<anadon>nckx: Because the build sometimes hung, I've resorted to tee'ing all the output to a log, then have been looking at that. <anadon>It never told me (I think?) where it put any log. <nckx>If you use --keep-failed one of the last ~4 lines will be ‘note: …’ <nckx>Every kept failed build will create a new directory. <anadon>But the tests hung, so it wouldn't ever get to any "last line". <nckx>anadon: All right. As mentioned, mainly working now, but will poke around. <nckx>anadon: You can kill things 😉 <anadon>OK. Please find out where I did the dumb thing :P <nckx>(From top or htop or whatever you prefer, I mean.) <anadon>I've been working most of today, and it is time for something else. <leoprikler>after reading the GWL docs... could this be used as a build system? <anadon>leoprikler: I think that has been a recurring joke. <leoprikler>IIUC GWL always runs all processes, whereas make et al. look if they can skip something, is this correct? <raingloom>soo i think static-package (from guix/build-system/gnu.scm) is buggy. i applied it to f2fs-tools and while the libraries are indeed static, the executables are not. not sure why. <dissoc3>im trying to get xrandr working again. all my monitors work (have output) but are arranged not how I want. so i try to use xrandr. when ran X closes and the message returned from xrandr is "X connection to :0.0 broken (explicit kill or server shutdown). <dissoc3>What logs should I be checking? /var/log/Xorg.*? <ZombieChicken>bavier`: docs claim one still needs to pass something grub understands in the target field of the bootloader-configuration. Are the docs out of date? <dissoc3>Blackbeard: i talked to you about this the other day and you said you dont have problems? do you have nvidia card? <Blackbeard>dissoc3: what is the xrandar command you are running ? ***vagrantc_ is now known as vagrantc
<Blackbeard>dissoc3: I do but I only have two monitors, not seven :/ <dissoc3>Blackbeard: xrandr --output DP-1 --mode 1920x1080 --rotate right <dissoc3>Blackbeard: well I disconnected all but one just to try to isolate problem. but it still exists <dissoc3>Blackbeard:so im down to one video card and one monitor for testing <bavier`>ZombieChicken: I may have misspoken. I recall uuids working, but have not actually used them for grub myself <ZombieChicken>seems like more than one bootloader is currently working in GuixSD. Can't find a convienient list of which ones are there, however <dissoc3>Blackbeard: i tried many xrandr commands but they all result in the same. X closes and it takes me to slim (or gdm) <Blackbeard>dissoc3 try this simple one $ xrandr --output DP-1 --mode auto <dissoc3>it says xrandr: cannot find mode auto <dissoc3>is that the same as --auto? because that doesnt crash. but i also think it doesnt invoke any changes <dissoc3>it spits out the connected monitors and interfaces <dissoc3>Blackbeard: i tried just simple transformations. rotating, resolution changes etc <Blackbeard>dissoc3 try this simple one $ xrandr --output DP-1 --mode 1920x1080 <dissoc3>works no problem. but it also does not change the screen as it's already in that mode <Blackbeard>dissoc3 try this simple one $ xrandr --output DP-1 --auto --rotate right <dissoc3>Blackbeard:any time a change is invoked it breaks <dissoc3>Blackbeard: i tried with varying resolutions <dissoc3>Blackbeard: it always breaks. but it boots with all monitors working. the only problem is they are all aligned in one row <dissoc3>Blackbeard: okay. i am skimming through the code. it looks simple <dissoc3>Blackbeard: do you think it would be worthwhile to try to manually add the modes and specifiy them directly? <Blackbeard>dissoc3: I don't know, but I would give it a try. <dissoc3>Blackbeard: well for whatever reason that software you linked works. at least it seems to. <dissoc3>Blackbeard: it at least let me arrange two monitors. so I will dig into it a little further. <dissoc3>Blackbeard: i very much appreciate your help. i've been banging my head on the wall with this. <Blackbeard>dissoc3: wonderful !! Seems like it should be packaged for guix then <plstohelp>i know theres some... GNU reason firefox isn't packaged <vagrantc>i think there's a page on gnu.org about the various reasons <dissoc3>Blackbeard: i spoke too soon. it works with 2 monitors but not 3. but at least i have a starting point <bandali>firefox itself as is is not completely free <bandali>so that's one reason for existence of gnu icecat <bandali>another is that icecat ships with additional add-ons like LibreJS and additional privacy tweaks to better protect your freedom and privacy <witty_name[m]>How is Firefox nonfree? Is it like Linux where there's binary blobs from third parties or is Mozilla doing some listening/branding stuff? <plstohelp>no, the software is free, but they occasionally engage in practices like encouraging nonfree extensions <plstohelp>debian is free software, but they make nonfree software readily available <witty_name[m]>Yeah like there was some Mr Robot "promotion" or something right? <witty_name[m]>> debian is free software, but they make nonfree software readily available <witty_name[m]>I don't want to get into philosophical debates but if the package manager is free software it's not really unfree just because it can be used to install non free packages? <bandali>afaik the software itself is also considered nonfree as is, because of mozilla's trademark policy <vagrantc>it's more because debian hosts infrastructure and has documentation directing people towards the non-free stuff <witty_name[m]>> afaik the software itself is also considered nonfree as is, because of mozilla's trademark policy <Blackbeard>dissoc3: I will still look at package it for guix. Seems like it would be helpful <bandali>iirc mozilla's trademark policy prohibit distributing modified versions of firefox (and their other products, like rust) without explicit permission and/or commercially <bandali>which is in violation of the free software definition <bandali>afaik debian pretends that's not an issue because they got explicit permission from mozilla to using the official firefox name/branding <bandali>at least that's my recollection. i may be rusty on the details <witty_name[m]>So does every Linux distro have to get permission to package it? <Blackbeard>There is a difference between the Free Software Foundation not considering debian Free and not endorsing it <vagrantc>in fact, the maintainer of the firefox packages in debian was paid by mozilla to work on packaging iceweasel in debian for almost a decade before they changed their mind on the policy <bandali>witty_name[m], if you package it unmodified, and don't charge money for it, then no you can package firefox without explicit permission <Blackbeard>But they won't endorse it because it implies using windows <vagrantc>trademark can be used to break the spirit of free software without breaking the letter of the license <Blackbeard>They won't endorse Debian not because it is not free <witty_name[m]>> But they won't endorse it because it implies using windows <Blackbeard>But because it allows for easy installation of proprietary software <Blackbeard>They Do know Debian is free and recognize it as such. Same with Firefox <bandali>Blackbeard, the issue with firefox is somewhat different than debian though <vagrantc>so if you rename it, you can do whatever you want with trademarked free software <vagrantc>well, within the constraints of the license <usney>What is the best single board to use with free software or almost totally free? <bandali>yes, the trademark issue is worked around by using a different name for the fork <vagrantc>the name, the branding, etc. are trademarkable ... the software itself is copyrightable (and sadly, patentable in some cases) <bandali>as i've mentioned before, i think this issue applies to rust as well, but we don't have a fork of rust just yet <bandali>for one thing, same issue with the trademark policy iirc <usney>you need to use rust and cargo now to use firefox? <bandali>yes, rust has been a build dependency of firefox for some time now <usney>I heard google chrome is the most used web browser right now according to wikipedia <witty_name[m]>So if you want to apply (and distribute) any patches to Firefox you can't use any of the branding? <bandali>witty_name[m], i think so far mozilla hasn't tried to enforce