I just uploaded to unstable an important modification to the code for
the Kalman smoother. Re-factoring existing code, correcting some bugs
for the diffuse smoother (nonstationary models), and eliminating
specialization of the code meant to improve efficiency.
In the future, we will expand the existing DLL for likelihood for the
smoother and dealing with special cases.
Please, report any problem that you could encounter with the new code.
Best,
Michel