IRC channel logs


back to list of logs

***Server sets mode: +Ccntz
<the_tubular>How's 2022 civodul, lots of commits there ?
<marusich>hello civodul
<marusich>hello unmatched-paren
<marusich>hello the_tubular and everyone
<karrq_>network split ?
<sneek>Welcome back karrq_, you have 2 messages!
<sneek>karrq_, podiki[m] says: are you using ./pre-inst-env within a guix shell? (all from the manual, but just checking)
<sneek>karrq_, podiki[m] says: for reference, I update packages/build local guix on a foreign distro sometimes, without any extra steps compared to a guix system
***f1reflyylmao is now known as f1refly
<civodul>the_tubular: not a single commit in sight :-)
<civodul>hey marusich!
*the_tubular sad
***sam__ is now known as sam_
<florhizome[m]>\o/ happy new year, guix \o/
<florhizome[m]>may your updates never freeze again (;
<GNUHacker>Happy GNU Year!
<jab>Hey guix!
<nckx>🎉 , Guix! 🎊
<RuZzz>Do I need to add mapped-devices to config.scm? If I don't have a RAID and I don't know - will I use encryption and LVM?
<nckx>For sure there is little point to having an empty mapped-devices field.
<nckx>(Which is the default anyway.)
<nckx>My return key no longer works. Great, 2022 sucks. Good night Guix.
<lilyp>nckx: Use C-j :)
<RuZzz>Will I be able to add LVM and encryption support in the future without a new installation and move files again?
***luke-jr- is now known as luke-jr
***califax- is now known as califax
<florhizome[m]><jpoiret> "one thing to remember, a service..." <- component configurations I guess?
<florhizome[m]>it’s something guix as a guile library does for you
<gnoo>there's a typo in info node of guix on info "(guix-cookbook) A Scheme Crash Course". there is an extra ! at the output of Hello Guix
<rekado>gnoo: I don’t see it. What info reader are you using and what version of the manual is it?
<gnoo>i'm using emacs, wait lemme check the 'info' command
<gnoo>info can't find the node at all
<gnoo>oh it can
<gnoo>i'm talking about these two lines:
<gnoo> (display (string-append "Hello " "Guix" "\n"))
<gnoo> ⊣ Hello Guix!
<rekado>ah, I see
<rekado>would you like to send a little patch?
<gnoo>sure, but i haven't completed the copyright assignment yet, is it needed?
<gnoo>how do i know which variable has a specific package? i'm trying to remove dbus. is it in %base-packages ? and what about a arbitary package?
<clemens3>installed guix via git and according to handbook after starting daemon did: guix install glibc-locales
<clemens3>error paste here:
<clemens3>any idea?
<gnoo>did you do guix pull before that?
<gnoo>do that and then guix system reconfigure /etc/config.scm
<clemens3>ok, will be back..
<gnoo>wait, you're installing guix system or package manager?
<clemens3>i went into the docu starting with install guix from git..
<clemens3>maybe that was before..
<clemens3>guix build from git
<gnoo>it's the guix distro or a foreign distro?
<clemens3>foreign distro
<gnoo>oh, then wait for someone else to respond
<rekado>gnoo: we don’t do copyright assignment in Guix
<rekado>clemens3: are you building without binary substitutes on purpose?
<rekado>because that’s what you seem to be doing. You’re building very basic early bootstrap packages.
<gnoo>rekado: can you please explain this line: (list #$(file-append inetutils "/libexec/syslogd")) ? thanks
<gnoo>specially file-append and inetutils and "/libexec/syslogd" ones. what is the binary's name to be ran and what is file-append and why "/libexec/syslogd" ?
<rekado>sorry, I’m headed out
<clemens3>rekado: yes, on purpose
<clemens3>i build guix first, so why stop afterwards
<clemens3>i came looking into guix because building from source seemed to be first class citizen
<clemens3>and I am particularly interested in bootstrapping
<clemens3>plenti of distros with decent binary packages
<AwesomeAdam54321>gnoo: file-append concatenates whatever inetutils expands to into the "/libexec/syslogd" file
<gnoo>thanks, i've made a seatd service!
<gnoo>(i don't know what any of that means, but it works and that's all i want)
<florhizome[m]><gnoo> "how do i know which variable has..." <- you will probably still have dbus around somehow, as it’s a dependency of many package. I think you could do package transformations in order to remove it, where possible. But I think you want to remove dbus service, that initialized the dbus?
<unmatched-paren>i can't figure out how to run an application using my nvidia card _with nouveau_. all the instructions i've found are for the proprietary drivers, except for the advice to set DRI_PRIME=1, which doesn't work
<unmatched-paren>there are a couple of things about xrandr, but that doesn't apply for wayland
<gnoo>florhizome[m]: no, dbus service is not part of %base-services so i don't have it. but doing 'ps aux | grep dbus' gives dbus-launch so i want to remove dbus completely
<gnoo>can i make a virtual package that provides and replaces an existing package(dbus) ?
<gnoo>that way i can replace it with something that does nothing
<AwesomeAdam54321>gnoo: You can provide your own package definition for dbus that replaces the existing one
<AwesomeAdam54321>gnoo: You're welcome
<rekado>clemens3: arguably, with reproducible builds it doesn’t really matter whether you build and bootstrap from source or download a reproducible build result.
