IRC channel logs


back to list of logs

<mbakke>rekado_: you can lazily reference it in 'inputs', like e.g. "sambamba"
<lilyp>badcodec: you can get a rough equivalent of nightly with the --with-latest package transformer
<lilyp>note that it builds from the latest git commit without any checks whatsoever (other than the test phase)
<lilyp>note also that nightly builds are typically meant for huge CI machines, not your lone laptop
<badcodec>did "guix install rust --with-latest=rust" and it says "could not determine latest upstream release of rust
<badcodec>lilyp: yes I could set up a server to do building
<lilyp>okay, in that case you need to hack your own scheme script which inherits the package definition and bumps the package + hash
*badcodec will be exploring it, thank you
<lilyp>"package + hash" meaning version + hash
<badcodec>> COVID is failing... release the HIV!
<badcodec>wrong channel
<podiki[m]>any advice for a package that has a (basic) makefile, but is built with go?
<podiki[m]>is it better to translate the commands make would run to extra build phases (some extra "go" invoking, installing files) or to bring in make? (not sure how)
<podiki[m]>oh nevermind, just wasn't in the correct directory (go-build-system has make available, but source directories are in that go style)
***califax- is now known as califax
<the_tubular>Can I get gcc-emacs on guix ?
<ryanprior[m]>the_tubular: You can but it's not part of upstream yet as far as I know.
<ryanprior[m]>Here's the package I've been testing, which enables native-comp (the libgccjit feature,) pgtk, and libsqlite3.
<the_tubular>Nice, thanks :)
***aya is now known as gyara
<Ribby>Turns out that few moments became a few hours short of a day! Maybe around 12 hours plus a schedule?
<Ribby>There's a bit of speckering in net stability and blank/sleep screen interruption during update/pull and upgrade, but so far things looking up.
<Ribby>I got the package extraction down, but now how do I make it install a program? I was just going for the flathub, but apparently, no guix support? Should I just pick a choice? Source code compilation is another time, I couldn't find the dependencies anyways (I am not the said project developer anyways).
<Ribby>Hold on, I'll restart because it's usually a norm for updates.
<Ribby>Ok, I back.
<Ribby>Do I use debian?
<Ribby>ubuntu worked though, at least for ubuntu.
<ryanprior[m]>Not sure what you're asking, Ribby. What's the context?
<Ribby>Hold on1
<Ribby>I love monospace, it's too spacey, but it works! I could also bolden it.
<Ribby>Much better!
<podiki[m]>the_tubular: see also
<the_tubular>woo, much choiches!
<Ribby>size 10 bold or size 12 bold? Will get around flathub/flatpak compatibility with guix, but so far, I read that debian is the closest dna to guix.
<Ribby>I will write down the debian choice for guix 32 bit.
<the_tubular>nix is probably far closer to guix than debian
<the_tubular>But i'm not sure if I understand correctly what you are trying to say
<Ribby>Oh, this is for installing flathub/flatpak.
<the_tubular>Ohh, no clue about those.
<Ribby>I gave up on installing installer files/packages and compiling source code for now (dependencies can be tricky).
<Ribby>No problem.
<Ribby>Oh yes, debian record blah blah on notes.
<Ribby>I have a flathub question, but I think it will fall flat in terms of its GUI visibility, prefix (custom installation directory path), and uninstallation questions.
<podiki[m]>flatpak works fine on guix, is that what you are getting at?
<podiki[m]>(I'm using it right now)
<Ribby>Well, I installed it, after some time with update/pull and upgrades.
<Ribby>I can run it thru the terminal, but is there a .lnk or graphical icon to start that chrome variant?
<podiki[m]>you mean a flatpak version?
<Ribby>That be ungoogled chromium.
<Ribby>I'm trying to find its directory to see if there's a gui to mouse click activation.
<podiki[m]>flatpak makes desktop files, that can be used from launchers or other things like application menus
<podiki[m]>you will need to add it to xdg_data_dirs
<podiki[m]> is what I do, but that looks more complicated than it is
<podiki[m]>you'll have in your profile, like ~/.guix-profile a directory with a script you can use, in etc/profile.d
<podiki[m]>or you can do it manually, doing flatpak run <your app> should print out a warning of the directories to add
<podiki[m]>probably ~/.local/share/flatpak/exports/share
<podiki[m]>gotta run, good luck!
<Ribby>So I just need to copy/write some files in that share directory?
<Ribby>there's a .bash_profile and .zprofile. they are in readable text file format.
<podiki[m]>no, that share directory should have an applications subdirectory, the .desktop files are there, which is what launchers and menus will look for typically
<podiki[m]>so in there should be whatever flatpak application you installed, its .desktop file
<podiki[m]>the point is you need that ../share in XDG_DATA_DIRS to be useful, probably set it in your .profile and log back in
<podiki[m]>ok really going now, but if you don't figure it out I'm sure someone can help (at maybe busier time)
<Ribby>Np, is there a contact or someone will take up answers?
