IRC channel logs


back to list of logs

<civodul>dustyweb: heheh :-)
<civodul>in other news, LVM support pushed!
*civodul -> zZz
<lfam>Classic civodul. Add a major feature and immediately leave :)
***philippee is now known as philippee`
***philippe` is now known as philipper905
<nckx>Aww, this lil' thing?
*nckx has pushed nothing significant, also goes to bed.
<philipper905>Why is there an emacs-org-mode package? Org-mode comes included in emacs
<lfam>"guix show: error: emacs-org-mode: package not found"
<lfam>I'm curious, where do you see this package, philipper905?
<philipper905>oh my bad it's under emacs-org I believe
<lfam>Hm, well not sure about the answer to your original question! Somebody will have an answer though
<philipper905>To answer my own question, it seems like it downloads from ELPA, which means ORG version 9.4 instead of 9.3. I don't know if anything else is different though
<guixy>hi guix
<guixy>I'm not the only one having trouble with ssh-copy-id am I?
<guixy>" If the redirection operator is '<<-', then all leading tab characters
<guixy>are stripped from input lines and the line containing DELIMITER. This
<guixy>allows here-documents within shell scripts to be indented in a natural
<guixy>There seems to be something wrong with ssh-copy-id. Am I the only one having trouble with it?
<guixy>It fails to see the delimiter EOF.
<guixy>Can't replicate it.
<lfam>Thanks for the dav1d fix, mbakke
<mbakke>lfam: np, I just noticed I was building a "git-checkout.drv" :-)
<tsmish[m]>guixy: can't replicate on debian with version 8.3p1 and 8.4p1 from official repo. Can you compare checksums with source tarball?
<mbakke>guixy: it's a known problem in the current version of OpenSSH:
<mbakke>if you want a challenge, you can try applying the patch to the Guix package
<mbakke>I have to go now, but otherwise I can try fixing it tomorrow
<lfam>guixy: Would you like to try fixing it? If not, I can make a patch for you to test
<guixy>I wonder if we can make this change in a simple g-exp
<lfam>You'd basically want to do something like this: <>
<guixy>Just add the patch to the repo? That's simple.
<guixy>I'm testing the patched openssh source right now.
<lfam>Yes, add the patch to the repo, use it in the source field of our openssh package, and register the new patch file in 'gnu/'
<xelxebar>Oh, sweet! LVM support now!
<guixy>no luck with `guix install --with-source=...`
<lfam>guixy: You'll need to follow the steps described in the Contributing section of the manual. Specifically the sections Building from Git and Running Guix Before It Is Installed
<guixy>I know. Since the patch is so simple I figured I would try it with the result of `guix build --source openssh`
<guixy>That way I would know it's what we need.
<lfam>But it didn't work?
<guixy>not with `guix install --with-source=...`
<guixy>That could be because system packages like openssh are in PATH before installed packages
<guixy>It works with `guix environment --ad-hoc --with-source=...`
<lfam>Substitution feels a lot faster in 1.2.0
<guixy>Building guix taks so long...
<lfam>Yes, it takes a while
<guixy>Whenever I send a patch there is always something the maintainers want me to fix. Unless I get a response tonight I won't be able to fix anything because I'm only off for tonight and tomorrow, and I have other things to do tomorrow.
<guixy>Spelling, capitalization, indentation, comments... something alwyays slips through `guix lint` and indent-code.el
<lfam>I am waiting for your patch :)
<lfam>I can also fix little things like that for you
<guixy>If I downloaded the email patch from upstream gitlab, should it apply correctly?
<guixy>Numbers might be off.
<lfam>I don't know
<lfam>You should download it from the openssh-portable repo
<lfam>That is what our package is based on
<lfam>I know that ssh-copy-id comes from that other repo, but that's not relevant for us
<guixy>Having trouble grabbing the raw patch from github...
<lfam>You can append '.patch' to the commit URL, like this: <>
<vits-test>sneek later tell luis-felipe: i imagined a shirt with the l. notice header before actual lyrics. better safe than sorry :|
<sneek>Got it.
<guixy>patch sent
*vits-test 's machine cannot wait to see it in guix proper, and segfaults beforehand!
<guixy>When I installed guix on a few old (but surprisingly free) laptops earlier, I noticed that if I take out the external drive with the installer on it when it says I can, it kernel panics.
<guixy>I've had to carefully time it so I remove the device when it is powered down for a restart.
<guixy>Anyone have a problem like that?
<guixy>It was a local build of the installer iso and I've run `guix pull` since making it.
<vits-test>guixy: did u `umount` it before removing?
<vits-test>not that kernel should segfault, ofc
<guixy>graphical installer...
<vits-test>Alt + F2...
<vits-test>:P ?