<rekado>clemens3: but the build error you stumbled into looks odd and is worth reporting, in my opinion.
<unmatched-paren>??? 'guix install: warning: at least 18,446,744,072,889.5 MB needed but only 270,230.2 MB available in /gnu/store'... when installing 0ad
<AwesomeAdam54321>unmatched-paren: I think that means there isn't enough disk space?
<rekado>unmatched-paren: that’s an integer overflow
<rekado>there’s a bug report for that IIRC
<unmatched-paren>oh, thanks :)
<unmatched-paren>hm, 0ad doesn't build anyway; it's not got a substitute. there's failures on the ci pointing to 0ad-data, but the log there doesn't have any information... just a lot of 'substituter-succeeded':
<unmatched-paren>ah well
<florhizome[m]><AwesomeAdam54321> "gnoo: You can provide your own..." <- sure that wouldn’t crash builds linking against dbus?
<florhizome[m]><gnoo> "florhizome: no, dbus service..." <- i mean the question is what does it/why is it there, right? what does busctl show?
<florhizome[m]>if a program launches with dbus–launch you might be able to substitute that out.
<florhizome[m]>fetching a patchset with wget from issues/–set/N fails for me, is that the right url?
<simendsjo>I'm having problems with a package which breaks my foreign distro. Tried both Manjaro KDE and Kubuntu, and it breaks them both. `sbcl` is the offending package. When installed, plasmashell doesn't launch correctly. Pretty sure this is what broke my Manjaro Gnome install too. See some details here:
<AwesomeAdam54321>florhizome[m]: It would probably crash builds that depend on dbus, unless the builds can be configured to build without it
<AwesomeAdam54321>s/build without/not use
<jpoiret>simendsjo: isn't this similar to ?
<jpoiret>you could try the `export XDG_DATA_DIRS=/usr/local/share/:/usr/share/` workaround?
<gnoo>florhizome[m]: which package provides busctl?
<jpoiret>florhizome[m]: it's
<jpoiret>gnoo: i don't think you should try to replace the dbus package by nothing in the dependency trees, as applications that depend on it often do need them
<jpoiret>ie networkmanager in %desktop-services depends on it
<gnoo>i don't use networkmanager, instead use dhcpd
<jpoiret>you'd have to also look at each package in your whole dependency graph that depends on it, and see if it can be compiled and used without dbus
<gnoo>also not %desktop-services, only %base-services
<gnoo>hmm, i remember seeing something about graphs in guix manual
<phodina[m]>Is there a guide or package already on how to build a package using meson build system which in fact calls cargo in its guts?
<phodina[m]>I'm writing package definition for sqeekboard and don't know how to build it
<jpoiret>phodina[m]: i'd say it's a complicated topic, and i don't think you'd find a guide for that
<jpoiret>you could take a cue from packages with rust in them, off the top of my head i was thinking of icecat but that could be too big of an example
<phodina[m]>I know really complex and I already hit a wall when attempting to port sbctl to add secureboot manager. It was gnu make and go build system.
<phodina[m]>Unfotunately, I tried using a Docker as an example but that didn't work out. The other reason is that I'm not that familiar with go build system really.
<phodina[m]>So perhaps my question should be more general. Is there a plan to support nested build systems?
<phodina[m]>I guess for now it's better to use cargo build system as it patches the crates and in the phases add what is necessary to make meson work, right?
<phodina[m]>* I know it's really complex
<gnoo>how can i have guix graph -t reverse-bag account for only the packages in the current profile(or enviroment) ?
<gnoo>(or reverse-package)
<florhizome[m]><gnoo> "florhizome: which package..." <- pretty sure dbus itself
<gnoo>florhizome[m]: guix environment --ad-hoc dbus -- busctl gives an error
<lilyp>phodina[m]: I once attempted to do this for fractal
<lilyp>I've used cargo-build-system as base and replaced configure etc. with the stuff from meson-build-system
<civodul>Hello Guix! :-)
<florhizome[m]><jpoiret> "florhizome: it's issues.guix.gnu..." <- when the second i* is an issue wget finds sth but downloads 0kB ...
<phodina[m]><lilyp> "phodina: I once attempted to..." <- I've seached for fractal but had no results. Is it in the main repo or is there a patch waiting to be merged?
<lilyp>There's a pretty old patch in the ML and the patch itself is not really useful
<lilyp>The only relevant thing is the main skeleton (using cargo-build-system for its #:cargo-inputs keywords)
<phodina[m]>lilyp: I do the same for the #:cargo-inputs keyword. But I guess every package will be different in the rest to some degree.
<florhizome[m]><florhizome[m]> "when the second i* is an issue..." <- * and when wget fetches a Textfiles git am doesn’t recognize the patch (patch is empty)
<Haider>do any of you all have a problem with qutebrowser?
<Haider>because WEBgl is blacklisted for me.
<lilyp>what's your problem?
<lilyp>(this was meant to phodina)
<lilyp>florhizome[m]: the first "patch" is probably the cover letter, which you can skip
<florhizome[m]>yes 0 is completely empty
<florhizome[m]>I’m not talking about that one anymore though
<florhizome[m]>N looks just like a mail with diffs in it
<lilyp>not all issues are git am friendly
<lilyp>perhaps you found a bad one?
