goal_expansion/2 transforms head of clause

[flexoron]

v2.83.15

?- [user]. b(('rule :- goal.')). goal_expansion(p(B),(B)) :- b(B). p(_) :- p(_). 'rule :- goal.'. end_of_file.
   true.
?- p(X). % is subject to goal_expansion, isn't it?
   error(existence_error(procedure,p/1),p/1). % unexpected
?- 

[infradig]

And yet...

➜  trealla git:(main) ✗ cat x.pl
:- dynamic(p/0).
goal_expansion(p,q).
p :- writeln(hello), p.


➜  trealla git:(main) ✗ tpl x.pl
?- listing.
p :-
   writeln(hello),q.
   true.
?- p.
hello
   error(existence_error(procedure,q/0),q/0).
?- 

[flexoron]

$ cat xx.pl
:- dynamic(p/1).
b((goal)).
goal_expansion(p(B),(B)) :- b(B).
p(_) :- p(_).

$ tpl xx.pl
?- listing.
b(goal).
goal :- % ???
   goal.
   true.

$ cat xxx.pl
:- dynamic(p/1).
b((b,u,g)). b. u. g.
goal_expansion(p(B),(B)) :- b(B).
p(_) :- p(_).

$ tpl xxx.pl
Error: permission error modifying user:(,)/2 % looking at issue the other way around.
?-

[infradig]

I'm not sure if goal expansion is required to apply to top-level input. All the dcumentation says compile time. However Scryer & SWI Prolog do it (Trealla doesn't), so i'm not sure now.

[flexoron]

I get the result with v2.84.3

$ cat xxx.pl
:- dynamic(p/1).
b((b,u,g)). b. u. g.
goal_expansion(p(B),(B)) :- b(B).
p(_) :- p(_).

$ swipl -q xxx.pl 
?- p(X).
X = (b, u, g).

$ scryer-prolog -f xxx.pl 
?- p(X).
   X = (b,u,g).

$ tpl xxx.pl
?- p(X).
   true.

[infradig]

As I said, Trealla doesn't do goal expansion at the top-level input.

…

On Mon, Oct 27, 2025 at 7:53 AM flexoron ***@***.***> wrote: *flexoron* left a comment (trealla-prolog/trealla#901) <#901 (comment)> I get the result with v2.84.3 $ cat xxx.pl :- dynamic(p/1). b((b,u,g)). b. u. g. goal_expansion(p(B),(B)) :- b(B). p(_) :- p(_). $ swipl -q xxx.pl ?- p(X). X = (b, u, g). $ scryer-prolog -f xxx.pl ?- p(X). X = (b,u,g). $ tpl xxx.pl ?- p(X). true. — Reply to this email directly, view it on GitHub <#901 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFNKSEU6HGBFPYJVFLI3GOT3ZUYD5AVCNFSM6AAAAACINHQNHSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTINBYHE2TANZUGQ> . You are receiving this because you commented.Message ID: ***@***.***>

[flexoron]

Still a bit uncertain about what it means: top-level input
When a progam is called like so:
tpl -g main xxx.pl with main :- pp(X,Y),write(X/Y),fail. pp(X,Y) :- p(X,Y), write(X/Y).
Is this top-level user input?

I mean trace shows that tpl does it right somewhere in the middle.

[infradig]

Yes. As is the REPL loop.

…

On Mon, Oct 27, 2025 at 10:25 AM flexoron ***@***.***> wrote: *flexoron* left a comment (trealla-prolog/trealla#901) <#901 (comment)> Still a bit uncertain about what it means: top-level input When a progam is called like so: tpl -g main xxx.pl with main :- pp(X,Y),write(X/Y),fail. pp(X,Y) :- p(X,Y), write(X/Y). Is this top-level user input? — Reply to this email directly, view it on GitHub <#901 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFNKSEQKFV37FTEKGFM3TQD3ZVJ7JAVCNFSM6AAAAACINHQNHSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTINBZGA3TEMJRHE> . You are receiving this because you commented.Message ID: ***@***.***>

[flexoron]

ah thnx. no longer bug.
b((w,o,r,k,s)). w. o. r. k. s.

You're not sure.
I'm not sure.
You close the issue.