<guixy>There's a screen that says something like "you may remove the install media and push the button to restart."
<guixy>Also, it's probably a bug that when guix starts it says success is an error.
<vits-test>it's tradition, not a bug.
<vits-test>people edit /etc/issue just to get that message.
<vits-test>on foreign systems, that is.
<guixy>How did that tradition get started?
<vits-test>i heard it started either when nckx got $partner, or rekado <strike>was put in submission</strike> married. but idk.
<vits-test>something like that for sure.
*lfam tests ssh-copy-id patch
<guixy>So it's a guix tradition?
<vits-test>ofc, no shit.
<WarmasterJindrit>Greetings, Guixers
<ryanprior>Language please vits-test =D
<ryanprior>Hi Warmaster Jindritch 👋️
<WarmasterJindrit>I recently installed guix and I am liking it so far, I just wanted to say good job and keep up the good work
<guixy>vits-test, It confused a friend of mine who tried guix and got a blank screen soon after seeing it. The system booted but the dm didn't work.
<guixy>WarmasterJindrit, Glad to hear you like it. May I ask what you use it for? Personal, professional, family, ...
<guixy>I've been using guix as my daily driver for nearly 2 years.
<WarmasterJindrit>Well, I primary use it as part of another project, a friend had me try Telega and OCaml
<guixy>Is there a general guix wishlist published somewhere?
<vits-test>yes, on libreplanet
<vits-test>guixy: idk the fine matters of dm and blank screen, but sway wm may work. it's only working thing on my box to start gui.
<guixy>I see...
<ryanprior>Warmaster Jindritch: thank you for sharing, that's great to hear!
<guixy>Too many things need node. December 20-30 I might work on something to import npm packages and their dependencies.
<ryanprior>guixy: check out pnpm if you want something pragmatic.
<ryanprior>It uses a system-wide immutable store and creates a node_modules folder full of symlinks
<guixy>Is node-build-system insufficient?
<ryanprior>I think it's sufficient for some things. But many common packages have hundreds of transitive dependencies, so it's going to take you a lot of reviewing work even with a good importer to get them into Guix.
<lfam>Patch pushed, guixy
<lfam>Thanks a lot!
<guixy>vm-manager doesn't let me eject a cd while an os is running...
<ryanprior>For example, jest lists 3 dependencies, so a lot of people are like "oh that's not too bad." But when you resolve indirect dependencies? 476 deps in total.
<ryanprior>So node-build-system is gonna need packages for all of those before you can build anything that requires jest, which is a ton of things, and that's just for jest alone.
<guixy>Yet another reason to dislike node :(
<ryanprior>TypeScript has 928 transitive deps (just checked)
<ryanprior>That's why I'm intensely interested in creating a rome-build-system for Guix, because Rome will be able to replace Jest and TypeScript and other things, and it has no deps other than Node itself.
<ryanprior>The rome-build-system will hopefully let us chop off thousands of transitive deps off and let us reasonably package a bunch of great free JavaScript and TypeScript software =D
<lfam>That would be awesome, ryanprior
<guixy>"Rome has zero dependencies"... except Typescript and its 928 dependencies.
<guixy>Unless Rome will package Rome?
<ryanprior>Yes Rome is self-hosting, and it can be compiled by esbuild (which is already in Guix)
<ryanprior>So the bootstrap story is solid
<jackhill>ryanprior: woah, that's awesome. I'm also encouraged that these tools seem to be written by JavaScript practitioners and not just distro packagers. It's like we were right in wanted to have reasonable ways to bootstrap and build from source :)
<ryanprior>yes 100%!
***sneek_ is now known as sneek
<txgvnn>Hello Guix
<txgvnn>Could you guys help me? My icecat browser render fonts incorrect
<txgvnn>here is log
<txgvnn>I installed font-dejavu
<txgvnn>failed to create cairo scaled font, expect ugly output. the offending font is 'DejaVu Sans 9.9990234375'
<txgvnn>font_face status is: file not found
<PotentialUser-73>Hello everyone! Trying to install on Guix on a new laptop (dos). Laptop is Lenovo on AMD Ryzen 7. After booting gives me the next message AMD-Vi: Unable to read/write to IOMMU perf coun
<ZombieChicken>Hello. Uh, is there not an arm64 install image as of yet?
<ZombieChicken>I see a binary, but not an installer for it...
<PotentialUser-73>Sorry, I'm complete newbie. What does it mean? It's impossible or I should use another image?
<leoprikler>PotentialUser-73: It means, that the IOMMU performance counter can't be read. I don't think that's a fatal error, but it might be a problem for energy saving. Does your boot succeed afterwards or does it get stuck?
