*raghavgururajan peeps in <cbaines>nckx, out of interest, what are you building on bayfront at the moment? <nckx>LibreOffice & Linux-Libre. <cbaines>are you just building them to see if they build? <nckx>Yes, post updating libsomethingorother. <nckx>cbaines: Does seem like bayfront is less derelict than it used to be. Your doing? I'll have to find another place to send LO test builds :) <cbaines>well, I got it running an instance of the Guix Build Coordinator a few weeks ago, and it's building things as well <cbaines>I need to tweak the guix-daemon default --cores so that one build can't use all of them <nckx>Ah, no more playing idle head node to an (IME slower) milano. That's good. <nckx>(I didn't actually benchmark either, it was just a feeling.) <cbaines>testing builds is something I want the Guix Build Coordinator to be able to facilitate, ideally people would be able to just submit builds and then track the results, and the builds could be executed across many machines <nckx>That is basically how I use it. Having a nicer way to do so would be, well, nice. <cbaines>nckx, the build farm is currently bayfront, milano, harbourfront, monokuma (ARM) plus a Power9 machine of mine *nckx always forgets about harbourfront and will do so again. <cbaines>I'm waiting for it to surpass ci.guix.gnu.org in terms of x86_64-linux substitute availability, currently it's at 90.3% compared to 92.1% for ci.guix.gnu.org, so maybe by tomorrow it'll be ahead <cbaines>milano and bayfront both suffer from having hard disks rather than SSD's, but I mounted /tmp as tmpfs on milano, and I think that's helped <roptat>drakonis, I just sent my scala package to guix-devel, to make sure everyone is confortable adding an unbootstrapped package <hwpplayer1>Sorry friends I installed the System on a real hardware by selecting msdos option <sneek>hwpplayer1, you have 1 message! <sneek>hwpplayer1, nckx says: Yes, I think there are some known bugs. We know it to have some unknown ones too. Reports of either are very welcome, but please include exactly what you ran, what you did, and what happened: ideally with error messages & log files, or else a screenshot with all the information. <drakonis>roptat: okay i dont see scala 3 here but this will do <roptat>drakonis, yeah, I have a binary scala3, but missing many deps for rebuilding it, so didn't send <drakonis>and you can look that up with guix search <hwpplayer1>I installed it already drakonis but I am not sure how to start it like GNOME under X <terpri>hwpplayer1, i'd be surprised if exwm supported wayland. i *think* gdm can fall back to loading ~/.xsession files, which is how i'd expect exwm to be run (no special service listed in the manual) <terpri>should be enough to have, e.g., (services (append (list (service gnome-desktop-service-type) (set-xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout)))) %desktop-services)) <terpri>... as the "services" entry of your operating-system <terpri>(plus whatever other services you use) <hwpplayer1>As I understand i will modify .xsession in Home directory user hwpplayer1 ? <terpri>then you can test gnome and, if that works, write an ~/.xsession for exwm <terpri>yes, when you have exwm installed it'd be whatever the "standard" xsession for exwm is, most likely <hwpplayer1>I tried to write a lisp command series that you have mentioned but I can not run GNOME <terpri>hwpplayer1, did you apply the changes with "guix system reconfigure /etc/config.scm" (assuming you're using the standard location)? you might also need to reboot for gdm to start (shepherd can probably start it but i barely use herd) <hwpplayer1>Okay I am checking it guix runs some process righ now I need to wait <terpri>i don't see a service for exwm -- hypothetically exwm-desktop-service-type -- which is why i suggested you'd have to write an ~/.xsession <terpri>it's a bit tricky to get a "startx"-equivalent command in guix so you may have to experiment a bit with the ~/.xsession (but it looks like people mostly, in effect, just exec emacs within a dbus session, sometimes with gnome-control-center in the background?) <hwpplayer1>I can manage GNU Hurd I have read the manual for Debian Hurd ***iyzsong-- is now known as iyzsong-w
<terpri>oh yeah, it might take a while if this is your first reconfigure after installation (due to the big backlog of package updates) <terpri>hwpplayer1, it's a kind of patching, yes, but a pretty specific method and guix already had the concept of patch-files (to adjust source for guix compatibility, etc.), so probably just used to be unambiguous <hwpplayer1>guix system: error: symlink: Permission denied: "/var/guix/profiles/system-3-link.new" <hwpplayer1>guix system reconfigure /etc/config.scm running with root <roptat>hwpplayer1, when you start your computer, you should see the GDM login screen, there should be a cog wheel next to the password field where you can select GNOME or other window managers <terpri>hwpplayer1, that's...odd, almost sounds like a profile-management-related bug <terpri>i gtg, roptat will probably be able to help better :) <roptat>well, I'll be gone in ~20 minutes too <roptat>what did you reconfigure with last time? Can you share your /etc/config.scm? <hwpplayer1>;; This is an operating system configuration generated <hwpplayer1>(use-service-modules desktop networking ssh xorg) <djames>ZNC will surely kick me because it thinks the chat is being spammed <hwpplayer1> (list (uuid "1b37da7a-f95f-457d-8650-35c7ffd000ed"))) <hwpplayer1>I do not have gui which is why i had to paste it directly <roptat>understood, can you try and use wgetpaste next time? (from the same-named package) <roptat>hwpplayer1, so you're missing desktop services, you'll need at least to change %base-services to %desktop-services to get GDM <roptat>and then run "guix system reconfigure /etc/config.scm" as root <hwpplayer1>Where to write %base-services to %desktop-services ? <roptat>in /etc/config.scm, I see you have %base-services, replace it with %desktop-services instead <roptat>the base services are the bare minimum to get a running system, it doesn't contain the necessary services for a desktop computer <hwpplayer1>guix system: error: service 'networking' provided more than once <roptat>hwpplayer1, ah, you need to remove the dhcp-client-service <roptat>your services field should be something like (services (append (list) %desktop-services)) <roptat>(to be faire appending an empty list is not necessary, but just to preserve the structure :)) <roptat>%desktop-services brings in a lot of services, including a networking