Matched data etc is stored in the blessed object.
For all predicates, there is a "master" method that dispatches calls to other predicates of the same name, depending on number of parameters.
If a clause referenced a predicate, then code is inserted into the RE which calls the sub. Depending on the return value of the sub, the code fails/succeeds matching.
At least, that was how far I have gotten in my planning.
I don't know how to unsets unification when it do RE backtracking (there is no support for that in the (?{ }) construct?). And I probably have missed other stuff.
This idea probably reminds more than me about the idi.. uhm, heroes which implemented logical circuits in the game of Life. :-)
Update: Please don't flame me if I got the vocabulary wrong; it was a long time ago I read Prolog. :-)
In reply to Re^2: Using RegExp backtracking to implement Prolog? :-)
by BerntB
in thread Using RegExp backtracking to implement Prolog? :-)
by BerntB
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |