IRC channel logs


back to list of logs

<atka>proper way to remove gdm? I tried to remove it from service in config.scm but still boots into gdm.
<calher>atka: Guix has GDM now? Rock on!
<dstolfa>atka: use (modify-services)
<drakonis>it had gdm for a few releases now
<calher>Last I remembered, Guix used LightDM for GNOME.
<leoprikler>Guix used LightDM?
<calher>leoprikler: Something that wasn't GDM, which Ubuntu used a lot.
<calher>Thought that was LightDM.
<leoprikler>the first known gdm issue predates the first known lightdm issue
<leoprikler>not sure if that's a metric or not
<atka>dstolfa: can you explain some of the syntax for that? I'm new to guix, currenly I tried using (services (remove (lambda (service)...
<dstolfa>atka: ah :). see the example here -- slim-service-type has an example on how to use (modify-services) to remove gdm-service-type
<atka>dstolfa: thanks!
<dstolfa>the (modify-services %desktop-services (delete gdm-service-type)) part
***Kimapr1 is now known as Kimapr
<leoprikler>speaking of modify-services, given that we have delete, we should probably also think about making it possible to add services using it
<dstolfa>leoprikler: good idea
<leoprikler>so you could write (modify-services %desktop-services (delete gdm-service-type) (add slim-service-type [config]))
<atka>ok, trying it out, we'll see if I did it right.
<leoprikler>and under the hood that would be translated into (cons* (service slim-service-type [config]) (old-modify-services ...))
<atka>also when issuing guix reconfigure after a change to config.scm does it always take so long? I just did it a minute ago and now I only changed a line yet it is downloading a bunch of stuff again?
<leoprikler>another neat, albeit probably obscure feature would be (old-service-type => [new-service-type] new-service-config)
<jeko>is there a way to specify additionnal channels in a system definition ?
<jeko>like the one I have in .config/guix/channels.scm
<dragestil>since guix can be used for configuration management, does this mean people normally don't use ansible to configure guix machines?
<jeko>dragestil: i do believe it haha
<jeko>and i think it s what I do when I run "guix deploy new-machin-conf.scm"
<dragestil>does this mean you need guix installed on your host machine as well to be able to deploy
<dstolfa>dragestil: if you want to do it from your machine instead of ssh and do it there, yes. it's more akin to salt than it is to ansible (as it requires a daemon)
<dragestil>hmm ok, is the requirement of a daemon due to guix requiring guix daemon?
<dragestil>ok thanks
<drakonis>daemon rewrite, eventually.
<dragestil>drakonis: what do you mean
<drakonis>there's work to do be done to make it entirely in guile
<mroh>calher: we don't have a lightdm-service (yet), see
<calher>mroh: Odd. I wonder what it was, then.
***califax- is now known as califax
***apteryx_ is now known as apteryx
<drakonis>hm, libglvnd is required for multigpu setups
***sturm is now known as bsturmfels
<famubu>Hi. I had been trying to install guix on my computer. But I'm getting some error while doing `guix pull`. log show this: It looked like some certificate couldn't be fetched from Does anyone know how this can be fixed. Found a similar issue at so I tried edition /etc/hosts and
<famubu>/etc/resolv.conf (with nameserver but the issue remains.
<ecraven>if I define my own custom packages, they should also be in (gnu packages ..) right?
<rekado_>ecraven: they don’t have to be
<ecraven>don't the cli tools only search for packages in that namespace?
<tissevert>hello guix !
<ecraven>hello ;)
<sneek>mothacehe, you have 1 message!
<sneek>mothacehe, bricewge says: Would it be possible to have RSS feed from Curiass per build history? So that a person interested in, say the "hello" package, can follow updates and failing builds of that package.
<mothacehe>bricewge: sure, that a nice idea. There's also: that provides a per-manifest dashboard that is still pending.
<bricewge>Great! Your patch match somewhat the usage case I was looking for.
<ecraven>can I get guix to tell me which package provides a given file? aclocal, for example ;)
<ngz>ecraven: Not yet.
<PurpleSym>rekado_: Could you have a look at the (minor) R patch at bug#49740 sometime?