service that conflicts with the simpler dhcp-client-service <roptat>I'd appreciate a feedback if you have time to test and give your opinion <drakonis>i'm going to fiddle with the gzdoom package because i got the cravings for some demon slaying <hwpplayer1>It is working somehow the guix system reconfigure /etc/config.scm command <roptat>hwpplayer1, hm... are you secretly a French speaker? ^^' <roptat>oh, we have a lot of French speakers here (I'm one of them) and the grammar of your last sentence looked familiar, sorry I got it wrong ^^' <hwpplayer1>I just improvise I do not focus on native American in the meantime <roptat>but that explains your timezone :D <roptat>yeah, you have to be patient, guix is often very slow <roptat>we're working on it and have made a lot of improvemets recently, but we're still not there yet <hwpplayer1>Yes but It is better to have a GNOME first right ? <roptat>you don't have gnome in your current setup <roptat>you don't need it if all you want is EXWM <hwpplayer1>building /gnu/store/cz6da7vdy7wiiiybycpfgswak4hqnwx7-guix-1.3.0rc2-checkout.drv... <roptat>did you run "guix pull" as root first? <roptat>can you cancel the build (control-C), and run "guix pull" as root first, then reconfigure again <roptat>right now you're downgrading guix to rc2, which is before 1.3.0 release <hwpplayer1>building /gnu/store/cz6da7vdy7wiiiybycpfgswak4hqnwx7-guix-1.3.0rc2-checkout.drv... <roptat>(the issue being that guix itself can only contain a package of a less recent guix, so unless you run guix pull, you're downgrading all the time) <roptat>please cancel that and run guix pull as root <hwpplayer1>hint: After setting `PATH', run `hash guix' to make sure your shell refers to `/root/.config/guix/current/bin/guix'. <roptat>ideally, check that "type guix" refers to /root/.config/guix/current/bin/guix <roptat>if not, try "hash guix" (no output is expected), and "type guix" again <hwpplayer1>guix is hashed (/run/current-system/profile/bin/guix) <roptat>that's incorrect, so run "hash guix" and try again <hwpplayer1>guix is hashed (/run/current-system/profile/bin/guix) <roptat>hm, the guix system is supposed to set the environment for you... <hwpplayer1>guix system reconfigure /etc/config.scm should i run that ? <roptat>then try "export PATH=/root/.config/guix/current/bin:$PATH; hash guix" <roptat>no we need to make sure guix refers to the right location, or you're going to downgrade again <hwpplayer1>export PATH=/root/.config/guix/current/bin:$PATH; hash guix okay I did <roptat>what does "type guix" tell you now? <hwpplayer1>guix is hashed (/root/.config/guix/current/bin/guix) <roptat>now you can run "guix system reconfigure /etc/config.scm" <roptat>yeah, you can't acccess /root as unprivileged user <roptat>hwpplayer1, it won't be faster but at least you won't be downgrading ^^' <roptat>good luck, I'm tired now, so I'm going to bed, see you later! <hwpplayer1>ice-9/boot-9.scm:1685:16: In procedure raise-exception: <drakonis>hmm, does git-fetch offer the ability to get submodules? <apteryx>yay, got disarchive built for i586-pc-gnu <apteryx>some new Guile library added to Guix, to help dealing with SWH archives I think <drakonis>aha i found the option to do recursive fetching <apteryx>it failed to cross-compile for Hurd, but luckily this being Guile it was easy (just fine tuning its configure.ac a bit) <drakonis>but wait a minute, there are no submodules in gzdoom ***iyzsong-- is now known as iyzsong-w
<drakonis>aw yeah i successfully built gzdoom 4.5.0 ***devnull[m]3 is now known as p10r[m]
<efraim>the oom-test for guile-3.0.7 is still failing on powerpc-linux-gnu <vivien_>I don’t understand why, and I don’t know where the log file is <vivien_>Do you know where I can find the log file? <efraim>vivien_: what architecture are you using? <efraim>guix build --no-grafts guile --log-file <vivien_>I mean, I can’t reproducibly build guile on my machine because of the failing test, I would like to see the log of the test to understand why <efraim>the log of the test isn't printed when it fails, and gets deleted when the build finishes. You'll have to build with with '-K' to keep the output directory when it fails <efraim>hmm, I would think it'd be in test-suite/standalone <vivien_>And of course, now it passes the test <vivien_>OK, so since a few days ago I get an error with guix pull: no code for module (system repl error-handling) <efraim>I think it was just fixed, go ahead and try pulling again <vivien_>efraim, no luck, the error is still there <vivien_>On my other machinne, the error is slightly different: no code for module (rnrs io simple) <brown121407>When running a Guix command that needs to download some substitutes, why does it display "substitute: updating substitutes from..." multiple times? Is the substitutes info downloaded in chunks? <leoprikler>it's downloaded as-needed and gets called for multiple items <meo>so I install trivial packages, like tmux or some fonts, and guix downloads 100+ packages, including e.g. glib, mesa and a bunch of others, which are then immediately removed by guix gc. Is this a mistake in package dependencies? <rekado_>does this also happen when you use “guix environment --ad-hoc tmux”? <rekado_>if you install things to an existing profile, Guix may need to build some profile hooks <rekado_>and depending on what packages the profile contains those hooks can require quite a bit of software <rekado_>this is also part of the reason for why Guix may download packages when you ask it to *remove* packages from your profile. <meo>cbaines: my concern isn't that it removes something it shouldn't, rather than the dependency graph blows up unreasonably and it seems like the stuff it downloads is actually irrelevant <meo>let me look at tmux dependency graph <meo>i just tried installing shotwell and it pulls in samba, liboauth and gnupg <meo>i'm clearly missing something <nckx>meo: shotwell requires libsoup which requires samba. <nckx>Same for, e.g., ‘guix graph --path shotwell liboauth’. *nckx goes to get headphones. *nckx agrees with roptat's reply. <tissevert>we've gotta start a band to make the release songs : ) *nckx updates some softsynths to celebrate. <tissevert>great ! do you play the synths ? I tried to at a time but I always get lost, I browse sounds for hours and never get anything done <nckx>tissevert: That's very recognisable. I can play with them for hours, journeying to very strange places, but it's purely in the moment. I have to get behind a piano to get anything ‘done’--which is seldom the point. <liltechdude>Hello! Why