Computer Science Canada

Quit #100
Author:  Insectoid [ Tue Dec 02, 2008 12:22 pm ]
Post subject:  Quit #100
So, I've been working on a triangular collision detection program. I wrote a line of code to return the number of degrees in an angle, but it quits the program on that line with an error 'Quit #100'. There are no exit, quit, etc. statements in the entire program. Anyone know what this is?

code:

%length variables created with Math.Distance

%cosine law to get angle
angleC := arccosd ((lengthA **2)+(lengthB**2) - (2*lengthB*lengthC)*(lengthC**2))


Remember, I need help with the error, not the actual triangular collision
Author:  The_Bean [ Tue Dec 02, 2008 1:49 pm ]
Post subject:  Re: Quit #100
The reason for the problem is because arccosd(num:real) where -1<num<1 because cosine is adjacent/hypotenuse and since adjacent is always smaller than the hypotenuse num will have to be between -1 and 1.
I'm guessing your looking for arctand.
Author:  Insectoid [ Tue Dec 02, 2008 4:17 pm ]
Post subject:  RE:Quit #100
That could be it...

Funny, the equation given to me by my math teacher uses cos...(cosine law).

EDIT: Yeah, I do know trig. Just forgot that equation (I asked specifically for the one that returns angles given only side lengths). Just thought I mentioned that in case people were wondering why I'm asking my math teacher for trig equations...
Author:  The_Bean [ Tue Dec 02, 2008 10:31 pm ]
Post subject:  Re: Quit #100
Your teacher was right with using the cosine law, but you messed up the formula.
code:

a ** 2 = b ** 2 + c ** 2 - 2 * b * c * cosd (A)
2 * b * c * cosd (A) = b ** 2 + c ** 2 - a ** 2
cosd (A) = (b ** 2 + c ** 2 - a ** 2) / (2 * b * c)
A = arccosd ((b ** 2 + c ** 2 - a ** 2) / (2 * b * c))

So replace it with
Turing:

angleC := arccosd ((lengthA ** 2 + lengthB ** 2 - lengthC ** 2) / (2 * lengthA * lengthB))
Author:  Insectoid [ Mon Dec 08, 2008 4:28 pm ]
Post subject:  RE:Quit #100
Ah! Thank you! (this was really rushed).

I had put this away for a bit to work on Perl, but I've decided to use it as my FP for CS. After I get it working in Turing, I'll change it to Java (because I want to concentrate on getting it to work without worrying about the complexities of that hideous language).

So, I need to make at least a simple triangular collision detection program, and then perhaps expand it to provide irregular collision detection based on combinations of triangles and rectangles, and then maybe do Asteroids.
Author:  Insectoid [ Mon Dec 08, 2008 6:10 pm ]
Post subject:  RE:Quit #100
Okay...more issues.

Quit #100 again, on this line:
code:

dotAngleA := arccosd (((dotLengthB **2)+(dotLengthC**2)-(lengthA **2))/(2*dotLengthB*dotLengthC))


This calculates the angle between 2 lines connecting the point to angles b and c. I'll try to supply a diagram to fully explain, as soon I figure out how to add text to an image in Pixen.

My way of doing this is, if any 1 of angles dotAngleA, B, or C are bigger than 180, the dot is not inside the triangle.
Author:  The_Bean [ Mon Dec 08, 2008 7:25 pm ]
Post subject:  Re: Quit #100
All i can come up with for you still getting an error is when:
code:

b**2+c**2-2*b*c>a**2

because your numerator would become greater than your denominator making a value greater than 1 or less than -1

EDIT

Now that I see the triangle (you managed edit before i put my post up), I don't think arccosd with produce a value greater than 180.
Author:  Insectoid [ Mon Dec 08, 2008 7:34 pm ]
Post subject:  RE:Quit #100
hmm...in theory that shouldn't happen...I know it's becoming greater than 1 or less than negative 1. I don't know why though.
Author:  The_Bean [ Mon Dec 08, 2008 7:36 pm ]
Post subject:  Re: Quit #100
Just tested it.
Turing:

for i : -10 .. 10
    put arccosd (i / 10)
end for

arccosd only produces between 0 and 180 because a triangle cannot have an interior angle greater than 180.
Author:  Insectoid [ Mon Dec 08, 2008 7:38 pm ]
Post subject:  RE:Quit #100
ah! thank you! Now how can I make it calculate over 180? is there a definite number range that will always produce over 180?

EDIT: lol, my 'hmm, that shouldn't happen' was before you edited your post. Razz

EDIT 2: Wait, cos values follow a wave, so there isn't a definite set, is there?

EDIT 3: I'm on the IRC right now, wanna go on there? (if you have time to help me off-board)
Author:  The_Bean [ Mon Dec 08, 2008 7:43 pm ]
Post subject:  Re: Quit #100
Well since it cannot calculate over 180 were going to have to use something else. So...
You can use my idea from before.
Turing:

if b**2+c**2-2*b*c>a**2 then
   %numerator is greater than denominator and
   %angle will be greater than 180
end if
Author:  Insectoid [ Mon Dec 08, 2008 7:44 pm ]
Post subject:  RE:Quit #100
Here's my diagram of my other concept: (making this way more complicated than it needs to be)
Author:  The_Bean [ Tue Dec 09, 2008 1:28 pm ]
Post subject:  Re: Quit #100
Last night after getting a working concept myself I realized another problem with both of ares, you can have 2 triangles overlapping, but without any of their vertices inside the other. Although depending on the speed and size at wich you have them moving this probably wouldn't happen.
Author:  Insectoid [ Tue Dec 09, 2008 4:07 pm ]
Post subject:  RE:Quit #100
Oh, crap...Didn't even think of that. Hmm...This is going to be fun! We could try to modify the 'perfect' oval collision into triangles, and thus check previous frames to see if a collision occurred mid-frame.

:


You can syndicate this boards posts using the file backend.php or view the topic map using sitemap.php.

Terms of Use | Privacy Policy