[m-rev.] diff: minor improvements to builtin, private_builtin

[Julien Fischer]

On Fri, 1 Aug 2025 at 02:19, Zoltan Somogyi <zoltan.somogyi at runbox.com> wrote:
>
> This should not need review.

It broke the ROTD builds, it's getting a review ;-)

> Minor improvements to builtin modules.
>
> library/builtin.m:
> library/private_builtin.m:
>     Abort if a predicate intended only for C is used
>     when targeting Java or C#.
>
>     Fix obsolete module qualifier and obsolete comments.
>
>     Improve english and indentation in comments.

...


> diff --git a/library/private_builtin.m b/library/private_builtin.m
> index ca5baac96..9677e33ca 100644
> --- a/library/private_builtin.m
> +++ b/library/private_builtin.m

> @@ -1608,7 +1605,14 @@ nyi_foreign_type_compare(Result, _, _) :-
>  ").
>
>  unify_remote_arg_words(_, _, _, _) :-
> -    semidet_fail.
> +    % This predicate should only ever be called when targeting C,
> +    % in which case, its definition will the foreign_proc above,

will *be* the foreign_proc

> +    % not this clause.
> +    ( if semidet_fail then
> +        true
> +    else
> +        error("unify_remote_arg_words called")
> +    ).

unify_remote_arg_words is semidet. The new body you have added
is det, which breaks complilation in the non-C grades.

> @@ -1636,6 +1640,9 @@ unify_remote_arg_words(_, _, _, _) :-
>  ").
>
>  compare_remote_uint_words(_, _, _, _, Result) :-
> +    % This predicate should only ever be called when targeting C,
> +    % in which case, its definition will the foreign_proc above,

Again, will *be* the foreign_proc. And in other spots throughout
this diff.

Julien.