IRC channel logs


back to list of logs

<luis-felipe>First personal experience with Androind, and it was horrible. I don't want to touch that thing any more.
<str1ngs>luis-felipe: I can't seem to replicate this see.
<str1ngs>luis-felipe: I hear ya, nomad should be usable just keep in minds it's still alpha. I do use daily myself.
<luis-felipe>str1ngs: Well, that's weird.
<luis-felipe>str1ngs: I know it's alpha, I tried it before, and wanted to check what's new.
<luis-felipe>str1ngs: It seems to be lighter, and that's nice. In my case, browsers take most of the RAM of this computer.
<str1ngs>luis-felipe: can you maybe take a screenshot might help.
<luis-felipe>Give me a sec.
<str1ngs>luis-felipe: nomad is pretty light and very much emacs centric. so if you are a emacs user should be pretty intuitive
<luis-felipe>That's one of the reasons I want to use it, and because its Guile.
<str1ngs>luis-felipe: that's a whole other fun aspect.
<luis-felipe>When I hover over the icon that should be Emacs, it says "Nomad", and hovering over the icon corresponding to Nomad, it says ".nomad-real".
<terpri>str1ngs, that's pretty cool that it's engine-agnostic
<str1ngs>luis-felipe: I'm not sure if this is related to nomad. it's definitely odd though.
<luis-felipe>Yeah, who knows...
<str1ngs>luis-felipe: ha think I found the issue
<luis-felipe>Oh, really :)
<str1ngs>sneek: later tell nly, can you check the desktop file for nomad. seem StartupWMClass=Emacs is wrong.
<sneek>Got it.
<str1ngs>sneek: later tell nly, could you also update the command to be more descriptive. thank you
<str1ngs>luis-felipe: that's definitely the issue. thanks for reporting this. and sorry for the inconvenience
<str1ngs>luis-felipe: do you mind if the fix for this waits abit. I have some other additions that I'd like to include in the next nomad guix declaration. mainly the new header bar.
<luis-felipe>Oh, no. Thanks for taking a look at it.
<luis-felipe>str1ngs: That's fine. I know nomad is not "production-ready". Take your time, your're the developer :)
<str1ngs>luis-felipe: there is a guix way to use a git version but still requires cloning nomad.
<luis-felipe>It's OK, I can wait.
<str1ngs>luis-felipe: thanks for your patients and for reporting this.
<luis-felipe>You're welcome.
***allan is now known as Guest96914
<Guest96914>Hello, can someone help me? I am trying to make a package for software called obs-gphoto but when I tried to build it it says gphoto2 is missing, despite me including it in the code. What is going on?
***daviid is now known as Guest60468
***Guest60468 is now known as daviid
<guixy>Hello again guix
<guixy>Can anyone with a relevant article contribute it to the guix blog?
<guixy>In other words, is the guix blog open to contributions?
<guixy>Guest96914, I think you need to make sure gphoto2 is a native input.
<guixy>But it depends on how it's used...
<guixy>It looks like you need libgphoto2, the library. Make sure it's an input.
<roptat>guixy, yes the blog is open to anyone who has something relevant to show :)
<roptat>I think there's a mail alias like guix-blog@ or something like that
<guixy>I don't see any thing related to contributing to the blog on the contact page.
<roptat>you can always send to guix-devel
***catonano_ is now known as catonano
<PotentialUser>Hello. Im trying to install dotnet core with the script that microsoft provides, said script installs at /usr/share but there is no such directory in GUIX, any suggestions? (i can provide a different install directory with their script, but i dont know wich one is the replacement for /usr/shared)
<PotentialUser>alextee: looking at the logs i see that you were asking about dotnet too, did you find any workaround?
<alextee[m]>no sorry
<PotentialUser>aww, and i assume you dont know what is the equivalent of /usr/shared in guix system
<raghavgururajan>Hello Gu...Gu...Gu...Guix!!!
