Very often I use associative array as the simple syntactic way of
representing of pairs of some values. But AAs are slightly
complicated for this task and they don't preserve order inside
it. I could use some struct for example
struct P(T, N)
{
T first;
N second;
}
and write something like
auto pairs = [ P(1, "one"), P(2, "two"), P(3, "three") ];
But this looking not very pleasant and it's not obvious what P
is, but writing words Pair is verbose. So what I was thinking
about it could we have some syntactic sugar for this simple type
of data. In my experience of using D I needed it very often and
emulated it with AAs. Problem is that I don't always need all the
features of it (I meen searching by hash). But syntax like
1. auto pairs = [ 1: "one", 2: "two", 3: "three" ];
2. auto pairs = [ 1 => "one", 2 => "two", 3 => "three" ];
or
3. auto pairs = [ 1 -> "one", 2 -> "two", 3 -> "three" ];
4. something else
It could be useful in my opinion. Resulting type for this could
be a simple struct with 'first' and 'second' properties and index
operators [0], [1].
Also this could be expanded to 'threes' of values and etc
auto threes = [ 1 -> "one" -> "I", 2 -> "two" -> "II", 3 ->
"three" -> "III" ];
In process of writing of this letter I remembered about tuple
literal proposal. It can also help to solve this problem, because
we can make it not verbose and also it will propose standard
struct name for it. For example it could look like this:
auto pairs = [ t{1, "one"} , t{2, "two"}, t{3, "three"} ] ;
auto threes = [ t{1, "one", "I"} , t{2, "two", "II"}, t{3,
"three", "III"} ] ;
In this code I assume that t{} is tuple literal. But all these
bracers are not that elegant as single -> or => or another
operator.
I'd like to see any comments