that, but i think it would be within their right to, according to their trademark policy <vagrantc>yeah, near the very end ... guix really should ask for permission to call it "rust" <bandali>if it is nonfree, guix really should not package it ;-) <bandali>but we haven't had that discussion in the FSDG distro community about rust yet iirc <vagrantc>literally, that might be all it takes to comply <witty_name[m]>Ok thanks everyone for explaining. Not trying to argue just trying to understand. <vagrantc>the problem with asking for permission, is then each of the end-users would also have to ask permission <vagrantc>which kind of kills the point of free software <bandali>yes, it's by definition not free software anymore <bandali>witty_name[m], it is, but no one has disputed/discussed its existence and freedom concerns around it yet <vagrantc>well, technically it is, it's just encumbered <vagrantc>but that's a nuance maybe not everone wants to get into <bandali>i think having the FSDG guideline on gnu.org take a concrete and firm stance about the topic and its specific cases would be useful in putting these ambiguities to rest <witty_name[m]>Is there any reason to believe Mozilla wouldn't give permission to distribute patches of you asked <witty_name[m]>Is the situation "we have some patches we want to include but can't" or more "we don't want to endorse the license by supporting it" <bandali>it's more the latter: for a piece of free software you don't have to ask for permission to do that <bandali>also, as vagrantc briefly mentioned, the licenses under which firefox's and rust's code are distributed under is free, but (at least part of) the issue is mozilla's trademark policy on their names, posing additional restrictions on the use/modification of the software <witty_name[m]>What happens to all the rust packages on guix if the FSF rules against rust? <bandali>mostly likely, end up being renamed and change all references of rust and cargo to other names <usney>all my guix software I have installed is still working great! <usney>they may since they package icecat <pkill9>nice to hear usney, that's my experience too <usney>I think they have to call it something else? <usney>qbittorrent is not crashing like the debian testing package was since I uninstalled it and then installed the guix it works way better <plstohelp>im running in a vm for now because ,y wifi card lacks free dirvers <usney>they have a compiler for rust it is just listed as rust plstohelp <usney>I think you can search for packages while guix is installing software <plstohelp>so i just booted into guix for the first time and tried to install neofetch <usney>just not install software while it is already installing software <plstohelp>and its installing, like 10 million packages <plstohelp>imagine allowing tweak tool to exist when you could integrate it into your settings menu <pkill9>does anyone use quassel out of interest? <plstohelp>it actually annoys me that guix has package names AFTER the random characters instead of befire <pkill9>it's a decent terminal irc client <usney>cool I heard it is pretty good <usney>it is recommended over irssi <pkill9>managing settings is a bit clunky though <pkill9>probably not any clunkier than other terminal clients though <pinoaffe>hi guix! I'm running a vm in guix using virtual-manager and I want it to have internet access, I tried the "Host device macvtap" bridging option, but that did not work and resulted in the host loosing its internet connection - does anyone know how to set up a network link between the two? <bavier`>pinoaffe: I think my virt-manager vms are typically set up with "usermode networking" <bavier`>but I understand that if you want the vm to be able to communicate with the host then things can get more complicated <pinoaffe>bavier: ah okay, I don't really need the guest to be able to communicate with the host, would be fine with just internet access <usney>how can I setup encrypted dns with guix? <usney>I mean not the distro but the packages it has <plstohelp>so doing the /etc/gdm/custom.conf will hopefully flip me to wayland <Blackbeard>plstohelp: not relevant, but if you are on a VM what's' the difference using x11 or Wayland? <plstohelp>i just want to learn how to force gnome to run in wayland mode <bdju>I saw a gif of the installer that showed i3, but is sway also an option? <plstohelp>idk if its preinstallable but you can get it from the rpos <ZombieChicken>if I install Guix on another distro, I can use it to generate a VM I can use for testing, right? <plstohelp>also, none of the programs ive installed have been integrating right with gnomes application launch <plstohelp>feel like im using a tiling wm all of a sudden <plstohelp>the option for gnome on wayland is missing in the login screen, as well <plstohelp>okay, the applications only show up in gnome after a relog <bavier`>Blackbeard: I considered packaging those a while ago but ran into licensing issues <plstohelp>IS there a way to run gnome on wayland in guix? it seems to be x11 by defualt, and gdm doesn't have the option like its supposed too <pkill9_>plstohelp: yes it's possible, if you use the sddm display manager, it shows an entry for running guix on wayland <Blackbeard>bavier`: ohh :/ I was just looking at them. I Haven checked the licenses, I assumed they were legit. Because I was looking at libregamewiki.org <thelounge>how can I configure guix to install and update packages using tor? ***pkill9 is now known as pkill9__
***pkill9_ is now known as pkill9
<bdju>plstohelp: yeah I run sway on my guix install, was just wondering if it was a preset option or if I'd have to edit the config on a new install <bdju>someone already mentioned sddm, but I'll second that. in my config I had gdm disabled (it was pulled in by default from %desktop-services) and then I replaced it with sddm <bdju>because of issues with wayland iirc <bdju>I had even more display manager problems on a nixos machine so I've just been running "exec sway" from a tty, not sure if that works with gnome <Blackbeard>bavier`: you are right. Seems like some songs do not specify a license :( <bavier`>though, it does appear to be in a "better" situation since I last checked, iirc <pkill9>you could remove the songs without a license <Blackbeard>pkill9: or change them for songs with a good license <bricewge>thelounge Ludovic did it a few days ago (search the IRC log) <bricewge>You need a tor service, privoxy talking to for and then "herd set-http-proxy guix-daemon" <Blackbeard>liberdiko: is that in the cookbook,? I think I'll try to add it <plstohelp2>im a bit new to shepherd, and have been struggleing to make sense of the documentation. how would one configure to use sddm rather than gdm? <Blackbeard>plstohelp2: that's not on Shepherd it is on guix manual <Blackbeard>plstohelp2: most stuff is configured in your config.scm file <Blackbeard>So you get familiar with all the things you can configure easily in your config.scm file ***plstohelp2 is now known as plstohelp
<brendyyn>plstohelp: it requires removing the gdm-service, and adding the sddm-service <plstohelp>i added the sdm service with display-server wayladn but it doesn't seem to be running <plstohelp>the gdm service is coming from %desktop-services <plstohelp>the manual says to use remove for removing stuff from %desktop-services <brendyyn> (remove (lambda (service) (eq? (service-kind service) gdm-service-type)) %desktop-services) <plstohelp>yeah haha i had to bring in rnrs lists for that but maybe it worked <plstohelp>the reconfigure went though, is it supposed to redownload everything on a reconf? <usney>where is the best mailing list to post to for a free software web extension to redirect amazon links to the products name into a privacy respecting search engine? <brendyyn>things already downloaded will be on in the /gnu/store until you guix gc theme <plstohelp>well, when i hit "reconfigure" it started pulling loads of stuff from ci.guix.gnu.org <plstohelp>stuff i obviously already had, like the kernel <plstohelp>my god, is all of this because i wans't root? <plstohelp>this really is a new paradigm, i actually have to think about this differently <cbaines>plstohelp, you need to reconfigure as root, I use sudo -E <cbaines>it won't need to download/build things again though, it's all the same store <plstohelp>then why did it just redownload