<florhizome[m]>how would I proceed?
<phodina[m]><lilyp> "what's your problem?" <- Well the end goal is Phosh running on ARM device such as Pinephone. 😁
<phodina[m]>Currently I face some issues with Phosh and when that's sorted out a virtual keyboard will definitely be handy. So I started looking also into squeekboard and that's why I asked here about composite build systems.
<phodina[m]>* Well the end goal is Phosh running on ARM device such as Pinephone. 😁
<phodina[m]>Currently I face some issues with Phosh and when that's sorted out a virtual keyboard will definitely be handy.
<phodina[m]>So I started looking also into squeekboard and that's why I asked here about composite build systems.
<florhizome[m]>phodina: do you have guix running else then that? ((:
<lilyp>Thanks for the context, but the question was rather in which way your build fails ;)
<phodina[m]>lilyp: Thought the context might be handy. Not atm as I'm on phone.
<phodina[m]>I'll paste the build log once I'm at my laptop.
<lilyp>Oh, I know it's going to be a GNOME Rust experiment.
<phodina[m]>florhizome[m]: I do have Guix System running on RPi and Pinebook Pro
<lilyp>GNOME folk are the only ones insane enough to try making Rust fit their agenda.
<phodina[m]>* I do have Guix System running on RPi and Pinebook Pro + Guix on PostmarketOS on Pinephone
<florhizome[m]>phodina have you tried running phoc of the vt like the person answered?
<phodina[m]>lilyp: Yeah, let's make it complicated. Though there are benefits using Rust and it's "easier" to integrate to Gtk compared to Qt.
<phodina[m]>florhizome[m]: Yes, phoc as a compositor is running. I can launch other wayland programs. However, there are some issues still with phosh
<florhizome[m]>I guess you switched the meson option for systemd off?
<phodina[m]>For some reason there the gsettings is not found
<phodina[m]>/gnu/store/pm0pwch3dn7k3hgzl53vnfk888x061ns-gnome-session-40.1/bin/.gnome-session-real: line 16: gsettings: command not found
<phodina[m]>Checked the package definition and the glib:bin is part of the wrap-program
<phodina[m]>florhizome[m]: There's patch for linking the systemd but now I get in the runtime error regarding the dbus
<florhizome[m]>No I mean as configure flags
<lilyp>phodina[m]: we should probably patch gsettings to an absolute command here
<lilyp>using glib:bin as regular input
<phodina[m]>florhizome[m]: Thanks, I do only substitutes in the source code. I'll check again the meson build options.
<phodina[m]>btw patches are ncluded in the issue ticket.
<phodina[m]>lilyp: Ok, I'll add them to the list
<civodul>--tune's in the house!
<lilyp>Not just that, you'll also have to write the substitute* code and check its correctness
<lilyp>wooo, SIMD instructions!
<phodina[m]><lilyp> "Not just that, you'll also..." <- I checked the patch and glib:bin is already part of inputs. So I need to substitute the command in the file
<phodina[m]>* I checked the patch and glib:bin is already part of inputs. So I need to substitute the command in the file
<phodina[m]>Though that has to be in gnome-session package, right?
<nckx>sneek: later tell unmatched-paren: Let's see what it does after a restart:
<florhizome[m]>everybody says rust is soo faaast but it’s downloads are very slooooow :/
<florhizome[m]><lilyp> "not all issues are git am..." <- still happy about tipps for dealing with those git am unfriendly patches (;
<clemens3>rekado: ok, thanks, will look for some mailing list to report
<avp_>Hello Guixers. What are possible reasons why Guix may keep telling that the public key is not authorized on 'guix offload test' even when I added my public key from the machine to a server and specified the server host key?
<nckx>florhizome[m]: patch -Np1 -i FILE && git commit with a copy/pasted commit message is a particularly icky one. :)
<florhizome[m]>just patch? not git?
<nckx>The second command is a git command.
<nckx>‘git commit -a’ to be exact.
<nckx>My return key still won't work. I think something crawled under it & died.
<avp_>Oh wait, I should have read the official documentation more carefully; the docs clearly say that I must authorize both machines before offloading.
<florhizome[m]>nckx well that’s handy
<florhizome[m]>it actually does something :D
<florhizome[m]>Can i just not worry about the rejects until I applied the whole set?
<taterbase>Hello, I'm trying to understand how I can create udev rules. I found this doc but it's not clear where I might add some udev rule info. Is there a better place to go to understand how this works under guix?
<pinoaffe>taterbase: here, you can see how I add udev rules from packages (brightnessctl and android-udev-rules) as well as "custom" udev rules (the whole "udev-rule" s-exp) to my guix system config
<pinoaffe>I have no clue whether this is a clean or idiomatic way of doing this, nor do I know where I found the info, but it works as far as I can tell
<taterbase>pinoaffe: this is very helpful thank you
<nckx>florhizome[m]: Rejects? Are you sure there was anything ‘wrong’ with the patch at all? From here it sounds like you're simply applying it to a conflicting version of Guix.\
<nckx>No tool will do that without throwing a tantrum.
