potential for better type inference with Vararg

[simeonschaub]

Since for example Vararg{Any,Int} throws an error during construction, would it be possible to take advantage of the fact that the N parameter in a Vararg is always an Int in type inference? It would be nice if the following inferred more precisely:

julia> f(::NTuple{N}) where {N} = N
f (generic function with 1 method)

julia> code_warntype(f, Tuple{Tuple})
Variables
  #self#::Core.Compiler.Const(f, false)
  #unused#::Tuple{Vararg{T,N}} where T where N

Body::Any
1 ─     return $(Expr(:static_parameter, 1))

Sorry if there is an issue about this already, but I couldn't find one.

[JeffBezanson]

Yes, it looks like we should add this.

[ranocha]

Closes ... #37316

It looks like this should have been closed by #38136?

[simeonschaub]

Looks like we still don't infer this on nightly:

julia> f(::NTuple{N}) where {N} = N
f (generic function with 1 method)

julia> code_warntype(f, Tuple{Tuple})
Variables
  #self#::Core.Const(f)
  #unused#::Tuple{Vararg{T, N}} where T where N

Body::Any
1 ─     return $(Expr(:static_parameter, 1))

[simeonschaub]

Ah, but it does infer now if the Tuple has a concrete eltype:

julia> f(::NTuple{N}) where {N} = N
f (generic function with 1 method)

julia> code_warntype(f, Tuple{Tuple{Vararg{Int}}})
Variables
  #self#::Core.Const(f)
  #unused#::Tuple{Vararg{Int64}}

Body::Int64
1 ─     return $(Expr(:static_parameter, 1))

[aviatesk]

Fixed by #51449.