<leoprikler>W.r.t. Guix compatibility you should try to search your notebook on, that should tell you whether you can run it with free software only or binary blobs are needed.
<PotentialUser-73>It stops at scheme@(guile-user)>
<PotentialUser-73>At least stops hear for me, as don't know what to do next
<tsmish[m]>PotentialUser-73: Are there other messages, I think initrd shows some errors before dropping to repl. Various sites recommend adding "iommu=soft" to kernel cmdline. Can you reboot to grub, press "e", find a line that starts with linux, go to the end of it, type "iommu=soft" and pressing ctrl-x to check if that works?
<PotentialUser-73>Yes, it gives another messages. I'll try
<PotentialUser-73>It worked. But there are others messages:
<PotentialUser-73>At least thanx for that tsmish
<leoprikler>Is there a message that indicates why the REPL is started?
<Zambonifofex>Hello! I’m not sure anyone remembers me, but I’m back! I was hoping to be able to wonder whether that issue regarding the `netdde` package for Hurd has been resolved since last time, and if not, if I could do anything to help resolve it. Also, I wanted to mention that it seems the Hurd download is giving me a 502 HTTP response, in case anyone cares to do anything about it.
<rekado_>Zambonifofex: sorry, I didn’t manage to continue after the build of the upgraded netdde failed.
<Zambonifofex>I see. That’s fair enough, there is no need to apologize, I think! I appreciate that you cared to try.
<PotentialUser-73>GC Warning: pthread_getattr_np or pthrea_ahhc_getstack failed for main thread
<PotentialUser-73>GC Warning: Couldn't read /proc/stat
<PotentialUser-73>then "welcome, this is gnu early boot Guile use '--repl' for an initrd REPL" after that goes
<PotentialUser-73>leading kernel modules... usb 1-2: device descriptor read/64, error -71
<PotentialUser-73>and some others
<PotentialUser-73>at the end there is ice-9/boot-9.scm:1669:16: In procedure raise-exception: failde to resolve partition "(a lot of numbers)"
<rekado_>PotentialUser-73: the last one seems like the actual error
<rekado_>not being able to resolve the root partition means that Guix can’t possibly boot the system.
<rekado_>so the handover from initrd to Linux never happens
<PotentialUser-73>rekado_ okey, so there is no clear ways to solve the problem?
<rekado_>we would need to find out why the disk partition isn’t found
<rekado_>in the REPL you might be able to run a few Guile snippets to clear up the situation and see where exactly it’s failing
<civodul>Hello Guix!
<vits-test>txgvnn: `fc-cache -rv`, try start icecat again.
<PotentialUser-73>rekado_ ok, which commands could I try?
<vits-test>ZombieChicken: which board do u have? we have not all bootloaders packaged, most notoriously raspberry's one.
<ZombieChicken>vits-test: Pinebook Pro
<vits-test>ZombieChicken: should work. my rockpro do.
<tsmish[m]>PotentialUser-73: I assume you are running installation/live image. How did you write it to flash/disk? Do you have another linux system so you can run "blkid" from shell and check if uuid's are correct (there should be one matching the one at boot). Also you can probably do something with repl, but I don't know what exactly.
<vits-test>ZombieChicken: easiest way is to install some other distro (like armbian), then install guix on top of it.. or so. i used to install on microsd (as it has lower boot priority by default), then `init` guix to emmc..
<wehlutyk>hello guix!
<ZombieChicken>vits-test: So basically use a uSD to bootstrap, in a sense?
<txgvnn>vits-test: Do you know which package has fc-cache?
<txgvnn>I wish Guix has command as `apt-file`
<vits-test>ZombieChicken: yes; "almost" my config, rockpro:
<vits-test>txgvnn: +1
<vits-test>ZombieChicken: about pinebook, by jan (see posts):
<leoprikler>txgvnn: fontconfig
<mothacehe>cbaines: civodul: hey! still available for the CI point?
<cbaines>mothacehe, yeah, did we decide to use BigBlueButton, or not?
<mothacehe>yup think so!
<txgvnn>Thanks vits-test, leoprikler. It works!
<vits-test>txgvnn: try nomad or qutebrowser; both have nice keyboard driven interface..
<PotentialUser-73>tsmish[m] That's right, I'm using flash that I burned with Etcher (I also tried to do it with rufus, but the problems were the same). And I don't have another linux on this machine, it's completely new laptop with dos.
<civodul>mothacehe, cbaines: yes, i'm here!
<cbaines>mothacehe, civodul do either of you know how to create a room in BigBlueButton?
<civodul>ah, i was about to ask :-)
<txgvnn>vits-test: Oh, Idk these packages. nomad, I thought you talk about
<txgvnn>I will try it. Thanks
<vits-test>txgvnn: this: `guix show nomad`
<txgvnn>vits-test: +1
<vits-test>it's alpha, but mostly in guile.
