Yeah, the issue is I forgot to account for the no_chromium_code business in making GN build the tests. Will go sort that out.

(The boringssl_*_test.exe binaries were never inheriting from the test template. They weren't being built at all on GN. BoringSSL test setup is different from Chromium's, so those are just straight-up helper binaries. I'm in the middle of trying to get these tests to actually run regularly.)

Do we have no Windows gn try bot? I wonder why this didn't get noticed...

It is annoying that our gn CQ coverage is so light right now. That will be improved very soon.

For the manifest issue you could either ignore the problem (and I will fix it) or follow the pattern in this CL:

https://codereview.chromium.org/1925493002

Again, the manifest issue is not high priority.

I believe https://codereview.chromium.org/1917223004/ will fix the warning/error.

Er, sorry, no it won't (at least, not in a chromium build). It would fix it in a webrtc build :).

To fix it in chromium land we need either the compiler warnings to be fixed or to swap the configs used to compile the code as discussed in comment #1.

We have windows gn trybots, but they don't build 'all', so we didn't catch this. I'm working on fixing that.

Yeah, I already have a CL uploaded to make it so we can swap the configs. It's just waiting on review and then I'll update DEPS and BUILD.gn in src.git to use it.

https://boringssl-review.googlesource.com/#/c/7795/

The following revision refers to this bug:
  https://boringssl.googlesource.com/boringssl.git/+/b3be1cf97da6c0c796f0ab9aebe7232b030d32b0

commit b3be1cf97da6c0c796f0ab9aebe7232b030d32b0
Author: David Benjamin <davidben@google.com>
Date: Wed Apr 27 23:15:06 2016

Add a 'configs_exclude' option to the generated GN 'create_tests' template.

Chromium uses GN's default configs feature which makes all targets default to a
set of configs. It then expects third_party code to take one of them
(chromium_code) out and put in a different one (no_chromium_code).

Because of that, we need a way to tell the template to emit -= lines. Add a
separate option for that.

(It may be worth making us clean against the chromium_code config rather than
the no_chromium_code one, but I'll explore that separately in case making the C
code clean ends up being a rabbithole.)

BUG= chromium:607294 

Change-Id: I2aa179665ab17439cc123fc86a7af9690cd4bcd6
Reviewed-on: https://boringssl-review.googlesource.com/7795
Reviewed-by: Adam Langley <agl@google.com>

[modify] https://crrev.com/b3be1cf97da6c0c796f0ab9aebe7232b030d32b0/util/generate_build_files.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f4ada8f2d80f1c13d7a387d561b620a17f09c491

commit f4ada8f2d80f1c13d7a387d561b620a17f09c491
Author: davidben <davidben@chromium.org>
Date: Thu Apr 28 18:16:51 2016

Roll src/third_party/boringssl/src a08142380..0a63b9653

https://boringssl.googlesource.com/boringssl/+log/a08142380981b366fb4f5eb61f9888f49342d388..0a63b96535dff86fc226e3a13e34252e702a45d0

This also modifies the GN build so the tests build with no_chromium_code.

BUG= 607294 

Review-Url: https://codereview.chromium.org/1931803003
Cr-Commit-Position: refs/heads/master@{#390422}

[modify] https://crrev.com/f4ada8f2d80f1c13d7a387d561b620a17f09c491/DEPS
[modify] https://crrev.com/f4ada8f2d80f1c13d7a387d561b620a17f09c491/third_party/boringssl/BUILD.generated.gni
[modify] https://crrev.com/f4ada8f2d80f1c13d7a387d561b620a17f09c491/third_party/boringssl/BUILD.generated_tests.gni
[modify] https://crrev.com/f4ada8f2d80f1c13d7a387d561b620a17f09c491/third_party/boringssl/BUILD.gn
[modify] https://crrev.com/f4ada8f2d80f1c13d7a387d561b620a17f09c491/third_party/boringssl/boringssl.gypi

This should be fixed now.