Code review - Issue 9697044: Removed checks for zero-length arrays because you can never define such arrays.https://codereview.appspot.com/2013-05-28T23:23:02+00:00rietveld
Message from unknown
2013-05-23T18:37:21+00:00Alok Priyadarshiurn:md5:38008d9a852eb137c27d027cec73f622
Message from alokp@chromium.org
2013-05-23T18:39:30+00:00Alok Priyadarshiurn:md5:5c3d8c2bc1fde8c79b0ff8c54ee2bb43
Message from alokp@chromium.org
2013-05-24T15:45:14+00:00Alok Priyadarshiurn:md5:65406876e6bd3ea04992a9fc98a7f796
Message from zmo@google.com
2013-05-24T17:35:00+00:00zmourn:md5:9c9c210a03b4cba7a0ab2d0bcf069c1b
Same here, nice cleanup, but add Shannon to figure out the bison version issue.
Also, can't seem to open glslang_tab.cpp (although that's less important because that file is generated)
Message from alokp@chromium.org
2013-05-24T18:57:17+00:00Alok Priyadarshiurn:md5:323cbcd0e8116d0ac56149738212059e
Committed patchset #1 manually as r2247 (presubmit successful).
Message from shannonwoods@chromium.org
2013-05-24T19:35:45+00:00Shannon Woodsurn:md5:d98fc662a3a4226b30ee06db7ac202ac
On 2013/05/24 17:35:00, zmo wrote:
> Same here, nice cleanup, but add Shannon to figure out the bison version issue.
>
> Also, can't seem to open glslang_tab.cpp (although that's less important because
> that file is generated)
This is probably the me you wanted to add.
Anything that doesn't break 64-bit safety is fine with me to check in to the trunk. A version of bison that's available on the same system as a 64-bit safe flex would be ideal, because then any codegen conflicts that come up in merges can get resolved without hopping back and forth.
If this changelist applies cleanly against dx11proto, doing so would also prevent some of the version conflict issues.
Message from zmo@google.com
2013-05-24T19:43:49+00:00zmourn:md5:45d88b900d5446795c9d8dfbd3e9149d
Ah sorry, didn't pay enough attention to the wrong email.
On Fri, May 24, 2013 at 12:35 PM, <shannonwoods@chromium.org> wrote:
> On 2013/05/24 17:35:00, zmo wrote:
>>
>> Same here, nice cleanup, but add Shannon to figure out the bison
>
> version issue.
>
>> Also, can't seem to open glslang_tab.cpp (although that's less
>
> important because
>>
>> that file is generated)
>
>
> This is probably the me you wanted to add.
>
> Anything that doesn't break 64-bit safety is fine with me to check in to
> the trunk. A version of bison that's available on the same system as a
> 64-bit safe flex would be ideal, because then any codegen conflicts that
> come up in merges can get resolved without hopping back and forth.
>
> If this changelist applies cleanly against dx11proto, doing so would
> also prevent some of the version conflict issues.
>
> https://codereview.appspot.com/9697044/
Message from kbr@chromium.org
2013-05-28T22:54:08+00:00kbr1urn:md5:4a628ec9026f4324d95264aaab299699
Do we have sufficient regression tests for this change? It isn't clear whether the original fix for https://code.google.com/p/angleproject/issues/detail?id=164 resulted in a WebGL conformance test.
Message from alokp@chromium.org
2013-05-28T23:05:28+00:00Alok Priyadarshiurn:md5:db48a51891b7749d00b3c68ca56703ee
These shaders in WebGL conformance tests have empty arrays:
https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array8_frag.frag
https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array10_frag.frag
https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array11_frag.frag
All of them are supposed to fail compilation but for different reasons. I think it will be good to add a minimal test just for zero-sized array. Is there a bug-tracker for WebGL conformance test?
Message from kbr@chromium.org
2013-05-28T23:06:01+00:00kbr1urn:md5:6528cbda77fe538d1acc0154ba38cb60
On 2013/05/28 23:05:28, Alok Priyadarshi wrote:
> These shaders in WebGL conformance tests have empty arrays:
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array8_frag.frag
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array10_frag.frag
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array11_frag.frag
>
> All of them are supposed to fail compilation but for different reasons. I think
> it will be good to add a minimal test just for zero-sized array. Is there a
> bug-tracker for WebGL conformance test?
Yes: https://github.com/KhronosGroup/WebGL/issues .
Message from alokp@chromium.org
2013-05-28T23:23:02+00:00Alok Priyadarshiurn:md5:5e065fdb6dcae2ff08d99e697bd09b79
On 2013/05/28 23:06:01, kbr1 wrote:
> On 2013/05/28 23:05:28, Alok Priyadarshi wrote:
> > These shaders in WebGL conformance tests have empty arrays:
> >
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array8_frag.frag
> >
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array10_frag.frag
> >
> https://www.khronos.org/registry/webgl/sdk/tests/conformance/ogles/GL/build/array11_frag.frag
> >
> > All of them are supposed to fail compilation but for different reasons. I
> think
> > it will be good to add a minimal test just for zero-sized array. Is there a
> > bug-tracker for WebGL conformance test?
>
> Yes: https://github.com/KhronosGroup/WebGL/issues .
https://github.com/KhronosGroup/WebGL/issues/282