<tissevert>is there some DNS herd service that got recently enabled by default in Guix ?
<ecraven>hm.. this daemon I'm trying to package calls ./tools/, which has a shebang of /bin/sh (which fails on guix build -f ...). how do I deal with this properly?
<tissevert>ecraven: not sure at all but isn't that what the patch-source-shebangs function was made for ? (more in the manual)
<ecraven>this happens in phase bootstrap, maybe that is before patch-source-shebangs :-/
<tissevert>ooohh I haven't used builds with a bootstrap phase, do you mean the patch-source-shebangs function isn't available then ?
<civodul>Hello Guix!
<ecraven>tissevert: bootstrap runs before it, so it is already too late. I just added a custom phase with a substitute*, but now I'm running into autotools problems.
<ecraven>Is there a way to "enter" the build directly and have an environment with everything set up like for building?
<ecraven>so I can run commands and see what happens?
<efraim>`guix build -K` will keep the directory when it fails, then you can go to /tmp/guix-build... and `source environment-variables`
<jeko>Yo Guix !
<tissevert>yo jeko
<jeko>tissevert: o/ Thanks for your last record hahaha
<tissevert>did the little one enjoy it ?
<jeko>tissevert: yeah, more to dance than to sleep but… enjoyed haha
<tissevert>ohh, too jazzy ? sorry : 3
<jeko>I am wondering if I can add a custom channel to a system definition ? to replace the need to edit ~/.config/guix/channels.scm
<ecraven>hm.. can I somehow get the guix builder to put the .git directory into the building directory? this tool uses git log to check for the current version. I have (method git-fetch) and (uri (git-reference ...) ...)
<efraim>you'll have to patch it. We delete the .git directory as part of git-fetch
<ecraven>how do I just replace an entire file?
<ecraven>how do I access a package's `version' inside a phase?
<efraim>,(package-version this-package) normally
<ecraven>should the build system type (--build for ./configure) really be "x86_64-unknown-linux-gnu"?
<efraim>yeah, that's an acceptable "triplet"
<ecraven>how do I find out which package provides what is usually in /bin/true?
<efraim>we don't have a way yet, so if I have it installed then normally `readlink $(which true)` will work, otherwise `find /gnu/store -name true`
<efraim>I also have apt-file installed on a debian box so I'll do `ssh debian-box apt-file search true | grep '/bin/true'`
<ecraven>ah, I found it in profile, but didn't understand what that means.. it's of course in coreutils.. thanks!
<ecraven>sorry about the many questions, please tell me if it's inappropriate here!
<efraim>which true returns the path to true, readlink follows the symlink
<ecraven>thanks a lot, it seems this has all worked now ;) I'll play around with it, and maybe I can contribute the knxd package to guix upstream if it turns out to work well
<rekado_>PurpleSym: will do
<jeko>Can I specify custom channels in a system definition ? Or is it only possible in ~/.config/guix/channels.scm ?
<mothacehe>jeko: you can use the /etc/guix/channels.scm file. Soon, you will also be able to declare the channels in the system configuration, see:
<rekado_>PurpleSym: the patch looks good to me. Odd that doesn’t show anything :-/
<PurpleSym>rekado_: Thanks! Push to core-updates or master? (I thought was just taking its time, but you’re right, still nothing…)
<rekado_>PurpleSym: maybe push to a new r-updates…?
<rekado_>based on top of master
***Kimapr6 is now known as Kimapr
<ngz>Is the CI stuck? I see no progress on master.
<mothacehe>ngz: yeah some SSL error while fetching Git repositories, just restarted it.
<PurpleSym>rekado_: Yeah, good idea preparing for the upcoming v4.1.1. Pushed. I also saw two references to /bin/sh, but not sure yet how to test this from the R side (R_system_timeout and R_popen_timeout).
<rekado_>PurpleSym: hmm, re the missing issue: it’s not mumi’s problem. The mails haven’t been transferred from the debbugs server.
<iyzsong>hello :~)
<rekado_>PurpleSym: restarted the rsync process; now it’s there
<civodul>hey ho
<civodul>mothacehe: thanks for pushing the search-input-file fix!
