Add UMAs for numerical approximation of table-based transfer functions.

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 6ffdf566a3bbb9aa52c68d745e4fa1dad3359ae4..ba4e0aaf8863a49c0d84a028d6d19eca8e280c6d 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -4409,6 +4409,15 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   <summary>Gamma properties of destination color space.</summary>
 </histogram>
 
+<histogram name="Blink.ColorSpace.Destination.LinearFitError">
+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    The L-infinity error (in 8-bit values) of the numerical approximation of
+    table-based ICC profile transfer functions, when falling back to a linear
+    approximation.
+  </summary>
+</histogram>

[Ilya Sherman, 2017/02/22 22:48:49]

You mentioned in the CL description that at least some of these histograms are
intended to be temporary.  Please indicate this within the <summary> or
<details>, including a target milestone for removal.  It's easy to lose track of
cleanup tasks, so it's helpful to document the intent.

+
 <histogram name="Blink.ColorSpace.Destination.Matrix"
     enum="ColorSpaceMatrixResult">
   <owner>ccameron@chromium.org</owner>
@@ -4418,6 +4427,24 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="Blink.ColorSpace.Destination.NonlinearFitConverged"
+    units="ColorSpaceNonlinearFitConverged">

[Ilya Sherman, 2017/02/22 22:48:49]

nit: s/units/enum

[ccameron, 2017/02/23 00:47:33]

Done.

+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    Whether or not the nonlinear least squares fit of the table-based ICC
+    profile transfer function converged.
+  </summary>
+</histogram>
+
+<histogram name="Blink.ColorSpace.Destination.NonlinearFitError">
+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    The L-infinity error (in 8-bit values) of the numerical approximation of
+    table-based ICC profile transfer functions, when the nonlinear approximation
+    converged.
+  </summary>
+</histogram>
+
 <histogram name="Blink.ColorSpace.Destination.Numerical"
     enum="ColorSpaceNumericalResult">
   <owner>ccameron@chromium.org</owner>
@@ -4427,6 +4454,34 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="Blink.ColorSpace.Destination.RawData"

[Ilya Sherman, 2017/02/22 22:48:48]

nit: Perhaps name this "ExtractedRawData", to clarify from the name that it's a
boolean histogram?

[ccameron, 2017/02/23 00:47:33]

Done.

+    units="ColorSpaceRawDataResult">

[Ilya Sherman, 2017/02/22 22:48:48]

nit: s/units/enum

[ccameron, 2017/02/23 00:47:33]

Done.

+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    Whether or not the output color space ICC profile was able to produce raw
+    transfer function tables.
+  </summary>
+</histogram>
+
+<histogram name="Blink.ColorSpace.Destination.TableSize">

[Ilya Sherman, 2017/02/22 22:48:49]

Please add a units attribute.

[ccameron, 2017/02/23 00:47:33]

Removed this one entirely, since it's less actionable (for the moment).

+  <owner>ccameron@chromium.org</owner>
+  <summary>The size of table-based ICC profile transfer functions.</summary>
+</histogram>
+
+<histogram name="Blink.ColorSpace.Destination.tMax">
+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    The maximum value of table-based ICC profile transfer functions.
+  </summary>
+</histogram>
+
+<histogram name="Blink.ColorSpace.Destination.tMin">
+  <owner>ccameron@chromium.org</owner>
+  <summary>
+    The minimum value of table-based ICC profile transfer functions.

[Ilya Sherman, 2017/02/22 22:48:48]

It looks like these values are actually multiplied by 255; please mention that
here in the histogram description.  Perhaps something similar to the description
of the Error histograms?  (And ditto for tMax.)

[ccameron, 2017/02/23 00:47:33]

Done (8-bit fixed-point).

+  </summary>
+</histogram>
+
 <histogram name="Blink.ColorSpace.Source" enum="Gamma">
   <owner>ccameron@chromium.org</owner>
   <summary>Gamma properties of image color space.</summary>
@@ -82766,11 +82821,21 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   <int value="1" label="Successfully extracted toXYZD50 matrix."/>
 </enum>
 
+<enum name="ColorSpaceNonlinearFitConverged" type="int">
+  <int value="0" label="Nonlinear fit did not converge."/>
+  <int value="1" label="Nonlinear fit converged."/>

[Ilya Sherman, 2017/02/22 22:48:49]

nit: I'd shorten these labels to "Did not converge" and "Converged".

[ccameron, 2017/02/23 00:47:33]

Done.

+</enum>
+
 <enum name="ColorSpaceNumericalResult" type="int">
   <int value="0" label="Failed to extract analytical transfer function."/>
   <int value="1" label="Successfully extracted analytical transfer function."/>
 </enum>
 
+<enum name="ColorSpaceRawDataResult" type="int">
+  <int value="0" label="Failed to extract raw transfer function tables."/>
+  <int value="1" label="Successfully extracted raw transfer function tables."/>

[Ilya Sherman, 2017/02/22 22:48:49]

nit: Likewise, I'd shorten these labels to "Failed to extract" and "Successfully
extracted".

[ccameron, 2017/02/23 00:47:33]

Done.

+</enum>
+
 <enum name="CombinedHttpResponseAndNetErrorCode" type="int">
 <!-- Generated from net/base/net_error_list.h -->