after installing `guile-chickadee`, chickadee does not installed in /run/current-system/profile/share/guile/site/3.0? <vivien_>You need to update your GUILE_LOAD_PATH to also include ~/.guix-profile/share/guile/site/3.0 <liltechdude>folder with this lib placed in /gnu/store/y9ndnfww24pic7p8lblr1276vff2jw8p-guile-chickadee-0.7.0/share/guile/site/3.0/chickadee <vivien_>So if you log out and log in back again it should be set up correctly <nckx>I think you also need Guile to be installed in the same profile (so if you ‘guix install guile-chickadee’, ‘guix install guile’). <vivien_>Or you could set up one terminal by doing . ~/.guix-profile/etc/profile <nckx>/run/current-system/profile is for your Guix System packages, you can't ‘guix install’ anything to it. <vivien_>You can also use guix environment: guix environmennt --ad-hoc guile-chickadee <vivien_>This will get you to a prompt where the path is set up correctly <tissevert>(wouldn't you need to include guile in that `guix environment` case to get the paths handled correctly ?) <nckx>vivien_: It won't unless the profile also contains guile. <liltechdude>thanks, I install guile (how strange, guile as executable was available) and reread profile <tissevert>liltechdude: it's more complicated than «things are installed» or not <nckx>liltechdude: Guile is in a different profile, probably /run/current-system/profile/bin/guile. Guix won't ‘see’ it when creating the profile that contains guile-chickadee, and will not apply its native-search-paths. This is... an inelegance of the current system, for sure. <tissevert>preparing an environment depends on the packages asked for in the environment, not on the existing «installed» packages <vivien_>That’s true for pure (and container) environments <nckx>vivien_: The above is true for all. <liltechdude>yes, some people in this chat write me previously that some program need to "installed" to change global variables <vivien_>I mean, I can run guix environment --ad-hoc indent -- blender <nckx>vivien_: This is about search paths and similar environment variables like GIT_SSL_CAINFO, not about what's in PATH. <meo>rekado_: when you said profile hooks, I assume you meant guix user profiles and package installation/removal triggers their update which may pull its own dependencies <nckx>Some variables like GUILE_LOAD_PATH, PYTHONPATH are requested by the ‘consumer’ of the variable (so Guile, Python, ...). So only Guile needs to declare ‘I use GUILE_LOAD_PATH, and this is how to construct a valid one’, instead of *every single* guile-foo package having to do so, or needing a global hook for each new variable. But if Guix doesn't see a consumer in the same profile as a provider, it doesn't have that information, and it doesn't <nckx>So it's an elegant system in some ways (it's definitely clever) but it has its drawbacks as well. <nckx>Mostly that people don't expect it to work [that way] :) <tissevert>not bad, fixing some obscure test error in a haskell package : ) <hwpplayer1>Can I install Guix system for example on Ubuntu to delete or remove Ubuntu ? <hwpplayer1>I mean with script I will run Guix rather than bare metal Ubuntu <vivien_>You need to install it in an unmounted root <TheAsdfMan>Hello, I get an error while reconfiguring my system in the latest commit, i was on derstrumpf leomd leoprikler lewo` Lightsword <TheAsdfMan> liltechdude link2xt Linus[m] lispmacs lle-bout[m] <TheAsdfMan> lukaso666 luke-jr m1kr0[m] m6locks madage madalu <TheAsdfMan> MatthewAllan93 maxzor mbakke mekeor[m] melg8 meo <vivien_>Basically, in your ubuntu as a foreign distro, you can do the same steps as the system installation with guix system init and such <vivien_>But you need to install it in a separate partition that is not mounted <TheAsdfMan>Hello, i got an error while reconfiguring my system in the latest commit, i was on 0d85c8, The error is on b905ab <hwpplayer1>I did not see any issue now but maybe in 10 minutes <vivien_>hwpplayer1, I don’t know, maybe there’s a mafia of french people at the top of guix <hwpplayer1>mafia is mounted to machkernel which runs top of oldie UNIX :D <vivien_>Also that’s my theory but "guix" ending with "-ix" reminds me of asterix <vivien_>Sooooo maybe the whole project is a great display of chauvinism *hwpplayer1 trying to reconfigure <TheAsdfMan>Also, when i do a guix pull --roll-back i can reconfigure normally <ss2>TheAsdfMan: I'm getting the same errors too. <hwpplayer1>World is crying clouds are shouting Same Errors!!! <ss2>is there a bug report so far? <hwpplayer1>How can I mount a usb disk device to format the system ? <nckx>(Where /dev/sdXX is very probably /dev/sdc and not /dev/sdc2.) <ss2>TheAsdfMan: reported <nckx>You shouldn't dd to (or mkfs) a mounted file system. <nckx>vivien_, hwpplayer1: Yes, Guix is a French distro maintained by the French mafia, this is correct. <vivien_>That’s why it has all the ocaml stuff <roelj>Is there a package recipe that overrides gcc for clang in the gnu-build-system? *hwpplayer1 is pulling b905abf <nckx>roelj: CC=clang doesn't work? <nckx>I don't know the direct answer to your question. <hwpplayer1>nckx: I broke the only usb storage which I can format the system <vivien_>Also please make it work with crosspcompilation <nckx>hwpplayer1: How did you break it? Physically? Everything else can be fixed. <nckx>Otherwise, I suspect the French mafia had something to do with it. <nckx>vivien_: Would the approach to that be the same as for GCC, invoking different clang-$HOSTs for different systems? <roelj>nckx: No.. I put it in an extra phase before the 'build phase containing: (setenv "CC=clang"). But no luck. <nckx>roelj: Depending on the build system it might work as a make- or configure-flag. <roelj>nckx: Thanks for the pointers <nckx>hwpplayer1: I'm not sure what you're asking. <hwpplayer1>hint: After setting `PATH', run `hash guix' to make sure your shell refers to `/root/.config/guix/current/bin/guix'. <nckx>roelj: Which package is it? It doesn't sound like the gnu-build-system is involved. It doesn't choose the compiler. The package's own build scripts do. Guix just puts a gcc in $PATH, and will put clang there if you add it as a native-input, too. <hwpplayer1>guix is hashed (/run/current-system/profile/bin/guix) *hwpplayer1 is running reconfigure <nckx>hwpplayer1: That's it. You ran ‘hash guix’, and you've now made sure your shell is using the right Guix. Nothing more. <roelj>nckx: It's ldc-1.10.0, which has been failing to build for a while and I am trying to fix it. <hwpplayer1>hash is like cryptographic process