Author Topic: [TIGCC] Key Testing Too Fast? (Read 6865 times)

bfr · « **on:** July 31, 2006, 05:09:00 am »

I installed TIGCC .96 Beta 7, and built and ran a program that I had built and ran in TIGCC .96 Beta 6. But when I tested it out, it seemed that key presses were detected way to quickly, even after putting for(d = 0;d<200000;d++){} at the end of the key test loop to slow it down. When i built it with TIGCC .96 Beta 6 though, the key presses were detected at the normal speed. Would anybody know why this is happenning?

I'm using _keytest() to test for key presses, if that helps.

Liazon · « **Reply #1 on:** July 31, 2006, 09:00:00 am »

i'll check it out. Thanks for telling me about a new beta though.

Hopefully this isn't a bug.

edit:

hmm.... well, so far in VTI, my program seems fine.

MathStuf · « **Reply #2 on:** July 31, 2006, 09:18:00 am »

Umm...that sounds like something on the calc... I recompiled Cobra Command and it worked fine with _keytest. GCC may optimize empty loops now (unless in another function if it does), but I don't know why it would.

Liazon · « **Reply #3 on:** July 31, 2006, 11:19:00 am »

I typically generate random numbers instead of use an empty loop. It's something I read in a technoplaza tutorial.

DJ Omnimaga · « **Reply #4 on:** July 31, 2006, 11:46:00 am »

is that the new version of tigcc?

bfr · « **Reply #5 on:** August 01, 2006, 05:06:00 am »

TIGCC .96 Beta 7 is the latest beta version of TIGCC.

Well, I also tried it with a timeout type of thing. The actions that occurred when keys were press could only happen when the timeout variable had a certain value. Here is psuedo-code for what I mean:

c1-->

CODE

ec1
timeout = 0;
loop
if(_keytest(RR_UP) && timeout=29){
//do stuff
}
timeout++;
if(timeout==50)timeout=0;
endloop
c2

ec2

But I got the same results as when I did the delay thing. I guess I'll try the random number thing.

MathStuf · « **Reply #6 on:** August 04, 2006, 08:07:00 am »

Try for(d = 0;d<200000;d++);. That's how I make delay loops instead of empty braces. GCC may get rid of empty brace sets now, but not empty statements.

bfr · « **Reply #7 on:** August 04, 2006, 11:34:00 am »

OK, thanks.

I'll try that when I get on my computer that has TIGCC installed...

bfr · « **Reply #8 on:** August 12, 2006, 06:34:00 am »

I finally got it to work. I put:

b = random(20);

inside the loop. Thanks, everybody

Liazon · « **Reply #9 on:** August 13, 2006, 01:00:00 am »

no problem

Ranman · « **Reply #10 on:** August 13, 2006, 10:25:00 am »

Here is a decent sleep function that is can be used on both real and VTI.

The IsVTI() ASM function was written by Julien Muchembled, I found it on the TIGCC Headquarters message board a while back. The Sleep() function is code that I wrote. It basically puts the calc's CPU into an idle state for the specified number of milliseconds. Hence, it saves battery power -- which is a good thing!

The 68K Assembly code needed...
c1-->

CODE

ec1
|=============================================================================
| VTI detection routine (by Julien Muchembled)
|
| this routine utilizes a bug in VTI: you can set even non-existing flags in
| the status register and query them later!
|=============================================================================

| C prototype:
| short IsVTI(void);

.data
.globl IsVTI
.even

IsVTI:

Logged

Ranman
Bringing Randy Glover's Jumpman to the TI-89 calculator. Download available at Ticalc.