describe(psi); claims that Maxima can't evaluate the
psi (the logarithmic derivative of gamma). Indeed
psi[n](x); returns a simple expression involving
zeta(n+1) for integer arguments x, and just psi[n](x)
for other arguments.
But there is another implementation of psi in the bffac
module, so load(bffac); bpsi(n, x, precision); does
yield numerical results for general arguments.
It seems like the right thing to do is to have psi call
bfpsi for arguments it can't evaluate.
Another slightly less right thing is to change the
description of psi to say that bfpsi in bffac can
handle general arguments.
For what it's worth,