IRC channel logs


back to list of logs

<Luk6655>looks like /var is the localstatedir by default
<apteryx>Luk6655: /usr/local I think
<Luk6655>no, definitely /var/guix
<Luk6655>however, unfortunately the installer failed half way... and now of course I can't re-run it unless I cleanup everything it did manually... unless there is an uninstaller, perhaps?
<apteryx>Luk6655: err, /usr/local/var
<apteryx>Luk6655: uninstalling should be as easy as 'rm -rf /gnu /var/guix'
<Luk6655>perhaps it is different on different distros, for me it is _for_sure_ /var/guix
<Luk6655>also this search for thisd string "Usually, this localstatedir option is set to the value /var."
<Luk6655>apteryx: unfortunately before it failed it managed to create all its users and services. I suppose if I don't clean those up manually it will fail on subsequent rerun
<apteryx>ah right, the distribution may have a differen value for it, but GNU standards mention it should be /usr/local/var, and this is what Autotools chooses by default
<Luk6655>have a look in the manual, please, also the install script has it hardcoded as /var
<Luk6655>we are talking about localstatedir, right?
<apteryx>Luk6655: I don't think it'll fail, I've tested the installer with multiple runs in a row before, and simply wiping the above directories was enough
<apteryx>if you're talking about
<Luk6655>ok, good to know, yes I am
<Luk6655>it is line 344 in that puts it in /var/guix
<Luk6655>unless this is something else...
<Luk6655>btw, the manual says to run the install script as root, does this mean with sudo, or logged in as root?
<Luk6655>I've run it via sudo and it failed when I answered yes to authorise binary downloads, it failed to find substitute keys
<Luk6655>ok, I know what happened, I had /gnu and /var/guix in the fstab (not mounted) prepared to copy the store and localstatedir to other partitions later. When it installed the service and mounted /gnu/store it must have mounted those entries in fstab overwriting the default profile it placed in the localstatedir. That's why it failed to find the keys
<Luk6655>it seemed to have finished sucefully on rerun afet I wiped those folders
<Luk6655>has anyone seen guix pull to fail like this: guix pull: error: bind: Permission denied ? (this is first guix pull after installing on a foreign distro)
<pkill9>Luk6655: possibly something owned by root in I believe ~/.cache/guix/<something> as a result of running a guix commadn as root with sudo
<pkill9>possibly ~/.cache/guix/checkouts
<Luk6655>yes, it seems likely...
<pkill9>run ls -l ~/.cache/guix/checkouts and see
<Luk6655>I'm just running guix pull via sudo to see if it works like this
<Luk6655>pkill9: have you used guix on a foregn distro? If yes, I wonder if you installed via sudo or logged in as root
<pkill9>I have ages ago
<Luk6655>the manual just says: run it as root
<Luk6655>so I have run it via sudo, but now I think I should've logged in as root
<Luk6655>as guix pull run via sudo did work
<pkill9>well have you checked that directory I mentioned?
<Luk6655>I did just now, everything in it is owned by my normal user
<Luk6655>perhaps one needs to run the first guix pull as root too?
<Luk6655>bind suggests it was trying to bind to a network socket for some reason
<pkill9>may be bind- mountinga directory
<Luk6655>I'm retrying without channels.scm so it only pulls from guix
<Luk6655>this will take a long while...
<Luk6655>should guix pull be mounting anything ?
<Luk6655>also the install script said something about it being recommended to install my distro's name service... perhaps it's absence has something to do with it?
<Luk6655>this is nscd service
<Luk6655>no, it fails exactly the same when pulling from the default guix channel
<Luk6655>that's the entire output (not much of it)
***jesopo is now known as jess
<Kabouik>I am getting this error when doing guix pull to refresh my private channel, but my private channel has nothing to do with that package. What am I missing? That is the private channel:
<apteryx>Kabouik: arg... seems I made another mistake
*apteryx looking
<apteryx>Kabouik: should be fixed with 0a92a64e59, apologies for the trouble!
