Re: [boost] [proto] design of expr<>

From: Larry Evans (cppljevans_at_[hidden])
Date: 2008-03-20 18:59:28

On 03/20/08 16:53, Eric Niebler wrote:
> Larry Evans wrote:
>> What about recording the arity *with* the tag instead of in
>> the expression associated with the tag?  IOW, instead of:
>>
>>    expr<Tag,Args,Arity>
>>
>> have:
>>
>>    expr<tag_arity<Tag,Arity>,Args>
>>
>> Why?  Well, it directly associates the arity with the Tag instead
>> of with the expression.  After all, sin(X) is not expressed
>> as sin(x,1). [or in lisp terms, (sin x) is not expressed as
>> (sin x 1) ].
> 
> What problem does that solve?
> 

I don't think it solves any problem that can't
already be solved, but it makes the solution
clearer.  It's clearer because the arity
is associated with the tag *not* the
expression.  The arity of a tag does not depend
on the expression; so, it should not be in
the expression.  Instead the arity of the tag
determines the expression validity.

OK, but you may say "the arity *still* is in
the expression:

   expr<tag_arity<Tag,Arity>,Args>

"; however, it's "more closely" associated
with the tag than the expression, so it's
really more a matter of style, I guess.
OTOH, I would guess it would make specifying
morphisms between algebra's easier since
any morphism would map equal arity's to
equal arities; and if the arity were
directly associate with the tag, as in
tag_arity<Tag,Arity>, then the mapping
would be easier to understand and verify
(with maybe a boost concepts class
which assured the pairs in the mpl map
had equal arities on both sides).