gamma - numerical evaluation at complex argument

Now when Maxima numerically evaluate all kinds
of Bessel functions with complex argument I wonder
can we do this for gamma function as well?

Now gamma(z) for real floating point z converts
it to complex, feeds in gamma-lanczos and converts
result back to real.

Can't we just use gamma-lanczos for complex floating z
directly?

-- 
     Vadim V. Zhytnikov

      <vvzhy@mail.ru> 
     <vvzhy@netorn.ru>