clpz + reif is a killer Scryer Prolog feature that is not really widely discussed

[jjtolton]

I'm new to Prolog, relatively speaking. I assumed all the varieties of Prolog were more or less interchangeable. And from discussions I assumed that the best among them was SICStus -- which makes sense, being that it is commercially supported.

This weekend while working through some difficult code, I actually spent the time to setup SICStus and their Spider IDE, and the results were surprising.

The most surprising thing was immediately how much I missed the combination of clpz + reif. I was able to port reif to sicstus, but I immediately began to miss the use of clpz.

The SICStus options are:

11.3.102 'if/3'
---------------

Synopsis
--------

'if(+P,+Q,+R)'

   _If_ P _then_ Q _else_ R, for all solution of P.

Arguments
---------

:P
     "callable", must be nonvar
:Q
     "callable", must be nonvar
:R
     "callable", must be nonvar

Description
-----------

Analogous to

     if P then Q else R

but differs from 'P -> Q ; R' in that 'if(P, Q, R)' explores _all_
solutions to the goal P.  There is a small time penalty for this--if P
is known to have only one solution of interest, then the form 'P -> Q ;
R' should be preferred.

   This is normally regarded as part of the syntax of the language, but
it is like a built-in predicate in that you can write 'call(if(P,Q,R))',
with identical effect if it does not contain any cuts.

   Cuts in P do not make sense, but are allowed, their scope being the
goal P.  The scope of cuts in Q and R extends to the containing clause.

'if_then_else(IF, THEN, ELSE, VALUE)   since release 4.8.0'

     True if IF=1 and VALUE=THEN, or IF=0 and VALUE=ELSE.  Corresponds
     to if-then-else expressions in most programming and modeling
     languages.

if/3 may seem similar to reif's if_/3 but it is not.

| ?- if(3 #< X, A=1, A=2).
A = 1,
X in 4..sup ? 

is NOT the same thing as

:- use_module(library(clpz)).
:- use_module(library(reif)).

?- if_(3 #< X,
       A=1,
       A=2
      ).
%@    A = 2, clpz:(X in inf..3)
%@ ;  A = 1, clpz:(X in 4..sup).

The fact that Scryer ships with this standard is a really, really powerful API combination that I don't think is widely known. I think it allows for much more interesting and powerful programs out of the box.

[jjtolton]

Not to mention that clpz_t/1 allows you to convert any clpz expression to a reified predicate is really nice. The mental agony of trying to convert a A #=< B into an if_then_else/4 compliant something like, T = min(A - B+1, 0) so I can use it with if_then_else/4 seems like unbearable mental agony. In the United States they teach "the alligator eats the bigger number", and to this date I still hate alligators because of it. But worse is trying to wrestle an alligator into some other kind of weird expression to satisfy if_then_else/4.

[UWN]

I was able to port reif to sicstus

In Appendix A of Indexing dif/2 you find reference to the original version which was for SICStus.