<raghavgururajan>> ‎PotentialUser‎: aww, and i assume you dont know what is the equivalent of /usr/shared in guix system
<PotentialUser>nice, i will try to use that for the install script
<roptat>you can't it's read only
<roptat>on purpose
<roptat>don't modify it, it'll break guix's assumptions about the store
<roptat>there's nothing like a global /usr/share directory in the guix system, it just doesn't work like
<PotentialUser>oh, that is inconvenient
<raghavgururajan>> $HOME/.guix-profile/share
<raghavgururajan>a.k.a /var/guix/profiles/per-user/rg/guix-profile
<raghavgururajan>Ah yeah, DO NOT modify
<PotentialUser>it seems like im in a pickle, hopefully once mono and dotnet core combine in .NET 5.0 someone wil pack it
<roptat>what you can do though is to install anywhere you like, it's just that there's no special directory named /usr/share
<roptat>you could always create a /usr/share and install stuff there
<PotentialUser>i will try that, i think some lookups made by dotnet might break but hopefully its easy enough to fix
<roptat>but then your binary will probably not know where to find its libraries, unless it bundles them all, down to the libc
<raghavgururajan>PotentialUser: If a package/program/software looks for it, patch the source to look for $XDG_DATA_DIR
<raghavgururajan>Which always points to available /share
<PotentialUser>hmmm, this look like too much trouble for little gain, i will use mono as much as i hate it until 5.0 comes, who knows, i might even pack it myself once its released
<roptat>also, you can unpack to any directory you want and create a guix container that shares it to /usr/share
<alextee[m]>PotentialUser: if you're writing a package, you can get the "out"'s directory
<alextee[m]>which will be something like /gnu/store/<longhash>/
<alextee[m]>that's your prefix and you should install everything there
<alextee[m]>something like this
<PotentialUser>Worry not, Im not making any packages yet until i finish reading the documentation, my questions were regarding the bash script that microsoft provides to install dotnet core
<alextee[m]>you need to write a package if you want to install anything system-wide
<alextee[m]>not sure if it will work if you install locally (under $HOME)
<PotentialUser>yeah, i think i will cope with mono until someone more knowledgeable saves the day
<PotentialUser>also, why the hash are at the start of the file instead of the end/middle?
<PotentialUser>it makes searching for stuff so annoying
<guixy>If you're searching in /gnu/store all you need to do is pipe your search to cut the hash out.
<PotentialUser>that is the annoying part
<guixy>cd /gnu/store; ls | cut -b34- ...
<PotentialUser>instead of placing the name at the start so that autocomplete can do its magic you have to use pipes, wich is a design choice i dont understant
<guixy>You could alter guix to put it at the end... but you would lose the benefits of substitues.
<alextee[m]>i was the same at first lol
<alextee[m]>it makes sense to be at the start, it can do lots of neat tricks because of it
<guixy>There are likely to be so many store items with the same name anyway
<guixy>If you want to find a built package, just build it again.
<guixy>cd `guix build hello`
<PotentialUser>nah, im not looking for alternatives i was just curious if it was a genuine mistake that is too late to fix or if there is a benefit that i dont see right now
<alextee[m]>oh wait you mean the part /gnu/store/<here>/
<PotentialUser>i mean the hash before the name, yes
<alextee[m]>yeah why couldn't it be /gnu/store/myprogram-#hash/ instead of /gnu/store/#hash-myprogram/ ?
<alextee[m]>im curious too
<PotentialUser>IIRC nix does it too for some reason
<PotentialUser>so maybe the inspiration comes from there
<alextee[m]><guixy "There are likely to be so many s"> i guess this
<alextee[m]>and also you're not supposed to tab there
<alextee[m]>there's other ways to do what you're trying to accomplish
<guixy>It would be nice if /gnu was a hidden directory.
