Add GPUDevice implementation and WebGPULayer

• Component: WebGL
• Status: RESOLVED FIXED

[Dean Jackson]

Foundation of the WebGPU implementation.

[Radar WebKit Bug Importer]

<rdar://problem/30879048>

[Dean Jackson]

Created attachment 303559 [details]
Patch

[Dean Jackson]

Not for review - just EWS. Writing API tests.

[Dean Jackson]

Created attachment 303640 [details]
Patch

[Dean Jackson]

Created attachment 303644 [details]
Patch

[Dean Jackson]

Created attachment 303706 [details]
Patch

[Sam Weinig]

Comment on attachment 303706 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=303706&action=review

> Source/WebCore/platform/graphics/cocoa/GPUDeviceMetal.mm:55
> +#ifndef NDEBUG
> +    [m_layer setName:@"WebGPU Layer"];
> +#endif

Do we really need to make these debug only still?

> Source/WebCore/platform/graphics/cocoa/WebGPULayer.h:33
> +#import <Metal/Metal.h>
> +#import <QuartzCore/QuartzCore.h>
> +
> +namespace WebCore {
> +class GPUDevice;
> +}

This file is not guarded by #if ENABLE(WEBGPU). Should it be?

> Source/WebCore/platform/graphics/cocoa/WebGPULayer.mm:35
> +
> +#import "GPUDevice.h"
> +#import "GraphicsContextCG.h"
> +#import "GraphicsLayer.h"
> +#import <wtf/FastMalloc.h>
> +#import <wtf/RetainPtr.h>
> +
> +using namespace WebCore;

This file is not guarded by #if ENABLE(WEBGPU). Should it be?

> Source/WebCore/platform/graphics/gpu/GPUDevice.cpp:32
> +#import "Logging.h"
> +
> +#if ENABLE(WEBGPU)
> +

When guarded like this, we usually put all the #includes/#imports below the main one, inside the guard.

> Tools/TestWebKitAPI/Tests/WebCore/mac/GPUDevice.mm:33
> +#include "config.h"
> +#import "Test.h"
> +#import <Metal/Metal.h>
> +#import <WebCore/GPUDevice.h>
> +
> +using namespace WebCore;
> +
> +namespace TestWebKitAPI {

This file is not guarded by #if ENABLE(WEBGPU). Should it be?

[Dean Jackson]

Committed r213541: <http://trac.webkit.org/changeset/213541>