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

Re: [Public WebGL] Re: ogles preprocessor conformance test disambiguated factoring



On Tue, Apr 24, 2012 at 2:59 PM, David Sheets <kosmo.zb@gmail.com> wrote:
> On Tue, Apr 24, 2012 at 2:51 PM, Kenneth Russell <kbr@google.com> wrote:
>> On Sun, Apr 22, 2012 at 5:57 PM, David Sheets <kosmo.zb@gmail.com> wrote:
>>>
>>> Here are a few more cases that should be added to preprocess2_frag.frag (valid):
>>>
>>> #define LOC 4 3
>>> #line LOC
>>> #line __LINE__
>>> #line __LINE__ 5
>>> #line __LINE__ __FILE__
>>
>> Hi David,
>>
>> Could you please incorporate the cases above into your patch and
>> regenerate it? It isn't obvious to me exactly what you intended.
>
> Attached.
>
> preprocess2_frag.frag contains all the valid constructions. The new
> preprocessor tests each contain an individual failing line directive
> test case. This ensures that implementations do not reject valid
> shaders and always reject invalid shaders.

Thanks. I've requested that these be incorporated into the ES conformance suite.

-Ken


> Thank you,
>
> David
>
>> I'll ask about incorporating these changes into the ES conformance
>> suite so that they can be more easily pulled down into the WebGL
>> conformance tests.
>>
>> Thanks,
>>
>> -Ken
>>
>>
>>> Thanks,
>>>
>>> David Sheets
>>>
>>> On Sun, Apr 22, 2012 at 5:35 PM, David Sheets <kosmo.zb@gmail.com> wrote:
>>>> Attached you will find an SVN repo unidiff patch to disambiguate the
>>>> various preprocessor errors that can occur in #line directives.
>>>> Specifically, macro expansion in line directives is now properly
>>>> tested. See <http://code.google.com/p/angleproject/issues/detail?id=183>
>>>> for a presently non-conforming implementation.
>>>>
>>>> Here is an accounting of compliant behavior:
>>>>
>>>> preprocess2_frag.frag should now PASS
>>>> preprocess11_frag.frag should FAIL
>>>> preprocess12_frag.frag should FAIL
>>>> preprocess13_frag.frag should FAIL
>>>> preprocess14_frag.frag should FAIL
>>>> preprocess15_frag.frag should FAIL
>>>> preprocess16_frag.frag should FAIL
>>>>
>>>> I would have integrated these tests into the test suite except to
>>>> regenerate the autogenerated test suite you must have a working copy
>>>> of the private OpenGL ES conformance tests so I didn't do this. I
>>>> couldn't do a single iteration of browser testing because of this.
>>>> Sorry, Gregg, Khronos is making you do more work by playing semantic
>>>> games with their "open" standards.
>>>>
>>>> I would like to see these tests used up-stream in the OpenGL ES 2.0
>>>> GLSL WG's (private) test suite. I release my contributions attached
>>>> into the Public Domain.
>>>>
>>>> Failure to properly test for these bugs (and features) and enforce
>>>> conformance will lead to divergence between WebGLSL and OpenGL ES 2.0
>>>> GLSL. Divergence of these languages appears to be inevitable as the
>>>> WebGL WG produces fundamentally public standards and the OpenGL ES 2.0
>>>> WG produces fundamentally private standards (the standards
>>>> specification documents do not match ground truth or the test suites
>>>> and discussion is private).
>>>>
>>>> Warm World Wide Web Wishes,
>>>>
>>>> David Sheets
>>>> 2nd Class Netizen
>>>>
>>>> P.S. When is the WebGLSL specification going to be stored in the
>>>> public WebGL SVN repo? Can the official document be converted into
>>>> HTML like the vast majority of other web standards?
>>>
>>> -----------------------------------------------------------
>>> You are currently subscribed to public_webgl@khronos.org.
>>> To unsubscribe, send an email to majordomo@khronos.org with
>>> the following command in the body of your email:
>>> unsubscribe public_webgl
>>> -----------------------------------------------------------
>>>

-----------------------------------------------------------
You are currently subscribed to public_webgl@khronos.org.
To unsubscribe, send an email to majordomo@khronos.org with
the following command in the body of your email:
unsubscribe public_webgl
-----------------------------------------------------------