<nckx>If conflicts are the problem, git is vastly superior, and you should not be using patch at all. Patch is only useful in the rare case where there's something about the patch itself (e.g., it got mangled in transit in some minor way — haven't figured it out myself) that Git doesn't like but the contents still apply cleanly.
<Karthik[m]>vagrantc: i need help with starting guix-daemon @ debian sid
<vagrantc>Karthik[m]: i've only used the daemon that ships with the package
<vagrantc>Karthik[m]: did you manually configure the systemd unit for guix-daemon? or are you saying the package is broken?
<vagrantc>Karthik[m]: at any rate, i've gotta head out very shortly ... good luck!
<Karthik[m]>vagrantc: I just install guix via apt get, no changes done
<vagrantc>Karthik[m]: please file a bug in the debian bug tracker
<Karthik[m]>vagrantc: Sure, Thank you!
<vagrantc>Karthik[m]: at a quick glance, the .service file in the package references /usr/bin/guix-daemon ... so somehow you must have something configured locally...
<vagrantc>Karthik[m]: Loaded: loaded (/etc/systemd/system/guix-daemon.service
<vagrantc>Karthik[m]: that's something locally configured, not what the package ships
<Karthik[m]>vagrantc: Now i remember, I once install guix manually a couple of months ago.
<vagrantc>remove that and convince systemd you've removed it and hopefully that'll work, then :)
<Karthik[m]>yeah, sure, thanks for the tip !
<vagrantc>first problem solved for the year for me! :)
*vagrantc waves
<porcupirate>0ad fails, something to do with untested minor version of SpiderMonkey. Relevant log section
<porcupirate>guix commit: 77fc94ef1fbd2f43706a4a0374cc07265629bfa3
<podiki[m]>porcupirate: this issue perhaps?
<porcupirate>Looks like it. Strange... I looked up phrases in the error in the archive and didn't find that.
<porcupirate>"0ad untested minor version is:open"
<porcupirate>At least I don't need to send my bug report.
<the_tubular>Ohh, podman has been merged!
<podiki[m]>cool. never used it but I see in the commit a note of what a user needs to do for it to work, shouldn't that be in the description so users will see it?
<lilyp>florhizome[m]: there's no tooling for those sadly, hence why people don't like them
<lilyp>my personal method is to export the patches from my mail reader
<the_tubular>Could someone help me with my variable list problem ?
<the_tubular>I'm trying to make a list of package, importing that list in another file and install those packages
<the_tubular>But it says the module isn't found
<porcupirate>the_tubular: The module isn't part of Guix?
<porcupirate>Check GUILE_LOAD_PATH
<Kolev>How do I modify keyboard layouts? I want to add → ↓ ← ↑ symbols to the UK Dvorak keyboard layout.
<the_tubular>No, I'm trying to make that module
<the_tubular>porcupirate, so (add-to-load-path (dirname (current-filename))) this would have to be part of my config.scm ?
<porcupirate>the_tubular: Is everything you need defined either in guix or in your config.scm?
<porcupirate>What module does it say doesn't exist?
<the_tubular>A file I've called my-modules, containing a list named my-packages
<the_tubular>this file is in the same folder as my config.scm
<porcupirate>the_tubular: Does it work with that code? `(add-to-load-path (dirname (current-filename)))`?
<the_tubular>I just slap that anywhere in mm config.scm ?
<porcupirate>Probably before (use-modules (my-packages))
<the_tubular>Let me give that a shot
<SeerLite[m]>Kolev: Try KMonad
<Kolev>SeerLite[m]: What is that?
<the_tubular>porcupirate : config.scm:4:0: error: module (gnu services my-variables) not found
<porcupirate>the_tubular: Is my-variables.scm in `(dirname (current-filename))`/gnu/services?
<the_tubular>Umm, I've added both the ` in your code
<the_tubular>And now I get config.scm:4:1: error: use-service-modules: unbound variable
<porcupirate>use-service-modules is a shortcut for use-modules
<porcupirate>expecting things in (gnu services x)
<SeerLite[m]>Kolev: It's a program that lets you customize the keyboard mapping and use set up layers (kinda like QMK). It takes over your keyboard and creates a fake one for the OS with whatever remapping you set up.
<Kolev>SeerLite[m]: That sounds incompat. with Guix.
<the_tubular>What is it expecting exactly ?
<porcupirate>the_tubular: (use-service-modules x y z) => (use-modules (gnu services x) (gnu services y) (gnu services z))
<SeerLite[m]>Kolev: No it's not, why do you think so? I'm actually using it right now. It is a package in Guix
<the_tubular>But which services exactly ?
<Kolev>SeerLite[m]: I declare my keyboard layout in config.scm.
<porcupirate>the_tubular: What is your call to (use-service-modules)?
<the_tubular>Sorry I'm still very bad with guile
<porcupirate>Don't worry
<the_tubular>Let me check
<the_tubular>(use-service-modules linux mcron virtualization)
<porcupirate>Those look valid.
<SeerLite[m]>Kolev: KMonad is an user program, it runs on top of the layout you declare in config.scm. Think about it like sxhkd but specifically for remapping and not restricted to the display server
<porcupirate>the_tubular: How are you importing my-variables?
<the_tubular>Yes, my bad. here it is after my git commit (use-service-modules linux mcron my-variables virtualization)
<the_tubular>I guess it's still my-variables that is causing issue right ?