<podiki[m]>there's here and the help-guix mailing list
<Ribby>I don't really mind typing in commands. I just need a text file to remember it, lol.
<Ribby>Ok, thx. See you laters and thank you for helping.
<podiki[m]>welcome! catch you next time
<Ribby>I have to say that for v1.3.0, wifi capability might not be sufficient enough. However, I do have access to ethers.
<Ribby>Watch out guys, I touched a b43 wireless driver site, and the default web browser goes kapoof!
<abcdw>Ribby: it's very likely that your wifi cards uses proprietary blobs, which are not supported by linux libre.
<Ribby>It's true, it's probably true.
<Ribby>Still, I think it's a good deal at the least.
<Ribby>It's likely that Atheros wifi usb will work for guix v1.3.0?
<abcdw>Idk, better look up it in the internet. The first links from ddg:
<abcdw>The list is not complete, but knowing the exact adapter model, probably, it is relatively easy to check somewhere else if your card is supported by Linux-Libre.
<Ribby>Yeah, I'm just going to do my shopping research with the variable at hand.
<Ribby>Of course, of course.
<apteryx>Ribby: have you checked
<xelxebar>apteryx: Nice! This site is definitely going in the bookmarks.
<Ribby>Sounds like a start! :)
<kocio>Hi, when I try to build gtg, I get the problem with meson not finding python3
<kocio>which is strange for me, because i can see python3 in my path
<gnoo>how can i use the system services as user services in user shepherd? e.g. there's a postgresql-service-type, how can i use that from a user's shepherd ?
<apteryx>you can't, but I hear that Guix Home intends to bridge this gap
<apteryx>kocio: try adding python-wrapper
<apteryx>as a native-input
<gnoo>apteryx: thanks
<kocio>hm, how should I do it?
<kocio>you mean that I should somehow edit gtg deinition?
<apteryx>yes; there was a change in meson where python is no longer propagated
<apteryx>gtg must have gone broken since
<gnoo>i have (service dnsmasq-service-type ...) but the dns server doesn't seem to be written to /etc/resolv.conf; dnsmasq itself is running. i also have (service dhcp-client-service-type) in there
<gnoo>this is my (services ...) part, do i need anything extra?
<kocio>apteryx: is there some guide how to fix (deive?) some package like meson?
<apteryx>the build is still broken, right? it's using an invalid flag
<apteryx>I'd suggest trying to update gtg to its latest version, if that ain't already the case.
<apteryx>that'd probably fix it; else look at upstream issues
<apteryx>finally look at the code/changelog and adjust the obsolete flag
<kocio>trying to install gtg just downloaded latest available sources (0.5) and was trying to build it which failes
<kocio>but since it's medon build system which i
<kocio>is broken, I think it should be fixed
<kocio>where should I find issues for y
<kocio>as you said, this is some meson problem in guix, not upstream
<kocio>with guix powers it should be also possible to modify and use my version of meson in the meantime
<gnoo>while defining a user service i get: unbound-variable(#f "Unbound variable: ~S" (mkdir-p) #f)
<gnoo>which module provides mkdir-p ?
<efraim>(guix build utils)
<roptat>hi guix!
<roptat>fosdem is starting, and you should be in the declarative and minimalistic computing devroom already :)
<phf-1>Hello Guix! I'm trying to package `python-psycopg2-binary' but it gives errors I dont' know how to resolve. Here is an error: Here is the package definition: I cannot share the full trace: too big.
<jpoiret>roptat: Mathieu Othacehe is also doing a talk right now in the the distribution devroom, although the video is having some issues apparently
***janneke_ is now known as janneke
<gnoo>phf-1: try removing the tests, it looks like the tests connect to a running postgresql instance
<lilyp>was there no minimalist devroom yesterday or are those simply not available after they're done?
<lilyp>phf-1: alternatively try to start a postgresql server that exists only in the build directory or somewhere else in /tmp
<jpoiret>lilyp: only sunday, yes
<lilyp>okay, then I hope I'll have some more time after they're done
<phf-1>gnoo, lilyp I'm ok with not running the tests. The idea is to contribute the package to Guix and so, maybe it's worth it to try to make the tests work. If it's ok not to make the tests work then I can send it to Guix with #:tests? #f maybe.
<gnoo>phf-1: try sending with #:tests? #f and explain why the tests are disabled. i think they recommend to always to the test but if it's a hassle setting that up then...
<gnoo>maybe there's some other postgresql-related package that runs similar tests?
<phf-1>gnoo, Will search, thanks.
<gnoo>phf-1: both ecl-postmodern and postgis have `(#:tests? #f...)
<phf-1>gnoo, well, `(#:tests? #f...)' it will be then, thx!
<jpoiret>phf-1: remember to add a comment indicating why tests are disabled
<jpoiret>oh, the libera->fosdem matrix chat is only one way, my question didn't go through :(
<phf-1>jpoiret, Ok.
<phf-1>Here is the license of the code: What should I fill the `(license ?)' field with in the package?
