1958023 - App startup crash end timeout is kind of strange

Status: NEW

(Toolkit :: Startup and Profile System, defect)

Description

[:Gijs (he/him)]

We have some code that detects when startup crash tracking "ends" and "normal browser operation" starts, and when we deem the startup "successful". It is used to distinguish startup crashes from others. This code is pretty old.

The thing that sort of puzzled me here is that the scheduling of when that method is invoked is farmed out to consumers, and in Desktop Firefox's case, we do it based off a 30 second setTimeout in the middle of the startup idle tasks. I haven't investigated whether Android or Thunderbird do different things.

ISTM that should either be 0 seconds after all the idle tasks or 30 seconds after some earlier point in startup. I'm also not sure how much of the logic in app startup (the incomplete startup file, the pref saving and how it interacts with session restore...) is in the right place.

Gabriele/Florian, do you have opinions here? Maybe I'm barking up the wrong tree and this is fine as-is?

Comment 1

[Gabriele Svelto [:gsvelto]]

IIRC the rationale was that we wanted to let tabs reload before we stopped that code, in case one would cause a startup crash. That logic is probably steeped in our original single-process model where a tab crash would take down the whole browser. Nowadays that cannot happen so it's definitely something we might revisit.