Trace FrameTreeNode Snapshots

content/browser/frame_host/render_frame_host_impl.cc

Index: content/browser/frame_host/render_frame_host_impl.cc
diff --git a/content/browser/frame_host/render_frame_host_impl.cc b/content/browser/frame_host/render_frame_host_impl.cc
index 39e9fc5c4d805792d3c1c0e2739b5b8ac5143f44..1c4794db14310c76d99a487dc7cc765767b09e89 100644
--- a/content/browser/frame_host/render_frame_host_impl.cc
+++ b/content/browser/frame_host/render_frame_host_impl.cc
@@ -922,6 +922,8 @@ void RenderFrameHostImpl::OnDidCommitProvisionalLoad(const IPC::Message& msg) {
   accessibility_reset_count_ = 0;
   frame_tree_node()->navigator()->DidNavigate(this, validated_params);
 
+  frame_tree_->TraceSnapshot();

[clamy, 2015/10/13 11:48:17]

I would rather have this execute inside NavigatorImpl::DidNavigate. This way we
can have it not executed if the renderer did not actually navigate (eg in
page-navigations).

[benjhayden, 2015/10/22 17:24:21]

I put it after the !did_navigate early return, but there's a comment saying not
to put anything else there, so I'm not sure if I did it right.

[nasko, 2015/10/23 15:52:04]

Now that I think about this. Why do we need to hook this into the FrameTree
itself? Why can't the tracer just be a WebContentsObserver, which tracks
RenderFrameCreated/RenderFrameDeleted/DidFinishNavigation?

If we go that route, it will be possible to even dump trees of nested
WebContents, which is going to be even more awesome for cases of Chrome apps
using <webview>, once we move it to out-of-process iframes.

+
   // For a top-level frame, there are potential security concerns associated
   // with displaying graphics from a previously loaded page after the URL in
   // the omnibar has been changed. It is unappealing to clear the page