[bitc-dev] Addition of DEFREPR

[Swaroop Sridhar]

Jonathan S. Shapiro wrote:

> I think that the following is a (small) counterexample to what you say:
> 
> (defrepr dr
>    ((tag swizzled unswizzled)
>     d:double
>     (tag left right)
>     (case (swizzled (i:int32 j:int32))
>           (unswizzled (b:bool)))
>     (case ((swizzled left) (k:int32)))))
> 
> [Note that the 'd' field has moved up between the tags. Also note that
> the (invariant ...) form was an error in the spec. Such forms have no
> special annotation.]
> 
> Because in this case the constructor would be written:
> 
>   (dr swizzled 2.0 right 5 6 7)
> 
> However, this may not actually be important, because the ordering
> between the tags and the values is independent. That is, the left to
> right sequencing of tag-ids must be preserved, and the left to right
> sequencing of values must be preserved, but (at least conceptually) we
> *could* decide that the following two constructor calls are equivalent
> in meaning to the one I gave above:
> 
>   (dr swizzled right 2.0 5 6 7)
>   (dr 2.0 5 6 7 swizzled right)

Or, we can say that all constructor labels are collected to he beginning 
and written as:

   (dr#swizzled#right 2.0 5 6 7)


> However, I think this can only be rationalized if we define the tags to
> be some sort of label (i.e something other than expressions), because it
> violates the usual rules for application.
I think this is OK. In ML, constructors are treated specially from 
normal application anyway (for example, they can appear in case 
statements, but normal applications cannot).


>>In the above scheme, the case tags are scoped within the defrepr, 
>>because they are always mentioned along with the defrepr's name. So, it 
>>is as if we wrote these tags as dr-swizzled-right, etc, which are 
>>guaranteed to be unique at top level. Right?
> 
> 
> I also thought that this could be made to work, but it has strange
> consequences. Consider:
> 
>   (let ((yes 1))
>     (dr yes 2.0 left 5 6))
> 
> where the intention is that this will turn into:
> 
>   (__con_dr-swizzled-right __tag_dr-swizzled-yes 2.0
>         __tag-dr-direction-left 5 6)
> 

The above scheme solved this problem as well.

    (let ((yes 1))
      (dr#swizzles#left 2.0 5 6))


Swaroop.