Re: What is the difference between ‘cl-labels’ and ‘named-let’?

> The Emacs Lisp reference manual does not describe the
> difference between the new-ish (since Emacs 28) ‘named-let’
> and ‘cl-labels’ (other than the syntax, which allows
> ‘cl-labels’ to define many recursive local functions, while
> ‘named-let’ can only define one function per call).  Are
> these two different in some significant way?

As it so happens, no: `named-let` is a fairly thin wrapper around
`cl-labels`:

    (defmacro named-let (name bindings &rest body)
      [...]
      (let ((fargs (mapcar (lambda (b) (if (consp b) (car b) b)) bindings))
            (aargs (mapcar (lambda (b) (if (consp b) (cadr b))) bindings)))
        [...]
        `(funcall
          (cl-labels ((,name ,fargs . ,body)) #',name)
          . ,aargs)))

> The description for ‘named-let’ says that recursive calls in the tail
> position are guaranteed to be optimized tail calls, while the
> description for ‘cl-labels’ does not mention this.
> Is this not the case for ‘cl-labels’, too?

No and yes:

- No, that guarantee does not hold for `cl-labels`.
- Yes, `cl-labels` optimizes tail-calls just as well as `named-let`
  (simply because `named-let`s tail-call-optimization is actually
  implemented in `cl-labels`), but that covers only those `cl-labels`
  that define a single function (hence, it's not guaranteed in general
  for `cl-labels`).

>   (info "(cl) Function Bindings")
>
> Also, the Emacs Lisp reference manual describes ‘named-let’
> as a special form (like ‘if’), while its docstring describes
> it as a macro.  Is this an error in the reference manual?

Sounds like it, yes.


=== Stefan