OK, thanks. But, a lot of what I said still stands: You should first get your base sketch worked out. Create your base sketch and your user parameters, and then exercise the heck out of it, making sure that it behaves as you expect before proceeding. In this case, that does not happen. So, that should still be your first step. In your simple example, I would not use those diagonal dimensions to control the shape, I would probably use angle dimensions, but that is preference. The key is testing your sketch. If it behaves as you want, then great.
Regarding offset: I understand that offset is sometimes useful, even necessary. I just tend to avoid it myself, whenever I can, because I know that downstream references to the offset geometry can be fragile. See: not-sure-why-this-explodes-so-much . So, I avoid it when I can. Sometimes Thin Extrude can work around this - just draw one set of curves, and use Thin Extrude to generate a thickened version of that set of curves. Sometimes it is easier to draw a second set of geometry. You have to consider the whole cost here. Yes, the initial creation might take longer, but if you spend more time later fixing things, is that initial shortcut worth it?
Regarding broken projections. I do not see any broken projections in the sample model. I may have done something differently, but the projections here are solid in my testing. It really helps if your projections come from a face, and not edge-by-edge. Edges are more fragile. If you just sketch on the top face of Extrude1, I expect the projections will never fail. Now, what I do see is that the additional lines for the dovetails come loose. I suspect that will be better if you fix your initial sketch behavior. The reaction of that first sketch to the parameter changes was so severe, and involved errors along the way, that I suspect that there is some correlation to those error states.
Jeff Strater
Engineering Director
Fusion