<txgvnn>A emacs-like web browser
<txgvnn>I have a error `Webkit encountered an internal error`
<vits-test>txgvnn: on my aarch64 i use fairly old version. if u wish u can report it on #nomad-browser.
<vits-test>anyway, be that nomad or qutebrowser (not alpha), it's may be better than mouse centered ones like icecat.
<tsmish[m]>PotentialUser-73: I'm currently assuming that writing program borked uuid's when writing. I'm not familiar with Etcher (or rufus for that matter), but rufus site says it has something called "DD mode", did you try that? Also, were the numbers displayed "1970-01-01-19-49-46-83"?
<txgvnn>vits-test: I consider using icecat, chromium because I want to prevent `Emacs pinky` when live so long with emacs :(
<vits-test>txgvnn: then qutebrowser is vim-alike. ':', 'o',, 'RET'.
<vits-test>aside bindings, keys from a-z are "always there", no need to point.
<vits-test>it's super comfortable way to deal with web.
<vits-test>but anyway, i tried.
<PotentialUser-73>tsmish[m] The numbers are 31393730-3031-3031-3139-343934363833. Okey, I'll try DD mode in rufus
<PotentialUser-73>tsmish[m] Tried DD mode but the problem is same
***apteryx_ is now known as apteryx
<vits-test>PotentialUser-73: sorry, idk but maybe u'll try make some gui stick with trisquel or so, install it to laptop (if it'll boot), then ?
<tsmish[m]>PotentialUser-73: okay, numbers are actually pretty interesting, cause I found this bug which may be related to what you are experiencing. So, now I really don't know what to do, but as a last measure, can you edit kernel cmdline again and remove root=... option, initrd should have uuid built-in from what I see, so you'll probably be fine, can also try with
<tsmish[m]>PotentialUser-73: but well, going with another may be your best bet in the long run. If you can, you should probably mail with description of your bug, laptop model and maybe a photo of boot process.
*vits-test *clap-clap*
<PotentialUser-73>vits-test Thanx, I'll try
<PotentialUser-73>tsmish[m] Okey, I'll try. Thanx anyway
<txgvnn>Anyone use libvirt-service-type service. I can not access on non-root user
<txgvnn>I added user to libvirt group
<txgvnn>error: Failed to connect socket to '/var/run/libvirt/libvirt-sock': Permission denied
<leoprikler>Did you relog after group change?
<txgvnn>even restart system
<txgvnn>And I see /var/run/libvirt directory under root user and group
<txgvnn>is it correct?
<cbaines>Still a work in progress, but I'm playing with performing system tests
<mothacehe>cbaines: woooh! I've been dreaming to see something like that!
<civodul>cbaines: yay, neat!
<civodul>you're not using the jobs coming from (gnu ci), right?
*civodul feels a growing will to replace Cuirass job alists with manifests
<mothacehe>manifests + channels would be nice :)
<cbaines>civodul, no, I'm using the derivations listed on that page. I'm aware that they differ from what Cuirass is building
<cbaines>civodul, or rather yes, I'm not using the jobs from (gnu ci)
<cbaines>In other news, is currently processing new revisions more slowly than usual, I think because it's failing to build glib for armhf-linux for every revision it processes
<civodul>cbaines: are you using the system-tests.scm manifest?
<cbaines>See for example
<cbaines>I think someone else spotted that this glib issue is what's probably broken
<cbaines>civodul, no, the Guix Data Service calls (all-system-tests) and looks through the records
<cbaines>Searching through the Guix DAta Service logs, suggests this is the point that it broke 129b9b16d9b588316cc997cf8f4fefe30961a417
<cbaines>Which is these commits
<cbaines>Which does include one commit that mentions glib
<civodul>oh, that's not good
<civodul>i think "make assert-binaries-available" doesn't check for replacements on the grafts
<tsmish[m]>so, somewhat offtopic, does someone know how used ram can be 3gb higher than sum of resident memory of all processes on the system aside from kernel memory leak.
<cbaines>I think the only other thing I can figure out is that the derivation that's failing does involve the glib-appinfo-watch.patch
<civodul>cbaines: that patch triggered a rebuild of GLib, and it could be that the test fails non-deterministically regardless of the patch
<civodul>(or so i hope)
<civodul>on berlin there were 0 test failures but 3 "timeouts", which i think means tests were too slow or something
<civodul>i'll try again
<vits-test>tsmish[m]: maybe it's a buff/cache from `free`?
<tsmish[m]>vits-test: here's htop if you are interested. doesn't look like cache, I think used doesn't include it.
<rndd>hi everyone ! i dont have $HOME/.guix-profile/lib/locale but i have $HOME/.guix-profile/lib/locales . is it right?
