Refine UMA quality for warm-start app list cases
Currently "warm start" app list cases (i.e. shown via the process singleton) have noisy UMA data. Whether the profile to show is loaded or not has a big impact on the metric. We are only interested in the case that the app list view is "warm", and can be immediately shown. Recording cases that first must load a profile from disk, or switch the app list view to a different profile, count as noise and skew the metric. This change filters out the noise, fixing an incorrect assumption in r270615 which resulted in only the noise being recorded (and a failed DCHECK otherwise). Before r270615, the UMA included cases requiring a profile load as well as those that didn't. After r270615, only cases that required a profile load were being recorded. Note that ShowAppListInteractiveTest.ShowAppListFlag was updated in r281213 to include both cold and warm codepaths via StartupBrowserCreator, to cover the aforementioned DCHECK. BUG=382793 Review URL: https://codereview.chromium.org/366233002 git-svn-id: svn://svn.chromium.org/chrome/trunk/src@282245 0039d316-1c4b-4281-b951-d872f2087c98
Showing
Please register or sign in to comment