IRC channel logs

<gnucode>ieure: I'll try that.

<gnucode>M-x toggle-debug-on-error gave me an error: string-prefix-p: Wrong type argument: stringp, (toggle-debug-on-error)

<gnucode>even M-x snake is not working. maybe my spacemacs is just borked.

<mange>Can you M-: (setq debug-on-error t) RET instead of toggle-debug-on-error?

<gnucode>I can try that sure.

<gnucode> https://paste.debian.net/hidden/22eaacf8 -> that's the bug that pops up when I try to run M-x geiser

<gnucode>I set (setq geiser-guile-binary "guile3.0") 'cause on OpenBSD guile3.0 is the binary name.

<mange>Hmm. That list ((geiser . 6) ...) looks wrong. What does (counsel--M-x-externs) return for you? (You can use M-: again to find out. If it's a list, the first value or two will be enough.)

<gnucode>mange: (counsel--M-x-externs) -> https://paste.debian.net/hidden/fd60a402

<mange>Hmm, okay, we must have different versions of counsel. I was looking at the source of counsel-M-x to find out whether the second argument of ivy-read was right, because it seems iffy to me.

<gnucode>I just updated spacemacs and all the packages an hour or two ago. So maybe it is a spacemacs issue.

<dsmith>gnucode, I've had issues with emacs where the .elc file are out of sync with the .el, or maybe it was an package upgrade that needed a newer dependency

<dsmith>Or something like that. ISRT removing some packages and getting them fresh.

<dsmith>gnucode, "Wrong type argument: stringp" was familar

<dsmith>Wasn't with geiser for me. magit or eglot I think.

<gnucode>I did update packages recently. I can try to remove one and see what happens...

<dsmith>gnucode, Sorry I can't be more specfic. Had annoying messages, and then... I didn't

<gnucode>no worries. I just created a github issue for spacemacs.

<gnucode>We'll see what happens. :)

<gnucode>as a fun aside, I spent the last few days trying to make a "periodic table of numbers". It's super cool defining a procedure like (prime? 7) via recursion.

<gnucode>It's surprising how short one can write such a function.

<gnucode>and it's perhaps worth noting that I wrote that silly trivial program without the traditional guile backtraces.

<gnucode>good old printf debuging. In this case (display "error\n") debugging.

<gnucode>scheme is super awesome!

<dsmith>Indeed

<jab>I wish guile had the equivalent of "typed racket".

<jab>is there an srfi for adding type declarations to functions ?

<jab>hmm srfi-253

<jab>apparently guile supports it too...

<lilyp>jab: you would have to install the srfi-253.scm script separately tho, it's not shipped with guix

<lilyp>s/guix/guile/

<jedb>how do you correctly set %auto-compilation-options? the manual says it should be defined in ~/.guile but is less than clear on what exactly it recognises as valid, and I can't seem to get the auto-compile optimisation level for a script to change no matter what I define it as

<mwette>jedb: I put ((@ (system base compile) default-optimization-level) 1) in .guile

<mwette>you can also set tree-il and cps optimizations separately

<mwette>level 0 is tree-il bytecode compiler only

<dsmith>Not sure of this is the issue, but guile only uses ~/.guile when "interactive" (or something like that).

<lloda>afaik guile always loads .guile unless you pass -q

<dsmith>Ah, depends on if you are running a "script" or not: https://www.gnu.org/software/guile/manual/html_node/Command_002dline-Options.html#index-init-file_002c-not-loading

<dsmith>For -q "Do not load the initialization file, .guile. This option only has an effect when running interactively; running scripts does not load the .guile file."

<lloda>i stand corrected!

<old>I also have this in .guile: (repl-default-option-set! 'optimization-level 0)

<jedb>I guess my next question is then: how do you change the auto-compile optimisation level for a script if .guile is the wrong place to put things?

IRC channel logs

2026-02-13.log