like sha256 ? <nckx>roelj: It's gone Clang-only? :-/ <roelj>nckx: Hopefully not, but I want to try out building with clang. <nckx>hwpplayer1: Yes. Guix uses sha256 by default. We use slightly different characters for the base32 encoding but it's just plain sha256. <hwpplayer1>building /gnu/store/cz6da7vdy7wiiiybycpfgswak4hqnwx7-guix-1.3.0rc2-checkout.drv... <nckx>hwpplayer1: Can you ^C that? Something's not right. <nckx>I didn't notice your /run/current-system above. Does `/root/.config/guix/current/bin/guix' not exist? <nckx>Before the ‘hint:’ above. <hwpplayer1>guix system: warning: Consider running 'guix pull' followed by <hwpplayer1>'guix system reconfigure' to get up-to-date packages and security updates. <nckx>Did it not tell you do do something involving GUIX_PROFILE=...? <hwpplayer1>I can understand what you say but I am not familiar with package manager and system <nckx>It is strange that your PATH is messed up. ‘guix pull’ should just work out of the box on Guix System. What does ‘echo $PATH’ say? <hwpplayer1>/run/setuid-programs:/run/current-system/profile/bin:/run/current-system/profile/sbin <nckx>This is an installed system, not the ‘live’ ISO installation environment, right? <nckx>hwpplayer1: Could you execute ‘. /etc/profile’? (That's a dot.) <nckx>That should add /root/.config/guix/current to $PATH. <hwpplayer1>/run/setuid-programs:/root/.config/guix/current/bin:/root/.guix-profile/bin:/run/current-system/profile/bin:/run/current-system/profile/sbin <nckx>hwpplayer1: Yep, ‘source’ and ‘.’ are synonyms, but ‘.’ is more compatible (‘source’ is bash-specific) and, well, shorter :) <nckx>hwpplayer1: OK, now you can reconfigure. <nckx>And it should not start building 1.3.0rc2. <nckx>Run hash guix again, first, though. <hwpplayer1>ice-9/boot-9.scm:1685:16: In procedure raise-exception: <hwpplayer1>guix is hashed (/root/.config/guix/current/bin/guix) <nckx>No problem, I do hope that error goes away 😛 <nckx>OK, that's the right Guix. <hwpplayer1>ice-9/boot-9.scm:1685:16: In procedure raise-exception: <nckx>Well fuck. hwpplayer1: It depends on what's actually going on. Could you paste the entire backtrace to paste.debian.org? <vivien_>Can I get the output of a process invoked with invoke from (guix build utils)? <nckx>‘You could just install--’ oh, right. hwpplayer1: If ‘guix install’ still works, guix install wgetpaste. <nckx>then try something like ‘guix pull 2>&1 | wgetpaste’. <nckx>hwpplayer1: After that, could you log out & back in & try to pull again? It might be due to some stale variables left over in your session. <ss2>nckx: I just reported a bug, if it is the same issue. Current checkouts can't reconfigure the boot loader. <nckx>ss2: It's an upstream bug on master? Oh. Thanks. <ss2>Reverting back to a checkout from two days ago works though. <nckx>I see your bug report (thanks!). <ss2>others higher up where experiencing it too, so it may well be one. <nckx>...that's not good. no code for module (system repl error-handling) <nckx>Was Guile recently updated? <ss2>nckx: 0ac95b4490f2b661370b29980f6f01d28d03db6a adds Guile 3.0 <hwpplayer1>hashed typed and reconfigure returns 1.3.0rc2-... *hwpplayer1 is running guix pull as a normal user unprivileged <ss2>are you trying to pull to current checkout? What fails then? <katco>is there some setting to the elogind service to set `/sys/power/pm_async` to 0? <hwpplayer1>Authenticating channel 'guix', commits 9edb3f6 to b905abf (17,347 new commits)... <nckx>Great, now I get it to, whilst running some system tests: no code for module (rnrs io simple) <nckx>hwpplayer1: That's OK. When you first pull guix to a new location (here: user), it will authenticate all GPG-signed commits. <nckx>Root is just a regular user here. And yes, I'm afraid so, until the cause is found & fixed upstream. <katco><nckx "katco: No."> ty. i guess i'll just write a one-shot service. <nckx>hwpplayer1: Try ‘guix pull --commit=df48a11e1b3e842f54603a6ef170e77aca47062a’. That will pull a (totally random) commit from some days ago that should work. <hwpplayer1>hint: Use `--allow-downgrades' to force this downgrade. ? <nckx>It's a signed upstream commit, it's just old. <hwpplayer1>I was returning the system to the old version when I was running Windows XP <nckx>If Windows had that wonderful feature sooner than Linux, good on them. <nckx>hwpplayer1: You just got very unlucky trying this today. It would have worked flawlessly yesterday, or indeed most other days. <nckx>(What more do you need? An Emacs on TTY2? GUIX CAN DO THAT.) <nckx>If pull succeeded, yes, go ahead and reconfigure. You might need to add an option there too, but you're not missing any cool new features. <hwpplayer1> guix-1.3.0-1.771b866 35.2MiB 511KiB/s 00:10 [## ] 13.8% as hwpplayer1 rather than root <hwpplayer1>To complete the upgrade, run 'herd restart SERVICE' to stop, <hwpplayer1>upgrade, and restart each service that was not automatically restarted. <hwpplayer1>Run 'herd status' to view the list of services on your system. <djames>After guix system reconfigure I get an error "no code for module (system repl error-handling)" <nckx>djames: Thanks. It's been reported. <nckx>hwpplayer1: You can, that's all I know. There's an emacs-exwm package mentioned in the manual. Never used it. <nckx>That example assumes you're using (services %desktop-services), so make sure you are. <hwpplayer1>how to change keyboard layout like setxkbmap tr ? <nckx>hwpplayer1: Under X, or on the Linux console, or...? <hwpplayer1>There is no X though I configured and Installed GNOME <nckx>hwpplayer1: sudo loadkeys trq (or tr+something else, I don't know what the q means). <nckx>hwpplayer1: How did you ‘configure and install GNOME’? For example, just running ‘guix install gnome’ won't pop up a GNOME desktop. However, if you chose GNOME during the installation (ISO) and only got a TTY, that's a bug. <hwpplayer1>nckx: I installed GNOME guix install gnome and someone help me to configure config.scm <nckx>If you share your complete configuration file someone might be able to help. I have to go o/ <tissevert>I don't see any obvious reason why gnome wouldn't start with your services configuration although I'm a bit confused you can declare both gnome and xfce desktop services without getting a warning <tissevert>(a quick tip though : the keyboard-layout setting lets you change your keyboard layout globally, and it should affect both your tty (so you won't need to manually loadkeys upon login and X session, when it finally starts) <hwpplayer1>Okay TTY is enough for a long time TTY Emacs Git SSH GPG GCC and programming languages and frameworks <tissevert>by hand, yeah, if you want to make it permanent you probably want to add an element to the (file-systems …) entry <tissevert>or someplace in /media/ like /media/hwpplayer1 <tissevert> /media is more «modern automated access with things like desktop services that dynamically create per-user folders» ***ece0 is now known as ece
