Use mojo for Chrome Loading, Part 1

content/common/url_loader.mojom

Index: content/common/url_loader.mojom
diff --git a/content/common/url_loader.mojom b/content/common/url_loader.mojom
new file mode 100644
index 0000000000000000000000000000000000000000..ac7a307681ce855480b273f46aca00836df053af
--- /dev/null
+++ b/content/common/url_loader.mojom
@@ -0,0 +1,45 @@
+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+module content.mojom;
+
+[Native]
+struct URLRequest;
+
+[Native]
+struct URLResponseHead;
+
+[Native]
+struct URLLoaderStatus;
+
+interface URLLoader {
+  // Loads the given |request|, asynchronously producing |response|. Consult

[mmenke, 2016/05/25 16:17:20]

|response|?  I don't see anything with that name here.

[yhirano, 2016/05/26 15:42:44]

Ah, sorry, I forgot to update the description when I updated the signature.

+  // |response| to determine if the request resulted in an error, was
+  // redirected, or has a response body to be consumed.
+  // |request_id| is for compatibility with the existing ChromeIPC.

[mmenke, 2016/05/25 16:17:20]

I don't think ChromeIPC is one word?

[yhirano, 2016/05/26 15:42:44]

Done.

+  Load(int32 request_id,

[mmenke, 2016/05/25 16:17:20]

Is there a reason this is "Load" and not "Start".  Think it's best to use
consistent terms across wrapped APIs, unless there's a good reason not to. 
Also, "Load" seems to have more of a "Do this <entire> thing and get back to me"
implication.

[yhirano, 2016/05/26 15:42:44]

Done.

+       URLRequest request,

[mmenke, 2016/05/25 16:17:20]

Hrm...I'm not sure the world can handle yet another class named URLRequest.  Not
that I have a better suggestion.

[yhirano, 2016/05/26 15:42:44]

I'm not particularly fond of names in this file but I don't have better idea
either.

+       URLLoaderClient client);
+
+  // If the request passed to |Load| had |auto_follow_redirects| set to false,
+  // then upon receiving an URLResponse with a non-NULL |redirect_url| field,
+  // |FollowRedirect| may be called to load the URL indicated by the redirect.
+  FollowRedirect();
+
+  // Cancels the request.

[mmenke, 2016/05/25 16:17:20]

Can we still get data after calling Cancel, or does it instantly cut off the
data stream?  I'm assuming the former?

[yhirano, 2016/05/26 15:42:44]

I added some comments. Is it clear now?

+  Cancel();
+};
+
+interface URLLoaderClient {
+  // Called when the response head is received.
+  OnReceiveResponse(URLResponseHead head);
+
+  // Called when the loader starts loading response body.
+  OnStartLoadingResponseBody(handle<data_pipe_consumer> body);
+
+  // Called when the loading completes. No notification will be dispatched for
+  // this client after this message arrives.
+  OnComplete(URLLoaderStatus completion_status);
+};
+