Re: IndexedDB transactions are no longer durable by default, and other changes

2015-04-04 Thread ben turner (bent)
On Saturday, April 4, 2015 at 9:02:32 AM UTC-7, Jonas Sicking wrote: > > "crash" in that sentence means "kernel panic" not "Gecko crash", right? > > That's correct as far as I understand it. Ben should confirm. Yep, should have said "system crash" or "OS crash" there. -bent _

Re: IndexedDB transactions are no longer durable by default, and other changes

2015-04-04 Thread Jonas Sicking
On Sat, Apr 4, 2015 at 3:15 PM, Gian-Carlo Pascutto wrote: > On 4/04/2015 12:22, Jonas Sicking wrote: > >> I tried to explain this, but I don't think it was particularly clear. >> >> Once the "complete" event has fired, the data has been completely >> transferred to the OS. At that point, any geck

Re: IndexedDB transactions are no longer durable by default, and other changes

2015-04-04 Thread Gian-Carlo Pascutto
On 4/04/2015 12:22, Jonas Sicking wrote: > I tried to explain this, but I don't think it was particularly clear. > > Once the "complete" event has fired, the data has been completely > transferred to the OS. At that point, any gecko process crashing or > getting killed will not be a problem. The

Re: IndexedDB transactions are no longer durable by default, and other changes

2015-04-04 Thread Jonas Sicking
On Fri, Apr 3, 2015 at 11:47 PM, Gian-Carlo Pascutto wrote: > >> The IndexedDB API does not currently have a way to say "no, really, I >> want to make sure that this important data is saved to disk before I >> continue". > > Do our internal APIs offer this? > > Android can kill Firefox at any time