RFC: replace the type keyword

[thestinger]

Calling this type is very confusing and has led to many newcomers thinking that it makes a new type when really it's just to save on typing (ha!) by making an alias.

I'm tempted to say it should just be like C++11 where you can do this:

using foo = int;
template<typename T> bar = std::set<T, std::greater<T>>;

If it's not feasible to extend use this way, calling it alias would still really be an improvement.

[metajack]

+1

[bstrie]

As long as we're bikeshedding type declarations, I want to register that I don't find the different uses of struct to be at all intuitive. For example, I know that "tuple structs" exist, but I honestly couldn't tell you what they are or how they are declared. Are they just named tuples? Because that really has nothing to do with structs at all. Same with newtype'd structs, I fear. It all feels very haphazard. If nothing else, we really need some better docs here.

[mzabaluev]

As things stand, type does not fully work as an alias due to e.g. #5635.
What's even more curious, a type made out of an instantiated generic like A<int> does not work in expressions where A<int> would result in syntax fail and must instead be written A::<int>. Does it mean that type works on syntactic level?

[graydon]

nominating for backwards compat. if we're going to do this, we need to decide to do it. if not, close.

[graydon]

(Incidentally, I'm not in favour of adding alias, but would be ok with extending use if we could make it work. Otherwise I think it's not a big enough deal to warrant the code-breakage.)

[graydon]

closing as WONTFIX