I'm familiar with the standard error, which usually means that I've got some variable I'm using that just hasn't been declared yet, but my problem is a little more complicated. I've got two files I'm dealing with. The first file is a binary file that just holds a long string of 0's and 1's that can be translated into a Matrix of values. The second file is a text file that tells me what data type those values are; they could be 8-bit, 16-bit, 32-bit, or 64-bit. I am then trying to read in 100 numbers from the first file, but I can only tell when to stop reading if I know the byte-size of the data in the file (like, if the data is stored as 64-bit numbers, then I've gotta read in 800 bytes of info, whereas I'd only have to read in 100 bytes if the values were 8-bit). To accomplish this task, I'm currently doing this (where dataTypeSize is already known to be the number of bytes used to define an individual value in the first file):

if(dataTypeSize == 1) //Because a char is 1-byte
    char * tempBands = new char[100];
else if(dataTypeSize == 2) //because a short is 2-bytes
    short * tempBands = new short[100];
else if(dataTypeSize == 4) //because an int is 4-bytes
    int * tempBands = new int[100];
else if(dataTypeSize == 8) //because a double is 8-bytes
    double * tempBands = new double[100];
else //Since the default size is 2-bytes
    short * tempBands = new short[100];
memset(tempBands, 0, 100*dataTypeSize);

The problem is that, at the memset() function, my compiler is yelling at me, saying that tempBands isn't declared (which makes sense, since there's no guarantee that those if statements will ever evaluate to true), but *I* know that there are no other options. dataTypeSize will ALWAYS be one of those 4 values, so tempBands will ALWAYS be declared. So, how do I convince my compiler that tempBands is definitely declared? Evidently my compiler doesn't understand my version of Trichotomy! (quadchotomy?)

Would there be a way to declare my own type of variable, like create a data type called MyData and declare it to be 4, 5, 6, 7, 8, 16, 64, 128 or however many bits I needed it to be, then say MyData * tempBands = new MyData[100]?

You could use a templated class that stores a type T and returns its size.

Quote:Original post by smitty1276
You could use a templated class that stores a type T and returns its size.

I'm curious to know what you mean by this. I did some looking around to figure out how Templated Classes work, and this is basically what I got (a simple example provided, for explanation):

template <class A_Type> class calc{  public:    A_Type multiply(A_Type x, A_Type y);    A_Type add(A_Type x, A_Type y);};template <class A_Type> A_Type calc<A_Type>::add(A_Type x, A_Type y){  return x+y;}int main(int argc, char * argv[]){	int dataTypeSize = 4;	if(dataTypeSize == 4)	{		calc<int> abc;		int a =2;		int b =3;		int c = abc.add(a, b);	}	else if(dataTypeSize == 8)	{		calc<double> abc;		double a =2;		double b =3;		double c = abc.add(a, b);	}	/*cout<<a<<endl<<b<<endl<<c<<endl; HAS ERRORS*/	return 0;}

This all works fine, compiles and all, but the problem is that I want to be able to access a, b, and c outside of those if-statements... and that's the problem I can't quite figure out how to solve =(

That isn't possible because C++ is a statically typed language -- the type of a variable cannot be decided at run-time. Consider polymorphism, or a variant type such as boost::variant.