> Let me mention that I like the rules for unwrapping.  I am sure it was not an
> easy compromise, but I think it is an elegant solution to the problems you
> point out with prior approaches, and will make macros significantly less
> painful to write.

I appreciate your saying so.

>    - Just to confirm: (car (syntax (1 2))) --> an error.

Yes.

>    - Could one write
>
>        (define-syntax or
>          (lambda (x)
>            (syntax-case x ()
>              [(_) #f]
>              ...
>
>      instead of (syntax #f)?  The definition of syntax objects
>      seems to allow it.

Yes.

>    - Related, is (quote (1 2 3)) a valid output of a transformer,
>      as opposed to (syntax (1 2 3)).  Again, the spec appears to
>      allow it.

Yes.  Of course, the resulting expression should result in a run-time
exception.

You might get better source-object correlation in some implementations
if you use #'#f or #'(1 2 3), but whether that's true is outside the
scope of the SRFI.

Kent