[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] WebVulkan

On Fri, Aug 5, 2016 at 9:01 AM Florian Bösch <pyalot@gmail.com> wrote:
The only thing you accomplish is making an API, that's like vulkan, but isn't vulkan, because... reasons.

Because Apple. That's a pretty damn compelling reason.

To be clear, I don't think it's smart to completely discard the existing work that's gone into the modern standards. The point is that we can't do Vulkan in its entirety without explicitly excluding a large chunk of the web. We could do Vulkan but with several functions stripped out (or made into extensions), a different model for memory management, and some limits on shader capabilities. That's not really Vulkan anymore, though, and if we've already made the call to break broad compatibility with the existing API then it's probably worthwhile to see what we can do to make it feel more at home on the web anyway.

The very fact that we'd be in _javascript_ rather than C would allow for some conveniences that could reduce verbosity. For example: Rather than making use of linked lists of structs where you have to re-declare the type of the object you just created in _javascript_ it's perfectly reasonable to just use a single dictionary that describes the object that you want. Less verbose, more webby, doesn't alter the fundamental concept of the API. The question then becomes: Are we still comfortable calling that Vulkan?