tools/metrics/histograms/histograms.xml

Index: tools/metrics/histograms/histograms.xml
diff --git a/tools/metrics/histograms/histograms.xml b/tools/metrics/histograms/histograms.xml
index ba8bf977d74849dcaf4ae68e7a3e32354bbb3ebc..3f730f85fc54a361ab2aa1d2d4883448db166028 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -70541,6 +70541,17 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="SiteInstance.ReusePendingOrComittedSiteCouldReuse"

[Ilya Sherman, 2017/06/07 19:53:15]

It looks like you're adding a new top-level category name, "SiteInstance".  This
might be appropriate, but I wanted to ask: Is there an existing top-level
category that it would make sense to record this metric under?  I'm a bit
protective of the global namespace =)

[Ilya Sherman, 2017/06/07 19:53:15]

Optional nit: I'd add a dot between "Site" and "Could"

[Charlie Reis, 2017/06/07 20:15:33]

We don't have a ProcessModel category, which might make sense at some point. 
It's most similar to metrics we track in SiteIsolation (e.g.,
SiteIsolation.*ProcessCountEstimate) or Memory (e.g.,
Memory.RendererProcessCount).

I could see using SiteIsolation here, even though it's not strictly about that.

On 2017/06/07 19:53:15, Ilya Sherman wrote:
+1

[clamy, 2017/06/13 15:04:47]

Done.

[clamy, 2017/06/13 15:04:47]

I have put the histogram in SiteIsolation.

+    enum="BooleanEnabled">
+  <owner>clamy@chromium.org</owner>
+  <summary>
+    Whether SiteInstances with a ProcessReusePolicy of
+    REUSE_PENDING_OR_COMMITTED_SITE could reuse an existing RenderProcessHost or
+    they had to create a new one. True when they reused an existing process,
+    false when they created a new on.

[Ilya Sherman, 2017/06/07 19:53:15]

nit: s/on/one

[clamy, 2017/06/13 15:04:47]

Done.

+  </summary>
+</histogram>
+
 <histogram name="SiteIsolation.AllResponses">
   <owner>creis@chromium.org</owner>
   <summary>