<porcupirate>Don't use (use-services) to import your custom code unless it's in gnu/services
<the_tubular>How should I import my custom code ?
<porcupirate>(use-modules (path to custom-code)) where path/to/custom-code.scm is in the load path.
<porcupirate>Where is my-variables relative to your config?
<the_tubular>This is my-variables
<the_tubular>Both in a directory in my home folder
<porcupirate>(use-modules (my-variables))
<Kolev>SeerLite[m]: Does it work in Emacs too?
<Kolev>SeerLite[m]: I read “an user program“ as “an oozer program“ because of “an“ and not “a.“
<the_tubular>Same problem porcupirate
<porcupirate>the_tubular: pastebin config.scm?
<SeerLite[m]>Kolev: Haha sorry I missed that, English is not my first language. Yeah it should work just fine anywhere, it emulates a real keyboard with the configured mappings
<florhizome[m]><lilyp> "florhizome: there's no tooling..." <- A bizarre situation, when you can not extract code from the ML by at least the same means it was added to it...
<florhizome[m]>I installed emacs-debbugs but I haven’t found how to export patches from it. Right now I still prefer doing git things from the cli, and maybe switch for a merge conflict.
<florhizome[m]>nckx pointed me to the “patch” command, it was able to apply (most of) the patches from the downloaded file so I’m actually good now.
<florhizome[m]>and I learned another thing^^
<porcupirate>the_tubular: What is the error?
<the_tubular> error: use-service-modules: unbound variable hint: Did you forget a `use-modules' form?
<lilyp>florhizome[m]: git send-email/git am ought to work in pairs.
<lilyp>If they don't, you found a major bug :)
<porcupirate>the_tubular: I think (use-modules (gnu)) should fix it.
<porcupirate>But you already have that
*the_tubular is confused.
<the_tubular>Is it cause of the line I added above ?
<SeerLite[m]>the_tubular: The first line has backquotes and Guile may be interpreting it as quasi-quoting. Maybe that's it?
<porcupirate>the_tubular: line 1 still has the quote.
<porcupirate>Get rid of quotes at beginning and end of line 1
<the_tubular>ice-9/boot-9.scm:3329:6: In procedure resolve-interface: no code for module (my-variables)
<porcupirate>my-variables.scm is in same directory as your config? Should work...
<the_tubular>Fixed it, forgot an "S"
<the_tubular>Now I got another one ... 7:18: error: linux-libre-5.15: unbound variable
<the_tubular>hint: Did you forget `(use-modules (gnu packages linux))'?
<porcupirate>(use-package-modules linux)
<the_tubular>What is the difference between use-modules and use-package ?
<porcupirate>(use-package-modules x y z) => (use-modules (gnu packages x) (gnu packages y) (gnu packages z))
<porcupirate>You already have (use-package-modules certs)
<the_tubular>So I don't need (use-package-modules certs) and (use-modules) ?
<the_tubular>Is it the same with use-services and use-services-modules ?
<porcupirate>I don't think there's a use-services or use-packages, is there?
<the_tubular>I don't know, I'm asking cause I have this : (use-service-modules linux mcron virtualization)
<the_tubular>And I'd like to use either one or the other, so I don't get confused in the future
<porcupirate>use-service-modules and use-package-modules are shorthand for different things.
<the_tubular>But they work the same ?
<the_tubular>Got it, let me try now
<porcupirate>It gets repetitive typing (gnu packages ...) and (gnu services ...)
<porcupirate>Looks like my-packages is a list of strings. The operating system packages should be a list of packages.
<porcupirate>Try (map specification->package '("bash" "bash-completion" ...))
<the_tubular>no code for module (certs)
<the_tubular>I try to put it in the same line as : (use-modules (certs) (gnu) (guix) (guix-package) (linux) (srfi srfi-1))(use-package-modules certs)
<porcupirate>(use-modules (gnu packages certs))
<porcupirate>(use-modules (gnu packages linux))
<the_tubular>Why isn't this not working : (use-modules (certs) (gnu) (guix) (guix-package) (linux) (srfi srfi-1)) ?
<the_tubular>Ohh I need the gnu package before ?
<porcupirate>certs and linux are supposed to be package modules, so they are in (gnu packages)
<porcupirate>That's why I just use-package-modules when I need a lot of package modules.
<the_tubular>no code for module (guix-package)
<simendsjo>jpoiret: I'm not sure if it's the same issue. I have fontconfig installed without any problems. Someone (manjaro..?) put a /etc/profile.d/ that sources the profile, but I tried modifying the file to reference /usr/local/share:/usr/share, but that didn't help. Ref
<porcupirate>Replace - with a space
<the_tubular>How would you write my modules and package for it to be clean and minimal porcupirate ?
<the_tubular>This is getting weird real fast
<mekeor[m]>the_tubular: can you paste your imports on, please?
<the_tubular>I did above mekeor[m]
<mekeor[m]>i don't think there is a module called (guix packages), is there?
<porcupirate>Maybe. Try importing it in the guix repl
<porcupirate>The way guix's code is organized, there's /gnu for the stuff in the system and /guix for stuff used by the package manager.
