🚀 go-pugleaf

cduguay@lcraft.ersys.edmonton.ab.ca (Claude Duguay) writes:

>This may seem silly or I might be missing part of this conversation but
>would it not be easier and just as portable to use a simpler 
>implementation of retractall/retract_all rather than creating elaborate 
>solutions to emulate compiler directives?

The problem is a general one, so we want a general solution.
retract_all is just one example, there are undoubtably many others.

Also a user's implementation of retract_all may be much less efficient than
the built-in implementation.

-- 
Fergus Henderson             fjh@munta.cs.mu.OZ.AU      
This .signature virus is a self-referential statement that is true - but 
you will only be able to consistently believe it if you copy it to your own
.signature file!

In article <uLBXRB2w165w@lcraft.ersys.edmonton.ab.ca>, cduguay@lcraft.ersys.edmonton.ab.ca (Claude Duguay) writes:
> This may seem sill or I might be missing part of this conversation but
> would it not be easier and just as portable to use a simpler 
> implementation of retractall/retract_all rather than creating elaborate 
> solutions to emulate compiler directives:

>         myretractall(Term) :-
>                 retract(Term),
>                 fail.
>         myretractall(_).

There are two snags:
(1) There may be substantial performance advantages to using the built in
    method.  (For example, Quintus Prolog uses the Lindholm/O'Keefe model
    for data base changes, and retractall/1 could count as one "clock tick"
    rather than several.)
(2) In a Prolog with modules, such as Quintus Prolog or ALS Prolog, you will
    have to add some kind of declaration to myretractall/1, while the built
    in version is already module-aware.
(3) The really big problem is that the indicated code for myretractall/1 is
    wrong.  For example, if I have
	p(a).
	p(b) :- q(c).
    then retractall(p(_)) is supposed to retract _both_ clauses, but this
    definition will only retract the first.  You would need

	myretractall(Term) :-
		clause(Term, _, Ref),
		erase(Ref),
		fail ; true.

    Because some Prolog vendors have not provided clause/3 and erase/1, you
    would have to do something like

	myretractall(Term) :-
		clause(Term, Body),
		my_retractall_aux(Term, Body),
		fail ; true.

	my_retractall_body(Head, Body) :- Body == true, !,
		retract(Head),
		!.
	my_retractall_body(Head, Body) :-
		retract(Head :- Body),
		!.

It would all really be much simpler if vendors would just provide retractall/1
with the same name and semantics. 

-- 
RMIT policy requires the use of Gender Inclusive Language in all publications
and in all information conveyed to students, staff, and potential job
applicants.  Lapses into English are the sole responsibility of the author.
The Constitution of Australia does _not_ guarantee Freedom of Speech.