<apteryx>(i.e., try guix pull now, it should be fixed)
<Kabouik>Oh cool, so that was not on my end! That's hard to believe, I have been struggling with these packages all night and suddenly when I finally get them working, there's an issue elsewhere that's not my fault! I didn't even consider that, I was trying to find what was wrong again in my packages. :p
<zamfofex>Would you look at that?! I (kinda) did it!
<apteryx>Kabouik: ah :-)
<apteryx>ah, this is cool. if you have virt-viewer installed in your user profile and pass the '-display spice-app' arguments to the qemu (or script, it launches in virt-viewer.
<zamfofex>sneek: later tell civodul: I was able to set up a simple substitute server for the Hurd! 🎉 <> For now, it only provides Hello with those tests disabled, but that should be enough to allow people to skip much of the bootstrapping process when trying it out, I feel. Also, I assume you’re partaking on the birthday event, so I enjoy it (or have enjoyed it)! 🎉
<vivien>Dear guix, I had to install evolution-data-server in my profile to get evolution to run. It did not remember any of my configured accounts, and when I tried to create a new one, I got a message box with something like The name org.gnome.evolution.dataserver.Sources5 was not provided by any .service files
<vivien>I grepped for that file in my store, found out it was provided by evolution-data-server, installed that to my profile, and i got my accounts back.
***taiju is now known as Guest3934
***taiju` is now known as taiju
***taiju_ is now known as taiju
***taiju` is now known as taiju
***iyzsong-alt is now known as iyzsong
***iyzsong-alt is now known as iyzsong
<cbaines>is anyone watching the 10 years event livestream?
<cbaines>the first session is about to start
<jonsger[m]>cbaines: seems to work :)
<cbaines>thanks goes to the people doing the video and who setup the livestream!
<Kabouik>How can I alter this line so that there is a space between the two configure options? I checked the guix build logs and that is what it does: configure: WARNING: unrecognized options: --enable-login--with-plugindir
<jpoiret>Kabouik: :configure-flags takes a list of arguments, here you're just supplying one that's the concatenation of all the strings you wrote down
<Kabouik>So it should be (list "--enable-login" (string-append ...))?
<jpoiret>i suggest (list "--enable-login" (string-append "--with-plugindir=" (assoc-ref %outputs "out") "/lib/sasl2"))
<jpoiret>exactly :)
<Kabouik>Let's try. I hope it works, I'm stuck with Outlook Web App since days because of that, the pain is real
<silver->hey is there a low quality stream
<xelxebar>Happy Birthday, Guix!
<sneek>Welcome back xelxebar, you have 1 message!
<sneek>xelxebar, rekado says: the code is here
***Dynom_ is now known as Guest5747
<polyex>guix fast and secure enough to use on a server?
<polyex>high traffic sites, ecommerce, stuff like that
<Kabouik>That worked jpoiret. :] I'll update the patch #57021 later today.
<unwox>polyex: guix should be as fast as any other distribution in this regard. not so sure about security though
<Unthahorsten>What is the recommanded way to deploy and use packages builded by GUIX on a closed cluster without neither open network, nor GUIX nor root access and only through a shared directory ?
<Luk6655>is there docker available?
<Luk6655>there is also guix pack
<rekado>Unthahorsten: on our HPC cluster we host Guix like this: a high-availability NFS file server that exposes /gnu/store and /var/guix to all cluster nodes; the active node on the file server runs guix-daemon, which listens for connections from cluster nodes on TCP.
<rekado>other cluster nodes mount /gnu/store and /var/guix from main server, set GUIX_DAEMON_SOCKET to guix://<server>:1234, and talk to the daemon there.
<rekado>when installing packages a cluster node really just tells the main server what to do and then sees the packages in the shared /gnu/store
<rekado>only one server needs to have root permissions: the main server needs to have permissions to host /gnu/store and export over NFS; and it needs root to run guix-daemon.
<Unthahorsten>Thanks, I cannot mount to /gnu/store I have another mount point, and I cannot install guix on the cluster. Is there a way to still use packages binaries ?
