Roll mojo SDK to 734c6e1652ff2f3b696e441722838f453f4f9b42

Roll mojo SDK to 734c6e1652ff2f3b696e441722838f453f4f9b42

- Remove usage of interface_impl.h
- Implements URLRequest::only_from_cache

R=ppi@chromium.org, viettrungluu@chromium.org

Committed: https://github.com/domokit/monet856b05974956fc1dfc850d8c296be96c17bbc816

[viettrungluu, 2015-07-09 16:41:53]

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/cooki...
File mojo/services/network/cookie_store_impl.h (right):

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/cooki...
mojo/services/network/cookie_store_impl.h:31: mojo::StrongBinding<CookieStore>
binding_;
Here and elsewhere: I have a slight preference for having binding_ be declared
first, basically matching the order of the arguments to the constructor
(logically, the request "gets turned into" a binding).

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/tcp_b...
File mojo/services/network/tcp_bound_socket_impl.h (right):

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/tcp_b...
mojo/services/network/tcp_bound_socket_impl.h:17:
TCPBoundSocketImpl(InterfaceRequest<TCPBoundSocket> request);
explicit

[qsr, 2015-07-10 07:53:17]

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/cooki...
File mojo/services/network/cookie_store_impl.h (right):

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/cooki...
mojo/services/network/cookie_store_impl.h:31: mojo::StrongBinding<CookieStore>
binding_;
On 2015/07/09 16:41:52, viettrungluu wrote:
> Here and elsewhere: I have a slight preference for having binding_ be declared
> first, basically matching the order of the arguments to the constructor
> (logically, the request "gets turned into" a binding).

Done. But I personally have a preference for the other way around. My reasoning
is that a bound Binding has a reference to the object, and if something makes it
do a call, it should not happen with an partially destructed object. Now, I
agree that this should not happen, but having it destroyed first ensure it will
not.

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/tcp_b...
File mojo/services/network/tcp_bound_socket_impl.h (right):

https://codereview.chromium.org/1227373002/diff/1/mojo/services/network/tcp_b...
mojo/services/network/tcp_bound_socket_impl.h:17:
TCPBoundSocketImpl(InterfaceRequest<TCPBoundSocket> request);
On 2015/07/09 16:41:52, viettrungluu wrote:
> explicit

Done.

[qsr, 2015-07-10 15:53:05]

I also updated the error handlers on mojo/services/network.

[qsr, 2015-07-15 12:33:11]

gentle ping?

[qsr, 2015-07-16 12:18:32]

qsr@chromium.org changed reviewers:
+ ppi@chromium.org

[ppi, 2015-07-16 12:35:55]

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
File mojo/services/network/url_loader_impl.cc (right):

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
mojo/services/network/url_loader_impl.cc:271:
interceptor.set_connection_error_handler([this]() { delete this; });
Is this intentional change that we now `delete this;` instead of calling
OnConnectionError()?

[qsr, 2015-07-16 12:37:18]

On 2015/07/16 12:35:55, ppi wrote:
>
https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
> File mojo/services/network/url_loader_impl.cc (right):
> 
>
https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
> mojo/services/network/url_loader_impl.cc:271:
> interceptor.set_connection_error_handler([this]() { delete this; });
> Is this intentional change that we now `delete this;` instead of calling
> OnConnectionError()?

Yes, the other behavior was a bug. If an interaceptor is disconnected, this
class should go away, interrupting any current request.

[ppi, 2015-07-16 12:39:46]

lgtm % comment below

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
File mojo/services/network/url_loader_impl.cc (right):

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
mojo/services/network/url_loader_impl.cc:271:
interceptor.set_connection_error_handler([this]() { delete this; });
On 2015/07/16 12:35:54, ppi wrote:
> Is this intentional change that we now `delete this;` instead of calling
> OnConnectionError()?

I'd suggest to make this fix separately for history clarity.

[qsr, 2015-07-16 12:41:40]

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
File mojo/services/network/url_loader_impl.cc (right):

https://codereview.chromium.org/1227373002/diff/70001/mojo/services/network/u...
mojo/services/network/url_loader_impl.cc:271:
interceptor.set_connection_error_handler([this]() { delete this; });
On 2015/07/16 12:39:45, ppi wrote:
> On 2015/07/16 12:35:54, ppi wrote:
> > Is this intentional change that we now `delete this;` instead of calling
> > OnConnectionError()?
> 
> I'd suggest to make this fix separately for history clarity.

Done.

[qsr, 2015-07-16 12:42:01]

Committed patchset #6 (id:90001) manually as
856b05974956fc1dfc850d8c296be96c17bbc816 (presubmit successful).