commit 6f0b34abee8dba611c253738d955c59f703c147a
Author: ShrekShao <shrekshao@gmail.com>
Date:   Tue Apr 23 17:40:58 2019 -0700

    Empty glsl fragment shader fixes (#2856)
    
    * drawBuffers(gl.NONE]) for empty es 300 fragment shader
    
    * let es100 shader writes something
    
    * capture error to avoid it gets capture by next glErrorShouldBe
    
    * add a dummy output in fs for deqp test emptyFragmentSource

commit eb0be146e63e743a2c587c9b6d9dbbd73078075e
Author: Kagami Sascha Rosylight <saschanaz@outlook.com>
Date:   Wed Apr 24 09:39:46 2019 +0900

    Split conflicting overloads as WebGLXRenderingContextOverloads (#2859)
    
    * split conflicting overloads as WebGLXRenderingContextOverloads
    
    * lint fix
    
    * defer python changes as requested

commit 233a529f2c414e81902cb3f3b2bdceb1b94cb87a
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Apr 22 17:22:05 2019 -0700

    Fix markup in OVR_multiview2. (#2868)
    
    Follow-on to #2867.

commit 2aef04eaf84f3ac089f3bd74e6df07082ebbadfe
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Apr 22 17:14:15 2019 -0700

    Add rationale for baseViewIndex matching requirement. (#2867)
    
    Also wrap this super long line.

commit 80f6908ac1c086d3577d6eacae802d010e462794
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Apr 22 10:53:51 2019 -0700

    In WEBGL_depth_texture test, check filtering results. (#2864)

commit 90c832eee80b6357ec7f54e88f0416527d808a1c
Author: Yunchao He <yunchao.he@intel.com>
Date:   Tue Apr 16 22:31:37 2019 +0800

    Fix a typo in Copyright for WebGL 2 compute (#2861)

commit 4a170b886d6768e78da2254e7b778e6a3699ff20
Author: ShrekShao <shrekshao@gmail.com>
Date:   Mon Apr 15 13:05:31 2019 -0700

    Add gl_PointSize = 1.0; in vertex-id.html (#2858)

commit 3998a13a8817ec6a1390bd30ff6bd44691875d11
Author: Hao Yunfei <yunfei.hao@intel.com>
Date:   Tue Apr 16 04:04:00 2019 +0800

    Up particle number to 64K (#2860)

commit 11be14a4d63bd3c4d7b7d2b8d35c6466bf337e0a
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Fri Apr 12 00:34:24 2019 -0700

    tex-mipmap-levels.html: more descriptive print messages (#2857)

commit 3f6583d3fee4ab71866ade794504a20eb6f63f88
Author: ShrekShao <shrekshao@gmail.com>
Date:   Wed Apr 10 13:20:42 2019 -0700

    Fix context-attribute-preserve-drawing-buffer.html (#2854)
    
    There's no way to guarantee the compositor finished preparing the frame.
    
    So we are gonna change the test to the following behavior:
    
    Do an infinite for loop for `await nextFrame()` and keep doing readpixels and checking within the loop until we reach some time limit (say 0.5s). For testing `preserveDrawingBuffer: false` we could early terminate if the pixel data matches. And when testing `preserveDrawingBuffer: true` we could assert failure once the pixel doesn't match the expected ones.

commit 034ba2123a3f423ca30565ee43b9708e57c3043b
Author: Yunchao He <yunchao.he@intel.com>
Date:   Wed Apr 10 12:54:04 2019 +0800

    Remove Program Pipeline from WebGL 2.0 compute (#2853)
    
    Program Pipeline is not a core feature for compute shader, it is
    not very important. Furthermore, program pipeline is not very
    useful. We didn't find that some demos or applications rely on it
    and state that it is a must-to-have feature. In addition, we didn't
    see a simple way to implement this feature on Direct3D 11. So we tend
    to remove this feature from WebGL 2.0 compute.

commit 975c77ba724b9c4f3469dd748371cfd089b633b2
Author: ShrekShao <shrekshao@gmail.com>
Date:   Tue Apr 9 15:26:40 2019 -0700

    depth and depth stencil texture completeness parameter setting (#2852)

commit 4e04d185a09deac86e0e276e0f1c79ebe9ba8038
Author: ShrekShao <shrekshao@gmail.com>
Date:   Mon Apr 8 12:29:17 2019 -0700

    Fix rendering-sampling-feedback-loop test for old drivers (#2850)
    
    * fix rendering-sampling-feedback-loop test for driver without WEBGL_draw_buffers

commit de364b2b2ef56fdd40f06e7e48793bfe74654572
Author: ShrekShao <shrekshao@gmail.com>
Date:   Thu Apr 4 19:35:18 2019 -0700

    update depth stencil feedbackloop test behavior (#2849)

commit f56580965dd5ff4057c9e6fc2abc3391e1e69762
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Thu Apr 4 17:16:47 2019 -0700

    Remove OffscreenCanvas commit() requirement (#2848)
    
    The commit() method of OffscreenCanvas contexts was controversial and
    has not been shipped by any browser. Rewrite the OffscreenCanvas tests
    to not require it.
    
    For background, see:
    https://github.com/w3ctag/design-reviews/issues/288
    https://groups.google.com/a/chromium.org/d/topic/blink-dev/hRZ_P2o-aEk/discussion

commit 34c1b95d7e594a9c385b9a60740328f6b2fc1fcd
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Apr 4 16:01:35 2019 -0700

    Add conformance/rendering/blending.html. (#2846)
    
    Also:
    * Add wtu.destroyContext and wtu.destroyAllContexts to facilitate
    cleanup.
    * Add maxArrayDiff and expectArray.

commit efd473349e273e8eef6334b94f280db741311c23
Author: Hao Yunfei <yunfei.hao@intel.com>
Date:   Fri Apr 5 06:32:04 2019 +0800

    Add Fluid example (#2844)
    
    This example is ported from
    https://github.com/walbourn/directx-sdk-samples/tree/master/FluidCS11.
    
    Try this example using latest chrome canary with below commands:
    Chrome.exe --use-cmd-decoder=passthrough --enable-webgl2-compute-context --use-gl=angle

commit ebe94601020f9903d99d8ad3eb0dabf27a3d49e2
Author: Rijubrata Bhaumik <rijubrata.bhaumik@intel.com>
Date:   Fri Apr 5 00:20:15 2019 +0300

    Expose extension for EXT_texture_norm16. (#2837)
    
    * Expose extension for EXT_texture_norm16.
    
    As suggested by @jdashg, that even with incomplete support on mobile,
    this is still probably nice to expose.
    
    @lexaknyazev succinctly summarized the rationale
    in https://github.com/KhronosGroup/WebGL/issues/2810#issuecomment-466387837
    
    1. Integer formats do not support filtering.
    2. Floating point formats with comparable memory consumption provide less precision.
       float16 can represent only about 15K distinct values in 0.0 .. 1.0 range.
       float11 and float10 - fewer than 1K and 500 respectively.
    3. DOM image sources (e.g. PNG) with 16 bits per channel could be copied to GPU
       without intermediate floating point processing
    
    Fixes #2180

commit 31c2f2e5c75471cde4ccdef8dd489989c64e12c3
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Sat Mar 30 13:50:25 2019 -0700

    Replace dispatchTask with awaitable dispatchPromise. (#2840)
    
    Makes it really easy to write code like:
    ```
    async function Test() {
      for (const x of TEST_ARGS) {
        await wtu.dispatchPromise(() => {
          TestOnce(x);
        });
      }
      finishTest();
    }
    Test();
    ```

commit ca07c9628d207b5cc560ee6eac32ceeb1c39a2ed
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Sat Mar 30 13:37:33 2019 -0700

    Add LICENSE.txt. (#2839)
    
    Copyright blocks can now be abbreviated to:
    ```
    Copyright (c) [year] The Khronos Group Inc.
    Use of this source code is governed by an MIT-style license that can be
    found in the LICENSE.txt file.
    ```

commit 291335b884b946118892647041b35ff6c0617217
Author: y-novikov <ynovikov@chromium.org>
Date:   Fri Mar 29 23:04:14 2019 -0400

    Test that vertex attribute is preserved across context switches (#2843)
    
    See http://crbug.com/920033 for rationale for this test.

commit 7e8ca9f9f60f42ef38685c67d1e8cbbcf45f7eb8
Author: ShrekShao <shrekshao@gmail.com>
Date:   Fri Mar 29 12:52:36 2019 -0700

    use highp for vertex id (#2841)

commit 321fb82973fbb39d6760c49465a040ffb9d86543
Author: Vladimír Vondruš <mosra@centrum.cz>
Date:   Tue Mar 26 23:39:15 2019 +0100

    Test compressed texture upload from (a subrange of) a PBO. (#2831)

commit c5d223c457b617ff88ba36d6e67657405f851272
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Mar 25 17:06:03 2019 -0700

    Add "desynchronized" flag to WebGLContextAttributes. (#2753)
    
    Developers of stylus-based drawing applications (e.g. Google Keep)
    have found this hint to be critical in order to be competitive with
    native applications. It has been prototyped in Chromium and yields
    significant latency improvements there.
    
    This is the WebGL analogue of https://github.com/whatwg/html/pull/4360
    and partially addresses https://github.com/whatwg/html/issues/4087 .
    
    Principally authored by @yellowdoge, Miguel Casas-Sanchez <mcasas@chromium.org>

commit 688fbfe33779392aa210d67d4aa12cb012f112c2
Author: ShrekShao <shrekshao@gmail.com>
Date:   Mon Mar 25 15:29:40 2019 -0700

    fix conformance2/extensions/00_test_list.txt (#2836)

commit 9ba538dbf5d55576c541ccd3118ef58bf7a1487e
Author: ShrekShao <shrekshao@gmail.com>
Date:   Fri Mar 22 14:02:29 2019 -0700

    fix test to sync up with spec update #2780 (#2835)

commit cba6ff57169c0a38d2823e93d1ce352ce29377bc
Author: Greggman <github@greggman.com>
Date:   Fri Mar 22 14:01:25 2019 -0700

    experimental-webgl to webgl (#2834)

commit 001bacd966136341c1f3b583150aa93da7071c2c
Author: mingyuhu <mihu@microsoft.com>
Date:   Mon Mar 18 18:08:55 2019 -0700

    multiview rename spec and testfile changes (#2832)

commit bbb96dc57ab5da5c486588675016967c0edee2be
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Mar 18 16:27:57 2019 -0700

    Fix XML for EXT_float_blend extension. (#2833)
    
    Prevented building of the extension registry. Follow-up to #2830.

commit fae1d54fe7d749cb49f4c215c1908e810b650d5f
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Mar 18 15:30:49 2019 -0700

    Implicitly enable EXT_float_blend and promote it to Community Approved. (#2830)
    
    Update tests to require implicit activation accordingly.

commit 2895add3e3388160a39460b3a0ff13b208af8fa0
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Fri Mar 15 17:34:49 2019 -0700

    get-buffer-sub-data-validity.html: regression test for crbug.com/941930

commit ea45aae7d0a2680e6784bc9b7bcaec439195dcb5
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Fri Mar 15 11:59:21 2019 -0700

    get-buffer-sub-data-validity.html: asyncify

commit c4222426da0d8d4a5eaa7125dd2e8f9965436377
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Thu Mar 14 16:43:02 2019 -0700

    Test that sampler objects override texture state (#2823)
    
    A sampler object can override a texture object's TEXTURE_COMPARE_MODE
    state. Chrome was not correctly accounting for this when validating
    texture/sampler compatibility.

commit 3e26ae228bbe7705a292757aebd8d7bb900b7d3f
Author: mbrodesser <48510213+mbrodesser@users.noreply.github.com>
Date:   Wed Mar 13 18:47:10 2019 +0100

    Remove temp_externs/html5.js's references to 'mozHidden' and 'mozVisibilityState' (#2826)

commit 7e1ff7282a418cc4861b547ec1dc08bda65b2acf
Author: Aaron Krajeski <krajeski@gmail.com>
Date:   Mon Mar 11 16:54:51 2019 -0400

    Test webgl canvases as well in projlod test (#2824)
    
    * Test webgl canvases as well in projlod test

commit f9e736fab5eaec878bf81b374143fed112187300
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Fri Mar 8 06:12:01 2019 +0400

    Set gl_PointSize in the vertex shader (#2819)

commit 502339ef1874b9d36b137c456311b207db08d6d2
Author: Greggman <github@greggman.com>
Date:   Thu Mar 7 18:09:52 2019 -0800

    remove unused parts of more tests (#2820)

commit f76c52cb219e57efa91b9611c9caddd403897a8b
Author: Yunchao He <yunchao.he@intel.com>
Date:   Fri Mar 8 10:04:33 2019 +0800

    Add specification for WebGL 2.0 compute (#2769)
    
    * Add specification for WebGL 2.0 compute
    
    * addressed feedback from Jiajie
    
    * Add a few sub-section to clarify undefined behaviors
    
    * Declared draft status and implementations may not shipped
    
    * Move spec status and copyright from W3C to Khronos
    
    * Addressed feedback from Alexey Knyazev
    
    * Addressed one more comment from Alexey Knyazev

commit f6ede11da440afe4f0fa8bdfb420c26f8c2c9593
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Mar 4 17:04:19 2019 -0800

    Update conformance submission readme with new PR process. (#2817)

commit 90f4899357a4537202c8bfda1f53849dbcc39f78
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Mar 4 15:55:13 2019 -0800

    Require each draw buffer with an attachment has a matching frag output type. (#2780)

commit c2419a295f3e4299eb9ade4c845344f8636b0edd
Author: ShrekShao <shrekshao@gmail.com>
Date:   Thu Feb 28 18:07:07 2019 -0800

    add more ext-float-blend tests (#2815)
    
    * add more ext-float-blend tests
    
    * mediump
    
    * style

commit 2ace8df9e4240c14f324abf807401c9893dc1395
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Tue Feb 26 03:55:42 2019 +0400

    Specify DOM image conversion (#2805)
    
    * Specify DOM image conversion
    
    * Address feedback

commit 4c80306d4c044b1a5c6c983d3eb65b89d5d76edb
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Feb 15 16:41:50 2019 -0800

    Add test for gl_VertexID. (#2807)
    
    In response to #2770, tests drawArrays, drawElements, and drawArrays+transform-feedback.

commit 4e2352d99d927bba090b7e744700efc786bceb43
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Feb 15 14:42:33 2019 -0800

    chmod +x server_localhost.py. (no PR)

commit 9b5457dc025b3219e3b1e75f55bac8275e5a59f9
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Feb 14 15:40:28 2019 -0800

    Add default transform feedback object test (#2802)
    
    * Test for using the default transform feedback object.
    
