From bug 528657 comment 4. The idea is to measure the total time the app has been running since the last crash, including multiple runs of the app. This would give us a distribution of times-to-failure, which would allow calculation of MTBF, crashes/time, startup crashes, and other interesting reliability-related stats.

Pretty sure we already track this data... We track a "time since last crash" last I recall, we just don't publish it on the site right now. In any case, if we don't we need to do the tracking part in this bug. If we *do*, this bug should morph to Webtools::Socorro for them to publish the information. (Note that we can already determine start-up crashes with uptime, which is currently reported in Socorro.)

This is different, as described in bug 528657 comment 4. This would be the sum of application runtime since the last crash, not just the absolute time difference. Consider: User runs browser, crashes on startup, doesn't restart. Goes on vacation for a week, comes back, runs browser again, crashes on startup. Results: Time since last crash = 1 week Run time since last crash = 1 second (or so). Uptime doesn't tell the full story either. If a user crashes once, then restarts the browser, runs for a week straight, restarts and crashes after a short while, we'll record a short uptime despite the user having a long browser usage in between. The "time since last crash" can be misleading.

Yeah, I'm saying I thought we had "time since last crash", the one you're talking about. "Uptime" is the same as "run time since last crash", in my book.

No, we currently report: * Time of last crash * Time of current crash * Time application started in current run from which we calculate: * Absolute time since last crash * Uptime of current application session