Histogram definitions for Stacking Container opt-in measurements.

Histogram definitions for Stacking Container opt-in measurements.

Along with https://codereview.chromium.org/23455031/, this 
adds and makes use of Renderer.CompositedScrolling UMA histogram.

This records the total number of RenderLayers that are scrollable
areas, need to become stacking containers, and will use composited
scrolling.

Each of the buckets are intended to be recorded at most once per 
RenderLayer. The ratio between these stats should give 
us an idea of how effective stacking containers are at 
allowing scrollable divs to become composited.

BUG=

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=221255

[hartmanng, 2013-09-03 19:37:38]

Alexei, please take a look.

[Alexei Svitkine (slow), 2013-09-03 19:41:57]

On 2013/09/03 19:37:38, hartmanng wrote:
> Alexei, please take a look.

Please update the histogram descriptions to mention when those histograms are
sampled (e.g. does it happen once per page load, once every draw, once every
layer creation, or something else?).

[hartmanng, 2013-09-03 19:48:12]

On 2013/09/03 19:41:57, Alexei Svitkine wrote:
> On 2013/09/03 19:37:38, hartmanng wrote:
> > Alexei, please take a look.
> 
> Please update the histogram descriptions to mention when those histograms are
> sampled (e.g. does it happen once per page load, once every draw, once every
> layer creation, or something else?).

done, PTAL.

[Alexei Svitkine (slow), 2013-09-03 19:54:49]

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12687: +<histogram
name="Renderer.IsScrollableArea">
Add enum="Boolean" here and below.

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12691: +    becomes scrollable.
It may be useful to also log the "non-scrollable" case, so that you have a
baseline to compare this too - e.g. it would let you answer "what percentage of
layers become scrollable". (You can do this in the same histogram with a value
of 0).

Otherwise, I'm not sure how you're looking to make sense of the absolute
number...

[hartmanng, 2013-09-03 20:23:06]

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12687: +<histogram
name="Renderer.IsScrollableArea">
On 2013/09/03 19:54:49, Alexei Svitkine wrote:
> Add enum="Boolean" here and below.

Done.

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12691: +    becomes scrollable.
On 2013/09/03 19:54:49, Alexei Svitkine wrote:
> It may be useful to also log the "non-scrollable" case, so that you have a
> baseline to compare this too - e.g. it would let you answer "what percentage
of
> layers become scrollable". (You can do this in the same histogram with a value
> of 0).
> 
> Otherwise, I'm not sure how you're looking to make sense of the absolute
> number...

The goal I have in mind is finding the ratio of scrollable areas that we're able
to promote to stacking containers. Only scrollable areas can become stacking
containers.

So the idea was to just count the absolute number of scrollable areas, and the
absolute number of stacking containers, and then the ratio between those two
numbers is what we're really interested in.

Hopefully that makes more sense, but do let me know if you still think I'm going
about this the wrong way :)

[Alexei Svitkine (slow), 2013-09-03 20:25:51]

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12691: +    becomes scrollable.
On 2013/09/03 20:23:07, hartmanng wrote:
> On 2013/09/03 19:54:49, Alexei Svitkine wrote:
> > It may be useful to also log the "non-scrollable" case, so that you have a
> > baseline to compare this too - e.g. it would let you answer "what percentage
> of
> > layers become scrollable". (You can do this in the same histogram with a
value
> > of 0).
> > 
> > Otherwise, I'm not sure how you're looking to make sense of the absolute
> > number...
> 
> The goal I have in mind is finding the ratio of scrollable areas that we're
able
> to promote to stacking containers. Only scrollable areas can become stacking
> containers.
> 
> So the idea was to just count the absolute number of scrollable areas, and the
> absolute number of stacking containers, and then the ratio between those two
> numbers is what we're really interested in.
> 
> Hopefully that makes more sense, but do let me know if you still think I'm
going
> about this the wrong way :)

Ah, that makes sense. In that case, I think it would be better to combine them
into a single histogram with different enum values for the two cases - so that
they can easily be compared together.

[hartmanng, 2013-09-03 23:34:33]

Thanks! PTAL

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/23874003/diff/5001/tools/metrics/histograms/h...
tools/metrics/histograms/histograms.xml:12691: +    becomes scrollable.
On 2013/09/03 20:25:51, Alexei Svitkine wrote:
> In that case, I think it would be better to combine them
> into a single histogram with different enum values for the two cases - so that
> they can easily be compared together.

Yeah, that makes a lot of sense - thanks for the suggestion! (Done).

I also added a third bucket on the histogram, which maintains a count of the
number of RenderLayers that use composited scrolling. At first, this will be the
same as the count of the ones that become stacking containers (which is why I
wasn't going to add it right now at first). Now that we're combining it all into
one histogram, though, I might as well just add this bucket now too. Soon (after
we land a couple patches in the next week) we'll also be interested in the ratio
of scrollable areas : composited layers.

[Alexei Svitkine (slow), 2013-09-04 15:18:45]

LGTM

[hartmanng, 2013-09-04 15:47:41]

thanks!

[commit-bot: I haz the power, 2013-09-04 16:38:53]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/hartmanng@chromium.org/23874003/19001

[commit-bot: I haz the power, 2013-09-04 20:24:00]

Change committed as 221255