IRC channel logs
2022-12-31.log
back to list of logs
<nckx>More than 11 days, apparently. <nckx>‘Adam Faiz wrote on 20 Dec 02:25 +0100 — It sounds like a good idea.’ <graywolf>Is there something like invoke but that would wrap system instead of system*? <nckx>(invoke "sh" "-c" string) <sner>graywolf: is there a reason you want to invoke `system`? Do you need shell interpretation for some reason? <sughosha>Is anyone currently able to see music files in gnome-music? For me it is just empty. <nckx>graywolf: I like it because it's more explicit than system, where I'm never sure which one is which. <graywolf>sner: I'm writing a wrapper script to automate installation of guix, and some things are easier in shell ("head -c 8b </dev/random >/some/file", ...) <nckx>Aw c'mon, that's a trivial fifteen-liner in Guile. <graywolf>I need to share some values between the (operating-system ...) declaration and the wrapper script, so having it in one language seemed like a good idea. <graywolf>It was originally shell, so there might be constructs like (format #f "[ \"$(lsblk -no path ~a | wc -l)\" -eq 1 ]" device)) <graywolf>I know I know, another trivial 15 lines in guile probably :D <nckx>I have a few server-side mcron jobs that I *still* need to port from shell. Since 2016. <mirai>does anyone here use mpd-service-type but on a headless guix system (i.e. no desktop-services or login managers) <mirai>I suspect that the service is broken when configured with pulseaudio output in headless systems. <mirai>since pulseaudio will start looking for a /run/user/<id> directory that won't exist <omlet[m]>Its possible replace gdm to ly lightdm ? <mirai>I assume you're using %desktop-services <omlet[m]>it's because I don't know what the hell it is to exchange a dm <mirai>in that case you will have to use the (modify-services) macro <mirai>add lightdm like any service <omlet[m]>mirai: what is the easiest forn to do this for inesperients? <mirai>I'll copy paste directly from a git diff <mirai>this is how I swapped gdm with sddm <omlet[m]>É porque neste monento acabei de criar uma conta matrix numa instância brasileira matrix <khronos_>Hi, I am trying to add a package to my config.scm file, but the example that is there in the package append line doesn't seem to allow me to specify anything other than what is there. <khronos_>(packages (append (list (specification->package "nss-certs")) <khronos_>I would like to add emacs and a few other packages to this command but the system doesn't allow me too. <khronos_>When reading the manual there seems to be a different syntax for this type of process. <nckx>(packages (append (list (specification->package "nss-certs") (specification->package "my-package") …) …)) <nckx>In practice, you'd write (map specification->package (list "nss-certs" "my-package" "foo" "bar" "yay" "efficient")) but it expands to the exact same thing as that. <khronos_>Can I have both keywords and add the map entry for my extra packages vs the original line that was put by the installer? <khronos_>I am just trying to change little things in the original /etc/config.scm file instead of making drastic changes with a system that I do not know all that well. <nckx>I'm not sure what you mean, but (packages (append (list (specification->package "nss-certs")) (map specification->package "my-package" "foo" "bar" "hmm" "okay") …)) certainly works too. <nckx>so you'd insert your map as a line in between the two existing ones. <nckx>It's not exactly idiomatic style but that can come later ☺ <mirai>(packages (append (map specifications->package '("pkg1" "pkg2" "pkg3")) %base-packages)) <mirai>will preserve the "original" config and is straightforward to add more packages as you go <nckx>I think they wanted to keep the nss-certs line as-was. <khronos_>It appers that none of these examples are working for me. <khronos_>I opied the lines and change the package names and the program does not take the config. <khronos_>I want to install screen lftp and emacs as global packages. <mirai>preserves that nss-certs line and adds an easy to modify line <nckx>khronos_: (packages (append (map specifications->package (list "nss-certs" "screen" "lftp" "emacs")) %base-packages)) <nckx>Seriously, don't bother trying to ‘keep’ lines unmodified, it's a recipe for pain. <nckx>Sorry, typo mine: specifications → specification. <mirai>didn't last more than 4 commits <nckx>Oh, there is a specifications->packages too. <nckx>khronos_: (packages (append (specifications->packages (list "nss-certs" "screen" "lftp" "emacs")) %base-packages)) <nckx>If you get errors, please share them, as well as your configuration (use paste.debian.net for more than 2-3 lines). <khronos_>I get a did you forget to use a use modules form error? <khronos_>I suspect my line parsing is not correct. <khronos_>Should all the text be on one line or two? <nckx>Lines have no meaning, only (((brackets))). <nckx>In extremis, a photo of your screen is fine if possible (and readable). <nckx>But first make sure your opening and closing brackets match our examples above. <nckx>An editor that highlights them somehow is not required, but beyond handy. <khronos_>I am using nano right now as this is the editor that is on the system. <nckx>OK, specificationS->package doesn't exist, that was a typo of mine. <nckx>Use ‘(specificationS->packageS (list "a" "b" "c"))’ — without map. <nckx>The ‘S’ is just for highlighting purposes, of course. <nckx>Because some unfunny bugger disabled bold in this channel. <nckx>ACTION trusts that it won't be abused. <khronos_>Well am still a little lost as I am blind so bold won't really work for me. <khronos_>So pasting the exact text might e most helpful for me. <khronos_>Right now I am using Debian in console mode ith Irssi and want to eventually use Guix for my main system once I get use to it. <nckx>(packages (append (specifications->packages (list "nss-certs" "screen" "lftp" "emacs")) %base-packages)) <nckx>I don't know how accessible Guix is for blind users. <khronos_>The espeakup package is there along with the kernel modules and espeak-ng as the audio output engine. <khronos_>My idea is to install espeakup and start it manually to get it to run and see if the system will read the console. <nckx>A new Guix subsystem maintainer 😉 <nckx>ACTION finds a lot of tyre-related products… <nckx>Their ‘Store’ has Google Chrome in it, so that at least would have to be disabled; but where the hell's the actual linglong source code… <nckx>(The Chinese version of their site works here, very strange.) <omlet[m]>nckx: Yes, i think the repo can desactivate <nckx>Am I an idiot or is there no Download/下载 button anywhere on the site? <nckx>ACTION notes that both can, indeed, be true. <omlet[m]>nckx: The bundle work in fhs other distros <omlet[m]>But for terminal apps or for install with terminal its with ll-cli <nckx>I can't say if it's free because I can't find the source code, and there's another project called linglong (with sources) muddying the search results. <nckx>But those ‘download’ links are for the individual ‘apps’, not the actual software, no? <nckx>Yes, it downloads a ‘.uab’ file which I don't want. <omlet[m]>nckx: Its actual, you can run the bundle softwsre in guix shell i think <omlet[m]>Its necessary the linglong tool or tools <nckx>Yep, I'm browsing it, thanks. <omlet[m]>nckx: The question its only about the repositories <nckx>Which question? Your original ‘Linglong is freedom software?’? <nckx>We'll only know if it's complete when someone tries to package and build it, but what I can see is indeed LGPL-3 and marked as such in the source files (and not merely in some vague README). That's good. <nckx>The ‘store’ is obviously nonfree, but maybe it could be mitigated somehow, I don't know. <omlet[m]>nckx: Yes the store its the sane in compararion with flathub.org <nckx>I'm not familiar with flathub.org either, so I can't say. <omlet[m]><nckx> "The ‘store’ is obviously nonfree..." <- For use fathub in guix i add the repository <omlet[m]>Yes flathub not is freedom, but flatoak only is the tool <omlet[m]>I only add the gnome repository kde elementary or more <omlet[m]>> <@omlet:matrix.org> For use fathub in guix i add the repository <omlet[m]>> Yes flathub not is freedom, but flatoak only is the tool <omlet[m]>> I only add the gnome repository kde elementary or more <omlet[m]>If linglong its the same, can package for guix <omlet[m]>But the true its, the bundle aare compatible with fhs <omlet[m]>> <@omlet:matrix.org> For use fathub in guix i add the repository <omlet[m]>> Yes flathub not is freedom, but flatoak only is the tool <omlet[m]>> I only add the gnome repository kde elementary or more <nckx>At first glance, yes, you should be able to package linglong for Guix with minor patching. <nckx>I'm not sure what you FHS point is, but Guix has rudimentary FHS emulation now that will only improve. <nckx>That is luckily not a requiment for inclusion in Guix ☺ (We prefer stable releases, but if there are none, there are none.) <omlet[m]>nckx: I speak with nix shell fhs its possible run linglong bundles <omlet[m]>nckx: launched half a year, so I can't tell if it's already complete stable <nckx>If I did nothing else in this life, I somehow convinced staff in #libera-newyears to spam Guix logo ASCII art. Twice. I can sleep in peace. <bdju>I have a syncthing shepherd user service set to run on login but since a while back it never returns me to my prompt, so I can't start sway unless I C-c out, which then stops syncthing. anyone run into this? <bdju>I set it up based on an old blogpost example <bdju>I guess I can C-z and then fg the shepherd process also <florhizome[m]><florhizome[m]> "nckx i add 'noresume' at the end..." <- nckx: this didn’t help. my btrfs seems to be really f*ed up, will take some time to restore data and try to repair stuff <PotentialUser-15>Does Guix offer something like Parabola's Iceweasel browser (a renamed version of the latest Firefox release)? I can only find Icecat and ungoogled-chromium, but the former is based on an older LTS version and comes with too many extensions and weird changes for my taste and I don't want to use a chromium-based browser. <rekado>(I misread this and thought you meant that *icecat* is a chromium-based browser) <rekado>icecat in Guix is probably the most up-to-date variant of icecat out there. <bjc>this may be an unpopular choice, but i run firefox from flatpak <rekado>bjc: you can also get firefox from nonguix. <panosalevro>first time i successfully installed guix and system is extremely slow. is that a video driver problem? <mirai>panosalevro: what kind of slow <mirai>and have you done a pull+upgrade+reconfigure after installing guix? <panosalevro>mirai: cursor is laggy, keyboard input has a delay of ~3 seconds, graphical applications take several seconds to