<guixy>Then you wouldn't really think about it
<apteryx>PotentialUser: you can use find /gnu/store -maxdepth 1 -name '*icecat*', which is probably going to return the results faster than what a bash autocompletion would be able to do on thousands of entries
<alextee[m]>but still why would you even do that?
<alextee[m]>there might be lots of icecats anyway from older generations
<PotentialUser>hey, autocompletion is definetively more fast than typing all of that
<alextee[m]>usually you want to find your store hash by listing your generations
<PotentialUser>ALSO, i almost forgot
<PotentialUser>why are there 2 pipewire packages?
<alextee[m]>`guix package --list-generations | grep icecat` or something
<PotentialUser>like, the same package but different versions
<apteryx>alextee[m]: that's true, but suppose your current guix is borked and your trying to find an older one to run from... dunno.
<guixy>Usually old packages are kept if some packages haven't updated yet...
<PotentialUser>might be that, huh
<guixy>guix package -A | cut -f1 | sort | uniq -d
<alextee[m]>PotentialUser: different versions
<alextee[m]>0.2.7 and 0.3.10
<alextee[m]>some apps might need the previous one
<PotentialUser>oh, i definitively didnt thought about that
<alextee[m]>you define the version in the package definition if you want
<alextee[m]>if you dont, it uses the latest
<PotentialUser>any of you guys use bspwm?
<PotentialUser>i cant get it working
<PotentialUser>gdm doesnt seem to detect it
<apteryx>PotentialUser: people or folks is better form than 'guys' here :-)
***capnick is now known as cap
<PotentialUser>welp, i guess rules are the rules, i will try to keep that in mind
<alextee[m]>apteryx: "guys" is used for girls too :P
<apteryx>it's not a strict rule, but I think it helps in keeping the place welcoming to all.
<apteryx>alextee[m]: I know, but that doesn't make it any more inclusive :-)
*alextee[m] shrugs
<PotentialUser>oh, i think its because WM dont have services by default
<PotentialUser>doesnt explain why it crashes in tty though
<apteryx>PotentialUser: don't quote me on that, but I think gdm uses .desktop files to discover the available WM, IIRC. Perhaps bspwm doesn't have any?
<PotentialUser>that is a secondary problem, it doesnt even work in a tty
<wleslie>does your tty have a DISPLAY set?
<peanutbutterandc>Quick question: How do I use emacs-geiser with guix? Preferably, I want run-geiser to start `guix repl` instead of the normal `guile`. How could I go about doing it?
<sneek>Welcome back peanutbutterandc, you have 2 messages!
<sneek>peanutbutterandc, leoprikler says: did you test Julien's .asoundrc? If so and it didn't work, then fine.
<sneek>peanutbutterandc, leoprikler says: Your hack can probably be made a little nicer by using (assoc-ref %build-inputs "alsa-plugins:pulseaudio"), but it still remains a hack around a somewhat different problem. Perhaps you might want to contribute towards documentation regarding the use of ALSA/Pulseaudio on foreign distros.
<peanutbutterandc>leoprikler, I am trying to polish up the change, yes. Yes, I see that this is quite a huge issue. But hopefully we'll have a solution/more-elegant-fix soon.
<wleslie>should `make` have clobbered all the po files in the guix repo?
<peanutbutterandc>does anybody here use geiser? How do I get a `guix repl` running with geiser?
<apteryx>C-c C-a
<apteryx>this will load the current module in your buffer
<apteryx>you'll still need to add your built guix source tree to the geiser-guile-load-path variable
<wleslie>how do I name my package in `guix lint` ? do I use the guix package name (like `gcc-10`) or the guile package name (like `(gnu packages gcc)`) or some other way?
<wleslie>it's the guix package
<peanutbutterandc>apteryx, I opened a package, (emacs -nw stand-alone.scm) and then did C-c C-a, which loaded the guix repl, and then went to the stand-alone.scm buffer and did M-x geiser-load-current-buffer and it complains about "no code for module (guix packages)". I am not sure what is going on
<apteryx>what does entering %load-path at the REPL prints?
