standups: finishing up patch to segment top-level view by channel irccloud.mozilla.com/file/dmtSMkyC/image.png

wlach: that is a _sweet_ view

wow

wlach: where do the change %s come from?

frank: this major release vs. the last one, adjusted for how long it's been out

frank: you can get more data if you click through on the individual cards

wlach: that is beyond cool

wow

frank: yeah I think that is pretty good context to have

wlach: fyi no pretty pictures on mine :( screenshots.firefox.com/n8iTkWre1xi…/data-missioncontrol.dev.mozaws.net

404 on those images

frank: yeah I just realized the icons didn't work in production. I have a PR above to fix that ^^

heh of course you do!

one interesting thing that those numbers suggest is that the current release (59) is doing much better than the previous one (58) was, at least on windows

need to verify that

man so that increase content_shutdown_crashes is pretty obvious

if this view doesn't replace arewestableyet, I don't know what will

wlach: when comparing against the previous release at the same point in time, how is that done?

e.g. is it the sum of all until that point in time in the release

or comparing the snapshots at that point in both releases?

frank: the former. we figure out when the first releases of a new major release went out, and start measuring from then, up to the relative amount that the current release has been available

beautiful

it's kind of confusing, but seems to correct for the fact that things are crashier for the first while after a new release is in the field

wlach: the only other difference being uptake rates, right?

yeah, we don't correct for those either

I have a bit of red warning text for the first bit after a new major release goes out, to warn people that the numbers can't really be trusted yet

I'm wondering what a correction for uptake rate would even look like

e.g. instead of a set time window, the time window up until N total session_hours are reached

so if the previous release had a slower uptake, then it would have a longer time range

anyways total speculation, this is neat

frank: yeah, might be worth exploring at some point-- I am planning on showing the number of accumulated usage hours in various places in the ui which should provide some of this context