Shiro Kawai wrote:
> SRFI-10 could be used to introduce literals for user-defined
> types.  Chicken, Gauche, Guile and STklos support it, AFAIK.
>
> Unfortunatley it has some issues, though: (1) The notation (#,)
> conflicts with R6RS.  (2) The semantics of read-time evaluation
> is rather defined in ad-hoc way, and it isn't clear what
> context the constructor is evaluated.  (3) Macro expansion comes
> after read-time evaluation, so there's no way to write a macro
> that generates #,-form, etc.

Thanks, but it's a nonstarter for me if you can't write '(x . y) to
construct a constant pair.  Beyond that, you give good reasons why SRFI
10 is not adequate in this setting.

David