Jump to content

  • Log In with Google      Sign In   
  • Create Account

Interested in a FREE copy of HTML5 game maker Construct 2?

We'll be giving away three Personal Edition licences in next Tuesday's GDNet Direct email newsletter!

Sign up from the right-hand sidebar on our homepage and read Tuesday's newsletter for details!


C++ macro return? (converting C++ macro to C# code)


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
4 replies to this topic

#1 Moe   Crossbones+   -  Reputation: 1248

Like
0Likes
Like

Posted 23 January 2014 - 07:07 PM

First of all, please forgive me, as my C++ is really quite rusty. It's been... seven years or so since I've really had to use it.
 

I"m in the process of converting some C++ code to C# (in particular some code for checking an axis-aligned bounding box against a triangle, in 3D).  The original source in question can be found here.

 

The particular portion that I'm having a bit of trouble is this:

#define AXISTEST_X01(a, b, fa, fb)			   \

	p0 = a*v0[Y] - b*v0[Z];			       	   \

	p2 = a*v2[Y] - b*v2[Z];			       	   \

        if(p0<p2) {min=p0; max=p2;} else {min=p2; max=p0;} \

	rad = fa * boxhalfsize[Y] + fb * boxhalfsize[Z];   \

	if(min>rad || max<-rad) return 0;

Now from what I remember, I thought C++ macros were basically pre-processor bits that get substituted in at compile time.  The thing I'm perplexed at is the return statement in the macro.  In this case, what happens if min < rad?  Does the statement still return any sort of value, or is there some sort of default return value on a macro that I'm not aware of?

Here's what the rough C# equivalent is, as far as I have been able to figure:

 

	//Assume variables are declared/initialized properly.  Unlike what I've done here...
        float p0;
        float p2;
        Vector3 v0;
        Vector3 v1;
        Vector3 v2;
        Vector3 boxHalfSize;
        float min;
        float max;

        private int AXISTEST_X01(float a, float b, float fa, float fb)
        {
            p0 = a * v0.Y - b * v0.Z;
            p2 = a * v2.Y - b * v2.Z;

            if (p0 > p2)
            {
                min = p0;
                max = p2;
            }
            else
            {
		min = p2;
		max = p2;
            }

	    float rad = fa * boxHalfSize.Y + fb * boxHalfSize.Z;

	    if (min > rad || max < -rad)
	    {
		return 0;
	    }
	
            //Otherwise... return something else?
           return 0;
    }

Any thoughts/ideas on how it should actually look?  I can't say I've ever dealt much with macros, and this one has me a bit stumped.



Sponsor:

#2 ApochPiQ   Moderators   -  Reputation: 15984

Like
6Likes
Like

Posted 23 January 2014 - 07:16 PM

Think of the macro as a really, really stupid search and replace. The importance of the return is not within the macro itself; the importance of the return is in the surrounding code where the macro is used.



#3 Hodgman   Moderators   -  Reputation: 30885

Like
4Likes
Like

Posted 23 January 2014 - 07:20 PM

1) Now from what I remember, I thought C++ macros were basically pre-processor bits that get substituted in at compile time.
2) what happens if min < rad?  Does the statement still return any sort of value, or is there some sort of default return value on a macro that I'm not aware of?

1) Right - it's basically an automated copy & paste of the ASCII text before the compiler actually runs.
2) Because of (1), this means it's not "returning from the macro", it's copy&pasting that text directly into the function body. If the if is true, the parent function returns 0, else the parent function continues running.
 
I would first kill off this macro in the C++ code, so it looks something like this:
int OldParentFunction()
{
	float v0[3], v2[3], boxhalfsize[3];
	float min=0, max=0, rad=0;
	AXISTEST_X01(1, 2, 3, 4)
	AXISTEST_X01(5, 6, 7, 8)
	return 1;
}
... -> ...
inline bool AXISTEST_X01( float a, float b, float fa, float fb, float* v0, float* v2, float& min, float& max, float& rad, float* boxhalfsize )
{
	float p0 = a*v0[Y] - b*v0[Z];
	float p2 = a*v2[Y] - b*v2[Z];
	if(p0<p2) {min=p0; max=p2;} else {min=p2; max=p0;}
	rad = fa * boxhalfsize[Y] + fb * boxhalfsize[Z];
	return (min>rad || max<-rad);
}
void NewParentFunction()
{
	float v0[3], v2[3], boxhalfsize[3];
	float min=0, max=0, rad=0;
	if( AXISTEST_X01(1, 2, 3, 4, v0, v2, min, max, rad, boxhalfsize) ) return 0;
	if( AXISTEST_X01(5, 6, 7, 8, v0, v2, min, max, rad, boxhalfsize) ) return 0;
	return 1;
}

Edited by Hodgman, 23 January 2014 - 07:23 PM.


#4 frob   Moderators   -  Reputation: 22206

Like
3Likes
Like

Posted 23 January 2014 - 07:41 PM

Wow, whoever wrote that macro should get a stern warning, as should whoever let it through a code review.

There are side effects everywhere that would be really difficult to reproduce in any other language. The most immediately obvious are that the block isn't wrapped in brackets to become its own statement, the macro parameters are used more than once (meaning any side effects from them may run many times) and the parameters are not properly parenthesised when used, plus the control structures and the return.

For bugs in your code, it looks like the comparison of p2<p0 was reversed, the lines "min=p2; max=p2;" should be fixed inside the else block with p0 for the second line, the variables min, max, and rad should probably be made local variables, and you are right about the problematic return at the end. It should return a different result (perhaps 1) when the test would have returned zero, and then when you call the function it should be if(AxisTest_X01(...)==1)return 0; That would give you pretty close to the expected behavior if it were used in well behaved code.

Of course if there were other creative values used, such as being run within some other block or a function parameter or yet another complex macro as a parameter or a carefully crafted template as a parameter, a whole lot of really strange things could happen because of the search-and-replace magic that macros represent. Those would likely to be impossible to duplicate in C# except on a case-by-case careful examination.

Check out my book, Game Development with Unity, aimed at beginners who want to build fun games fast.

Also check out my personal website at bryanwagstaff.com, where I write about assorted stuff.


#5 Moe   Crossbones+   -  Reputation: 1248

Like
1Likes
Like

Posted 23 January 2014 - 08:44 PM

Awesome guys.  Many thanks on the uber-quick reply.  Also, frob, good catch on my mistake there.  Thinking about it, it now makes sense that the macro isn't returning anything, but is calling return in the code where it is being used.  I also agree that this is a pretty not-nice use of a macro.  

I haven't seen too many other samples out there that use the separating axis theorem for AABB-triangle detection.  The only other sample I've seen had issues as well.  

Again, thanks a bunch you guys.  

(I feel like it's been quite a while since I've been around here, and I'm glad to see the community is still alive and kicking, like back in the good ol' days.)






Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS