Best Practices: QAbstractItemModel font metrics computations

Question about Best Practices in relation to QAbstractItemModel font metrics-based computations:

I was surprised to see that Item View MODEL implementations are supposed to generate presentation-oriented data, e.g. SizeHint (-Role) from the data() method. We also need to use font metrics to generate QPixmaps for the DisplayRole.

I'm now registering a sample QWidget with one of our QAbstractItemModel implementations to make font and font metrics available to that implementation's geometry computations. This doesn't feel like the right direction.

These presentation-oriented roles are for the default item delegate. You are free to implement your own delegate that controls all aspects of the presentation without accessing the model.