Mysterious use counter data

[Philip Jägenstedt]

http://www.chromestatus.com/metrics/feature/timeline/popularity/231

This is DOMImplementationHasFeatureReturnFalse, a counter I added to
check if the DOM spec could be simplified to have hasFeature() always
return true. [1]

Can someone explain how the data could look like this? The initial
climb to around 0.06% is because of the M34 stable release on April 8.
On April 24 it took a dive to around 0.004% again, which seems to
coincide with another stable channel update.

How should this be interpreted? Can someone with internal access have
a look at some channel which is known to have this counter and see
what the numbers look like?

(Whether or not to actually change hasFeature() is a separate
question, and this mail is not an intent to do so.)

Philip

[1] https://www.w3.org/Bugs/Public/show_bug.cgi?id=25428

[Ojan Vafai]

Looking at the Windows stable release (34.0.1847.116):

4/13: 0.09%

4/26: 0.005%

That roughly matches what chromestatus shows. I have no idea what's going on though. The total number of pageviews measured for the two days is pretty close (off by ~15%), so I think it's unlikely that this is due to a bug where we're failing to report pageviews. Happy to gather other data to dig in more. Not sure what other information would be useful.

[Philip Jägenstedt]

I've checked all the counters introduced in M35 (222-273) and none of
them show a similar pattern. One possible explanation is that the data
is accurate, i.e. that a somewhat popular site stopped using
hasFeature. Unless someone has further ideas, maybe it's safest to
just wait for a few weeks to see where the data goes...

Philip