[Webkit-unassigned] [Bug 237921] New: See if we can get rid of all the WGPU stub structs

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Tue Mar 15 14:30:33 PDT 2022


https://bugs.webkit.org/show_bug.cgi?id=237921

            Bug ID: 237921
           Summary: See if we can get rid of all the WGPU stub structs
           Product: WebKit
           Version: WebKit Nightly Build
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: WebGPU
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: mmaxfield at apple.com

The shared WebGPU.h header has:

typedef struct WGPUDeviceImpl* WGPUDevice;

However, it would be convenient if we could have our own names like this:

namespace WebGPU {
class Device {
    ...
    id<MTLDevice> m_device;
    ...
}
}

I thought of 3 possible solutions:

1. Use reinterpret_cast<>() to freely convert between WGPUDeviceImpl* and WebGPU::Device*. This breaks C++'s type system, and is technically UB (I think) so I was hoping to avoid this option.

2. Don't have nice names, and don't use "namespace WebGPU":

struct WGPUDeviceImpl {
    ...
    id<MTLDevice> m_device;
    ...
}

This is unfortunate because it foregoes C++'s built-in namespace support in favor of prefixes, and isn't the pattern we use for the rest of WebKit

3. Do a double-pointer-indirection:

struct WGPUDeviceImpl {
    Ref<WebGPU::Device> device;
};

This isn't great for performance (though it's probably negligible) and has a somewhat bad smell about having two types which mean the same thing.

For the initial implementation, I chose option 3, under the assumption that it would be easy/mechanical to change later if a better option presents itself.

We should see if there is a better solution.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20220315/0b1cff5e/attachment.htm>


More information about the webkit-unassigned mailing list