Flush out initial [un]registerServiceWorker roundtrip

content/renderer/render_frame_impl.cc

Index: content/renderer/render_frame_impl.cc
diff --git a/content/renderer/render_frame_impl.cc b/content/renderer/render_frame_impl.cc
index 2cec72d37c399fbaeae723f83b4fe5d3e3ebcfca..5984329d9e109f1230be0cfeb6ad193f2d0c6511 100644
--- a/content/renderer/render_frame_impl.cc
+++ b/content/renderer/render_frame_impl.cc
@@ -4,6 +4,8 @@
 
 #include "content/renderer/render_frame_impl.h"
 
+#include <string>
+
 #include "base/command_line.h"
 #include "base/strings/utf_string_conversions.h"
 #include "base/time/time.h"
@@ -213,6 +215,13 @@ WebKit::WebCookieJar* RenderFrameImpl::cookieJar(WebKit::WebFrame* frame) {
   return render_view_->cookieJar(frame);
 }
 
+WebKit::WebServiceWorkerRegistry* RenderFrameImpl::serviceWorkerRegistry(
+    WebKit::WebFrame* frame) {
+  // am I supposed to be using render_view_ here or
+  // RenderViewImpl::FromWebView(frame->view()) ?

[michaeln, 2013/09/30 23:41:27]

I'd vote to make the WebServiceWorkerRegistry registry NOT frame specific and to
just have the a single instance of that thing. Would that simplify things?

Platform::serviceWorkerRegistry() sort of accessor.

[alecflett, 2013/10/01 00:17:04]

I went this way after some consultation with abarth@. I'm not a fan of the
single instance - I don't think it would simplify things as we'd have to keep
our own map and looking up the current origin to find an object that would
otherwise already be attached to the frame or view. 

I would personally like the notion of being per-origin but that concept doesn't
really exist in the WebKit API...  (though maybe it should) I'm expecting to use
the frame as a way to get to the current security origin... I feel like this is
analogous to the ApplicationCacheHost and related classes.

+  return render_view_->serviceWorkerRegistry(frame);
+}
+
 void RenderFrameImpl::didAccessInitialDocument(WebKit::WebFrame* frame) {
   render_view_->didAccessInitialDocument(frame);
 }