Thread: Testing some code, lots of errors...

Originally Posted by itCbitC

and shouldn't this invocation

Code:

op[k].f1(pop(stack, stackSize))

be

Code:

(*op[k].f1)(pop(stack, stackSize)) /* as f0 f1 f2 are all pointers to functions */

Ah, let me post the updated code so people aren't confused... I fixed that when laserlight told me about the evaluation problems...

Code:

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>

double pop (double stack[], int *stackSize);
void push (double stack[], int *stackSize, double x);
double multiply1 (double x, double y);
double add1 (double x, double y);
double neg1 (double x);
double divide1 (double x, double y);
double square1 (double x);
double squareRoot1 (double x);
double min1 (double x, double y);
double max1 (double x, double y);
double expo1 (double x, double y);
double pi (void);
double L (double x);
double E (double x);

typedef struct Operator
{
	char symbol;
	int operands;
	union
	{
		void *f;
		double (*f0)(void);
		double (*f1) (double x);
		double (*f2) (double x, double y);
	};
}Operator;

//double evaluate (const Operator op[], int opSize, const char * expression);

Operator mult2={'*',2,multiply1};
Operator add2={'+',2,add1};
Operator neg2={'-',1,neg1};
Operator divide2={'/',2,divide1};
Operator square2={'S',1,square1};
Operator sqrt2={'s',1,squareRoot1};
Operator min2={'m',2,min1};
Operator max2={'M',2,max1};
Operator expo2={'^',2,expo1};
Operator pi2={'pi',0,pi};
Operator loga2={'log',1,L};
Operator expon2={'e^',1,E};

int main (void)
{
	char input[81]={'\0'};
	struct op[13]={mult2, add2, neg2, divide2, square2, sqrt2, min2, max2, expo2, pi2, loga2, expon2};
	printf("Enter your calculations (blank to quit): \n");
	fgets(input, sizeof(input), stdin);
	//printf("Answer: ", evaluate(op[], 13, input));
}

/*double evaluate (const Operator op[], int opSize, const char * expression)
{
	int i;
	int indexSize;
	int *stackSize;
	int stackSz=0;
	double temp=0;
	double temp2=0;
	indexSize=strlen(expression);
	stack
	for (i=0; i<indexSize; ++i)
	{
		int k;
		switch(expression[i])
		{
			case '1':
				push(stack, stackSize, 1.0);
				break;
			case '2':
				push(stack, stackSize, 2.0);
				break;
			case '3':
				push(stack, stackSize, 3.0);
				break;
			case '4':
				push(stack, stackSize, 4.0);
				break;
			case '5':
				push(stack, stackSize, 5.0);
				break;
			case '6':
				push(stack, stackSize, 6.0);
				break;
			case '7':
				push(stack, stackSize, 7.0);
				break;
			case '8':
				push(stack, stackSize, 8.0);
				break;
			case '9':
				push(stack, stackSize, 9.0);
				break;
			default:
				for (k=0; k<opSize; ++k)
				{
					if(expression[i]==op[k].symbol)
					{
						switch(op[k].operands)
						{
							case 0:
								push (stack, stackSize, op[k].f0());
								break;
							case 1:
								temp=pop(stack, stackSize);
								push (stack, stackSize, op[k].f1(temp));
								break;
							case 2:
								temp=pop(stack, stackSize);
								temp2=pop(stack, stackSize);
								push (stack, stackSize, op[k].f2(temp, temp2));
								break;
						}
				
					}
				}
		}
	}
}*/

double multiply1 (double x, double y)
{
	return x*y;
}

double add1 (double x, double y)
{
	return x+y;
}

double neg1 (double x)
{
	return -x;
}

double divide1 (double x, double y)
{
	return x/y;
}

double square1 (double x)
{
	return pow(x,2);
}

double squareRoot1 (double x)
{
	return sqrt(x);
}

double min1 (double x, double y)
{
	return x>y?x:y;
}

double max1 (double x, double y)
{
	return x>y?y:x;
}

double expo1 (double x, double y)
{
	return pow(x,y);
}

double pi (void)
{
	return 3.1415926535;
}

double L (double x)
{
	return log(x);
}

double E (double x)
{
	return exp(x);
}

double pop (double stack[], int *stackSize)
{
	double value=0;
	value=stack[*stackSize];
	--*stackSize;
	return value;
}

void push (double stack[], int *stackSize, double x)
{
	++*stackSize;
	stack[*stackSize]=x;
}

As you can see the evaluate bits are commented out but I've been working on them as well... Using temp variables now for the pops... Seems like an all around good idea.