<civodul>hey there! everyone tuned in for FOSDEM? :-)
<phf-1>civodul, hi! Please, could you help me with that? Not sure what to fill the license field of the package with. Here is the link of the license: Here is the package so far:
<civodul>phf-1: hi! that looks like LGPLv3+, but check the source file headers to see if that matches
<civodul>great talk by ekaitz on RISC-V minutes ago
<phf-1>Great, thank you!
<KE0VVT>How do I get a media server set up on the system?
<jpoiret>wdym by media server?
<jpoiret>i'm afraid we don't have the big ones such as plex/jellyfin, although that could change with some effort
<KE0VVT>jpoiret: I do miss Jellyfin, but I will settle for minidlna.
<jpoiret>it's apparently called readymedia now, but guix has it!
<KE0VVT>jpoiret: But setting up the server req. messing with config.scm, and I'm scared of it.
<KE0VVT>abcdw: I'm trying to get RDE to work for me. Here are my changes so far:
<abcdw>KE0VVT: Looks legit. BTW, we hang out on #tropin channel with some folks using rde.
<lasnesne[m]>Are there anyone not having issues with pyqt dependent softwares? anki and e-reader in calibre are also broken for me; most of texts are invsible.
<vivien>lasnesne[m], did you install fonts?
<lasnesne[m]>vivien, yes, I already checked my font installation before. Also, Toolbar and some others work flawlessly, so i think it isnt font problem.
<vivien>I’m a bit clueless, sorry.
<lasnesne[m]>Well, maybe i should poke it around with virtual machine and ‘guix shell’
<florhizome[m]>Might be a qt issue.
<florhizome[m]><jpoiret> "i'm afraid we don't have the big..." <- would love to have a channel for selfhosting
<mroh>florhizome: there is
<florhizome[m]>No, a guix channel :D
<davidl>Hi I noticed that when you specify --image-size for guix system image, the size of the outputted image is actually larger. More specificially it outputs an image that is exactly the additional size of the root offset and the esp-partition as defined here: Anyone who wants to
<davidl>make a quick fix? mbakke: this also causes a problem in the create script of ganeti-instance-guix leading to failures with the qemu-img command therein which copies the guix system outputted image to the target device ("cannot grow device file" or similar).
<davidl>just remove echo $(( $(( 2**20 * 40)) + $(( 512*2048)) )) in from image-size in the procedure here:
<mbakke>davidl: can you report a bug (or patch :-)) about --image-size?
<mbakke>davidl: for ganeti-instance-guix I'll make it not use --image-size at all, because it wastes space in the store
<mbakke>but then, VMs will have to resize their file system after boot
<mbakke>or during boot, "cloud-init" style
<davidl>you could just add a line removing the exact bytes that you calculated with blockdev command against the target device.
<davidl>Uploaded file:
<davidl>mbakke: ok thats a good reason and a good idea. Here is some proof and fix for the create script if keeping the image-size argument (--image-size=10G and ganeti storage was also set to 10G):
<davidl>root@node2 ~/src/ganeti-configs# du -b /gnu/store/abvy7jqp06jg0q0lxgs7plsxcravldwc-disk-image
<davidl>10780409856 /gnu/store/abvy7jqp06jg0q0lxgs7plsxcravldwc-disk-image
<davidl>root@node2 ~/src/ganeti-configs# DEVICE_SIZE=10780409856
<davidl>root@node2 ~/src/ganeti-configs# blockdev --getsize64 /dev/vg-drbd/622610a0-cbab-4c92-9eeb-d76ca47d0980.disk0
<davidl>root@node2 ~/src/ganeti-configs# echo $(( DEVICE_SIZE - $(( $(( 2**20 * 40)) + $(( 512*2048)) )) ))
<davidl>root@node2 ~/src/ganeti-configs# echo 'thus we can do this: DEVICE_SIZE=$(( DEVICE_SIZE - $(( $(( 2**20 * 40)) + $(( 512*2048)) )) )) directly after setting DEVICE_SIZE in the create script in ganeti-instance-guix'
<mbakke>davidl: subtracting the EFI partition size from DEVICE_SIZE seems like a good workaround for now, would you like to submit a PR? I can relase a hotfix version with it.
<davidl>Yeah, I can submit by tomorrow probably is that fine? I have to leave my computer about now.
<davidl>mbakke: and thanks!
<dlowe>I'm trying to do a complicated dovecot configuration. Is there a break-glass configuration option where I can just specify a key value?
<mbakke>davidl: sounds great, I don't have much time to test today anyway :-)
<dlowe>[looks at configuration source] apparently not
<dlowe>Is there something like guix edit but for services instead of packages?
<vivien>Hi, software that I want to package requires access to a data set to run the tests. Problem is, there seems to be no license for the data set:
<vivien>I think it’s not OK to package the test data set, so I just disable the tests in the final package; what do you think?
<civodul>dlowe: no, but "guix system search" shows clickable source locations
<civodul>vivien: yes, with a comment indicating what the situation is
***jonsger1 is now known as jonsger
<iung>osbooted x200 crashes (something i915) on newest linux-libre.
<dlowe>civodul: okay, thanks!
<dlowe>Hmm. clickable in what?
<civodul>dlowe: in terminals that support hypertext links
<civodul>so not xterm, not Emacs shell-mode, but something like GNOME Terminal
<dlowe>okay, I don't get hypertext links here, just /gnu/services/mail.scm:lineno/colno
<civodul>looking at the code, that part should be a hyperlink on terminals that support it
<tschilptschilp23>Is it fine to have 'guix system' and 'guix home' at different states (guix home at a newer state than system, by means of 'reconfigure' calls)? I'm asking because I just had some strange gnome-behaviour, that somehow seems to have been solved by a system reconfiguration...
<abcdw>tschilptschilp23: In most cases yes, I even have different channels for System and Home.
<tschilptschilp23>abcdw: thanks! maybe it just was time for a few reboots...
<tschilptschilp23>abcdw: basically my programs were running, but I couldn't use them, because in not-overview-mode I had this grey 'oops, something went wrong that couldn't be resolved by the system, please logout and try again' fullscreen-window covering everything else!
<tschilptschilp23>(hadn't seen this on guix before!)
<tschilptschilp23>But i started with guix-home just a short while ago, so I wanted to make sure, that I'm basically on the right track!
<abcdw>Generally, it should be ok to have Home and System slightly 'out of sync', however, if it happenning repeatedly, please write to bug-guix.
<abcdw>I mean the issue you had :)
<tschilptschilp23>abcwd: I will have an eye on that. They were some 9 pulls apart from each other because I was changing a lot in Home and forgot a bit about System :)
<jpoiret>TIL macOS has a sandboxing feature that can restrict access to filesystem/network and other things!
<form_feed>I'm working on a system configuration for my first Guix real installation. I want a LUKS>LVM>BTRFS setup. In the manual, I see that the `swap-space` declaration accepts a `dependencies` list, but I see none for `mapped-devices`. My question is: is it enough to have the `mapped-device` declarations listed in the expected order to indicate a dependency? So `(mapped-devices (list (mapped-device LUKS)
<form_feed>(mapped-device LVM)))`?
<jpoiret>form_feed: for now, there's no such field, and so i think it's the order that matters
<jpoiret>although I think the order that actually matters is that of the (dependencies ) field of the root file system
<form_feed>jpoiret: oh, that makes absolute sense. I was overthinking it. Thank you once again :)
<jpoiret>although I must ask, why LVM>BTRFS?
<jpoiret>BTRFS has subvolumes that might be good enough for your uses
<jpoiret>that's what i'm using myself
<form_feed>jpoiret: Well, the short answer is that this needs to be a bulletproof "work" system and, at this point, I'm not sure if I'll be able to get Guix the way that I need in time. So I'd like to ensure that, if needed, I'll be able install something else quickly to get something done before I resume my migration to Guix. I like the flexibility LVM gives me there.
<jpoiret>right, just saying that subvolumes can achieve the same here
<jpoiret>as an example, I had /guix_root and /arch_root on my btrfs partition
<form_feed>With multiple distros? Oh, never saw that. O_O
<jpoiret>with 2 completely different systems on them while I was moving to guix
<jpoiret>basically, you mount the same partition but with a different root
<jpoiret>right now, my home is the subvolume at /arch_root/home while my root is /guix_root
<jpoiret>and I mount the btrfs's / to /btrfs
<jpoiret>(meaning my guix root is both at / and /btrfs/guix_root)
<form_feed>That sounds really good. Makes sense to move the root from / to something named. Thanks for this tip!
***jonsger1 is now known as jonsger
<KE0VVT>I forgot the env. var. that makes Alacritty work on the X200.
<jgart>Hi Guixers! Is this something I can ignore when compiling Guix? "Please consider running po4a-updatepo to refresh it."
<unmatched-paren>hello guix; i've been meaning to do this for a while, but was scared off because last time i did it, i bricked everything and now i've forgotten; how would i remove GDM without removing the whole of %desktop-services?
<unmatched-paren>(also for some reason both pipewire and pulse are running...)
<unmatched-paren>also, since i use wayland, can i safely remove the (set-xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout))) thing?
<podiki[m]>use modify-services and delete
<jgart>unmatched-paren, do you have your system configs online somewhere?
<podiki[m]>(modify-services %desktop-services (delete gdm-service-type)) in your services list
<unmatched-paren>podiki[m]: thanks
<unmatched-paren>jgart: not currently, no
<podiki[m]>I don't have a keyboard configuration line on X, but I also don't use a custom layout (I use xmodmap for some tweaks instead)
<crackedwheat>I installed guix on a laptop following davidwil and I find I am missing .guix-profile dir in my home directory.
<roptat>jgart, yes you can ignore this
<roptat>it's po4a being confused because we create two files from the same po
<lfam>crackedwheat: That directory should be created the first time you install a package as your user
<crackedwheat>From, I am supposed to have this directory.
<roptat>crackedwheat, it'll be created after you create the first generation
<podiki[m]>crackedwheat: perhaps you don't have a user profile yet? is that possible? what happens if you just try to install something, like "guix install hello"
<crackedwheat>oh, thanks
<jgart>roptat, THNX, got it
<podiki[m]>right, makes sense
<crackedwheat>yes, installation of hello did create that directory
<unmatched-paren>now i'll just hope my login doesn't blow up like last time...
<unmatched-paren>hm, i just noticed i have both networkmanager and wpa_supplicant installed too, aren't they mutually exclusive? or does one build o the other?
<jpoiret>iirc networkmanager uses wpa_supplicant, either one internal to nm or another external one
<vivien>In guix, you need to have a wpa_supplicant service for networkmanager to run
<vivien>(run is not the appropriate term; guix system reconfigure won’t allow a configuration with a network manager service but no wpa_supplicant)
<phf>Hello Guix! I try to follow but I keep having this issue: Essentially, it says: `guix environment: warning: variable 'PKG_CONFIG_PATH' is missing from shell environment', except that I do not touch this variable in .bashrc
***phf is now known as phf-1
<jpoiret>phf-1: is bash your shell?
<jpoiret>do you do any funky things in your bashrc?
<jpoiret>bashrcs should not have anything beside customize the shell itself when it starts
<jpoiret>customizing *
<phf-1>jpoiret, no:
<jpoiret>nothing in .bash_aliases or .bash_functions that could mess this up (who knows)?
<phf-1>`.bash_functions' :
<jpoiret>looks like this might be
<jpoiret>well, you can check manually that PKG_CONFIG_PATH is properly set inside the shell and proceed from there, ignoring the --check warning
<phf-1>jpoiret, Ok, thank you!
<Haider>Even though I have installed gcc and sqlite and emacsql, I cannot get OrgRoam to work and make it's database. Do you know what I could do?
<jpoiret>if you're confident that .bashrc isn't messing things up, then it should work without any issues :) it's not mandatory to --check
<jpoiret>Haider: do you have any error message?
<Haider>I get this in the minibuffer: emacsql-sqlite-ensure-binary: No EmacSQL SQLite binary available, aborting
<Haider>I have installed gcc though.
<jpoiret>phf-1: FTR, are you on Guix or foreign (and if so, which distribution)?
<phf-1>jpoiret, Yes. Debian11
<Haider>Im on Guix (non foreign)
<jpoiret>phf-1: seems like this happens for people on non-guix, no idea why though
<phf-1>jpoiret, Ok, thanks for looking into it!
<phf-1>and yes, you were correct. It's that issue. PKG_CONFIG_PATH is correctly defined wo the --check flag.
<jpoiret>Haider: what package exactly did you install?
<jpoiret>gcc-toolchain should be the one
<Haider>ill try that
<jpoiret>can you try eval'ing (executable-find "gcc")?
<jpoiret>also, we have emacs-emacsql packaged in guix, maybe this would work better
<Haider>gcc-toolchain works!
<jpoiret>great! :)
<Haider>ur a legend thank you :)
<Haider>btw, How did u know that
<Haider>anyway, thank you for your help.
<jpoiret>Haider: it's classical Guix folklore :)
<jpoiret>the gnu compiler toolchains are always called xxx-toolchain
<jpoiret>although it might be better to hide "gcc" from the user-facing package list
<roptat>it's already what we do
<roptat>gcc is hidden, and instead it installs gcc-toolchain
<roptat>nevermind, it's no longer the case
<jpoiret>right, a simple hello world doesn't compile with gcc
<jpoiret>Haider: basically, the difference is that gcc is only the gcc compiler, without glibc or the binary utils that it might invoke
<jpoiret>the reason there's a separation between both might be because of bootstrapping, but I don't actually know
<Haider>Makes sense. Thanks all of you :)
<podiki[m]>question about how to refer to a filesystem, e.g. in swap:
<podiki[m]>what is the variable that goes in the list (see last examples)? how do you refer to a filesystem declared earlier?
<roptat>I'm also quite confused by that example...
<podiki[m]>I've seen some people's examples where they select from the full operating-system filesystems list....?
<podiki[m]>currently I have a needed-for-boot? in my btrfs swap subvol, but I believe that can be dropped and folded into the swap dependencies...
<podiki[m]>but the example is lacking
<podiki[m]> (looking to close that out, probably with a better manual example)
<phf-1>Building Guix from source gives many of these: `;;; no code for module (semver)'
<rekado>phf-1: semver is provided by a separate package. It is optional.
<podiki[m]>there's no name field or something like that for a file-system right?
<podiki[m]>as in, name to refer to it in the swap
<phf-1>So, I've one error after running make check building Guix from source (master branch): Am I supposed to send a bug report?
<rekado>phf-1: which of tests failed?
<phf-1>rekado, this one:
<form_feed>jpoiret: is your system configuration publicly available?
<unmatched-paren>as expected, removing gdm causes the boot process to hang (and also loads more logs to appear for some reason...)
<unmatched-paren>i'd like to completely remove gnome, but looks like that won't happen anytime soon...
