2D Rotation problem

Finally my 2D rotator seems to rotate, at least sort of.
Still having huge problems to make it work right. After a while (angle value 150+), my x values goes out of the screen for no good reason

I got the same routine working with blitz basic (without sintable and with calculated radians). So my lucky guess is that my sintable still sucks. It's created with following: IS 0,450,1280,256,0,w,1,y,y
This should create sintable with 1.25 perioids and max value 256, so i can easily divide the result with 256 (asr.l #8). Anyone got ideas what i'm still doing wrong?

Full source:
http://paradise.untergrund.net/tmp/2dRot.s

[StingRay]

Didn't check the code but my best bet is: sub.w d3,d2 is cause of your problem, it should be sub.l d3,d2.

[Leffmann]

The muls.w instruction produces a 32 bit product, and when you tell ASM-One to create a sinus table for this purpose you should select no half correction.

Originally Posted by StingRay

Didn't check the code but my best bet is: sub.w d3,d2 is cause of your problem, it should be sub.l d3,d2.

Now that i think of it, it makes perfect sense. And it works fine now.

Leffmann:
Changed the sinustable also. Could you please explain a little why half correction doesn't work here?

[pmc]

Originally Posted by StingRay

Didn't check the code but my best bet is: sub.w d3,d2 is cause of your problem, it should be sub.l d3,d2.

Originally Posted by Rebb

Now that i think of it, it makes perfect sense. And it works fine now.

More Jedi bug fixing from StingRay!

You sensed a disturbance in the code!

Gonna start calling you Obi-Wan

[Leffmann]

Originally Posted by rebb

Now that i think of it, it makes perfect sense. And it works fine now.

Leffmann:
Changed the sinustable also. Could you please explain a little why half correction doesn't work here?

I think it adds half a degree to each degree for which it calculates a sine. There are times when you want this but for this sort of lookup table you want exact values, and if you inspect your table you will see there are no values equivalent to -1 and 1, which you would expect for all the right angles.