load <jonsger>panosalevro: whih desktop environment are you using? <florhizome[m]><PotentialUser-15> "Does Guix offer something like..." <- I had a look at it once. It’s a very custom build but it should be similar to other Mozilla projects . Also they use some alternative programs (tauthon instead of python). So it would be a longer effort. <jgart[m]>We could package it but it would be a larger maintenance burden <jgart[m]>PotentialUser-15: If you want to take on that task/maintenance then go for it <jgart[m]>You can always start by putting it in a channel so that others have easy access to use it and then upstreaming it when ready <mirai>panosalevro: what are the system specs of this <panosalevro>i just pulled up htop and it seems that all my cpu cores are working at 100% <jgart[m]>has anyone tried using send-notification-command with libnotify from the earlyoom service? <panosalevro>it is sway that's eating up my cpu cores. do i need earlyroom? <jgart[m]>you would add sway as a regex in the field to ignore processes <jgart[m]>given that your display manager is not running <jgart[m]>panosalevro: what version of sway are you on? <nckx>Why would you want earlyoom at all, this is not memory-related. All it does is add CPU load. <nckx>It's likely that your AMD GPU doesn't support free software so your CPU is taking over. <jgart[m]>I've found earlyoom to be useful at times in general <jgart[m]>not that earlyoom is the solution to panosalevro:'s bug :) <KarlJoad>Does Guix use something other than DHClient for DHCP? My desktop sends its hostname to my router, but my CI server does not. The only difference I found that could explain that is that my desktop uses %desktop-services, whereas the CI server uses (modify-services %base-services ...). <Kolev>I'm happy to see Guix has Hebrew fonts! <KarlJoad>Nevermind. I skipped over the dhcp-client-service-type in the CI's config. <omlet[m]><Kolev> "I'm happy to see Guix has Hebrew..." <- I want learn hebrew for create my semitic conlanflg <omlet[m]>The semiric structure its better in my opinion <nckx>I don't know enough about it to understand why, but that sounds fascinating as heck. <KarlJoad>There must be a way to write a config file in /etc that I am just not seeing. I need to write /etc/dhclient.conf <KarlJoad>Yeah I found that... I just didn't read the manual closely enough the first few times through. <nckx>Note that there's a dhcp-client-configuration too. etc-service-type is a bit of an escape pod. <nckx>Is the name unfortunate? I'm not sure. I think I think it is. <KarlJoad>dhcp-client-configuration does not let me change dhclient.conf in any way. It does not have a config-file parameter or other fields I can hook into. So I have to manually write the config using a simple-service extending etc-service-type. <KarlJoad>Kind of. But I also only need one line in the config. I want dhclient to send the device's hostname when sending a DHCP request, so that the router's internal DNS allows me to resolve by the device's hostname, rather than need to remember IP addresses. <KarlJoad>Seems like Network Manager already does that by default, which is why my desktop did not need additional work. <nckx>ACTION runs NM on one of the servers here like an absolute madperson. <rekado>we usually arrange for escape hatches in service configurations <rekado>perhaps dhcp-client-configuration should also have one <KarlJoad>Yeah, I was surprised by how little customization dhcp-client-configuration had. <haugh>I find it really annoying to manage complex SSH configs with Guix <nckx>What does a complex SSH config look like? <nckx>Isn't that a client option? <nckx>I always forget that Guix Home exists. <haugh>Ah, I should have specified. <haugh>Yes, I'm currently just smashing everything into a single file <nckx>I wish Guix {,Home} services just wrapped the complete upstream syntax, then built optional simple-foo-configuration's on top of those. I've never warmed to most of the Guix services because they started off in the (when authors-thinly-disguised-preference? (display "7_lines_of_random_settings = true")) style. <nckx>[Slight exaggeration alert.] <nckx>s/optional/opinionated/ if you will. <haugh>Ha! I have the same feeling, but I understand that there's a tradeoff with simplicity/accessibility and also contributors are most likely to implement what they actually use. <haugh>I'm working on syntax that makes it cleaner/terser to generate line-oriented strings. I also need a better understanding of the store monad and how the "pseudo-fs" is instantiated. I think giving advanced users the ability to populate complex trees of config files without having to shell out or string-append will make a big difference in terms of confidence <nckx>I'm odd in that all my ‘Guix’ system configuration files are in the native syntax, but still Guix provides tremendous value in deploying them. And I can still programmatically generate them, which is nice when you're faced with YAML hell. <nckx>haugh: In #guix, we are all cool. <nckx>I like you. You flatter me. <nckx>ACTION needs to go party. Happy New Year, everybuddy. May all your Guix come true. <rekado>I always thought that maybe we could import existing augeas lenses to generate our Scheme DSLs. <rekado>bleh, I played myself: kodi 18.x had a failing test on aarch64; thought I’d update it to 19; noticed some bundled jars, replaced them and had to add groovy to the inputs; groovy fails to build for aarch64. <nckx>(The only Greek I speak is 2000 years old.) <mjw>nckx, haha, sorry, I assumed based on your top-level domain. I never learned old-Greek and thank you is the only thing I know in new-Greek. <nckx>Oh, you're not the first ☺ Those are just my initials, disappointingly. <mjw>that is funny, I should have actually checked your website. <nckx>I do know a few ‘new’ words, and being able to read the alphabet at speed is quite handy when travelling. <nckx>mjw: We met, if briefly :) <Guest67>there was a command to download the packages to the destination hard drive <mjw>Which tells me "I don't have a Web site." You are very confusing :) <Guest67>instead of to the hard drive running the live iso, anyone know what im talking about? <nckx>mjw: I am quite the paradoxical scamp! <nckx>Guest67: Do you mean ‘guix system init’? I'm not quite sure. <nckx>ACTION has to do some more social stuff. <mjw>nckx, this is very embarrassing. You mean, in person. Was this at guix days 2 years ago? <mjw>ACTION hates having such a bad memory. sorry <Guest67>guix fetches the packages but stores them on the current hard drive <Guest67>not the hard drive being installed to <Guest67>there was something u could do to change the download destination <nckx>Guest67: herd start cow-store /mnt <nckx>mjw: Inderdaad! But I don't think we talked much, it's all right :-) Was it only 2 years ago? It seems longer ago. Maybe this year? <nckx>ACTION has a worse memory, trust me. <Guest67>do you peeps use guix as a daily driver? im replacing my main arch system with guix <lilyp>both for personal use and work <nckx>ACTION included themselves in that oddly vague answer. <Guest67>i have used the guix system image command to create some isos <Guest67>but it seems they're not bootable by qemu <Guest67>they are bootable when i dd them onto a usb though <nckx>Hum, I'm not sure. Did you use -t iso9660 for guix and -cdrom for QEMU? <nckx>(They are ‘dual-personality’ but I've never tried to boot them as HDDs in QEMU.) <nckx>I use -cdrom a lot, it should work. <nckx>I don't know what the default is but I don't think it's that. <nckx>(The ‘iso9660’ output can actually be dd'd to a USB drive as well! It's a magic lie!) <nckx>I guess the non-iso9660 ones would work with different QEMU arguments, but I don't use those. <Guest67>is it possible to package my dotfiles as a guix package <nckx>You could. I don't know if it would be user-friendly. Maybe with git-checkout. <nckx>ACTION now really going to do those IRL social things before they are ostracised for being a total nerd. Keep my secret safe. <Guest67>is the package manager guix written in guile as well? so is it possible to add hooks to installations for example? like emacs <Guest67>sorry im asking lots of questions lol <nckx>The only part that isn't can be comfortably be regarded as a black box. Everything that's useful to hack is in Guile. <bjc>now guix is deep computer science <nckx>(A Guile daemon would be mildly useful in attracting new features/interest, but you'd never modify the daemon *as part of your system configuration*.) <lilyp>what do you mean by "the daemon" here? <lilyp>the currently running instance? <nckx>It's a nonsensical question. <lilyp>it's not, because guix already offers capability to alter how the daemon behaves at the system config level <lilyp>obviously limited capability, but still <nckx>You're confusing configuring with modifying. <lilyp>what counts as modifying in your definition? <KarlJoad>nckx: I have thought about rewriting the deamon-side code into Guile in the past... <nckx>Did you come across reepca's attempt? <nckx>Oh, it's in Guix.git proper. guile-daemon branch. <nckx>No idea how salvageable it is. <KarlJoad>I have no idea either. This is something I thought about doing but not something I have started yet. I need to look at the C++ code to figure out what is going on. <KarlJoad>Can the cuirass coordinator and cuirass workers in the new cuirass system be on the same machine? <wonko-the-sane>\∞/ happy orbit since the last arbitrary time offset, may your invocations of ((gεnεrατε-&-τεsτ good-enuf? mακε-βετα-lαmβδαs) Θκαyish-Θr-βαδlαm) return a value(s). ∞ blam da βlαm βlαm! βlαmdα! λ <wonko-the-sane>i may have woken up slightly drunk... less bad than a hangover i suppose <wonko-the-sane>when my less-wrong judgement prevails i release i dont need more, failing that i feel like Ford Prefect: "I need a strong drink and a peer group" <oriansj>wonko-the-sane: the peer group is here, the drink however is a different problem <devmsv>so I don't now if anyone is going to see this but... my guix system is broken since a lot of time. What I mean by this is that I have even use gnu/system/example/desktop.tmpl (change just the necessary bits) and I don't have any icons. In my system I usally use EXWM and also, no icons on the systray or in the apps <devmsv>It's really weird...And I think this is causing me other throubles with emacs-xwidgets reporting an error about GtkPixbuf and just closing emacs-no-x-toolkit works great <devmsv>ACTION tested with emacs from master WITHOUT xwidgets and emacs works <devmsv>Definetly the problem is with emacs-xwidgets: http://ix.io/4k5H can anyone test on their guix system? I know in foreign distro it works <ph03n1xaim>Hello, guix. I'm facing segmentation fault while running cargo run and cargo built with dependencies added