On Wed, Oct 12, 2016 at 2:41 PM, Mauro <mauro...@runbox.com> wrote:
> On Wed, 2016-10-12 at 20:03, Yichao Yu <yyc1...@gmail.com> wrote:
> > On Wed, Oct 12, 2016 at 1:22 PM, esproff <espr...@gmail.com> wrote:
> >
> >> Consider the code:
> >>
> >> abstract AbstractFoo
> >>
> >> type Foo <: AbstractFoo
> >> end
> >>
> >> f(x::AbstractFoo, y::Integer) = "method 1"
> >> f(x::Foo, y::Real) = "method 2"
> >>
> >> foo = Foo()
> >> f(foo, 1)
> >>
> >> This code results in an ambiguity error, since both methods contain one
> >> argument with a more specific type declaration than the other. Now
> >> consider this code:
> >>
> >> abstract AbstractFoo
> >>
> >> type Foo <: AbstractFoo
> >> end
> >>
> >> f(x::AbstractFoo, y::Integer) = "method 1"
> >> f{T<:Real}(x::Foo, y::T) = "method 2"
> >>
> >> foo = Foo()
> >> f(foo, 1)
> >>
> >> This code, in my opinion, should work, since the second method spawns a
> >> bunch of sub-methods, one for each concrete subtype of Real in the
> second
> >> argument, and thus one of these sub-methods should be
> >>
> >
> > No the set of signature the second method matches in the second case is
> the
> > same as that in the first case.
>
> However, somewhat similar, this does not error:
>
> julia> g(x) = 1
> g (generic function with 1 method)
>
> julia> g{X}(x::X) = 2
> g (generic function with 2 methods)
>
> julia> g(4)
> 2
>
> So here the parameterized method does seem more specific.
>
This is missing method overwite. i.e. a bug.