and build again when i pulled today <plstohelp>is it supposed to be pulling from the store? <cbaines>If you're reconfiguring with a different revision of Guix, you'll probably need to download/build some new store items <cbaines>because the package definitions will have changed <plstohelp>well, all i did was turn one service on and another off <cbaines>Are you reconfiguring with sudo, or have you logged in as root? <cbaines>Ok, keep in mind that root and your normal user might have different revisions of guix <cbaines>So the effect when you reconfigure your system as root, compared to your normal user might be different <cbaines>All I'd suggest is be consistent, either always reconfigure using your normal user, or root <plstohelp>i tried as normal, but it threw a "permission denied' <cbaines>You'll need to use sudo -E as your normal user <plstohelp>aw man, at automatically holds the old config in grub? gotta love guix <plstohelp>so i run the reconfig, and it pulls me into sddm <plstohelp>but it wont boot into gnome, and when i reboot, black screen <plstohelp>ive got the lamba remove to return evrything in desktop-services EXEPT gdm-serive type <plstohelp>ive got service sddm-service type. configuration, display server wayland <plstohelp>okay, so in my new config it says "new session c1 of user gdm" before it hangs <plstohelp>i would love some pointers because i can't find anything wrong in my config.scm <Blackbeard>plstohelp: can you share your config.scm? Use a pastebin <cbaines>plstohelp, I'd maybe try restarting if you haven't tried that. And if sddm isn't working, try and find some relevant logs. <plstohelp>basically, sddm lauched when i reconfed, but on reboot, i SEE gdm start, and then hang <ATuin>plstohelp: can you get a text console when that happens, or it just hangs the whole machine? <ATuin>btw question about the ASDF build system, what's the purpose of having the build-system for binaries used by packages? <ATuin>is that the recommended way? or just better to provide the source and let the users to compile it? <plstohelp>this config compiled, but black screen on boot <plstohelp>like i said, i saw the message of session c1 started for user gdm, so something is going wrong <plstohelp>but at the same time, the reconf launched me right into sddm, but it wouldnt login to gnome from there <plstohelp>going to sleep, will check logs in the morning ***apteryx_ is now known as apteryx
<PotentialUser-50>Hi there! It's my first time to install guix. I am getting a build of `/gnu/store/v5p240zm3w45qf30da7rn5zmpl41d2w0-profile.drv' failed <xelxebar>Is the error giving you any more hints? Or is it one of the tight-lipped "failed to build" errors? <PotentialUser-50>build of /gnu/store/079awjky5kg0ai24pzrs5xb4jhxxa67i-guix-packages-base.drv failed <ATuin>maybe you have more info in the log "var/log/guix/drvs/07/9awjky5kg0ai24pzrs5xb4jhxxa67i-guix-packages-base.drv.bz2" <xelxebar>PotentialUser-50: Maybe you're running out of memory? I've had very similar issues on resource-constrained systems when pulling updates guix-packages-base. <xelxebar>IIRC, someone mentioned you need at least 2gb laying for base. Don't quote me on that though. <xelxebar>Yeah, looks like you have about 1gb free. My hypothesis is that an OOM (out of memory) condition is killing the build. <xelxebar>PotentialUser-50: Just cat the path to that log file <ATuin>less would it, normally it's configured to deal with compressed files <xelxebar>If my hunch is right, I bet it contains just a single line <ATuin>in the *worst* case you can decompress it <ATuin>xelxebar: probably, but it's a good advise to look always at the logs, just in case :d <ATuin>seems that asdf-build-system produces wrong ASDF versions <cbaines>PotentialUser-50, you can use bzcat to view it <ATuin>yep, your less is not properly configured to deal with bz files <cbaines>I don't think less can automatically decompress things, at least that's never worked for me <xelxebar>PotentialUser-50: The point being, that log file is just a normal file, so you can deal with just like you would any other *leg.bz2 <ATuin>cbaines: LESSOPEN should do it <xelxebar>PotentialUser-50: Yeah, looks like the build is just dying in the middle. If you have a way of bumping up the memory easily, it's worth a try. Otherwise... wait for a substitute to pop up and try again? <xelxebar>I've had luck waiting around for a while and trying to pull again <ATuin>anyway forget about thhe less thing, as xelxebar said it's a normal file <ATuin>but looks like it failed downloading, is that all the output you get? <ATuin>ahh no sorry, not downloading <xelxebar>PotentialUser-50: I'm successfully running guix on a machine with only 512MB memory. Do occasionally encounter exactly this problem when pulling, though. <xelxebar>Usually, waitaing around a while and trying to pull again later will eventually end up working, because it happens to find a substitute <cbaines>Is it guix pull you're trying to run? <brendyyn>im pretty sure there are other x200 users <PotentialUser-50>should I install anything new in the meantime? Won't it affect my rollback if I have to do it? <cbaines>PotentialUser-50, I'd try guix pull --commit=f0da92cb42f99cddadd9cea373758355cb7c6351 <xelxebar>cbaines: As a matter of fact. I just had the same thing happen ond my tiny f1-micro gcp machine. guix pull fails on guix-packages base... <cbaines>That commit should have substitutes from ci.guix.gnu.org <xelxebar>cbaines: Oh, nice. That's the resource I need. Why modular-master instead of master, btw? <xelxebar>I should check the repo for documentation about the different branches <cbaines>Just naming, the "guix-master" Cuirass specification builds packages and system tests. "guix-modular-master" builds the derivations related to guix pull. <xelxebar>Hope PotentialUser is just migrating usernames to NewUser... <ATuin>mmmm why asdf sources are installed in a directory prefixed with "sbcl-"? <ATuin>are not they source, so independent of the lisp implementation <xelxebar>Also, just want to say how awesome it is to see wip-hurd getting continued attention. Slowing trying to bootstrap myself so I can join the ranks of these devs <ATuin>ok, i have a fix for the vesioning in asdf, so now let's see how to contribute it. First time i do it. <leoprikler>and to build asdf, you need some cl package, and I assume sbcl is the one guix uses as baseline <ATuin>in this case i was building explicitly sbcl-alexandria so it used asdf-build-system/sbcl <ATuin>so yeah, that makes sense, but still i'm a bit confused i ended up with 3 asdf files <leoprikler>yeah, guix can be confusing at times, but it helps if you have a look at the graphs <davidl>Does anyone know how to install conda packages when you have installed conda to /gnu/store using the guix package for it? I need something similar to pip install pkg --user, so I made a conda profile with conda create -y -n conda_mlcourse ; source activate conda_mlcourse ; conda install tensorflow, but Im just getting a stacktrace error. <ATuin>one in "lib/sbcl", one in "share/common-lisp/sbcl-bundle-systems" and one in "share/common-lisp/sbcl-source" <ATuin>the one in sbcl-source seems to be just the original source code, since the asd file does not contain the version generated by guix, that's why i thought that it would make more sense to call it just source <ATuin>the problems i'm finding is getting different asd files for the same system in the asdf:*source-registry* <leoprikler>davidl: I think the preferres solution is to use `guix [install|environment --ad-hoc] tensorflow` <leoprikler>specifically, prefer (pure) environments over pip/conda/node/package.el/... <ATuin>leoprikler: anyway i will submit a patch for the versioning problem first, since asdf is complaining all the time that we are breaking the version format and returning NIL as version <ATuin> FileNotFoundError: [Errno 2] No such file or directory: '/home/user1/.conda/pkgs/openssl-1.0.2l-0/info/no_link' <ATuin>that looks the original problem davidl <ATuin>dunno how conda works though <cbaines>assuming /home/user1/.conda/pkgs/openssl-1.0.2l-0 doesn't exist, maybe you can ask conda to install openssl-1.0.2l-0 <ATuin>question about guix patches: i guess before submmiting one i need to open a bug report right? <davidl>leoprikler: I agree that would be nice, Im following a course where I dont have as