<Luk6655>you can generate a binary tarball that contains all dependencies, or a docker image
<Luk6655>there is also --relocatable option so those binaries can be run without root (this is what I would use)
<Unthahorsten>Thanks very interesting ! But I dont have user namespace :(
<rekado>without user namespaces you can use a different backend, such as the proot backend
<rekado>it’s slower, but still allows you to run relocatable applications
<Unthahorsten>Is there a way with script using chroot or pivot_root ?
<rekado>“guix pack” generates archives that you unpack anywhere and then use.
<rekado>it’s not the preferred way of using Guix, because there’s no sharing.
<Luk6655>that's what the --relocatable option is for, so the resulting tarball canbe unpacked anywhere on the filesystem (although I never used it myself)
<rekado>I used it for deploying web applications to a system where I don’t have Guix
<rekado>it works fine, but obviously having a full Guix installation on the target is much more flexible.
<Unthahorsten>Thanks that should do the job
***taiju_ is now known as taiju
***califax_ is now known as califax
<Kabouik>Thank you for pushing shirah cbaines! That was my first accepted patch. :]
<PotentialUser-24>Hey, can someone please give me a hand with something real quick?
<tschilptschilp23>Hi guix! Is it possible to add a channel definition right into a manifest to be used with 'guix shell -m MANIFEST.scm'?
<tschilptschilp23>* manifest file
<lechner>Hi, i know how to create derivative packages via inherit, but i use a new package name. how can i allow my patched package to use the original name in the package name space, please?
*tschilptschilp23 thinks to have found a use-case for 'guix time-machine'
*abcdw is trying to fix super key handling problem in emacs-next-pgtk
<the_tubular>Is guix days over ?
<TopExpert>why does this Guile code work (set-difference '("one" "two") '("one" "one")), but not this? (set-difference %base-initrd-modules '("ahci"))
<TopExpert>looks like a bug to me, but is it something specific to Guix?
<TopExpert>(use-modules (gnu)) to get %base-initrd-modules
<TopExpert>is it just for me set-difference isn't working?
<TopExpert>I tried, (lset-difference) (lset-difference!), none work on the %base-initrd-modules variable, the only thing that works is (delete "ahci" (delete "xyz" (delete "three" %base-initrd-modules))
<jpoiret>lechner: I don't really understand what you mean by "use the original name"? If you inherit from another package, you'll also inherit the name field
<jpoiret>TopExpert: where did you find set-difference? I only could find lset-difference in (srfi sfri-1)
<TopExpert>jpoiret: (use-modules (ice-9 common-list))
<jpoiret>note that lset-difference needs an equality predicate
<TopExpert>yes (lset-difference eqv? %base-initrd-modules '("ahci")) not working
<TopExpert>wondering if my system is messed up or something
<jpoiret>TopExpert: that module doesn't seem to be documented in the Guile 3 docs
<TopExpert>yeah kind of running into all corner cases as usual
<oriansj>well thank goodness, we need people to hit those corner cases so that we know about what we missed.
<jpoiret>TopExpert: eqv? will not check for string equality
<jpoiret>see 6.9.1 in the Guile manual
<jpoiret>you will need string?
<TopExpert>(ice-9 common-list)
<TopExpert>jpoiret: even with (lset-difference eq?), it doesn't work for me
<jpoiret>sorry, equal?
<jpoiret>eq? will not work either
<TopExpert>jpoiret: they look like strings, I checked with (class-of (car %initrd-modules))
<TopExpert>and (delete "ahci") works
<jpoiret>eq?, eqv? and equal? have different behaviour depending on the specific object you're looking at
<TopExpert>jpoiret: oh my god thank you
<TopExpert>equal? worked
<jpoiret>your first example works because of common subexpression elimination (most likely), since all the strings will only be computed once and thus will be actually equal through eq? and eqv?
<TopExpert>yeah I understand now, thanks a bunch :)
<oriansj>the functional differences between them is for performance reasons and generally most of them should be avoided if you don't know what you are doing.
