Re: Dollar sign.

On 29/01/2014, at 10:02 PM, Jan Wielemaker wrote:

> See
> 
> http://www.swi-prolog.org/pldoc/doc_for?object=section%28%27sec:topvars%27%29

"Reuse of top-level bindings" says what $X means
when X is a variable.

From that description, I for one expected this to be
a *lexical* thing, so that
	$X	saved value of X
	$(X)	error
	$foo	error
etc.

It is way too confusing if $foo is '$'(foo)
but $Foo is not '$'(Foo).

Consider also
	?- ?- Y = true, $Y.
ERROR: variable `Y' does not exist

I see that, and say "WTH?  Y most certainly does exist,
look, there it is".  It would be clearer if this message
were
	ERROR: the interactive level has not yet saved
	any value for Y.

I suggest that $<variable> should be handled in the parser,
and that $ should *not* be a predefined operator.  The
reason I suggest that is this example:

	?- V =.. [$,_], portray_clause(g(V, V)).
=>	g($A, $A).
	V = $_G1611.

The system should not (by default) *produce* $<variable> in output.
The simplest way to ensure that is to stop $ being an operator.