    Note: 2 lines are marked as FIXME and should be removed when/if
    impls and the spec are updated to make `gl.bindBuffer` less
    strict.
    
    * Update to match modern TF behavior.
    
    Also move copyright block to the bottom of the file.

commit ecdb40de223aad8307ee421ce1013c505421265a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Feb 14 15:40:15 2019 -0800

    Add serve_localhost.py for more robust local test runs. (#2808)
    
    This invokes a modified basic http.server that adds anti-caching
    headers. Otherwise, browsers cache test files, making testing and
    debugging less reliable.

commit 8a96bff0a360f65b53357a7b47773d4768e866f4
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Feb 13 22:55:38 2019 -0800

    Add quick mode support to more tests. (#2804)
    
    * Add quick mode support to more tests.
    
    Also spew time elapsed when finishTest() is called from deqp if not in
    quiet mode.
    
    * Forward-port to ToT.

commit f21a352ab16391f866de24cd5611bfdfb897e7fb
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Wed Feb 13 18:08:35 2019 -0800

    Add correctness tests to tex-unpack-params-imagedata.html (#2649)
    
    * Add correctness tests to tex-unpack-params-imagedata.html
    
    Follow-up to https://github.com/KhronosGroup/WebGL/pull/2646
    
    * var->const/let

commit 7fd63b943aa7af1943de5adc0527fb5ff085a317
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Feb 13 17:49:53 2019 -0800

    Backport #2661 (wtu.dispatchTask) to snapshots (#2803)
    
    * Backport #2661 (wtu.dispatchTask) to 2.0.0.
    
    This should improve run times.
    
    * Also backport wtu.dispatchTask to 1.0.3.

commit effc32f0a0a5e06d389e107c84ed27463c06fb8a
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Tue Feb 12 16:41:24 2019 -0800

    integer-cubemap-specification-order-bug.html: add more test cases (#2806)
    
    Makes this test more likely to catch the intended driver bug in the presence of attempted workarounds. (This caught that a few of my attempts at workarounds were not good enough. The eventual fix is (probably) to just preallocate POSITIVE_X, level 0 before everything else.)
    
    These tests were generated by shuffling the initialization order
    thousands of times and picking specifically the cases that failed
    on the affected driver.
    
    crbug.com/905003

commit a0f51b2e123f39c9ff12e621b0b47dd28dd64424
Author: Austin Eng <2154796+austinEng@users.noreply.github.com>
Date:   Thu Feb 7 17:28:00 2019 -0800

    Promote WEBGL_multi_draw and WEBGL_multi_draw_instanced to draft (again) (#2798)

commit 94a5a696a655a192f8b1e1d64aa364b5d41568c4
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Feb 7 13:56:37 2019 -0800

    Add &quick=1 option for deqp/functional/gles3/fbocolorbuffer. (#2792)
    
    * Add QUICK_TEST option for deqp/functional/gles3/fbocolorbuffer.
    * Add &quick=1 option to pages.
    * Add checkbox for toggling on webgl-conformance-tests.html.

commit a94a3ec951c9923bb076490ff6f49027bcf9093e
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Feb 5 18:02:46 2019 -0800

    drawElements with a zero-byte buffer should call with count=0. (#2793)
    
    If validating args, only the arg being validated should have an invalid
    value. Thus, don't call drawElements(mode=-1, count=1) with a zero-byte
    buffer, or you can get either error.

commit f85396bae84af5b701661e68739b8960cb647454
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Tue Feb 5 14:30:39 2019 -0800

    Disable incorrect builtinprecision tests (#2797)
    
    The builtinprecision tests are not correct. Errors were introduced when
    porting the tests from C++ to JavaScript. Tests that fail here pass in
    the C++ version of dEQP ported to WASM. This change disables the tests
    until they can be fixed.

commit dcbb9e5d9e37857d57a9fc711577cf2489ca6197
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Feb 4 19:57:33 2019 -0800

    Backport COMPARE_MODE fix to 2.0.0/deqp/functional/gles3/shadertexturefunction/texturesize.html from #2758. (#2794)

commit ec594f52bc6d5718959b513494d5df4df5b2bb42
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Mon Feb 4 19:57:17 2019 -0800

    Remove skipCorner logic from texImage *_integer-unsigned_byte tests (#2796)
    
    Seems to no longer be necessary to skip these cases/checks!
    Tested on 10.13.6 (Chrome, Firefox) and 10.12.6 (Chrome).

commit 19cc31b495f0bb742b4d1dd6f6e890a9e964d661
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Mon Feb 4 17:56:58 2019 -0800

    Skip deqp derivate.dfdy highp tests on 2.0 ToT (#2795)
    
    (These suppressions were already in 2.0.0, this just adds them to
    top-of-tree 2.0 as well.)

commit 1054b8711dfd58b8bd51fc6a32d4c7c40ca1eb0a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Feb 1 16:32:19 2019 -0800

    Repair 2.0.0 snapshot. (#2790)
    
    conformance-suites/2.0.0/conformance/misc/webgl-specific.html
    conformance-suites/2.0.0/conformance2/rendering/element-index-uint.html
    conformance-suites/2.0.0/conformance/more/conformance/constants.html

commit c4e43f30afb5e9b88b36717f1304ee2af2ea4b60
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Feb 1 12:50:51 2019 -0800

    Add basic test for EXT_float_blend. (#2779)
    
    * Add basic test for EXT_float_blend.
    
    Also test that float32 blending is forbidden without EXT_float_blend.
    
    * Backport EXT_float_blend to WebGL 1.

commit 899ea6fbf1e429be63ff881d396adb170070c732
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Feb 1 12:48:16 2019 -0800

    Allow arbitrary enums for readPixels via IMPLEMENTATION_COLOR_READ_*. (#2776)
    
    * Allow arbitrary enums for readPixels via IMPLEMENTATION_COLOR_READ_*.
    
    This would allow, for example, BGRA readPixels, even though BGRA would
    normally be an INVALID_ENUM for webgl.
    
    * Allow INVALID_OPERATION for potential IMPLEMENTATION_COLOR_READ_*.
    
    * DEPTH_STENCIL and UNSIGNED_INT_24_8 are never valid for ReadPixels.

commit b28d4ee6610dd766d5d554195005a968b1b2353f
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Jan 31 22:44:05 2019 -0800

    Test for A8-on-R8 swizzle being applied to the wrong tex unit. (#2778)
    
    Firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1503813

commit 5ddec66b9433726f42efb2a095ab254edfd9dcb4
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Jan 31 22:42:03 2019 -0800

    Stencil mismatch is only an error if there are active stencil bits. (#2781)
    
    The 1.0.3 version of webgl-specific.html should request stencil:true, and enable(STENCIL_TEST).
    
    Follow-on to #2583.

commit 21f95aa41c58bb21baab4208e562257bad8baaa0
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Thu Jan 31 21:58:43 2019 +0400

    Revision 2 (#2786)

commit 028ad45e934d0bcdeac8f2e34497a4d771503f93
Author: Jie Chen <jie.a.chen@intel.com>
Date:   Thu Jan 31 02:44:42 2019 +0800

    Remove maxShaderCompilerThreadsKHR (#2772) (#2787)

commit 05ee65bb4f9af60614973e20881dacdd4ac62cee
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Wed Jan 30 00:00:37 2019 +0400

    Fix uri in href (#2785)

commit 5ffba0300bc1ad6b8274b764187cbf0e3b855244
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Mon Jan 28 23:51:46 2019 +0400

    Fix XML (#2783)

commit 488af984b533b7cdb07f34b59f959039e16b3b48
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Jan 28 11:51:16 2019 -0800

    Test blitting RGBA8 non-1.0 alpha to a singlesampled RGB backbuffer. (#2782)

commit 91adeb715ae16588b42f7656a3bf0170455b0552
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Mon Jan 28 21:53:03 2019 +0400

    Fix typo in WEBGL_blend_equation_advanced_coherent (#2784)

commit 1a331cc99ea77cbf09b18c067aaf1f3aad3a47f0
Author: Alexey Knyazev <lexa.knyazev@gmail.com>
Date:   Sat Jan 26 01:02:36 2019 +0400

    First draft of EXT_draw_buffers_indexed (#2777)

commit db52df17f0d012983dc281e4864c71485a86bd0e
Author: Jamie Madill <jmadill@chromium.org>
Date:   Tue Jan 22 15:24:23 2019 -0500

    Support robust buffer access for drawArrays. (#2771)
    
    Currently out-of-bounds drawArrays is tested to verify that the implementation
    returns INVALID_OPERATION. The WebGL spec also allows for robust buffer access
    when accessing array buffers out-of-bounds. WebGL already supports robust
    access behaviour for drawElements. Because there are some significant perf
    advantages to robust buffer access for drawArrays as wel this is worth
    supporting.
    
    See https://www.khronos.org/registry/webgl/specs/latest/1.0/#4.5
    Also see http://anglebug.com/3000
    
    A few WebGL 1.0 test changes are neccessary:
    
     * rendering/out-of-bounds-array-buffers: test added.
       * Verifies new behaviour much in the same way as the indexed test.
       * Mirrors out-of-bounds-index-buffers test for drawElements.
     * js/tests/out-of-bounds-test: accept NO_ERROR.
       * Fixes potential TDR with huge size and accepts NO_ERROR.
       * Mirrors f51f2b7120c for drawElements support.
     * more/functions/drawArraysOutOfBounds: test removed.
       * Not relevant when robust access is accepted.
       * Mirrors 714ccda0287 removal of drawElements test.
     * more/functions/drawArrays: removed out of bounds error checks.
       * Not relevant when robust access is accepted.
     * rendering/out-of-bounds-index-buffers: updated message.
    
    Existing implementations shouldn't need any changes to pass the updated tests.

commit 708092d7e7ffb6213c7ba98275d79e83f0c95d41
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Tue Jan 22 11:46:24 2019 -0800

    Skip texture_functions.texturelodoffset.sampler3d_float_vertex (#2774)
    
    It is skipped in native dEQP. See:
    https://android.googlesource.com/platform/external/deqp/+/40ff528%5E%21/
    https://bugs.chromium.org/p/angleproject/issues/detail?id=3094

commit b329259eab7eda0b9c1a0919d1dc162db28c450c
Author: Austin Eng <2154796+austinEng@users.noreply.github.com>
Date:   Tue Jan 22 11:45:12 2019 -0800

    Promote WEBGL_multi_draw and WEBGL_multi_draw_instanced to draft (#2773)

commit 55795a7eadbdd58d8d76f2d5c97ecf029f6a2eab
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Tue Jan 22 11:35:24 2019 -0800

    Runner: semi-optimize layout for smaller screens (#2763)

commit b4b0d6f0d59364bd05125572702552d107d5abbb
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Jan 18 09:46:50 2019 -0800

    Fix webgl2 support for oes-texture-float-linear test. (#2768)

commit b4dc13cac080f7109bfc143b6f2fcefef754411b
Merge: e93e6410 6dd142c5
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Jan 17 16:48:52 2019 -0800

    Merge pull request #2767 from natashalee/master
    
    Use glEnumToString helper for format and type names

commit 6dd142c59f91853b7a8ca45eca6eb18e9e23d6c5
Author: Natasha Lee <natlee@microsoft.com>
Date:   Thu Jan 17 14:56:01 2019 -0800

    Use glEnumToString helper for format and type names

commit e93e6410202b2c3424afe593a8af7d0d835c2149
Merge: 37f9e10f e1571399
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Jan 17 12:14:37 2019 -0800

    Merge pull request #2766 from natashalee/master
    
    Add framebuffer tests to oes-texture-float

commit e1571399c25a22348ccc908d3eca01a533f702ee
Author: Natasha Lee <natlee@microsoft.com>
Date:   Thu Jan 17 11:05:00 2019 -0800

    Remove trailing white spaces

commit 75c3bba3d838334d45c09942868f6f8eccb6c7e5
Author: Natasha Lee <natlee@microsoft.com>
Date:   Wed Jan 16 15:37:41 2019 -0800

    Add framebuffer tests to oes-texture-float

commit 37f9e10fd7e2061cd900dead152a2116c3d71e22
Merge: f22b9ad8 cbf9db27
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Jan 9 18:01:27 2019 -0800

    Merge pull request #2765 from kenrussell/reset-vertex-attrib-binding
    
    Allow vertexAttribPointer to reset binding points' state.

commit cbf9db2787b12d3a8bef9956953b607376899d0c
Author: Kenneth Russell <kbr@chromium.org>
Date:   Tue Jan 8 16:23:00 2019 -0800

    Allow vertexAttribPointer to reset binding points' state.
    
    This was previously done in #2228 but a couple of places were missed.
    
    The specified behavior is already tested by the conformance suite.

commit f22b9ad8d75bf99652dfc3b1db40e73d9fdd3840
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Jan 7 17:53:08 2019 -0800

    Add deleted object test; strengthen transform feedback test. (#2764)
    
    * Add deleted object test; strengthen transform feedback test.
    
    Framebuffers and programs are the primary container objects in the
    WebGL 1.0 specification. Test deleting attachments of both bound and
    unbound framebuffers. Deleting programs and shaders is well tested
    elsewhere.
    
    Strengthen the WebGL 2.0-specific transform feedback test to cover
    deleting output buffers, as well as active transform feedback objects.
    
    Vertex Array Objects (VAOs), the last container type, are already well
    tested by conformance/extensions/oes-vertex-array-object.html and
    conformance2/vertex_arrays/vertex-array-object.html .
    