<TopExpert>yep, trying to learn guile on the go while configuring Guix
<tschilptschilp23>I'm just trying to run 'guix time-machine -C my_channel.scm -- guix shell -m testmanifest.scm', but get the answer "guix: guix: command not found\nhint: Did you mean `git'?". Do I want too much, or am I using a wrong approach? What I want is to create an environment as defined in 'testmanifest.scm', but with package versions from an earlier commit. 'my_channel.scm' contains 'testmanifest.scm' con
<tschilptschilp23> The contents are chosen 'randomly' -- I just needed a package that was updated on the current guix (3ba8c2b5076cff874f12e2880632ec5e5507acff).
<flypaper-ultimat>tschilptschilp23: after the -- of time-machine you dont need to type guix again
<flypaper-ultimat>you'd type
<flypaper-ultimat>guix time-machine -C my_channel.scm -- shell -m testmanifest.scm
<tschilptschilp23>flypaper-ultimat: whoooo! thanks a lot, holy grail alarm :)
<apteryx>are the /var/guix/drvs logs ever cleaned up?
<apteryx>mine weighs 9.1 GiB
<dthompson>hey the 10 years of guix livestream is awesome thank you!!
<dthompson>I'm in the US so I miss a lot before I can tune in but I caught pjotr's talk and it was great
<apteryx>agreed! there's also the ##guix10years channel in case you missed it
<dthompson>I did miss that
<tschilptschilp23>where would I find the livestream?
<tschilptschilp23>ggoes: thank you, I guess I'm already late for today...
<apteryx>it's restarting
<tschilptschilp23>just noticed :)
<Kabouik>What is the recommended way to use printers on Guix? I have the cups service running, installed system-config-printer and from there set up a printer, but I am getting when trying to print a test page.
<Kabouik>I know printers are not easy to set up on some distributions, I don't know if Guix is one of them
<dirtcastle>After installation finally my first full system configuration just completed. I just wanted to say I'm so happy.
<tschilptschilp23>Kabouik: I remember an issue like that, and it turned out to be gnome -- have you tried printing a real document already?
<tschilptschilp23>in my case the missing file actually was the testpage itself
<Kabouik>I haven't no, let's try. I'm not using Gnome though (well, there's just this UI for system-config-printer). If that doesn'twork, I'll try that:
<tschilptschilp23>I guess I was mixing up gnome and system-config-printer, good luck!
<Kabouik>Unfortunately no, real documents don't seem to arrive in the job list of the printer
<Kabouik>Self test pages work, though. Weird.
<tschilptschilp23>Printers are quite of a miracle to me anyway, so I guess I'm not the best consultant on that. At a certain point things started to work nicely, but in the meantime my printer itself died. Which reminds me, that I actually need a new one. To be able to co-battle on this topic again...
<Kabouik>The `lp` program could perhaps help debugging by sending files to the printer from CLI but we don't have it in Guix
<Kabouik>I got to go, I have to give up on that for now.
*abcdw found out that bin/ffplay is missing in ffmpeg@4
<lechner>jpoiret: will my customized package replace the original as an input elsewhere even if i assigned a new variable name to it ("define-public")?
<ulfvonbelow>readlink -f $(which lp) => /gnu/store/4g8fi4ii4ikdbfns46cd1a21bpx32fi5-cups-2.3.3op2/bin/lp
<unmatched-paren>lechner: no
<lechner>unmatched-paren: thanks, is there a way?
<lechner>hi, i am trying to test an updated package definition in git but ./bootstrap cannot find autopoint. what is the proper fix, please?
<lechner>okay, i am supposed to run guix environment guix --pure first, right?
<apteryx>my system seems to be confusing user ids for a service
<apteryx>/var/lib/gdm is owned by id 955, group 967, which no longer exist
<apteryx>and it contains entries owned by the opendht user (I used the opendht-service-type before).
<apteryx>it seems to recycle old user IDs that are no longer active on my system
<lechner>Hi, and then outside 'guix environment' ./pre-inst-env guix build xkeyboard-config does not use my new package definition (i can tell from the version) while inside 'guix environment' the same command results in this error: ./pre-inst-env: line 55: exec: guix: cannot execute: Is a directory What am I doing wrong, please?
<morganw>apteryx: I think if you remove and re-add gdm is will break for that reason, because the permissions for logging will likely be wrong. I think there was an open issue about how to handle user IDs but the solution wasn't clear.
