[Pharo-project] System time
avalloud at smalltalk.comcastbiz.net
Mon May 2 21:24:49 CEST 2011
> The way I read it, each _call_ to timeGetTime should be accurate to
> within 5 milliseconds.
The docs say 5ms or more, depending on the machine, and there is no
indication of what affects the precision on the machine. But even if
that's irrelevant, your argument assumes nobody else in the machine
called timerBeginPeriod with an argument of, say, 100. Basically,
relying on timeGetTime is bound to cause problems because it forces you
to assume (guess) what every other program is doing, which opens the
door to wrong assumptions and weird, hard to reproduce behavior.
> The current time is calculated
> baseTime + (newTimeGetTime - baseTimeGetTime) , so precisions not adding
> up does not explain drifts of 36 seconds...
How many times is timeGetTime() called during the accumulated drift?
More information about the Pharo-project