    Regression test for http://crbug.com/739604 .

commit a2b35635aaef3e9301d69f77f9a0a3fd99291b08
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Tue Dec 18 13:45:02 2018 -0800

    Typo: gl.clear(gl.COLOR_BUFFER) -> gl.clear(gl.COLOR_BUFFER_BIT) (#2760)

commit d0276441b5bf335fef312bfc999fd78868498935
Merge: b59c749a 254388e9
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Dec 18 16:24:40 2018 -0500

    Merge pull request #2759 from kenrussell/change-context-creation-attribs-to-boolean
    
    Change WebGLContextAttributes to boolean from GLboolean.

commit 254388e94568465b75d091f0ab719c0fb89f0fee
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Dec 18 01:56:06 2018 +0000

    Change WebGLContextAttributes to boolean from GLboolean.
    
    Based on review feedback from #2753.

commit b59c749ad949319074d06ead5c678346e2cb7a22
Merge: 283cbc3a 7c957ce3
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Dec 14 17:34:53 2018 -0800

    Merge pull request #2758 from jdarpinian/master
    
    COMPARE_MODE is required to use shadow samplers

commit 7c957ce31901581fcf9d66c3744e0895d57417a3
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Fri Dec 14 17:30:27 2018 -0800

    COMPARE_MODE is required to use shadow samplers
    
    This test uses shadow samplers, so it is required to set
    TEXTURE_COMPARE_MODE to avoid INVALID_OPERATION. We can set it
    unconditionally because it is ignored for non-depth textures.

commit 283cbc3aa76713462f00b10b6f53bfb7b4782c10
Merge: 1576a5be a0dedc3a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Dec 14 17:13:09 2018 -0800

    Merge pull request #2757 from kenrussell/rerun-idl-generator
    
    Re-run IDL generator after OffscreenCanvas changes.

commit a0dedc3a98b170a82ca4c6196da33f3a568ecebc
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Dec 14 16:51:32 2018 -0800

    Re-run IDL generator after OffscreenCanvas changes.
    
    Neglected to do this in #2699.

commit 1576a5bed57437bab4c5bd4a74ba7a3c69a259bf
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Thu Dec 13 23:26:12 2018 -0800

    Test uploading texture from animating transparent WebM (#2755)
    
    Regression test for https://crbug.com/912842

commit 6ff0ec6f39d9d67f0a3d6c16662e8f5dffd815b8
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Dec 13 16:53:13 2018 -0800

    Add more variants of implicit-clear test. (#2756)
    
    Test against all variants of antialias and premultipliedAlpha.
    
    Follow-on to #2751. Addresses additional scenarios from
    http://crbug.com/913033 .

commit 9017b1b69940dfb6e2c3b426743685c0a90233ed
Author: zhenyao <zmo@chromium.org>
Date:   Tue Dec 11 18:01:26 2018 -0800

    Add a regression test for crbug.com/913301 (#2754)
    
    On MacOSX, if we set a large TEXTURE_BASE_LEVEL on an immutable texture, then
    call GenerateMipmap() on it will trigger a driver crash.

commit ab0136705cf603cc0d2aab5686e643220d63fb97
Merge: f35b2483 0c1173e5
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Dec 10 13:54:59 2018 -0800

    Merge pull request #2750 from jdarpinian/texture-type
    
    Test shadow samplers and TEXTURE_COMPARE_MODE

commit f35b24831e318d3477f10c28b77b9a4314351292
Merge: ce08b9be eff1352e
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Dec 10 11:38:35 2018 -0800

    Merge pull request #2752 from null77/amend-multiview-xfb-test
    
    Update multiview transform feedback validation.

commit ce08b9bede2de2f3cad9ef2603078d86bdbaf71e
Merge: 5f234540 4e3fac2a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Dec 10 11:37:33 2018 -0800

    Merge pull request #2748 from jchen10/parallel_benchmark
    
    Add KHR_parallel_shader_compile benchmark

commit eff1352eccf4883776ff63c08002cef43c1be25f
Author: Jamie Madill <jmadill@chromium.org>
Date:   Sun Dec 9 10:16:42 2018 -0500

    Update multiview transform feedback validation.
    
    This updates the CTS to be in line with the recent extension wording
    change. It allows multiview rendering when Transform Feedback is bound
    and paused.
    
    See http://anglebug.com/3012

commit 5f234540c41911b1b4e85d53195032efe9060512
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Dec 6 18:59:35 2018 -0800

    Color mask must be preserved during implicit clears. (#2751)
    
    * Color mask must be preserved during implicit clears.
    
    Catches a bug that manifested on Chrome on macOS, but could have
    affected any other platform. This new test already passes on Firefox.
    
    Regression test for http://crbug.com/911918 .

commit 0c1173e509ccbc3a1135f86ea6f3b3e4757bb96c
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Wed Dec 5 17:52:19 2018 -0800

    Test shadow samplers and TEXTURE_COMPARE_MODE
    
    Setting TEXTURE_COMPARE_MODE should cause an error when depth textures
    are used with float samplers. Shadow samplers should accept only depth
    textures and only when TEXTURE_COMPARE_MODE is set.

commit 4e3fac2ad2887ef1b563ae9b0e463d16a6a84383
Author: jchen10 <jie.a.chen@intel.com>
Date:   Thu Nov 29 16:12:17 2018 +0800

    Add KHR_parallel_shader_compile benchmark
    
    Chromium bug: http://crbug.com/849576

commit 620ae9dbdff6320f7f10a8bbd62c086ec3765e73
Merge: 5b6cbd78 c9fd3485
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Nov 29 13:46:39 2018 -0800

    Merge pull request #2746 from null77/multiview-xfb-paused
    
    Amend ANGLE_multiview to allow for paused XFB.

commit c9fd34859b72939c68a8ccba7a049911f108c855
Author: Jamie Madill <jmadill@chromium.org>
Date:   Thu Nov 29 16:10:00 2018 -0500

    Amend ANGLE_multiview to allow for paused XFB.
    
    Rationale: a paused XFB object is essentially invisible to the
    rendering engine. The base ES spec says: "When transform feedback
    is paused, no vertices are recorded." Moreover, many operations
    that fail when an XFB is active are allowed when it is also paused.
    This change makes the multiview behaviour consistent with the rest
    of the spec.

commit 5b6cbd789b9b91b4e46dde883c9f2ecb31eddade
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Nov 15 15:20:15 2018 -0800

    Run canvas-teximage-after-multiple-drawimages test in onload handler. (#2744)
    
    The JavaScript execution was racing with document parsing, causing the
    test to fail intermittently. Follow-on to #2706.
    
    Chromium bug: http://crbug.com/905682 .

commit 5e45a4931159424e818fce4a5fd662967f9589ab
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Nov 13 23:51:13 2018 -0800

    Cleanup in framebuffer-texture-changing-base-level. (#2738)
    
    Also check that each level in turn is actually cleared to green.
    Overall failure is still present in all browsers.

commit 61846ed93959110eabf31151862efc4b3ec56b97
Merge: 25bf3474 23ee3cc9
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Nov 12 19:28:28 2018 -0800

    Merge pull request #2735 from austinEng/master
    
    Replace proposal EXT_multi_draw_arrays with WEBGL_multi_draw and WEBGL_multi_draw_instanced (#2717)

commit 25bf347425439e422b1355f9d7ed0bc8f20313ee
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Nov 9 17:37:47 2018 -0800

    Clear background of 2D canvas as originally intended. (#2743)
    
    * Clear background of 2D canvas as originally intended.
    
    Another small follow-on to #2741 .

commit b01245d5b14402b7e768fa5ae96bf5636e1c495a
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Nov 9 16:42:34 2018 -0800

    Clean up unused variable. (#2742)
    
    Small follow-up to #2741.

commit da239df8decc27d4b61f49efe51906a585b22b5b
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Nov 9 12:00:53 2018 -0800

    Test repeated rendering of WebGL canvas to 2D canvas. (#2741)
    
    Each draw should show only the most recently rendered content from the
    WebGL canvas. This was not the case when rendering via SwiftShader.
    
    Regression test for http://crbug.com/894021 .

commit 23ee3cc9b4189e60dd4ace7624e1144db3cef150
Author: Austin Eng <austineng.inr@gmail.com>
Date:   Fri Nov 9 10:11:53 2018 -0800

    Add tests for WEBGL_multi_draw and WEBGL_multi_draw_instanced
    
    These tests share much of the same functionality. They are
    tested in the same file.

commit a86bbf69e08e0c45d35972b8dd95f17f9c176fc0
Author: Austin Eng <austineng.inr@gmail.com>
Date:   Fri Nov 9 10:11:38 2018 -0800

    Add proposals WEBGL_multi_draw and WEBGL_multi_draw_instanced (#2717)
    
    Based on previous proposal EXT_multi_draw_arrays.
    Add additional entry points for Instanced draw call variants.
    Add additional builtin gl_DrawID.

commit 8dbc733322a4e75e6246a82bd2930f296fb53c36
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Nov 7 20:18:48 2018 -0800

    Promote BPTC and RGTC to Community Approved. (#2739)

commit 07072d2b78157c011d00656820128f624435b1fd
Merge: c987f075 f0f73e24
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Nov 6 23:13:48 2018 -0800

    Merge pull request #2740 from KhronosGroup/kainino0x-patch-1
    
    Deduplicate getDefault3DContextVersion logic

commit f0f73e244ebb3c74b32716800fc115031753e8bd
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Tue Nov 6 15:08:37 2018 -0800

    Deduplicate getDefault3DContextVersion logic

commit c987f075bfdca44119175c41f547d849c08983a3
Author: He Yunchao <yunchao.he@intel.com>
Date:   Fri Nov 2 03:09:54 2018 +0800

    Don't skip deqp tests in WebGL 2.0.1 if the corresponding tests (#2307)
    
    are not driver/hw bugs in c++ deqp

commit f43cafe974d84690e53e03cf43e62aa2a16bcdc2
Author: Ilia Mirkin <imirkin@alum.mit.edu>
Date:   Fri Oct 26 21:28:35 2018 -0400

    Clarify texStorage*D initialization for compressed formats (#2737)
    
    The reference is made to ES 3.0 section 3.8.4 pseudocode, which in
    turn is prefaced with
    
      If internalformat is a compressed texture format, then references to
      TexImage* should be replaced by CompressedTexImage*, with format, type
      and data replaced by any valid imageSize and data.
    
    Include a clarification to this effect directly in the WebGL spec to
    avoid requiring the reader to dig through to see if online compression
    of a "zero" color is implied.
    
    This is a copy of commit d0612c40c67 for the latest file.

commit bb64ef4e3c11b945e649d903113e3bdc9e25331d
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Oct 24 14:10:50 2018 -0700

    Follow up fixes for lifetime changes from 2731. (#2736)
    
    Test that isBuffer always returns false after deletion.

commit 22174ac6ee372bd5834bf35bc2172f43bd24980c
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 23 15:28:47 2018 -0700

    Changes to conformance2/textures/misc/copy-texture-image.html (#2226)
    
    * Sized formats can still drop components. (e.g. RGBA->RGB)
    * *_SNORM formats are INVALID_ENUM for CopyTex*Image*.
    * Depth formats may generate INVALID_OP in addition to INVALID_ENUM.

commit 0d55c887e92b645f6effe753528323ab2ffd94c2
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Oct 22 18:06:51 2018 -0700

    Split up shaderoperator/common_functions.html. (#2733)
    
    This test suite was taking close to the maximum timeout in Chromium's
    CI system.
    
    Addresses http://crbug.com/820225 .

commit 9d1b42e87d22ac323c49f525f5ae409d3c5227b2
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Fri Oct 19 16:15:34 2018 -0700

    Mute all videos so they don't get caught by autoplay blocking. (#2732)

commit cfdbb91e930e130829dd6d3e7395e8e3ab68813d
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Oct 19 14:14:15 2018 -0700

    *TexSubImage on an undefined texture image is INVALID_OPERATION. (#2730)
    
    Since this can happen before we check all our enums, this subtest
    should accept either error.
    
    (Regression from #1320)

commit abce2ab39eafdb0a39466117a0e111a6a9effde5
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Oct 17 15:39:46 2018 -0700

    Add EXT_multi_draw_arrays extension proposal. (#2717)
    
    * Add EXT_multi_draw_arrays extension proposal.
    
    Upon request from multiple CAD customers.

commit 83dee100808acef87189fe65c69ef0ea6c5130be
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Oct 17 15:23:02 2018 -0700

    DEQP lifetimes test should detach (not delete) buffer from VAO. (#2731)
    
    The spec used to forbid detaching using vertexAttribPointer.
    Also remove keep-alive requirement in
    conformance/extensions/oes-vertex-array-object.html.

commit da3ffa69c24132648de4f1b13bef1007a0065422
Author: Tarek Sherif <tareksherif@protonmail.com>
Date:   Tue Oct 16 21:23:28 2018 -0400

    Dirty flag tests (#2600)
    
    * WebGL 1 Disabled vertex attribute update test
    
    * Add to test list
    
    * Vertex buffer update after draw test
    
    * WebGL 2 versions of dirty flag tests
    
    * VAO disabled attributes test
    
    * Linted
    
    * Remove rendundant WebGL 2 ports
    
    * Fix comments

commit 7ea36c1f110392d4bfb59007eb943d10fd30ea30
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Tue Oct 16 17:23:16 2018 -0700

    Test readPixels from multisampled framebuffers. (#2729)
    
    The readPixels tests were all being run with {antialias: false}. There's
    a bug in Qualcomm's Adreno drivers (http://crbug.com/890002) that only
    manifests with {antialias: true}. This change tests both configurations.

commit 5454f9aaeedc833de525135bad36d899ca0dd240
Merge: 9e95241e 286cb63c
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 16 15:01:20 2018 -0700

    Merge pull request #1320 from jdashg/integrate-webplat
    
    Integrate tests from web-platform-tests.

commit 286cb63c5d1944d568f5c4c4a3f97bd6e7d59790
Author: jdashg <jdashg@gmail.com>
Date:   Thu Nov 19 16:50:22 2015 -0800

    Integrate tests from web-platform-tests.

commit 9e95241ef80215d2770d907156c243f335fbbdba
Merge: 1da2c306 ec309215
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 16 13:55:50 2018 -0700

    Merge pull request #2726 from imirkin/master
    
    ASTC all-zeroes data decodes as error color

commit 1da2c306a632b8f09022b19c16d1b4d10a05d4e7
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 16 13:54:14 2018 -0700

    DrawBuffers has no effect on sampling feedback. (#2727)
    
    * DrawBuffers has no effect on sampling feedback.
    
