Adds browser-side component of Java Bridge for a TabContents

content/browser/tab_contents/tab_contents.cc

Index: content/browser/tab_contents/tab_contents.cc
diff --git a/content/browser/tab_contents/tab_contents.cc b/content/browser/tab_contents/tab_contents.cc
index 371cf891217f87e543b765d0bb2a5fd743d8f085..048864ff02d2442f8911525b84eb7731d80ec32b 100644
--- a/content/browser/tab_contents/tab_contents.cc
+++ b/content/browser/tab_contents/tab_contents.cc
@@ -210,6 +210,9 @@ TabContents::TabContents(content::BrowserContext* browser_context,
   // tab contents (normally a tab from the same window).
   view_->CreateView(base_tab_contents ?
       base_tab_contents->view()->GetContainerSize() : gfx::Size());
+
+  java_bridge_dispatcher_host_manager_.reset(

[jam, 2011/10/21 16:18:21]

we don't need to add this for all of chrome, do we? i.e. it seems we don't need
to link this in unless a port is using this java feature

[Steve Block, 2011/10/21 16:27:48]

No, this isn't needed unless we want to use the Java Bridge. I was going to ask
about this.

What's the accepted practice regarding flags? The complete feature will only
build on platforms with a JVM present, so we'll need a compile-time flag. Should
I add a run-time flag too? If so, I guess it makes sense to guard as little as
possible with the compile-time flag, and use the run-time flag where possible?

+      new JavaBridgeDispatcherHostManager(this));
 }
 
 TabContents::~TabContents() {
@@ -1471,6 +1474,7 @@ void TabContents::RenderViewGone(RenderViewHost* rvh,
 
 void TabContents::RenderViewDeleted(RenderViewHost* rvh) {
   render_manager_.RenderViewDeleted(rvh);
+  FOR_EACH_OBSERVER(TabContentsObserver, observers_, RenderViewDeleted(rvh));
 }
 
 void TabContents::DidNavigate(RenderViewHost* rvh,