*jonsger has sent in all required updates for guix-1.2.0.@opensuse. let see if they go through :)
<abcdw>Somebody review the patch please (:
<civodul>jonsger: yay!
<civodul>abcdw: hi! i'll take a look if nobody beats me at it (though i'm not a Wayland person)
<abcdw>civodul, Thanks a lot!)
*jonsger did some investigation on the BigBlueButton issues on GuixDay...
<efraim>I suppose for a service I should use with-output-to-file and read from /dev/random and not use dd if=/dev/random and system*
<nckx>Good morning, Guix.
<nckx>efraim: Yes.
<efraim>with any luck we'll have munge, (basic) slurm and (limited) lizardfs services soon-ish
<soheil>Hello friends, in connection with this issue ( I edited the config file like the image ( Where does the problem come from?
<jonsger>efraim: oh nice
<nckx>soheil: Did you add the (supplementary-groups '("libvirt")) part too?
<nckx>What does GNOME Boxes say now?
<vits-test>tsmish[m]: try `free -h` instead of htop? i sure it's buff/cache.
<tsmish[m]> total used free shared buff/cache available
<tsmish[m]>Mem: 7,6Gi 5,5Gi 611Mi 453Mi 1,5Gi 1,3Gi
<tsmish[m]>Swap: 0B 0B 0B
<nckx>soheil: You're closing your (list…) too early, after (set-xorg-configuration…(…(…keyboard-layout)))) <- one bracket too many. So you're trying to append that list to (service virtlog-service-type), which isn't a list, which won't work. You need to move that extra ) to behind your last service.
<soheil>Many thanks to nckx. The problem was solved.
<vits-test>tsmish[m]: not that i understood anything, but gnome-software claims 1.7 % of memory whole gnome-shell use 2.5 %.. rm gnome-software?
<vits-test>about memory usage, idk what to think. seem pretty normal, i'd my memory actively used back in fedora (with gnome, especially).
<tsmish[m]>vits-test: yeah, gnome takes like 500mb in memory. ps thing actually is a bit wrong, but you can "awk 'BEGIN {s=0} {s += $1} END {print s/1024/1024}'" it. You get around 1.64 gigs as sum of rss, free otoh reports 3.4 gigs as used + 3.7 gigs available which clocks around 8gb which I have in total. So somebody ate 2 gnomes and is hiding now.
<nckx>sneek: was it you? no more botsnack.
<leoprikler>sneek confessed their crime with a smile
<nckx>Guilty look.
<roptat>ok, i managed to build a first package with my soong-build-system :)
<roptat>although there are still some questions I need to answer, like how do I map blueprint module names to inputs
<nckx>roptat: What's soong? I'm enough of a nerd to know the reference but not enough to know the software.
<roptat>it's the build system used by android
<nckx>Oh cool, a custom build system.
<roptat>I'm trying to upgrade our adb/fastboot packages for now, and they are stuck at android 7 because now we need to use soong to build them
<leoprikler>I don't know the reference. Someone enlighten me please.
<nckx>leoprikler: Nunian Soong built androids in Star Trek.
<nckx>E.g. Data.
<nckx>And Evil Data.
<guixy>You mean Lore?
<guixy>That was an interesting story arc.
<guixy>Happy thanksgiving everyone
*nckx notes it's Noonien, apparently, although even ner^Wfans don't seem to agree.
<nckx>guixy: Same to you ☺
<leoprikler>so the "don't do evil" folks who gather all your data have an evil data build system.
<leoprikler>surely there's nothing wrong about that :)
<nckx>For once I was subtle.
<leoprikler>You were… they… not so much.
<guixy>I'm going to try guix deploy on two computers. Is there a way to specifiy the correct drive setup for both systems? One boots /dev/sda, the other boots something like /dev/mmcblck1p0
*civodul can't find a way to "grep -r" with ivy/counsel
<leoprikler>guixy: alists probably :)
*rekado_ published the blog post
<leoprikler>fill in one operating system template with common stuff, then (a (operating-system (inherit ...) ...)) and (b (operating-system (inherit ...) ...))
<civodul>rekado_: yay! \o/
<rekado_>will the website update automatically?
<civodul>yes, but let me kick the mcron job so we don't have to wait for too long
<rekado_>sweet, thanks!
<guixy>I'll let you know how it goes.
<civodul>(it still takes time tho, as i wrote on the list)
*nckx cries nepotism but can't wait to see pix.
<civodul>cbaines: guix-modular passes now, which confirm my theory of a non-deterministic failure
<guixy>machine-ssh-configuration is missing documentation for 'host-name', and the example says "localhost". It's a little confusing.
<civodul>👍 rekado_!
