Hacker News Reader

Prolog is cool but has exponential complexity which makes it impractical for real world use.

The rete algorithm was supposed to be a solution but has anyone applied rete to a prolog implementation yet?

Isn't RETE a forward-chaining algorithm while Prolog is a backward-chaining language? I'm not sure you could use RETE for the actual Prolog semantics. Currently, fast implementations of Prolog use a virtual machine, the most popular is WAM, that can be compiled then to machine code (GNU Prolog does it). Also some implementations have JIT indexing, which improves performance too.

Prolog itself does not have exponential complexity. It depends on the problem you're solving.

Reply on news.ycombinator.com