    Incomplete textures should be ignored by feedback detection also.
    
    * Combine webgl 1 and 2 tests for rendering/sampling feedback loops.
    
    * Review fixes.

commit 96b30d21d102e4b2dd60e05ec56b9c0612857838
Merge: 26fe7d28 09f8f330
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 16 13:47:07 2018 -0700

    Merge pull request #2724 from jdashg/zero-depth-resources
    
    Clarify that all depth resources are cleared to 1.0 by default.

commit 26fe7d28db303981be59786bfab9fd294e273dfe
Merge: ac1664bb bfbe124a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Oct 16 13:46:38 2018 -0700

    Merge pull request #2728 from jdashg/no-zero-gen-mipmap
    
    Forbid GenerateMipmap from operating on zero-size textures.

commit bfbe124a7bb92eed34ea7cc18694167ed66b1030
Author: Jeff Gilbert <jgilbert@mozilla.com>
Date:   Wed Oct 10 17:53:07 2018 -0700

    Forbid GenerateMipmap from operating on zero-size textures.

commit ac1664bb9cb4a2191f5d01fe4f47c8f00d5b8952
Author: Khronos Group Webmaster <webmaster@khronos.org>
Date:   Wed Oct 10 15:43:50 2018 -0400

    Create CODE_OF_CONDUCT.md

commit ec3092155258c9c1b9f6596d3d3b58eba7d30c3a
Author: Ilia Mirkin <imirkin@alum.mit.edu>
Date:   Sun Oct 7 19:25:48 2018 -0400

    Expect error color after texStorage2D initialization of an ASTC texture
    
    A block with all zeroes is invalid, so sampling from it will generate
    the magenta error color. As per the GL_KHR_texture_compression_astc_*
    specification, the block mode, which is the first 10 bits, is marked as
    reserved when the low nibble is zero. (See Table C.2.8.)
    
    Fixes #2725.

commit d0612c40c67ab86bc6d6b0e7cc0d89c871d441b0
Author: Ilia Mirkin <imirkin@alum.mit.edu>
Date:   Sun Oct 7 19:16:26 2018 -0400

    Clarify texStorage*D initialization for compressed formats
    
    The reference is made to ES 3.0 section 3.8.4 pseudocode, which in
    turn is prefaced with
    
      If internalformat is a compressed texture format, then references to
      TexImage* should be replaced by CompressedTexImage*, with format, type
      and data replaced by any valid imageSize and data.
    
    Include a clarification to this effect directly in the WebGL spec to
    avoid requiring the reader to dig through to see if online compression
    of a "zero" color is implied.

commit 78e2fa62ac3f061bfcc7feeec98dad4ad103ca59
Merge: 816dbcc8 2bb8be26
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Oct 4 13:53:33 2018 -0700

    Merge pull request #2721 from jdashg/ext-bptc
    
    Rename BPTC and RGTC extensions to EXT_.

commit 816dbcc84a68946e0bd4680490c4526862d5add2
Author: Jiawei Shao <jiawei.shao@intel.com>
Date:   Thu Oct 4 08:47:16 2018 +0800

    Add test to expose an alpha blending issue on Intel macOS drivers (#2716)
    
    * Add test to expose an alpha blending issue on Intel macOS drivers
    
    This patch intends to add a conformance test to expose an alpha
    blending issue on Chromium on Intel macOS drivers. The test is
    based on the case provided in https://crbug.com/886970.

commit 09f8f3302013a518b35f3d46e7628a21ac3d404e
Author: Jeff Gilbert <jgilbert@mozilla.com>
Date:   Wed Oct 3 16:43:01 2018 -0700

    Clarify that all depth resources are cleared to 1.0 by default.
    
    Not just renderbuffers.

commit 2bb8be266c718add7e2ccb262c4733fe00ad19f2
Author: Jeff Gilbert <jgilbert@mozilla.com>
Date:   Fri Sep 28 12:59:21 2018 -0700

    Rename BPTC and RGTC extensions to EXT_.
    
    Also retarget BPTC from the ARB_ ext to the EXT_ one, since EXT_ is the
    ES version.

commit c560071cac9aef0b26f089c4ead84b5b9e39c368
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Sep 27 03:29:23 2018 +0300

    Fix WEBGL_compressed_texture_bptc extension number (#2719)

commit 42c423dc6fcda34394439caf55f860aae97bc7a3
Merge: f894e92c 2d6bddbe
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Sep 26 10:15:15 2018 -0700

    Merge pull request #2720 from Oletus/ext-rgtc
    
    Add a draft RGTC compressed texture extension

commit 2d6bddbeaca9d22492940ab3f8a96c060ef65c81
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Thu Sep 13 20:54:01 2018 +0300

    Add a draft RGTC compressed texture extension
    
    The extension simply exposes the native compressed texture extension
    EXT_compressed_texture_rgtc. The extension contains red and red-green
    compressed texture formats.
    
    The extension should be supportable on both OpenGL and D3D11. In D3D11
    the compressed texture formats exposed here are called BC4 and BC5.
    D3D11 also has signed and unsigned variants of these formats.
    
    The red-green formats from this extension are particularly suited for
    storing normal maps.
    
    The native extension requires OpenGL ES 3.0, but the WebGL extension
    is intended to be exposed starting from WebGL 1.0. The native
    extension does specify sampling from the missing channels and
    explicitly mentions that it doesn't add uncompressed red-green
    formats, so it's reasonably straightforward to interpret against the
    OpenGL ES 2.0 / WebGL 1.0 spec.
    
    The included test verifies that the enum values are correct and that
    the size restrictions are respected, but does not cover decoding yet.

commit f894e92ce3056919ea525ede2c7978d016098c1f
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Mon Sep 24 08:50:10 2018 +0300

    Add a test for instanced rendering with large divisors (#2715)
    
    This test covers a recently fixed ANGLE bug http://anglebug.com/2832
    that was present on the D3D11 backend.

commit c797a4f79589ed3cddd5f06fa0ea553d369ab292
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Sep 21 20:09:38 2018 +0300

    Add a test for WEBGL_compressed_texture_bptc (#2714)
    
    The test is not very thorough yet, as it's not testing for correct
    decoding of the compressed data. However, more complete tests will be
    easier to develop once there is a prototype implementation available.
    
    The extension is moved to draft so that prototyping can proceed.

commit 105938fa0ad3caa2a34502e4e3e724aea2e322e2
Author: Ashley (Scirra) <ashley@scirra.com>
Date:   Thu Sep 20 19:07:15 2018 +0100

    Delete WebGL_blend_equation_advanced proposal (#2713)
    
    Delete WebGL_blend_equation_advanced in favour of the new WEBGL_blend_equation_advanced_coherent proposal.

commit 8061947f20c994a605dc18cf16a640fb97986be6
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Sep 20 21:06:09 2018 +0300

    Improve compressed texture tests (#2712)
    
    * Improve compressed texture tests
    
    A new test utility collection for compressed texture tests is added in
    sdk/tests/js/tests/compressed-texture-utils.js. It can validate common
    functionality in compressed texture extensions, such as enumerating
    formats and size restrictions on data that is passed in to texture
    uploads.
    
    This also removes duplicate testing of querying
    COMPRESSED_TEXTURE_FORMATS when no texture compression extensions are
    enabled. This is tested in textures/misc/compressed-tex-image.html so
    it doesn't need to be tested in individual compressed texture
    extension tests. compressed-tex-image.html is now enabled also for
    WebGL 2.0.
    
    New coverage is added for trying to use all the different compressed
    texture formats when the extensions are disabled.
    
    Coverage is added also for trying to use more varied incorrect
    dimensions for ASTC textures.
    
    * Add a conformance test for WEBGL_compressed_texture_etc1
    
    This extension is enabled in some WebGL implementations but has not
    had proper conformance tests until now. Add a simple conformance test
    with some enum tests and negative tests to have at least some
    coverage.

commit be85b36242a47a49820b193c233e6a6cdaba00ac
Author: Ashley (Scirra) <ashley@scirra.com>
Date:   Fri Sep 14 19:05:28 2018 +0100

    Propose WEBGL_blend_equation_advanced_coherent (#2709)
    
    Proposal to expose KHR_blend_equation_advanced_coherent to WebGL. Forked from the previous proposal for KHR_blend_equation_advanced according to #2704.

commit a50f779ed49d33f4d10c2868105565162b7133ba
Author: Jie Chen <jie.a.chen@intel.com>
Date:   Sat Sep 15 02:03:41 2018 +0800

    Promote KHR_parallel_shader_compile to draft. (#2690) (#2711)

commit 324f1e21f1c07525c8bf069a2bc76a43464d6787
Author: Jiajia Qin <jiajia.qin@intel.com>
Date:   Sat Sep 15 01:52:04 2018 +0800

    Add ComputeBoids example (#2707)
    
    This example is ported from
    https://github.com/google/nxt-standalone/blob/master/examples/ComputeBoids.cpp.
    
    Try this example using latest chrome canary with below commands:
    Chrome.exe --use-cmd-decoder=passthrough --enable-webgl2-compute-context
               --use-angle=gl  --use-gl=angle
    
    Currently, it still can't get correct result on d3d backend since SSBO
    for d3d is still under development in ANGLE.

commit f88e25f549a4b764674f6aeb415772fb912ea66a
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Sep 14 15:01:07 2018 +0300

    Add a proposal for a BPTC compressed texture extension (#2710)
    
    The proposed extension simply exposes the native compressed texture
    extension ARB_texture_compression_bptc.
    
    The extension should be supportable on both OpenGL and D3D11. In D3D11
    the compressed texture formats exposed here are called BC7 and BC6H.
    D3D11 also has signed and unsigned variants of the float format and
    linear and SRGB variants of the fixed point format.
    
    The extension is possible to support on both WebGL 1.0 and 2.0.

commit 6d2f3f4cb8bac1f7c4a945c73d07a33df74f22f9
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Sep 13 13:46:00 2018 -0700

    Add canvas-teximage-after-multiple-drawimages.html test. (#2706)
    
    * Add test case uploading a 2D-rendered canvas which has had a series of
    drawImage operations (from both images and other 2D-rendered canvases)
    applied to it.
    
    Regression test for http://crbug.com/878545 .

commit a24c704e2f2e0056acdaa057b18359f6151c46cd
Merge: 05c20036 3298307b
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Sep 11 17:34:13 2018 -0700

    Merge pull request #2708 from nox/named-getter
    
    Avoid using window's named getter on a couple of tests

commit 3298307b35efbde78beb9832cf0a00c22fb0f190
Author: Anthony Ramine <n.oxyde@gmail.com>
Date:   Wed Sep 12 01:14:39 2018 +0200

    Avoid using window's named getter on a couple of tests
    
    Servo doesn't implement the named getter yet and thus those tests fail for
    reasons unrelated to our WebGL support.

commit 05c200369ee11b96f6794cff556b488d7c23113e
Merge: 3aa718b8 4c4bcdef
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Aug 31 13:09:00 2018 -0700

    Merge pull request #2703 from kenrussell/better-specify-vertexattrib-v
    
    Better specify errors for vertexAttrib*v entry points.

commit 3aa718b8aad395d84b8954e35ca453fbc6dc6c8e
Author: Anthony Ramine <n.oxyde@gmail.com>
Date:   Fri Aug 31 21:52:39 2018 +0200

    Make some tests fail more gracefully and cooperatively. (#2702)
    
    Those tests time out when run in Servo because of missing support for a
    bunch of features, such as preserving the drawing buffer or loading SVG
    images. Instead, make the tests fail, and finish the tests so that the
    next ones are started.

commit 4c4bcdef3b0d53ba1c449e6280e4cdcfeb2f4390
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Aug 31 11:56:53 2018 -0700

    Better specify errors for vertexAttrib*v entry points.
    
    This behavior is already tested by gl-vertex-attrib.html.
    
    Fixes #2700.

commit 9a64910fc61fa3c71e6d055416a489bd35a6ea18
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Aug 30 15:31:46 2018 -0700

    Add references to OffscreenCanvas. (#2699)
    
    OffscreenCanvas is documented in the WHATWG's current specifications,
    and browsers are beginning to ship support for fetching WebGL contexts
    against them, as well as use them as texture sources. Document this
    properly in the WebGL specification.
    
