basis=mixed or atomtype in oniom calculations

basis=mixed or basis=atomtype
does not seem to work in oniom calculations since
the consecutive lines of the input file with the basis assignment to the atoms, e.g. def2-TZVP 2 or O: def2-TVP are missing in the input file for the calculation of the model system.

While I am at it; Are there still plans for partial rather than full geometry optimizations, constraints, frozen coordinates? 

Dear Peter,
Indeed, the subsequent lines of the basis keyword in the case of 'basis=mixed' and 'basis=atomtype' are not copied into the input of the model system calculations. However, there is a workaround with the use of subminp keyword (see attachment). When 'subminp=temp' (see also the t+t, t+m options), you can define template files (MINP.X.tpl), where X is the number of the subcalculation, which will be copied to the top of the input file of the subcalculations. As a result, the setup created by our oniom modul can be overridden. Note that this way the printed results can be somewhat misleading: it can be seen that the last calculation of the example uses a 6-31G* basis set, but actually, it uses aug-cc-pVTZ for the oxygen atom and cc-pVTZ for the hydrogen atoms.

Regarding your second question, I think there are plans for constrained optimization but Prof. Kállay is more competent to answer.

Best regards,
Bence