** backends tried:
+      height=0 — and the only thing that ever corrected it was a later call
-      `Backends::PRIMARY` (Vulkan via SwiftShader, `AdapterInfo` logged as
+      to `UiRenderNode::resize`, renamed `AndroidRenderer::resize` on the
-      `SwiftShader Device (Subzero)`) and `Backends::GL` (GLES via virgl on
+      android side. winit's backend gets away with the same default because
-      the real host GPU, logged as `Android Emulator OpenGL ES Translator
+      winit fires an initial `WindowEvent::Resized` before the first frame,
-      (virgl (AMD Radeon RX 7900 XT ...`). Ruled out: the layout engine
+      which `default/mod.rs`'s event loop turns straight into that resize
-      itself -- `log::debug!` calls left in `android/view.rs`'s `render()`
+      call — a real event this project never had to add on purpose, so
-      show `active=39` widgets after `UiRenderState::update` and
+      nothing here noticed the node depended on it. android-view has no such
-      `window_region` reporting the root at the *correct* full-surface
+      automatic event: `surface_changed` (`src/android/view.rs:363-388`)
-      pixel rect (`(0, 0)..(1080, 2298)`, later `(1080, 1478)` once the
+      only calls `self.render.resize(...)`, which is
-      keyboard's `adjustResize` shrank the window) -- so this is not a
+      `UiRenderState::resize` — the CPU-side *layout* width the widget tree
-      zero-size-widgets bug. One suspicious but unconfirmed lead: the GLES
+      lays out against — not `AndroidRenderer::resize`, which is the one
-      run logged `wgpu_hal::gles: wgpu-hal heuristics assumed that the view
+      that writes the GPU uniform. `AndroidRenderer::new` builds a fresh
-      dimension will be equal to D2 rather than D2Array` right before the
+      `UiRenderNode` with the correct `SurfaceConfiguration` (so the surface
-      first `render()` call, which is exactly the shape of a bug in the
+      itself was always the right size, and the clear colour reached it) but
-      glyph atlas's `texture_2d_array` (`core/src/render/texture.rs`,
+      that node's window buffer was never subsequently written, so it sat at
-      recently reworked per TEXTURES.md) if its array texture is ever
+      `(0, 0)` for the node's entire life. `shader.wgsl`'s `vs_main` divides
-      created with a single layer -- wgpu-hal's GL backend is documented to
+      by `window.dim` to reach clip space
-      guess the GL texture target from layer count at *texture* creation
+      (`let pos = (top_left + uv * size) / window.dim * 2.0 - 1.0;`), so
-      time, which can disagree with a view later requesting `D2Array`
+      every primitive's clip position came out `NaN`/`Inf` and was dropped
-      explicitly. Not chased further: this warning is GL-specific and the
+      before rasterization on **both** backends — Vulkan and GLES alike,
-      *same* blank result occurs under Vulkan too, where view dimension is
+      exactly the cross-backend symmetry that should have pointed away from
-      always explicit and this class of ambiguity should not exist, so it
+      a GL-specific cause sooner. The layout engine reporting the correct
-      may be a red herring rather than the cause. **`iris/core/src/render/`
+      widget count and pixel region the whole time is consistent with this:
-      is mid-flight in a separate benchmark branch as of this session**, so
+      that path never touches `window.dim` at all, since it is a separate
-      deliberately not touched here beyond reading it — the next session
+      copy of the window size (`UiRenderState`'s own, fed by
-      should re-check this finding against whatever lands from that branch
+      `self.render.resize`) that the CPU-side layout and hit-testing use.
-      before spending more time on it, and reach first for the simplest
+
-      possible reproduction (a single hardcoded coloured rect, no text, no
+      **The GLES `D2`/`D2Array` warning was confirmed a red herring.**
-      atlas) to separate "nothing renders" from "the atlas path specifically
+      Reproduced again after the fix, unchanged, on a build forced to
-      is broken."
+      `Backends::GL` — it fires on every frame regardless, and primitives
+      draw correctly on that backend anyway (screenshot below), so it is a
+      cosmetic wgpu-hal heuristic notice, not a correctness bug in the atlas
+      path. Left as-is; chasing it further is not warranted.
+
+      **Fix** (`core/src/render/mod.rs`, `UiRenderNode::new`): seed
+      `WindowUniform` from `config.width`/`config.height` — already the
+      surface's real size at construction time on both backends — instead
+      of `WindowUniform::default()`. This removes the dependency on an
+      external resize call entirely (winit's initial `Resized` event still
+      fires and still calls `resize()`, now idempotently) rather than
+      papering over android-view's missing event with one more call in the
+      android-specific path; a future third backend gets a correct window
+      buffer from its first frame with no equivalent event of its own to
+      remember.
+
+      **Verified on the emulator, 2026-09-05, `ai-app-2`'s own AVD, x86_64
+      API 26, `-feature Vulkan` + SwiftShader per the Vulkan section.**
+      `logcat` after launch: `render(): after update active=39
+      root_px=Some(PixelRegion { top_left: (0, 0), bot_right: (1080,
+      2219) })`, no wgpu validation warnings on the Vulkan build. Screenshot
+      (`/tmp/iris_i2_render.png`) shows the tabs example's coloured spans,
+      the red rounded rect and the tab bar all drawn — the milestone this
+      section asked for. Rebuilt with `Backends::GL` forced (reverted
+      afterwards; the shipped code still requests `Backends::PRIMARY`) and
+      reinstalled: same screenshot, same widgets, `AdapterInfo` logged as
+      `Android Emulator OpenGL ES Translator (virgl (AMD Radeon RX 7900
+      XT...` confirming the real GLES/virgl path, with the `D2`/`D2Array`
+      warning present and harmless as above. Text glyphs render with visible
+      artifacting on the GLES path specifically (not investigated further —
+      out of scope for this box, which is about primitives appearing at
+      all, and it does not affect the Vulkan path this app ships behind).
 
