add UMA for no. of pages with no distilled data

components/dom_distiller/core/viewer.cc

Index: components/dom_distiller/core/viewer.cc
diff --git a/components/dom_distiller/core/viewer.cc b/components/dom_distiller/core/viewer.cc
index 6d569c46e6268301695f198b8aa1ce02fa34f589..44c3055277ecd3f8257534302172d6396241c6d6 100644
--- a/components/dom_distiller/core/viewer.cc
+++ b/components/dom_distiller/core/viewer.cc
@@ -9,6 +9,7 @@
 
 #include "base/json/json_writer.h"
 #include "base/memory/scoped_ptr.h"
+#include "base/metrics/histogram.h"
 #include "base/strings/string_util.h"
 #include "components/dom_distiller/core/distilled_page_prefs.h"
 #include "components/dom_distiller/core/dom_distiller_service.h"
@@ -95,6 +96,7 @@ void EnsureNonEmptyTitleAndContent(std::string* title, std::string* content) {
   if (content->empty()) {
     *content = l10n_util::GetStringUTF8(
         IDS_DOM_DISTILLER_VIEWER_NO_DATA_CONTENT);
+    UMA_HISTOGRAM_COUNTS("DomDistiller.Statistics.NoDistilledDataCount", 1);

[nyquist, 2014/09/26 21:52:54]

It doesn't feel right to do this as a histogram to me, given the number is hard
coded. If you  do a histogram, how about counting 0 on the outside and 1 on the
inside? And if you do that, you might want to use a boolean instead? See
UMA_HISTOGRAM_BOOLEAN. I think that makes sense

If you want to do a count in general, you could do that too, but then you would
have to compare it to some other count. Actions are sadly defined in //content
though:
https://chromium.googlesource.com/chromium/src.git/+/master/content/public/br...

Is that why you chose histogram?

Anyway: I vote for boolean about empty and non-empty histogram. That should give
us all the numbers we need.

[kuan, 2014/09/26 22:46:55]

Done.

   }
 }