<ulfvonbelow>lechner: that sounds quite strange! What does `./pre-inst-env sh -c 'echo $PATH'` say?
<unmatched-paren>lechner: i have no idea what the best way is
<mbakke>libgc segfaults when booting a 'guix system vm' from core-updates
<ulfvonbelow>it looks like scripts/guix probably hasn't been created from scripts/ yet. That's done when you run "make". If you try running './pre-inst-env guix ...' before you run "make" for the first time, it'll look for ./scripts/guix, not find it, and try looking for ./guix instead (because that's the next entry in PATH that pre-inst-env adds), but that's actually a directory.
<ulfvonbelow>lechner: have you run './configure --localstatedir=/var' and 'make' yet?
<lechner>ulfvonbelow: no. also,
<lechner>ulfvonbelow: i am following
<lechner>ulfvonbelow: i see you already found the problem above. thanks!
<lechner>unmatched-paren: there should be a way to replace an offical package with my own
<ulfvonbelow>lechner: if you're using a local checkout, you could just directly substitute it in the code. Give your package definition the same name and delete the other one.
<ulfvonbelow>alternatively, if you just want to "try it and see" without having to delete the original, you can use package transformations wherever you want it replaced, e.g. 'guix build <somepackage> --with-input=<original-package>=<my-modified-package>' will build <somepackage> with every occurrence of <original-package> replaced with <my-modified-package>.
<lechner>ulfvonbelow: I would like to add a custom layout to xkeyboard-config but do not think the modification is necessarily interesting for others. Can I make the change locally?
***meo is now known as protomeo
<lechner>hi, building guix from Git i see warnings like ;;; no code for module (semver ranges) is that expected?
<ulfvonbelow>lechner: changes made locally (by editing the file in the git repository) won't affect others' versions
<ulfvonbelow>according to secton 2.2 of the manual, guile-semver is an optional dependency necessary only for the 'crate' importer
<lechner>ulfvonbelow: thanks for the semver answer
<lechner>ulfvonbelow: i would like to use a custom keyboard layout but not necessarily a custom version of guix
<ulfvonbelow>in that case, you'd need to tell xorg to use your custom xkeyboard-config package. You can do that by setting your xorg-configuration (via the set-xorg-configuration service in your system configuration) to have the 'xorg' field point to an xorg that's been transformed by replacing its xkeyboard-config input with your custom package. You would also need to define your custom xkeyboard-config outside of the guix repository, either in a
<ulfvonbelow>channel to be used by 'guix pull' or somewhere GUIX_PACKAGE_PATH can find it or somewhere you specify with -L on the command line
<ulfvonbelow>you'll want to look at package-input-rewriting in section 9.3 "Defining Package Variants" of the manual, and at set-xorg-configuration and the xorg-configuration record type in section 12.9.7 "X Window" of the manual
<lechner>ulfvonbelow: thanks for those hints! do i have to do something else for grub or console?
<ulfvonbelow>that's a bit tricky. keyboard-layout->console-keymap in (gnu system keyboard) does accept an optional keyword argument to specify the xkeyboard-config package to use, but where it's called in (gnu bootloader grub) and (gnu system linux-initrd) it doesn't pass along any argument.
<ulfvonbelow>because it's evaluated at the time the procedure is called, however, there is probably a cheeky solution you can use: in your system's config.scm, just add (set! xkeyboard-config my-xkeyboard-config), where my-xkeyboard-config is your custom package's scheme variable name.
<ulfvonbelow>it might be good for a <keyboard-layout> to be able to specify an xkeyboard-config package
<lechner>Hi, am I supposed to run 'guix style' when updating a package version even if it obscures my changes? should i submit two back-to-back patches instead? without applying style: with applying style:
<lechner>ulfvonbelow: thanks for that pointer. my skills right now are probably too low to add an argument or named field for the package, but i will try
<Cairn>Is there a way to source an earlier substitute, if the current substitute failed?
<Cairn>Like, would time-machine use the earlier substitute if I found the right commit?
