🚀 go-pugleaf

Read the doco ... SetTimer only has 55ms resolution (rounded up)
so (int)(1000/30) = 33 , which becomes 55ms which means if the
WM_TIMER message is never left in the queue then 90 WM_TIMER
msgs will take approx. 5 sec.  These are *not* real time OSs.
You should always check clock-ticks if accuracy is important. If you
really require 33 ms timers use multi-media timers, they can trigger
at 1ms intervals (but have associated overheads).

Good Luck

ChrisB2

Robin Burrows wrote in message <3619553C.11D5@bigfoot.com>...
>Hello,
>
>I'm trying to set up a timer with the following line:
>SetTimer (hwnd, 1, 1000/30, NULL);
>(30 interrupts per second) but the interval is nothing like that. For
>example, I process 90 interrupts which should take 3 seconds, but it
>takes more like 15. I am using a fprintf in my WM_TIMER, but the file
>remains open so the buffering should absorb it, there is no other
>lengthy code in there. What is the minimum gap between interrupts? Or am
>I just being dense? Do I really have to use GetTickCount and compensate
>for it myself?
>

A couple of things:

1) The standard timer is driven by the standard BIOS timer tick, so you
can't generate more than 18.2 timer events per second using SetTimer. To go
faster, you'll have to use the multimedia timers (see the timeSet*
functions, etc.)

2)If the timer generates another tick before your app has processed the last
WM_TIMER message, it will NOT put another one on the queue. Also, WM_TIMER
messages are low priority, and will only be given to your app if there are
no other messages on the queue.

--
//
// reply to:
// BgPorter at acm dot org
//
Robin Burrows wrote in message <3619553C.11D5@bigfoot.com>...
>Hello,
>
>I'm trying to set up a timer with the following line:
>SetTimer (hwnd, 1, 1000/30, NULL);
>(30 interrupts per second) but the interval is nothing like that. For
>example, I process 90 interrupts which should take 3 seconds, but it
>takes more like 15. I am using a fprintf in my WM_TIMER, but the file
>remains open so the buffering should absorb it, there is no other
>lengthy code in there. What is the minimum gap between interrupts? Or am
>I just being dense? Do I really have to use GetTickCount and compensate
>for it myself?
>

Robin Burrows wrote:
> I am using a fprintf in my WM_TIMER, but the file
> remains open so the buffering should absorb it, there is no other
> lengthy code in there. What is the minimum gap between interrupts? Or am
> I just being dense? Do I really have to use GetTickCount and compensate
> for it myself?


Greetings.

I share the opinions of the people who already responded. I'ld like to
add though that I find this a strange way to print. Probably your
concern is to to not block the main thread - which would result in
'blank' pieces of screen waiting to be updated during the printing
process, and the user would not be able to do anything. There are
several other, proper, ways to take case of this. Do the printing in
another thread. This is by far the best way, but also not very easy if
you didn't do this before. The ProcessMessages is a far easier way out:
encapsulate the timer code in in a loop, and do a ProcessMessages after
each loop.



Joris Van Damme
ecogreen.joris.vandamme@ping.be