Confusing handling of crashes long after they happened

[Matthew Paul Thomas]

We may giggle at Apport for saying that an application crashed "(now or in the past)", but it raises a valid point.

If whoopsie crashes or is killed, so that a subsequent crash log isn't dealt with until Ubuntu next restarts, it does not then make sense to put up the standard application crash alert inviting you to reopen an application that might not have been open for days.

<https://wiki.ubuntu.com/ErrorTracker#error>: "Silent error types ... consist of: ... * Any error that occurs over a minute before the error alert could appear, where you have interacted with the session during that interval."

[Martin Pitt (pitti)]

FWIW, I'm fine with ignoring old errors. I don't think that crash reports are so precious that we need to bother the user about very old ones, we get plenty of them. This sounds similar (dup?) to bug 1033932.

[Matthew Paul Thomas (mpt)]

We should still collect old errors for statistical purposes. But we can treat them as the silent error type, so they get aggregated with the next non-silent error, and we're not bothering people with them individually. Specification updated. <https://wiki.ubuntu.com/ErrorTracker?action=diff&rev2=145&rev1=144>