IRC channel logs

<lechner>old: OK, thanks!

<mwette>old: do you know for sure if guile CPS optimizer translates a call-with-values form to a multi-valued continuation ?

<old>mwette: idk off hand. Why asking ?

<mwette>looking at srfi-189

<lechner>Hi, why would in Emacs (get 'with-exception-handler 'scheme-indent-function) be nil, please?

<old>mwette: AFAICT, it is a known primitive in tree-il and gets translate into a primcall

<old>ah it's suppose to be converted into a let-values in peval

<mwette>old: thanks

<mwette>the sample impl of srfi-189 seems to be turning multiple values into lists

<mwette>and using apply for mv calls

<old>which function?

<mwette>maybe-ref, for example

<mwette>note my use of 'seems': not really sure yet

<old>just looking at it, but using pair? is slower than null?

<old>well only true if the object is actually a heap object

<old>nvm what I said, I gues it's fine

<mwette>Is `(values 1)' equivalent to `1' ?

<mwette>... as returned by a procedure?

<mwette>hmm ... `(+ (values 1) (values 2))' => `3'

<old>mwette: I think so yes

<old>at optimization level 1, this is optimized to 3

<old>through peval

<old>655 ;; Statically-known number of values.

<old>656 (($ <primcall> src 'values vals)

<old>657 (and (<= nmin (length vals)) (or (not nmax) (>= nmax (length vals)))

<old>658 (make-call src (make-lambda #f '() consumer) vals)))

<mwette>thank you

<old>that peval stuff is such a best

<old>I don't even know how to read all of that

<mwette>What? The first comment in peval is "This is a simple partial evaluator." :)

<mwette>JohnCowan: The srfi-189 sample implementation exports `values'. Is that intended? (I'm not up on r6rs lib's.)

<mwette>old: my attempt to start looking into srfi-189 for values: https://paste.debian.net/hidden/c723f779

2026-09-25.log