<apteryx>there should be your guix source tree there
<peanutbutterandc>wleslie, I think package name. Only, you need to have it available to guix. So, for eg: `guix lint -L . my-package-that-is-defined-in-current-directory-as-a-public-variable-in-a-module`
<peanutbutterandc>wleslie, or, if I'm in my guix source checkout, from the root, `./pre-inst-env guix lint name-of-package-that-i-changed`
<apteryx>if not, you need to set it in your .emacs, with something like: (setq geiser-guile-load-path '("~/src/guix")) or the likes.
<peanutbutterandc>apteryx, I'm on a random directory (my channel).... is it possible to set it while running emacs (and how - I'm an emacs n00b). I move around a lot.
<wleslie>thanks, I figured out why that didn't work for me when I first tried it
<apteryx>peanutbutterandc: there should be a menu
<apteryx>Geiser -> Add to load path
<peanutbutterandc>wleslie, If there's anything wrong in the package: any syntax error, anything at all.... it just says 'package not found'. In case of `guix lint -L /path/to/all/personal/.scm-files`, if any of the .scm files have an issue, it fails. Hope that helps!
<peanutbutterandc>apteryx, checking it out now...
<apteryx>wleslie: some packages, like gcc are hidden
<wleslie>yes my package isn't really gcc
<peanutbutterandc>for hidden packages can be linted using '(@ (module) hidden-stuff)'. I've done that a couple times. :D
<peanutbutterandc>I mean '(@@'
<wleslie>I'll make all of mine public until I get them building through (:
<peanutbutterandc>apteryx, Hmmm.... I think this could be because I'm on scheme@(mypackagemodule) right now? o.O
<peanutbutterandc>and even %load-path is an unbound variable
<apteryx>ah, you're trying ot get a REPL on your channel. Hmm. I guess it should simply appear on your load path as well, *after* your guix load path?
<apteryx>remember the file names must match their module name
<peanutbutterandc>apteryx, could you please try this out: I open `emacs -nw chordpro.scm` and then I load guix repl using `C-x C-a` then M-x geiser-load-current-buffer kinda' doesn't work
<peanutbutterandc>whatismore, in C-x C-a, I can't even (use-modules (guix packages))
<peanutbutterandc>,m guix and ,m guile followed by ,use (guix packages) also fails
<peanutbutterandc>I think it's a Guile REPL and not a guix repl
<peanutbutterandc>I want geiser to run `guix repl` and not `guile`
<peanutbutterandc>(I am on a foreign distro, BTW. Emacs, guile, emacs-guix and geiser were all installed via guix)
<scottviteri>I am currently trying to install the guix system with an arch dual boot
<scottviteri>My setup is BIOS/MBR
<scottviteri>I have used the installation media to put guix on one of my primary partitions, but I cannot seem to find the grub config file that was generated
<scottviteri>Where should I be looking?
<scottviteri>My bootload config is standard: (bootloader (bootloader-configuration
<scottviteri> (bootloader grub-bootloader)
<scottviteri> (target "/dev/sdX")))
<apteryx>peanutbutterandc: checking
<peanutbutterandc>apteryx, thank you. I've hit a wall regarding this long ago.
<midnight>peanutbutterandc: What's the 'c' stand for?
<peanutbutterandc>midnight, it's peanutbutterandcrackers but truncated because of freenode's nick length limit
<midnight>\o rad. Sorry for the interruption.
<peanutbutterandc>midnight, not at all. I like the interaction. Esp when I'm despairing over things not going how they're supposed to go. :D
<mroh>peanutbutterandc: If you have emacs-guix, try M-x guix-switch-to-repl
<apteryx>peanutbutterandc: also failing at making the REPL happy so far
<midnight>peanutbutterandc: :-) Well, from this particular tiny chunk of Canada, someone's rooting for you in case it helps.
