Is there some good intuition for the last result of this:
julia> struct AA{T} end
julia> issametype(::AA{T}, x::T) where T = true
issametype (generic function with 1 method)
julia> issametype(::AA, x) = false
issametype (generic function with 2 methods)
julia> issametype(AA{Int}(), 1) #Ok, types are the same
true
julia> issametype(AA{Int}(), 1.0) # Ok, types are not the same
false
julia> issametype(AA{Integer}(), 1) # Uhm...?
true
It happens to do what I want in this case, but I though I needed to define issametype
like this:
issametype(::AA{T1}, x::T2) where {T1, T2<:T1} = true
to get the same result as above.
julia> 1 isa Integer
true
so (AA{Integer}(), 1)
matches issametype(::AA{Integer}, ::Integer)
the T
in your definition is not constrained to concrete types, after all
Makes perfect sense. I guess I tripped up on how one is used to writing e.g. isinteger(::AA{<:Integer}) = true
, since AA{Integer}
is concrete and forgot about the other argument.
Last updated: Nov 06 2024 at 04:40 UTC