Arthur Smyles wrote:
> You currently have a private procedure named ra:car+cdr. It's proven useful by
> the fact that you use it to implement car, cdr, caar, etc. I think that this would be
> useful for other types of deconstructions of con cells and would like to see it
> made public. Secondly, I suggest calling it decons.

The implementation of car and cdr are likely to change because cdr may
allocate, while car never does.  The current implementation in terms of
car+cdr means that car may allocate even though it does need to.

Now, that's not to say car+cdr (a name which has precedent in SRFI 1) is
not useful.  I just don't think you've motivated it sufficiently, so I'm
not convinced it needs to go into the specification.

David