<porcupirate>So /guix/packages.scm defines useful methods for dealing with packages
<porcupirate>"/gnu/packages.scm" defines useful methods for searching defined packages.
<porcupirate>So yes, there is a (guix packages) module.
<the_tubular>error: %default-modules: unbound variable hint: Did you forget `(use-modules (gnu services shepherd))'
<porcupirate>Either add (gnu services shepherd) to your use-modules list or add shepherd to your use-service-modules list
<the_tubular>This is so confusing lol
<the_tubular>I'm so sorry
<porcupirate>Don't be sorry for trying new things
<the_tubular>error: %my-packages: unbound variable
<the_tubular>Do I need the "%" sign ?
<the_tubular>I'm stuck with vi editor and I'm hating it lol
<porcupirate>That isn't meant to be an emoji lol
<porcupirate>% usually means it's a public variable.
<the_tubular>There's no emoji from my side, but I don't understand it
<the_tubular>How do i import the list in my-variables called my-package to my config.scm ?
<the_tubular>my-packages *
<porcupirate>It's a sed command you can run in vi...
<porcupirate>You've imported it. Use my-packages instead of %my-packages
<the_tubular>Ohh so just remobe the "%" correct ?
<the_tubular>Got it
<the_tubular>error: openssh-service-type: unbound variable
<porcupirate>guix system search openssh
<porcupirate>the location field says it's in gnu/services/ssh.scm
<the_tubular>Yep, I see this
<the_tubular>Is it just something not imported ?
<porcupirate>So add (gnu services ssh) to use-modules or add ssh to use-service-modules
<the_tubular>Or it's another error
<the_tubular>Got it
<porcupirate>How long have you done gnu/linux?
<the_tubular>Too long lol
<the_tubular>Like 10 years ish
<porcupirate>But not much guile?
<the_tubular>I'm pretty good with ctl, but guile is kicking my ass
<the_tubular>cli *
<the_tubular>guix system: error: service 'networking' requires 'wpa-supplicant', which is not provided by any service
<the_tubular>Got this one now
<nckx>sneek: later tell unmatched-paren: Success has been forcibly achieved.
<the_tubular>This just means that wpa-supplicant would need to be in my "my-packages" list correct ?
<nckx>sneek: Botsnacks.
<nckx>sneek: Botsnack?
<nckx>Diet sneek.
<porcupirate>the_tubular: you need a wpa-supplicant service
<the_tubular>Can i get a human snack ?
<nckx>the_tubular: Human snack, made from 100% genuine humans.
<nckx>Ya weirdo.
<the_tubular>I mean, if this is cannibalism, then botsnack is too ...?
<nckx>sneek is a bot…!?
<nckx>I thought it was just a very helpful person.
<porcupirate>Sometimes I see messages that would make great fortune messages...
<porcupirate>Speaking of which, is fortune up-to-date?
<nckx>I mean, it's more up to date than its home page appears to be.
<porcupirate>latest github release 3.12.0... guix has 2.28.0
<nckx>I noticed. I'm trying a bump.
<porcupirate>It has new dependencies
<nckx>(‘games.scm’? Really? Much game. Very play.)
<porcupirate>Better in toys.scm?
<lilyp>probably predates toys.scm
<porcupirate>I've been complaining about guix package organization for years now.
<nckx>Yes! Much toy! Very play!
<nckx>And yet it didn't seem to magically help? Thus, porcupirate was enlightened.
<porcupirate>I've suggested moving things, but everyone seems fine with using `guix search`. So I use `guix search`
<lilyp>we haven' found the enlightenment that is gentoo style folders yet :P
<nckx>One can bee *too* enlightened.
<notmaximed>Patch series for git-fetch support for "guix refresh -u" (tested with 'generic-git' and 'minetest'):
<sneek>notmaximed, you have 1 message!
<sneek>notmaximed, mbakke says: making native-inputs accessible for native builds would also solve
<the_tubular>In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): "bash"
<porcupirate>the_tubular: map specification->package
<the_tubular>What does that mean ?
<notmaximed>Does someone know the nick (if any) of Xinglu Chen
<nckx>Looks up a string ("bash") name as a package object (=struct, here).
<notmaximed>sneek: later tell civodul, iskarian: new git-fetch updater patch series: <>
<sneek>Got it.
<notmaximed>sneek: later tell iskarian, civodul: new git-fetch updater patch series: <>
<sneek>Will do.
<notmaximed>(not sure if sneek supports multiple recipients)
<notmaximed>sneek: botsnack
<nckx>sneek: tell notmaximed, nckx: I wondered the same thing.
<sneek>notmaximed,, nckx says: nckx: I wondered the same thing.
<nckx>That's what I feared.
<nckx>sneek is enlightened… but with a dimmer.
<the_tubular>I like guix a lot but god damn does it feel like black magic
<porcupirate>sneek: what is black magic?
<the_tubular>I replaced it porcupirate and now I get an old error : error: my-packages: unbound variable
<the_tubular>hint: Did you forget `(use-modules (my-variables))'?
<the_tubular>But I have it in that line : (use-modules (gnu) (guix) (guix packages) (srfi srfi-1) (my-variables))
<nckx>This is the second time in a few days that someone has reported that kind of seemingly-obviously-wrong thing.
