I have tried to simultaneously fit weighted datasets and (unless I have the syntax wrong) it seems this is not supported, because the weights are lost during the fit and as a result the uncertainties come out wrong. It would be very useful to have this feature.