<mothacehe>civodul: yw :) Cuirass is has almost caught up!
<civodul>yeah, pretty cool! i was looking at it
<mothacehe>I had to restart Cuirass this morning because of this error: "Git error while fetching inputs of 'version-1.3.0': "SSL error: syscall failure". It would be nice if Cuirass was robust to this kind of errors. I'll report it on bug-guix I guess.
<ecraven>how does guix system docker-image work? I'm trying 'guix system docker-image ./config.scm'
<ecraven>but that fails
<sss1>hi all, is ti possible to store distfiles cache on separate location in guix ?
<ecraven>this is the config and the output. any ideas?
<civodul>ecraven: hi! could you check the build log at /var/log/guix/drvs/18/13v6mnmc2fjs2g5wrgskp625vbd6xb-guix-docker-image.tar.gz.drv.bz2, as noted in the error message?
<civodul>mothacehe: weird, what's that SSL error about?
<ecraven>hm.. does that build start a VM? ERROR: In procedure readdir: In procedure readdir: Cannot allocate memory (I have about 10G free RAM)
<civodul>yes, it starts a VM
<civodul>so it looks like that VM lacks RAM, hmm
<ecraven>hm.. can I just install a rootfs from a running guix (not guixsd) into a block device?
<civodul>sure, see "guix system image", which is the generic command to build images
<ecraven>there's also this: [ 6.868143] 5ijdndxv9107lzn: page allocation failure: order:7, mode:0x40dc0(GFP_KERNEL|__GFP_COMP|__GFP_ZERO), nodemask=(null),cpuset=/,mems_allowed=0
<ecraven>civodul: I don't want an image, just the root directory
<civodul>"guix system build", then
<civodul>well, "image" is to be taken in a broad sense
<ecraven>guix system build doesn't actually put things into a directory though, does it?
***Kimapr4 is now known as Kimapr
<ecraven>say I'm running on arch linux and want to install guixsd into /mnt (which is actually /dev/sda2 or whatever)
<ecraven>can I do this?
<ecraven>I don't need grub or anything, I just want the root directory initialised, so that I could boot from that partition
<civodul>you could run "guix system init config.scm /mnt", but that also installs the bootloader, unless you tweak the 'bootloader' field of the config
<ecraven>I tried removing it, but that didn't work, it complained
<civodul>right, you need to derive, say, 'grub-bootloader', and annihilate its "install" procedure
<civodul>mothacehe: do you have the right incantation for that at hand? :-)
<civodul>we should define 'null-bootloader' somewhere to make it easier
<mothacehe>guix system init config.scm /mnt --no-bootloader ?
<civodul>oh, is it this simple?
<ecraven>ah, nice, thank you!
<ecraven>sorry for not seeing this
<civodul>ah indeed, nice
<ecraven>I'm trying to create an LXC container template for guix
<ecraven>but I have no knowledge of guix *or* of lxc containers, so this is rather complicated :D
<ecraven>I'd like to run guix as a container in proxmox
<civodul>be sure to turn it into a section of the cookbook when you reach the end of the road :-)
<ecraven>hehe, not sure I'll ever reach that.. but if I do, I will document it
<ecraven>does guix have an /etc/os-release file?
<jeko>mothacehe: Thank you for your answer ! Can't wait to see it available in a system configuration ! I'll go with /etc/guix/channels.scm in the meantime
<ecraven>what is the init process on guixsd?
<ecraven>i.e. what should I start if running inside a container?
<rndd>hi guix!
<rndd>i have a problem. i write some bash scripts and i want them to be runnable on different distros. what is the best way to identify guix system using bash?
<rekado_>PurpleSym: FYI
<boeg>does guix have any philosophy around for example that the "source of truth" should be the manual pages, or the only documentation or something third?
<boeg>contrary to other linux distributions, I mean. Does it set itself in another position regarding that?
<maximed>rndd: Why do these bash scripts want to know whether they are running on Guix System?
<ecraven>does "#!/usr/bin/env bash" work on guix?
<ecraven>otherwise, those scripts don't have any way to run on guix anyway, there being no /bin/bash, right?
