> Well you can call printSqured(b) and have its T inferred and that makes all
> the difference.
True. Two different ways around this:
* calling it like this `printSquared()(b)` to indicate the inferred first
parameter list (sucky IMHO) or
* defining it in upper-case
proc PrintSquared(T)(o: A(T)) =
echo o.x * o.x
Run
in analogy to the "upper-case names for types (i.e. classes of concrete
instances ), lower-case names for values (i.e. concrete instances)" rule. If we
don't want to rely on just identifier case, a new keyword would help:
procclass PrintSquared(T)(o: A(T)) =
echo o.x * o.x
Run
More typing, but only in the definition.
None of these are perfect, but "`()` is for parameter tuples, `[]` is for
dereferencing by index" is just too nice to lose it to generics.
As for generic types, independently of generic procs, parentheses instead of
brackets could be used right away.