to it. It was advised to have libgit2 on the mailing list. But currently on my system, I have openssl installed and it prevents me from installing libgit2. Anyone else facing this issue? I'd be really grateful if you could help me. I have been running guix for a few months but I think I need to still grasp on small things here and <ph03n1xaim>there. Would be glad to provide any additional information required to help me debug this issue. <ph03n1xaim>The way I did was use development environment for bat (cat written in rust) <Guest56>does it make sense that guix is compiling more than 3 versions of rust right now? and its taking ages? <Guest23>does it make sense that guix is compiling more than 3 different versions of rust?? <stikonas>presumably once could skip rust stages by not rebuilding version n with version n. Now it's using version n - 1 -> n -> n -> n +1 -> n+1. Might be quicker to do just n - 1 -> n -> n + 1 <nckx>It makes sense if you dissabled or otherwise expect substitutes. <Guest23>i didnt know firefox depended on rust lol <nckx>It's the entire reason Rust exists. <stikonas>Guest23: rust is invented by mozilla for use in firefox <nckx>But if you're using a channel, and the channel runs a substitute server (the only channel with FF that I know of does), make sure it's added and authorised. If you want. <nckx>Happy 2023 everyguix. May it bring you freedom in many ways. <efraim>ACTION adds a gratuitous 'or else' to the above <nckx>Freedom will be stringently enforced. <omlet[m]><robin> "feliĉan novjaron!" <- Forgejobthe esperanton word for communoty gitea fork <omlet[m]>* Forgejo the esperanton word for communoty gitea fork <Moritz64>Hi, I am trying to use guix to install packages that debian does not offer. However, in most cases (python modules) the package can't be installed using guix either as the package file is missing. What is the correct way to proceed for e.g. a pypi module? I am currently creating a package file myself, but is this the way it's meant to be? Is there <Moritz64>a standard place to store such custom scm files? <Moritz64>I don't see any messages (i.e. none at all, not only no answer to my question) - am I online? <nckx>sneek: Later tell Moritz64: You were on-line, but IRC is a bursty medium and keep in mind it that was new year's day… <jonsger>maybe 2023 is the year of the Guix System desktop :) <nckx>Those have been around for years. <jonsger>ACTION dreams of a Guix System live ISO/image :) <reza>Hi, I want to replace the syslog with rsyslog, to do so I cloned the syslog-service-type and call the the rsyslogd instead. however after reboot, I get stuck on login prompt. <reza>since I'm trying to modify the logging system itself and I possibly broke something during this modification. does anyone knows how I can trace about the possible failures of this modification? <mirai>must the package source be "uploaded" somewhere? <mirai>I'd like to use a local directory as the source uri for a package <mirai>(at least I'd like to test build it and try running it a bit before uploading it somewhere) <lilyp>you can also use --with-source <mirai>with file uri it complains: error: cannot run ssh: No such file or directory <mirai>well, with file:/// it starts picking up but <mirai>and it complains that it doesn't "appear to be a git repo" <mirai>even though it can be cloned fine with git clone? <devmsv>so I'm still facing this problem with emacs-xwidgets. Even running it with guix shell container complains about the gtkpixbuf <Guest5481>isnt it ironic how guix is all about privacy yet we dont have a torbrowser package? <Guest5481>is this not an essential piece of software :( <bjc>the issue is firefox, not tor. guix does have tor <nckx>Guix is not all about privacy. <rekado>bjc: Guix also has firefox, just not in the main channel. <mirai>incidentally that will include privacy but they're not equivalent <bjc>rekado: if the channel it exists in is called "non-guix" it's safe to say it's not in guix ;) <mirai>but firefox has trademark restrictions <rekado>bjc: I wrote “Guix has”, not “it’s *in* Guix’ <mirai>you can't just build it from source and distribute it as firefox <nckx>You can't do that with any software without explicit permission. <bjc>rekado: same logic appies: non-guix means, literally, "not guix" <ecbrown>i would argue that "guix has" and "its in guix" are virtually the same <ecbrown>firefox is a stone's throw away from kosher guix <nckx>Any software with a trademark, anyway, and a trademark doesn't need to be registered to exist. <mirai>nckx: that's the basis for iceweasel or icecat or ???? (name != firefox) <nckx>Let's leave the anti-Firefox FUD for #fsf or wherever it festers. <jonsger>ACTION still hasn't received any letter from Mozillas lawyers, even after distributing Firefox for a year or so :) <Guest5481>well jeez we have a firefox package but not a torbrowser package <nckx>Wait till they find out that GNU is a trademark, without the written relaxations that Mozilla gets bullied for. <mirai>torbrowser should(?) be fine if you can build it from source <mirai>its not named firefox for one <nckx>They don't want you to, but you should be able to. <bjc>i believe the issue is the mpl, specifically. you can't distribute a modified version of firefox and call it firefox. that's why iceweasel exists, since it's modified firefox <nckx>They prefer you run the binary. <bjc>guix doesn't modify rust, so rust is distributable as rust <rekado>did you know that the TeX license also forbids modification when distributing as TeX? <nckx>Do we not modify Rust? I'm surprised, but it's more of a coincidence than anything. <nckx>Nothing to do with trademarks. <mirai>rekado: you have to call it something else yes, <bjc>firefox has freedom issues™, like playing drm media, which is antithetical to gnu, and thus needs to be stripped <lilyp>ACTION quickly deletes rustc-1.54.0-src.patch <mirai>bjc: those can be stripped (well, if we want to from source its not like we can build those anyway) <twopubsolar[m]>there is a package torbrowser-launcher, which downloads the binary at startup, but it's not in guix <mirai>*to build firefox from source <bjc>but if you strip them, you can't call it firefox, which is why guix has iceweasel <nckx>bjc: Which part of the MPL? I'm not as familiar with it as with the GPL. <bjc>nckx: i'm not sure. this is what i've read previously around rust licensing, so i may be off base <rekado>there is no iceweasel package for Guix. <nckx>Yeah, it's like you can't modify Guix and call it Guix if the Guix project objects to your use of the name. <nckx>This, at least, has nothing to do with the MPL or Mozilla. <lilyp>I mean, you probably can distribute Microsoft Guix ;) <bjc>i think it's specifically in the mpl. i'll see if i can dig up the issue on the rust tracker <ecbrown>nckx: sounds like trademark. i thought you said its not trademark. ? <nckx>ecbrown: What's not a trademark? <ecbrown>that guix proper can object to guix knock-offs <mirai>nckx: if not trademark then I don't know what it is, its just that they get fussy if you distribute your builds and call it "firefox" <ConvolutedSquare>I would like to have my Guix server behave as a kind of caching substitute proxy for my LAN, so that other machines get substitutes through it and it caches them. How would I go about it? And to have Guix pull from that server, I assume I just need to set up a git mirror? <nckx>If you have a trademark, you can object to usage that you, well, object to. <rekado>ConvolutedSquare: you can run ‘guix publish’ on your server <ecbrown>ConvolutedSquare: "substitue server" <nckx>There seem to be some weird but unstaded definitions of ‘trademark’ floating around here that confuse me. <rekado>ConvolutedSquare: I don’t think there’s a preconfigured way to set up a substitute proxy, though you could have your other machines use the server as an offloading machine <ConvolutedSquare>I want all substitute requests to go through the server. So if the server doesn't have a substitute, it will fetch it and serve to the client. <rekado>i.e. all builds get offloaded to the server, which is then free to fetch them or build them <ConvolutedSquare>Ah, I didn't realize it can fetch substitutes for offloaded builds. I thought that meant it would always build them from source. <lilyp>I'm p sure the Linux trademark only exists so that you can't sue them over washing powder <nckx>Why it exists is not relevant to your legal options. <rekado>ConvolutedSquare: you can also configure the daemon to perform 0 builds <rekado>ConvolutedSquare: not sure if this affects downloads as well, thoughh <rekado>but yes, if substitutes are set up they will be downloaded <nckx>To your local Guix checkout. <ecbrown>Guest5481: not to where guix pull downloads to. you can apply patches to a git clone of the guix source <bjc>nckx: i can't find the issue on github, but other reading indicates you're right: it's a bog standard trademark issue, and mozilla just happens to be fairly protective of it <ecbrown>then ./pre-inst-env (i recommend reading the manual, knowing about ./pre-inst-env etc.) <Guest5481>hmmmmm so i need to maintain my own version of guix source code? <bjc>if you want to patch guix with your additions, you need a copy of the source code to do it. if you want to just add some packages for private use, you can create your own channel for that <lilyp>nckx: Also source on "GNU is a trademark" <nckx>bjc: I haven't seen any evidence of them being protective of it, though. That's also far too vague a reason to bar them from a distro, akin to ‘the dev was mean to me on an ML once’. <bjc>nckx: wikipedia has some history on the icecat browser. there's artwork that's problematic, apparently <singpolyma>Every project name is a trademark, but most are unregistered and undefended, so not legally relevant <mirai>so unless tor browser also suffers from some trademark sensitivities, it is fine for guix <nckx>Hehe, why did my mind go straight to furry artwork. <bjc>you've been on the internet too long <nckx>‘We don't want to be trademark bullies, but we will use trademark to protect the public's ability to recognize Tor Project software. Trademark law helps us to assure that the name is used only in connection with genuine Tor anonymity software and for accurate description of software and services.’ <nckx>Bog standard stuff, just like Firefox. <Guest5481>so after i patch guix source code, how would i go about installing only a specific package of it? <nckx>Build it, then ./pre-inst-env guix install PACKAGE <Guest5481>yes but there is no standalone package its a patch <nckx>twopubsolar[m]: You could do a hacky with search-paths, but it's not what manifests are for. <nckx>ACTION knows there's a patch for setting arbitrary search-paths, but not if it's been merged yet. <mirai>heh, then if tor browser is to be packaged in guix, it will have to be called something else, say scallion-browser <mirai>ecbrown: that's if Tor Project has the same kind of reaction that Mozilla has with firefox <singpolyma>TBB probably still has add-ons search so violates free distribution <singpolyma>mirai: Mozilla doesn't have that reaction, it's a myth <mirai>singpolyma: those can be stripped? <bjc>in fact, it's the only reason i install flatpak <mirai>singpolyma: I don't know, from what I know it's the reason why iceweasel came to be many $SOLs ago <singpolyma>Iceweasel was a misunderstanding and has been reversed <nckx>Any misunderstandings that remain are very deliberately kept alive. <nckx>(Not referring to y'all here.) <nckx>It is not fair to blame Mozilla for this. <mirai>it doesn't seem so (Open Source Project Guidelines / You Must NOT:) <nckx>Yeah, don't modify Firefox and call it Firefox. <nckx>Not making X build on Y. <mirai>> give the impression your software is actually published by, affiliated with, or endorsed by Mozilla. <nckx>A Guix Firefox distro package would not be. <mirai>building from source and distributing it seems to fall into the 'publish' category <mirai>if they're 100% reproducible I guess its a non-issue <mirai>unless you want to argue "this bit is different than my bit" <nckx>Doesn't need to be, but of course that would be awesome for other reasons. <singpolyma>Reading trademark policies too strictly is a popular passtime, but Mozilla has even explicitly said "no, Debian, you can call it Firefox of course don't be silly" <nckx>Binary hash has no bearing on the trademark discussion though. <singpolyma>It's unfortunate that they were forced to do that, but such is life <ecbrown>singpolyma: did mozilla's permission extend to debian derivatives? <nckx>‘Don't be silly’ is such a hard guideline for some folks to follow. <singpolyma>We have a similar problem in XMPP space where some people are afraid to say "Jabber" because Cisco hrs a registered trademark. A trademark they have not enforced against the community in 20 year and they actively support the tech etc, but some people like to be afraid by hypotheticals they think they understand <singpolyma>ecbrown: it's not special permission for Debian. It's just clarity that the trademark policy has never applied to what Debian does <mirai>the legal equivalent of "don't be silly" can vary between the legalese-fluent and the common folk (it even varies among themselves) <nckx>mirai: True, but it's not really the issue here. There is no grey area or danger zone for distributions. Just a very vocal set of people determined to spread misinformation. And I have no idea why, but it's a thing. <ecbrown>singpolyma: how long did that ambiguity last? years, if i recall <nckx>Which makes me wonder: any legitimate reasons FF is not in Guix? Is it too hard to disable the store/whatever it's called? <nckx>Presumably, if IceCat exists, it can't be. <singpolyma>nckx: I'm not sure if disabling the store and drm downloader is a material change or not <nckx>Ah, now that's an excellent point. <singpolyma>I think it's just deemeed not worth it since icecat exists? <mirai>I don't blame people being afraid for hypotheticals, the fact is that the thing they're afraid of CAN materialise (and has happened). This can be meaningfully and definitively fixed if the part "with the stick" eliminates the ambiguity <nckx>Trademark FUD aside, that's still true. <bjc>by "store" do you mean the extensions/themes download page? <ecbrown>a "firefox" that does not play drm is no firefox. can't watch netflix <singpolyma>mirai: you can always be sued. Following some imagined "letter of the law" doesn't protect you <bjc>why would guix care about access to the store, then? because it may provide non-free software? <nckx>‘Eww, Mozilla is eevil because they will sue you if you call your build Firefox!!’ ‘Ehm, actually, we asked them (why didn't you?) and they say it's fine.’ ‘Uh… No, our point was that they are eevil… for… reasons! So you can't trust what they say!’ <nckx>Motivated reasoning is one hell of a drug. <bjc>that's a shame, because the extensions are the only reason i install firefox at all, even through flatpak, rather than use iceweasel <mirai>singpolyma: non-sequitur, otherwise you may just have the judge be a roulette wheel <singpolyma>bjc: icecat can use all the same extensions, it's the same codebase <nckx>ACTION builds their own FF, with disabled DRM and telemetry and hookers, called ‘Firefox’ because they like the logo, but doesn't distribute it. <singpolyma>mirai: once you see a judge you have already been sued :) <bjc>singpolyma: the extensions page link in settings brings you to an empty gnu project page <nckx>mirai: It's a pretty sane statement, whether you like it or not. <ecbrown>i'm glad that they've been fastidious about (TM) and the like. otherwise, my $work would still be installing SCO Unix <bjc>it may be able to use them, but it gets in the way <singpolyma>bjc: sure but you can use the Mozilla add-ons website <nckx>You can be sued for anything. You might be sure to win, and then you'll only be much poorer and 2 years later. <bjc>at the point where the browser would rather show me nothing than something useful, i feel like it's trying to tell me that my needs aren't welcome and i should be using something else <mirai>nckx: the legal departments (or the equivalent for communities) do have to do something in order to justify their existences (and expenditures) <nckx>Heh. No disagreement. Alas, dislike for legal depts doesn't mean the law is a computer programme that you can run to completion beforehand and debug at home. <bjc>nckx: tell that to the blockchain!!1! <bjc>too late, it's already declared that statement illegal <nckx>Code is la—wait, that's my money, give it back. <bjc>it did go well. very well. for certain people, at least <mirai>bjc: malformed input perhaps <bjc>it felt very normal to me <nckx>Need finger-quote markup in IRCv3. <bjc>i'm sure matrix has it <mirai>Is there not a unicode character for it <mirai>given that it has lots of stuff <nckx>I wanted to use the peace sign but it is not good. <oriansj>anyone else having an issue with emacs magit-mode failing to prompt for password when signing a commit? <nckx>[guess] Missing pinentry? (In gpg-agent.conf?) <nckx>I don't know if it still is, but you used to have to specify the full file name (I still do). <oriansj>currently have: pinentry-program /home/oriansj/.guix-profile/bin/pinentry-qt in there; so it should be able to find it <apteryx>nckx: supposed to not be needed if pinentry is in your user profile <oriansj>well it is both in my profile and it was all working yesterday (without any updates done nor changes to my configuration) <viivien>How does qa pick the order of issues to check? At https://qa.guix.gnu.org/patches, I see some month-old issues that have not been picked up yet, and some very recent issues that are already green. <nckx>apteryx: Good to know! Mine (both gpg and pinentry-gtk-2) are in my system profile, I guess that explains it. Or used to. <nckx>I'm used to ‘same profile’ requirements but not ‘user profile’ ones. <nckx>Submitter is best placed to do so. <cbaines>viivien, builds are submitted for the 150 issues where patches have most recently been sent <cbaines>there are multiple reasons why an issue might show up with an "unknown" (grey) status, and unfortunately that isn't displayed very well at the moment <cbaines>it could be anything from the patches not applying, to the data service failing to process the revision, to the builds not yet being complete <viivien>I think my issue is stuck in the gray state because it reads: Yet to process revivion, job failed with a link that says the revision failed. <cbaines>viivien, right, next step is to probably check the log linked to from that Guix Data Service page and try and work out why it failed <cbaines>viivien, could you share the issue number? <viivien>The log is about postgresql doing its thing and boom you found a bug :) <viivien>Oh no that’s not the postgresql thing <viivien>I’m mixing up things in my head sorry <viivien>Can I download /gnu/store/s43d6fgnhxhdsy3yyqvl8z960rc4n0gw-compute-guix-derivation.drv on the server? I’d like to check that derivation. <viivien>(well, download the derivation and its inputs) <viivien>It’s not in a machine I can query for substitutes <twopubsolar[m]>viivien: if it's on another machine, you can use `guix archive` command <nckx>Very much worth the one-time key exchange. <cbaines>viivien, unfortunately substitutes aren't available for that derivation. They are meant to be for the channel instance derivations, but the revision loading process doesn't get that far <twopubsolar[m]>the translations got merged, then somehow rolled back, and it also caused them to roll back in weblate? <cbaines>viivien, I can reproduce the failure locally, so I think this is actually picking up an issue with the patches <viivien>cbaines, how do you reproduce the failure? <cbaines>(there's also the issue that the data service doesn't flag the error well, and fails trying to insert 0 records, but that's tangential) <cbaines>I think there's an similar guix pull thing in contributing.texi: guix pull --url=/path/to/your/checkout --profile=/tmp/guix.master <cbaines>I'm not sure what the difference is, but I think both of those should be similar to what the data service is trying (and failing) to do <viivien>make as-derivation fails with: no code for module (guix) but it’s the same error on top of origin/master <cbaines>viivien, hmm, I think I get past the failure by removing #:use-module (guix git) from gnu/packages/build-tools.scm <cbaines>I think there may be some restrictions on using (guix ...) modules from (gnu packages ...) <cbaines>viivien, is the (guix git) module needed there? I don't think I can see any use of it in the build-tools module <viivien>I think it is a remnant of the previous iterations of the patch <viivien>I will try without it but I just ran "make clean" so I have to wait until guix is rebuilt :) <atka>hello guix, is anyone running rootless podman containers on guix these days? I see the podman package has received some love recently. It was broken last time I tried <viivien>cbaines, thank you for your help :) I pushed a new version of the patch series, but Guix QA did not update its revision so it’s still marked as failed. Is it possible to re-run it with the v11 patch? Anyway, I managed to run make as-derivation on my end (I had to pass a couple of -L args to the guile process for as-derivation) so I guess the code is good enough. <cbaines>viivien, when you say you "pushed", what do you mean? I don't see new patches being sent to the issue <cbaines>if you do send some new patches, they should be used and tested <viivien>But if I click "view Git branch" on the QA page, it shows the commits from an earlier version <viivien>And the message "Yet to process revivion, job failed" is still there, with the link pointing to the old log <cbaines>viivien, right, Patchwork is used to take the emails understand what's going on, but here it seems very confused <cbaines>it doesn't seem to be joining up the patches that should be in the same series <cbaines>viivien, are you able to use git send-email to send patches? <cbaines>OK, does that handle sending the emails? <viivien>No, I use my email client to send them <cbaines>viivien, right. I don't know how to send patches without git send-email, so unfortunately I can't really help here. <cbaines>viivien, if you want to try replicating what it does, I'd maybe suggest making sure to send a cover letter when a patch series has more than one patch, and sending the patches as replies to the cover letter <cbaines>that might help, but I really don't know <viivien>I have an option to create a cover letter in magit <viivien>I think I messed up my git configuration because all commits have a commit date of Mon Sep 17 00:00:00 2001 <viivien>Maybe that’s why patchwork can’t group them but that’s a problem on my side <Kolev>I still don't know how to use Magit. I use Git from the command line. <viivien>Oooh if I choose the "deep" thread style it creates message IDs and each commit becomes a reply to the previous commit, and the first is a reply to the cover letter. <morganw>> I think I messed up my git configuration because all commits have a commit date of Mon Sep 17 00:00:00 2001 <morganw>I think is just what git format-patch does to indicate that the content is from git format-patch, so it isn't a configuration editor <viivien>Right, the commit log shows reasonable dates <unmatched-paren>happy 2023 guix :) i'm still alive, i just don't have anything to contribute at the moment. <Kolev>unmatched-paren: Are you ( on the mailing lists? <florhizome[m]>unmatched-paren happy new year :) I guess you still don’t have commit access? <Kolev>unmatched-paren: Happy GNU year! <oriansj>well I solved my magit issue: my gnupg key expired last night <oriansj>I knew I forgot something last night <ham5urg>When config.scm is reconfigured, how is Guix elaborating how to go from the current state to 'none' to new. Wouldn't be a 'destructor/constructor' needed? <viivien>That’s a half success: patchwork picked up the v11, but unfortunately it only picked the last commit of the series. <jman>I can't figure out the syntax of home-shell-profile-configuration by reading the documentation. I get an error "source expression failed to match any pattern" when reconfiguring the home <jman>Does anyone have a usage example? What is the format of its "profile" parameter? <viivien>Looks like it worked now! I set the thread style to "shallow" (C-m s s) and added a cover letter (C-m l) and Patchwork now recognizes that there are 4 commits. Next time I will be sure to do that from the start. <viivien>The git branch does not seem to recover as well though. <omlet[m]>Sorry but gnuinos.org (devian fork) its is gnu project? <omlet[m]><rekado> "omlet: it’s not listed here..." <- But the name its gnuios <omlet[m]>odd, plrque still above uses (observed through in the script) the domain fsf.org in javascript <nckx>But note ‘freenode’, so either that August came and went without result or they are still seriously using freenode, which is not the best sign. <viivien>ACTION pictures the world in a few years when we detect unmaintained projects by their "follow us on Twitter" buttons <nckx>I think Guix HPC's presence was officially ended. <mekeor[m]>hello. how to install a minimal texlive? texlive-base seems to include many languages that i do not need. <mekeor[m]>(i only need english and german at the moment) <jgart[m]>mekeor: That would be a great thing to work on <jgart[m]>mekeor: See that manual section. Search for > In the future <jgart[m]>We need texlive collection packages that are well tested to work <viivien>From my very limited experience with texlive, there are a few obscure error messages that mean "you don’t have a big enough texlive installation", so I always have the largest texlive I can :) <jgart[m]>I also get frustrated when I install texlive packages and then I'm missing something <jgart[m]>I want a minimal install but then it become a goose hunt to find what you need <jgart[m]>Having smaller texlive collection packages that are known/tested to work without installing GBs worth of stuff would be cool <jgart[m]>void linux's philosophy is "just use tlmgr" and install what you need but I think we should collect and automate this knowledge for user convenience <jgart[m]>texlive dependency issues are a problem for a lot of people that guix can solve easily but it would require some research and making new packages for that as stated in the guix documentation <apteryx>omlet[m]: there are as many roadmaps as there are contributors to Guix :-) <viivien>On my roadmap, there is a minetest server service, emscripten, and the whole of NPM :) But there is so much easy and useful stuff to add to guix that these get pushed off the list. <vagrantc>omlet[m]: i doubt the wifi will work on linux-libre <KarlJoad>Am I missing something with the zabbix-server-service-type? I have a database password set, but when I try to start the server I get the error "database is down" "connection failed" "FATAL: password authentication failed for user "zabbix". <KarlJoad>I did manually set the password using "passwd zabbix", but I am still not getting it to work. <johnabs[m]>Hey Guixers, I'm trying to work with inkscape, but it keeps crashing when I try to import png images. Is this reproducible for anyone? <johnabs[m]>I checked the bug tracker, but I couldn't find any bugs that weren't already fixed. <johnabs[m]>...no joke, try to reproduce it with the dark theme on <johnabs[m]>I just turned the dark theme off...and it worked. <johnabs[m]>Using the "Colorful" Appearance, with Default Canvase, and default keyboard <johnabs[m]>When you open it, it should prompt you. At least that's what it does for me <johnabs[m]>Also, just a sec, it may be file picking instead of the theme. <apteryx>ah, Interface -> Theming (there was a search option) <johnabs[m]>Ah, actually, it's the "open-file" dialogue box that seems to cause the issue <johnabs[m]>I get this error though: (org.inkscape.Inkscape:27857): dconf-WARNING **: 19:51:04.946: failed to commit changes to dconf: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name ca.desrt.dconf was not provided by any .service files <johnabs[m]>Yeah, so if I "Open" it fails, but it adds the file to my "Open Recent" list. When I open directly from there, it works fine? <johnabs[m]>But when it opens a file browser to select the file, and I click open, it "encouters an error and needs to close". <apteryx>sounds like some GTK thing is not wrapped <apteryx>ACTION tries 'guix shell --pure inkskape' <omlet[m]><vagrantc> "omlet: i doubt the wifi will..." <- In pinetab 2 <Kolev>How do I print out a package definition? <pkill9>Kolev: `guix edit <package name> <pkill9>well, that will show you the package definition in default text editor <Kolev>pkill9: It says the file is write-protected and then shuts my editor down. <apteryx>perhaps 'EDIT=cat guix edit <package name>' then <apteryx>rather, 'EDITOR=cat guix edit <package name>' <apteryx>'EDITOR=less guix edit hello' works for me <apteryx>or if you intend to modify the package, ./pre-inst-env guix edit <package>, ran from a built guix checkout <Kolev>I'm trying to figure out where the fonts Keter and Frank Reuhl CLM came from. They are not packages. <pkill9>Kolev: you're not supposed to actualy edit it <pkill9>because that guix is from the store <pkill9>either do what apteryx suggest, or just clone the repo and find the pakcage definition in theere <oriansj>pkill9: well if we are not supposed to actually edit; then it probably shouldn't allow editing <podiki[m]1>it is handy to open the file and browse it in your favorite editor ( I use it all the time) <podiki[m]1>or as mentioned, in a guix checkout where you can indeed edit and save <Kolev>pkill9: Oh no, I just wanted to read it, and that helped. I wanted to see if SIL Ezra had Frank Ruehl CLM. <akirakyle>save it as guix.scm and build it first by running guix build -c 1 -f guix.scm <akirakyle>then you can enter a shell with guix shell --no-grafts --pure guile gtk@4 -f guix.scm which can run the examples <whereiseveryone>g-golf/guix.scm:21:0: error: #{ }#: unbound variable hint: Did you forget a `use-modules' form? <akirakyle>hmmm, that's weird, you sure you copied the pastebin correctly? <akirakyle>Oh it looks like there's some invisible unicode character on line 21 what guile doesn't like <daviid>akirakyle: nothing imprtnt, but when you paste using centos paste service, you can select the language 'Scheme', it helps us to read the code ... <akirakyle>whereiseveryone: it looks like this is NO-BREAK SPACE <akirakyle>yes, something between your cut and paste added it <akirakyle>or change the shebang to #!/usr/bin/env -S guix shell --no-grafts --pure guile gtk@4 -f /path/to/guix.scm -- guile -e main <whereiseveryone>hmmm works now but I had also tried that before and it didn't I might have done something wrong before <akirakyle>fwiw I don't get a segfault on hello-world.scm <daviid>akirakyle: could you try all other examples, please <akirakyle>I'm on c4025af8c11c3e9ea0c2747b6c475c916fb61d80 <whereiseveryone>I have also ran this: ./configure --prefix=/var --with-guile-site=yes <akirakyle>since this installed g-golf to the store, it should be irrelevant what you did in the repo <akirakyle>well I'd update my guix to your commit to debug, but unfortunately since I'm on aarch64, there's currently no binary substitutes available for linux so it'd take a day for me to update <daviid>whereiseveryone: these are guix specific steps, so not appropriate for any other distro .. <akirakyle>whereiseveryone: hopefully I'll get my x86-64 machine up and running again sometime soon, then I can see if I can reproduce on that commit <akirakyle>whereiseveryone, yes several months older since that's the last commit with a linux binary substitute available for aarch64 <whereiseveryone>daviid: guix is supposed to work on any distro. it's packaged in Debian, for example <akirakyle>I wish the guix build farm had some more/better configured aarch64 machines, since they currently can't keep up <daviid>whereiseveryone: debian users don't need to install guix to run g-golf ... those steps are only applicable to guix users - no pun, no fight, just explaining to you, the GNU G-Golf upstream package uses the gnu build system, and that doesn not need any specific steps ... <daviid>whereiseveryone: those