On Wed, Mar 4, 2009 at 12:50 PM, Heinrich Apfelmus
... The unusual thing about your implementation is probably that you're tying a knot by making both moveUp and moveLeft record fields. This reminds me of
Weaving a web. Ralf Hinze and Johan Jeuring. 2001. http://www.informatik.uni-bonn.de/~ralf/publications/TheWeb.ps.gz
The problem with knot-tying / sharing is of course that they are tricky to update. What about the crucial function
update :: ZContext a -> Maybe a -> ZContext a
that changes the data at a leaf? I think that with your current approach, you'd have to regenerate the whole context which pretty much defeats the purpose of a zipper.
Hi Heinrich and thanks for your reply. I haven't read the paper (which I'm going to read now) but I don't fully understand your point. I'd (and indeed I did) write 'update' as: update z x = z { this = this z >> Just x } exploiting the '>>' operator's logic. How would this differ from the corresponding 'update' in the original Huet's FP? Maybe I don't get how my update would impact performances. In both cases the update functions leave context unchanged, don't they? If my update function have to replicate the context since functional values are immutable, doesn't Huet's do the same? Thank you for any further comments. Cristiano