Some attempts to make PageLoadTest.Reliability faster and hopefully less flak...

Make PageLoadTest.Reliability faster and potentially less flaky in usage 1:

- Remove network requests for live websites, replace them with local resources.
- Lower timeout. Only about:crash hits the timeout, and it does not need such a
  large timeout value.
- Remove page down, as the sleeps between page down calls were adding several
  extra seconds to the test.

BUG=1078496
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=12344

[Patrick Johnson, 2009-03-24 00:18:21]

This should reduce the test time for PageLoadTest.Reliability usage 1 by about
an order of magnitude (from 48 seconds to 4.2 seconds).

http://codereview.chromium.org/42543/diff/1/2
File chrome/test/reliability/page_load_test.cc (right):

http://codereview.chromium.org/42543/diff/1/2#newcode180
Line 180: Sleep(sleep_timeout_ms());
What is the reason we need to wait 2 seconds after each page down key press? Can
this time be reduced?

http://codereview.chromium.org/42543/diff/1/2#newcode306
Line 306: page_down = false;
I'm not completely happy with this, as this means the page down code path won't
be tested by usage 1. However, it does save time in the test (8 seconds).

[huanr, 2009-03-24 04:28:03]

LGTM. Thanks for the change.

http://codereview.chromium.org/42543/diff/1/2
File chrome/test/reliability/page_load_test.cc (right):

http://codereview.chromium.org/42543/diff/1/2#newcode180
Line 180: Sleep(sleep_timeout_ms());
On 2009/03/24 00:18:21, Patrick Johnson wrote:
> What is the reason we need to wait 2 seconds after each page down key press?
Can
> this time be reduced?

Wait for page down to finish in VM environment which will be slow with full page
heap. I plan to make UI control synchronous.