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 e0710b1ea0815eacdddbe6b9b234a7154a2d3da0..bbb1563a571129a1d79871e08dd7f9d31a8f7666 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -13476,6 +13476,17 @@ Therefore, the affected-histogram name has to have at least one dot in it.
 </histogram>
 
 <histogram name="Net.FtpServerTypeCount" enum="FtpServerType">
+  <obsolete>
+    Replaced by Net.FtpServerTypeCount2.

[Alexei Svitkine (slow), 2014/06/19 19:02:38]

Nit: Specify a date when the changed landed. Same below.

[davidben, 2014/06/19 19:42:03]

Done.

+  </obsolete>
+  <owner>phajdan.jr@chromium.org</owner>
+  <summary>
+    Each bucket is the number of FTP server types the user has encountered
+    during the session.
+  </summary>
+</histogram>
+
+<histogram name="Net.FtpServerTypeCount2" enum="FtpServerType2">
   <owner>phajdan.jr@chromium.org</owner>
   <summary>
     Each bucket is the number of FTP server types the user has encountered
@@ -13540,6 +13551,17 @@ Therefore, the affected-histogram name has to have at least one dot in it.
 </histogram>
 
 <histogram name="Net.HadFtpServerType" enum="FtpServerType">
+  <obsolete>
+    Replaced by Net.HadFtpServerType2.
+  </obsolete>
+  <owner>phajdan.jr@chromium.org</owner>
+  <summary>
+    Each bucket is a boolean (0 or 1) indicating whether the user has had a

[Alexei Svitkine (slow), 2014/06/19 19:02:38]

Commenting on this description, but comment applies to the new histogram entry,
really.

This description isn't quite right. A bucket isn't a boolean, rather it's a
count of the number connections of that type that occurred. I'd also remove
"during the session" wording, since that's confusing and not quite correct
depending on one's definition of "session".

[Alexei Svitkine (slow), 2014/06/19 19:07:46]

Nevermind, I see the distinction this is trying to capture. Maybe a more clear
description that a given bucket's value is logged with a count of 1 the first
time an FTP server of that type is accessed within a session.

[davidben, 2014/06/19 19:42:03]

Yeah, that wording is really weird. How are these new ones?

+    connection with an FTP server of that type during the session.
+  </summary>
+</histogram>
+
+<histogram name="Net.HadFtpServerType2" enum="FtpServerType2">
   <owner>phajdan.jr@chromium.org</owner>
   <summary>
     Each bucket is a boolean (0 or 1) indicating whether the user has had a
@@ -37647,7 +37669,8 @@ Therefore, the affected-histogram name has to have at least one dot in it.
 
 <enum name="FtpServerType" type="int">
   <summary>
-    FTP server type as defined in net/ftp/ftp_server_type_histograms.h
+    Old FTP server type as previously defined in
+    net/ftp/ftp_server_type_histograms.h

[Alexei Svitkine (slow), 2014/06/19 19:02:38]

Can these be marked obsolete too? I don't recall if enums can be marked
obsolete, but I think they can?

[davidben, 2014/06/19 19:42:03]

Looks like it. Done.

   </summary>
   <int value="0" label="Unknown">
     Unknown (could be a server we don't support, a broken server, or a security
@@ -37667,6 +37690,18 @@ Therefore, the affected-histogram name has to have at least one dot in it.
   </int>
 </enum>
 
+<enum name="FtpServerType2" type="int">
+  <summary>
+    FTP server type as defined in net/ftp/ftp_server_type_histograms.h
+  </summary>
+  <int value="0" label="Unknown"/>
+  <int value="1" label="/bin/ls"/>
+  <int value="2" label="Windows"/>
+  <int value="3" label="VMS"/>
+  <int value="4" label="Netware"/>
+  <int value="5" label="OS/2"/>
+</enum>
+
 <enum name="GaiaSessionRestoreOutcome" type="int">
   <int value="0" label="Undefined"/>
   <int value="1" label="Success"/>