steps should be included in the g-golf guix package and or g-golf guix 'specific' doc <akirakyle>whereiseveryone and daviid I think that the make issue which requires the -c 1 (which causes make -j1) should be fixed in g-golf, then the package should be updated in guix, and a comment should be added to the package description about requiring --no-grafts <akirakyle>the rest of those commands are just "generic guix" <whereiseveryone>daviid: which is the g-golf guix 'specific' doc that you're referring to? <akirakyle>maybe the thing to do would be for me or someone else interested in this to write up and post somewhere a "getting started with g-golf in guix guide", although that should probably wait until more of these bugs have been ironed out <whereiseveryone>you mean that that info should be in the guix manual or somewhere else? <daviid>akirakyle: right, a g-golf receipe for guix users might be the best solution <whereiseveryone>It would be awesome to have a tested getting started guide with guix that is well tested <daviid>akirakyle: i'll look into '... the -c 1 (which causes make -j1) <Kolev>I'd love some G-Golf apps to read. <Kolev>GNOME apps written with G-Golf. <daviid>Kolev: not many but there are a few examples in g-golf <Kolev>I just hope GNOME Builder can handle Scheme projects. <akirakyle>I'll tentatively say that I'll write up some guide at some point, contingent on getting myself up to speed on how to write such apps :) <daviid>and you may grab the gtk4-demo and the libadwaita demo and start to write some using g-golf ... remember though g-golf still is an alphapackage, be readfy to hit bugs <akirakyle>I know us guix users desire more scheme apps :) <Kolev>whereiseveryone: GNOME Builder is not just an IDE. It builds/runs the app, too. <akirakyle>Kolev: why GNOME Builder when we have emacs ;) jk <Kolev>Press Play and the app magically builds. <Kolev>akirakyle: GNOME Builder makes it accessible to non-Guix users. <Kolev>whereiseveryone: Yeah, Guix GUI. <daviid>all link you to their respective source code ... <Kolev>A full-featured GNOME HIG app written in Guile would be a dream... <Kolev>whereiseveryone: Human Interface Guidelines. The specification for how GNOME apps behave. <Kolev>Not just a GTK app, but a GNOME app. <Kolev>With headerbars and all that. <Kolev>I could never get into Tilix. <Kolev>It was nice, back when Terminal did not follow the HIG. <cronoStar>Hey im working on getting a big emacs-guix bug out of the way. I wanting to know if anyone knows how guile with a socket works. <cronoStar>Its basically just guile in a socket, but is there any non-standard (repl) behavior I should be aware of? Its not reading gexp in the #~ form, and want to see if guile (socketed) can read gexp in that form. <nckx>whereiseveryone: Just idly scrollin' by: /var is --localstatedir, not --prefix! <ham5urg>How do I uninstall the packages which were installed via a manifest file? Is there a way which invokes the manifest file? <zacchae[m]>ham5urg: Like, you want to install eveything from the manifest except one package? I think the way to do that is to modify the manifest <ennoausberlin>ham5urg: I am not aware of a way removing packages which are given by manifest declaration. <ham5urg>I see, thanks guys. Is there a way to group desired packages? I would like to install/remove groups of packages. <ennoausberlin>ham5urg: You can have multiple manifest files. And install them at the same time. Just use guix package -m file1 -m file2 etc <zacchae[m]>When you install via a manifest, it will automatically remove any packages NOT in the manifest, so no need to remove existing packages <ham5urg>Is it possible to have 'multiple' .guix-profile/ per user? E.g. 'guix-profile/' as default and a 'guix-profile-namespace-x'? <florhizome[m]>If you use guix home you will have another profile automatically <rekado>mekeor[m]: you cannot do without the stuff that is in texlive-base <rekado>you may not care for the hyphenation patterns, but they will become inescapable necessities in the future <rekado>not only are they a requirement for a minimal latex installation, they also determine features of the different *tex binaries <rekado>currently we actually have a problem with our latex binaries, because they were not built in an environment where these hyphenation patterns were available <rekado>I’m going to work on this in a separate branch; the result will be another build cycle to rebuild all ‘formats’ once they have been used to build babel and default hyphenation patterns <mekeor[m]>that might explain some issues i ran into, while using latex yesterday <ham5urg>Here https://paste.debian.net/1265932/ I try a "guix home reconfigure ~/manifest/default/main.scm" and next to it I have a "~/manifest/default/files/init.vim" which I want to copy to ".config/vim/init.vim" but I get an "invalid character `/' in name". What am I doing wrong? <rekado>ham5urg: plain-file doesn’t allow /. <rekado>that’s for the file name in the store <rekado>openjdk9 only builds on aarch64 with GCC 7, not GCC 10 <rekado>it builds fine with a package transformation, but sadly I haven’t been able to apply that transformation in code <rekado>the first problem is that the gcc-toolchain are available in commencement only, and we shouldn’t import that module if it can be avoided <sughosha>ham5urg: Just in case if you don't know yet,`home-bash-configuration` also has `environment-variables` property. Eg. `(environment-variables '(("ENV" . "value")))` <rekado>using just an alist `(("gcc" ,gcc-7)) as the toolchain argument to build-system-with-c-toolchain caused a variable lookup error on ‘guix pull’ <rekado>sughosha: I looked at the patch; could you add a comment why tests are disabled? <rekado>I also recommend using a gexp for the phases <rekado>(arguments (list #:phases #~(modify-phases … ))) <rekado>then you can use #$output instead of binding the ‘outputs’ keyword argument and looking up the output with label ‘out’. <rekado>sughosha: is it really enough to install the .so? Shouldn’t the headers also be installed? <rekado>sughosha: why are origin and commit separated as a global definition? That seems unnecessary. <sughosha>rekado: Sorry, my computer crashed and I relogged in, I may be missed what you asked before. 1. I forgot to add comment for disabling tests. 2. It is just that there is no test suite for libswell, as far as I know yet. 3. I will use gexp instead. <sughosha>Regarding headers, I don't know if they have any use after compiling. I will take a more detailed look into that. And the only .so file needed is libswell, which provides interface for reaper (https://www.reaper.fm/, non-free) <sughosha>I seperated origin because libswell (or swell) is not the only thing it provides. It has many other things but I don't know which of them are for linux or Windows or MacOS, since it is cross platform. <sughosha>The only library that I know from this repo is linswell. Maybe in future if someone wants to build any other, they can use the same origin. <rekado>sughosha: I added my review to the issue for the record. <rekado>users of the .so will need to have some sort of interface declaration, which is what the headers provide <rekado>I’d prefer to separate the origin later if necessary. <rekado>it’s not something I’d do for just one package. <rekado>PSA: I’m rebuilding the Xapian database for issues.guix.gnu.org; the search will be affected until that’s done. <sughosha>rekado: Thanks for the review. I will take a deeper look into headers and come up with new version of patch. <sughosha>Also try to see if there is other package that could be built from wdl repo, otherwise don't seperate the origin. <rekado>PSA: maintenance of issues.guix.gnu.org is over. <rekado>please report any problems you might encounter <ham5urg>Can a user installed application be started after boot automatically? <ham5urg>Or has the user to login and do it manually? <ham5urg>ennoausberlin, yes, spawning a daemon (which was installed by a user in his profile) at start. <toledo>Does the binary installation of guix works on NixOS? <ennoausberlin>ham5urg: pm.scm might be a good start if you want to define your own daemon <mirai>is it possible to have multiple uri for a package source (think of mirrors but user defined) <mirai>itd: yes, that's what I was looking for <mirai>does git-reference support bare-repos as source? (used with file:// scheme) <efraim>rekado: I can bang on openjdk9 on aarch64 for a bit, try to finesse the guile code <sneek>Welcome back efraim, you have 1 message! <sneek>efraim, zacchae[m] says: Oh wow, and just like that my qutebrowser is fixed (as far as I'm concerned)! Thanks a ton <rekado>efraim: just FYI: I’ve built it successfully on pankow with this command: guix time-machine --commit=8e883dc8210d4a7c3f09961994685ed54942fd73 -- build --system=aarch64-linux --timeout=99999 --max-silent-time=999999 --with-c-toolchain=openjdk=gcc-toolchain@7 openjdk@9 <rekado>I set up a git checkout of Guix on pankow to find the right set of changes to make this permanent <rekado>a proper fix exists for later versions of the JDK but I wasn’t able to backport it. <isf>here user of icecat? <isf>Someone can tell me the names of the non-librejs/non-jshelter addons of icecat? <efraim>rekado: which version is that diff for? <ham5urg>Does autostart means a service-file must be written? <Guest54>now i can have backup iso's with any software i'd like <Guest54>if only packaging my own dotfiles with my guix config was possible <rekado>it’s apparently been backported down to JDK 11, but I can’t find any actual patches for that <efraim>I've seen that page before, I didn't get further than that. I wouldn't be suprised if Debian has a patch for jdk11 <Guest54>which software do you guys use for music? <efraim>I'll see about applying it by hand, i've been working on riscv64 and node backporting which is similarly fun backporting <rekado>Guest54: playing, recording, or performing? <isf>Guest54, I use rhythmbox for play music, radio, podcast <efraim>made it to file 8, that one doesn't exist yet in this version of openjdk <rekado>Guest54: I use mpd as the only music player in the home network. <isf>where I can find a list of the icecat addons? <rekado>efraim: I deleted a bunch of hunks and renamed all locations <rekado>(hotspot/src/cpu/aarch64/vm and hotspot/src/os_cpu/aarch64/vm) <efraim>yeah, I saw the added vm directory <rekado>and I only left hotspot/src/cpu/aarch64/vm/macroAssembler_aarch64.cpp and removed all the other variants (e.g. _log) <efraim>macroAssembler_aarch64_log and macroAssembler_aarch64_trig don't exist <Guest54>do any of you peeps here use org-roam with org mode? its the ultimate environment for note-taking <efraim>time for some grep fun to find [[:digit:]]+d\) <rekado>for control of mpd I use simple-mpc from