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 5265c80f2b4c090461ae32840ccadf8dee69e850..c465b102bb197658aa39d29dac6e37739268432b 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -9220,6 +9220,15 @@ Therefore, the affected-histogram name has to have at least one dot in it.
   <summary>Chrome OS login success reason.</summary>
 </histogram>
 
+<histogram name="Login.UsersActiveWeekly" units="users">
+  <owner>alemate@chromium.org</owner>
+  <owner>nkostylev@chromium.org</owner>
+  <summary>
+    Chrome OS histogram that keeps track of number of users who has logged in in

[Ilya Sherman, 2014/05/30 23:03:55]

nit: "users who has" -> "users who have"

[Alexander Alekseev, 2014/05/30 23:20:39]

Done.

+    the last 7 days. Reported on every boot and once a day after that.

[Ilya Sherman, 2014/05/30 23:03:55]

Note that this will bias your metric toward users who reboot more frequently. 
Is that intentional/acceptable?

[Alexander Alekseev, 2014/05/30 23:20:39]

This metric actually counts number of user home directories with modification
time less than 7 days. So the actual value doesn't depend on report time.

What is more important, it is always reported with total number of user
home directories. So these two metrics match.

[Ilya Sherman, 2014/05/30 23:24:35]

My concern is that the metric is *reported* on every boot.  Hence, users who
reboot more frequently will contribute more samples than users who reboot less
frequently.

However, if your goal is to match an existing metric that already has this bias,
then LGTM.

+  </summary>
+</histogram>
+
 <histogram name="Login.UserType" enum="LoginUserType">
   <owner>cmasone@chromium.org</owner>
   <summary>