Test alternate Job failing but request succeeding.

Test alternate Job failing but request succeeding.

If the alternate Job fails but the Job to the origin succeeds, the request
should succeed.  This CL adds an HttpNetworkTransaction unittest to document
this.

BUG=474217
Committed: https://crrev.com/40448a53491fa82e1c66c4b0fc82327ea798b76d
Cr-Commit-Position: refs/heads/master@{#329221}

[Bence, 2015-05-08 17:39:01]

bnc@chromium.org changed reviewers:
+ rch@chromium.org

[Bence, 2015-05-08 17:39:02]

Ryan: PTAL.  This is the unittest you proposed in https://crrev.com/1121043002. 
(Note that the case when the connection to origin fails and the alternative one
succeeds is already covered in AltSvcCertificateValidationTest.)

[Ryan Hamilton, 2015-05-08 21:20:26]

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
File net/http/http_network_transaction_unittest.cc (right):

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
net/http/http_network_transaction_unittest.cc:9340:
refused_data.set_connect_data(MockConnect(ASYNC, ERR_CONNECTION_REFUSED));
I was thinking that this would be testing the "don't use http/1" path. Does this
test still pass if you remove the connect data, and use
ssl.SetNextProto(kProtoHTTP11) instead? (Because we should not use the socket
for alt-svc)

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
net/http/http_network_transaction_unittest.cc:9398: }
Looks great. Can you add a second transaction and confirm that it goes over the
existing http/1.1 socket and does not fire up a new alt-svc job (and hence no
new socket).

[Bence, 2015-05-11 16:11:29]

Ryan, PTAL.  Thanks.

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
File net/http/http_network_transaction_unittest.cc (right):

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
net/http/http_network_transaction_unittest.cc:9340:
refused_data.set_connect_data(MockConnect(ASYNC, ERR_CONNECTION_REFUSED));
On 2015/05/08 21:20:26, Ryan Hamilton wrote:
> I was thinking that this would be testing the "don't use http/1" path. Does
this
> test still pass if you remove the connect data, and use
> ssl.SetNextProto(kProtoHTTP11) instead? (Because we should not use the socket
> for alt-svc)

Oh okay.  Sorry it wasn't clear to me from https://crrev.com/1121043002/#msg13.

Done.

https://codereview.chromium.org/1136753002/diff/1/net/http/http_network_trans...
net/http/http_network_transaction_unittest.cc:9398: }
On 2015/05/08 21:20:26, Ryan Hamilton wrote:
> Looks great. Can you add a second transaction and confirm that it goes over
the
> existing http/1.1 socket and does not fire up a new alt-svc job (and hence no
> new socket).

Done.

[Ryan Hamilton, 2015-05-11 18:01:13]

lgtm

[Bence, 2015-05-11 19:06:38]

The CQ bit was checked by bnc@chromium.org

[commit-bot: I haz the power, 2015-05-11 19:07:10]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1136753002/40001

[commit-bot: I haz the power, 2015-05-11 19:13:22]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-05-11 19:15:24]

Patchset 3 (id:??) landed as
https://crrev.com/40448a53491fa82e1c66c4b0fc82327ea798b76d
Cr-Commit-Position: refs/heads/master@{#329221}