Basic network service implementation for iOS

Basic network service implementation for iOS

R=eseidel@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/68ecbedbcb2c3b96bf27bb981074e39e51f0c7b0

[Chinmay, 2015-06-08 20:42:44]

chinmaygarde@google.com changed reviewers:
+ eseidel@google.com

[Chinmay, 2015-06-08 20:42:45]

@abarth: I am having trouble properly submitting the buffered contents when
|WriteDataRaw| returns anything other than success. Will post a subsequent
review addressing that FIXME.

[eseidel, 2015-06-08 20:49:36]

eseidel@chromium.org changed reviewers:
+ eseidel@chromium.org

[eseidel, 2015-06-08 20:49:36]

Filed a bug about this using Obj-C: https://github.com/domokit/mojo/issues/218

[eseidel, 2015-06-08 20:53:53]

This seems reasonable.  Please do an update with the mojo:: changes before
landing.

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/network...
File sky/services/ns_net/network_service_impl.h (right):

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/network...
sky/services/ns_net/network_service_impl.h:21: mojo::NetAddressPtr
local_address,
We're in the mojo: namespace, we shouldn't need to use mojo:: inside the mojo
namespace.

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/network...
File sky/services/ns_net/network_service_impl.mm (right):

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/network...
sky/services/ns_net/network_service_impl.mm:13:
mojo::InterfaceRequest<mojo::URLLoader> loader) {
Same problem here.  No need for mojo:: inside namespace mojo {

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/url_loa...
File sky/services/ns_net/url_loader_impl.mm (right):

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/url_loa...
sky/services/ns_net/url_loader_impl.mm:18: mojo::URLResponsePtr _response;
Is objc++ smart enough to run destructors on these?  I would think so, but it's
been probably 10 years since I wrote Obj-c++

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/url_loa...
sky/services/ns_net/url_loader_impl.mm:136:
(URLLoaderConnectionDelegate*)connection_delegate_;
Are you sure you want to c-style cast here?

[Chinmay, 2015-06-08 22:03:43]

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/url_loa...
File sky/services/ns_net/url_loader_impl.mm (right):

https://codereview.chromium.org/1165283002/diff/1/sky/services/ns_net/url_loa...
sky/services/ns_net/url_loader_impl.mm:18: mojo::URLResponsePtr _response;
On 2015/06/08 at 20:53:53, eseidel wrote:
> Is objc++ smart enough to run destructors on these?  I would think so, but
it's been probably 10 years since I wrote Obj-c++

Yeah, unless `-fno-objc-call-cxx-cdtors` is specified

[eseidel, 2015-06-08 22:13:54]

lgtm

https://codereview.chromium.org/1165283002/diff/20001/sky/services/ns_net/url...
File sky/services/ns_net/url_loader_impl.mm (right):

https://codereview.chromium.org/1165283002/diff/20001/sky/services/ns_net/url...
sky/services/ns_net/url_loader_impl.mm:119: [(id)pending_connection_ release];
Why is (id) necessary here?

Do we have a smart pointer we should be using instead?  RetainPtr<> perchance?

[vtl, 2015-06-09 16:59:55]

On 2015/06/08 22:13:54, eseidel wrote:
> lgtm
> 
>
https://codereview.chromium.org/1165283002/diff/20001/sky/services/ns_net/url...
> File sky/services/ns_net/url_loader_impl.mm (right):
> 
>
https://codereview.chromium.org/1165283002/diff/20001/sky/services/ns_net/url...
> sky/services/ns_net/url_loader_impl.mm:119: [(id)pending_connection_ release];
> Why is (id) necessary here?
> 
> Do we have a smart pointer we should be using instead?  RetainPtr<> perchance?

See base/mac/scoped_nsobject.h.

[Chinmay, 2015-06-10 23:16:13]

Committed patchset #2 (id:20001) manually as
68ecbedbcb2c3b96bf27bb981074e39e51f0c7b0 (presubmit successful).