<apteryx>peanutbutterandc: nevermind, it works
<apteryx>you must have both your channel's root (I cloned your repo at: /tmp/guix-packages/), and Guix's checkout root in your %load-path
<apteryx>and then navigating to chordpro.scm and C-c C-a worked.
<apteryx>one thing which annoys me is this error I sometimes get: arguments: ("icecat-use-older-reveal-hidden-html.patch")
<apteryx>it's because of the use of the local-file macro which doesn't work in Geiser.
<apteryx>that occured when attempting (add-to-load-path "/tmp/guix-packages") followed by ,use (chordpro)
<apteryx>not sure how C-c C-a can work around it.
*apteryx zzzz
<peanutbutterandc>sorry I was away. Had some chores.
<peanutbutterandc>mroh, with asdf.scm open, I only have M-x guix-switch-to-buffer availabe as the solve completion....
<peanutbutterandc>midnight, Well, thank you very much! (:
<peanutbutterandc>apteryx, Hmmm.... I will try this route out then. Thank you very much. Sweet dreams.
<peanutbutterandc>I think I'll just ask things manually for now. Where do I get subsitute-keyword-arguments from?
<scottviteri>I think I have found the answer to my question: I should run guix system init without the no-bootloader option, accept my guix partition as making my primary grub configuration, and add a menu-entry for my arch partition in the guix config
<peanutbutterandc>I could have brute-forced through the #:use-module in a grep-ed package definition and would have found this out already but no: I had to figure out a way to do it with geiser. :[
<str1ngs>peanutbutterandc: (guix utils)
<peanutbutterandc>str1ngs, thanks! Just had that figured out. lol. If you have any tips regarding guix and geiser, please do give me some.
<str1ngs>peanutbutterandc: I use helm-git-grep with something like defin subst key. scheme is easy to grep with flex if you prefix with define
<peanutbutterandc>str1ngs, that is still a little above my knowledge level. But thank you very much nevertheless....
<peanutbutterandc>question: (susbstitute-keyword-arguments) seems to substitute things... I need to extend the same argument.... are there any good examples that anyone knows about which does that?
*guix-vits bwaaah!
<str1ngs>peanutbutterandc: what do need to extend? phases configure-flags?
<peanutbutterandc>str1ngs, Yes!
<peanutbutterandc>I need to add an extra --parameter=stuff
<str1ngs>peanutbutterandc: but build system does it use?
<peanutbutterandc>Ah! it seems substitue-configure-flags does more than what it's name suggests. It seems it can delete and stuff and everything!
<peanutbutterandc>str1ngs, gnu-build-system
<str1ngs>peanutbutterandc: this snippet would modify the configure flags
<peanutbutterandc>str1ngs, wow. That is beautiful. Thank you very much.
<str1ngs>peanutbutterandc: note this handles removing and adding flags as well. you might not need delete or cons*. if you are just adding something you can just use cons
<peanutbutterandc>str1ngs, Yes. It's really elegant. I probably would have written something ugly.
<peanutbutterandc>what's the difference between cons and cons*? (looks up reference manual)
<str1ngs>don't sweat it. I'm assuming you are using this in a inherited package?
<peanutbutterandc>Yes sir
<peanutbutterandc>I just need to add another configure flag that's all
<str1ngs>peanutbutterandc: cons adds one item and the tail is a list. cons* adds multiple items and the tail is a list.
<peanutbutterandc>str1ngs, Oooooh! I can see how that'd be super useful. I could clean up my channels.scm to use cons*, too (if I used multiple channels).
<str1ngs>peanutbutterandc: since you are only adding something you don't need delete either.
<peanutbutterandc>understood. thank you very much.
<str1ngs>no problem
<Brendan[m]2>when i create a new working tree and build, i end up with all these po files as Unstaged Changes for some reason
<scottviteri>What is the expected workflow of adding packages to one's distribution? Are we supposed to install packages and update the guix config? Or can we just add package names to the guix config? I am assuming that we have to do the former, because otherwise the boot sequence would also have to know how to install packages/connect to the internet. Thanks in advance.
