typealias bug ?

[Jutho]

Since this did not get answered on the user forum, I expect nobody knows and this is a potential bug.
Starting with the following definitions

abstract Vartype
type VT1<:Vartype
end

abstract Graph{T<:Vartype}
abstract Subgraph{T<:Vartype} <: Graph{T}
type Vertex{T<:Vartype} <: Subgraph{T}
end
type Block{T<:Vartype,N} <: Subgraph{T}
end

abstract GraphObject{T<:Vartype}
type SubgraphObject{T<:Vartype,S<:Subgraph{T}} <: GraphObject{T}
end

typealias VertexObject{T<:Vartype} SubgraphObject{T,Vertex{T}}
typealias BlockObject{T<:Vartype,N} SubgraphObject{T,Block{T,N}}
typealias VertexBlockObject{T<:Vartype} Union(VertexObject{T},BlockObject{T})

O1=SubgraphObject{VT1,Vertex{VT1}}()
O2=SubgraphObject{VT1,Block{VT1,3}}()

I did expect all of the following to be true, but nevertheless got the results in the comment

isa(O1,VertexObject) # false
isa(O1,VertexObject{VT1}) # true
isa(O1,VertexBlockObject) # false
isa(O1,VertexBlockObject{VT1}) # true

isa(O2,BlockObject) # false
isa(O2,BlockObject{VT1}) # false
isa(O2,BlockObject{VT1,3}) # true
isa(O2,VertexBlockObject) # false
isa(O2,VertexBlockObject{VT1}) # false

I don't see how this is different to isa(randn(3,3),StridedMatOrVec), which returns true with or without full specification of the argument.

Is this a bug or expected behaviour?

[kshyatt]

Is this 2.5 year old bug still a problem?

[JeffBezanson]

Yep, but is fixed by #18457.

[vtjnash]

@StefanKarpinski I'm pretty certain we won't backport that fix

[StefanKarpinski]

then put the 0.6 label on it