<jpoiret>form_feed: no, not yet unfortunately
<jpoiret>unmatched-paren: I don't think it should, i've seen plenty of users that don't use GDM
<jpoiret>can you access other ttys?
<jpoiret>podiki[m]: here's the relevant part of my config if you want some inspiration
<podiki[m]>unmatched-paren: I don't use gdm (use sddm) but on X and with Xmonad
<podiki[m]>jpoiret: thanks. I would think the manual example should be more useful (what does it work for?) and even better the dependencies of swap-space should be easier to use
<kozo[m]>unmatched-paren: For instance, this is my setup... (full message at
<kozo[m]>If you are removing GDM, you need to add something to take it's place
<jpoiret>podiki[m]: what do you think is difficult in the dependencies mechanism?
<phf-1>So, I've setup a small blog with tutorials about Guix: (username: guix-channel, pwd: guix). It's very rough and are quick notes I took during my journey.
<phf-1>for example:
<podiki[m]>not difficult exactly, but a you should be able to more directly specify the file-system
<podiki[m]>the example in the manual is short and no idea for what case it works
<phf-1>Some things are obviously wrong, others less so.
<jpoiret>right, i think a more complete example that would also define the variable names would be clearer
<jpoiret>especially for people not acquainted with Guile
<jpoiret>i'll try working on that
<podiki[m]>I think it would be nice to have a way to refer to a file-system specified directly, not sure how
<jpoiret>well, you can do it with something like what i do
<podiki[m]>maybe just a simple function that would basically due what you did
<phf-1>So, question: if this can be of any help I can work on it a bit more, take inputs into account to make it better.
<jpoiret>right, maybe a helper would be nice, for people without much Guile knowledge
<phf-1>can this be of any interest?
<podiki[m]>phf-1: only took a very quick look, but in general I always vote for more examples and notes from people out there
<jgart>phf-1, awesome! but asking me for a login..
<podiki[m]>jpoiret: presumably such a thing would be useful in other ways, when you need a file-system reference (I don't know what, but seems like something that should be available in the configuration system)
<phf-1>podiki[m], It's very much in flux. Trying to centralize all the inputs I get from the channel and aggregate them somewhere.
<jgart>phf-1, is there a git repo with the posts?
<phf-1>jgart, not yet. I can setup that and take inputs from issues I guess.
<phf-1>jgart, is that the idea?
<jgart>just wanted to read the posts
<jgart>the blog posts that you mentioned above, that is
<podiki[m]>jpoiret: thanks in advance for potentially working on this; better manual example would be step 1, a little synctatic sugar even better :)
<phf-1>jgart, you cannot access the website?
<jpoiret>synctatic sugar i'm not sure, but a helper function is definitely possible
<jgart>phf-1, nope
<jpoiret>i wrote the swap dependencies feature so that's why :p
<jgart>it's asking me to login
<phf-1>jgart, and now?
<phf-1>I've just done a `guix system reconfigure'
<jgart>yup, it's back!
<phf-1>jgart, It's full of horrors, but wanted something out now. Any inputs welcome. anyway, now trying to use `guix deploy' to manage that small server
<jgart>I'm glad someone made a TLDR on how to bootstrap a guix checkout and get started hacking ;)
<jgart>that's great!
<jgart>phf-1, might be nice to add a version of your blog post to the cookbook
<jgart>the way ambrevar had added his blogpost on using profiles
<podiki[m]>jpoiret: well thanks for the swap dependencies! I think it is a nice step
<jgart>phf-1, I think that would be very useful for new users to get started with a TLDR
<podiki[m]>jpoiret: and happy to help out with anything
<phf-1>jgart, yes, I think so. And in case of questions, it's a nice reference to have: "step ... did not work for me"
<jgart>"step ... did not work for me" what is step?
<jgart>phf-1, I got Guix running on alpine linux yesterday. I might make a TLDR install guide soon
<phf-1>jgart, "step" is a position in the TLDR.
<phf-1>like "in post ..., the `make' step did not work because ..."
<jgart>TLDR on the alpine TLDR guide:
<phf-1>haha great. I can publish that in 5min
<jgart>I forgot whether glibc-utf8-locales failed to be useful on alpine linux or whether I ended up installing glibc-locales in the end...
<jgart>will have to test that next time I'm on alpine
<jgart>phf-1, what static site generator are you using?
<phf-1>jgart, guix and emacs
<unmatched-paren>i like having a login screen, but there's none that:
<phf-1>jgart, guix and emacs and nginx
<phf-1>that's it
<unmatched-paren>- aren't basically associated with one of the big bloated desktop environments
<unmatched-paren>- are based on wayland
<phf-1>I can put the code out there if interested.
<phf-1>jgart, I add a few things and publish your TLDR. I'll be right back.
<jpoiret>unmatched-paren: ehrm, unfortunately that won't give you many results
<jpoiret>greetd + gtkgreet might do it for you, but it's not merged yet
<jpoiret>although you could help merging it by testing the patchset
<unmatched-paren>yeeeaahh... that's why i'd like to remove my dm until one appears (or i find the time/knowledge to make one...)