much time as I would like though, but eventually I would want to do the course with guix env instead. <leoprikler>fyi tensorflow is already packaged, so you can try that whenever you want/need to <cbaines>davidl, hmm, unfortunately I don't know anything about conda <leoprikler>out of curiosity, what happens if you do conda install xz? <xelxebar>cbaines: BTW, your --commit=<blah> flag fixed the failing pull on my system. Hope it did the same for PotentialUser as well. <davidl>leoprikler: thta did seem to work actually, while conda install openssl gave a similar stacktrace error as conda install tensorflow. <leoprikler>it appears just the openssl package in conda is broken then <g_bor[m]>I am runnign guix system in a vm right now. <g_bor[m]>I can tell that the installation went smoothly form the 1.1.0 ISO <g_bor[m]>I am tryng to go for a setup to share certain files with the host. <g_bor[m]>Do I need to do anything fancy, or just declare it in my os confgi with fs type 9p? <cbaines>g_bor[m], providing you get the mount parameters right, I think it just works <u8t6>I tried to install Guix System distribution yesterday, turns out 1GB of RAM is not enough <TZander>u8t6: if you end up having to compile some stuff you run through ram very quickly. <cbaines>u8t6, out of interest, did you have any swap? <u8t6>> u8t6, out of interest, did you have any swap? <g_bor[m]>I have notice something... if I try to reboot or shutdown my system from virt-manager it freezes. <cbaines>Were you using the graphical installer u8t6 ? <u8t6>anyway, I installed debian right away, will try it over debian <cbaines>u8t6, Hmm, I'm not sure how that handles swap. <g_bor[m]>my guess is that sheperd gets the info to restart the machine, but somehow fails to bring it down... <u8t6>Hello g_bor[m], if no one answers your question here you may direct it to the mailing list at bug-guix@gnu.org <g_bor[m]>Actually this is more of a proble if someone really don't want to hand manage vms. <Kimapr>i guess it's a stupid question and probably already answered before, but what is alternative of build-essentials on guix? <g_bor[m]>The debug log revealed a RIP fault in the drm module. <Kimapr>already found guix package names for everything but build-essentials <cbaines>Kimapr, as far as I know, build-essentials in Debian is just a metapackage that depends on common packages required for building things <g_bor[m]>it looks like it is fixed in newer kernels. <g_bor[m]>For the time being I guess you can just disregard this. <Kimapr>scons: Reading SConscript files ... <Kimapr>Checking whether the C compiler works... no <Kimapr>compiler not correctly configured <Kimapr>i guess build-essential were needed to actually build it <cbaines>Kimapr, so if you need a C compiler, use the gcc-toolchain package <Kimapr>should have think of this myself <Kimapr>i think it is weird that installing gcc also downloads gtk+ <raghavgururajan>sneek. later tell apteryx: Thanks for pushing the patch. Btw, I tested it on master and the application worked fine. I did not get Mesa related errors. I tested by `./pre-inst-env guix environment --ad-hoc linphoneqt --pure` and `linphoneqt &`. <MYMYRSRS>I used the --no-substitutes option to build my package, but it is too slow to download. Can I change the download address it uses? <ecbrown>well, there's your justification to go fibre ;-) <MYMYRSRS>I do n’t think it ’s a transmission media issue, it ’s a regional issue. <cbaines>If it's downloading from www.python.org that's slow, I'm not sure there's much that we can do <cbaines>MYMYRSRS, is python the package you're trying to build, or is it something that uses Python? <MYMYRSRS>Due to regional issues, the fastest way for me to download the source code is to use gentoo's distfiles (there are mirror sites in my area). <cbaines>MYMYRSRS, are you using --no-substitutes because you want to build python from source, or some other reason? <cbaines>I ask because fetching the substitute (if available) for python might be faster. <cbaines>that's unfortuate. Guix isn't great in terms of mirror/network access <raingloom>more reason to work on torrent/ipfs/whatever integration? <cbaines>well, those things might help, or at least provide paths to making the situation better <MYMYRSRS>I think Guix only needs to provide the rsync service of ci.guix.gnu.org to improve this situation. <MYMYRSRS>If so, I can build a mirror site in the area <MYMYRSRS>Then the question comes back, is there really no way to change the download address of the source code? <cbaines>I'm not sure what the state of the rsync service on ci.guix.gnu.org is <cbaines>MYMYRSRS, you can change the download address in the source code, Guix does support mirrors as well (although I don't have a great understanding of that) <cbaines>The other approach would be to edit the right .scm file in a checkout of the Guix Git repository <MYMYRSRS>Thanks, I will try to add support for other distributed distfiles in the source code. <raingloom>static-package did make it compile libraries as static, but the executables still link dynamically to a bunch of things. <raingloom>i tried forcing GCC to use the -static flag, which it seems to do, but the result is the same. <TZander>each item in your 'ldd' it links to should be compiled as a static lib instead of a dynamic one. All need to be compiled that way. <TZander>I'm no guix expert, not sure if there is any help for that. But that may be something you missed <bricewge>MYMYRSRS: I also have speed speed issue with ci.guix.gnu.org for a few days. <MYMYRSRS>I can't ping ci.guix.gnu.org now, only hydra.guix.gnu.org is used in the morning (GMT+8) <MYMYRSRS>Before that, I always visit ci.guix.gnu.org in the morning and it has a speed of 300kb / s, other times it is 10kb / s <ecbrown>seems like i get burst from ci.guix and then it tapers off to 1.4 MiB/s <bricewge>I also have a brust at first but it tapper at 40KiB/s after ... <raingloom>....oh. duh. i was using both util-linux and util-linux:static as inputs. <raingloom>....but removing util-linux breaks the build.... <jonsger>MYMYRSRS, bricewge: in which countries do you reside <bricewge>jonsger: France. I've come to believe that my ISP (Free) is fucking with me: small substitutes are downloaded fast but for big ones it's utterly slow <TZander>bricewge: try downloading something from another place. Like an ISO from kubuntu,org or so. <TZander>or, just anything big from a company like microsoft :) <bricewge>TZander: Downloading Kubuntu or the latest Guix ISO, I've good speed 2MiB/s <jonsger>oke, I get 10MB/s at Vodafone Germany <civodul>bricewge: i have the same ISP as you (with FTTH) and i got 65MB/s for the URL above <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, nckx says: Wow, that was fast. Thanks! <MYMYRSRS>The speed of accessing ci.gnu.org a few months ago was barely available, but it is now very slow <civodul>MYMYRSRS: there was initial work to rsync from ci.guix.gnu.org to a server in China <civodul>if you're interested and in a position to help with that, let us know! <bricewge>civodul: Argh, so this really regional... <greenrd>I am trying to install guix and the installation failed because the keyboard layout is called different things when it comes to the Linux console and Xorg <greenrd>so I tried to set it manually by (xorg-configuration (keyboard-layout (keyboard-layout "gb"))) <bricewge>MYMYRSRS: Well then, we don't have an issue of the same magnitude I guess <greenrd>hold on, it's still printing the other keyboard layout name in the error message, so I must have a syntax error <MYMYRSRS>To be honest, my programming level is not very good. What contribution can I make to Guix's servers in China? <civodul>bricewge: you're on "real" Internet, not 4G? <greenrd>maybe the problem is that I am trying to call the procedure called keyboard-layout, but guix thinks that I am trying to refer back to my previous keyboard-layout definition <civodul>MYMYRSRS: we should check with rekado_, but i think we need someone to host a server in China <civodul>greenrd: you didn't use the installer, did you? <greenrd>civodul: no, the installer fails on Qubes because on Qubes I need to set a static IP configuration <greenrd>Is there an example of xorg-configuration that sets a keyboard layout using a string? <MYMYRSRS>In this case, I may provide little help, because I do not have a