making @polyvar generic in dimension

[dahtah]

I have some code involving polynomials that I'd like to make generic in dimension. To be precise, consider the following bit of code:

@polyvar x[1:2]
p = sum(x.^2)
#... do some computation involving p

Now say we want to make this generic, ie. have

@polyvar x[1:d]
p = sum(x.^2)

for dimension d, which we may assume known at compile time.
The current macro doesn't authorise that and I'm struggling to find a way around the problem (but my knowledge of metaprogramming in julia is poor). For instance, the following doesn't work:

function test(Z::Val{N}) where N
    @polyvar x[1:N]
    sum(x.^2)
end

even though N is known at compile time. Is there a workaround?

[rdeits]

You can just construct the Variable instances yourself rather than trying to use the macro. For example, you can create a tuple of variables with the built-in ntuple function:

julia> d = 4
4

julia> ntuple(i -> Variable{Symbol("x", i)}(), d)
(x1, x2, x3, x4)

If you use @macroexpand, you can see that the @polyvar is doing essentially the same thing:

julia> @macroexpand @polyvar x[1:2]
quote
    x = ((TypedPolynomials.Variable){Symbol("x[1]")}(), (TypedPolynomials.Variable){Symbol("x[2]")}())
    x
end

[dahtah]

Very helpful, thanks. Maybe this should be in the doc somewhere?