Expansion of double backticks and commas

[Daniel Kochmański]

Hello Andrea,

Generally ,,@ (and in a matter of fact ,@,@) are unspecified behavior because the specification says[1] that both , and ,@ operate on a FORM[2] while in these constructs they are invoked on something that under normal circumstances produces many values.

The behavior exhibited by clisp and gcl (and ecl) relies on the assumption that what appears after , (or ,@) is a form so it can
transform `(,X) to (list X), hence ``(,, at X) to `(list , at X) and that leads to (list 0 1 2 3).

Having that in mind CLtL2 book in one of appendices proposes the reader algorithm where splicing behaves the same as in SBCL. This
may be a bit cumbersome if we play a fool with the reader:

CL-USER> (let ((x '(1 2 3)))
           (car (second ``(,@, at x))))
, at 1

Under ECL this will signal an error: ,@ or ,. has appeared in an illegal position just as you'd expect from `, at foo.

Generally recursive splicing seems more intuitive but it allows results that are not valid lisp forms. Shallow splicing
on the other hand yields simpler (context-free) reader implementation and does not allow such values. All in all
not having the form after comma (or comma atsign) is an UBI.

[1] http://www.lispworks.com/documentation/HyperSpec/Body/02_df.htm
[2] form n. 1. any object meant to be evaluated. 2. a symbol, a compound form, or a self-evaluating object. 3. (for an operator, as in ``<<operator>> form'') a compound form having that operator as its first element. ``A quote form is a constant form.'' 

Best regards,
Daniel

--
Daniel Kochmański ;; aka jackdaniel | Przemyśl, Poland
TurtleWare - Daniel Kochmański      | www.turtleware.eu

"Be the change that you wish to see in the world." - Mahatma Gandhi



------- Original Message -------
On Friday, February 10th, 2023 at 21:13, Andrea Monaco <andrea.monaco at autistici.org> wrote:


> I see that SBCL expands this way:
> 
> 
> ``(,,@'(0 1 2) ,3) -> `(,0 ,1 ,2 ,3)
> 
> 
> 
> so apparently the second comma is consumed while the first one is copied
> before each element of the spliced list.
> 
> This behavior seems peculiar to sbcl, as clisp evaluates that form to
> (LIST* 0 1 2 '(3)) and gcl to (LIST 0 1 2 3): they both discard the
> other backtick-comma pair.
> 
> 
> I wonder what ANSI says on the matter. Does any of the two behaviors,
> the sbcl's and the clisp/gcl's, follows from the basic rules, for
> example those in
> http://www.lispworks.com/documentation/HyperSpec/Body/02_df.htm?
> 
> I'd appreciate any help. Thanks,
> 
> 
> 
> Andrea Monaco