    In Chromium, this work was tracked under issue http://crbug.com/563816
    and many dependent bugs.

commit fba3728a1bd6cbe027f7f7a5be3a40ed1bb69727
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Aug 30 14:45:52 2018 -0700

    Test webgl2 vs. webgl canvas context modes. (#2701)
    
    This test covers the following HTML spec pull request:
    https://github.com/whatwg/html/pull/3971

commit 296a72447f297df4a75a0cb9b59d4f3f931e271e
Author: Ashley (Scirra) <ashley@scirra.com>
Date:   Thu Aug 30 02:35:57 2018 +0100

    Propose WEBGL_blend_equation_advanced (#2697)
    
    * Propose WEBGL_blend_equation_advanced
    
    Proposal to expose KHR_blend_equation_advanced to WebGL.

commit 40e4ce679cde38632341404e39d2471dd011c216
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Aug 22 09:59:25 2018 -0700

    Fix href of rejected WEBGL_get_buffer_sub_data_async extension. (#2698)
    
    This was causing the extension registry to fail to rebuild
    successfully.

commit 5fed0610cd7c5212f9646bf5dea211c5f5fd8baa
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Tue Aug 14 10:40:59 2018 -0700

    Document non-blocking getBufferSubData (#2696)

commit 1fae933adbaf340636c3bdec8a074bbd9341e9ca
Author: Jie Chen <jie.a.chen@intel.com>
Date:   Fri Aug 10 08:18:26 2018 +0800

    Expose KHR_parallel_shader_compile. (#2690) (#2691)
    
    Add the extension proposal.

commit 9d810e5f222d29e388e97020d1616ecced53c992
Merge: 7ca87fb1 8bc2c423
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Aug 9 14:17:05 2018 -0700

    Merge pull request #2695 from Oletus/more-webgl-multiview-tests
    
    Add WEBGL_multiview tests for depth buffer and transform feedback

commit 8bc2c423dfec6de53da1254a71a05cb777c6d5be
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Thu Aug 9 14:10:34 2018 +0300

    Add a WEBGL_multiview test for interaction with transform feedback
    
    The test passes on the latest prototype implementation in Chromium.

commit 9f0e3c6a75a4ed4e591dab97347be954599e2f8a
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Thu Aug 9 12:07:18 2018 +0300

    Add a WEBGL_multiview test that uses a depth buffer
    
    The test passes on the latest prototype implementation in Chromium.

commit 7ca87fb1d3da3b3d2060886e8c58e726d74c8219
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Aug 9 10:49:28 2018 +0300

    Add a test for passing a struct from a const array (#2693)
    
    This covers bug http://crbug.com/871434 . The root cause seems to be
    in the NVIDIA OpenGL driver.

commit 6c8c70151a2c4d2d171fb82532c767b643cc821c
Merge: a3a59b1d ba0e8e2c
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Aug 8 16:13:44 2018 -0700

    Merge pull request #2694 from kainino0x/texture-from-camera
    
    Add stress test for camera-to-texture upload

commit ba0e8e2cfd724f91cefbc8327821f0691bddd35b
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Wed Aug 8 11:54:57 2018 -0700

    Add stress test for camera-to-texture upload

commit a3a59b1d1f22d2fdb9afb34b9e58342f0092d0c6
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Aug 3 10:23:11 2018 -0700

    Update Chrome troubleshooting URL. (#2687)
    
    Make it point to the top-level help for Chrome. The previous URL has
    been redirected to a "tips and tricks" page which is not very useful.
    
    Thanks to webmaster James Riordon for raising the issue.

commit 216296b89ea62f6fc840602d0315512bc3fe6dac
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Aug 1 13:39:32 2018 -0700

    Regenerate WebGL 1.0 IDL. (#2686)
    
    Follow-on to #2685. Forgot to regenerate the IDL after backing off
    from my first change.
    
    Related to #2683.

commit 2125fb72c727a9b5df09928762aa63c87b88d126
Merge: beddcab1 e58ea0f7
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Aug 1 11:23:00 2018 -0700

    Merge pull request #2685 from kenrussell/expose-canvas-member-only-to-window
    
    Expose "canvas" attribute only to Window context.

commit e58ea0f7808b2ff56e0e1eb0db19ebd64d1673d8
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Aug 1 10:14:04 2018 -0700

    Expose "canvas" attribute only to Window context.
    
    HTMLCanvasElement doesn't exist on workers.
    
    Fixes #2683.

commit beddcab16e9a4abd8a9ba29336a98458e94643f1
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Wed Aug 1 19:51:18 2018 +0300

    Test WEBGL_multiview when the extension is disabled (#2684)
    
    The added tests make sure that none of the WEBGL_multiview
    functionality is enabled if the extension is disabled.

commit f32b09c7ede24b31cfc9866592c326fe4f18a159
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Wed Aug 1 10:58:59 2018 +0300

    Add WEBGL_multiview tests for single view operations (#2682)
    
    This adds two test files. One concentrates on drawing to multiview
    framebuffers with shaders that don't contain a GL_OVR_multiview
    extension directive. The other one does different operations on
    multiview framebuffers containing just a single view.
    
    All of the added tests pass with the latest version of ANGLE, however
    latest Chromium still needs to integrate the latest ANGLE changes with
    fixes to multiview read functionality.

commit 4c2bd94357af70a829e3165b6503c1a48fd689f2
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Jul 31 15:08:18 2018 -0700

    Explicitly set image's src in tex-video-using-tex-unit-non-zero.html . (#2680)
    
    It's not clear whether there's any guarantee that an img tag whose
    source is a data: URL will be fully loaded before inline scripts below
    it execute. Instead, set an onload handler for the image and set its
    src in script.
    
    Intended to address intermittent failures in this test per
    http://crbug.com/830901 .

commit ba80f238630a3cf6ec8114342f016d0d96a3d178
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Jul 31 15:06:34 2018 -0700

    Simplify WebGLTestUtils.startPlayingAndWaitForVideo. (#2681)
    
    Chrome's media team points out that their own internal tests only
    watch for video.currentTime > 0, and those tests are reliable. Use the
    same structure in WebGLTestUtils. This change should work across all
    web browsers.
    
    Addresses feedback given in http://crbug.com/834933 . Attempting to
    eliminate test flakiness observed in http://crbug.com/830901 .

commit 7be5f9b10ff71d49ebee0a2be7a55738c6a4ec36
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Jul 31 11:12:00 2018 +0300

    Add a test for draw framebuffer completeness (#2671)
    
    The test checks that the draw framebuffer can still be rendered to
    even if the read framebuffer is incomplete.
    
    The new test exposes a bug in Chrome 67. It seems to be specific to
    ANGLE's D3D11 backend. Bug link: http://anglebug.com/2737
    
    The latest stable version of Firefox is not affected by the bug.

commit 21bf7a8843da78e6cdf066f3165c1b8f7905d04b
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Jul 31 11:09:47 2018 +0300

    Add more WEBGL_multiview tests (#2679)
    
    Add more WEBGL_multiview tests
    
    This adds tests for different features that interact with
    WEBGL_multiview:
    
    1. Multiple draw buffers.
    2. Using a flat varying in a multiview shader (this covers a bug that
       was recently worked around in ANGLE).
    3. Instanced draw calls. Multiview can be implemented by leveraging
       instancing together with some other GL / D3D11 features, so it's
       important to test this.
    
    All of these tests pass when run against Chrome Canary with WebGL
    draft extensions enabled.

commit bf125f823ef41098b0c72e049f1fc9bae196170b
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Jul 31 00:27:32 2018 +0300

    Add WEBGL_multiview support to workload simulator (#2678)
    
    This demonstrates the performance benefits from using WEBGL_multiview.

commit 81b864a3b73a10a6e9c1540105a0409cf24b2484
Author: Justin Novosad <justin.novosad@gmail.com>
Date:   Mon Jul 30 17:22:55 2018 -0400

    Add OffscreenCanvas cases to texImage tests (#2496)

commit 95e34c25cb5b65f5d86fa6ca3b7d3359cb956ba7
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Mon Jul 30 12:26:31 2018 +0300

    Add a basic WEBGL_multiview test for WebGL 2 (#2669)
    
    Add a basic WEBGL_multiview test for WebGL 2
    
    The test tests multiview framebuffers based on 2D array textures.
    
    The test covers framebuffer queries as well as rendering into a
    multiview framebuffer. The test has been tested against a prototype
    implementation for Chromium.
    
    There's a separate test file for testing WEBGL_multiview interaction
    with EXT_disjoint_timer_query_webgl2.
    
    Also fixes an off-by-one error in the WEBGL_multiview spec. The same
    issue was fixed in the OVR_multiview spec and WebGL should follow
    suit.
    
    Separate tests could still be added for:
    
    1. drawing into a multiview framebuffer with a non-multiview shader
    (with num_views both 1 and > 1).
    2. Instanced draw to a multiview framebuffer.
    3. Multiple color render targets in a multiview framebuffer.

commit 945c872fd6c90173c125bff4048de06167ee156e
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Jul 26 15:16:04 2018 -0700

    Add [Exposed] Web IDL attribute to all WebGL interfaces. (#2673)
    
    * Add [Exposed] Web IDL attribute to all WebGL interfaces.
    
    This extended attribute has been required by Web IDL since
    heycam/webidl#423 . In order for WebGL to be accessible by web workers
    using OffscreenCanvas, the attribute being added to all WebGL
    interfaces is:
    [Exposed=(Window,Worker)]
    
    Fixes #2571 .
    
    * Address review feedback from jgilbert

commit 677a0e8fef00c34f02232c5f08a6ceeb3147a209
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Jul 20 17:51:35 2018 -0700

    Add license to webgl.idl and webgl2.idl. (#2674)
    
    Use the same MIT license that's used throughout the WebGL SDK, as it
    seems a better choice than Khronos' specification license.
    
    Remove obsolete comments referring to Khronos' typed array spec; it
    was folded into the ECMAScript specification some time ago.
    
    Fixes #2668 .

commit e6b8f7a0d64edbadb1c8025d9216b3f4d6122afe
Author: Jeff Leger <jleger@qti.qualcomm.com>
Date:   Fri Jul 20 16:16:22 2018 -0400

    Fix the ERROR_EPSILON for GL struct tests (#2676)
    
    Affects the ogles/struct/struct_049_to_56.html tests.  The ERROR_EPSILON
    value of 0.1 was too strict for FP16 implementation of medium precision.
    This change relaxes the epsilon to 0.125.

commit 21dbf06b5aa6c7dc8cf56314d4a3f96f57956c53
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Jul 19 13:50:58 2018 -0700

    Improve logging for tex-video-using-tex-unit-non-zero.html . (#2672)
    
    Use WebGLTestUtils.comparePixels, and include failing video type in
    error messages.
    
    Intended to diagnose intermittent failures of this test in
    http://crbug.com/830901 .

commit 6a1722fa0ed15b52563da09989115c729dca21c9
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Jul 17 13:57:33 2018 +0300

    Specify binding one buffer to two transform feedback outputs (#2658)
    
    This case used to be underspecified, so behavior varies between
    platforms. Particularly on D3D having two stream outputs written to
    the same buffer is disallowed, and on OpenGL ES there's some undefined
    behavior. Prohibit having the same buffer bound to two transform
    feedback outputs when calling BeginTransformFeeback to avoid
    non-portable behaviors from now on.
    
    Currently the test doesn't yet generate the expected error on any of
    the platforms tested. On Chromium or Firefox running on ANGLE's D3D
    backend, the buffer is not touched by transform feedback.
    On Chromium's GL backend, only one of the outputs is written to the
    buffer by transform feedback.
    
    The behavior is the same on recent HW/drivers from both NVIDIA and
    Intel.

commit c00e3f947a8c1a12861413c9637c7f464d4dc78c
Merge: eb12d5d1 b3f6a419
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Jul 11 12:24:47 2018 -0700

    Merge pull request #2662 from Oletus/fix-instancing-spec
    
    Fix WebGL 2.0 spec about non-zero divisor requirement

commit eb12d5d19790fe91d7ee0514459295ad42b636eb
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Jul 9 17:27:25 2018 -0700

    Added test cases for structs as out and inout parameters. (#2663)
    
    The highp variant of these tests fails on certain GPUs.
    
    Regression tests for the Chromium bugs http://crbug.com/851870 and
    http://crbug.com/851873.
    
    Test cases originally from
    https://github.com/mrdoob/three.js/issues/14137 .

commit b3f6a419edca22865feebc51867eaa0c374a7e98
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Mon Jul 9 17:45:42 2018 +0300

    Remove ENABLED_ATTRIBUTE section entirely and add test
    
    Add a test to verify that rendering works even if there's no enabled
    attribute arrays.

commit 10b0ae8b64e47daae420e751fd9024481e1d0bca
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Wed Jun 27 10:58:39 2018 +0300

    Fix WebGL 2.0 spec about non-zero divisor requirement
    
    The conformance tests were previously fixed to remove the requirement
    that there's at least one non-zero divisor, as it is not needed in
    D3D11-backed implementations of WebGL. This change brings the spec in
    line with the tests.
    
    The test is also expanded a bit to cover drawArrays and drawElements,
    not just their instanced variants.
    
    Currently Chromium passes the test, but Firefox follows the old 2.0.0
    spec rather than the newer version of the test.

commit 1a20624ad90ae0c6f7d27dbfea6443ec4fd84c5a
Author: zhenyao <zmo@chromium.org>
Date:   Wed Jul 4 18:09:29 2018 -0700

    Add a regression test for crbug.com/859400 (#2665)

commit a5c263cc63ffc2cc189b5214074c8792067c1853
Merge: cd43e6b0 8f5bf236
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Jun 26 16:17:09 2018 -0700

    Merge pull request #2659 from jdarpinian/tf-generic-binding
    
    Transform feedback generic binding point should not block buffer usage

commit cd43e6b0ea4e4cbe16c31d24eacb84420f1e79e0
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Jun 25 18:46:40 2018 -0700

    Add wtu.dispatchTask for use instead of waitForComposite. (#2661)
    
    Texture upload tests in particular take a long time because they run a max of
    60 tests per second, because they rely on waitForComposite to yield execution.
    waitForComposite should only be used when necessary. Use dispatchTask if we
    want to yield to the event loop.
    
    Before:
    all/conformance/textures/canvas (Passed: 3880/3880 in 48.68 seconds)
    conformance/textures/canvas/tex-2d-rgb-rgb-unsigned_byte.html (Passed: 1122/1122 in 14113.0 ms)
    
    After:
    all/conformance/textures/canvas (Passed: 3880/3880 in 8.90 seconds)
    conformance/textures/canvas/tex-2d-rgb-rgb-unsigned_byte.html (Passed: 1122/1122 in 2097.0 ms)

commit 963ec081162e4738fc6273cca797eb13ec88ce16
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Jun 25 18:32:49 2018 -0700

    Fix bug where comparison contradicts comment. (#2660)

commit 8f5bf23661eee2c0bb3042d03094f8b4581a4818
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Fri Jun 22 18:07:49 2018 -0700

    Transform feedback generic binding point should not block buffer usage
    
    The transform feedback generic binding point is not part of the
    transform feedback object and is not used for transform feedback. Only
    the indexed binding points are used. A buffer that is bound to the
    generic binding point should be usable for either transform feedback or
    non-transform-feedback purposes.

commit c39d3f7b6b88bc62a9851d6b81e309c320d26f83
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Jun 21 06:15:37 2018 +0300

    Test constructing a struct with a single member (#2657)
    
    This is a regression test for a Swiftshader bug.

commit e3ee0b7b66d076ccca0617d28c8993292ddece80
Merge: fb97619b bd6c4b1b
Author: Dean Jackson <dino@grorg.org>
Date:   Fri Jun 8 07:42:15 2018 +1000

    Merge pull request #2656 from JusSn/webkit-raf-update
    
    WebKit (now) supports requestAnimationFrame

commit bd6c4b1b248cd1a510d29e41d20cbea87f78ad1a
Author: Justin Fan <justin_fan@apple.com>
Date:   Thu Jun 7 12:20:46 2018 -0700

    WebKit actually supports requestAnimationFrame

commit fb97619baa91b3ae2322fccfa838efb419efde58
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Jun 6 11:55:20 2018 -0700

    Clarify errors when objects from different contexts are passed in. (#2654)
    
    This situation uniformly generates INVALID_OPERATION except for the
    is* queries, but this was not well specified or tested.
    
    Cover all API entry points accepting WebGLObject arguments in
    incorrect-context-object-behaviour test. Add WebGL 2.0 version of this
    test. Remove now-redundant resource-sharing-test.

commit 5e83981c2c44f24109b96b555d6865d481e0c609
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Tue Jun 5 08:36:18 2018 -0700

    Remove getBufferSubDataAsync tests from 00_test_list.txt (#2655)
    
    See #2653

commit 25a501c8ade78e334b108212fd18fc75595f8517
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Tue May 15 17:25:56 2018 -0700

    Add tests for getBufferSubData correctness and performance
    
    With the new "nonblocking getBufferSubData" optimizations,
    getbufferSubData needs these additional tests to make sure it doesn't
    return data that is too old or too new.

commit 19fe51d80f8decb662e218d0433e5220edd165b1
Author: Kai Ninomiya <kainino@chromium.org>
Date:   Fri May 4 23:01:34 2018 -0400

    Remove WEBGL_get_buffer_sub_data_async

commit 8ea92581353d18f50d25159489897ae634eb23b1
Merge: 602e0d14 d7cad2af
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu May 31 17:38:44 2018 -0700

    Merge pull request #2607 from jdashg/oob-fetch
    
    Spec drawElements behavior with null ELEMENT_ARRAY_BUFFER binding.

commit 602e0d14156817b7a7702b526bbd4b6adf02480e
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Wed May 30 18:09:07 2018 -0700

    Add WebGL workload simulator (#2652)
    
    A tool that provides a UI to select various kinds of WebGL work to
    perform each frame. Useful to test the effect of various different
    workloads on performance.

commit 198d637dd3e21d837fac6b3186cc6bc72e2f7219
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed May 16 19:21:43 2018 -0700

    Test texStorage[2,3]D with non-square sizes. (#2651)
    
    Regression test from Firefox where e.g. level=9, width=512, height=1
    fails.

commit 8a323892ef16706d3913ce61a3cc1513bd9deffe
Merge: e87fa9ff 3dc6e321
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed May 16 17:13:35 2018 -0700

    Merge pull request #2650 from Richard-Yunchao/blittingAreaOverflow
    
    Add tests for blitting to cover width/height overflow.

commit 3dc6e3217a52041343c9bdbafb13367e6731afdd
Author: Yunchao He <yunchao.he@intel.com>
Date:   Wed May 16 16:55:46 2018 +0800

    Add tests for blitting to cover width/height overflow.
    
    The range of an integer is [-0x7fffffff - 1, 0x7fffffff].
    We restrict the width/height to the max value that can be
    store in an integer in latest WebGL 2 spec. If we
    reverse the data order for blitting, an invalid value like
    0x7fffffff + 1 would become a valid value for an integer:
    -0x7fffffff - 1. But it is invalid for blitting.
    
    The new added tests cover this corner case to make it clear.

commit e87fa9ffca05dc5cdeb7563711eeed6c52be12f1
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue May 15 18:28:59 2018 -0700

    Revert "Test texStorage[2,3]D with non-square sizes. (#2647)" (#2648)
    
    This reverts commit 5a9f360fd5becea418f04920dbe6f6de342bc3e2.

commit 5a9f360fd5becea418f04920dbe6f6de342bc3e2
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue May 15 16:58:25 2018 -0700

    Test texStorage[2,3]D with non-square sizes. (#2647)
    
    Regression test from Firefox where e.g. level=9, width=512, height=1
    fails.

commit 2e7b0deac9bfdc41677396eac3f496dcbf74e136
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Thu May 10 22:57:09 2018 -0400

    Add a test for texImage3D from ImageData with unpack params (#2646)
    
    Regression test for https://crbug.com/804123

commit 24d8b3821cdca60e8f0892d20d1d3b0e3bf02313
Merge: b003a35a 94feeacc
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu May 10 12:03:55 2018 -0700

    Merge pull request #2642 from kenrussell/reject-webgl-shared-resources
    
    Reject WEBGL_shared_resources.

commit b003a35a6996fd57d0e7918fe6a7090145844ce1
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Wed May 9 12:11:01 2018 +0300

    Add a test for declaring both gl_FragColor and gl_FragData invariant (#2643)
    
    Declaring a variable invariant doesn't access it, so it should not
    really count as using the variable. Earlier versions of ANGLE would
    treat declaring both gl_FragColor and gl_FragData invariant an error,
    but recent versions have been fixed to allow this usage.

commit 94feeacc3985cd33f21ec0caa8445ced9e8937d6
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu May 3 14:19:00 2018 -0700

    Discussed alternatives on recommendation from kainino

commit 9c49f1700ac43db8fc011e4a2b2ae5ed6235985a
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed May 2 20:20:41 2018 -0700

    Reject WEBGL_shared_resources.
    
    At this point in the WebGL API's development it is not profitable to
    invest the time to implement this extension. There are other higher
    priorities, and all browsers are committed to making it possible to
    draw WebGL-rendered canvases into other canvases quickly, which
    essentially solves this problem.
    
    Provides a bit more description than #826.
    
    Removes test case added in #300.

commit bdb3165df40e27beaa1d2ad5615ce98cda04beb2
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Apr 26 16:08:41 2018 -0400

    Reject WEBGL_compressed_texture_atc and remove associated tests. (#2641)
    
    Addresses #2310.
    Also per discussion on public_webgl:
    https://www.khronos.org/webgl/public-mailing-list/public_webgl/1702/msg00054.php

commit 8ba2a913b9616fc3b9ff4ef0ddf071e463a45487
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Apr 20 20:56:34 2018 +0300

    Limit blitFramebuffer source and destination rectangle size (#2639)
    
    * Limit blitFramebuffer source and destination rectangle size
    
    BlitFramebuffer bugs that trigger when using large rectangle
    dimensions seem to be widespread on all desktop OpenGL drivers. They
    have been seen on at least:
    
    AMD Windows OpenGL
    Intel Windows OpenGL
    Intel Linux OpenGL
    NVIDIA OpenGL (all platforms)
    
    In addition there are some issues with handling large dimensions in
    existing blitFramebuffer workarounds inside ANGLE.
    
    In some cases these issues can be worked around by clipping the
    blitFramebuffer rectangles so that their width and height fit a signed
    32-bit integer. However, this approach doesn't work in case
    blitFramebuffer is used to scale the framebuffer and the greatest
    common divisor of the source and destination dimensions is 1.
    
    Since these issues likely don't have any impact on real-world
    applications, it's reasonable to limit the dimensions in the spec so
    that rectangles taller or wider than the maximum GLint value are not
    allowed.
    
    * Add more coverage for valid blitFramebuffer calls with large size

commit 8f662c6551a58ee53597f01ec873bfa8c11c4f63
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Thu Apr 12 15:46:49 2018 -0700

    Test transform feedback buffers bound to disabled vertex attribs and unbound VAOs
    
    This makes the tests for http://crbug.com/696345 a little more complete.

commit 7b676e3d3380aeda31255c0efd1924d075d0ea69
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Apr 13 15:20:25 2018 -0700

    Add test verifying minimal quality of rendered lines. (#2637)
    
    Regression test for http://crbug.com/820461 and
    http://crbug.com/swiftshader/99 .

commit 0f89b74f839de2693bd09585207bb4a56ec2a366
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Thu Apr 12 19:12:47 2018 -0700

    Add more stencil validation tests (testing validation caching) (#2635)
    
    Add new tests testing
    - various transitions from valid to invalid states (in order to test
      invalidation of the cached validation in Chrome)
    - drawing to the default framebuffer
    
    Also modifies dummySetProgramAndDrawNothing to keep one WebGLProgram
    _per WebGL context_.
    
    https://crbug.com/806557

commit 7c0541da63f571512c49758cbc0767117997a270
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Apr 10 15:54:15 2018 -0700

    Fix IDL of WEBGL_shared_resources draft extension. (#2633)
    
    Checked with https://www.w3.org/2009/07/webidl-check (plugging in some
    needed definitions above this extension's IDL).
    
    Fixes #2596.

commit 5c6e631e4a8a6306d74e49d4c6c51425f5a33a21
Merge: 3d0a94e4 17fc5257
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Apr 7 15:09:51 2018 -0700

    Merge branch 'zhenyao-copyTex'

commit 17fc5257d8dd505e8680d4fa57a5e2496df6aef3
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Apr 7 15:09:22 2018 -0700

    Fixed line endings and whitespace error

commit 3d7cc71804009149c9261191792127a633cd0fa0
Merge: 3d0a94e4 a6dcc8d6
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Apr 7 15:08:07 2018 -0700

    Merge branch 'copyTex' of https://github.com/zhenyao/WebGL into zhenyao-copyTex

commit 3d0a94e483d54f4c57470867a0da09abdd80c8ee
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Sat Apr 7 09:55:00 2018 -0700

    wtu.loadShader's gl.shaderSource should be treated as infallible. (#2631)

commit 07a5275d89cadce1c18d5497846c3b11857f4fad
Author: Jeff <26851121+etisdew@users.noreply.github.com>
Date:   Sat Apr 7 09:48:18 2018 -0700

    Consistent use of HTTPS (#2632)
    
    This is a very light PR amending http to https to the readme.

commit fefa32a950127f6e01a5cdd1536d639a949cd82b
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Apr 6 20:00:43 2018 +0300

    Add a test for having a case statement inside a block (#2630)
    
    This was not being validated correctly by ANGLE until recently, though
    this would not cause issues if the native drivers were working
    correctly.

commit ae6e95459340f198e6fc5350de494b71ac1e7bb6
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Apr 6 19:44:31 2018 +0300

    Add a test for using non-constant texture offset (#2629)
    
    The existing texture offset tests did not seem to cover this invalid
    usage of the texture offset functions yet.

commit 51813339bb17da50881369cfdcf4062ebf4e2ce2
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Apr 5 16:07:18 2018 -0700

    Use fixed number of iterations in sync-webgl-specific. (#2627)
    
    * Use fixed number of iterations in sync-webgl-specific.
    
    Like the conformance2/query/occlusion-query test, use a fixed number
    of iterations, as on slow automated test machines, a fixed duration
    based on Date.now() leads to test flakiness.
    
    Improve the synchronization with the GL server by using readPixels
    instead of flush or finish.

commit fcbee0da70ba81b80d922e0534bfbd24f47d31af
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Apr 5 15:36:29 2018 -0700

    Test combinations of clearBuffer*(), clear() and draws. (#2628)
    
    With the preserveDrawingBuffer:false context creation option, Chromium
    was incorrectly tracking the effects of clearBuffer on the default
    back buffer.
    
    Modified various quad setup functions in webgl-test-utils to allow the
    quad to be more easily scaled.
    
    Regression test for http://crbug.com/828262 .

commit 2e8a4bf2581aa48d6976421fac67144359184889
Author: Deyi Li <deyi.li2015@gmail.com>
Date:   Wed Apr 4 17:35:00 2018 -0700

    Test queueing multiple context restored event after context is lost (#2611)
    
    When calling restoreContext() multiple times after context is lost,
    all queued context restored events should be merged before an event is fired.
    Therefore, calling restoreContext() multiple times should not cause error or crash.

commit c012dc26b0e7ef5c7c2f5cc1cbc3ad19f8f5e71b
Author: Patrick To <patrto@microsoft.com>
Date:   Wed Apr 4 17:22:57 2018 -0700

    Add test to verify DYNAMIC_DRAW bufferData functionality (#2610)
    
    * Add buffer-data-dynamic-delay.thml test to verify DYNAMIC_DRAW functionality for buffers
    * Change shaders and draw to use vec2 and TRIANGLE_STRIPS

commit aef0b3a79983db6f553f35f6ccd7b9303a72d2fa
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Wed Apr 4 16:59:53 2018 -0700

    Change dummySetProgramAndDrawNothing to draw >0 elements (#2626)
    
    * the program now sets gl_Position and gl_FragColor to vec4(0.0)
    * drawArrays now draws one triangle instead of 0
      (but that triangle is of zero size)
    
    This is necessary to make sure the draw call doesn't early out before it
    does validation.

commit 221bb2dcdc712c7bc0a4daa8db997c2dfd57bd27
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Wed Apr 4 15:13:36 2018 +0300

    Clarify WebGL 2 spec for attribute aliasing (#2616)
    
    GLSL ES 3.00.6 already has strict rules when it comes to attribute
    aliasing, so WebGL 2.0 can simply refer to those rather than spell out
    its own.
    
    A new test is added to fully cover the attribute aliasing rules of
    GLSL ES 3.00.6. It shares some code with the older attribute aliasing
    test for WebGL 1.0.

commit ca7b24622038f002da5a203c1399983d48cc1658
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Tue Apr 3 15:36:17 2018 -0700

    Fix usage of glEnumToString in webgl-specific-stencil-settings (#2625)
    
    and add a safeguard against future glEnumToString usage mistakes

commit f8324ad7fd8faa60a9201fbf7e02df1e379ea39c
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Mon Apr 2 19:19:42 2018 -0700

    Report skipped tests as both skipped and failed rather than just failed. (#2624)
    
    Skipped tests also show as gray instead of red.

commit 25b0cd6380bd406a1c4cddee91168a82b59b5009
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Apr 2 18:11:33 2018 -0700

    Revamp text results, include json block for result details. (#2617)

commit e899c531173da8ba956df8de6dd74fb7d9b1fa9c
Author: Patrick To <patrto@microsoft.com>
Date:   Fri Mar 30 13:18:38 2018 -0700

    Add unpack flip y test for depth textures (#2609)

commit aa27baddb0f5662987164c34b79150786f2118e6
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Mar 30 13:16:51 2018 -0700

    Replaced invalid "enum" with "GLenum" in disjoint timer query extensions. (#2623)
    
    Verified that these are the only places in the extensions' IDL which
    have this mistake.
    
