On Tuesday, 2 June 2015 at 08:10:27 UTC, rsw0x wrote:
exactly what is the difference here?
I have a rather large CTFE-generated TypeTuple in one of my
structs in my project, and I can seemingly replace it with a
Tuple with absolutely zero differences... except I compile
about 60-70% slower.
The tuple page is even confusing me
http://dlang.org/tuple.html
A variable declared with a TypeTuple becomes an ExpressionTuple:
alias TL = Tuple!(int, long);
is it using Tuple!(T...) and TypeTuple!(T...) interchangeably?
Regular tuples are simply structs with a few methods like
opIndex. Ex. `Tuple!(A,B,C)` is mostly equivalent to `struct { A
_0; B _1; C _2; }`.
TypeTuples (whose name is IMO a misnomer) are special
compile-time-only objects whose values are passed in the template
parameters and can be types, aliases, or literal values. They
aren't first-class types, and shouldn't be used as data storage,
but they do have some unique properties that make them useful for
metaprogramming.
Here's a small demonstration of TypeTuples:
import std.stdio;
import std.typetuple;
void foo(int x, int y, int z) {
writefln("x = %d, y = %d, z = %d", x, y, z);
}
void main() {
int a, b, c;
// Declare `vars` to be a typetuple containing aliases to a,b,c
alias vars = TypeTuple!(a, b, c);
vars[0] = 1; // Actually assign to a, b, c
vars[1] = 2;
vars[2] = 3;
foo(a,b,c);
foo(vars); // Same as above call
alias test_values = TypeTuple!(123, 456, 789);
foo(test_values); // Same as foo(123, 456, 7890)
// won't work; test_values[0] is a literal and can't be
assigned to.
//test_values[0] = 321;
foreach(ref some_var; vars) {
// static foreach; this loop body is unrolled for each item in
vars.
some_var += 5;
}
foo(vars);
}
Output:
$ rdmd ~/test.d
x = 1, y = 2, z = 3
x = 1, y = 2, z = 3
x = 123, y = 456, z = 789
x = 6, y = 7, z = 8