<civodul>you rock
<civodul>great to see a picture of the Stick
<civodul>two of them, even
<civodul>12-string uh
<guixy>inherit looks like useful syntax, but I don't see where it's defined, and it isn't formally documented in guile or guix info.
<rekado_>the 12-string instrument is the Grand Stick; the rosewood instrument has 10 strings, that’s the Alto Stick.
<guixy>I see it's provided by define-record-type*
<nckx>guixy: It's a feature of all (guix records).
<rekado_>they aren’t normally mounted on microphone stands, but I like them better this way
<nckx>‘info guix inherit’ briefly describes it but it's not ‘formal’.
<nckx>It is in the docstring.
<rekado_>more strings just mean more redundancy and thus more ways (including hopefully a comfortable one) to play the same harmonies or lines; you don’t play them all at once
<guixy>Is there an easy way to see docstrings?
<nckx>,describe <foo> at the guix/guile repl.
<guixy>without looking at the code
<nckx>Here ‘,describe define-record-type*’. Drawback is it works only after importing the module: ,use (guix records).
<rekado_>you can also use (procedure-documentation the-procedure)
<guixy>Yes... but that doesn't work on syntax.
<guixy>Thanks for the tip though
*nckx goes looking for 'phones to listen to the Ode again.
<civodul>rekado_: ok, got it
<civodul>the WP page on the Chapman stick could be expanded ;-)
<guixy>I want to specify host-key. Is that the entire content of /etc/ssh/ or just the base64 blob?
<nckx>Should be everything if it's like other users in Guix.
<nckx>If not, let's make it so.
*jonsger feels more like Franz Schubert when writing 20 lines of lilypond and feeling like a music hacker :P
<mroh>trying to add our wifi modules to the installation-os in gnu/system/install.scm: -> "invalid field specifier". %host-type isn't available there?
<efraim>Initial munge service
<nckx>mroh: No, it's bad syntax. You can't use WHEN here: you can't make the presence of the field conditional [like this], but you can change its value. Use something (k-l-m (if (blah blah) '("blah" "bloo") '())).
<mroh>ah, ty!
<civodul>efraim: oh munge, the thing that SLURM uses, right?
<cbaines>civodul, did you succeed in building glib on native armhf, or using QEMU?
<cbaines>I've tried a number of times to build it locally using QEMU, but those two tests take large amounts of memory and then timeout
<guixy>guix deploy isn't working.
<guixy>guix deploy: error: unauthorized public key
<Formbi>why is it that some packages don't get substituted when I have guix publish set up?
<guixy>I don't know
<cbaines>guixy, I haven't used guix deploy, but should the key be authorized?
<guixy>cbaines, I think so
<cbaines>Formbi, substitutes are unobvious, it could be that the key guix publish is using for signing is unauthorized, or maybe that a substitute isn't available yet
<jonsger>Formbi: or has to do with caching
<Formbi>it is authorized
<cbaines>Formbi, if you have an output like /gnu/store/XXX-foo you can check for the narinfo by requesting http://EXAMPLE/XXX.narinfo
<civodul>cbaines: hmm i didn't actually succeed, 2 "timeouts" this time
<jonsger>Formbi: but if I would start from scratch I would go guix build coordinator + nginx. I guess it's more comprehensible
<Formbi>how to request the narinfo?
<cbaines>civodul, I'm suspicious that -c1 might help, as the tests do run in parallel, and maybe it's more likely to pass on actual hardware, but these tests look like they might be worth disabling (maybe the "slow" group could be skipped)
<civodul>hmm lemme see
<civodul>trying with -c1 on guix-x15
<Formbi>can I somehow force the caching of something?
<civodul>(apparently the one it was offloaded to before)
<cbaines>Formbi, did you manage to check if there's a narinfo available?
<civodul>Formbi: how did you set up "guix publish"? using --cache?
<Formbi>cbaines: how to do it?
<cbaines>Formbi, I believe I described that previously, if you have an output like /gnu/store/XXX-foo you can check for the narinfo by requesting http://EXAMPLE/XXX.narinfo
<Formbi>civodul: no
<Formbi>But how to request it? In a browser?
<cbaines>Formbi, whatever way you like, a web browser or wget are fine
<cbaines>Formbi, it's a text file, but what's more interesting is whether it's available, or not
<Formbi>resource not found
<jonsger>Formbi: I think guix build coordinator has post-build hooks to do this
<civodul>Formbi: to test, i usually do "wget -O - http://localhost/xxxxx.narinfo", replacing "xxxx" with the hash part of the store item that exists in the store
<civodul>does that make sense?
<efraim>civodul: yeah. still have to test it against munge on debian but it should work. I have to figure out how to do slurm, i want to reuse fields when it makes sense
<Formbi>civodul: ahh, like this
<Formbi>so it does exist
<samplet>Hi Guix!
