[Public WebGL] Signed Int to Float Convertion Problem

Kenneth Russell [email protected]
Tue Aug 15 16:21:26 PDT 2017


Uploaded https://github.com/KhronosGroup/WebGL/pull/2486 documenting the
differences in behavior. Kirill, based on your testing, is the change to
the WebGL 2.0 specification needed, or could that one be left alone (and a
conformance test added)?

Please review. Thanks.

-Ken



On Tue, Aug 15, 2017 at 1:42 PM, Kenneth Russell <[email protected]> wrote:

> This is in the fixed function part of the hardware. There's no possibility
> for emulating the behavior. Since WebGL's hosted on top of a large range of
> OpenGL versions, it's impossible to guarantee OpenGL ES 2.0's behavior in
> WebGL 1.0 contexts.
>
> To be honest since this hasn't come up until now I'm inclined to simply
> mention this in the differences section of the spec and not change the
> spec. Adding a query to determine the underlying behavior would only slow
> down startup time for most WebGL apps and add yet more code to the browser
> for a corner case.
>
> -Ken
>
>
>
> On Tue, Aug 15, 2017 at 12:47 PM, Mark Callow <[email protected]> wrote:
>
>>
>> On Aug 15, 2017, at 12:34, Zhenyao Mo <[email protected]> wrote:
>>
>>
>> But WebGL1 could be implemented on top of ES2 or ES3, so the question
>> is do we want to emulate ES2's behavior for WebGL1 implemented on top
>> of ES3 or desktop GL.
>>
>>
>> Yes and IIRC it is acceptable for ES implementations to return an ES3
>> context even when an ES2 context is requested.
>>
>> I don’t know how practical emulation is.
>>
>> Regards
>>
>>     -Mark
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://khronos.org/pipermail/public_webgl_khronos.org/attachments/20170815/e5593656/attachment.html>


More information about the public_webgl mailing list