***dongcarl7 is now known as dongcarl
<tissevert>a bit, yes, but I'm not very familiar with GCC or assembly, I'm more of a haskell and functional programming girl <tissevert>great ! : ) I love talking (and especially about functional programming) <hwpplayer1>I came here to run Guix because it is like GNU Emacs <tissevert>after all, they say Emacs is a good OS but only lacks a text editor, guix could be that editor ^^ <leoprikler>it was supposed to be boring (it's a play on SaaS) <hwpplayer1>I offer kyebo which has a meaning key book in Japan <tissevert>(but then people might say it only lacks a good OS and… well we don't want that) <hwpplayer1>And also we need to write that in English format <hwpplayer1>Could you please suggest me a system tool runs on terminal ? <hwpplayer1>For example on Ubuntu I run setxkbmap but here loadkeys <leoprikler>oh, you can use htop on Guix, just guix install htop <leoprikler>if you're in real dire situations, plain top might also work <tissevert>hwpplayer1: wait, no, loadkeys was about setting your keymap in tty, setxkbmap works as long as you're in a X session and have setxkbmap installed <vivien_>If you’re tired of htop, I’ve discovered bashtop but it doesn’t work well on light terminals <nckx>hwpplayer1: If by ‘system tools’ you mean ‘system monitors’ like htop, you could try s-tui, glances, or bashtop. <nckx>So many ways to burn CPU cycles to monitor your CPU cycles. <vivien_>It’s something in the line of cryptsetup luksOpen ... <nckx>hwpplayer1: You ‘open’ (unlock) the device with ‘cryptsetup open --type luks ...’, then mount the resulting /dev/mapper/<device> normally. See man cryptsetup from the cryptsetup package. <nckx>vivien_'s answer is also correct, they changed the syntax a while ago but luksOpen & friends are still supported. <hwpplayer1>I only worry about this is a new system for me not a standard Linux or GNU/Linux <nckx>If my test results are any indication, guix pull && guix system reconfigure should work again now. <nckx>I'm hopeful but confirmation is appreciated :) <djames>nckx: Confirming it works for me *nckx will never not forget something when pushing, sigh. <nckx>That's where it belongs. <tissevert>(out of curiosity when got it broken ? because I updated like sunday and haven't seen anything) <muradm>today all day I was not working for sure <tschaubless>Is there a way to add a WebDAV file system in the (file-systems ...) entry? If not, is there a more or less straight-forward alternative? <nckx>tschaubless: I don't think there's a way (yet), but davfs works from the command line. <tschaubless>nckx: cheers, luckily the command line option already works for me... out of curiosity (and lazyness to go through the manual) is there a way to append (or somehow modify) lines to a file, say /etc/fstab...? I'm talking about the system configuration file of course, using guile <MysteriousSilver>Hi! Running `guix upgrade` upgrades all packages installed. How do i only upgrade packages that aren't up-to date? <vivien_>MysteriousSilver, sometimes other packages get re-built because their dependencies changed; if so, upgrading only the dependency will not actually upgrade anything, it will just use some space on the store <Ikosit>Which package has the ldconfig executable? <nckx>Ikosit: It's not available in Guix. <nckx>Because it never makes sense. <nckx>It would be installed by glibc but we explicitly disable it. *Ikosit just found an issue about that <Ikosit><nckx "Because it never makes sense."> What makes never sense? <bdju>how would one go about disabling messages on TTY1? stuff about ntpd and all that <nckx>‘There doesn't seem to be an easy way to get the true list of directories to scan for shared libs, but this works for our particular cases.’ -- From Setup.hs. Except, of course, when it doesn't. <nckx>I don't speak enough Haskell to understand what they're trying to do, but it's based on some questionable assumptions. *Ikosit puts guix integration on his todo list <nckx>‘There is something like a true list of directories that contain all shared libraries, and we can get this list by peeking inside a system wide cache file used by the linker.’ <nckx>That would be wonderful, Ikosit. <Ikosit>there's already nix integration so it can't be that hard, can it? <nckx>From what I've heard Stack seems to be a pretty basic part of doing a Haskell, and missing it is probably bad. <tissevert>I know only cabal to manage haskell project, and when I came to Guix, I discovered I didn't really need it anymore <tissevert>because guix import hackage and then haskell-build-system and I never need to use cabal anymore <Ikosit><nckx "‘There doesn't seem to be an eas"> Which setup.hs? <nckx>It's in src/Stack/Setup.hs. <tissevert>I only installed it to generate fresh .cabal file for new projects and that's it <Ikosit>I don't know much about stack myself <tissevert>I've never understood exactly how stack replaced cabal and to what extent both were compatible <tissevert>can stack build a «cabal project» (if there is such a thing) <Ikosit>i think stack is a wrapper for cabal <tissevert>I know cabal's shortcoming have been severely criticized until Cabal released their «nix-style» build <tissevert>which seem essentially superseded by nix- and guix- packaging <Ikosit>cabal is the advantage ofr guix, that it has incremental builds (i think that's how they're called) <Ikosit>If you change only one file, cabal doesn't recompile your complete project <tissevert>oh, that ^^ yeah, that's right, now guix compiles everything all the time <tissevert>I thought I could live with it but now I'll want to fix it <nckx>I wanted to read the Nix package for Stack but I cannot find such a thing. <apteryx>civodul: if a shepherd service extends the account-service-type, are the new users and groups guaranteed to exist (as in, sync'd to disk) before the service that depends on them starts? <jackhill>(thanks to repology for finding it for me) <jackhill>Althought I get the impressing that using the Nix tooling instead of the stack tooling is a common thing to do. <jackhill>nckx: of course, the top of that file says, "/* hackage-packages.nix is an auto-generated file -- DO NOT EDIT! */" so