Add histograms to record the severity scores for SSL

Add histograms to record the severity scores for SSL

patch from issue 400323002
BUG=403978

[radhikabhar, 2014-08-15 00:11:54]

@felt - Can you PTAL

[felt, 2014-08-15 15:11:05]

The only real code in this CL is in RecordSSLInterstitialSeverityScore, right?

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_b...
File chrome/browser/ssl/ssl_blocking_page.h (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_b...
chrome/browser/ssl/ssl_blocking_page.h:1: // Copyright (c) 2012 The Chromium
Authors. All rights reserved.
the changes in this class are only because you've patched in the captive portal
CL, right?

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
File chrome/browser/ssl/ssl_error_classification.cc (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
chrome/browser/ssl/ssl_error_classification.cc:74: 0, 100, 20);
is "20" the number of buckets? what kind of granularity does that give us? is it
enough?

[radhikabhar, 2014-08-15 18:28:28]

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_b...
File chrome/browser/ssl/ssl_blocking_page.h (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_b...
chrome/browser/ssl/ssl_blocking_page.h:1: // Copyright (c) 2012 The Chromium
Authors. All rights reserved.
On 2014/08/15 15:11:04, felt wrote:
> the changes in this class are only because you've patched in the captive
portal
> CL, right?

Yes

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
File chrome/browser/ssl/ssl_error_classification.cc (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
chrome/browser/ssl/ssl_error_classification.cc:74: 0, 100, 20);
On 2014/08/15 15:11:04, felt wrote:
> is "20" the number of buckets? what kind of granularity does that give us? is
it
> enough?

That's right 20 is the number of bucket counts. We are expressing it in 
percentage so buckets will be divided according to 0-5,5-10.

Should we wait for some data, or should I increase the bucket count. I think it
is better if we wait for some data to come in.

[felt, 2014-08-15 19:06:19]

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
File chrome/browser/ssl/ssl_error_classification.cc (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
chrome/browser/ssl/ssl_error_classification.cc:74: 0, 100, 20);
On 2014/08/15 18:28:28, radhikabhar wrote:
> On 2014/08/15 15:11:04, felt wrote:
> > is "20" the number of buckets? what kind of granularity does that give us?
is
> it
> > enough?
> 
> That's right 20 is the number of bucket counts. We are expressing it in 
> percentage so buckets will be divided according to 0-5,5-10.
> 
> Should we wait for some data, or should I increase the bucket count. I think
it
> is better if we wait for some data to come in.

my understanding is that it's logarithmic or some such, so the buckets get much
larger closer to 100. is that right?

[radhikabhar, 2014-08-15 20:28:57]

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
File chrome/browser/ssl/ssl_error_classification.cc (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
chrome/browser/ssl/ssl_error_classification.cc:74: 0, 100, 20);
On 2014/08/15 19:06:19, felt wrote:
> On 2014/08/15 18:28:28, radhikabhar wrote:
> > On 2014/08/15 15:11:04, felt wrote:
> > > is "20" the number of buckets? what kind of granularity does that give us?
> is
> > it
> > > enough?
> > 
> > That's right 20 is the number of bucket counts. We are expressing it in 
> > percentage so buckets will be divided according to 0-5,5-10.
> > 
> > Should we wait for some data, or should I increase the bucket count. I think
> it
> > is better if we wait for some data to come in.
> 
> my understanding is that it's logarithmic or some such, so the buckets get
much
> larger closer to 100. is that right? 

I don't think so it is logarathmic

[palmer, 2014-09-09 00:13:51]

palmer@chromium.org changed reviewers:
+ palmer@chromium.org

[palmer, 2014-09-09 00:13:51]

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
File chrome/browser/ssl/ssl_error_classification.cc (right):

https://codereview.chromium.org/474103002/diff/20001/chrome/browser/ssl/ssl_e...
chrome/browser/ssl/ssl_error_classification.cc:74: 0, 100, 20);
It looks like there is a standard macro for counts 1 – 100, so let's just use
that (50 buckets, which should give us very good granularity):

#define UMA_HISTOGRAM_COUNTS_100(name, sample) UMA_HISTOGRAM_CUSTOM_COUNTS( \
    name, sample, 1, 100, 50)

...

#define UMA_HISTOGRAM_CUSTOM_COUNTS(name, sample, min, max, bucket_count) \
    STATIC_HISTOGRAM_POINTER_BLOCK(name, Add(sample), \
        base::Histogram::FactoryGet(name, min, max, bucket_count, \
            base::HistogramBase::kUmaTargetedHistogramFlag))

[palmer, 2014-09-09 00:14:53]

I've put my version of this CL up at https://codereview.chromium.org/549363002