<mubarak>Hi guix
<apteryx>OK, my gdm issue is definitely a permission issue
<apteryx>I've added a ls -alR /var/lib/gdm into the activation script, and it looks like:
<apteryx>(i.e. it's a mess)
<ulfvonbelow>lechner: adding an extra field for <keyboard-layout> would require modifying guix, so probably best not to worry about it right now as long as the workaround with set! works
<ulfvonbelow>regarding 'guix style', I don't know.
<lechner>ulfvonbelow: okay. thanks so much for your help today!
<apteryx>there's already some logic to straighten permissions, but it only kicks in if /var/lib/gdm has the wrong permissions, which is not the case here
<apteryx>it also would only fix the directories, not files anyway
<apteryx>ok, no, it would have fixed directories + files
<mubarak>I would like to ask a question. I installed multible packages"guix install vlc qbittorrent icecat hexchat", because of slow internet speed, guix start downloading icecat source code and started compiling. when I compile with my laptop(Core i3) it start freezing. I waited for it for almost four hours and it didn't finish compiling icecat. Every time
<mubarak>I want to install icecat it start compiling. So my question is how to make guix download and install only binary packages and never compile a package? sometimes I'am in hurry and I cant wait 2 days to compile one package.
<lechner>ulfvonbelow: now i think i need some help with my fonts. which codepoint did you send, please?
<zamfofex>mubarak: Usually, if there are substitutes available, it’ll prefer downloading them instead of compiling the packages from source.
<mubarak>zamfofex, is there any option i should specify to guix so it search for substitutes only?
<ulfvonbelow>codepoint 128077 (dec) / #x1f44d (hex), name: THUMBS UP SIGN
<ElishaEmmaunel[m>Do you have a Bitcoin wallet or Coinbase wallet?... (full message at <>)
<drakonis>ha, matrix.
<zamfofex>mubarak: What do you get when you type ‘guix describe’? I’m not sure if you can request for it to only use substitutes for packages.
<ulfvonbelow>make sure to check the manual application notes for fonts (section 2.6.3 "X11 Fonts")
<ElishaEmmaunel[m>Do you have a Bitcoin wallet or Coinbase wallet?... (full message at <>)
<ElishaEmmaunel[m>Do you have a Bitcoin wallet or Coinbase wallet?... (full message at <>)
<apteryx>rekado: hi!
<lechner>ulfvonbelow: thanks!
<lechner>Hi, is it okay two send two patches to -patches in one message if the second is only for style?
<Cairn>Rush packaging is wild. I'm just going down the dependency chain and breaking things, just got bump a version
<Cairn>*just to
<lechner>Hi, is it okay to ignore this lint comment, please? gnu/packages/xorg.scm:3894:19: xkeyboard-config@2.36: label 'gettext' does not match package name 'gettext-minimal'
<lechner>Hi, what can I do about this, please? gnu/packages/xorg.scm:3884:12: xkeyboard-config@2.36: source not archived on Software Heritage and missing from the Disarchive database
<lechner>Hi, do I need to split this string to make the line shorter, or is it okay to leave it?
<podiki[m]>source not archived message you can ignore (is informational, I think might trigger a fetch to be archived automatically?)
<apteryx>lechner: about labels, you could remove them entirely for xkeyboard-config
<apteryx>then the warnings will be solved.
<podiki[m]>you can make that string shorter by using the name field (instead of writing out xkeyboard-config)
<podiki[m]>and yes, you can remove the labels to the new style, much nicer :)
<apteryx>gdm works after making /var/lib/gdm a tmpfs
<apteryx>(I had permission issues)
<apteryx>could someone tell me if it remembers their session correctly? for me, it alawys defaults to the first session available in the gear icons
<apteryx>gear icon menu*
<polyex>any infosec researchers use guix? curious how its security potential is compared to other distros
<apteryx>hm, perhaps because I do not use GNOME as a whole
<lechner>Hi, how can I add the %xorg-updater to the package definition for xkeyboard-config, please?
<jpoiret>polyex: i'm not sure there are many different pros and cons compared to other distros, except for the clearer chain of trust and being source based
<polyex>ok so basically same as other distros, cool
<lechner>apteryx podiki[m]: hi, about labels, just drop the quasiquoters and use a straight list as is done elsewhere?
