"Datapoint is the median of n tests" view should still include pushlog links about areweslimyet

[staktrace]

I was browsing around the mobile data a bit more today and a lot of the pushlog links and ranges given don't make much sense. For example, I was trying to figure out which change introduced the small bump in "Start" memory usage around Feb 8. After zooming in I see there is a gap in the data, with the last datapoint before the gap at 67.66 MB and the first datapoint after the gap at 68.45 MB. Clicking on the "before" shows one build with the cset 5e735fee82 and clicking on the "after" shows one build with the cset d60526b97a. So now what I want do do is get a pushlog of everything that changed between these two, but there's no link that I can see that does this. At the top of the "after" data box I see this:

68.45MiB Δ N.aNGiB
build 85751b3b03b6 .. d60526b97a17 (pushlog)

which has a number of problems: (1) is the NaN, (2) is that the build at the start of the range is not the last build for which we have data, (3) is that the pushlog link actually links to https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=3f864c99009d&tochange=d60526b97a17 which has a different start than what is displayed.

IMO the text should say "build 5e735fee82 .. d60526b97a17" and the pushlog should link to https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=5e735fee82&tochange=d60526b97a17

[Nephyrin]

The problem is the build range is showing all the builds that have any test data, rather than all the builds that have test data for that line.

As for pushlog links, they link from the last cset of the previous point to the last cset of the current point - the fromchange parameter is exclusive, rather than inclusive. So that part is correct - or would be if we properly excluded builds that have no test data.

I'm working on some patches that will fix this up