<maximed>When /usr/bin/env exists and bash is in PATH, then yes, #!/usr/bin/env bash works
<maximed>this is the case by default
<maximed>/bin/bash doesn't exist however
<rndd>just wanna know is there way to identify guix system using bash
<roptat>the presence of /run/current-system?
<efraim>ecraven: I have an /etc/os-release file if you need:
<ecraven>efraim: turns out that proxmox doesn't like os-releases it doesn't know about, so that wouldn't help :-/ thanks though
<rndd>roptat: ohhh, but is there /run/current-system if i install guix on top of foreighn distro?
<ecraven>there's /gnu/store :D
<roptat>if you want to detect the package manager, maybe /gnu/strore
<maximed>or ~/.guix-profile
<roptat>~/.guix-profile will not exist right after installation though, because there's no profile yet
<rndd>well, i definately have /gnu/store on another distros which have guix install as pm
<rndd>thank you
<roptat>the question was about detecting the guix system, though :)
<roptat>because if you only want to detect guix, you could use "which guix" ;)
<rndd>bye, guix!
<therealdonotshak>5.3.3 Getting Substitutes from Other Servers shows an example of how you add substitutes to your system config.
<ekaitz>hi maybe someone can help me
<ekaitz>i'm trying to run a program and I always get file not found
<ekaitz>i greped objdump and i found it needs:
<maximed>are these absolute paths?
<maximed>(/gnu/store/.../lib in RPATH or RUNPATH)
<ekaitz>nope this program was compiled from outside
<maximed>then you'll need something like 'elfedit' to make the paths absolute
<maximed>or maybe setting LD_LIBRARY_PATH appropriately will work
<roptat>the "file not found" comes from the fact that the kernel can't find the linker
<roptat>it's probably set to /lib/ or similar, but there's no such file on the guix system
<ekaitz>roptat: and there's any way to make the kernel find the linker for them?
<ekaitz>maximed: LD_LIBRARY_PATH... let me try it!
<maximed>The kernel probably does not look up the linker in LD_LIBRARY_PATH though ...
<roptat>you can create an environment with the linker in /lib, or you can run patchelf to change the linker and set a runpath
<roptat>I use the former to run android studio:
<ekaitz>roptat: i'm basically trying to make android stuff but using the command line tools
<ekaitz>I had tons of problems with android studio :D
<roptat>ah, then you'll need that blog post, even if you don't run android studio
<roptat>there are some stuff that gradle will download that contain a binary, and you can't modify them, gradle will verify the checksum and replace them everytime you try to patchelf them...
<ekaitz>yeah... i've been reading it, I just was trying to make it easier but looks the emulator doesn't want to run without the environment
<ekaitz>so yeah...have to make the environment
<roptat>I think it was related to the tools that create the apk
<ekaitz>this is the most hostile software I used in my life
<PurpleSym>rekado_: Ah, nice. I’ll replace references to /bin/sh tomorrow. `guix environment -C` has a symlink, but I think consensus is to replace it.
<ekaitz>roptat: i'm getting the same issue :S
<roptat>mh.. what does ldd tell you?
<roptat>is it a 32-bit binary?
<roptat>(if so, you'll need a 32 bits libc)
<ekaitz>oh where's the ldd binary?
<roptat>I think it's part of glibc
<roptat>yes, glibc
<ekaitz> => not found
<ekaitz>but the others are pointing to the correct path!
<ekaitz>ugh i needed to set LD_LIBRARY_PATH by hand and now it finds the path but it's not running still
<roptat>it's still missing the linker
<roptat>even if ldd shows it, it only shows it because ldd has a linker
<roptat>/lib64/ => /gnu/store/fa6wj5bxkj5ll1d7292a70knmyl7a0cr-glibc-2.31/lib/ is not correct
<roptat>the second path is only there because ldd itself knows about its linker, but when you run it, the kernel doesn't know, it's looking for /lib64/ only
<roptat>that's why I was suggesting to use patchelf, or a container where you link /lib to the profile's lib directory
<ekaitz>oh i understand
<ekaitz>I'm trying with the container but I'm unable to make it work :(
<roptat>do you have glibc in the container?
<roptat>and did you manage to get a /lib/ld* in it?