<ryanprior>When you add package names and `guix system reconfigure`, it fetches the data ahead of time and updates your profile before you reboot.
<ryanprior>You can even launch your new reconfigured system in a VM to try and get a sense for how it will behave.
<efraim>hello guix!
<guix-vits>Hello efraim.
<scottviteri>how can one launch in a VM
<scottviteri>is it this qemu-system command from
<ryanprior>Right, the manual should be helpful ':D
<scottviteri>haha message received, I'll figure it out
<raghavgururajan>sneek, later tell nckx: The system freeze I was telling you about, it happens like this
***mmohammadi98129 is now known as mmohammadi9812
***apteryx is now known as Guest64583
***apteryx_ is now known as apteryx
<mfg>Hi guix :)
<mfg>so i just wanted to build guix from git, and inside a guix env guix --pure environment the configure phase fails, stating that it can't find guile development packages. Has anyone had the same problem before? if so how did you solve it? :|
<cbaines>mfg, can you share the error you get via ?
<mfg>cbaines: solved it, i forgot to rerun configure and some old profile paths didn't exist anyomre...
<cbaines>Good good :)
<moesasji>mfg. I definitely remember hitting this and it seems to be a known issue when using guix on a foreign distro. My recollection is that you need to add something to the environment command to avoid using the system guile.
<moesasji>@mfg ignore comment...
***daviid is now known as Guest77937
<mfg>moesasji: i also use guix on a foreign distro so good to know what i have to do if i encounter the problem there :)
<mfg>Is it expected that grafting takes /very/ long when building with --system=armhf-linux? (ca. 15 minutes)
<mfg>and still running
<roptat>grafting is supposed to be fast, but for very big packages, it can take some time
<roptat>I haven't counted, but for a ~500MiB package it can take maybe 10 minutes on my x86_64 machine
<mfg>okay then something must be hanging
<cbaines>mfg, because of --system=armhf-linux, you might be grafting using QEMU, which is going to be slower than native grafting
***jonsger1 is now known as jonsger
***jonsger1 is now known as jonsger
<civodul>hey hackers!
<civodul>↑ a childhurd in a Guix VM :-)
<guix-vits>civodul: How to connect to childhurd?
<civodul>it's all documented, but basically "ssh root@localhost -p 10022" with the default settings
<civodul>or: vncviewer localhost:5900
<guix-vits>Thank U.
*guix-vits OvO red eyes forever.
<civodul>those image-dir.drv take ages to build
*guix-vits bwaah.
<nly>weird, ssh localhost -p 10022 says Connection reset after a while
<sneek>Welcome back nly, you have 2 messages!
<sneek>nly, str1ngs says: can you check the desktop file for nomad. seem StartupWMClass=Emacs is wrong.
<sneek>nly, str1ngs says: could you also update the command to be more descriptive. thank you
<nly>sudo herd status childhurd
<nly>Status of hurd-vm: It is started.
<guix-vits>civodul: ^ meow?
<civodul>nly: try restarting it
<civodul>there seems to be race conditions or something with the secret service
<nly>civodul: same thing
<nly>stuck at "start ext2fs: Hurd server bootstrap: ext2fs[device:hd0s1] exec"
<civodul>oh, then it fails to boot
<nly>ah, crap
<PotentialUser-77>Hello! Does anyone have a recommended way to run virtual machines in Guix? GNU Boxes sort of works, but it won't let me pass through USB devices (trying to enable pass-through for devices just immediately switches off when I click it) and I'm having trouble figuring out how to configure file sharing. virt-manager, on the other hand, isn't connectin
<PotentialUser-77>g to KVM unless I run it with sudo.
<PotentialUser-77>I already have my Guix installation configured to run libvirtd on startup and my user is part of the kvm group.
