Did the build files change in between patch set 1 and 2? analyze_details (under step "analyze" in the incorrectly-passing https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win_angle_deqp_rel_ng/3342 ) indicates that the following test targets were analyzed:

    "angle_deqp_egl_tests",
    "angle_deqp_gles2_tests",
    "angle_deqp_gles31_tests",
    "angle_deqp_gles3_tests"

but it's angle_deqp_egl_vulkan_tests which would have caught the bug. Not sure why that target wasn't analyzed.

`angle_deqp_egl_vulkan_tests` as a file doesn't actually exist. Perhaps that's the problem? The test for angle_deqp_egl_vulkan_tests is really running `./angle_deqp_egl_tests --use-angle=vulkan` (at least that's how it's done locally).

Nevertheless, those files should have been rebuilt anyway between patchsets 1 and 2.

Ken, those might be the isolate names. The _vulkan_ tests use the same isolate as the _d3d11_ or _gl_ variants. They specify different arguments and show up as a different test step despite using the same binary.

It's a bit mysterious to me as to why this step no-oped. There certainly were impactful CL changes between revisions 1 and 2. Maybe someone with a bit more recipe expertise could diagnose what happened here.

It's weird that analyze ignored angle_deqp_egl_vulkan_tests. That suite is run on at least one of the waterfall bots it runs: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20dEQP%20Release%20%28NVIDIA%29/14441 shows that test suite running.

Let me look at how those targets are chosen.

A dEQP roll just landed that didn't run any tests. This ended up breaking the GPU.FYI waterfall and ANGLE CQ.

https://chromium-review.googlesource.com/c/angle/angle/+/1304993

https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win_angle_deqp_rel_ng/3349

So I think keeping this pri-1 makes sense.

I'm working on  bug 899908  which is a P0, so I can't look at this right now. Will look more closely when things calm down.

Going to investigate this while martiniss works on the P0.

#0: I'm wondering if something strange w/ gerrit happened in your PS2 runs. For some reason, the output of `gerrit fetch current CL info` only refers to PS1: https://logs.chromium.org/logs/chromium/buildbucket/cr-buildbucket.appspot.com/8931324011033803024/+/steps/gerrit_fetch_current_CL_info/0/logs/json.output/0

#6: that's https://bugs.chromium.org/p/chromium/issues/detail?id=850100 again and is not the same as #0. analyze still doesn't understand the implication of DEPS (https://bugs.chromium.org/p/chromium/issues/detail?id=813932), which is why chromium's DEPS file is in the analyze blacklist: https://codesearch.chromium.org/chromium/src/testing/buildbot/trybot_analyze_config.json?rcl=2720c7f0968163e37dcb41ca7f50fc6d7f850872&l=25

OK, I should probably fix  issue 850100  then. That seems like the more pressing issue. I imagine this could be done by editing the trybot_analyze_config to add an 'angle' section for exclusions?

#10: yep. we'll need to hook that up on the recipe side w/ analyze_names; I can handle that if you want to handle the additional of the angle section to trybot_analyze_config.json?

Sounds good. Patch up at http://crrev.com/c/1307734 . Although I may have been mistaken - I think we might need to update the chromium section because these builds are inside a Chromium checkout + recipe.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/724f1e75e90f94cd6cebfe2e045bd01f031984c6

commit 724f1e75e90f94cd6cebfe2e045bd01f031984c6
Author: Jamie Madill <jmadill@chromium.org>
Date: Tue Oct 30 20:48:40 2018

Add ANGLE's DEPS to trybot exclusions.

This should force DEPS changes in ANGLE to trigger all trybots.
ANGLE's bot config is a bit special since we use recursive DEPS.

Bug:  850100 
Bug:  899737 
Change-Id: Iab6e996663386f66f1fa05f2ee10531d6bd85224
Reviewed-on: https://chromium-review.googlesource.com/c/1307734
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#604005}
[modify] https://crrev.com/724f1e75e90f94cd6cebfe2e045bd01f031984c6/testing/buildbot/trybot_analyze_config.json

jmadill resolved the scarier of the two issues. #0 appears fluky, and I was unable to reproduce despite multiple attempts using the same change (and accidentally modifying the change...).

syoussefi: if you see this again, please reopen. Until then, tentatively WontFixing, as I don't think there's anything to do here.