<ekaitz>not sure about the second
<ekaitz>the libc yes
<roptat>actually, you should link /lib64, not /lib I think
<roptat>or both
<ekaitz>i'm basically running the same command you use for the environment
<ekaitz>now I managed to add the ld-* but it still fails :D
<ekaitz>I should be able to run `ls /lib` from inside the container right?
<ekaitz>ugh forget this I'll start over this is depressing
<ecraven>is there a description anywhere of how guixsd boots up? which process is PID 1, things like that?
<roptat>the shepherd is pid 1
<ecraven>how is it started? does it need any parameters?
<roptat>the bootloader passes the arguments to the kernel, it lets it know about its location
<ecraven>I'm trying to understand how I would run the shepherd in a container, so without actually booting
<roptat>you could use "guix system docker-image" or "guix system container", that would be simpler I think
<ecraven>so guixsd just starts '/gnu/store/655wq4zm0j9fh05xzzr1hc6a2z5y8dra-shepherd-0.8.1/bin/shepherd'?
<ecraven>that doesn't produce an lxc image that I can run
<ecraven>also, guix system docker-image doesn't actually work here, it fails while running inside the VM for building
<roptat>not exactly, in my case, it uses an initrd that contains the shepherd and stuff, and runs the boot script from it (/gnu/store/3ssf75i8dz1zqzyszlkckwcairdy55cy-system/boot, but the location changes at every update)
<ecraven>hm.. so in the end, it might be rather hard to actually boot without a full boot?
<ecraven>I'll try to get a docker image built, and see what that does
<roptat>I managed to run a guix system from a chroot once...
<roptat>but the shepherd was not working, I had to run my daemons by hand
<ecraven>why does guix system docker-image start a VM? shouldn't it just start adding layers?
<roptat>don't know
<rekado_>PurpleSym: I thought /bin/sh is okay, because it’s a POSIX requirement.
<rekado_>PurpleSym: I’ll merge r-updates later tonight because it has already been fully built by
<rekado_>(I’ll remove the two commits where I accidentally upgraded r-mosaic and then reverted the commit.)
<BlackMug[m]>hi there
<BlackMug[m]>i tried to search for instructions in the docs on how to install software for linux downloaded from website
<BlackMug[m]>the example i used is TorBrowser
<BlackMug[m]>where to extract it in order to make it work?
<BlackMug[m]>(normal extraction in desktop or so doesnt work)
<roptat>it doesn't work because the binaries hard-code a location for the glibc/libs that doesn't exist on the guix system
<roptat>there's no location where you can extract them that would make it magically work
<roptat>you can try and use patchelf to get the linker location and rpath right, or run in a guix container, ...
<BlackMug[m]>i see, so guix container allow the execution of the software without the need to be from guix package manager
<roptat>not really, but it lets you "mount" packages on /lib and more common locations that the binary expects
<BlackMug[m]>so appimage , and using guix as a server is like work in hell
<roptat>I mean, you can run that binary with "guix environment --container", specifying the required libraries, and that would work, even though it's not the main use case
<the_tubular>I've got a weird problem, when I guix pull I always get this at the end : New in this revision: and a list of "new package" my problem is that the list is always the same
<roptat>this is how I run android-studio from the binaries:
<the_tubular>Even if I read it with guix pull --news, the list is going to stay the same
<roptat>mh... what does "type guix" tell you?
<roptat>and "which guix"?
<the_tubular>guix is hashed (/home/vm/.config/guix/current/bin/guix)
<roptat>are you running "guix pull" or "sudo guix pull"?
<the_tubular> /home/vm/.config/guix/current/bin/guix
<the_tubular>Both, in fact what are the difference between the 2 ?
<roptat>guix pull updates the guix for your user, sudo guix pull for the root user
<the_tubular>I noticed it happens when I sudo guix pull
<GNUtoo>hi, does someone already had that error or does someone has a clue on what could be the cause of that error: 'guix system: error: cannot close compressed log file (BZip2 error = -6)'
<roptat>the_tubular, I suppose it updates the root's guix, but compares with your user's
<roptat>I think it's not recommended to run sudo with guix pull
<roptat>I mean guix pull with sudo
<roptat>the recommended commands to update the guix system are "guix pull" then "sudo guix system reconfigure /etc/config.scm"
<roptat>that last sudo uses your user's guix, so you only have to update the user's guix
<roptat>GNUtoo, never seen that
<GNUtoo>ok, -6 is probably ENXIO (No such device or address) which looks strange
<roptat>is /var/log a separate partition?