<guixy>How can I make sure the key is authorized?
<guixy>The public key on the host (the one I'm deploying to) is
<civodul>howdy samplet!
<guixy>Am I supposed to copy it from ~/.ssh/known_hosts ?
<civodul>efraim: neat! so soon we can have clusters running Guix System, right? :-)
<guixy>server at '' returned host key 'AAAAC3NzaC1lZDI1NTE5AAAAIJfu5OWDYAn5bWgQe++cwK0yqTzB+cqe5OErM1yknGFA' of type 'ed25519' instead of 'AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBM/ulFt3e/wBvbhbn2wH77wFFais8LW0NLp3WQj/nFqC4aWhSBltpUPQbGDwfUkIYYjC9/lTCT0EqcqWbzHKOwY=' of type '#f'
<efraim>civodul: that's the plan :) new cluster with pjotr, intial setup with Debian, plan is to switch it over as its ready
<Formbi>the substitute shenanigans seem to happen more often (but not always) when I'm trying to download something from a computer with Arch to a computer with Guix
<samplet>I’ve come with a bit of Cuirass frustration. I have been running an instance for Disarchive. It builds the first round of metadata for the first evaluation, but it never does anything after that.
<samplet>I’m wondering if it always uses the same checkout of Guix or something.
<samplet>(It’s built 108 evaluations, but maybe it sets the load path wrong.)
<civodul>efraim: really cool; be sure to write about it on when you're ready :-)
<Formbi>I changed the publish's launch options in the Arch computer to those from Guix and it still doesn't work
<guixy>Still no good...
<guixy>Guix deploy unauthorized public key
<cbaines>guixy, I'd maybe check the logs on the remote machine, that should show something about why access is denied
<cbaines>Formbi, what bit doesn't work, if there's a narinfo available, I'm guessing you're maybe not seeing the store item being fetched from guix-publish?
<guixy>cbaines, where does the default guix install keep its openssh logs?
<cbaines>guixy, probably syslog, so I'd check /var/log/messages or /var/log/debug
<guixy>Looks like it isn't on the host.
<cbaines>looks like what isn't on the host?
<guixy>the problem.
<guixy>Nov 26 11:04:24 localhost sshd[22247]: Accepted publickey for admin from port 44552 ssh
<guixy>2: RSA SHA256:05orBrSJqDkovWs25+bkHxIt98q38AQWGfhSHwAfzpY
<guixy>Things seem to be working fine.
<guixy>ssh works fine from a shell.
<cbaines>Do you see that "Accepted" message appear when you run guix deploy?
<guixy>yes. Multiple times
<guixy>ssh to the target machine
<guixy>watch tail /var/log/messages
<guixy>Don't know what else to do.
<vits-test>guixy: is admin has NOPASSWD?
<guixy>$sudo cat /etc/sudoers
<guixy>root ALL=(ALL) ALL
<guixy>%wheel ALL=(ALL) ALL
<guixy>admin ALL = NOPASSWD: ALL
<guixy>that's on the target machine
<vits-test>guixy: and u'd ran `guix archive --authorize < key` on target?
<guixy>I forgot about that
<guixy>I did that now.
<guixy>still didn't work...
<guixy>sudo cat /etc/guix/ | ssh admin@ sudo guix archive --authorize
<guixy>From the deploying computer
<Formbi>cbaines: it's not fetched
<Formbi>when i guix build something, it starts building from source and not substituting
<cbaines>Formbi, from the command line output, does it look to be checking your guix publish for a substitute?
<jonsger>Formbi: sudo rm -rf /var/guix/substitute/cache/ can do wonders :)
<Formbi>jonsger: didn't work :(
<jonsger>Formbi: do you have a cache-bypass-threshold set?
<Formbi>how to check it?
<a_v_p>Hello everybody. Does Guix have NextCloud as a package? I've searched through the package list on the official site, but with no success.
<a_v_p>Although it seems that someone at least tries to use NextCloud in Guix:
<jonsger>a_v_p: no we don't have. Although I use it on my server (still version 17 or 18). You can inspire you by
<a_v_p>That's nice, thanks jonsger
<rekado_>huh, the blog post is on the HN front page and Paul Davis (primary author of Ardour) has posted a comment criticizing my endorsement of the Calf plugins.
<guixy>$ cat /etc/guix/ | ssh admin@ sudo guix archive --authorize --; ssh admin@ sudo cat /etc/guix/acl
<guixy>The key that it says is unauthorized is listed in the target /etc/guix/acl
<philippe`>Hi guix
<philippe`>I've been working through the CONTRIBUTING section of the docs and when i run `make authenticate` I get the error `guix: command not found`
<philippe`>Am I not supposed to run it from within the guix environment?
