Take ownership of net.HttpIdentSrcURL statistic.

Take ownership of net.HttpIdentSrcURL statistic.

Committed: https://crrev.com/38c214270e19d5f95f10dc0173a49311f03966a0
Cr-Commit-Position: refs/heads/master@{#316967}

[Tom Sepez, 2015-02-13 21:55:54]

tsepez@chromium.org changed reviewers:
+ isherman@chromium.org

[Tom Sepez, 2015-02-13 21:55:55]

+isherman for OWNERS review.

[Ilya Sherman, 2015-02-17 22:54:26]

https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:17388: +<histogram
name="net.HttpIdentSrcURL" units="requests">
Please associate an enum rather than a units attribute -- the units attribute is
meant to describe the x-axis, not the y-axis.

https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:17392: +    embedded in the URL itself.
It looks like this is a histogram with only a single bucket.  How do you plan to
interpret the results?  Are you just interested in understanding whether the
number is zero vs. non-zero?

https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:17394: +</histogram>
Was this histogram previous no defined in any histograms.xml file at all, or
does it also need to be removed from the internal copy of the file?

[Tom Sepez, 2015-02-18 00:29:35]

>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> tools/metrics/histograms/histograms.xml:17392: +    embedded in the URL
itself.
> It looks like this is a histogram with only a single bucket.  How do you plan
to
> interpret the results?  Are you just interested in understanding whether the
> number is zero vs. non-zero?
> 
We want a count of N day active users who hit this condition.

What would you label something like this?

>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> tools/metrics/histograms/histograms.xml:17394: +</histogram>
> Was this histogram previous no defined in any histograms.xml file at all, or
> does it also need to be removed from the internal copy of the file?
Not sure.  How would I tell?

[Ilya Sherman, 2015-02-18 01:16:33]

Note: It helps to reply to the comments directly, rather than to the email, for
the sake of nesting when I reply to your replies, and you reply back, and so on.

On 2015/02/18 00:29:35, Tom Sepez wrote:
> >
>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > tools/metrics/histograms/histograms.xml:17392: +    embedded in the URL
> itself.
> > It looks like this is a histogram with only a single bucket.  How do you
plan
> to
> > interpret the results?  Are you just interested in understanding whether the
> > number is zero vs. non-zero?
> > 
> We want a count of N day active users who hit this condition.
> 
> What would you label something like this?

What baseline are you comparing against?  Or are you not too concerned about the
count itself, but mostly just interested in how it shifts over time?  Even
still, I'd expect that you'd want to normalize by number of N day actives, for
some definition of N-day active users; otherwise, the count would increase just
due to user population growth.  So, picking a baseline is critical in order to
have a meaningful metric; and usually, it helps to build that baseline into the
histogram itself.  I can give better guidance if I know what baseline you're
planning to use =)

>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > tools/metrics/histograms/histograms.xml:17394: +</histogram>
> > Was this histogram previous no defined in any histograms.xml file at all, or
> > does it also need to be removed from the internal copy of the file?
> Not sure.  How would I tell?

I tried to do a quick code search, and didn't find it.  I guess I'm curious why
the histogram was added 3+ years ago (if I read git blame history correctly),
but you're only adding the histograms.xml mapping now.  Was this added, never
used, forgotten about, and is now being resurrected because it's become relevant
again?  Or have you being viewing the data already; and if so, how?

[Tom Sepez, 2015-02-18 17:32:03]

On 2015/02/18 01:16:33, Ilya Sherman wrote:
> Note: It helps to reply to the comments directly, rather than to the email,
for
> the sake of nesting when I reply to your replies, and you reply back, and so
on.
> 
> On 2015/02/18 00:29:35, Tom Sepez wrote:
> > >
> >
>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > > tools/metrics/histograms/histograms.xml:17392: +    embedded in the URL
> > itself.
> > > It looks like this is a histogram with only a single bucket.  How do you
> plan
> > to
> > > interpret the results?  Are you just interested in understanding whether
the
> > > number is zero vs. non-zero?
> > > 
> > We want a count of N day active users who hit this condition.
> > 
> > What would you label something like this?
> 
> What baseline are you comparing against?  Or are you not too concerned about
the
> count itself, but mostly just interested in how it shifts over time?  Even
> still, I'd expect that you'd want to normalize by number of N day actives, for
> some definition of N-day active users; otherwise, the count would increase
just
> due to user population growth.  So, picking a baseline is critical in order to
> have a meaningful metric; and usually, it helps to build that baseline into
the
> histogram itself.  I can give better guidance if I know what baseline you're
> planning to use =)
> 

It's to drive feature deprecation.  Or rather, to block the feature deprecation
by showing the feature is still in use.

