Scaling of geometry is simple stuff. Basic mathematical calculations.
Yet for some reason, we can't do it easily to parts or bodies. As it stands, you can scale imported geometry during import (sometimes useful), and you can scale geometry that you've derived into another part (useful, but a workaround at best).
To my knowledge, there is no easy way of simply scaling the geometry within the part, at a feature on the feature tree. Why not? It's clearly technically possible. So why force the user to duplicate (derive) a part every time they want to adjust the scale?
Let's say I'm doing some rough concepting work, and I've modelled a control box using several features. When I put it into the assembly, I find that the control box is a little bit too big for the buttons I need to put onto it. As it's only rough, I'd like to simply scale it down to say 80% of its size. How do I do this? Well I've got two choices. Edit each of the features in the box to adjust its size, or create a new derive part and adjust the scale on creation. The first option is a lot of effort for a simple change, and the second results in an additional, useless part model clogging up the vault. Why?! It should be so simple to provide a scale feature in part?
While we're about it, this feature could be effectively applied to multi-body parts, by allowing you to scale descreet bodies independently of each other.
Seems like a simple winner to me!