Hmmm, that seems like the tail wagging the dog. Grouping components by package types seems like a convenience only for manually kitting parts because it saves having to hand write long lists of prefixes on the label on a bag of the same valued parts. Most of the time people want components numbered by their association to a particular functional part of the circuit. For example in a non-inverting opamp circuit there may be 3 resistors all of different values and maybe package types but because they are all part of that opamp circuit, it makes sense to number them R(m+1), R(m+2), R(m+3). If there are two such circuits then it makes sense to number the parts for the second circuit would preferably be numbered R(n+1), R(n+2), R(n+3). (m and n are just arbitrary base numbers because they probably won't be numbered in a simple sequence such as R5, R6, R7, R8, R9, R10)) The component numbering will then also be grouped by functional association on the PCB which works really well once any component in a functionally associated group is found but it can be time consuming trying to locate a functional group on a complex board. Annotating by rows or columns in a schematic breaks that association because the XY grid of numbering in the schematic will almost certainly not be mapped onto the PCB so the reference designator (prefix) numbering gets scattered over the PCB. Being able to annotate by rows or columns in a PCB also breaks that association but it makes finding a component on the PCB during debugging much easier. Can you give an example of where your feature would be helpful?