Most Efficient Timer?

[Alex Tweedly]

At 13:33 29/11/2004 -0500, Frank D. Engel, Jr. wrote:

>This helps to avoid another problem as well.  If a one-second timer is 
>started at, say, 1:31:32, then the minute will be about half-over by the 
>time the display is updated, so that the time display is only accurate 
>about half of the time.
>
>Of course, the timer may drift somewhat if there is a delay in 
>message-processing, depending on how it has been configured...

You can solve both of those problems by doing something like

     send myTimer to me in (1000 - (the millisecs mod 1000) ) millisecs

OK - it looks ugly, but it should line you up onto the second boundaries 
quite simply.

In fact, the following script (with the addition of the obvious scrolling 
field, and a "Stop" button") shows that there is some systematic drift of 
up to 8 milliseconds (on my very SLOW Win2000 laptop).

>global gStop
>
>on mouseUp
>   put 0 into gStop
>   put empty into field "Field"
>   send myTimer
>end mouseUp
>
>on myTimer
>   put (the millisecs mod 1000) into t
>   put t & TAB & the secs & TAB & the millisecs & cr after field "Field"
>   if gStop = 0 then
>     send myTimer to me in (999 - (the millisecs mod 1000) ) millisecs
>   end if
>end myTimer

-- Alex.