server, <MYMYRSRS>There are many universities or server manufacturers in China that provide open source mirroring services. I think they may be able to contact them <bricewge>civodul: I've got the issue on both 4G and ADSL but not when using Orange's 4G <civodul>perhaps you could file a support request, just in case <greenrd>I tried ((symbol-fref 'keyboard-layout) "gb") but that didn't work either - I get Wrong type to apply: #f <davidl>When Im running guix import pypi pydotplus Im getting the error: "X.509 certificate of 'pypi.org' could not be verified: signer-not-found invalid" any ideas whats up with that? <cbaines>davidl, works for me. What do your SSL_CERT_DIR and SSL_CERT_FILE environment variables look like? <g_bor[m]>I have checked the situation in virt-manager. <g_bor[m]>The freeze on restart goes away when not using qxl but vga as graphics driver. <davidl>cbaines: SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt and SSL_CERT_DIR=/run/current-system/profile/etc/ssl/certs:/etc/ssl/certs <cbaines>davidl, OK, and does the $SSL_CERT_FILE exist, and does the $SSL_CERT_DIR contain files? <g_bor[m]>It still does not halt or reboot, it locks the screen instead. <g_bor[m]>But I believe that is a confgiuration issue. <davidl>cbaines: yes. However, I changed these to the profile-specific equivalent ones ~/.guix-profile/.. and now it works <davidl>so maybe the system-wide ones are outdated. <g_bor[m]>Currently I don't have a hipervisor with kernel 5.2+ to test the fix. <g_bor[m]>I assume this rarely comes up in guix too often, as most of our vw-s are run without qxl, or not managed from the libvirt reboot api. <greenrd>I think I need to call the procedure with @. How do you call a procedure with @ in Scheme? <sneek>Welcome back laokz, you have 1 message! <sneek>laokz, xavierm02 says: There is a nix package for guix, and then you can use nix normally. And there's guix import nix <greenrd>I tried a few ways, and I keep getting "invalid field specifier" errors <ATuin>to send a patch should i open a bug first sending an email to the bug-guix ml or just to the guix-patches one? <greenrd>The Guile reference manual says "To call a specific procedure from a given module, we can use the special form (@ (module) procedure)" <thvk-ivgf>greenrd: please paste (paste.debian.net) your file <cbaines>ATuin, if it's a single patch, then just sent it to guix-patches@gnu.org <cbaines>ATuin, if you have a series of patches to send, email guix-patches@gnu.org, then send the patches to the allocated bug address <ATuin>mmm actually i think it should be several, i'm adding a new function to asdf-build-system and then i need to call it in the packages defined in lisp-xyz <ATuin>so i guess it should be one patch per file, right? <raghavgururajan>Folks! How do you pass "no. of cores" as an arguments to `run-vm.sh` script generated by `guix system vm`? <ATuin>cbaines: ok, then i guess i need to open a new bug first :D <usney>How much space goes guix take up on a foreign distro? Just asking cause I might install it along side debian arm64 with limited space. <usney>I mean if you just install a few programs with guix with limited to no dependencies <cbaines>raghavgururajan, if you look in the run-vm.sh script, you'll see it's calling something like qemu-system-x86_64. If you run that with --help, you'll see the options. I think you want something like -smp cpus=n <cbaines>usney, guix size for one of the profiles created by guix pull says ~640MiB for me, so expect that guix itself will take up a bit of space. <civodul>note that this always uses XKB keyboard layout specs, both for the console and Xorg <cbaines>It mentions: all build users are currently in use <cbaines>I'm not sure if that's on the head machine, or on the machine it was trying to offload to <greenrd>civodul: yes, there are no examples of setting a different keyboard layout for X on that page <guixer>I'd like to see which packages I have installed via `guix install' and which I applied via manifest. Is there a simple way to do that? Or do I need to write a script for that? <greenrd>I mean, it's not really different, but it has a different name for some reason, maybe political <greenrd>"In the example above, for GRUB and for Xorg, we just refer to the keyboard-layout field defined above, but we could just as well refer to a different layout." <g_bor[m]>I believe this is not meaningful as stated. When you instantiate a manifest into a profile, then exactly the packages in the manifest will end up in that profile <lispmacs>hi, is there a metapackage that contains all the packages in the gnu build system <lispmacs>so you could do something like guix environment --pure ad-hoc gnubuildsystem <g_bor[m]>You can then do install, and thing like that, but once you apply the manifest again, it will get back to the state descibed by the manifest <thvk-ivgf>g_bor[m]: and `guix` can't check list of installed packages agains the manifest? <guixer>g_bor[m]: So the manifest applies the desired state for the profile? All packages installed via `guix install xy' are not available in the profile anymore? <res453>It doesn't seem to be gpg signed or includes a sha <res453>should I be using the asdf-build-system/source procedure <civodul>greenrd: oh right, sorry; to avoid the name clash, move the 'keyboard-layout' field of 'operating-system' below the 'services' field *civodul fixed embarassing issues in the Shepherd <ATuin> res453 have you experience with ASDF? <res453>I know it is the common lisp build system but I am a newb in common lisp <res453>I can get around reading docs about asdf though <res453>I would like to package cmn to then be able to package slippery chicken <ATuin>ok, i have been reading the code for asdf build system last week since i found some problems with stumpwm <res453>slippery chicken also depends on cm2 (i.e. common music 2 <ATuin>anyway, i think the asdf-build-system/source should work as long as you configure your ASDF system to look for it <res453>it's a really cool package. It would be great to be easily able to install this cl package via guix <res453>the ultimate goal is to package slippery chicken <res453>slippery chicken works with clisp and other implementations of common lisp so it might be nice to add the guix "use-flags" for those alternative common lisp implementation in the guix package definition <res453>Would someone be able to mentor me on this? It would be greatly appreciated! <ATuin>i'm quite new in cl but if you use asdf-build-system/source it will be independent of the implementation <res453>How do I deal with a package that doesn't have a gpg signature or sha file in guix? How should I package that? <ATuin>that build systems seems that just copies the sources so if the cl that you want to use supports ASDF it should be fine <ATuin>i guess you want to use git dunno <greenrd>civodul: now I get In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #<procedure keyboard-layout ... <TZander>thats ok, you just package it and put the sha256 in that you can calculate with guix download <greenrd>civodul: I don't think moving it is going to work, because other stuff depends on it... wait, unless I move them too. <res453>I did that and got 1f4ivhg1vq2fwklmwjj30bj7ij1yab45rnjjwps3qpsh0lpd6yj3 <ATuin>exactly, that way you compute the hash, then try to use the /source variant of asdf <res453>so just use that for the base32 sha? <ATuin>that should be the easiest path <TZander>if there is no gpg, then this is a good fallback <ATuin>well if there is not gpg signature it's hard to verify it :D <res453>Sounds like this one should be pretty easy then <res453>Would this common lisp package be acceptable as an upstream guix package given that? <rekado_>the only thing missing for rsync access to the cache is a fix to “guix publish” <rekado_>it generates files with very restrictive permissions. <rekado_>sneek: later tell MYMYRSRS you can’t ping ci.guix.gnu.org because the server doesn’t respond to pings. <greenrd>civodul: ok, you were right - I just used the xkb name for both - gb - and it seemed to work. I didn't realise that guix would automatically translate from the xkb keyboard layout name into the console keyboard layout name. <TZander>res453: there are many people that slowly give up on gpg as it becomes clear that there isn't really any benefit to the general public. You downloading it and putting