Type unions in D

[Jeremie Pelletier]

Justin Johansson Wrote:

> What's the best way of emulating a system of quantified type unions in D  (D1)?
> 
> By type unions (sometimes called algebraic sums in the literature), I don't mean the C/C++ "union" facility.
> 
> Suppose there is some class X, I'd like to have a first class type for representing the following quantified types (using RegExp notation):
> X?  zero or one instances of an X  (call this type XQ)
> X a single instance of X
> X* zero or more instances of X  (call this type XS)
> X+ at least one X (call this type XP)
> 
> To unify the whole thing, now let X, XQ, XS and XP all be derived from a common base type T.
> 
> Note that Scala has the Option, Some and None classes which kind of handles (but not exactly) the XQ case.
> 
> Brendan Eich, JavaScript inventor, apparently has suggested type unions for that language.  Some FP languages also have type systems that support type unions.
> 
> http://wiki.ecmascript.org/doku.php?id=meetings:minutes_oct_13_2005
> 
> "Type Unions
> Brendan thinks we should have some ability for type unions in the language.
> 
> var x : Number | Null
> 
> It would be good to have something like this because otherwise developers end up enforcing their invariants on their own, with the potential for buggy code."
> 
> As always, thanks for all replies.
> 
> Justin Johansson
> 

What you want sounds a lot like a variant type. Check std.variant in phobos, it has a template to let you define custom variants through type tuples.