Math Error with rotations

Hi all,

I have a script that gives a math error when assigning a rotation to another rotation. The error can occur at TotalRot = Rot1 and/or TotalRot *= Rot2.
basicaly it does:

Rot1 = ZERO_ROTATION;
Rot1 = Rot1 / llAxisAngle2Rot(Axis1, Angle1);

Rot2 = ZERO_ROTATION;
Rot2 = Rot2 / llAxisAngle2Rot(Axis2, Angle2);

//step1
TotalRot = Rot1;
//step2
TotalRot *= Rot2;

When I cut out the function that gives the error into another script, it sometimes generates the same error or just works! So I don't have an example code that generates the error (yet). I checked many times if there are any changes of Rot1 before it gets assigned to TotalRot but it doesn't. They are definied in the scope if a function

I am wandering about the rotations. I suppose using llAxisAngle2Rot() always gives a valid quaternion for Rot1 and Rot2. So a math error when assigning one rotation to the other cannot give an error.
However it does. Does anyone have suggestions ?

(I'll try to get and example her that generates the error)

Sure you aren't running into a bounds check error?

As in:

vector total;

default
{
	touch_start(integer total_number)
	{
		vector a = <123,65,86>;
		vector b = <89,23,286>;
		total = a;
		total *= b;
		llSay(0, (string)total);
	}
}

(BTW, lslint will not check a vector dot product. "Invalid operator: vector = float.". I'll post it in the lslint thread)

Returns:

"[8:33] Object: <123.00000, 65.00000, 5590.00000>
[8:33] Object: Script run-time error
[8:33] Object: Bounds Check Error"

even switching to this:

vector total;

default
{
    state_entry()
    {
        vector a = <4,5,8>;
        vector b = <2,3,2>;
        total = a;
        total *= b;
        llSay(0, (string)total);
    }
}

will return:
[8:40] Object: <4.00000, 5.00000, 40.00000>
[8:40] Object: Script run-time error
[8:40] Object: Bounds Check Error

dang dyslexia and selective vision. Read the wiki wrong and lslint was right

Would have expected dot product to be a * b = <a.x * b.x), (a.y * b.y), (a.z * b.z)>

But it doesn't. Look at the wiki again:

* Vector dot product <1,2,3> * <4,5,6> = (1 * 4) + (2 * 5) + (3 * 6) = 32

Hi Jesse,

Yes, that is a tricky error. I checked the types and they are all rotations, no unexpected conversions to float or some other type.

I have separated the faulty function in 2 functions and that works fine. Still no idea what went wrong because the sequence of calculations and the values passed are exactly the same.

Thanks for your help

Sorry, I don't always state myself clearly.

Check the output of these lines by using a say:

TotalRot = Rot1;
llOwnerSay((string)TotalRot);
It is going to return a rotation

But
TotalRot *= Rot2;
llOwnerSay((string)TotalRot);
Is going to return a float, if it is going to return anything
It shouldn't even compile because you already have typecast
TotalRot as a rotation and you can not change it to a float

To check yourself then do something like this:
TotalRot = Rot1;
float output_test = TotalRot * Rot2;
llOwnerSay((string)output_test);