<guixy>Has anyone here been able to successfully run guix deploy?
<guixy>guix system list-generations lists them, so I guess they have been working after all?
<nckx>mroh: Do you use the aircrack rtl driver?
<mroh>nckx: No, I don't have this hardware.
<guixy>$ guix deploy --dry-run deploy.scm
<guixy>guix deploy: error: dry-run: unrecognized option
<guixy>but dry-run is an option listed in guix deploy --help
<guixy>bug report pending...
<guixy>bug report sent
<Rovanion>So I want to submit a stupid little patch, is the mailing list the way in?
<philippe`>If build.xml is in a subdirectory how to I tell ant-build-system to look for it there?
<Rovanion>Gotta get mu4e working again I guess :P
<civodul>philippe`: hi! re your previous question about "make authenticate", the assumption is that the "guix" command is available from your system
<civodul>probably you should run it in a enviornment that's no --pure or --container so that "guix" is visible
<civodul>regarding Ant and build.xml, i'm not sure
<civodul>perhaps roptat can answer
<philippe`>civodul: got it
<philippe`>civodul: looking at the other packages I might be able to do something with modify-phases
<mroh>philippe`: try something like `(arguments '(#:make-flags (list "-f" "dir/build.xml")))`
<mroh>ups, stupid matrix client, sorry ;)
<roptat>does sort work for a partial order? like if two elements are not comparable?
<philippe`>mroh: progress!
<mroh>yay ;)
<civodul>roptat: the 2nd argument to 'sort' does not have to define a total order in practice
<roptat>so, what should I answer when two elements are not comparable?
<civodul>#t or #f? :-)
***lle-bout_ is now known as lle-bout
<civodul>i think you've gotta try
<civodul>i don't know what guarantees we have though
<civodul>i just know that "in practice it works", which is perhaps mildly reassuring
<civodul>sneek: later tell vagrantc efraim pushed dpkg! \o/
<guixy>hmmm... guix deploy did not properly set up changes in /etc/sudoers
<civodul>guixy: oh really? normally that happens as part of "activation", which is one of the last things it does
<roptat>almost there... I need to please llvm-ar, and I'll be able to install a library with my soong-build-system :)
<civodul>you mean song-build-system, does that have to do with rekado_'s post?...
<roptat>no, I mean soong-build-system, soong is a tool used to build android
<mbakke>hmm, gpg-agent suddenly stopped working for me, presumably after the secret key timed out, and does not ask for password again (instead just hangs)
<mbakke>probably a regression since 2.2.25 as I've never had that happen before
<roptat>it worked!
<guixy>From the manual " ‘git-http-service-type’; can currently be used to configure Nginx trough ‘git-http-nginx-location-configuration’.
<guixy> Shouldn't that be "through"? I'm not familiar enough with Nginx, and I wouldn't be surprised if trough was their name for a some local service...
<nckx>No, it's a typo.
<nckx>guixy: I'll fix it unless you want to submit a patch.
<guixy>Go ahead. I'll report if I find any more.
<guixy>Not all cgit-configuration fields have documented defaults. I would expect at least the "package" field to default to the cgit package.
<guixy>Gitolite is configured through a special admin repository which you can clone, for example, if you setup Gitolite on ‘’, you would run the following command to clone the admin repository.
<guixy>That looks like a run-on sentence to me.
<guixy>Gitolite is configured through a special admin repository which you can clone. For example, if you setup Gitolite on ‘’, you would run the following command to clone the admin repository.
<nckx>Doesn't really address the redundant main flaw of redundancy that needs to be addressed.
<ryanprior>Anybody know what guix package has "hostname" in it?
<ryanprior>Thank you!
<OriansJ>guix version b3ece8db770340bb8b94563548d57f63f5d2dec2 fails to build qemu at /gnu/store/ryad0k15xl0i33fdsbq7bidr45b83wna-mpg123-1.26.3.drv with the error message: libtool: error: '/gnu/store/r5qflfi674s336pbaqkphiadqyj2psq0-pulseaudio-13.0/lib/pulseaudio/' is not a valid libtool archive
<civodul>mbakke: never experienced that and i haven't upgraded yet so i'd say it's a regression
*civodul -> zZz
<cbaines>OriansJ, what system is that with? With x86_64-linux I get /gnu/store/12cavia06d2mjm6886zmpxzgfwdnqrpf-mpg123-1.26.3.drv as the derivation
<nckx>OriansJ: Do the contents look sane?
<nckx>Mmm, cavia.
<guixy>Gitolite is configured through a special admin repository which you can clone. For example, if you setup Gitolite on ‘’, you would run the following command to clone the admin repository.