enabling `gn check` on the whole WebRTC repo

webrtc/BUILD.gn

Index: webrtc/BUILD.gn
diff --git a/webrtc/BUILD.gn b/webrtc/BUILD.gn
index f8e0e8de15ee49cf10059cc0aa4fde27086d8210..1a79a12db22afe2365fbb31e95bc55c0fae5d05f 100644
--- a/webrtc/BUILD.gn
+++ b/webrtc/BUILD.gn
@@ -229,6 +229,12 @@ rtc_source_set("video_stream_api") {
     "video_receive_stream.h",
     "video_send_stream.h",
   ]
+  deps = [
+    ":webrtc_common",
+    "api:transport_api",
+    "base:rtc_base_approved",
+    "common_video:common_video",
+  ]
 }
 
 # Contents of video_frame.h is moved from top-level down to common_video/.
@@ -349,6 +355,13 @@ if (!build_with_chromium) {
 }
 
 rtc_static_library("webrtc_common") {
+  # TODO(mbonadei): Remove (bugs.webrtc.org/7745)
+  # Enabling GN check triggers cyclic dependency error:
+  # //webrtc:webrtc_common ->
+  # //webrtc/api:video_frame_api ->
+  # //webrtc/system_wrappers:system_wrappers ->
+  # //webrtc:webrtc_common
+  check_includes = false
   sources = [
     "common_types.cc",
     "common_types.h",
@@ -361,6 +374,9 @@ rtc_static_library("webrtc_common") {
     # Suppress warnings from the Chromium Clang plugin (bugs.webrtc.org/163).
     suppressed_configs += [ "//build/config/clang:find_bad_constructs" ]
   }
+  deps = [
+    "base:rtc_base_approved",

[kjellander_webrtc, 2017/06/01 13:39:01]

Since we have the check disabled, don't you think it makes sense to remove this
too? Tons of target depends on webrtc_common...

[mbonadei, 2017/06/01 13:44:30]

I don't know... It is still an improvement respect to 0 dependencies.
Let me know what do you prefer, I don't have a strong opinion since
with check_includes = false things will probably move without any
notice.

[kjellander_webrtc, 2017/06/01 14:01:17]

I'm just thinking that since it's something that shouldn't be here in the first
place, maybe the person fixing the dependency forgets to remove it when doing
so.

[mbonadei, 2017/06/01 19:08:48]

I am a little bit confused. It is a real dependency (webrtc/config.h includes
webrtc/base/basictypes.h) so we will not remove it even when the cycle
will be fixed. Am I missing something? :)

[kjellander_webrtc, 2017/06/01 19:45:17]

Leaving it in fine. I just thought we were about to refactor it out, but it'll
probably take a while anyway.

[mbonadei, 2017/06/01 19:55:58]

Ah ok, now I got it. I was missing that webrtc:5876 is about moving
webrtc_common to webrtc/api. :)

+  ]
 }
 
 if (use_libfuzzer || use_drfuzz || use_afl) {