The second example in the first section is a bit misleading. They say that to create a column matrix, ie. an (n, 1) matrix, the syntax is

    [1 2 3]'

but that will give you the (lazy) hermitian adjoint of the column vector, not a column matrix.

    julia> [1 2 3]' isa Matrix
    false

Instead, if you really need a true (n, 1) matrix you should 1) not use adjoint because that will also take complex conjugates (unless that's what you wanted) and 2) use collect to turn the lazy transpose (or adjoint) into a true matrix

    julia> collect(transpose([1 2 3]))
    3×1 Array{Int64,2}:
     1
     2
     3

However, you almost never need to do this sort of thing and you should be fine using either transpose on a row vector or just using a real column vector.

Also, near the end they talk about closures but don't actually show any closures unless one is to assume the code snippets they're showing are actually inside a function body themselves.