what is the difference between...

[sail3005]

template <class X> and
template <typename X>

????

[Zaei]

For example?

template<int x>
and
template<class t>

The first one <int x> would take in a number (or char, it it were <char x>, etc), while <class t> takes in a class type, int, char, etc. So, say you have a template struct:

template <class t>
struct someStruct
{
     t x;
     t y;
};

You can use t in place of a type name. You can use the <int x> for things like template metaprogramming (using the compiler to find values at compile time, instead of at run time (fibinocci (sp?) numbers are a good example)).

Z.

[kedaman]

Zaei, can you post the partial evaluation for fibonacci series using templates again? the search doesn't work :/

[Zaei]

Here ya go kedaman:

template <unsigned int x>
struct fib
{
  enum
  {
       fib = fib<x-1>::fib + fib::<x-2>::fib
  };
};

template <> struct fib<0> { enum { fib = 0 };};
template <> struct fib<1> { enum { fib = 1 };};

Z.

[kedaman]

doh! I wanted a template with two parameters, but i forgot that you only need one for fibonacci.

Thanks anyways.

here's an integer power template that doesn't work in MSVC(because of the other parameter), and I want to know if anyone has a similar that will work for MSVC
http://www.extreme.indiana.edu/~tveldhui/papers/pepm99/

[CornedBee]

This is cool. First I thought this must be impossible because the template parameter must be a constant. But then I realized that constant - constant IS a constant. Really cool.

[kedaman]

It sure is cool, but it would be even cooler if this did work in MSVC

[Zaei]

try it this way kedaman:

template <int x, int y>
struct pow
{
  enum
  {
      pow = x* pow<x, y-1>::pow
  };
};

template <int x, 0> struct pow { enum { pow = x }; };

Not tested, just a quick rewrite of the code below... Try it out, and let me know what happens =).

Z.

[Zaei]

uhh, should have been "code above" =).

Anyway, doesnt work, I just tested it out. Maybe in VC.NET ::shrug::

Funny thing is, I was thinking about a pow template earlier =).

Z.

[kedaman]

You can make a power of 2 template but not a power of n :/
So much for template specification when you have to specify all parameters

[kedaman]

Ok, I got around it nicely by forcefully resetting x parameter so that they will both line up at x,1 to 0,0.

But, it doesn't seem to work using constants, why?!?

template <int x, int y>
struct pwr2{
    static const int p;};
template <int x, int y>    const int pwr2<x,y>:: p= x*pwr2<x*(y>1),y-1>:: p;
template <>                const int pwr2<0,0>:: p=1;

template <int x, int y>
struct pwr{enum{      
    p = x* pwr<x*(y>1), y-1>:: p  
};};
template <> 
struct pwr<0,0>{enum{ 
    p = 1 
};};

int main(){ 
    cout << pwr<2,3>:: p << endl;
    cout << pwr2<2,3>:: p << endl;
    return 0;
}; 

IT would be nicer to have any type of constants than enumerated short integers.

[Zaei]

Enumerations compile to the smallest possible data type for the number of parameters inside of them. For almost all enumerations that I create, I have a "FORCE_DWORD = 0x79999999" value to force the enumeration to compile to 32 bits, so it shouldnt be a problem to have large enums in your templates.

Z.

[kedaman]

Can I have say complex value in an enumeration?

[Zaei]

Eh? Explain?

Z.

[kedaman]

something like
enum { x=cmplx(y,z) }
which would make x a complex value (a struct with 2 floats)

[Zaei]

I have no idea. You might try to cast the cmplx() result into an __int64, and assign to the enum, and then get the values out using a union, etc.

Z.

[sail3005]

cool, thanks for the replies guys. heh, thanks for the fibonacci code too