I think that the following code should work (sorry if there is any typo):
type Stack[T] = concept c
c.push(T)
c.pop(var T) is bool # or something like this (this is: var t:T;
c.pop(t); )
proc f(x: Stack[int]) = ...
but the following should fail to compile:
type Stack[T] = concept c
c.push(T)
c.pop(var T) is bool # or something like this (this is: var t:T;
c.pop(t); )
proc f[T](x: Stack[T]): T = ...
because `c.push(T)` should not infer the type. It is not unique, and one type
could implement both `Stack[int]` and `Stack[float]`.
>From the proposed manual it seems like you are suggesting that the second
>example should also work. Am I right in this? What do you think?
Thanks,
Peter
