parse_expr_to_symbolic can't see function scope variables

[orebas]

This seems to be a difficulty in pointing parse_expr_to_symbolic to the right scope. In the below MWE, parse_expr_to_symbolic() silently "misses" the @variables declared in the function test(). I believe it creates new, non-time-dependent variables in Main scope as well.

I guess this make sense if one understands the difference between module scope and function scope, but is a potential source for errors and is probably not the desired behavior for most users. The documentation has a nice warning about this, so thanks for that.

using ModelingToolkit
using ModelingToolkit: t_nounits as t, D_nounits as D

#GLOBAL SCOPE

@variables x(t) y(t) a b
str_one = "a + b*b + x + y*y"
println(Symbolics.parse_expr_to_symbolic(Meta.parse(str_one), @__MODULE__))

#function scope
function test()
	@variables u(t) v(t) c d
	str_two = "c + d*d + u + v*v"
	println(Symbolics.parse_expr_to_symbolic(Meta.parse(str_two), @__MODULE__))
	println(Symbolics.parse_expr_to_symbolic(Meta.parse(str_one), @__MODULE__))
end
test()

OUTPUT:

a + x(t) + b^2 + y(t)^2
c + u + d^2 + v^2
a + x(t) + b^2 + y(t)^2

[orebas]

I am trying to find a way around this using a "temporary submodule" or something, but I'm not really sure what Julia allows.