Consistently use LoFi for an entire page

chrome/renderer/page_load_histograms.cc

Index: chrome/renderer/page_load_histograms.cc
diff --git a/chrome/renderer/page_load_histograms.cc b/chrome/renderer/page_load_histograms.cc
index 32a8363d6e29d906541d94f98f308695fd4b3afd..6259c0125bfb3d5e1e2e1d64e985956887085e6b 100644
--- a/chrome/renderer/page_load_histograms.cc
+++ b/chrome/renderer/page_load_histograms.cc
@@ -907,12 +907,15 @@ void PageLoadHistograms::Dump(WebFrame* frame) {
 
   MaybeDumpFirstLayoutHistograms();
 
+  content::RenderFrame* render_frame =
+      content::RenderFrame::FromWebFrame(frame);
+
   // Metrics based on the timing information recorded for the Navigation Timing
   // API - http://www.w3.org/TR/navigation-timing/.
-  DumpHistograms(
-      frame->performance(), document_state, data_reduction_proxy_was_used,
-      false /* TODO: render_frame->IsLoFiOn() */, came_from_websearch,
-      websearch_chrome_joint_experiment_id, is_preview, scheme_type);
+  DumpHistograms(frame->performance(), document_state,
+                 data_reduction_proxy_was_used, render_frame->IsUsingLoFi(),

[nasko, 2015/10/01 18:28:57]

Is it possible to get this bit somehow through the DRP instead of RenderFrame?
It will be great if we can keep it internal for now.

[megjablon, 2015/10/01 19:43:25]

DRP doesn't have any idea what the state would be for a frame and only knows
when the state changes from one ShouldEnableLoFiMode call to the next. Asking
the DRP would just keep the same hackiness that we're trying to get rid of.

[nasko, 2015/10/01 20:13:57]

Acknowledged.

+                 came_from_websearch, websearch_chrome_joint_experiment_id,
+                 is_preview, scheme_type);
 
   // Old metrics based on the timing information stored in DocumentState. These
   // are deprecated and should go away.