<apteryx>lechner: exactly
<apteryx>if there are non "out" outputs involved, then they'll have to be like `(,glib "bin), for example
<apteryx>and you'll have a lint warning again, which you can ignore this time...
<lechner>apteryx: thanks!
<gnucode>anyone using sway?
<gnucode>gnome-clocks is not starting for me...
<GNUtoo>gnucode: yes, I use sway
<GNUtoo>I had to do a hack to be able to start it from kdm
<gnucode>GNUtoo: do you wanna try 'guix package -i gnome-clocks; gnome clocks;'
<gnucode>I am getting an error.
<gnucode> Couldn't open cannot open shared object file: No such file or directory
<GNUtoo>I've a multi-boot system and right now I'm booted on Parabola with Guix on top, would that help?
<GNUtoo>Also I'm on i686
<GNUtoo>I can reboot if needed but it'll take a bit of time
<jpoiret>gnucode: are you on Guix system? How is gnome-clocks installed?
<gnucode>jpoiret: I am on guix system
<gnucode> commit: a7af25ad313bbc10d42801c48cd16dfda53290a1
<gnucode>jpoiret: I just did a 'guix install gnome-clocks'
<gnucode>I am using sway.
<gnucode>GNUtoo: sure. See what happens.
<GNUtoo>On top of Parabola that doesn't make that error, though I'm not sure what it's supposed to do
<GNUtoo>ahh it's a bit slow to start
<GNUtoo>I'll reboot at some point and try that in Guix
<GNUtoo>*guix system
<gnucode>sounds good.
***rgherdt_ is now known as rgherdt
<jpoiret>gnucode: libGLES missing seems weird
<jpoiret>i'm not very up to speed with how openGL should be loaded though, I know there are some libglvnd shenanigans on top of the already hard to understand mesa
<jpoiret>but it isn't surprising given that error message that it works well on parabola: it just loads the host's .so file
<gnucode>jpoiret: my computer only supports openGL 1. Not 2. I wonder if that is the issue.
<gnucode>at least I think it doesn't support GL2. I've got a thinkpad T400.
*GNUtoo needs to rest, I'll try to test later
<GNUtoo>The error looks strange though
<GNUtoo>(even T400 is supposed to support GLES AFAIK, otherwise a lot of things woudn't work)
*GNUtoo only had issues with some games with an X200
<GNUtoo>gnucode: did you try not to use guix package -i but add it in the system.scm instead?
<tschilptschilp23>I'm trying to define a go package with this definition:, but it fails with this backtrace: I guess it's not really meant to be built with guix' go-build-system from git source, as upstream just writes about choco-, brew- and npm-package managers on their github-site. Anyway, does anyone see what might be going wrong?
<ElishaEmmaunel[m>Do you have a Bitcoin wallet or Coinbase wallet?... (full message at <>)
<ElishaEmmaunel[m>Do you have a Bitcoin wallet or Coinbase wallet?... (full message at <>)
<polyex>biggest guix server installation known of?
<rekado>polyex: big in what way?
<polyex>number of servers and traffic
<rekado>we have a couple of really big deployments at HPC centers
<polyex>more looking for public facing stuff powering web sites
<rekado>it’s ~25 servers and serves substitutes to all Guix users.
<polyex>im wondering what it's like to run a large fleet of servers for a public web site. how it handles varied loads, network attacks, how deployment/cm/monitoring and all that stuff are
<polyex>guix servers*
<polyex>as compared to like RHEL and BSD's
<polyex>i think the declarative nature of guix could make it amazing for that
<rekado>I don’t see any difference between Guix and RHEL in terms of reliability for serving web stuff.
<polyex>and nothing about guix that could maybe make its security worse?
<polyex>seems like a linux distro is just whatever stuff is bundled with the kernel, so all you really need to start is the kernel (librelinux, deblobbed ofc) the declarative system manager (guix) and a system config to feed into it
<mbakke>apteryx: awesome that you upstreamed the CMake certificate patch :-)