Re: a bug in current_functor

On 10/29/2013 04:37 PM, Ulrich Neumerkel wrote:
> The manual reads:
>
> | current_functor(?Name, ?Arity)
> | Successively unifies Name with the name and Arity with the
> | arity of functors known to the system.
>
> While accurate, the locution "known to the system" does not seem to
> ring a bell to many. It means: access to system internals, don't.

Not entirely. It allows generating an upperbound, just like
current_atom/1 does. Indeed, I wouldn't really know a sensible usage for
this upperbound except for debugging purposes.

> Is there any use to it?  Except for debugging purposes, if the number
> of registered functors goes up for no reason?  But then a better place
> would be some system module.
>
> It cannot be for backwards compatibility, since in DECsystem 10
> Prolog, the second argument used to be a structure.
>
> current_functor(Name,Functor)
>     Generates  (through backtracking) all currently known functors, and
>     for each one returns its name and most general  term  as  Name  and
>     Functor  respectively.  If  Name  is given, only functors with that
>     name are generated.

I really do not recall why it got there. I assume because I was working
sort of systematically through compatibility with C-Prolog or Quintus or
whatever Prolog. I see little reason to remove it or change it either as
long as nobody comes up with some sensible usage scenario. Possibly a
comment that this is only sensible for debugging and not all systems
provide it (I think quite a few systems do not have the concept; they
simply use atom/arity pairs) and have no clue whether there are or ever
have been compounds with that pair.

	Cheers --- Jan