within Emacs. <mirai>rekado: do you use mpd to stream audio over the network? <rekado>mirai: I don’t stream. I have a little home server that takes care of backup and storage, and it’s also connected to my living room speakers+amp. <rekado>I *could* stream with mpd but my living space is too small for this to make much sense. <ham5urg>Must any daemon be started via a service-module? <mirai>rekado: do you use the mpd service as a system service or is it a "home service" <mirai>< has no idea if services under gnu/services will also double as home services <mirai>ham5urg: "no", at least they don't have to be configured as service-modules <mirai>rekado: does it have a pulseaudio output? <mirai>ham5urg: you can configure the daemon manually but you will have to find a way to autostart it <rekado>mirai: I don’t use pulseaudio with mpd; the machine is headless, so I have no need for it. <mirai>could be a simple-service or adding it to xinitrc or bashrc, etc. <rekado>(I’m using ladspa plugins with alsa for effects like equalizers) <mirai>rekado: oh ok, I was wondering how you were using it if you're not streaming and not using pulse <efraim>ran out of space on the offload machine, I'll try with --no-offload first <mirai>I've been wondering if there were any mpd service users running it systemwide with pulse <mirai>ACTION suspects that mpd-service-type is subtly broken in many ways but in non-apparent ways <sam-d[m]>I have a package install in profile a and want the same package added to profile b. I am offline. I was expecting guix install -p ... to first check the store before polling the substitute servers. However, the install fails because no substitute servers can be reached (because I am offline). This did not match my expectation. How do I achieve this in guix? <mirai>I suspect it falls apart if you use it systemwide with a pulse output as it sets a env variable that is non-system user specific but no one complained about it yet <mirai>will take a look more carefully to it when I get the time to <ennoausberlin>ham5urg: You can pass the --network command line option to share the hosts network <ham5urg>ennoausberlin, I would need more, a different network as the host would serve multiple networks at the same time <ham5urg>With paired NICs, one in the container, one in the host and the routing done by the host (so my plan) <superkamiguru>So I have been trying to generate a docker image using "guix system image --image-type=docker --system=aarch64-linux /etc/config.scm", but during the build it keeps installing powerpc crossbuild packages and then the builds fail due to ppc not being supported. Is there any reason why building for aarch64 is pulling ppc dependecies? <ham5urg>ennoausberlin, danke anyways you helped a lot. <superkamiguru>For context, if I use the system tag with "guix pack" and generate a docker image that way, then it downloads the aarch64 dependencies correctly. Not sure if there is something I am missing that is different between guix system and guix pack <jonsger>superkamiguru: are you on an aarch64 host system or on x86_64? <jonsger>cross-compiling would require --target=aarch64-linux-gnu, but I'm not sure if this is possible for docker images... <superkamiguru>what is the main difference between --target and --system? Under the qemu-binfmt-service documentation it gives an example of using -s to cross compile <superkamiguru>"guix build -s armhf-linux inkscape" after enabling the service. -s also works for cross building with guix pack but --target failed <superkamiguru>Sorry if I am being confusing at all. To sum up my experience here after enabling the binfmt service: <superkamiguru>guix system image -s aarch64-linux : tries to build for ppc and fails <jman>Hi, what is the preferred way to lint a Scheme/Guile file and see all errors/warnings? I run `guile-tool compile myfile.scm`. Is there a better way? <rekado>ham5urg: ‘--network’ shares the net namespace <rekado>you can give the container its own net namespace and manually create a network device <rekado>ham5urg: see the cookbook for an example of a Guix System container with separate networking <jman>thanks rekado - good point. I'm in Scheme mode in emacs but `flycheck-list-errors` returns no errors. I'll investigate why <nikken>is someone already working on porting Gnome to 43? would that ideally end up in one big patch set or is incremental work welcome (or feasible)? <rekado>jman: you’ll need emacs-flycheck-guile <rekado>nikken: this would be done in several separate commits that would be applied together. <jgart[m]>akirakyle: Are you planning to upstream that guile-g-golf@0.8.0-a.1? <ham5urg>rekado, I saw it but the example is not guix-ish I would say. <rekado>ham5urg: how so? Because the network interface is built manually after starting the container? <rekado>to make it Guixier one would have to extend coverage of guile-netlink and extend the network services in Guix. <nikken>rekado: thanks, i thought as much :) there's no gnome/desktop team afaict, so is there anyone i could reach out to if i want to help? <rekado>I once upgraded Gnome, so you could reach out to me, but it’s probably better to just ask anyone here for help <jgart[m]>nikken: You can also reach out to raghavgururajan <jgart[m]>nikken: They worked on GNOME as part of Outreachy internship and afterwards as a volunteer <mirai>is there any special procedures needed to make asan work with a guix package? <rekado>ham5urg: our networking services are built on guile-netlink <rekado>when I wrote the cookbook entry guile-netlink did not support bridge devices. <KarlJoad>I am setting up a zabbix-server-service-type. I have a database password set, but when I try to start the server, it errors out with "database is down" "connection failed" "FATAL: password authentication failed for user "zabbix". I did manually set the password using "passwd zabbix". <attila_lendvai>i have trouble connecting to protonmail.com from my laptop, regadless of my upstream. their servers seem to be blocked from my guix (as per `telnet ip 443`). any ideas how to debug this? ip route seems fine, and the ip address is ok, too. i'm baffled. <podiki[m]1>I haven't had any protonmail problems from my guix machine <attila_lendvai>podiki[m]1, what's baffling is that it also happens with telnet. maybe i have misconfigured something, and things have changed meanwhile, but the IP address is correct. <mirai>is there any purpose for git-daemon-service to exist when git-daemon-service-type is also exported? <mirai>its one of the few services that is not used as (service foo-service-type) but rather as (git-daemon-service #:config ....) ! (and the keyword can't be omitted as well) <podiki[m]1>attila_lendvai: yes that's strange...sadly my network and routing experience is too limited (well I'm good at breaking my local DNS server and then fixing it at least) <KarlJoad>mirai: I believe the answer there is git-daemon-service allows you to more easily construct the git-daemon-service-type more easily in the simple case, but I am not sure. <mirai>git-daemon-service is "helpfully" defined as (define* (git-daemon-service #:key (config (git-daemon-configuration))) (service git-daemon-service-type config)) <mirai>it saves you from typing (service ...-service-type) at the cost of typing #:config <mirai>but is it worth having a mix of (service foo-....-type ...) and some stray (foo-service #:config ...) ? <mirai>I'm guessing this is a relic from before the services got the syntax they have today <nckx>Certainly not! A patch updating it to the new style would be warmly welcomed :-) <mirai>nckx: I'll prepare it shortly then <mirai>do I just erase it from existence or do I add a deprecation warning as usual <mirai>and of course, edit the documentation to reflect the new syntax <nckx>That style has been deprecated for aaages, but since I see it's still the officially documented one, yes, best deprecate first. <nckx>Two totally unconfusing uses of deprecate but you catch my drift. <nckx>attila_lendvai: What happens with telnet? Immediate disconnection? <nckx>So even if you hit return? <nckx>Oh, enter text and hit return. <nckx>You said you used the IP, so can't be DNS. <attila_lendvai>nckx, well, i used the domain name, but it resolves to the right IP, as per a web based dns lookup tool <attila_lendvai>and i haven't messed with the firewall, or at least not that i know of. i have a wireguard VPN, but it deosn't have anything special, and all other sites work that i have tried. baffling. <nckx>OK, but does using the IP directly work? You said telnet ip 443 iirc. <nckx>My understanding of wg is that it is not 'that' kind of VPN that magicks DNS settings (which is good) but is not deep. <attila_lendvai>nckx, telnet prints "Trying [ip address]", so... but i have tried everything i could, including copy-pasting the ip address. <attila_lendvai>i have reached a point of even considering that maybe someone at Guix got angry of them, and grepped the source... <nckx>Ping working is justt great; bye-bye mtr... :-/ <bjc>if ping works, but telnet doesn't, it sounds like a firewall somewhere <nckx>(To be clear: the facebook blacklist is (a) weird (b) extremely opt-in.) <attila_lendvai>bjc, yep. but i didn't even have iptables installed. i installed it, and looked at iptables -L... any other firewalls in guix? <bjc>firewall on the wireguard end? <nckx>And you can reach it from other machines connected to the first hop? <bjc>i don't think guix puts in any firewalls by default <nckx>(He's always disconnecting me on IRC so we have beef.) <bjc>that's what you get for being an alt-right troll <attila_lendvai>bjc, wg is only for a specific subnet, and turning it off doesn't change anything, unfortunately <attila_lendvai>nckx, yep. browser on my phone, on the same wifi, just opens the page right on <attila_lendvai>bjc, i thought of that, but telnet goes through ipv4, and the ipv4 address pings. if i copy-paste the ipv4 address into the browser, it remains an hourglass forever. <attila_lendvai>plus i have ipv6 turned off (as per ifconfig, the wireless has no ipv6 address) <futurile>Beginner Q: I'm a bit confused about 'profiles' and guix pull. Imagine I have a couple of profiles under ~/.guix-extra-profiles. Do I do one `guix pull` and then guix upgrade for each profile (e.g. `guix package --upgrade --profile=/home/blah/.guix-extra-profiles/example/example`). OR do I do a `guix pull` for each profile (e.g. guix pull --profile=/home/blah/.guix-extra-profiles/example/example` and <futurile>then guix upgrade for each profile. When I do a guix pull, with a `--profile`` - the second option - I'm getting odd results. <bjc>futurile: ‘guix pull’ doesn't upgrade anything other than the guix recipes <bjc>you still need to manually upgrade, post-pull <nckx>futurile: The former. Guix lives in its own profile, alon, this is the --profile that ‘guix pull’ accepts. <futurile>OK - I must be confused about what guix pull does then. I knew the `guix` command lived it its own profile. I thought `guix pull` updated the package versions from the repository (as well as updating the guix command). So when I was updating a `profile' I was thinking I have to do `guix pull` to get the latest package versions for that specific profile. <nckx>(And by Guix I mean the guix command. You'll see that ‘type guix’ shows a location in your ~/.current, different from any other package your user installed, which are in ~/.guix-profile. That's also a profile on its own.) <nckx>Guix is just guix, the guix in your $PATH, there is no separate ‘guix’ for each separate ~/.guix-extra-profiles. <vivien>If I submit a package to guix, am I to add my name in the copyright section on top of the file? <vivien>Does it apply for package updates? <vivien>Is there a threshold, like 15 lines of patch? <attila_lendvai>damn! after a pull/reconfigure/reboot, i can once again connect to protonmail. i have a hypothesis: at one point i tried to connect to proton vpn, but ultimately failed and gave up, and something may have remained half-configured <lechner>vivien / i think it's up to you. from a copyright perspective, the most important thing is that the current year is represented somewhere. you should always add your name and year if you are very young. in the us, copyright is the life of the author plus seventy years, but we have git now <lechner>in case of doubt, it's better to add yourself <lechner>at least for yours truly, copyleft is better than public domain <vivien>I’m not very young but I don’t understand your last message, guix is not public domain <lechner>vivien / without a copyright marking your contribution might be considered as being placed in the public domain, but i think that risk is low given the customs in free software projects <vivien>Is there a place in the world where that might be the case? <lechner>vivien / where which might be the case, please? <vivien>Do you know where someone could claim that my code is "public domain" if I don’t add a copyright line? <vivien>(In my country it’s not even clear whether you are allowed to release something into my country’s own definition of "public domain") <vivien>(I mean, before the author’s death) <vivien>The package definition for guix that I’m talking about was written after 1989 :P <nckx>Copyright is something you must explicitly disavow in the US, and in many places that's not even possible at all. <vivien>So if the commit log has my name and a date, I don’t have to update each file I’m touching? <lechner>i think there is a real risk of code being published under expat/MIT on Github by default, for example <nckx>Well, no, but those aren't copyrightable anyway. <nckx>Yes, but what is copyrightable is not quantifiable. Some things obviously aren't (changing a version and hash), some things obviously are (a programme), but most lies in between. <vivien>The threshold is the amount of $$$ I have to pay a lawyer to prove that the code is mine I guess :) <nckx>That is actually the correct answer :) <nckx>There was a time where I deliberately added my name for changes that probably don't qualify. If somebody ever wants to steal Guix, let them do the busywork of browsing the log. <nckx>I do not think they will. <Kolev>I recently re-licensed my stuff under GPLv3 since Fedora and others kicked out CC0. <Kolev>My stuff isn't worth anything, but I guess I still have to GPL it. <nckx>And now it's more valuable. Win-win. <nckx>ACTION reboots, which will probably fail. <lechner>Microsquirt will marry ChatGPT to GitHub. Any code you need for any problem, conveniently licensed under MIT <mirai>you could dual license your work <Kolev>I was using CC0 because I wasn't even sure if my stuff could be copyrighted. <mirai>in theory, if the original project wants to relicense to something else, that 'OR' clause can save them some trouble if that's what you're worried about <lechner>please go GPL all the way. i still remember the old time, and it was not good <vivien>It would be cool for guix to warn: "Warning! You linked that GPL library to this non-GPL program! Are you sure you have enough lawyers?" <mirai>vivien: that's the job for tools such as REUSE, etc. <Kolev>lechner: I've settled on GPLv3+. <rekado>efraim: oh, got openjdk9 to build after all. <Kolev>Even if it's just my little scripts, it's getting GPLv3+. <rekado>efraim: must have made a mistake earlier when I added gcc-7 <mirai>going GPL has subtleties, you can see the importance of the 'or-later' clause in effect with freecad and libredxf <efraim>rekado: I found I needed gcc-7 for openjdk10 also <mirai>(and there's the license compat too) <efraim>openjdk11 is still supported, so hopefully we won't need to add it there too <Kolev>mirai: The lack of "or later" is awful. <mirai>Kolev: a matter of opinion (see linux kernel licensing debate) <lilyp>People working on the Linux kernel can be wrong too. <Kolev>mirai: As someone who had no problem licensing my stuff public domain, I'm not that scared of GPLv4 being bad or anything. <GalaxyNova>lilyp: do you dissagree with the inclusion of rust code in the kernel? <Kolev>GPLv3 still applies, even if GPLv4 is bad. <mirai>lilyp: I'm surprised they went with rust when there's Ada and SPARK already <Kolev>The Government loves Ada. Write in Rust. <GalaxyNova>rust uses llvm and is hyped up for os dev a lot nowadays <Kolev>Write in Rust, write in Rust. Write in Rust, write in Rust. Don't even mention JavaScript. Write in Rust. <mirai>Kolev: personally, I prefer to use AGPL on non-trivial things, the trivial stuff gets a 'MIT or CC0' just to appease the "patent concerns" <lilyp>Write your JavaScript^H^H^H^H^H^H^H^H^H^HWebAssembly in Rust. <mirai>and where the project is GPL but my changes are trivial, I'd put my patches as 'GPL or MIT or CC0' <Kolev>mirai: I thought about using Expat, but it wreaks of corporate Silicon Valley hipsterdom. <lechner>Kolev / not sure GPLv3 matters with a bad v4 <sughosha>Hi! I am having trouble in packaging something. This package builds successfully with the old style of build arguments, but when I use gexp, a warning says "'replace' may only be used within 'modify-inputs'" and this package doesn't appear in pre-inst-env. <sughosha>I want to replace build phase, by the way. <Kolev>lechner: People who do GPLv3-only are usu. concerned about a future GPL that they disagree with. They don't want their work to fall under it. But it's no worse than if you licensed your work Expat. <mirai>Kolev: copyright topics usually revolve around limiting something that is practically infinite (not scarce) <nckx>sughosha: Sounds like you might not have gexped modify-phases? Share the code to find out. <mirai>personally I find that incredibly silly (and at conflict with the concept of money as you're basically selling something that's endless = near infinite money cheat code) <Kolev>I'm not concerned about evil GPLv13. GPLv3 will always apply to my work. <sughosha>nckx: I have gexped modify-inputs: #~(modify-inputs). <lilyp>modify-inputs doesn't need gexps <sughosha>It is not modify-inputs. It is #~(modify-phases) <nckx>sughosha: Don't pay attention to that error message. <sughosha>nckx: But the package doesn't show up in pre-inst-env. <nckx>(Probably) because m-p isn't gexped, Guix wants to see replace in the context of m-i. <mirai>if it were not for the "monopoly" provision that copyright grants to the holder, I personally wouldn't care whether or not I want my changes labelled as X or Y or that you could grab them and not share back. It's the exclusive rights that irk me the most <nckx>sughosha: If you can't share the code, just gexp m-p and do not gexp m-i, and try that. If you get more errors, share those. <nckx>s/can't/& or are unwilling to/ <sughosha>nckx: I can only see my new package in my pre-inst-env, if I keep arguments in the old way. Only when I gexp arguments, I am getting the above warning and an error saying unknown package. <nckx>Please correct the error and try again. <nckx>Or share the code. (Best deal.) <lilyp>sughosha: are you sure that you properly transition your code? <lilyp>In particular, are you sure that you don't quote the gexp <nckx>And don't #~ any part of any inputs. <apteryx>cbaines: hi! is QA happy? Most statuses seem to appear as 'unknown' <nckx>Add (guix gexp) to the imported modules. <sughosha>nckx: Oops, silly mistake.. Thank you!!! <nckx>I guess because of the way things are implemented, Guile will try to interpret the ‘replace’ syntax before it ever gets to the unknown (because not imported) ‘#~’ syntax. <nckx>I don't see anything else wrong, but I didn't build it. <nckx>Those who back up daily or more often: what do you use? <Kolev>I use Borg but don't run it as often as I should. <nckx>mirai: No, just wait for a review. You can send a reminder ping by now. I cannot review it on here, but I see no obvious flaws. Is the size significant enough to warrant a separate output? <ConvolutedSquare>nckx: by the way, what I ended up doing is installing Guix onto a USB drive and booting off that. Is there any danger to creating a script that mounts over /gnu/store after booting? I want the store on my BTRFS array. <nckx>Thanks for your back-up answers by the way. <nckx>I have too many questions. <nckx>I just have to leave soon, so can't. But this is a working set-up? <ConvolutedSquare>The USB boot is working perfectly, yes. Didn't try mounting over the store yet <nckx>OK. I assume this system is safe to experiment with (at least now :). I really have too many questions, but many of those will be answered by your (mis)adventures. Good luck. <elevenkb>hello there, `emacs-org-roam` has a bug but I know how to patch it to fix the bug. <elevenkb>until I come up with something suitable for upstream, what is the best way to modify the package so that it includes my fix? <piethesailor>Hello all! I 'guix pull'ed and 'guix upgrade'ed my machine recently. rebooting errors out and I land immediately in: grub> <elevenkb>piethesailor: what hapens if you press ESC? <piethesailor>I also took video of the boot process and I see there is an issue after 'Continuing boot since secure mode is disabled'. this error is 'error: failure reading sector 0x1d2 27438 from `hd0'. <piethesailor>or as the screen says. GNU Grub version 2.04 Minimal bash line editing is supported.... grub> <ieure>It sounds like your Grub configuration got eaten, so all you have is the grub CLI. <ConvolutedSquare>piethesailor: If I were you I would boot a rescue USB and check that the drive is healthy before anything else. <rekado>efraim: unfortunately, java-testng fails. Has test failures. <rekado>(I’m just interested in building groovy on aarch64)