<cbaines>PotentialUser-77, is your user in the libvirt group?
<PotentialUser-77>I think it gave me an error when I tried to do that?
<PotentialUser-77>Although I might've tried to put my user in a "libvirtd" group
***jonsger1 is now known as jonsger
***zava_ is now known as zva
<joshuaBPMan>Does anyone know if is libre software?
<apteryx>rekado: any idea why one patch is missing from: I see the 1/3 patch in debbugs.
<mfg>cbaines: yes i expected it to be slower, but i didn't expect it to be that much slower :D
<janneke>civodul: oh, a childhurd in a VM! Also a wonderful patch series, what a great present on this Sunday.
<scottviteri>I know there must be simple answer for this, but I am somehow missing it in the documentation. How do I set the password to run sudo?
<joshuaBPMan>janneke didn't we already have a childhurd running in a vm as a service?
<joshuaBPMan>scottviteri : Have you set the root password? gpasswd root RET password RET password
<scottviteri>I used "passwd"
<scottviteri>Does it matter
<joshuaBPMan>scottviteri you did it right then. I mispelled.
<joshuaBPMan>You might need to add your regular user to your suoders file. I think you can do that via the config file too.
<joshuaBPMan>scottviteri under the operating-system you can specify a suoders-file.
<joshuaBPMan>scroll down a little bit and you will see an example
<janneke>joshuaBPMan: sure, so maybe i misread -- i thought that through the staged /dev/kvm fix (possibly together with --cpu base) we would be able to run a hurd-vm inside a linux vm
<janneke>anyway a great patch series by civodul to mature the childhurd service
<joshuaBPMan>janneke You would probably know more about it than me. I was just curious what you meant. Thanks for explaining. :)
<civodul>janneke: ah yes, being able to test it in a VM is nice
<civodul>it's a series of small things that help
<joshuaBPMan>also did you guys see my email about setting up the Hurd vm to use the e1000 network driver?
*civodul didn't (yet!)
<joshuaBPMan>apparently, using the e1000 network driver is the "best" way to current set up networking in the Hurd.
<joshuaBPMan>I guess there is some bug in netdde...let me see if I can find the email.
<janneke>civodul: *love it*
<civodul>janneke: heheh, cool! :-)
<joshuaBPMan>Well bummer I can't find the email.
<joshuaBPMan>That's the recent email that says the e1000 driver is a good way to avoid a bug in netdde
***zava_ is now known as zav31
<rekado>apteryx: that’s a good question. I don’t know yet.
<rekado>apteryx: mumi works on the original debbugs data, so perhaps there’s a problem chopping up and parsing the so-called “log” file for this bug.
<rekado>apteryx: looking at the corresponding log file I don’t see anything unusual about it, so perhaps the parsing of the mail itself failed. I’ll look into this some time later.
<cbaines>after quite a bit of mutilation, I have managed to cross-compile a guix-build-coordinator package for the hurd...
<cbaines>I had to remove sqitch (because perl), guile-fibers and guile-prometheus (probably my fault) as inputs
<cbaines>I also had to add most of the Guile dependencies as native-inputs as well as inputs
<cbaines>but, after all that, it does actually build, and I don't think any of those removed inputs are used in the agent part
<cbaines>so there's a chance it might work in a hurd VM
<PotentialUser-69>Greetings, I am having an issue with guix copy. My server can to my host, but can't from. My host can't do either to or from.
<PotentialUser-69>I have created keys with guix archive and authorized on both machines each other's .pub
<PotentialUser-69>I also have guile-ssh installed on both machines
<PotentialUser-77>cbaines: So my user is now part of both the kvm and libvirt groups. Trying to connect virt-manager to KVM results in the following error:
<PotentialUser-77>libvirt.libvirtError: Failed to connect socket to '/var/run/libvirt/libvirt-sock': Permission denied
<cbaines>Ok, who owns that file?