[edit] - Sorry I must be sleepy just reread what you posted and realized you're probably right about that...
[edit2] - Edit well maybe not... Compiler doesn't seem to care... I guess your way is better programming practice however... My way might not work in every compiler.

Originally Posted by laserlight

Pointers to functions can be used with the same syntax as functions.

right but you still need to dereference the pointer in order to get at the function as in

Code:

(*op[k].f1)(temp)

Originally Posted by stevesmithx

I don't know whether it is a good idea to use multi-character constants.
I have never used them before.

Right-o, good call...

Down to 1 error:

1>c:\documents and settings\edward\my documents\visual studio 2008\projects\cstuff\cstuff\cpolishcalculator.c(52 ) : error C2143: syntax error : missing ';' before '['

it has something to do with these 3 lines of code:

Code:

	char input[81]={'\0'};
	struct op[13]={mult2, add2, neg2, divide2, square2, sqrt2, min2, max2, expo2, pi2, loga2, expon2};
	printf("Enter your calculations (blank to quit): \n");

Originally Posted by laserlight

The pointer named op is already dereferenced with op[k]. The function pointer does not need to be explicitly dereferenced.

However, you have highlighted something that I am otherwise unfamiliar with: unions. To get that part of the code to compile, I had to resort to this:

Code:

typedef struct Operator
{
	char symbol;
	int operands;
	union
	{
		void *f;
		double (*f0)(void);
		double (*f1) (double x);
		double (*f2) (double x, double y);
	} fun;
} Operator;

Sparrowhawk, notice the introduction of the name fun. You need to give the union a name, and I have chosen fun as the name. You would then invoke the function as: op[k].fun.f1(temp);

Incidentally, you probably should just remove the void* named f from the union.

For unions nested in structures they can be left unnamed but you need to have a bracket there... I'm not sure what happens if you name them really , more to type perhaps.

Also, have any ideas about the error I posted right before this post?

Originally Posted by laserlight

The pointer named op is already dereferenced with op[k]. The function pointer does not need to be explicitly dereferenced.

isn't op[k] the k-th element of the array of structs.

Originally Posted by laserlight

However, you have highlighted something that I am otherwise unfamiliar with: unions. To get that part of the code to compile, I had to resort to this:

Code:

typedef struct Operator
{
	char symbol;
	int operands;
	union
	{
		void *f;
		double (*f0)(void);
		double (*f1) (double x);
		double (*f2) (double x, double y);
	} fun;
} Operator;

Sparrowhawk, notice the introduction of the name fun. You need to give the union a name, and I have chosen fun as the name. You would then invoke the function as: op[k].fun.f1(temp);

good point about naming the union, guess it got overlooked in all this debugging effort and it's sleepy time here.
so imo it shoud be (*op[k].fun.f1)(temp)

Originally Posted by laserlight

Incidentally, you probably should just remove the void* named f from the union.

excellent point; no need for objects you aren't using.

Originally Posted by laserlight

Where does it state as such in the C Standard? The MinGW port of gcc disagrees with you in both C90 and C99 mode.


Yes. However, at this point there is no array, only a pointer to the first element of the array.


Let me ask you a simple question: is op[k].fun.f1 a function pointer?

Ah my mistake, it's specific to Microsoft... I'm using Visual C++

Source:
http://msdn.microsoft.com/en-us/libr...0d(VS.80).aspx

Didn't see that bolded "Microsoft Specific" text heh...

I've still got the:

1>c:\documents and settings\edward\my documents\visual studio 2008\projects\cstuff\cstuff\cpolishcalculator.c(52 ) : error C2143: syntax error : missing ';' before '['

With

Code:

	char input[81]={'\0'};
	struct op[13]={mult2, add2, neg2, divide2, square2, sqrt2, min2, max2, expo2, pi2, loga2, expon2};

Originally Posted by laserlight

Yes. However, at this point there is no array, only a pointer to the first element of the array.

not sure how you get at that but guess I don't share your point of view and of course I could be totally wrong.

Originally Posted by laserlight

Let me ask you a simple question: is op[k].fun.f1 a function pointer?

yes it's a pointer to a function.

Originally Posted by laserlight

Instead of "struct op[13]" you probably want to write "Operator op[13]". When you do that, however, the initialiser that follows is not correct since it only an aggregate of function pointers. I am not sure of the correct syntax where unions are involved, so all I can suggest is to manually assign the function pointers to the corresponding Operator objects.

Yup that worked.

Originally Posted by laserlight

Look at the context: the code is in a function whose declaration is:

Code:

double evaluate (const Operator *op, int opSize, const char * expression)

cool beans for the longest time i was thinking that its all contained in main(); makes sense now that op is passed as an argument to evaluate(); must be that time i suppose