the sha256 into the guix git history means that future changes of this version will be detected. <TZander>res453: I expect that the package will be accepted. <rekado_>we don’t require source archives to be gpg signed. <rekado_>if it’s free software it will generally be okay for Guix. <res453>rekado: thanks for that info. yes it is libre/free <res453>Ok, I will import a template and try to get the above to packages to build in my pure environment <res453>One other question. The author of slippery chicken is bundling the source code of cm-2.6.0 with his latest slippery chicken release <res453>should I modularize cm-2.6.0 as a separate guix package? <res453>cm-2.6.0 is a separate package on its own <janneke>to rebuild automake, we need to rebuild cmake...hmm? <TZander>janneke: that would be funny, but sounds like a bug. <glat-agent>Is GNU/Linux Activation Technologies available on Guix? <TZander>glat-agent: ae you being funny? Guix if free software, which means its gratis. <bandali>TZander, side note: being libre doesn't necessarily imply being gratis <ecbrown>Kimapr: awesome! i look forward to playing it <bandali>also, i'm like 99% sure that's a troll <sirgazil>I just found that Chris Marusich sent a patch a long time ago for supporting SSH authenticated repositories in package definitions, but the discussion was dropped. <TZander>bandali: I loved the statement at fosdem that everyone would have been saved so much trouble if 'free-software' would have been initially called 'freedom-software'. <sirgazil>I'd really like to know if it is really possible to support this in the future. <bandali>TZander, :-) that rings a bit more awkward to my ears, but yeah i guess <bandali>ideally, english would have separate words for libre and gratis <bandali>like french, german, persian, and many other languages do <TZander>sirgazil: maybe its best to make a copy of a git repo on a public place if you are interested in packaging software that is behind a password. <TZander>sirgazil: or, in the usecase of one of the commenters, just do a checkout in your private dir. guix packages work fine with file:///home/bla/mysoftware <sirgazil>TZander: I don't see how that would make the private packaging and distribution easier. <TZander>you distribute the build instructions and have a simple 2 line script where the first does a clone of your private repo into something like your homedir. <res453>what liscense argument should I define in the definition? <sirgazil>TZander: I meant, "Thanks, but that's not what I'm looking for"; not "Sorry, I'm not interested in what you have to say" (I'm not a native English speaker). <TZander>sirgazil: no problem, hope you find a solution. <sirgazil>Well, if this can't be included in Guix, I guess Marusich could package "git-fetch/impure" and distribut it as a Guile module, unless I'm missing something. <TZander>sirgazil: feels like calling it 'impure' is missing the point. All you need is integration with the ssh-agent. <TZander>have to say, I have no knowledge of the guix internals. I packaged 2 things, thats it <ThomasAlbers>Im trying out guix (the distro) and I was wondering if someone could help me <ThomasAlbers>I tried reading the documentation, but i wasnt able to do what im trying <ThomasAlbers>I would like to install bspwm but i dont want to use a login manager like gdm. Is that possible? <sirgazil>It should be possible. People here do that, but I don't have an example, so will let others help you :) *sirgazil looks for examples in Guix source code... <sirgazil>ThomasAlbers: Did you already installed the Guix System (the GNU distro)? <ThomasAlbers>I did not choose a desktop enviroment since I did not like the options <lispmacs>hi, I'm working on a package definition, but the source is inside a tar.bzip2 file that is inside a folder that is inside a tar.gz file <lispmacs>what should I add to my package definition to handle that? <pkill9>lispmacs: you'll want to replace the 'unpack phase probably <ThomasAlbers>is it possible to reboot and shutdown without being root? <pkill9>ThomasAlbers: i think `loginctl reboot` <lispmacs>pkill9: does somebody know of a similar example offhand? <pkill9>ThomasAlbers: it's part of elogind, maybe you're not using the elogind service? <ThomasAlbers>also a irc related question, when you reference my name and it appears with a color, is it a special command or erc just detects a user name <pkill9>I just press tab and it types your name (im using quassel client) <ngz>ThomasAlbers: it's an ERC feature (probably common to most IRC clients). See `erc-current-nick-highlight-type'. <ThomasAlbers>ah, ok. I don't use irc often, so i thought it was some kind of command <bandali>there's also an external emacs-erc-hl-nicks package, if you'd like to try <ThomasAlbers>yes, but is just an alternative to gdm, i would like to have just xinit <ThomasAlbers>I guess I'll just install gdm and play around with guix until I gain the required experience to try other options <plstohelp>Blackbeard- of course I tried w/o wayland, the point isn't just functionality, the point is to learn the inner workings by expiramentation <plstohelp>also im going to want wayand when I install it on an actual machine in the future <plstohelp>this is what I did, and I'm not sure what I'm doing wrong <ngz>jayspeer: yes, it is called guile-next <Blackbeard>plstohelp: I was asking if it worked without Wayland <jayspeer>ngz: thanks; I've been meaning to ask about "-next" naming scheme, ...why? emacs-next, guile-next... when will the next suffix on a release be dropped? when it's considered stable? <ngz>I think it is different from emacs-next, and guile-next. emacs-next is an unstable version of Emacs. There will probably always be an `emacs-next'. <ngz>OTOH, I think there is a guile-next because Guix is not fully ported to Guile 3, i.e., renaming guile-next to guile would move everything package using "guile" as an input to guile3, and it may not be possible yet. <ngz>Now, take it with a grain of salt, I may be missing the point :) <TZander>jayspeer: I'd say its the same as asking when people stop(ed) referring to 'python3' with the number and just packaged it as python. <jayspeer>I have no problem with python{2,3} naming, I think it's more readable than adding "-next" to package. I want to run emacs 27, but when that goes to stable, and 28 goes to -next, I might get some stuff broken after updating :/ <jayspeer>I believe the problem is that guix automatically installs latest version. On gentoo they mark package as stable/unstable/etc and by default the latest stable is installed. But I believe it's not currently possible on guix <lfam>jayspeer: You might get broken stuff with Emacs 27, too. That's why it's called "emacs-next" <ngz>jayspeer: Guix installs what you told it to install, i.e., you can install a specific version of a package. However, if you don't, you're right that the latest release is used. <bandali>i think installation of packages is currently broken for emacs-next <pkill9>im decent, how are you Blackbeard? <bandali>leoprikler and apteryx put some work into that; it's still WIP afaik <lfam>If we want a package for some reason but it's not fully working, we have to call it something else to reduce bug reports from people who wanted something that would work reliably <jayspeer>what I've been meaning to say -- guix install emacs should install latest stable, so 26.3, but guix install emacs@27.1 should install the newer one. But I think it won't work for now <lfam>It's not perfect but I think it's okay given the issues with the Emacs 27 package <bandali>that should probably only be done for stable releases <lfam>If it's not going to work right, I think it's okay to make people jump through hoops for it <bandali>and besides, from what i understand, guix has a policy of not keeping multiple/older versions around, unless in exceptional cases <Blackbeard>thvk-ivgf: seems like the games are hard coded right? <lfam>Right, bandali. There has to be some good reason for it <lfam>And some effort behind it, to maintian it <thvk-ivgf>Blackbeard: no, just the definition is messed up, and two games need some writable dirs. <lfam>It's not impossible to create a way to mark packages as "stable" or "unstable" <ngz>We can also mark packages as hidden. <lfam>I've been thinking about related issues lately. But I wonder if it's a good idea to