> >
>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > > tools/metrics/histograms/histograms.xml:17394: +</histogram>
> > > Was this histogram previous no defined in any histograms.xml file at all,
or
> > > does it also need to be removed from the internal copy of the file?
> > Not sure.  How would I tell?
> 
> I tried to do a quick code search, and didn't find it.  I guess I'm curious
why
> the histogram was added 3+ years ago (if I read git blame history correctly),
> but you're only adding the histograms.xml mapping now.  Was this added, never
> used, forgotten about, and is now being resurrected because it's become
relevant
> again?  Or have you being viewing the data already; and if so, how?

This is a case of putting the counter in, and checking it every few years.
https://uma.googleplex.com/p/chrome/histograms/?dayCount=1&histograms=net.Htt...

The immediate concern is that the counter might be removed per policy if there
wasn't a clear owner assigned in the .xml file.

[Ilya Sherman, 2015-02-18 22:17:33]

On 2015/02/18 17:32:03, Tom Sepez wrote:
> On 2015/02/18 01:16:33, Ilya Sherman wrote:
> > Note: It helps to reply to the comments directly, rather than to the email,
> for
> > the sake of nesting when I reply to your replies, and you reply back, and so
> on.
> > 
> > On 2015/02/18 00:29:35, Tom Sepez wrote:
> > > >
> > >
> >
>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > > > tools/metrics/histograms/histograms.xml:17392: +    embedded in the URL
> > > itself.
> > > > It looks like this is a histogram with only a single bucket.  How do you
> > plan
> > > to
> > > > interpret the results?  Are you just interested in understanding whether
> the
> > > > number is zero vs. non-zero?
> > > > 
> > > We want a count of N day active users who hit this condition.
> > > 
> > > What would you label something like this?
> > 
> > What baseline are you comparing against?  Or are you not too concerned about
> the
> > count itself, but mostly just interested in how it shifts over time?  Even
> > still, I'd expect that you'd want to normalize by number of N day actives,
for
> > some definition of N-day active users; otherwise, the count would increase
> just
> > due to user population growth.  So, picking a baseline is critical in order
to
> > have a meaningful metric; and usually, it helps to build that baseline into
> the
> > histogram itself.  I can give better guidance if I know what baseline you're
> > planning to use =)
> > 
> 
> It's to drive feature deprecation.  Or rather, to block the feature
deprecation
> by showing the feature is still in use.
> 
> > >
> >
>
https://codereview.chromium.org/924803003/diff/20001/tools/metrics/histograms...
> > > > tools/metrics/histograms/histograms.xml:17394: +</histogram>
> > > > Was this histogram previous no defined in any histograms.xml file at
all,
> or
> > > > does it also need to be removed from the internal copy of the file?
> > > Not sure.  How would I tell?
> > 
> > I tried to do a quick code search, and didn't find it.  I guess I'm curious
> why
> > the histogram was added 3+ years ago (if I read git blame history
correctly),
> > but you're only adding the histograms.xml mapping now.  Was this added,
never
> > used, forgotten about, and is now being resurrected because it's become
> relevant
> > again?  Or have you being viewing the data already; and if so, how?
> 
> This is a case of putting the counter in, and checking it every few years.
>
https://uma.googleplex.com/p/chrome/histograms/?dayCount=1&histograms=net.Htt...
> 
> The immediate concern is that the counter might be removed per policy if there
> wasn't a clear owner assigned in the .xml file.

Fascinating -- the fact that it shows up on the dashboard means that the
histogram was defined in the .xml file at some point, but then got deleted at
some later point.  Weird.

Anyway, LGTM.  Thanks for the clarifications.

[Tom Sepez, 2015-02-18 23:35:14]

The CQ bit was checked by tsepez@chromium.org

[commit-bot: I haz the power, 2015-02-18 23:36:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/924803003/20001

[commit-bot: I haz the power, 2015-02-19 02:04:27]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-02-19 02:05:12]

Patchset 2 (id:??) landed as
https://crrev.com/38c214270e19d5f95f10dc0173a49311f03966a0
Cr-Commit-Position: refs/heads/master@{#316967}