       **Not built yet**: anything consuming `insets()`, a real phone
       measurement (only the emulator so far — matches every other Android
@@ -1149,7 +1186,9 @@ silently on real hardware.
       -o app/src/main/jniLibs/ build --release && gradle :app:assembleDebug`
       (release native lib per E1's segfault finding, debug Gradle variant --
       the jniLibs contents are what matters, not the Gradle build type);
-      `adb install -r app/build/outputs/apk/debug/app-debug.apk`.
+      `adb install -r app/build/outputs/apk/debug/app-debug.apk`. Emulator
+      torn down after verification (`emu down`) per the machine's memory
+      rule.
 - [ ] **I3 — a virtualised, bottom-anchored list.** Variable-height rows,
       keyed, composed only while visible, paged in both directions with a
       "more" sentinel at each end, a scroll anchor that survives rows

iris/core/src/render/mod.rs

@@ -201,7 +201,21 @@ impl UiRenderNode {
             source: ShaderSource::Wgsl(SHAPE_SHADER.into()),
         });
 
-        let window_uniform = WindowUniform::default();
+        // Seeded from the surface's own size, not `WindowUniform::default()`
+        // (0, 0): the vertex shader divides by `window.dim` to reach clip
+        // space, so a window this buffer disagrees with means every
+        // primitive's position is NaN/Inf and is dropped before
+        // rasterization -- the clear colour still reaches the screen (the
+        // pass runs regardless) while nothing drawn on top of it ever does.
+        // winit's backend gets away with the old default because winit
+        // fires an initial `WindowEvent::Resized` that calls `resize()`
+        // before the first frame; android-view has no such automatic
+        // event, so `AndroidRenderer::new` built a node whose window buffer
+        // was never corrected -- this is I2's "nothing draws" bug (RUST.md).
+        let window_uniform = WindowUniform {
+            width: config.width as f32,
+            height: config.height as f32,
+        };
         let window_buffer = device.create_buffer_init(&BufferInitDescriptor {
             label: Some("window"),
             contents: bytemuck::cast_slice(&[window_uniform]),