put "policy" of that kind in the package definitions, especially since people may take Guix and build another distro with it. We already went through something similar regarding difficulties building certain packages on our previous build farm <lfam>We couldn't build some packages, but other people could build them just fine <lfam>Why mark them as something that must be built locally in that case? <ngz>Is there something wrong with the CI right now? I updated a package (xournalpp) two days ago, and it doesn't appear in ci.guix.gnu.org. <cbaines>and according to data.guix.gnu.org, there should be a substitute from ci.guix.gnu.org <cbaines>what bit of specific information do you mean? <ngz>cbaines: the page you pointed me at. How to reach it? <ngz>cbaines: (sorry for this naive question) <Blackbeard>thvk-ivgf: you can basically learn how to do everything in that patch haha <thvk-ivgf>Blackbeard: some of these games even, despite their age, have flags to specify the alternative data locations. <ngz>cbaines: Also, can you get the build output from that page? <leoprikler>tbh I'm not actively working on Emacs 27, I just passively help people when some issue arrives that I can solve <cbaines>Start at http://data.guix.gnu.org/ -> Click "master" (the branch) -> Click "Latest processed revision" -> Click "View packages" -> Search for the package -> Click "More information" on the relevant package -> Click "Version history" -> Click "Include outputs" <jayspeer>does running "guix pull" as non-root user updates system packages on guix system? <cbaines>You can also copy the output and run: guix build /gnu/store/baaa3fnjkr8fsdvn4mh2ly7n2s9fsiv9-xournalpp-1.0.18 and assuming you've got substitutes enabled, it'll download it to your local store <cbaines>jayspeer, no, you'd need to reconfigure with an updated Guix to effect system packages <jayspeer>so will running guix system reconfigure /etc/config.scm cause my system to update? <jayspeer>Blackbeard: I want to update my system install <g_bor[m]>jayspeer: usually you can do either of these thing: as root do a guix pull followed by guix system reconfigure. <jayspeer>g_bor[m]: thanks, that's what I've been looking for <g_bor[m]>Or you can do a guix pull as user, then sudo -E guix system reconfigure. Any sudo invocation is that preserves $HOME <g_bor[m]>There actually is a bit of confusion as how to do things as root in guix. <g_bor[m]>Most people come from systems where being root or running a command with sudo has very similar effects. This is a bit different when using guix. <plstohelp>because i followed what the manual said with regards to sddm <plstohelp>it is still TRYING to boot gdm but then it hangs <plstohelp>all i did was run the sddm service with some settings, and replaced "desktop services" with "desktop services but without gdm-service-type" <Blackbeard>plstohelp: are you having a gdm problem on wayland problem? <plstohelp>What I want to happen- I want sddm to be my login screen, so i can then launch gnome on wayland from boot <plstohelp>what is HAPPENING is, it boots normally up until I see "started session c1 for user gdm", and then it hangs <plstohelp>if I run the reconf in a live environment, it actually DOES launch sddm, but i can't login <plstohelp>my orginal config, with gdm and gnome on x11, works a-ok <plstohelp>what i want is a configuration with sddm and gnome on wayland <plstohelp>on an unrelated not, is there a GNU reason for tor browser not being packaged? or is it just that no one has done it <lfam>plstohelp: Basically, you should use only Tor browser from the Tor project. It's considered bad form to package it independently because distros could introduce changes that would make its network signature distinct and make it easier to identify users <lfam>We do offer a Tor service at the system level <plstohelp>delete the cached data, then try applying the config.scm? <Blackbeard>plstohelp: yes, seems like that's what being done. Let's hope it works <plstohelp>failed to execute child process "gio-launch-desktop" repeated dozens of times <plstohelp>tries to start gdm (even though it shouldn't) and hangs <plstohelp>it shouldn't even be trying to turn on gdn because i removed the service from my config.scm <greenrd>The manual installation instructions didn't remind me to set passwords before rebooting, so I didn't, and now I can't login <g_bor[m]>I will have to check the mails, I have seen something around gio-lauch-desktop <greenrd>I tried to fix the problem by replacing GDM with SLiM, and running guix system reconfigure, but I get guix system: error: failed to connect to '/var/guix/daemon-socket/socket': Connection refused <greenrd>I am trying to do this from within a chroot, I booted from the install image, since rescue mode is rather basic <plstohelp>but im going to go eat and then try it after that i suppose <greenrd>herd start cow-store within the chroot also gives the same error, and herd start cow-store outside the chroot seems to work, but doesn't help <g_bor[m]>plstohelp: they say it was fixed on core-updates. <Blackbeard>greenrd: you don't need to start cow-store if you already installed it <greenrd>yeah, that's the solution I just figured out <plstohelp>so just put that under operating system-packages? <g_bor[m]>I did not do this myself ever, but the message triggerd my memory :) <plstohelp>i keep getting syntax errors in the coonfig file <plstohelp>keep getting "end of file in string constant" <lfam>plstohelp: It means that there's a mistake in your config.scm <lfam>Probably, you forgot to close a "quoted" section <plstohelp>i put (list glib "bin") in my global os packages <plstohelp>wrong type argument in position 1 (expecting struct)= "bin" <plstohelp>man, i really am going to have to sit down and read this manual cover to cover <jayspeer>can I build hurd image without cloning guix git repo? <g_bor[m]>do you have any idea rearding the issue plstohelp faces? <g_bor[m]>I guess you are better into the gdm stuff than I am. <g_bor[m]>I will try to have a look at those issues meanwhile. <vagrantc>i seem to recall efraim had some configs that worked for me <vagrantc>recently, i've just started with %base-services instead of %desktop-services and adding what i need <vagrantc>rather than removing things i don't need <plstohelp>my config.scm has "desktop services, without gdm" <g_bor[m]>vagrantc: I am also doing that for a long while. <plstohelp>also the reason my reconfs kept rebuilding was because a kernel update released in the eman time <plstohelp>the sad part is gdm is SUPPOSED to have wayland option <g_bor[m]>plstohelp: actually the forward looking soultion would be to add proper wayland support to gdm. <plstohelp>iirc gdm is supposed to have a config file where you can force wayland <plstohelp>or does that only force gnome into wayland and not gdm itself? <plstohelp>but i seem to be stuck with 5.4.31 on my working config <plstohelp>also the wayland session for sddm STILL gave the gio-launch-desktop error <lfam>plstohelp: What command are you using to reconfigure? <lfam>Are you doing that as the root user? Your own user? With sudo? `sudo -E`? <lfam>Did you also do `sudo -i guix pull`? <lfam>`guix pull` is per-user so you need to make sure you are handling that correctly <lfam>If you already did `guix pull` as your regular user, you can use it to reconfigure with `sudo --preserve-env [-E] guix system reconfigure /etc/config.scm` <lfam>Either that, or do pull *and* reconfigure as root <plstohelp>id do everything as user but i can't do some things <plstohelp>guix is the most fascinating piece of software ive seen in ages <vagrantc>yeah, sudo -E guix system reconfigure ... <vagrantc>that uses the user's "guix pull" to reconfigure the system <vagrantc>i haven't don't guix pull as root for a year or two now :) <plstohelp>how do i get the config with my installed packages <plstohelp>iirc guix can do that, like a more sophisticated version of pacman -Qqen > packages.txt <plstohelp>also i have a bug but I'm not sure if i should report it to guix or gnome upstream <vagrantc>plstohelp: i only do "guix pull" as the user. then i do "sudo -E guix ..." for things that require root privledges <plstohelp>so why is my pull not updating me from kernel 5.4.31 to 5.4.33, but a reconf did? <plstohelp>im so stuck in the old fashioned package manager mindset lmfao <vagrantc>not quite understanding