@iensen 😁

Below is a CR-Prolog program:

predicates
a(). b(). c().

rules

a.

-a :- not b, not c.
b :+. c :+.

b :- c.
c :- b.

Sparc+DLV gives:

SPARC  V2.49
program translated
{a, b, c}

{a, b, c}

The answer set {a, b, c} is duplicated, possibly because it has two abductive supports {b :+.} and {c :+.}.

Allow 'mod' to occur in program literals.

SPARC performs badly when output answer sets are large. It works faster if output is taken from DLV run on the translated file.

This program:

1{p(X):#s(X)}1 should work.

Hi I'm having trouble with the solveropts argument "pfilter" and "filter" with sparc 2.48 and DLV Dec 17 2012. Somehow the argument is not being passed from sparc to DLV. In the screenshot I compile using sparc and pass the compiled file and desired filter to DLV and it gives the expected output. Thanks.

Allow to use arithmetic expressions in definitions of a new constant, e.g:

const m = 1.

const n = m+1.

E.g, this program:
sorts

s = {a}.

predicates
p(#s).
q(#s).
rules
p(a).
q(a).

show p.

When being run in SPARC with an option -A should only output p(a) in answer sets.

| java -jar sparc.jar

does not work..

Allow anonimous sorts (e.g, a predicate definition of the form p({a,b,c}).

:- #count{X1 : p(X1)} > 0, q(X1).
should produce an error or a warning

When an unsupported version of clingo is used (<=5), Sparc should produce an error.
Some code needs to be added in ClingoSolver.java.

The program with

maxint = 100000.

does not compile..

Make CR-rules work when clingo solver is used.

This program:

#maxint=20.
sorts
#s={f(5),6}.
predicates
p(#s).
q(#s).
rules
p(f(X+1)).

produced the following swi-prolog program:

:- use_module(library(clpfd)).
:- style_check(-singleton).
set_1(6).
set_1(f(5)).
p :- set_1(X_0),X_0 #= f(X+1),[X_0,X] ins 0..20,labeling([],[X_0,X]).

main :-(p -> writeln(yes) ; writeln(no)).

Which has illegal atom X_0 #= f(X+1).

If a sparc program has many answer sets, they are all stored in a string buffer.
This can be avioided by streaming them directly to the terminal window (when the option -A) is used.

Given a program with empty sorts, such as:

sorts

#booleans = {true, false}.

#garage = {r2}.

#john = {r1}.

#football = {r3}.

#activity = {}.

#carry = {}.

Running java -jar sparc.jar program.sparc produces the following output and then terminates:

SPARC  V2.56
sort activity defined at line 11 column 2 is empty

E.g, the program
sorts

s = {a}.

predicates
p(#s).
q(#s).
rules
p(a).
q(a).

show p.

Should only produce p(a) when run in sparc with an option -A.

Given program

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Template for a SPARC file
%% Author:
%% Description:
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
sorts
    #people = {f(a,sara), f(b, sara), f(a, bob)}.
predicates
    father(#people, #people).
rules
    father(f(a,bob), f(a,sara)).
    father(f(a, bob), f(b, sara)).
    
display
father(X, f(Z, sara)).

We have complaint for the statement in the display section:

SPARC V2.54
temp.sp: non-ground term "f(Z,sara)" occuring in program as 2 argument of predicate father/2 
is not a program term

Currently, both default negation and classical negation are allowed before sort atoms:

For example, the following program is valid:

sorts

s = {a,b,c}.

s2 = {a,b,d}.

predicates
p(#s).
q(#s2).
r(#s2).

rules

q(X) :- -#s(X).
r(X) :- not #s(X).

However they do not have intuitive behavior (non of the literals q(d) and r(d) are implied).

Current proposal is to prohibit negations.
(Alternatively, we can alter the semantics to make not #s(X) mean "X is not in s" for any X in the domain of the program)

The following program:

sorts

s = {0,1}.

predicates
p(#s).
rules
1{p(X)}1 :- X<1.

Incorrectly classifies variable X as unrestricted.
The translation should be
1{p(X):s(X)}1 :- X<1,s(X).

When I use -o together with -A option, the output is repeated twice.