AbstractArray(A) throw method error while AbstractArray{Float64}(A) does not.

[dlfivefifty]

The following seems inconsistent to me:

julia> A = rand(5,5)
5×5 Array{Float64,2}:
 0.302006   0.855053  0.661854   0.756174  0.163208
 0.954073   0.919937  0.499419   0.766859  0.69692 
 0.651032   0.18593   0.071093   0.144807  0.100126
 0.0723692  0.710081  0.0466832  0.864155  0.101641
 0.764091   0.134407  0.636173   0.121791  0.211467

julia> AbstractArray{Float64}(A)
5×5 Array{Float64,2}:
 0.302006   0.855053  0.661854   0.756174  0.163208
 0.954073   0.919937  0.499419   0.766859  0.69692 
 0.651032   0.18593   0.071093   0.144807  0.100126
 0.0723692  0.710081  0.0466832  0.864155  0.101641
 0.764091   0.134407  0.636173   0.121791  0.211467

julia> AbstractArray(A)
ERROR: MethodError: no method matching AbstractArray(::Array{Float64,2})
Closest candidates are:
  AbstractArray(::Any) where T<:AbstractArray at abstractarray.jl:21
  AbstractArray(::Union{QR, LinearAlgebra.QRCompactWY}) at /Users/solver/Projects/julia7/usr/share/julia/site/v0.7/LinearAlgebra/src/qr.jl:414
  AbstractArray(::QRPivoted) at /Users/solver/Projects/julia7/usr/share/julia/site/v0.7/LinearAlgebra/src/qr.jl:421
  ...
Stacktrace:
 [1] AbstractArray(::Array{Float64,2}) at ./abstractarray.jl:21
 [2] top-level scope

[nalimilan]

Sounds like a good idea in the perspective of supporting Vector(x) and AbstractVector(x) as alternatives/replacements to collect, see #16029 (comment).