the question ... they're entirely different operations <lfam>`guix pull` doesn't update anything you have installed or that is "in effect". It updates the list of available packages. <lfam>Similar to `apt-get update` <lfam>That's an oversimplification but close enough <sirgazil>It's about your OS config. Do you use a login manager or simply log in using that black screen log in your are presented with in TTYs? <sirgazil>I'm trying to set up a minimal system configuration that uses no graphical login manager and a WM (sway in this case)... <sirgazil>I already have an OS definition for that, but would like that after starting the computer, the system displays automatically a TTY where I can log in. <sirgazil>Currently, to log in, I have to press Ctrl+Alt+F2 and login there. <Blackbeard>sirgazil: I am using the default gdm with %desktop-services <vagrantc>i found at one point, even though i removed gdm from starting, it would still build gdm and gnome which was obnoxious <Blackbeard>sirgazil: check the example for "Scheme Variable: slim-service-type" it removes gdm <vagrantc>especially when there weren't substitutes <plstohelp>okay so i just did guix pull && guix upgrade <plstohelp>where did 5.4.33 even come from in the reconf <sirgazil>Blackbeard: Ha, that's the part I'm readying now :) <plstohelp>at this point, im just going to wair until x11 is no longer an option <vagrantc>it's a different packaging paradigm ... the system configuration is largely separate from what the user's have available <plstohelp>im tired of tabbing, does anyone have a recommendation for a good irc client in the guix package list? <vagrantc>there are lots of different ones that are good at different things *vagrantc likes irssi too as a commandline irc client <lfam>I use the paredit Vim plugin. It's good enough for balancing parentheses <plstohelp>there should be some place i can see the older version <lfam>It's up to the user to version it, plstohelp <g_bor[m]>The icon theme is also hardcoded in linphone? <vagrantc>in recent versions of guix, you can get a reference to the config used to build that generation with guix system list-generations <g_bor[m]>I belive that you can get the config from any system generation. <vagrantc>or guix system describe for just the most recent one <g_bor[m]>That means it is not versioned, but if you made a system from it, then you should be able to recover it. <plstohelp>what happens if you overwrote the same config file? <vagrantc>i've thought about putting my config in guix and setting up a git push hook to run guix system build <plstohelp>okay so, I have a bug where when I install a new program, it doesn't show up in gnome until i relog/reboot <g_bor[m]>I have quite an accurate guess what happens here. <plstohelp>i search in the gnome menu for icecat, no results <plstohelp>but then, on the next boot, i search for icecat and there it is <plstohelp>you know, the search bar in the activities menu <g_bor[m]>Most probably on other distros the files are placed in a path that is picked up by the environment when the dm is started <g_bor[m]>In guix every package lives in its own directory. <g_bor[m]>I do not now if it is possible at all to augment the searchpath of a running DM session... <plstohelp>it works on a reboot, just not on package install <g_bor[m]>It gets picked up from the shell initialization file the next time the session starts. <plstohelp>In that case, i will simply wait patiently for x11 to die <Blackbeard>You will need linux-Pam I am not sure it is on guix by default <thelounge>i am new to guix and would like to install packages via tor <rekado_>no, this is an extra command line argument <rekado_>you would use that to download package binaries over tor instead of downloading directly from ci.guix.gnu.org <civodul>thelounge: you first need to set up an HTTP proxy that'll redirect trafic through Tor <civodul>we should write about how to do that <civodul>ok, then you need to run guix-daemon under torsocks <joshuaBPMan>Also I'm trying to update our sway package by the way. <joshuaBPMan>I won't have it done anytime soon, but I'd like to jump sway up to 1.43 <thelounge>and after importing the key with command it still fails <thelounge>This script installs GNU Guix on your system <thelounge>[1587224925.973]: Starting installation (Sat 18 Apr 2020 11:48:45 AM EDT) <thelounge>[1587224925.978]: [ PASS ] verification of required commands completed <thelounge>[1587224925.988]: [ FAIL ] Missing OpenPGP public key. Fetch it with this command: <thelounge>gpg: key 090B11993D9AEBB5: 129 signatures not checked due to missing keys <thelounge>gpg: key 090B11993D9AEBB5: "Ludovic Courtès <ludo@gnu.org>" not changed <thelounge>gpg: key 090B11993D9AEBB5: "Ludovic Courtès <ludo@gnu.org>" not changed <plstohelp>so somthing about guix is REALLY throwing the gnome disk usage analysis tool for a loop <plstohelp>im already running out of space, but the disk usage tool can only actually find 600mb of data <plstohelp>what tools would you recommend for analyzing disk usage <civodul>thelounge: it should be "x86_64-linux", as mentioned in the manual <civodul>i recommend using the script, though <civodul>thelounge: did you import the OpenPGP key as root? <civodul>"wget ... | gpg --import -", all that as root <rekado_>thelounge: make sure that the user that runs the script is the same as the one that imports the key — both should be root. <Blackbeard>Minor detail I didn't think about. I always do $ sudo -s <Blackbeard>thelounge: please copy paste your last terminal commands on debían paste and share the link <noobly>is there an easy way to run android or iOS apps on Guix? <jayspeer> noobly: is there an "easy" way on any other distro? <noobly>jayspeer: it would probably be somewhat easier on a distro that has anbox prepackaged, but idk, I've never done it before <lispmacs>am working on a package definition. My build dies simply with message: 'command "autoreconf" "-vif" failed with status 127' <lispmacs>but I can run this command fine when I build manually from the source <lispmacs>is there some way to up the boot log verbosity? <Blackbeard>thelounge: ok please, seems like you are using sudo. Please run this <rekado_>lispmacs: you’ll need to add “autoconf” to the inptus <Blackbeard>gpg --verify guix-binary-1.1.0.x86_64-linux.tar.xz.sig <thelounge>gpg: assuming signed data in 'guix-binary-1.1.0.x86_64-linux.tar.xz' <thelounge>gpg: Signature made Wed 15 Apr 2020 07:43:14 AM EDT <thelounge>gpg: using RSA key 3CE464558A84FDC69DB40CFB090B11993D9AEBB5 <thelounge>gpg: Good signature from "Ludovic Courtès <ludo@gnu.org>" [unknown] <thelounge>gpg: aka "Ludovic Courtès <ludo@chbouib.org>" [unknown] <thelounge>gpg: aka "Ludovic Courtès (Inria) <ludovic.courtes@inria.fr>" [unknown] <thelounge>gpg: WARNING: This key is not certified with a trusted signature! <thelounge>gpg: There is no indication that the signature belongs to the owner. <thelounge>Primary key fingerprint: 3CE4 6455 8A84 FDC6 9DB4 0CFB 090B 1199 3D9A EBB5 <thelounge>Blackbeard: it is downloading guix-binary-1.1.0.x86_64-linux.tar.xz <jayspeer>so... I want to build a hurd vm image. I'm trying to it by following the blog post, however there are some things I don't understand <jayspeer>first - it's says the branch is merged, but the hurd.scm fails to build with normal guix <jayspeer>second - what is going on in the background? laptop fans are spinning like crazy, but the processes'es names are rather cryptic <jayspeer>second and half - can I somehow see what's going on in the background? -v 1 didn't help <thelounge>Blackbeard: there was core dump maybe because of torsocks <Blackbeard>jayspeer: how many CPUs ho you have? Maybe use more cores? <jayspeer>Blackbeard: whole 2 (4 with hyperthreading); how can I tell it to use more cores? <jayspeer>but I can see 4 processes with hurd.scm in htop, so imagine this is not the problem <Blackbeard>thelounge: what error did you get? Can you copy and paste in debian paste? <kkebreau>GNOME 3.34 patches from the wip-gnome3.34 branch have been merged with core-updates. Let the testing begin! >:-D <thelounge>Blackbeard: i closed the term and when I try to run the script again I get " [ FAIL ] A previous Guix installation was found. Refusing to overwrite" <plstohelp>how would one try core-updates instead of master <kkebreau>plstohelp: guix pull --branch=core-updates