IRC channel logs

2025-11-28.log

back to list of logs

<dsmith>rlb, Did you see this: https://lists.gnu.org/archive/html/bug-guile/2025-10/msg00015.html
<rlb>dsmith: nope. Sounds like it's at least an existing issue? i.e. 3.0.10 has it, so worst case, current main wouldn't be a regression, were we to release it.
<rlb>(Offhand, also suspect this may be something for Andy, until/unless more of us are better with the relevant facilities...)
<rlb>In any case, thanks for the heads up.
<dsmith>rlb, Seem to be an issue with other (all?) 32bit os's. Reminded me of some of funkiness you are seeing.
<rlb>Right -- 3.0.10 is just broken on 32-bit (and there have been big-endian issues), but Andy fixed much (I'd thought all) of that in main, and debian has cherry-picked what was needed to get all the relevant debian architectures working.
<rlb>Given the bug report, I wonder if main might work, or if it's still broken too (I don't think main was tested?).
<daviid>sneek later tell tohoyn I fixed gi-array->scm and scm->gi-array to support array of GVariant*, pushed on the devel branch ...
<sneek>Okay.
<daviid>sneek later tell tohoyn hum, something's wrong, but not sure wjat yet ... looking to find exactly what ...
<sneek>Will do.
<daviid>sneek later tell tohoyn actually I think it's ok, could you please pull, run the make/install danse, try and let me know
<sneek>Got it.
<daviid>sneek later tell tohoyn tested using this modified version of your code - https://bpa.st/IVPQ - here is what I changed: no rnrs nor srfi-43 imports, adding "bus_unown_name" (line 21), removing the dbg and make-c-pointer-array defs, calling g-golf scm->gi-pointers instead
<sneek>Got it.
<cow_2001>is there such a thing as mmap for guile? reading and writing from and to a.. maybe an mmapped bytevector?
<cow_2001>could there be?
<dsmith>cow_2001, Guile uses mmap internally, but I don't think it's exposed to scheme
<cow_2001>oh hmm
<dsmith>Sections of .go files are mmap'ed in, IIRC
<canterel>hi
<canterel>I'm having a heck of a time understanding how to get data from a port (in this case an HTTP response). I can make the request in cURL and it works, but Guile only ever returns an #eof
<ray1729>canterel: are you usings Guile's (web client) ? Hard to help without more info, but a random thought: if the server is returning a redirect, the body will be empty, it's possible cURL is automatically following the redirect.
<canterel>ray1729: I am, here's a pastebin I linked earlier but I think I was not logged in https://pastebin.com/6xBQLNiN
<canterel>I did check for a 200 response
<tohoyn>sneek, botsnack
<sneek>tohoyn, you have 4 messages!
<sneek>tohoyn, daviid says: I fixed gi-array->scm and scm->gi-array to support array of GVariant*, pushed on the devel branch ...
<sneek>tohoyn, daviid says: hum, something's wrong, but not sure wjat yet ... looking to find exactly what ...
<sneek>tohoyn, daviid says: actually I think it's ok, could you please pull, run the make/install danse, try and let me know
<sneek>tohoyn, daviid says: tested using this modified version of your code - https://bpa.st/IVPQ - here is what I changed: no rnrs nor srfi-43 imports, adding "bus_unown_name" (line 21), removing the dbg and make-c-pointer-array defs, calling g-golf scm->gi-pointers instead
<sneek>:)
<tohoyn>daviid: variants seem to work now, but sending a response to the method call causes a segmentation fault. See https://bpa.st/EO6Q for the new code and https://bpa.st/EDSQ for the output. "Muistialueen ylitys" means segmentation fault. Note that despite the segmentation fault the reply is sent (I checked this with dbus-send).
<tohoyn>daviid: in order to reproduce the bug launch the server in some console window and use dbus-send to send a MyMethod1 call to the server.
<ray1729>canterel: http-get returns two values: the response and the body, or if #:streaming? is true, a port for reading the body. The port you're trying to read will already have been read.
<ray1729>Here's the code I wrote last year to accomplish a similar task: https://pastebin.com/NFaYPH9X
<ray1729>I'm using the (srfi srfi-71) module which extends let to allow assignment of multiple values in the return from http-get
<xxd>Is there an LSP that can do document format as well for Guile?
<xxd>Emacs
<xxd>Probably with Eglot?
<xxd>"[eglot] Unsupported or ignored LSP capability `:documentFormattin..." This is with guile-lsp-server package on Guix.
<Arsen>maybe you need to update eglot, support for documentFormattingProvider exists in 31
<Arsen>(idk when it was added)
<xxd>Which LSP do you use?
<Arsen>for guile? none, I was using geiser last I wrote guile, but I hear https://git.sr.ht/~abcdw/emacs-arei may be decent too
<xxd>I'll have a look, ty
<mwette>cow_2001: if want something to try for mmap, i have this: https://github.com/mwette/guile-wl-play/blob/main/mmap.c
<mwette>loaded via extras.scm in the same dir
<kestrelwx>#chickadee
<kestrelwx>Omg
<kestrelwx>The russian roulette whether ERC will swap focus into another buffer.
<kestrelwx>cow_2001: There's mmap being implemented in Guix: https://codeberg.org/guix/guix/pulls/4173/commits/fdc13e85a6d95de6d5fb1dee2c7374618510700f.
<kestrelwx>mmap support that is*
<cow_2001>O_O
<cow_2001>let's see, mwette and kestrelwx
<cow_2001>wow.
<mwette>the finalizer is broken in mine; there was some discussion on the guix issue regarding how to do it
<yelninei>Hi, is there a way to query all (public) variables in a module?
<civodul>yelninei: hi! you can use (module-map proc (resolve-interface '(whatever)))
<yelninei>civodul: Thanks, works great. Is there a reason why it is undocumented, it feels pretty useful?
<sneek>tohoyn: Greetings :D
<kestrelwx>yelninei: It's in (guile) Module System Reflection
<kestrelwx>Oh, yeah, 'module-map' doesn't seem to be.
<civodul>it’s among the oldest parts of Guile, people wouldn’t write doc back then ;-)
<rlb>mwette: https://codeberg.org/guile/guile/issues/64 --- in case you hadn't seen it.
<rlb>(and wanted to)
<daviid>sneek later tell tohoyn please (a) update your coe so it uses the main I provided, so I canlauch the script with -d option; (b) provide a clear example as to where (in the console or ...) and how, exactly, to call dbus-send ... as said, I have zero dbus knowledge ... tx
<sneek>Okay.
<daviid>sneek later tell tohoyn it would be better if you used dimfi (a peek 'clone', that uses less space, less newline, less semi-colomn ...) to debug, and 'normal' display inside your methods ...
<sneek>Will do.
<daviid>sneek later tell tohoyn fwiw, dimfi always newline and returns its last arg ...
<sneek>Okay.
<daviid>sneek later tell tohoyn see how dimfi is used in some g-golf core functionality ... as examples as to how to also properly indent debug info ...
<sneek>Okay.
<rlb>(having guile-*.tar.gz in .gitignore surprises me a bit...)
<rlb>(I tend to think of it for things that will also be cleaned.)
<rlb>...and I suppose if we *are* going to have it, it should include more extensions now since distcheck builds .gz .lz and .xz now.
<rlb>s/ now\././ :)