Re: [eclipse-users] Predicate declarations in Eclipse

From: Joachim Schimpf (Independent Contractor) <"Joachim>
Date: Mon, 30 Apr 2007 17:07:57 +0100
Siddharth Angrish wrote:
> Hi
> 
>         I am trying to port a big prolog code (FSA6) to ECLiPse. The
> code is compatible
> with swi/sicstus/yap(SSY) prologs. I have a few queries regarding
> predicate delarations:
> 
> 1) SSY support the keyword, multifile. Does EClipSe use "dynamic" for
> the purpose?

Yes, the conmpatibility packages currently define multifile as dynamic.


> 
> 2) Does "dynamic" also serve the purpose of the keyword
> "discontiguous" which is there
> in all SSY

Well, Eclipse supports discontigous, but doesn't recognize the
module: prefix.  So you may be better off redefining it as dynamic
for your purposes.  Add the following to the compatibility package
you are using:

:- export discontiguous/1.
:- export op(1150, fx, (discontiguous)).
:- tools(discontiguous/1,discontiguous_/2).
discontiguous_(Pred, Module) :-
	dynamic(Pred)_at_Module.

and add discontiguous/1 to the existing directive
:- export eclipse_language except ...
to hide the original ECLiPSe definition.


> 
> 3)  SSY support the following declaration:
>       multifile user:help_info/4.
>    which I think says that-> To the predicate help_info/4,  "defined"
> in module "user",
>   new clauses can be added in the current file.
> 
> I modified it to: (only in the first file)
>     dynamic user:help_info/4
> 
>             However, this module does not exist anywhere in the code I
> am porting.

If you have loaded the quintus compatibility library (directly, or
indirectly through sicstus or swi), then the user-module exists.


> I suppose it is created on the fly in SSY, a feature which ECLIPSE
> does NOT support.
> 
> To circumvent the "non-availability" of this feature I created the user module
> myself as  following:
> -----------user.ecl------------------
> :- module(user).
> help_info(A,B,C,D).
> ----------------------------
>       Then in the module I am working on, I included the delcaration
> :- lib(user).
> 
> Upon compilation I am getting the error: ( I am using->
> lib(myLibraryName). to compile)
> [[type error in dynamic user : help_info / 4 in module fsa_array]]


Again, ECLiPSe's native dynamic directive does not recognise the module:
prefix, but the definition in lib(quintus)/lib(sicstus) does. You haven't
said which compatibility library you are using.  If you use swi, then
add in swi.ecl the line

	(dynamic)/1,

to the reexport ... from quintus directive.


Furthermore, ECLiPSe doesn't allow (for good reasons) program clauses
to be module-prefixed.  We can however add limited support (for dynamic
clauses only) to the compatibility package as follows:

% Support Quintus-style "qualified clauses" (for dynamic predicates only)

:- export macro((:)/2, t_colon_clause/2, [clause]).

t_colon_clause(QualClause, []) :-
	( QualClause = (Module:Head:-Body) ->
	    Clause = (Head:-Body),
	    functor(Head, N, A)
	;
	    QualClause = (Module:Clause),
	    functor(Clause, N, A)
	),
	( current_module(Module) -> true ; create_module(Module) ),
	export(N/A)_at_Module,
	assert(Clause)_at_Module.

I will add this to lib(quintus).


> 
> what can be the problem?
> 
> 4) How can one see all the "loaded modules " in the current instance of eclipse

Use current_module/1.


> 
> 5) Is there any way/need of getting a "public declaration" for a
> predicate in ECLIPSE

This should more or less equivalent to export/1.  If you use lib(quintus)
or sicstus, it is already defined.  lib(swi) is lacking a reexport directive
for it (i'll fix that).


-- Joachim
Received on Mon Apr 30 2007 - 17:08:09 CEST

This archive was generated by hypermail 2.3.0 : Thu Feb 22 2024 - 18:13:19 CET