On 11/09/2012 03:35 PM, Too Embarrassed To Say wrote:
> struct Parameterized(T, U, V, W)
> {
> T t;
> U u;
> V v;
> W w;
> this(T t, U u, V v, W w)
> {
> this.t = t;
> this.u = u;
> this.v = v;
> this.w = w;
> }
> }
You are obviously using this just as an example but I wanted say that
you don't need a constructor because structs provide an automatic one
that copies the arguments to the members one by one.
> Parameterized!(int, double, bool, char) p1; // compiles
> // or
> auto p2 = Parameterized!(int, double, bool, char)(); // must have the
> empty () to compile
> // or
> auto p3 = Parameterized!(int, double, bool, char)(57, 7.303, false,
> 'Z'); // compiles
That is my preference.
> // but not
> // Parameterized!(int, double, bool, char)(93, 5.694, true, 'K') p4;
> // Error: found 'p4' when expecting ';' following statement
> // nor
I don't expect that to work; the arguments must go after the variable there.
> // Parameterized!(int, double, bool, char) p5(93, 5.694, true, 'K');
> // Error: found 'p5' when expecting ';' following statement
Ok, that is strange. I don't understand what "statement" the compiler
sees there.
> // nor, OK this was a crazy try
> // Parameterized!(int 93, double 5.694, bool true, char 'K') p6;
> // Error: found '93' when expecting '.' following int
No, that shouldn't work.
There is also the common idiom of providing a convenience function to
help with template parameter deduction:
auto parameterized(T, U, V, W)(T t, U u, V v, W w)
{
return Parameterized!(T, U, V, W)(t, u, v, w);
}
/* ... */
auto p7 = parameterized(93, 5.5, true, 'K');
Ali
