Fixed flakiness in WebGL conformance, and presumably other tests, under Telemetry.

Fixed flakiness in WebGL conformance, and presumably other tests, under Telemetry.

Made the ChromeViewHostMsg_CloseCurrentConnections message synchronous so
that Telemetry can't issue further network requests until it's processed.

Removed call to util.CloseConnections from _RunCompoundAction in
page_test.py. This was aborting resource requests issued by the current
test, and was the root cause of the flakiness of the WebGL conformance
tests that loaded external resources.

BUG=280750
R=dtu@chromium.org, palmer@chromium.org, tonyg@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=221220

[Ken Russell (switch to Gerrit), 2013-09-03 23:27:40]

palmer: benchmarking_messages.h review please.

[tonyg, 2013-09-04 00:02:24]

lgtm

https://codereview.chromium.org/23533029/diff/1/chrome/common/benchmarking_me...
File chrome/common/benchmarking_messages.h (right):

https://codereview.chromium.org/23533029/diff/1/chrome/common/benchmarking_me...
chrome/common/benchmarking_messages.h:22:
IPC_SYNC_MESSAGE_CONTROL0_0(ChromeViewHostMsg_CloseCurrentConnections)
lgtm as long as you've verified that the comment stating this is only used for
debugging/testing is still accurate.

https://codereview.chromium.org/23533029/diff/1/tools/telemetry/telemetry/pag...
File tools/telemetry/telemetry/page/page_test.py (right):

https://codereview.chromium.org/23533029/diff/1/tools/telemetry/telemetry/pag...
tools/telemetry/telemetry/page/page_test.py:200: # Note that we must not call
util.CloseConnections here. Many tests
My preference would be to just delete it without adding the comment. But up to
you.

[dtu, 2013-09-04 00:04:22]

lgtm

[palmer, 2013-09-04 00:33:22]

lgtm

[Ken Russell (switch to Gerrit), 2013-09-04 17:18:59]

https://codereview.chromium.org/23533029/diff/1/chrome/common/benchmarking_me...
File chrome/common/benchmarking_messages.h (right):

https://codereview.chromium.org/23533029/diff/1/chrome/common/benchmarking_me...
chrome/common/benchmarking_messages.h:22:
IPC_SYNC_MESSAGE_CONTROL0_0(ChromeViewHostMsg_CloseCurrentConnections)
On 2013/09/04 00:02:24, tonyg wrote:
> lgtm as long as you've verified that the comment stating this is only used for
> debugging/testing is still accurate.

The comment's still accurate. This message is ignored unless the net
benchmarking extension's been enabled.

https://codereview.chromium.org/23533029/diff/1/tools/telemetry/telemetry/pag...
File tools/telemetry/telemetry/page/page_test.py (right):

https://codereview.chromium.org/23533029/diff/1/tools/telemetry/telemetry/pag...
tools/telemetry/telemetry/page/page_test.py:200: # Note that we must not call
util.CloseConnections here. Many tests
On 2013/09/04 00:02:24, tonyg wrote:
> My preference would be to just delete it without adding the comment. But up to
> you.

I've given this more thought and am leaving the comment in place. If this call
is added here the only consequence is flakiness, which is worse than a
deterministic test failure, so a breadcrumb to prevent accidental
re-introduction later is important.

[Ken Russell (switch to Gerrit), 2013-09-04 17:26:37]

Committed patchset #1 manually as r221220 (presubmit successful).