<jgart>phf-1, so you generated the html from org mode you mean?
<unmatched-paren>hahaha what: "swaplabel: /dev/nvme0n1p3: not a valid swap partition" <- this is my swap partition, sudo fdisk -l confirms it
<unmatched-paren>this is why my swap service stopped working i guess
<unmatched-paren>how should i replace the swap partition?
<phf-1>jgart, Yes.
<jpoiret>make sure you don't have anything important on it!
<unmatched-paren>well, it's certainly a swap partition, as i confirmed with fdisk
<unmatched-paren>but bizarrely it suddenly stopped working a while ago
<jpoiret>fdisk will not tell you if a partition is properly formatted, only the partition UID
<unmatched-paren>/dev/nvme0n1p3 645076992 652888063 7811072 3.7G Linux swap
<jpoiret>yes, the Linux swap here is only inferred because the partition is marked as such in the partition table
<jpoiret>that doesn't mean it's formatted as swap
<jpoiret>something may have corrupted it
<jpoiret>you can try to swapon manually and see what the kernel tells you
<unmatched-paren>i don't see any uuids or labels here, is that normal?:
<unmatched-paren>sudo swapon /dev/nvme0n1p3?
<jpoiret>lsblk doesn't output UUIDs for me either for btrfs fses, maybe it's a shortcoming of the command
<unmatched-paren>aaaaaah i know what happened
<unmatched-paren>so my ram is four times bigger and i tried to suspend
<unmatched-paren>s/bigger/bigger than my swap/
<jpoiret>oh yeah alright
<unmatched-paren>that did not go well :)
<unmatched-paren>swapon: /dev/nvme0n1p3: software suspend data detected. Rewriting the swap signature.
<unmatched-paren>ok i'll try reconfiguring now and see if shepherd restarts the swap service
<jpoiret>you can try to restart it now
<jpoiret>without reconfiguring
<jpoiret>it shouldn't change anything
<unmatched-paren>i need to reconfigure anyway -.o.-
<jpoiret>right :)
<unmatched-paren>so many gnome libraries being updated that i'd like to kill... gdm seems to be resisting its removal :)
<unmatched-paren>it's sentient
<unmatched-paren>the gnomes are rising up
<jpoiret>gnome can be pulled by a bunch of things unfortunately
<podiki[m]>would be nice to revive that lightdm patch series
<podiki[m]>it seemed pretty far along, just some details in the service definitions?
<phf-1>jgart, Here you got:
<unmatched-paren>guix system: warning: exception caught while executing 'start' on service 'swap-9e5a4ef1-6d5d-48b7-bbff-d26ec59dbc0b':
<unmatched-paren>In procedure swapon: "/dev/nvme0n1p3": Device or resource busy
<unmatched-paren>i guess that's only because i swapon'd it manually, right?
<unmatched-paren>maybe i should try to swapoff first
<unmatched-paren>woo! "shepherd: Service swap-9e5a4ef1-6d5d-48b7-bbff-d26ec59dbc0b has been started."
<unmatched-paren>thanks, jpoiret
<jpoiret>happy to help
<phf-1>jgart, What do you mean by "the way ambrevar had added his blogpost on using profiles" ?
<unmatched-paren>jpoiret: btw, i've now configured upower to shut down on low power instead of suspending, so this won't happen again
<unmatched-paren>emacs' auto save should prevent me from losing work
<jpoiret>why didn't suspending work?
<jpoiret>we don't have a very good resume from suspend-to-disk story right now i think
<unmatched-paren>because my ram is way bigger than my swap
<ngz>Aaah. Back.
<unmatched-paren>it made a horrific scratching noise when it tried to suspend and as we've just seen corrupted it
<rekado>phf-1: in the past we have repeatedly added articles to the cookbook that had previously been published as blog posts.
<phf-1>rekado, Ok, so how can I send you that post?
<rekado>phf-1: depends on how involved you’d like to be :) Ideally, we’d receive any addition to the cookbook as a git-formatted patch to the guix-cookbook.texi file.
<rekado>but if you think that this is too difficult to accomplish you can also send your article to with a subject that indicates that this is meant to be an addition to the cookbook.
<phf-1>rekado, Ok. Will try the texi file path first. Maybe org-mode can be exported to that somehow.
<ngz>phf-1: Org mode can export to Texinfo, indeed.
<unmatched-paren>M-x org-texinfo-export-to-texinfo
<unmatched-paren>is there a way to install a gcc cross-compiler for other architectures? i'd like to begin learning assembly for powerpc and riscv so i can help with porting things
<jgart>phf-1, ambrevar had originally made a blogpost on using guix profiles
<jgart>phf-1, then he re-published it as a cookbook entry
<unmatched-paren>and i'll do that by compiling random c files with gcc -S to output asm
<unmatched-paren>but obviously i need a cross-compiler to do that
<phf-1>rekado, How do I build the guix-cookbook.texi so that I can check locally if i'm doing ok?