<roptat>or bzip errors are not the same as errno
<ecraven>does shepherd have any way to increase the loglevel? if I start it (inside a container), it just does nothing and exits right away, no output
<ecraven>ah, actually, it complains about ../etc/shepherd.scm not being found.. who should create that, wouldn't guix system init do that?
<GNUtoo>oh ok
<GNUtoo>ahh maybe I get it, the guix daemon doesn't run as root
<GNUtoo>so while I still have some space, I'm low on space
<GNUtoo>so maybe it can't write because of that
<roptat>ecraven, yeah guix system init or reconfigure would create that
<ecraven>it seems it didn't..
<ecraven>would --no-bootloader prevent that?
<GNUtoo>(btw, /var/log is in my / rootfs)
<roptat>ecraven, I don't think so
<roptat>possibly, you're missing a -c argument to pass the correct configuration file
<roptat>maybe it's looking for the default one, in /gnu/store/...-shepherd/etc/shepherd.scm
<ecraven>this is what I run: guix system init config.scm root --no-bootloader [where `root' is the directory]
<ecraven>yes, that's what it's looking for. should that be generated?
<roptat>no, that file doesn't exist and shouldn't exist
<roptat>a store item can't be modified after it's been built. Instead, we create a shepherd.scm in a another store item, and point the shepherd to it when we start it
<ecraven>how is the initial one found?
<roptat>in the initrd, the shepherd is run with a reference to that file
<ecraven>so the initrd has the correct one hardcoded?
<ecraven>hm.. how would I find the right one with --no-bootloader?
<ecraven>could it be this? ./gnu/store/axxdbmc1nimc4421lkdnrjrbs85ssxgi-profile/share/guile/site/3.0/shepherd.scm
<roptat>maybe, who knows?
<ecraven>no.. what should be in that file?
<roptat>if you can find the boot file it generated, it has a reference to the configuration
<ecraven>is that even generated with --no-bootloader?
<roptat>I don't know
<roptat>I think so, --no-bootloader is supposed to stop just before installing the bootloader
<ecraven>do you know what that file is named?
<roptat>but you have no way to predict the exact name
<ecraven>I only have one file like that, but it doesn't look like configuration, just like normal code
<ecraven>is this it?
<roptat>yes, the shepherd is configured with guile
<roptat>ah no
<roptat>it should be one line
<roptat>oh, it's shepherd.conf, not .scm
<ecraven>ah, great, I have that!
<ecraven>sorry about the confusion
<roptat>no, it's my fault
<BlackMug[m]>@roptat your website onion url broken, it shows the urls like this:
<roptat>is that bad? what is expected?
<the_tubular>I mean /404.html is probably not expected :P
<BlackMug[m]><roptat "is that bad? what is expected?"> if the url changed from clearnet to onion should refer to the same path of the page
<BlackMug[m]>but instead it will redirect to 404
<roptat>hm... that's weird, I see "onion-location http://c25o7knygjm3m67jy27yuynvv4pkfi25naucscmh4ubq2ggiig3v57ad.onion/en/running-android-studio-on-guix.html" for instance
<roptat>(in the headers, maybe you're using something else?)
<roptat>oh I see! you're using the link on the bottom and it's missing the .html extension
<roptat>BlackMug[m], should be fixed now
<ecraven>great, now shepherd starts (and fails to start lots of services ;)
<BlackMug[m]>roptat: yep now working thx
<achow101>I'm having some issues with guix archive --import. On one machine it works, but on another, it doesn't, for the same archive.
<achow101>The error on the failing machine is "guix archive: error: corrupt input while restoring archive from #<closed: file 7fb0e58281c0>"
<GNUtoo>Making some space solved it.
<GNUtoo>Thanks for the help.
