Landin's Knot in Dogelog Player (Was: ANN: Dogelog Player 1.3.5 (Cyclic Terms))

Mild Shock <[email protected]> Fri, 15 Aug 2025 19:55:46 +0200
Newsgroups comp.lang.python
Message-ID <[email protected]>
Dogelog Player is a Prolog system written
in 100% Prolog for the target platforms JavaScript,
Python and Java. We recently introduced the
display of cyclic terms in the top-level.

As a next iteration we turned all built-ins
upside down, gave it either a visitor pattern
or union find, so that they can deal with cyclic
terms. The final step is then to untie Landin's Knot.

The new predicates term_decompose/2 and
term_canonical/2 have found applications in
library(sequence) and library(aggregate). The
result is primarily a structural compare among

cyclic terms, which the end-user can lift
to a canonical compare.

See also:

Landin's Knot in Dogelog Player
https://qiita.com/j4n_bur53/items/f096445f959998768316

Mild Shock schrieb:
> Dear All,
> 
> We are happy to announce a new edition
> of the Dogelog player:
> 
> - Enhanced Core:
> We provide internally unify_cyclic() which implements
> Hopcroft & Karp (1971) algorithm and can deal with
> cyclic terms. It is accessible via the built-in
> predicate (=)/2 and also used by the Prolog interpreter
> for solving clauses. Similar there is an enhanced
> (==)/2 available to the end-user.
> 
> - Enhanced library(tools):
> The debugger is now cyclic term aware and uses a term
> factoring based on a new same_term/2 before printing
> a goal. The display is simply a conjunction Equation_1,
> .., Equation_n, Goal, where the equations are necessary
> to break cycles. As before Prolog variables are still
> printed by their serial number.
> 
> - Enhanced library(session):
> The answer display is now cyclic term aware and uses
> the same term factoring as the debugger. Answer
> substitutions and cycle breaking equations are
> interspersed. We tried hard to not break old features,
> such as fresh singleton variable names, fresh multiton
> variable names and marked variables suppression.
> 
> Have Fun!
> 
> Jan Burse, https://www.herbrand.ai/ , 02.08.2025