I'll look into it, however I don't think the gles2_conform failure is related.

It looks like angle_end2end_tests always references libEGL.dylib even in the static library configuration. However, it doesn't appear that the rpath is set up on Mac, which is why it can't find the libEGL.dylib. Similar thing is set up on Linux, though: https://cs.chromium.org/chromium/src/third_party/angle/src/tests/BUILD.gn?q=cs+angle_end2end_tests&sq=package:chromium&dr=C&l=86

It seems to be a bot specific problem: https://build.chromium.org/p/chromium.gpu.fyi/builders/Mac%2010.10%20Release%20%28Intel%29/builds/13743 contains the GN switch patch and angle_end2end_tests and gles_conform succeed.

OK. Probably a bug with the non-swarming config.

It looks like the install_name and rpath aren't set up for these shared libraries, like they are in GYP: https://cs.chromium.org/chromium/src/third_party/angle/src/libGLESv2.gypi?type=cs&q=third_party/angle/src/libGLESv2.gypi&sq=package:chromium&l=697

https://chromium-review.googlesource.com/c/354302/ should fix that.

The following revision refers to this bug:
  https://chromium.googlesource.com/angle/angle/+/db8ae836731591f5db582545de403b3ca1c8edf7

commit db8ae836731591f5db582545de403b3ca1c8edf7
Author: Robert Sesek <rsesek@chromium.org>
Date: Tue Jun 21 15:47:03 2016

[Mac/GN] Properly set the install_name and rpath for shared_library targets and their consumers.

BUG= 621932 
R=cwallez@chromium.org

Change-Id: I10b0fd5a668d74b7062747002bc48ef0e0a6df9c
Reviewed-on: https://chromium-review.googlesource.com/354302
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

[modify] https://crrev.com/db8ae836731591f5db582545de403b3ca1c8edf7/BUILD.gn

Robert, it looks as though your CL fixed end2end_tests. Can you also look at how to fix gles2_conform_test on the same bot?

It looks like we're missing some kind of testdata based on the failure:

https://build.chromium.org/p/chromium.gpu.fyi/builders/Mac%2010.10%20Release%20%28ATI%29/builds/11261/steps/gles2_conform_test/logs/stdio

[ RUN      ] GLES2ConformTest.GL_abs_input_run
[11630:1299:0621/141120:5062800266650:ERROR:gles2_conform_test.cc(122)] 
gen/gpu/gles2_conform_support/gles2_conform_test_autogen.cc:6: Failure
Value of: RunGLES2ConformTest("GL/abs/input.run")
  Actual: false
Expected: true
[  FAILED  ] GLES2ConformTest.GL_abs_input_run (42 ms)

Same kind of issue as above, just a different dylib/executable:

% /Volumes/Build/src/out/gn-debug/gles2_conform_test_windowless -- -run=GL/abs/input.run
dyld: Library not loaded: ./libcommand_buffer_gles2.dylib
  Referenced from: /Volumes/Build/src/out/gn-debug/gles2_conform_test_windowless
  Reason: image not found
zsh: trace trap  /Volumes/Build/src/out/gn-debug/gles2_conform_test_windowless -- 

I think a spurious dependency snuck in to the GN build that caused that runtime dylib dependency to be created. I don't see that dylib as a load-time dependency in the GYP build. Potential fix: https://codereview.chromium.org/2082793007/

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/66c5718b28d1dff3d05628218fe5207477b9f923

commit 66c5718b28d1dff3d05628218fe5207477b9f923
Author: rsesek <rsesek@chromium.org>
Date: Wed Jun 22 15:25:31 2016

Remove dependency //gpu:command_buffer_gles2 on //gpu/gles2_conform_support/egl.

This causes a run-time dependency on libcommand_buffer_gles2.dylib that
does not exist in GYP.

BUG= 621932 
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel
TBR=piman@chromium.org

Review-Url: https://codereview.chromium.org/2082793007
Cr-Commit-Position: refs/heads/master@{#401284}

[modify] https://crrev.com/66c5718b28d1dff3d05628218fe5207477b9f923/gpu/gles2_conform_support/egl/BUILD.gn

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a51ae8640f42aabb260064275c73706219192049

commit a51ae8640f42aabb260064275c73706219192049
Author: cwallez <cwallez@chromium.org>
Date: Wed Jun 22 17:00:37 2016

Roll ANGLE 55deb36..be5a8a4

https://chromium.googlesource.com/angle/angle.git/+log/55deb36..be5a8a4

BUG=None,621932

TBR=geofflang@chromium.org

TEST=bots

CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.linux:linux_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2085323002
Cr-Commit-Position: refs/heads/master@{#401310}

[modify] https://crrev.com/a51ae8640f42aabb260064275c73706219192049/DEPS

Mac appears to have gone green and I don't *think* there are any other failures directly attributable to my CL.

Thanks Robert. I think this is fixed too.