[Haskell-cafe] Is Curry alive?

[Richard O'Keefe]

On 2/11/2010, at 1:27 PM, Gregory Crosswhite wrote:

> Hey everyone,
> 
> This is a little off-topic, but I just ran into a problem which might benefit from being attacked by a logic language,

Why not describe the problem?

> so I've been looking for a good one to try out --- and hopefully one that has a very efficient implementation since I want to iterate through billions and possibly trillions of nondeterministically generated solutions.

I think about the practical success of Model Checking, and wonder whether it
might be better NOT to iterate through so many.

> I've also been looking at Prolog but I am having trouble seeing whether I can process N non-deterministic solutions in O(1) space (rather than first generating a O(N) size list),

The point of backtracking search is that you need only space for the current
candidate solution, not for all solutions visited so far.  So much so that the
Iterative Deepening family of search algorithms cheerfully *revisit* graph nodes
in order to save time over all.

> and I checked out Mercury but the documentation for it is a bit sparse.
Packl
The Mercury documentation I downloaded in March comes to 773 pages.

faq.pdf			  6 pages
library.pdf		470 pages
reference_manual.pdf	150 pages
transition_guide.pdf	 10 pages (Mercury for Prolog programmers)
user_guide.pdf		137 pages

Packing the tutorial into a single HTML file gives another 19 pages.
Ralph Beckett's tutorial adds another 53 pages of PDF.

So that's 845 pages all up.  "sparse"?



"a bit sparse"?