<the_tubular>What nckx ?
<nckx>hint: Did you forget ‘(use-modules (module you already bloody used))’.
<porcupirate>echo "(use-modules (my-variables))" | guix repl -L $PWD
<porcupirate>I might have mismatched parens
<the_tubular>You might be too fast blaming guix, it might still be me
<nckx>Blaming Guix is more polite.
<the_tubular>porcupirate what is suppose to be the expected output ?
<the_tubular>I just get another unbound variable
<porcupirate>the_tubular: An error if there was a mismatched variable
<the_tubular>error: specification->package: unbound variable scheme@(guix-user)>
<nckx>porcupirate: You'll have to package the author's new pet documentation generator, at least:
<porcupirate>That's the error
<porcupirate>(use-modules (gnu packages))
<porcupirate>in my-variables.scm
<nckx>the_tubular: Were any other errors/warnings printed before ‘error: my-packages: unbound variable’?
<nckx>It's fair to blame Gui{le,x} if not.
<porcupirate>Just missing (gnu packages) module to define specification->package
<the_tubular>error: my-packages: unbound variable hint: Did you forget `(use-modules (my-variables))'
<the_tubular>No nckx
<porcupirate>the_tubular: prepend (use-modules gnu packages) to my-variables.scm
<the_tubular>I did
<porcupirate>Still not working?
<porcupirate>If there's a problem with guile, it's that it hides error messages that prevent modules from loading.
<simendsjo>jpoiret: The issue you linked was in fact my issue, thanks! I looked at the sbcl package, and I also needed the workaround for XDG_CONFIG_DIRS, not just XDG_DATA_DIRS
<nckx>porcupirate: Yes.
<the_tubular>porcupirate :
<the_tubular>This is correct right ?
<porcupirate>nckx: the_tubular: I've had to use cli magic. Like echo "(use-modules (my-variables))" | guix repl -L $PWD
<porcupirate>the_tubular: Move line 1 below line 3
<nckx>Have you reported this bug? I will otherwise, but it would be slightly better for everyone if someone actually affected would (since you understand the issue well enough to have found a work-aronud, at least).
<florhizome[m]><the_tubular> "I'm pretty good with ctl, but..." <- the wonders of guile begin when you do “guix repl” and then ,help
<florhizome[m]>but it’s debug messages are shit and syntax can get from elegant to a bit obscure pretty fast. And the man is not very good.
<the_tubular>Nope, porcupirate
<nckx>Ah, ‘the man’ is to blame, of course.
<porcupirate>nckx: I don't know if it's guix repl or guile that suppresses use-module errors.
<nckx>From a user perspective, who cares :)
<porcupirate>Send to both mailing lists?
<civodul>notmaximed: yay for the git-fetch updater patches, i'll take a look!
<sneek>Welcome back civodul, you have 1 message!
<sneek>civodul, notmaximed says: iskarian: new git-fetch updater patch series: <>
<nckx>Or start with Guix, (at least) Ludo' swings both ways anyway.
<nckx>…or ask civodul themselves, apparently :)
<florhizome[m]>of course, who would I be to blame but my own kind?
<florhizome[m]>who argues against a more gentle man?
<the_tubular>Any other ideas porcupirate ?
<florhizome[m]>plot twist civodul = the man but this time I mean it in terms of he’s epic
<porcupirate>the_tubular: What does this output? echo "(use-modules (my-variables))" | guix repl -L $PWD >/dev/null
<nckx>the_tubular: What does my-variables.scm look like? To me, it seemed like porcupirate was implying that errors in m-v.scm weren't ‘propagated’ when evaluating a module that imported it. Else I misunderstood.
<civodul>porcupirate: it's the Guile REPL that swallows use-modules errors
<civodul>howdy nckx :-)
<the_tubular>It outputs nothing porcupirate
<porcupirate>The import error should be output to stderr.
<porcupirate>the_tubular: What about without >/dev/null
<nckx>civodul: Has this been tracked already?
<the_tubular>error: specification->package: unbound variable
<porcupirate>or echo "(use-modules (my-variables))" | guix repl -L $PWD >/dev/null; echo $?
<nckx>the_tubular: Are you using specification->package in m-v.scm?
<the_tubular>Yes, nckx
<nckx>Are you importing (gnu packages) there too?
<civodul>nckx: i wouldn't necessarily call it a bug, but we can discuss :-)
<the_tubular>porcupirate told me to add it
<civodul>i didn't follow the whole discussion tho
<nckx>civodul: Same! It was just… surprising, that's all I know.
<porcupirate>civodul: the_tubular used a list of strings to name the desired packages
<porcupirate>Potentially easier to map specification->packages instead of use-package-modules for all the packages.
<the_tubular>Wait it's specification->packages ?
<porcupirate>the_tubular: Take a string. Output the first package named that string.
<nckx>But from the last message it might have been more confusion to which file the ‘hint’ applied. It applied to m-v.scm (is my guess), but this is not where the (use-modules) was added.
<nckx>the_tubular: No.
<nckx>Singular, both.
<porcupirate>(specification->packages "hello") searches for first package named "hello"
<nckx>And (map s->p (list "hello" "is" "it" "me")) applies it to each package "name" in turn, returning a list of packages.
<the_tubular>So what should I try next ?
