R help - Sep 2003 - Multivariate Kalman filter with time-varying coefficients

Hi

Does anyone know of any R code for estimating a *multivariate* state
space model using a Kalman filter where the output matrix H(t) is
time-varying but predictable (i.e. measurable w.r.t information at time
t-1) in the observation equation 

y(t) = H(t) z(t) + R w(t)? 

[Here y(t) are the observations, z(t) is the state variable, w(t) the
observation error and R R' the observation error covariance]

Cheers,
David