[NotLanded] Implement the fetching algorithm of the Web Manifest.

public/platform/WebManifestRequest.h

+// Copyright 2014 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.
+
+#ifndef WebManifestRequest_h
+#define WebManifestRequest_h
+
+namespace blink {
+
+struct WebManifest;
+
+// The WebManifestRequest interface is used when request a WebManifest from a
+// WebFrame. On failure, requestFailed() will be called with a FailureReason.
+// On success, requestSucceeded() will be called with the parsed WebManifest
+// object passed as an argument.
+class WebManifestRequest {

[Nate Chapin, 2014/06/03 18:32:42]

I think part of my confusion about this patch is that this probably shouldn't be
in public/platform/, as it's primarily a callback to notify high level code
about something blink is doing, not a low level API blink uses to do its job.

The way we usually do this is to have a ManifestLoaderClient next to the
ManifestLoader in Source/core/loader/. ManifestLoaderClient is then implemented
in Source/web/ by a ManifestLoaderClientImpl, which notifies a WebManifestClient
in public/web/. :-/

At least, that's how I understand it. eseidel or one of the other public/OWNERS
might have a better idea.

[mlamouri (slow - plz ping), 2014/06/03 20:53:10]

I will ping eseidel. My design sounds simpler but if I'm not doing it right, I
would definitely change it.

+public:
+    enum FailureReason {
+        NoLinkManifest,
+        NetworkError,
+    };
+
+    virtual void requestSucceeded(const WebManifest&) = 0;
+    virtual void requestFailed(FailureReason) = 0;
+};
+
+} // namespace blink
+
+#endif // WebManifestRequest_h