Vector is to Array as XX is to SubArray

[tlienart]

SubVector doesn't currently exist in Julia, would the following definition make sense?

const SubVector{T} = SubArray{T, 1}

similar to

julia/base/array.jl

Line 58 in 676ccf4

const Vector{T} = Array{T,1}

This could maybe be added there:

julia/base/subarray.jl

Line 40 in 056be22

happy to open a PR for this if people think this is useful? (there's this old issue: #3224 but I'm not sure it's relevant)

[oscardssmith]

I don't think this is especially useful. It's hard to think of any code that should be dispatching on SubVector. As such, I think this would probably just be clutter. That said I totally could be missing something here.

[simeonschaub]

One thing to note here is that SubArray{T, 1} is not a concrete type, so the parallels with Array and Vector may be a bit misleading, if you want to use it for annotating types in a struct. It would still work for things like dispatch though, so I don't have a strong opinion either way.

[tlienart]

One (possibly poor) reason why I find it useful personally to define this is that I like to use typing in some function signatures as I think it sometimes helps understand what the function does (I know this is not necessarily a shared opinion). In that sense, I prefer to annotate foo(sv::SubVector{Int}) than foo(sv::SubArray{Int,1}) (sometimes it seems to me it's justified to type like this bc you would want to take into account that it's a subarray in the body of the function).

Also there's the symmetry argument (Array/Vector, SubArray/SubVector) though @simeonschaub's argument makes sense.

Anyway, I'm happy with this issue being closed if no one else finds the suggestion interesting :)

[KristofferC]

foo(sv::SubVector{Int})

Why not use AbstractVector?

[tlienart]

Yes all these are working alternatives, this was merely out of expecting a SubVector to be there and not finding it.

I'll close this as it seems people don't see this as particularly useful, thanks for the comments!