YMMV <drakonis>they're looking into a way to get rid of auto-generated files and just directly import on the fly <pineapples>I quietly read all threads/messages on the guix-devel mailing list, and I must say that I'm excited for the future of GNU Guix. I love how performance and usability improvements to the existing codebase, as well as implementing new features are the focus of the project. It gives me hope that it is far from stalling in the near future :) <pineapples>The ideas for 1.4.0 brought up in „What's next?” are especially riverting <terpri>does anyone have a standard firefox package lying around with *just* the nonfree bits removed? (i think just disabling EME should sufficient, maybe pocket too since its server is nonfree, although i may be forgetting something) <apteryx>civodul: based on experiments, it seems to not be guaranteed. Polling until getgr and getpw succeeds seems to help. <terpri>icecat is getting a bit outdated for some of the (free) extensions i use, and stripping out the icecat patching seems nontrivial (maybe a change in the build system or something) <terpri>(and ofc there's the official flatpak, but i think that does include nonfree code) <terpri>yeah, it's buggy with multimedia playback sometimes (i'm about halfway to figuring out the root cause), and the different user-agent breaks some sites (though that's not really icecat's fault) <drakonis>oh yes, i'll be happy once multimedia playback works fine <terpri>but mostly its (understandably) being based on firefox ESR means it'll always be a bit behind standard firefox. i'm not 100% sure whether things like container extensions (Temporary Containers etc.) being broken is due to that, but it seems a likely enough reason <drakonis>to be fair, firefox esr doesnt have audio playback weirdness <ghosthell> Non Terrestrial Or Terrestrial Beings which can help me with Trans Universal Transportation (Please PM Me)011011011 <terpri>drakonis, exactly the problem i'm poking at in my spare time :) certain audio streaming methods are just broken, i'm not sure if it's due to using more system libraries (since moz has extensively patched lots of vendored libs) or something else <jackhill>I vaguely remember cwebber poking at a firefox package too <zeminger>Hi. I want to add support for LVM-in-LUKS booting in Guix. The problem is that it is not possible for lvm-device-mapping to say they rely on another mapped-device. Does anyone have an idea on how best to proceed. <drakonis>interesting to know that there's work on this <apteryx>zeminger: have you tried? Adding a dependencies field such as for file-system would probably be a good idea, but it may work already if the mapped-device are setup in the order they are listed? <zeminger>The problem is that the LUKS mapping is deemed to not be needed for booting as nothing directly relies on it\ <zeminger>So the GRUB config and the initrd do not attempt to open it <apteryx>can it be worked around by adding a dependency to the luks mapped device to a file system object? <zeminger>I have done so in the past, and I also had to manually edit the GRUB config as it did not add the correct modprobe lvm. But I would prefer to add a cleaner way. <terpri>zeminger, LVM-in-LUKS is supported now. i switched to plain btrfs-on-LUKS, so i don't have an example config handy, but it works perfectly well <terpri>zeminger, iirc mapped-device entries are simply processed in-order, so later entries can depend on earlier ones (luks-device-mapping precedes lvm-device-mapping) despite the lack of explicit dependencies <drakonis>jackhill: that whole issue is just :chefkiss: <zeminger>terpri did you need a fake file-system to depend on the LUKS partition to make GRUB and the initrd aware of the LUKS partiton? <aocusr>hi, is there a comparison of guix 1.3 to latest nix ? <terpri>zeminger, no, although v. (info "(guix)File Systems"): <drakonis>aocusr: hmmm, andrew tropin made a video comparing guix to nix <terpri>"When the source of a file system is a mapped device (*note Mapped Devices::), its ‘device’ field _must_ refer to the mapped device name—e.g., ‘"/dev/mapper/root-partition"’. This is required so that the system knows that mounting the file system depends on having the corresponding device mapping established." <drakonis>but its a 50 minutes video and who the heck has time to sit all the way through <zeminger>terpri yes the problem is that LVM is a mapped-device and not a file-system. My file-system(s) all point to the LVM mapped-device, and without a fake entry nothing points to the LUKS mapped-device. <terpri>so if i'm reading that correctly you cannot use uuids or labels for LVM-based filesystems, it has to name the LV device <terpri>zeminger, hrm, so you have (mapped-devices (luks-mapped-device ...) (lvm-mapped-device ...))), but guix doesn't open the luks device automatically? <ruffni>is there an equivalent for "guix edit foo" for services? how can i easily look up service definitions? <zeminger>terpri yes. guu/system.scm looks at the file-system(s) that use the mapped-device(s) and if the file-system is needed for boot so is the mapped-device <terpri>zeminger, can you post the relevant bits of your config? that sounds like either a bug or really unintuitive behavior <terpri>i was using btrfs-on-LVM-on-LUKS until just a few days ago, so i know it *can* work <zeminger>terpri Ah okay f00e68ace070fd5240a4b5874e61c26f6e909b6c happened, sorry I was a bit (5 months) behind :P <zeminger>Which means that Guix opens all crypto devices on boot (I think) <zeminger>PotentialUser-53 you could disable the tests, it seemed to compile fine <hwpplayer1>testing guix package manager on Milis Linux mps system <hwpplayer1>I am testing guix package manager on Milis Linux mps system <terpri>note to self: do not try to compile rust and firefox using 64 threads without adequate cpu cooling (a heatsink half the size of my head doesn't quite cut it with a single fan :p) <terpri>hwpplayer1, what are milis and mps? <hwpplayer1>Milis is a LFS OS Linux based, now working on wayland mps is the package manager <terpri>"/me ducks" doesn't have quite the same connotation as "/me googles" <hwpplayer1>guix pull: error: failed to connect to `/var/guix/daemon-socket/socket': Böyle bir dosya ya da dizin yok <terpri>i'm not sure why guix doesn't come with a sysvinit script... but it just needs to run "~root/.config/guix/current/bin/guix-daemon --build-users-group=guixbuild" which i guess you could put in /etc/rc.local (if i remember the name correctly) <ss2>nckx: just back; pulling and about to test