<cbaines>For me it's root, with libvirt as the group
<janneke>cbaines: that souns cool...almost afaid to ask: what is guix-build-coordinator?
<cbaines>PotentialUser-77, you'll probably need to log out and back in again if you've only just added your user to the libvirt group
<PotentialUser-77>cbaines: I already reconfigured and rebooted with my user now part of libvirt
<cbaines>janneke, no need to be afraid, it's something I've been working on for the last few months
<PotentialUser-77>cbaines: I went and checked the file, it's owned by root with root as the group.
<PotentialUser-69>PotentialUser-77: What does your libvirt service look like?
<PotentialUser-77>How do I check that?
<cbaines>PotentialUser-77, I have (unix-sock-group "libvirt") in my libvirt-configuration, which could be relevant
<PotentialUser-77>(list (service libvirt-service-type)))
<PotentialUser-69>You will want something like
<PotentialUser-77>Oh, sorry. That's the OS config
<janneke>cbaines: thanks -- ah, and the tool uses information from the corresponding guix service, right?
<PotentialUser-77>Where is libvirt-configuration?
<PotentialUser-69>(service libvirt-service-type (libvirt-configuration (unix-sock-group "libvirt")))
<PotentialUser-69>It will be in your services in your config.scm
<cbaines>janneke, assuming you mean corresponding Guix Data Service, it can, but not necessarily. I'm using it in conjunction with the Guix Data Service, as that provides information about what derivations can be built, as well as providing substitutes for the derivations
<PotentialUser-77>PotentialUser-69: Ah, okay. So I was looking at the right config. I guess my config of it is wrong though
<cbaines>PotentialUser-77, from the docs, the default unix-sock-group is root, I'd suggest changing it to libvirt
<PotentialUser-69>Great, add in that additional code and restart and it should work. I just checked my virt-manager and it does connect without that error.
<PotentialUser-77>Alright! Thank you both!
<cbaines>janneke, I'm excited about the Guix Build Coordinator because it can do things like automatic retries, building things more than once, and offers a different take on how to provide substitutes
<PotentialUser-69>PotentialUser-77: \o/
<drakonis>guix build coordinator
<drakonis>is this the hydra rewrite?
<cbaines>drakonis, nope, at least that's not my intention
<cbaines>Cuirass I think was intended to be in some way compatible, at least initially
<PotentialUser-77>Alright, so now I'm getting an error about the system_bus_socket
<PotentialUser-77>libvirt.libvirtError: internal error: Unable to get DBus system bus connection: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or directory
<PotentialUser-77>The file it's trying to access has root as its user and root for its group
<civodul>cbaines: kudos for the (degraded) coordinator on GNU/Hurd!
<lafrenierejm>Anyone available to help troubleshoot a CTAN package recipe?
<civodul>lafrenierejm: hi! i'm afraid i'm not the right person
<civodul>Paul Garlick has been playing with that recently it seems
<civodul>so you can try emailing guix-devel maybe?
<joshuaBPMan>It's probably best to email help-guix.
<PotentialUser-77>It looks like I actually have the same issue on gnome-boxes as with virt-manager. The difference is I didn't realize it with gnome-boxes because I only saw it when I started it from a terminal.
<PotentialUser-77>"(gnome-boxes:4822): Boxes-WARNING **: 17:00:03.330: libvirt-machine.vala:275: Failed to connection to system libvirt: Unable to open qemu+unix:///system: internal error: Unable to get DBus system bus connection: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or directory"
<PotentialUser-77>So I guess in general I have an issue with getting virtual machine managers to connect to the DBUS system bus socket
<cbaines>civodul, thanks, I've only attempted to build the package so far, but hopefully I can attempt to actually get it running sometime soon...
<cbaines>PotentialUser-77, does that file exist, or does it not as the error suggests?
<cbaines>It exists on the machine I'm currently using
<mroh>PotentialUser-77: try to add your $user to the "root" group, maybe?