Network connectivity - packet pacing and FEC tests for packet loss.

Network connectivity - packet pacing and FEC tests for packet loss.

After finishing the regular experiment for 21 packets loss test, we
start one of the following two new tests.

1) Pace the packets. The delay between the packets is calculated from
the total time to receive packets in the first experiment (total time
is the time from when we recieved the first packet to the time it has
taken to recieve the last packet). If the total time is 20 secs, then
we send each packet with a delay of 1 sec.

2) Non-paced test. Will send packets as soon as possible.

Deleted TCPStatsClient class and thus no TCP connectivity
tests. Merged UDPStatsClient into NetworkStats class.

R=jar
TEST=unit tests

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=155111

[ramant (doing other things), 2012-08-31 19:57:56]

Hi Jim,
  Made the changes we had discussed for packet loss testing. Would appreciate if
you could take a look at these changes.

thanks much
raman

[jar (doing other things), 2012-09-05 01:00:55]

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
File chrome/browser/net/network_stats.cc (right):

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:84: static const int32 kPorts[] = {6121,
999999};
Is this a sentinel? If so, use a named constant to check for it.

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:223: const int kSocketBufferSize =
packets_to_send_ * 2048;
Perhaps we want this to be double the packets_to_send_, as we commonly send a
first set of 21... and then repeat for a second set.  We probably won't proceed
to the second test until we get all the first-set responses... but if something
got "stuck," we should be able to handle as many as 42.

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:550: std::string load_size_string =
base::StringPrintf("%dB", load_size_);
nit: We only use this on line 556, so we might as well use this format string
%dB on line 552, and pass in load_size_.

[ramant (doing other things), 2012-09-05 04:28:54]

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
File chrome/browser/net/network_stats.cc (right):

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:84: static const int32 kPorts[] = {6121,
999999};
On 2012/09/05 01:00:55, jar wrote:
> Is this a sentinel? If so, use a named constant to check for it.

999999 is used by the unit tests. Added a comment to indicated port 999999 is
used for unit tests.
Done.

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:223: const int kSocketBufferSize =
packets_to_send_ * 2048;
On 2012/09/05 01:00:55, jar wrote:
> Perhaps we want this to be double the packets_to_send_, as we commonly send a
> first set of 21... and then repeat for a second set.  We probably won't
proceed
> to the second test until we get all the first-set responses... but if
something
> got "stuck," we should be able to handle as many as 42.

Done.

http://codereview.chromium.org/10913037/diff/6004/chrome/browser/net/network_...
chrome/browser/net/network_stats.cc:550: std::string load_size_string =
base::StringPrintf("%dB", load_size_);
On 2012/09/05 01:00:55, jar wrote:
> nit: We only use this on line 556, so we might as well use this format string
> %dB on line 552, and pass in load_size_.

Done.

[jar (doing other things), 2012-09-06 00:36:06]

lgtm

[commit-bot: I haz the power, 2012-09-06 00:38:13]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rtenneti@chromium.org/10913037/14002

[commit-bot: I haz the power, 2012-09-06 03:09:30]

Change committed as 155111