your commit. <terpri>(just guessing that an LFS-based distro might not use systemd) <hwpplayer1>Is it like installing many instances remote and local ? <ss2>sorry, I didn't confirm your question. :) <roptat>there's the maintainer collective <roptat>and commiters, but in general, it's not really a top-down structure <terpri>hwpplayer1, basically yes, "deploy" is for managing a cluster of guix machines iiuc (e.g. VPSes) <roptat>no, it's a group of people called the "collective of maintainers" <hwpplayer1>I do not offer to my customers a top-down structure but a structure like a Basketball Team's and Ubuntu African Culture way's <hwpplayer1>Collective means as in my understanding collective leadership <hwpplayer1>Okay One day maybe I am gonna become a maintainer :D <roptat>I don't think I'd say "lead", because their role is more to animate and make sure progress is made, rather than give directions and tell everyone what to do <hwpplayer1>When I say lead I say to take responsibility for good things <hwpplayer1>I work 24/7 and I lead 24/7 Support Team for example <nckx>jackhill: Thanks for the Nix link! ss2: Thank you for the confirmation, I'll close the bug. <terpri>hwpplayer1, i would say the leadership structure is a bit politically sensitive right now. PM me if you want my perspective <terpri>but i think roptat's characterization is a good one in terms of day-to-day work <nckx>drakonis: It was definitely jackhill, I didn't see any pings from you but will look again. <aocusr>the question is which is easier to understand and deploy like you want :) <nckx>terpri: Guix's leadership politically sensitive? I'm... flummoxed, but I don't think I want to know, so I won't ask. <nckx>More general thoughts about auto-generating expressions, right? <nckx>Of course I thank you for those as well :) <drakonis>i thought you were referring to the nix rfc about dynamic expressions <nckx>drakonis: On the Internet? I don't think that's allowed. <nckx>hwpplayer1: Sure, thanks for asking. <nckx>(For practical reasons as much as politeness -- I never notice msgs.) <leoprikler>Is the `eval echo $prefix` hack inside configure.ac copyrightable? <terpri>nckx, GNU governance might be a better way to put it. it's OT here, i only mentioned it because hwpplayer1 seemed curious <nckx>Oh. OK. That's absolutely correct. Thanks for clarifying. <hwpplayer1>I will get 1.5 TB RAM Xeon Server and Nvidia Tesla GPU x 4 <drakonis>there is work on that on the side, but its not something you can get on the main repos <hwpplayer1>Okay i will get a machine without GPU when i start <drakonis>well, it is findable but its not exactly on a highly usable state right now <drakonis>but again, this isnt the place for this topic <drakonis>the other less than legal stuff is up to you to find <civodul>ruffni: re editing services, there's nothing like "guix edit" yet, but "guix system search" shows you source code locations as clickable links <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, nckx says: I reverted commit a67c00f4f7ee0a70fce14a7e1907cce332c85813 to fix #48496. Pulling in sqlite3 for now seems relatively harmless, at least on privileged architectures... :o) <nckx>hwpplayer1: There are 5 maintainers who do, well, not that much, and I think that's a good thing. Mostly ‘people management stuff’. They also have final say over who gets commit access. Said commit access isn't a badge of honour (although it is a compliment), just an optimisation to keep the code flowing smoothly. TL;DR by talking here you've already joined the team. Welcome. <nckx>‘Not that much’ of loudly telling people what to do, I mean. <hwpplayer1>He teached me licensing, documentation and anything I need <nckx><hwpplayer1: FSF should only support Guix> lol. <nckx>TBH I think it's in everyone's best interest to not to do that. People who want Trisquel should get the best Trisquel possible, and conversely Guix should not feel pressured to conform to more traditional paradigms™ to appeal to everyone. <hwpplayer1>Kali Linux trainers and leaders run Gentoo and I will run Guix GNU <ruffni>civodul: thanks! but this just looks up package definitions, no? it looks pretty much like `guix search`? did i miss some option? <jorge[m]>Hola,que significa entrada corrupta durante la restauración del archivo desde socket ? <hwpplayer1>hola! but please speak in English for me at least <jorge[m]>Por alguna razón desde la nueva instalación de mi Guix System no he podido cargar ningun paquete,en uno de los intentos de deje descargando toda la noche y nada. <zimoun>“guix build hurd --target=i586-pc-gnu“ works fine (substitute) but then passing --check, it fails. Do I miss something? <apteryx>hwpplayer1: multiple languages are allowed here :-) <apteryx>eh, then you need not worry about what they wrote :-) If they speak another language and don't get an answer it's up to them to try again in another language if they wish. <hwpplayer1>My family members can speak Albanian Serbian Turkish English French and more <mdevos>jorge[m]: ‘corrupted archive etcetera socket’ can sometimes happening during substitution (e.g. as a result of "guix install", "guix build", "guix environment" ...). <mdevos>jorge[m]: I've seen it before and am pretty sure it is a bug. <mdevos>(A bug in the error reporting, in the substitute server, in the substitution code, or somewhere else. I wouldn't know where exactly) <mdevos>I don't know if it has been reported before <hwpplayer1>Future is Guix If it is good as a phrase or motto I will make a wallpaper for that <mdevos>In any case, you can try to work-around by retrying the command <jorge[m]><mdevos "jorge: Does that help?"