Re: Meaning of ':' and '{}'

Advait wrote:
Hello everyone,

I had a problem with ':' in the developement of expert shell in SWI



KNOWLEDGE BASE

:- op(900,xfx,':').
:- op(880,xfx,then).
:- op(870,fx,if).
:- op(550,xfy,or).



rule : if a or b or c or d or e then problem.
%% Note the ':'

END KNOWLEDGE BASE


c:/pl/test.pl compiled 0.00 sec, 1,088 bytes
Welcome to SWI-Prolog (Multi-threaded, Version 5.4.7)
Copyright (c) 1990-2003 University of Amsterdam.
SWI-Prolog comes with ABSOLUTELY NO WARRANTY. This is free software,
and you are welcome to redistribute it under certain conditions.
Please visit http://www.swi-prolog.org for details.

For help, use ?- help(Topic). or ?- apropos(Word).

1 ?- listing.

Yes
2 ?-


This shows no rule.
If we replace ':' by any other atom it gives proper result.
Does any thing wrong with ':' ?
Please, can any one give me the solution on it to use in SWI ?

Two issues here:

1. do you really want to give your rules directly to the Prolog
compiler? The Prolog compiler will just take them as a collection
of facts (for the predicate :/2 or then/2, see below).
I find it more likely that you will need a sort or rule-compiler,
i.e. a program that reads your rules as data structures, and
transforms them into something more useful (like a decision tree).
Such a program would use e.g. read/2 to read the rules, and in
that case you will not have any problem with using the colon,
because _your_ program decides what the colon means.

2. If you give your rules to a Prolog compiler with a Quintus-style
module system (as you have done), the colon will be interpreted as
a module qualification. So your example rule is taken as a fact for
the predicate then/2 in module 'rule', because it is equivalent to

rule:then(if(or(a, or(b, or(c, or(d, e))))), problem)).

Prolog compilers that don't treat the colon specially will take it
as a fact for :/2 instead, but that probably clashes with the name
of a built-in predicate. So you should either choose a different
functor, or change the precedences such that the colon is not the
toplevel functor.

However, i really think you do not want to compile these rules as
facts at all.

-- Joachim

.