In principle, I would suggest this, which is going to be the correct way to do this in Julia in the future:
load("distributions.jl")
rand(Uniform(-1, 1))
Unfortunately there is currently a bug in it that I won't have time to fix today. If someone else is motivated to learn how Julia's RNG's work, you can look in extras/distributions.jl. You can also confirm that the generator is not working as I expect it should based on this loop:
for i = 1:100
println(rand(Uniform(-1, 1)))
end
On my machine that produces 100 negative numbers, an event with probability 2^-100 if the generator worked.
-- John