1

I read on msdn that although timers cannot guarantee to fire at the exact interval (in my case 1 second) they will not fire before the interval.

The timers on one pc is working fine (Windows 7) while on the other (Windows Server 2003) fires every 0.99999936 seconds.

I'm really interested in why this is happening.

I noticed this because I had code counting seconds to newSeconds = newSeconds + delta.Seconds Where delta was DateTime.Now - lastTime

The seconds part was showing 1 on Windows 7 and 0 on Windows Server 2003.

Solution was to just read totalseconds, but still I wonder why it's firing before.

Can anyone elaborate on this?

Edit

I actually have it happening on two different windows 2003 pcs. My wondering goes deeper into the areas off is there a difference between os's, is the .net framework 4 different for 7 vs 2003? Or any other deviations people might know of? How are the timers implemented, could it be a hardware related issue?

And as oppose to this one: C# timer getting fired before their interval time I have it happening all the time, on every tick. No need for long running. Thanks

Edit

public void OnTick(object sender, EventArgs e)
{
    var delta = DateTime.Now - _lastTime;

    DoStuff

    _lastTime = DateTime.Now
}
Community
  • 1
  • 1
Dashu
  • 337
  • 2
  • 15
  • Wild guess; difference in rounding in win7 and WS2003. Most likely due to the underlying datatype chosen. – lboshuizen Oct 31 '12 at 01:06
  • Have a look at this: http://stackoverflow.com/questions/2432160/c-sharp-timer-getting-fired-before-their-interval-time – FrostyFire Oct 31 '12 at 01:08
  • Rounding is not an issue. A second has not passed therefore the TimeSpan.Seconds is still 0. – Dashu Oct 31 '12 at 01:21
  • I understand that timers are not accurate, and that reference link was a good post by Hans Passant. Alot of people have "trouble" with the interval being more than a second, but in my case I would expect that behaviour but it's the opposite – Dashu Oct 31 '12 at 01:22
  • `fires every 0.99999936 seconds`? How did you "measure" this? There are some reasons to think about hardware matters but when the deviation is **that small** I bet it is just an accuracy matter, rounding errors accumulated. – Arno Oct 31 '12 at 07:51
  • I "measured" this by looking at the value held in the delta variable (timespan) – Dashu Oct 31 '12 at 12:34

2 Answers2

0

This line:

newSeconds = newSeconds + delta.Seconds

Could and probably will drift further and further away from a true measure. Imagine measuring 1 mile with a yardstick: You will eventually drift off.

The actual time discrepancy may be the result of the small amount of time between when the event fires and when you grab the time. There must be a few CPU cycles in between those two points.

Steve Wellens
  • 20,506
  • 2
  • 28
  • 69
0

I would recommend you to use reactive extensions. Read the "It's all about time" chapter in this blog post.

m0sa
  • 10,712
  • 4
  • 44
  • 91