IRC channel logs

<dsmith>rlb, Is that that the gnu date args issue? There is a bug for NetBSD on that. Just saw it yesterday I think..

<dsmith>rlb, https://debbugs.gnu.org/cgi/bugreport.cgi?bug=26121

<rlb>No, I meant the compilation/load issue.

<dsmith>rlb, No idea, sorry

<rlb>OK, I suppose I'll see about creating one later.

<daviid>any fluidsynth user by any chance? i now have a low level api binding to a few of the fluidsynth api, enough to play a sound (to mke this msgshort) - but i fail to change the instrument, I am wondering if anyone would know (afaict, there is unfortunately no irc nor matrix channel/room to discuss fluidsynth api quiz)

<euouae>Hello

<sneek>euouae, you have 1 message!

<sneek>euouae, mwette says: I believe ,si is the right meta command. You can also (use-modules (system vm vm)) (vm-add-next-hook! your-hook) if I read right

<euouae>I'm reading some bytecode instructions but I cannot comprehend them

<euouae>In particular things like (static-ref 2 16356). This supposedly loads ip[16356] into sp[2]. There's a few of these instructions with these random ip offsets. What are they referencing??

<euouae>I'm looking at this disassembly: <https://termbin.com/uhl5>. There's a single jump, and the procedure ends with tail-call. If so, when does this ever finish? I don't understand

<euouae>ah in this case it just reuses the stack frame, so it probably configures some other procedure to be called

<euouae>Hey all, I've been messing around with the brainfuck implementation to get an idea

<euouae>I can do something like: (use-modules (language brainfuck parse)) (define bf-program ",+.") (define bf-IL (call-with-input-string bf-program read-brainfuck)) (define tree-IL (compile bf-IL #:from 'brainfuck #:to 'tree-il))

<euouae>That works fine. But (compile bf-IL #:from 'brainfuck #:to 'scheme) does not work. Why not? I get "no way to compile #<<language> ...>"

<euouae>In the language spec, there is (scheme . compile-scheme). The compile-scheme procedure works: (compile-scheme bf-IL '() '()) is okay.

<euouae>ah this patch fixes both my issues <https://lists.gnu.org/archive/html/bug-guile/2025-06/msg00005.html>

<euouae>ArneBab: This is a good/easy patch <https://lists.gnu.org/archive/html/bug-guile/2025-06/msg00005.html>, consider applying it!

<dsmith>rlb, Oh! The <= or < thing? Yeah, no bug# for that as far as I know.

<euouae>Hello

<ArneBab>euouae: can you turn it into a pull request?

<euouae>ArneBab: sure

<ArneBab>euouae: thank you!

<euouae>ArneBab: Here it is <https://codeberg.org/guile/guile/pulls/30>; one patch [2/2] was from Bolle himself, the other patch [1/2] I had to edit myself to make it apply.

<euouae>plus I changed it a bit

<euouae>I haven't tested it yet, but I'll test it soon

<ArneBab>euouae: it looks good to me, and without risk to other parts. Please tell me when you tested it.

<euouae>yup I will; I'll add a unit test too

<euouae>ACTION runs that make -j6 and waits

<euouae>ArneBab: okay, it seems to work. but let me work on this patch a bit more.

<euouae>I want to fix the docs (include examples of how to compile brainfuck) and I also want to add one more compiler (the 'value compiler that "runs" brainfuck programs)

<euouae>I also want to include unit tests

<euouae>ACTION is confused as to show test-suite/tests/brainfuck.test passes if at all

<euouae>Wait, lol. Is anyone ever running make check on the Guile repo?

<euouae>Is brainfuck just ignored?

<euouae>ah wait, my patch altered something. ignroe the baove

<euouae>above *

IRC channel logs

2025-10-13.log