After figuring out Blitting (caching everything onScreen into a single bitmap), I have been struggling with how to tackle collision detection. Since everything has been compacted into a single bitmap you would have to figure out the collisions by using the data alone. After picking Jesse Warden's brain for a bit, it seems that you either have 2 options.
1. You use a tile based system. This would be great tile-based games... (duh). And it would be great for the background/foreground of a sidescroller since they are going to be tiles anyways.
But I am not excited about using this for the characters/enemies/missles/etc of a side scroller. I feel it would be too restrictive to the characters to contain them within a bounding box and it would be a bigger hassle than it would be worth.
2. This brings us to option 2. Which is to use the blitting for the background as I mentioned above, but then use sprites/bitmaps for everything else. I will still try to cache as many animations into bitmaps as possible, but I will let them remain as their own display object on the screen and not merge them with the background. That way I can still do the hitTest and I will not be restricted to bounding boxes that a complete tile based system would create.