[Webkit-unassigned] [Bug 119761] Replace currentTime() with monotonicallyIncreasingTime() in all possible places

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=119761





--- Comment #8 from Arunprasad Rajkumar <arurajku at cisco.com>  2013-08-17 11:26:59 PST ---
(In reply to comment #7)
> It almost feels like I'm missing something given the above discussion, but I think that the importance of using monotonicallyIncreasingTime for things like timeouts is that currentTime() can go back, e.g. for DST changes or NTP adjustments.

currentTime() returns UTC time(atleast in linux currentTime uses gettimeofday, which return UTC time), I think only localTime() is prone to DST adjustments. I agree that NTP correction is a problem here.

> 
> You don't want your 100 ms timer to take 1 hour + 100 ms at the moment when daylight savings changes happen.
> 
> Clock time is also more likely to jump forward under normal use conditions, for the same reasons. Of course, your process can be suspended and miss the exact time even when using monotonicallyIncreasingTime.
> 
> Also, I think that "monotonically increasing" may not be precise enough - some monotonically increasing time functions are unrelated to actual time, e.g. they could return raw number of CPU ticks since boot, which of course depends on constantly changing CPU clock speed. But I don't think that naming discussion should block fixing the behavior.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.