Multiple level (say, N-level) undo in Mathematica would perform the same function single-level undo performs now, only more than once. The idea is to put the whole state of the application and its top document back to what it was up to N steps back. In a significant number of cases, this would require some sort of disk operation to save the old state before the evaluation of each line. N old states would have to be saved. In some cases, this act of saving old states would slow Mathematica down. However, if the feature were skillfully implemented, the slowdown would typically be quite small.
Building this feature is a pedestrian accomplishment in todays software marketplace. As an example, Adobe Photoshop, which in this respect faces just as many (and very similar) undo implementation challenges as Mathematica, has effectively infinite undo levels.
On Fri, Sep 7, 2012 at 4:56 AM, Bill Rowe <firstname.lastname@example.org> wrote:
> On 9/5/12 at 3:07 AM, email@example.com (Murta) wrote: > > >1- More Undo! (1 Undo is a shame) > > The key issue here is what a multiple level undo would undo. If > the thought is simply to be able to undo text typed into the > front end with no attempt to undo computations or concern > regarding variable state reflect the result of undoing changes > to text input, that is one thing. But to have variable state, > computations etc reflect the undone changes to text input is > quite something else and probably not possible to do in general. >