hessian_sparsity fails with registered funcitons

[baggepinnen]

Due to unknown linearity

ERROR: Failed to apply rule (~f)(~(x::isidx)) => if haslinearity_1(~f)
        #= /home/fredrikb/.julia/packages/Symbolics/J8IHJ/src/diff.jl:586 =#
        combine_terms_1(linearity_1(~f), ~x)
    else
        #= /home/fredrikb/.julia/packages/Symbolics/J8IHJ/src/diff.jl:588 =#
        error("Function of unknown linearity used: ", ~f)
    end on expression Base.FastMath.exp_fast(Symbolics.TermCombination(Set([Dict(139 => 1), Dict(139 => 2)])))

How would one go about registering the derivative for a function as well?