> Lo intente varias veces por ejemplo el guix intall y tarda mucho y da el error. <zimoun>hum, if I build several time in row, then it works. Hurd mysterious :-) <mdevos>My attempt has changed [...] for example "guix install" now only gives the error much later? *hwpplayer1 says ((eval I am a GNU Geek, I can hack anything) (Future is GUIX) t) <mdevos>That is my experience as well at times when I try this work-around. <mdevos>jorge[m]: Tip: "guix install --keep-going STUFF" will ignore the error and ‘keep going’, trying to download the rest. Of course, since the package could not be substituted, eventually "guix install" will still fail. <mdevos>If you try "guix install --keep-going STUFF" one or two times or so, I think the command will succeed. <jorge[m]><mdevos "If you try "guix install --keep-"> ok gracias, lo intentare mas tarde <mdevos>hwpplayer1: what are you trying to do? <mdevos>hwpplayer1: what are you trying to do? Do you want to change some parts of guix? <mdevos>(packages, service definitions, build phases, ...) <mdevos>I think I know what you are trying to do, but it is not completely possible at the moment <mdevos>But for a large part it is possible, it's just not all automated <mdevos>"guix build --sources=all this-package that-package" <mdevos>That will download the source code required to build the packages <hwpplayer1>guix mirror --sources=all target=server_ip_address or hostname <mdevos>‘Someone’ could implement a "guix mirror source-code" command that does the equivalent of "guix build --sources=all this-package that-package and-all-other-packages" <hwpplayer1>One package manager to apply and deploy all software <zimoun>the question is what needs to be mirrored? The store? The Git repo of Guix? Something else? <hwpplayer1>like in US in Canada in Europe or wherever whenever <mdevos>I would suggest mirroring the git repo, and the source code of all source code you need <mdevos>(Run "guix build --sources=all interesting-packages" on the mirror server) <mdevos>You still need to ask the ‘clients’ of the mirror to actually use the mirror for source code. <mdevos>I don't know how that would work, maybe adding the mirror to the list of 'substitute servers' of each client is sufficient, or maybe things are more complicated <hwpplayer1>It will be like ("chat.freenode.net" "chat.x.net") <mdevos>IIUC, source code is downloaded from the following, in order (not sure though): The substitute server list, (untrusted) substitute servers discovered on the local network (if configured), ci.guix.gnu.org even if it isn't in the previous list, the upstream location, Software Heritage and Software Heritage+Disarchive <mdevos>So the ‘clients’ of the mirror service don't need to do anything new, they just need to add your mirror to the list of substitute servers. <mdevos>hwpplayer1: I think you may be interested in setting up your own substitute server. <mdevos>A substitute server provides both the source code, and the binaries. <mdevos>I recommend reading the section ‘5.3 Substitute’ of the manual, and look at cuirass <zimoun>apteryx, mbakke or any core-updates mergers: when is the next core-updates merge planned to start? Updating Julia to 1.6.1 requires an update of pcre2 and utf8proc <http://issues.guix.gnu.org/48325#12>. In order to know what does it mean “wait”? Or instead use a “trick”. :-) <mbakke>zimoun: the core-updates freeze will start soonish, but will likely take a while to merge because it is huuuge. So it's probably best to give Julia newer versions on pcre and utf8proc on 'master' if you want it in the next few months. <zimoun>mbakke: ok, cool! Thanks. IIUC, I should submit a patch to core-updates for updating pcre2 and utf8proc soonish, and add to master extra prce2 and utf8proc symbols with their update to be able to update julia. These 2 extra would be removed by the merge. Does it make sense? <keo[m]1><mdevos "E.g., here are a few separate su"> Do they complement default substitute server or just duplicate it for the sake of trust? I'm asking because my machine rebuilds too much every time (and on i686 it's really painful) and i'm searching on how to eliminate that <iyra>I need some help with Guix System, please. I've tried both X11 (i3) and Wayland (sway/hikari), and I get my screen flashing quite often - about once a minute at least, maybe a little more often. <iyra>This doesn't happen on the TTY, and it wasn't always happening. I'm fully upgraded on everything <iyra>joshuaBPMan: Ryzen 5, AMD HD7850 (using nonguix linux with microcode) <iyra>the flashing seems to happen more often when the system is doing something under stress - for example, when compiling Linux it seemed to flash way more often than normal. icecat and Youtube makes it flash quite often too <iyra>joshuaBPMan: Thanks, I'll check that out - but at the moment it happens regardless of icecat running :< <iyra>as far as I know, yes - but it's been a few months since I was using other distros. I was using NixOS before with no flashing. I'm considering having a look by liveUSB booting another distro <iyra>In that time, something in my PC may have started failing, so it really could be a hardware problem <joshuaBPMan>I actually do not run gnome on guix system. I personally just have issues with it. I use sway, because it's the only environment on guix system that I've found that works well. (I have not tried all desktop environments or vms). <joshuaBPMan>hmmm. Possibly a hardware issue. I'm sure there is some way to go about debugging your graphics drivers or screen flashing, but I would not know how to do about doing that. <iyra>Yeah, I didn't enjoy gnome either. I try to stick to tiling WMs like sway when it works or xmonad. But I did try and take sway out of the equation this time to check <joshuaBPMan>iyra: hmmm. I'm running guix system on a thinkpad T400 running osboot. Sway works well for me. <joshuaBPMan>I suppose that you could always use guix on a foreign distro, but that thought personally sickened me. :) I wouldn't get the cool declarative nature of guix. <iyra>Yeah, I know, sway works fine on guix, and I've used it before. It's just since some upgrades a few weeks ago maybe that things don't work any more, and I keep having the same problem even if I choose older revisions in GRUB <joshuaBPMan>iyra: I was having a hard time using guix system on my macbook. The touchpad would only move up and down, which was weird. <joshuaBPMan>that's partially why I switched to the T400. I'm also not a gamer. :) <mekeor[m]>yippie yay, i just managed to run gitlab inside docker on guix-system, working nicely together with nginx and letsencrypt :D <mekeor[m]>joshuaBPMan: to be a bit more precise and humble, i used gitlab's "omnibus" docker image which afaik is not ideal but, eh, well, it works ~shrug~ <mekeor[m]>next step is to package gitlab for guix hahaha, jk, i have other priorities unfortunately <yjftsjthsd>Do guix channels have to use git, or is it possible to use, say, hg or fossil? <mekeor[m]>yjftsjthsd: afaik only git is possible right now <yjftsjthsd>Okay, thanks. Not a terrible burden, but I prefer alternatives when I can get them <drakonis>with effort it should be possible to plug other vcses <yjftsjthsd>A glance at guix-1.3.0/guix/channels.scm makes me think that it would be technically okay but painful in terms of underlying assumptions (the code/comments tend to imply that channel==git repo). No biggie; it isn't my first pick but git is fine so I'm not inclined to spend effort on it