<roptat>achow101, could be an authorization problem, or maybe the archive is really corrupt on one machine (transfer issue?)
<achow101>roptat: It seems like the problem was with the installation. I had used guix pack to make a tarball the install script could install, but if I use the release tarball, it has no problem
<achow101>I'm not sure why guix pack would have made an installation that didn't work. I've done this before without issue, and can do it repeatedly without issue (using VMs to test)
<ekaitz>roptat: i'm back to this but using your tutorial
<ekaitz>are you able to run mksdcard?
<ekaitz>it's giving me issues
<ekaitz>android-studio runs perfectly btw
<roptat>I haven't tried it by hand, but I suppose it's working if the device manager works
<roptat>what's the issue?
<ekaitz>when I start android studio the first time it says: unable to run mksdcard
<roptat>hm... I don't get that error, but I'm using the last command, that adds some more stuff to LD_LIBRARY_PATH
<roptat>maybe that's required?
<ekaitz>I'll start over again just in case
<ekaitz>the avd is not working neither :((((((
<civodul>hey ho! Guix-Jupyter 0.2.2 is out! (and one "guix pull" away :-))
<jackhill>yay, thanks civodul!
<dstolfa>civodul: nice!
<roptat>great :)
<ecraven>how does guix deal with loading .so files dynamically? how does a program find them?
<ecraven>if shepherd says "failed to start service", where can I see *why* it failed?
<Gooberpatrol66>are you supposed to use and as substitute servers?
<Gooberpatrol66>the keys are in the directory with and but they aren't enabled by default
<cybersyn>hiya guix, i just want to report that while my first install of guix was a bit confusing last November (having first tried the TUI installer then, but having to resort to manual config), I just did a clean install on the same machine (2010 iMac) with the TUI interface and it installed without a problem or a gnuissance, and that feels like a real massive leap for the system.
<civodul>ecraven: .so are searched for in the "normal" way: $LD_LIBRARY_PATH first (if defined), then DT_RUNPATH for ELF files
<ecraven>civodul: thanks!
<civodul>ecraven: re shepherd, you ahem, have to find out...
<civodul>usually the service's log file would have clues
<civodul>if not, "strace -p1" remains a great tool :-)
<ecraven>;) that sounds like there's room for improvement there :D
<civodul>yeah, i think so :-)
<jorge[m]123>Hola,he colocado un posting en Pastezone To link to your entry use:
<jorge[m]123>mañana realizare otro por error de reloj en el kernel.
<civodul>jorge[m]123: me parece que todo está bien; ¿hay un problema?
<jorge[m]123><civodul "jorge: me parece que todo está b"> Si,creo que ese error y el que escribiré mañana son la causa de no realizar guix pull.
<jeko>Yoo Guix
<civodul>jorge[m]123: pero no veo ningún error aquí
<roptat>Gooberpatrol66, you installed guix on a foreign distro with the script? It should have asked whether you want to use substitute servers or not. If you said yes, it should have authorized the keys, otherwise try "guix archive --authorize < /path/to/the/"
<roptat>also ci and berlin are the same thing
<Gooberpatrol66>roptat: i'm using guix system
<Gooberpatrol66>there are 4 keys in /root/.config/guix/current/share/guix but only 2 are used by default for substitutes
<roptat>oh right, you can configure that in /etc/config.scm
<roptat>you'll notice that berlin is a symlink to ci (or the other way around)
<roptat>mh, oh no it's not
<roptat>but at least, berlin ci and ci are the same file (see sha256sum /root/.config/guix/current/share/guix), so that's why you have only two authorized keys
*roptat needs to go
<the_tubular>Anyone is running guix system in WSL2 successfully ?
<GNUtoo>civodul: I think we need something in guix for handling adding files to the system images: the issue with the out of band issue is that it's the wrong abstraction in my opinion:
<GNUtoo>(1) we can do it by hand or code tools that do it but to properly integrate it we'd need to handle all the image formats that guix handles
<GNUtoo>(2) the partition scheme and so on is handled by Guix, so I'm not sure that auto-detecting the configuration with some external tools will always work
<GNUtoo>The idea I have would be to have some abstraction that would construct the final image from an existing image and for instance a tarball