    Fixes #2596.

commit 6b68a24befc2b537f6c905299a6f1db521351a5e
Author: Jiajia Qin <jiajia.qin@intel.com>
Date:   Sat Mar 31 01:15:49 2018 +0800

    Add a test case in for loop condition (#2620)
    
    The test in for loop expression can't expose an Intel driver bug. We
    add another test in for loop condition to conver it.

commit d7cad2afebc236c647e52e393f990a624590a90e
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Mar 8 17:00:57 2018 -0800

    Spec drawElements behavior with null ELEMENT_ARRAY_BUFFER binding.
    
    Issue #2500.

commit da5abe6e9bae583e4ae73aa553547fcaa637d099
Merge: ca2d5236 ae936c9f
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Mar 28 16:04:15 2018 -0700

    Merge pull request #2621 from jdarpinian/master
    
    Explanatory comment for transform feedback buffer ownership change.

commit ae936c9fba7d70c208833fd0890fdce61840cea4
Author: James Darpinian <jdarpinian@google.com>
Date:   Wed Mar 28 14:54:34 2018 -0700

    Explanatory comment for transform feedback buffer ownership change.

commit ca2d523604bcd6d4e4210f5f7a5f3b481f214a41
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Mon Mar 26 20:01:11 2018 +0300

    Explicitly specify uniform and attribute name collisions (#2615)
    
    WebGL has long enforced behavior that goes against GLSL ES 3.00.6
    section 12.47. WebGL requires that uniform and attribute names don't
    collide, whereas GLSL ES specifies that there's no collision in this
    case. The WebGL test that enforces this is:
    
    conformance\glsl\misc\shaders-with-name-conflicts.html
    
    The test refers to discussion in OpenGL ES working group, but it seems
    that the results of the discussion were not put into writing outside
    of the test case.
    
    Explicitly specify the WebGL behavior to clear confusion around this.

commit 55905c6f5fc0b8677b2e46cdf6c11571d47f6612
Author: zhenyao <zmo@chromium.org>
Date:   Fri Mar 23 09:43:40 2018 -0700

    Add a regression test for Chrome's DeleteBuffers behavior. (#2614)

commit b06d0845682f15896a1043100a039d1dca03f6e3
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Thu Mar 15 13:35:21 2018 -0700

    Test that transform feedback calculations don't overflow. (#2613)
    
    * Test that transform feedback calculations don't overflow.
    
    * Note that the buffer is too small.

commit a29b472481c45c1135a149942b4cf590f6f29581
Author: Patrick To <patrto@microsoft.com>
Date:   Wed Mar 14 13:39:41 2018 -0700

    Change GLSLConformanceTester.runTests() to be synchronous instead of using setTimeout (#2612)

commit 63cf94f8801d617e5068e6b86a9fed7a6b86b4e5
Author: Patrick To <patrto@microsoft.com>
Date:   Tue Mar 13 13:00:43 2018 -0700

    Remove premature finishTest from constant-precision-qualifier.html (#2608)

commit dba66c45745a6508d8f311ac57d7ba0e90845f94
Merge: f51f2b71 14dc1854
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Mar 9 16:50:02 2018 -0800

    Merge pull request #2583 from jdashg/stencil-masking
    
    Only complain on stencil writemask/value_mask/ref front/back mismatch…

commit 14dc185472c2679ceba016521d2275a0b149ed10
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Dec 18 19:23:43 2017 -0800

    Only complain on stencil writemask/value_mask/ref front/back mismatch for the `s` stencil bits of the draw framebuffer.
    
    - WRITEMASK and VALUE_MASK mask, whereas REF clamps
    - Add and use drawNothing call

commit f51f2b7120c320383a86c0529c6b8c16b5ef6f1d
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Mar 8 23:48:33 2018 -0800

    Support robust_buffer_access_behavior implementations. (#2606)
    
    In particular, don't try to draw $LARGE numbers of triangles. An RBAB
    implementation will not reject these 'invalid' draws, and will try to draw
    all the triangles, which, for N=0x7fffffff trips the TDR and incurs context
    loss.

commit 3faeace7755a8e9327c82919b71ef35bf6e6714d
Merge: bfeae430 5e584a9a
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Tue Mar 6 15:54:23 2018 -0800

    Merge pull request #2604 from jdarpinian/too-small-buffers
    
    Test transform feedback into too small buffers.

commit 5e584a9a1428c6bf38a103ee43c5907b18eccc08
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Tue Mar 6 12:49:32 2018 -0800

    Use constant sizeOfFloat instead of 4.

commit f1369ed1a7242f0f614d1d61aac5997130ae896b
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Mon Mar 5 17:38:47 2018 -0800

    Make valid output nonzero to distinguish from default zeroed state

commit 4fc5d5f8fcd0bef02f336a9a1dbcd8571ef972c4
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Mon Mar 5 15:42:22 2018 -0800

    fix copy/paste error

commit bfeae430a584553eff6553fb694025e448bb3d09
Merge: 3c1cb020 fffff31f
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Sat Mar 3 00:25:45 2018 -0800

    Merge pull request #2602 from jdarpinian/transform_feedback
    
    Expanded transform feedback tests

commit fffff31fbba0f4fadd9730253609b64620a5ed8f
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Fri Mar 2 18:27:14 2018 -0800

    Also test draw*Instanced and drawRangeElements.

commit 8b76f94ad47aaf0ccea96a888bfae0dfec9820ba
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Fri Mar 2 18:09:14 2018 -0800

    Remove unused helper functions

commit 35c0c2875304c2353ba6d4b70319dc007975f48e
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Fri Mar 2 18:05:49 2018 -0800

    Test transform feedback into too small buffers.

commit 3c1cb0203b6cfc10389e85a350b2ea6ca29d01ce
Author: vikas soni <36902690+vikaschromie@users.noreply.github.com>
Date:   Fri Mar 2 12:50:33 2018 -0800

    Include file path fix. (#2603)
    
    This test was passing when using python SimpleHTTPServer even if the include
    paths were incorrect. This is because python smartly searched for these include
    files in other folders. The test failed when command line was used to run test
    by bots or when running test directly on chrome without using the simplehttpserver.

commit 8755324177728b95abf6f4d41604ddde5030d1c1
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Thu Mar 1 19:17:17 2018 -0800

    update title

commit 94d5fb4e2ca805663ece4c237225cce68306ae75
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Thu Mar 1 19:00:41 2018 -0800

    Rename states-and-errors test to switching-objects

commit b3b94b28982ab8cd814cf9ee305e5156ee728926
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Thu Mar 1 18:51:20 2018 -0800

    Uncomment code

commit 3ed07c5817e76f7de836ba5d4b454d56d35e23c7
Author: James Darpinian <jdarpinian@chromium.org>
Date:   Thu Mar 1 18:47:39 2018 -0800

    remove commented code

commit 3b1a79834b02fc5bef0d7bf482f11ca4d4c8229d
Author: James Darpinian <jdarpinian@google.com>
Date:   Mon Dec 11 18:42:05 2017 -0800

    Test the effects of switching transform feedback objects

commit 79e283f383b33950fcb02a5e3974c8bab24c9704
Author: vikas soni <36902690+vikaschromie@users.noreply.github.com>
Date:   Thu Mar 1 14:19:47 2018 -0800

    Multiple WebGL2 Context sharing texture2darray/texture3d data bug test. (#2601)
    
    * Multiple WebGL2 Context sharing texture2darray/texture3d data bug test.
    
    This test verifies that 2 different contexts both using 2d array texture
    or 3d texture does not share/leak the texture data among them due to context
    save/restore bug.
    
    * Fixed trailing whitespaces, copyright date. Added link to original bug
    in test description.

commit 5edec046d84d2c7227c4b98ab7f7c5fc594781d8
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Wed Feb 7 18:35:38 2018 -0800

    Define that sync objects do not become signaled until control returns... (#2598)
    
    to the event loop.
    
    Follow the same semantics for sync objects as were previously used for
    queries. Add a test for the new behavior.

commit a182a9ad3078aca566d8355eabf2d9f56f70ee82
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Feb 6 19:37:52 2018 +0200

    Assign attribute locations explicitly in luma texture copy test (#2597)
    
    The test is not querying attribute locations, so use the test
    utilities to bind the attributes to locations 0 and 1. This ensures
    that the test runs consistently.
    
    This change is backported also to conformance suite version 2.0.0.

commit 98a3d3b4355644bb1187a858fdd2aff6b2bc84ec
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Feb 1 21:07:55 2018 +0200

    Reset vertex attrib divisors between instancing tests (#2595)
    
    The newest version of the ANGLE_instanced_arrays test runs different
    subtests with different shaders, which may have overlapping attrib
    locations. Make sure that the different subtests don't interfere with
    each other by resetting vertex attrib divisors after each subtest.
    
    Previously the test could pass if attrib locations were automatically
    assigned in a certain way, but changes to attribute location
    assignment in the WebGL implementation could make it fail when it
    should have passed.
    
    This change does not need to be backported to earlier conformance suite
    versions nor the WebGL 2 test for instancing, as they only use one
    shader and always bind vertex attrib locations explicitly, so the
    locations can only be set one way.

commit 2b36e6230fff116a7480e4ada1039556be8d8125
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Jan 30 21:26:11 2018 +0200

    Add a test for a HLSL bug with if, return and else if (#2594)
    
    An if statement containing a return together with an else if produces
    wrong output on Chrome and Firefox on Windows.
    
    This is a minimal test case for this bug: http://anglebug.com/2325

commit a6dcc8d6388d95c9086fcd4b87a0bb58cd4d2ed5
Author: Zhenyao Mo <zmo@chromium.org>
Date:   Mon Jan 29 17:11:13 2018 -0800

    Add a regression test for crbug.com/804118

commit 1f55c35cf302f0575584c83a32d041dd79952e5c
Author: James Riordon <jriordon@outofcontrol.ca>
Date:   Mon Jan 29 10:16:32 2018 -0500

    Expand page with additional links

commit db4649010bb7d3a8b17003139626056701b9dcc6
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Fri Jan 26 20:54:42 2018 -0800

    Probe for max texture level for each format separately. (#2579)
    
    * Probe for max texture level for each format separately.
    
    * Remove now-undefined references, pass null for texImage data.

commit a6440561426d6779e4eefa3bcef3d51ff33657d4
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Thu Jan 25 11:22:48 2018 -0800

    Increase tolerance to 3. (#2586)
    
    Variances of 2 have now been seen on MacBook Pros with AMD GPUs:
    https://ci.chromium.org/buildbot/chromium.gpu/Mac%20Retina%20Release%20%28AMD%29/22175
    
    Fixes http://crbug.com/804887 .

commit 68301bc35c9162ae6680294f47a4c308b747f48a
Author: Greggman <github@greggman.com>
Date:   Fri Jan 26 02:23:03 2018 +0900

    add test for incompatible texture types for samplers (#2587)
    
    note: shadow sampler types are not tested here.
    they probably need a separate test as they have
    more requirements related to texture parameters

commit 4181ad31b5ce4f40dee512eb764b61d7bb6131ae
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Jan 19 11:48:07 2018 -0800

    Update new test harness's use of Telemetry API. (#2584)
    
    It seems this WebDriver based harness also uses the Chrome-specific
    Telemetry API when running on ChromeOS.
    
    Fixes http://crbug.com/803843 . Thanks to nednguyen@ for the patch.

commit 2c59b0382d65c6778a4328ce542cfe11475e2eb2
Author: He Yunchao <yunchao.he@intel.com>
Date:   Wed Jan 17 06:08:32 2018 +0800

    Remove unnecessary GLenums about coverage (#2581)

commit 075a5110d2e8857c0007e2e1501c321b55cb51ba
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Fri Jan 12 22:28:17 2018 -0800

    Add GLSL 3.00 test for forward declarations of functions. (#2580)
    
    Fails on Pixel C with Chrome because of flattening of precision
    qualifiers into variables' types, and because of a bug in the GLSL
    compiler on this device.
    
    Thanks to Robert Muth of art.muth.org for reporting.

commit efdb969820ea7e2bd73f6a00da5f983a7357cc09
Author: zhenyao <zmo@chromium.org>
Date:   Mon Jan 8 22:31:12 2018 -0800

    Add a test case for CopyTexImage2D behavior when source/target images (#2576)
    
    * Add a test case for CopyTexImage2D behavior when source/target images
    are from the same texture. Includes pixel verification.
    
    This is a regression test for crbug.com/797235

commit a6f99093cbc1e22112b0d36ae368438d268f2376
Author: Jinho Bang <zino@chromium.org>
Date:   Fri Jan 5 08:23:47 2018 +0900

    Fix misuse of mixin syntax (#2575)
    
    Introduced `interface mixin` syntax in #2566 but it's incorrect.
    A `interface mixin` SHOULD NOT includes another mixin.
    
    This fixes #2566.

commit 41f508c9bef01b280b6e139db2946c1ea372fcd3
Author: Tarek Sherif <tsherif@gmail.com>
Date:   Thu Jan 4 17:48:55 2018 -0500

    Regression tests for ANGLE UBO dirty bit bugs (#2573)

commit 2df151be79b1a44d41464b149d6044fbf6bbf2f7
Author: Jinho Bang <zino@chromium.org>
Date:   Thu Jan 4 13:34:46 2018 +0900

    Use Web IDL's new-ish interface mixins concept (#2566)
    
    WebIDL recently introduced dedicated syntax for mixins [1]. This
    replaces the existing [NoInterfaceObject] and "implements" syntax with
    "interface mixin" and "includes" in the appropriate places.
    
