gpui-ce v0.3.3 updates/refactors + performance optimizations#4
Draft
muktidaya wants to merge 10 commits into
Draft
gpui-ce v0.3.3 updates/refactors + performance optimizations#4muktidaya wants to merge 10 commits into
muktidaya wants to merge 10 commits into
Conversation
Ensures all windows are repainted after draining the main queue, preventing potential stale frame issues when the application wakes up.
The event loop ran with ControlFlow::Poll and unconditionally called request_redraw() in about_to_wait(), causing the GPU to re-present the same frame at display refresh rate (~60-120 fps) even when nothing had changed. require_presentation: true compounded this by ensuring present() was called every frame, which fired SurfacePresent -> request_redraw() in a self-sustaining loop. Switch to ControlFlow::Wait and remove the unconditional request_redraw() from about_to_wait(). Rendering is now driven by three sources: WakeUp events (sent by external threads via BackgroundExecutor::wake()), the SurfacePresent chain (for consecutive dirty frames), and OS events that flow through cx.notify() -> dispatcher -> WakeUp. Expose BackgroundExecutor::wake() so callers outside gpui (e.g. a tokio watch-channel task) can signal the event loop when new data is ready without polling. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ontrolFlow::Wait Same-thread event processing (mouse scroll, keyboard, clicks) can dirty a window via cx.notify() → invalidate_view, but nothing called request_redraw() afterwards. Under ControlFlow::Wait the event loop sleeps, so the dirty frame was never rendered. Add a request_redraw() call after every window_event dispatch. The on_request_frame handler already gates real work on is_dirty(), so this is a no-op when nothing actually changed — idle CPU stays near 0%. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace StdRng::clone() with StdRng::from_rng() in TestDispatcher::rng, as rand 0.10 removed Clone from StdRng - Add required depth_slice and multiview_mask fields to wgpu RenderPassDescriptor in the wgpu_surface example Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ream consumers of the library: - Add #[allow(dead_code)] to in-progress wgpu rendering infrastructure (VERTEX_ATTRIBUTES constants, bind-group structs, unused fields/methods) that is defined but not yet wired into the active rendering path - Replace overly-narrow cfg_attr on PlatformInputHandler with a plain #[allow(dead_code)], since its methods are platform-specific (x11/wayland) and legitimately unused on macOS builds - Fix irrefutable `if let SurfaceContent::Wgpu` pattern → `let` binding, removing the now-orphaned closing brace - Delete empty after_frame() method that had a TODO questioning its existence - Prefix unused `cx` and `depth_view` variables with `_` in wgpu_surface example
Sync with gpui-ce upstream (de080b0 "Re-re-fork (gpui-ce#23)"): - Add scheduler module with Priority, Scheduler trait, TestScheduler, and executor types for task scheduling and timing - Add PlatformScheduler bridging PlatformDispatcher to Scheduler trait - Replace executor.rs to delegate to scheduler types instead of using PlatformDispatcher directly - Rename gpui_macros -> gpui_ce_macros as local workspace crate in tooling/macros/ - Simplify RunnableVariant from enum to type alias - Update PlatformDispatcher trait (remove label param, RealtimePriority) - Fix queue.rs: probability() -> weight(), Realtime -> RealtimeAudio - Reorganize: fonts to resources/, scripts to scripts/, license update - Add workspace declaration to Cargo.toml with tooling/macros member Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add missing `type Result<T> = T` to AppContext derive impl - Remove `type Result<T>` from VisualContext derive impl (it belongs to AppContext) - Fix `Window::focus` call passing a spurious second argument - Declare `rust_analyzer` as a known cfg in Cargo.toml to suppress unexpected_cfgs warning Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
RedrawRequested is a WindowEvent in winit 0.30+. The unconditional request_redraw() added at the end of window_event by fad7c07 caused every RedrawRequested to immediately queue another, so the event loop never slept under ControlFlow::Wait — recreating the spin loop that a2ed2da had fixed. Return early after invoking on_request_frame so RedrawRequested does not chain itself. All other window events (input, focus, resize, etc.) still fall through to request_redraw() as intended. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
muktidaya
commented
Apr 7, 2026
muktidaya
left a comment
muktidaya
left a comment
Author
There was a problem hiding this comment.
Lot's of (rough WIP) work to bring the repo current with gpui-ce v0.3.3.
Admittedly I've found myself obsessively working on this to the point that this is approaching too much for a single PR.
muktidaya
changed the title
muktidaya
changed the title
Owner
|
I'll mark this as a draft until further updates, aside from that, looks good so far. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Re-request of #3 due to changes to my downstream fork from which the diff is provided.