No, they're just sort of rough concepts that I've thought about a little bit, but not tested to see how viable they are. I haven't written any down before, but here's a few options:
For example, say you wanted to carve a hexagon into a rectangle:
The default carve behaviour will start at one of the faces of the hexagon and simply iterate all faces in a dumb way. This usually means the carve will rotate around the solids:
One idea to improve this is to be smarter about picking the next face to carve - the one with the greatest angle between plane normals should produce a larger "chunk" per carve (I call this the "alternate" method, because it tends to alternate between sides):
A second idea is to draw "lines" from the corners of the carving solid to the corners of the carved solid. This would form a more pleasing "bevelled" type look to the carve result and might end up easier to manage. There are quite a few gotchas in this implementation, though. Some corners would need multiple "lines", depending on the positioning of the corners, and a naive implementation creates more objects than the classical method. It also might not work so well when the carved object doesn't completely contain the carver, though it's probably possible to code that in. (I call this the "corners" method)
Another idea is to first carve the bounding box of the object (keeping the result), and then do the full carve on that resulting object. This minimises the length of non-straight lines in the result, but potentially creates more brushes in the result. (I call this the "bounding" method)
Again, I'm not entirely sure how great these ideas are (especially the "corners" one), and there are probably other (better) methods that I haven't thought of. These different methods can probably be combined to improve the results further (e.g. the "bounding" method and the "alternate" method can probably be combined quite easily).
Edit: corners might work if both bounding boxes are intersected and carved before the method is applied:
First the two bounding boxes are intersected:
Then both objects are carved by the intersecting bounding box using the "bounding" method. After that, the result of that carve then has the "corners" method applied:
I dunno, it might work. "Corners" seems like it would produce the most desirable result in most cases, but it also feels like it won't work (producing invalid solids) in some situations, or that it might be the most difficult one to implement.