<nckx> is correct AFAICT.
<porcupirate>But it would fail because there is no "is" or "it" or "me" package.
<nckx>porcupirate: They are proprietary.
<the_tubular>I wonder if podman is gonna give me as much trouble
<nckx>the_tubular: Are you using specification->package in m-v.scm? --> so, did you add (use-modules (gnu packages)) there?
<the_tubular>I'm pretty good with docker, used podman once before
<nckx>I don't see what podman has to do with this.
<the_tubular>You pasted m-v.scm above nckx
<nckx>So I did.
<nckx>Never mind.
<the_tubular>This is pretty much what my file looks like
<nckx>I had remembered define- as use-.
<the_tubular>nckx podman has nothing to do with it. Just wondering if I'll be debugging stuff fore hours on end or not :P
<nckx>If you prefer podman, use podman.
<the_tubular>I do, but it just got merged
<the_tubular>And we don't have systemd
<the_tubular>Which podman really pretty heavily on
<the_tubular>rely ?
<nckx>I really don't understand this bug, and if it's not a bug, I sure don't see the typo.
<porcupirate>I can't make a dummy sample where guile suppresses (use-module) error.
<the_tubular>Are you sure the problem is in that file and not my config.scm ?
<nckx>the_tubular: What if you replace the first two lines with (define-module (my-variables)
<nckx> #:use-module (gnu packages))
<nckx>the_tubular: No.
<nckx>Well, I'm not, at least. I just bumbled in here 15 minutes ago.
<the_tubular>Is it really   #:use-module (gnu packages)) ?
<nckx>I can reproduce this only if the use-modules precedes define-module, but you already said above that you swapped the two.
<nckx>#:use-module should be equivalent to the correct order anway.
<the_tubular>Same error
<porcupirate>Gotta love how theere are a zillion ways to do something in guile
<nckx>I can't reproduce that. Sorry.
<the_tubular>Mind to throw me a debpaste, just to confirm nckx ?
<nckx>Of course:
<nckx>Command follows it.
<nckx>If I swap the two first lines I get your error, but not otherwise.
<porcupirate>Stuff at the end
<the_tubular>This is clearly not what I was seeing nckx
<nckx>I believe you. I just can't debug what I can't reproduce.
<the_tubular>Same error still ...
<porcupirate>What are you calling and what error are you getting?
<the_tubular>I'm calling sudo guix system reconfigure config.scm
<porcupirate>And it says it can't find m-p?
<the_tubular>I think it works
<the_tubular>Hold onm
<the_tubular>guix system: warning: cannot determine provenance for current system
<the_tubular>Should I ignore that ?
<porcupirate>It's just confused that you separated your config into modules.
<mekeor[m]>by the way, for the record, as far as i know, module names depend on the directory structure, too. if you import (foo bar) in a file /lorem/ipsum.scm, then there should be /lorem/foo/bar.scm, right?
<porcupirate>It might have problems if you have unattended updates.
<porcupirate>the_tubular: Make sure you include -L $PWD
<porcupirate>the_tubular: sudo guix system -L $PWD reconfigure config.scm
<porcupirate>Nevermind. Forgot we worked around that.
<the_tubular>Too late, should I ctrl c ?
<porcupirate>It's working?
<porcupirate>Building stuff
<the_tubular>Thanks a lot!
<porcupirate>Glad to help.
<the_tubular>Is there a noob to pro book for guix/guile ?
<porcupirate>info guix-cookbook?
<porcupirate>There might be some blogs in the wild.
<porcupirate>I learn everything from the source.
<porcupirate>I suppose you can use sneek to clarify things.
<porcupirate>sneek: what is specification->package
<the_tubular>sneek is sleeping
***Noisytoot is now known as ducks
***ducks is now known as Noisytoot
<porcupirate>Maybe it's because I asked it about black magic and it got scared...
<porcupirate>private chat it's only recognizing help
<porcupirate>sneek: botsnack
<oriansj>porcupirate: black magic doesn't exist, only understanding that isn't properly shared or improperly used.
*nckx ret. porcupirate, the_tubular how was it fixed?
<nckx>Hmm. :)
<porcupirate>the_tubular tried it again and it worked.
<porcupirate>I don't know if there were additional edits or what.
<nckx>Well, I agree in spirit with oriansj (and with Arthur C. Clarke for all the rest). I suspect a brain/computer desync of some kind.
<porcupirate>sneek: botsnack
<nckx>Maybe an edit that was thought saved that wasn't. It happens.
<nckx>sneek: good night, sweet botsnack.
<porcupirate>sneek Responds to botsnack, doesn't respond to my apropos.
<porcupirate>Oh, now it's responding...
<oriansj>nckx: perhaps but not for long if guix gets its way.
<oriansj>a real mapping between source and binaries finally become the standard.
<the_tubular>My new kernel is building, I'll go get some humansnack
*the_tubular Snacks for everyone!
<porcupirate>3 hours late for lunch. G2G
<the_tubular>Damn it just crashed lol
<the_tubular>Right when you left
<the_tubular>ice-9/boot-9.scm:1685:16: In procedure raise-exception: In procedure for-each: Not a list: "/boot/efi"
<the_tubular>Ohh, it has to do with target vs targets