    This fixes #2538, #2539 issues.
    
    [1] https://github.com/heycam/webidl/commit/45e8173d40ddff8dcf81697326e094bcf8b92920

commit 57590ed441d09dad8050def51162e3177334d884
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Wed Jan 3 13:09:05 2018 -0800

    Promote WEBGL_multiview to draft. (#2572)

commit deee02713ef79e0dd9b409fdaea75c19d0ff0730
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Wed Jan 3 21:51:08 2018 +0200

    Add a test for assigning twice to a swizzled vec inside a function (#2570)
    
    This reproduces a crash at least in Chromium running on recent NVIDIA
    OpenGL drivers.
    
    Chrome bug report that this test case is based on is here:
    http://crbug.com/798117

commit 8a5a77beb5bf9f213ab04ea36f0c55adce3af39e
Merge: 47b400b6 678861db
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Sat Dec 30 17:53:08 2017 -0800

    Merge pull request #2569 from kenrussell/fix-jdashg-review-comments-from-2523
    
    Fix review comments from @jdashg on #2523.

commit 678861dbd7ef1583ea6cbf95fbfef8681d1dc207
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Dec 30 15:42:22 2017 -0700

    Fix review comments from @jdashg on #2523.

commit 47b400b67281698c537263e9f00d1394550201a2
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Dec 28 23:17:02 2017 +0200

    Add a test for framebuffer tex attachment with changing base level. (#2568)
    
    This exposes a bug in ANGLE's D3D backend.
    
    http://anglebug.com/2291

commit 45bca00d00c9638c8fa71ba28f37a717657b3748
Author: James Darpinian <jdarpinian@gmail.com>
Date:   Fri Dec 22 13:55:13 2017 -0800

     Relax restrictions on transform feedback buffer use (#2560)
    
    * Relax restrictions on transform feedback buffer use. #2346
    
    * Add minimized conformance test for transform feedbacks' buffer state.
    
    Regression test for http://crbug.com/696345 .
    
    Thanks to @greggman for the test.
    
    * Test for having the same read and write buffer using a transform feedback
    
    note: there are 2 lines marked as "FIXME" that I believe should be removed
    assuming the spec changes so that `gl.bindBuffer` is less strict
    
    * Test for using the defaul transform feedback object
    
    Note: 2 lines are marked as FIXME and should be removed when/if
    impls and the spec are updated to make `gl.bindBuffer` less
    strict.
    
    * Update tests for new transform feedback binding behavior
    
    * Remove trailing whitespace
    
    * More [copy]buffer[Sub]Data tests.
    
    * Revert "Test for using the defaul transform feedback object"
    
    This reverts commit d96d44b45ed78e7d99d6c5a008a387067c7d3fe3.
    
    * Also update conformance suite for new feedback rules
    
    * add --min-version 2.0.1 to new tests
    
    * Match Firefox's behavior.
    
    * Clarify that unbound TF objects don't count

commit 09447f610c62c6b92a1ec75ecb20cdb5ab95c046
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Tue Dec 19 14:25:12 2017 -0800

    Propose WEBGL_texture_multisample. (#2562)

commit e98f0d0ae8d7dfb6703432bd7d0b6bc5eccf86ee
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Dec 19 19:23:09 2017 +0200

    Add a test for weird name conflict on NVIDIA GL driver (#2565)
    
    Some recent versions of NVIDIA GL drivers have an issue with a
    function in parameter being passed as an inout argument while there is
    a global variable with the same name.
    
    The issue will be fixed in a future NVIDIA driver update.
    
    http://crbug.com/792210

commit d458ada06171a85af00367251a4ed55db7fe2018
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Dec 18 15:53:56 2017 -0800

    Fix possible race condition in image-bitmap-from-blob tests. (#2561)
    
    Possible cause of http://crbug.com/795945 .

commit 1a8ed15c9c25f2b1ccd97149c94ea245ab982252
Author: zhenyao <zmo@chromium.org>
Date:   Fri Dec 8 13:10:35 2017 -0800

    Add a regression test for ANGLE bug (crbug.com/793307). (#2559)

commit c74600c018560088c9172b9c6022a910f2fbf89c
Author: Deyi Li <deyi.li2015@gmail.com>
Date:   Thu Dec 7 11:08:50 2017 -0800

    Tests should accept multiple valid error conditions (#2552)
    
    * Additional tests change for texSubImage2d when format/type differs
    
    These tests should also accept INVALID_OPERATION for the same reason we discussed in https://github.com/KhronosGroup/WebGL/pull/2550
    
    * change conformance-suites test results

commit 47422e0de87baf65fb84bd3c167658ff2b9be2f3
Author: zhenyao <zmo@chromium.org>
Date:   Wed Dec 6 09:53:04 2017 -0800

    Add a test for FLIP_Y and PREMULTIPLY_ALPHA behaviors in WebGL2. (#2555)
    
    * Add a test for FLIP_Y and PREMULTIPLY_ALPHA behaviors in WebGL2.
    
    Also update the 2.0 spec.
    
    * Relax pixel comparison to allow premultiply_alpha error.
    
    * Per kbr comment
    
    * Add a regression test for Chrome specific bug
    
    * Fix

commit b7c786a9a8db12cea2faa3402c522ef6df430465
Author: Jie Chen <jie.a.chen@intel.com>
Date:   Wed Dec 6 14:35:32 2017 +0800

    Fix a bug in tcuFuzzyImageCompare.bilinearSample (#2557)
    
    This is a typo compared with the native dEQP source code.

commit 0a541bf404fc944edefcad82e456b248a428966d
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Mon Dec 4 11:19:57 2017 -0800

    More depth/stencil renderbuffer initialization tests (#2540)
    
    * More depth/stencil renderbuffer initialization tests
    
    Both multisampled for WebGL 2.0 and non-multisampled for WebGL 1.0.
    
    * revise

commit 17c2ce535da33093717cdb8612a1f4b50ff7583a
Author: Kai Ninomiya <kainino1@gmail.com>
Date:   Fri Dec 1 17:15:24 2017 -0800

    clarify "clip z" in multisampled-depth-renderbuffer-initialization.html (#2553)

commit b3fb9805bd5e3c920452a7f8fc51eacf2836b4f8
Author: gofr <32750931+gofr@users.noreply.github.com>
Date:   Sat Dec 2 01:40:23 2017 +0100

    Make test harness result reporter log success message to console (#2533)
    
    The wrapped logging function it uses only takes one argument.
    Concatenate the message to the success state string rather
    than passing it as an unused second argument.

commit 05591bbeae6592fd924caec8e728a4ea86cbb8c9
Author: JiangYizhou <yizhou.jiang@intel.com>
Date:   Fri Dec 1 07:44:02 2017 +0800

    Reproduce Intel driver bug on windows when rendering large viewport (#2551)
    
    On Intel windows platforms, rendering large viewport and stencil buffer
    enabled will lead to memory leak on d3d11 driver.
    
    Test for help reproduce bug http://crbug.com/782317

commit a2639880f84609fb6330bd2f7d1bd1c1016bd3c3
Author: Deyi Li <deyi.li2015@gmail.com>
Date:   Wed Nov 22 15:50:23 2017 -0800

    Accept INVALID_OPERATION for texSubImage2D when format/type differs (#2550)
    
    * Accept INVALID_OPERATION for texSubImage2D when format/type differs
    
    INVALID_OPERATION should also be an acceptable result because texSubImage2d could be called with format and type that differ from format and type that was set previously by texImage2D.
    
    * Update Tests Results in conformance-suites

commit 54f8b13411e86f4f5f3d115eb6a6cb26605694b7
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Nov 20 12:04:13 2017 -0800

    Simplify render-after-resize-test. Draw single unit quad. (#2549)
    
    Exposes the issue more simply. Follow-on to #2546.
    
    Still passes in Firefox, fails in Chrome and Safari, on macOS 10.13 on
    MacBook Pro with NVIDIA GPU.

commit e4919fa03c74bd561dcabf3e61668fa3c7e54353
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Nov 14 12:09:45 2017 -0800

    Fix fencepost errors covered up by wtu.checkCanvasRect. (#2546)
    
    Follow-on to #2544 and #2543.

commit aca2944238a652220a2ecab56de93597822cdf22
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Nov 14 19:37:53 2017 +0200

    Add test that initializes an array inside a loop (#2545)
    
    This attempts to reproduce a driver bug seen on Adreno 418.

commit 97f84e72990ad16eeef6e5e757e7946d11748d67
Merge: b34beaf6 f9c0cbb6
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Mon Nov 13 20:28:06 2017 -0800

    Merge pull request #2534 from Oletus/webgl-multiview-clarification
    
    Clarify WEBGL_multiview spec

commit b34beaf60f12a0023dd3c5ec91deb5403aa438d1
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Mon Nov 13 11:38:08 2017 -0800

    Simplify test: eliminate 2D canvas, draw triangles instead of point. (#2544)
    
    Fix latent bug in wtu.setupSimpleColorProgram. Fix copyright date.
    
    Follow-on to #2543.
    
    Acts as regression test for http://crbug.com/775202 .

commit e968ed426a18a309a104f2fe1bea307dc1f6da2c
Author: Greggman <github@greggman.com>
Date:   Tue Nov 14 01:28:36 2017 +0700

    Adds a test for rendering directly after resizing the canvas (#2543)

commit 4894e573dd5e106decfd50c105307e8cb7db4605
Merge: 12192b94 317361c3
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Sun Nov 12 12:49:19 2017 -0800

    Merge pull request #2542 from kenrussell/scissor-state-management-bug
    
    Add test for scissor rectangle maintenance bug during implicit clears.

commit 317361c35ba3a9af6a1a635acddbaf170736bce3
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Nov 11 17:31:06 2017 -0800

    Fixed tabs

commit dc79339399dba03e20a985da04bd0e57257dff19
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Sat Nov 11 17:24:49 2017 -0800

    Add test for scissor rectangle maintenance bug during implicit clears.
    
    Regression test for http://crbug.com/778770 .

commit 12192b948daa2d83269ab46cbf03d65e9f2a48a3
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Nov 7 19:47:45 2017 +0200

    Add a test for varying struct with an inline definition (#2541)
    
    This test exposes a bug that was recently fixed in ANGLE.

commit e0fd88e26a7531752733647c331f76ec59106a6f
Author: Brandon Jones <tojiro@gmail.com>
Date:   Fri Oct 27 15:04:35 2017 -0700

    Added a test to ensure preserveDrawingBuffer isn't leaking resources
    
    * Added a test to ensure preserveDrawingBuffer isn't leaking resources
    
    * Update 00_test_list.txt
    
    This test provokes a crash on Samsung Galaxy S6 devices running Android 6.0.1. Other devices seem to handle it fine.

commit c9c9c573432be33b983ed7218717df70fda94fdf
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Fri Oct 27 19:15:48 2017 +0300

    Add tests for switch/case validation corner cases (#2535)
    
    These tests expose bugs in ANGLE, most of which have been already
    fixed in ANGLE TOT.

commit f9c0cbb6c6fe09de443548cf43b9b90d1f6875f3
Author: Olli Etuaho <oetuaho@nvidia.com>
Date:   Tue Aug 22 17:51:47 2017 +0300

    Clarify WEBGL_multiview spec
    
    The multiview spec is now written against the latest revision of the
    native OVR_multiview spec. The native OVR_multiview spec has been
    tightened to specify more error conditions, so some of the addendums
    in the WebGL spec are not necessary anymore.
    
    Formatting and examples are also improved.
    
    This doesn't introduce any major functional changes.

commit 5d2cb3399b93abee71f9a4bb139399b512ad7b0a
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Oct 17 18:33:48 2017 +0300

    Add more tests for vector/scalar arithmetic inside a loop (#2531)
    
    Add more variations of shaders that reproduce the same bug. This adds
    better coverage for a workaround under development for ANGLE.
    
    Some more complex variations are put to a separate test file - the
    workaround in ANGLE will not be sophisticated enough to handle them,
    but they should be fixed by a future driver update.

commit e7fe9be6f15d19430994be25ae6f52afc2a87d0a
Author: Kirill Dmitrenko <kdmitrenko@gmail.com>
Date:   Mon Oct 16 19:27:10 2017 +0300

    WEBGL_texture_source_iframe: some rewording about promises (#2530)

commit ac794b3b5cc5c264f9a8b5128dbfaac9d1b50ed9
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Thu Oct 12 17:27:17 2017 +0300

    Add test for vector/scalar arithmetic inside a for loop (#2529)
    
    This exposes a bug in the NVIDIA GL driver.

commit d88d5742579f59a999037221b36435abfb84c055
Author: Olli Etuaho <olli.etuaho@gmail.com>
Date:   Tue Oct 10 19:23:53 2017 +0300

    Add a test for switch/case corner cases (#2528)
    
    This test case covers some bugs in ANGLE's HLSL translation.

commit 9536e757bd06f1e8401a5c93c332f81c7d143f99
Author: Yang Gu <yang.gu@intel.com>
Date:   Sun Oct 8 05:47:02 2017 +0800

    Add conformance test script (#2361)
    
    This script is to run any version of Khronos WebGL conformance test on
    various OSes (like Android, ChromeOS, Linux, MacOS and Windows) with
    various browsers (like Chrome, Edge, FireFox, Safari, etc.). Results
    will be compared with expectations and a final report will be generated.

commit 5eb2d21bf9747e9d18092691dd2be40815d9c93a
Author: Jeff Gilbert <jdashg@gmail.com>
Date:   Thu Sep 28 15:00:14 2017 -0700

    Update EXT_sRGB test to forbid GenerateMipmap as per parent spec. (#2524)

Cherries picked
================================================================================
Merge base from: origin/master

commit 8bf026028fe6f66454d18711daab7a26967b37cd
Author: Ken Russell <kbrussel@alum.mit.edu>
Date:   Tue Sep 26 15:09:24 2017 -0700

    Add test case performing repeated blitFramebuffer calls without clears. (#2523)
    
    Test case for crbug.com/768969 .