<phf-1>There should be a way to have an HTML file out of it
<jpoiret>podiki[m]: see for some modifications to the examples
<jeko>Yooo Guixters !
<jeko>Had a great FOSDEM ?
<podiki[m]>jpoiret: nice! personally I'm okay with plenty of code, but would it make sense to not have the "filter" needed explicitly in the configuration? so (file-system-mount-point "/btrfs" file-systems) instead, where that function incorporates the filter?
<podiki[m]>or are you thinking for other uses for that predicate?
<jpoiret>that's what i thought initially too, but i think it's just as readable and lets you see filter in action
<jpoiret>so it's a better learning opportunity
<civodul>jeko: the "minimalistic & declarative" track was pretty awesome :-)
<jpoiret>i don't think one's particularly better than the other though :)
<jeko>civodul: not a surprise ^^
<podiki[m]>yeah, I think either way it is a nice convenience rather than needing to define a function just for that in the swap configuration
<podiki[m]>and the examples are more useful now
<jeko>pulling from my custom channel is taking ages…
<char[m]>I think I messed up my internet from config.scm. is there a way to system reconfigure without using internet?
<phf-1>rekado, jgart, patch sent for the cookbook. Not sure about copyright but we will see @ the review stage I guess.
<podiki[m]>char: just roll back (in grub) to when it worked?
<podiki[m]>or possibly change system generation and restart services?
<jgart>phf-1, awesome! THNX
<phf-1>jgart, This is the TLDR I use to send packages:
<phf-1>works well for small things.
<jgart>ah great
<jgart>phf-1, these are the ones I use:
<char[m]>podiki: it is definitely dns related. Previous generations do not work either
<phf-1>jgart, rare to see janet :) I meant the TLDR version of the Guix manual regarding the packaging guidelines; not just sending the patch.
<jgart>ha yeah, I use it regularly
<phf-1>Anyway, let's `guix deploy' things
<jgart>gpd 53816 -1 -v2
<jgart>example looks like above
<jgart>first argument is the ticket number to send to
<jgart>-1 and -v2 are git flags
<phf-1>jgart, why janet and not Bash ?
<jeko>It's been 10 minutes I'm stuck right after executing `guix pull`
<jeko>from my custom channel
<phf-1>jgart, PEG instead of regex... interesting :) anyway, off topic for this channel I guess.
<jeko>could be a hint but …
<civodul>jeko: "sudo guix processes" should show you the derivation(s) being built; which one is that?
<jgart>phf-1, why not janet?
<phf-1>haha, good answer ;)
<jgart>tbh, I just did it for the sake of the hack
<jgart>and too give me a reason to practice learning janet
<jgart>mini project = scripts
<jeko>civodul: /gnu/store/28k5yj520zwqx4vn8ijpgz1irq23dp3r-guile-wrapper/bin/guile --no-auto-compile /home/jeko/.config/guix/current/bin/guix pull
<jeko>civodul: now there is ClientCommand: /gnu/store/28k5yj520zwqx4vn8ijpgz1irq23dp3r-guile-wrapper/bin/guile --no-auto-compile /home/jeko/.config/guix/current/bin/guix shell
<civodul>jeko: under "LockHeld", you should see the .drv being built for the "guix pull" command above
<jeko>>`SessionPID: 6093 ClientPID: 6075 ClientCommand: /gnu/store/28k5yj520zwqx4vn8ijpgz1irq23dp3r-guile-wrapper/bin/guile --no-auto-compile /home/jeko/.config/guix/current/bin/guix pull`
<jeko>it's all I have
<rekado>gah, the GHC bootstrap looks pretty hopeless
<rekado>GHC is really determined to only be built with GHC
<rekado>even as far back as the first public release it used non-standard syntax
<rekado>tried building parts of GHC4 with nhc98 and it immediately gets confused by all the non-standard code.
<mark_>Is hugs still a thing?
<rekado>hugs is the most likely way to build nhc98 without generated C files
*mark_ really has limited haskell knowledge, and really no idea about bootstrapping it.
<rekado>I just wanted to know if it was worth the effort of building nhc98 with itself through hugs.
<jeko>civodul: I'll let guix pull working until I go to sleep haha. Hope I can hack tomorrow
<rekado>but it looks like nhc98 is of no use for the GHC bootstrap
<rekado>the way I see it we can only slightly improve on what we have now by building an early GHC with generated C files.
*mark_ finds
<mark_>and notes there is no part-2
<rekado>yes, I’m still working on part 2
<rekado>pretty sure that with some more effort I could get nhc98 to build with nhc98 on hugs.
<rekado>the problems then were due to the more recent GCC toolchain
<rekado>(I’m now using GCC 2.95 and glibc 2.2.5)
<rekado>I also took a detour and built the modified hugs in GHC 4, but it’s also kinda useless as it’s unable to interpret the GHC 4 compiler sources.
<rekado>I’ll soon push packages for GHC4 (without the compiler and hslib) and nhc98 (built from generated .hc files), so that others can play with it.
<civodul>rekado: going this far is quite an achievement!