Re: "dict"ionary typing?

On 27-11-13 19:53, Alan Baljeu wrote:
> I'm eager for dictionaries to be a great feature in SWI,
> 
> including highly usable syntax.  Thus I'm pressing for 
> improvements before things get locked in.  Is anyone else 
> concerned about the proposed syntax?
> 
> I suppose things like this are the primary motive for 
> the new syntax.  
> 
> M.polar_radius() := D :-
>            D is sqrt(M.x*M.x + M.y*M.y).
> 
> ?- P = point{x:3, y:4}, X = P.x, D = P.polar_radius().
> 
> 
> 
> However, I'm struggling to accept this.  I feel it confuses 
> the standard meaning of =/2 as "unify terms".  Here the right 
> side feels like it ought to be a term but I gather it isn't.
> (And if it isn't a term, it can't be meta-programmed?)
> Newcomers will think functional notation and be baffled when 
> P = 2+3 doesn't work.

I'm also concerned about this. It feels inconsistent.

Couldn't is/2 be overloaded for the purpose of functional assignment, so
that the above would become

P = point{x:3, y:4}, X is P.x, D is P.polar_radius().

I realize the functional notation could not be used directly in goal
arguments then. But so what? That's Prolog. IMHO a consistent syntax is
well worth the slight added verbosity.

Kilian