IRC channel logs


back to list of logs

<brendyn>the deblob check script certainly takes a long time, just sayin'
<brendyn>weird. when guix tries to download a url, i get 404. when i download it manually with wget it works
<brendyn>guix refresh: error: HTTP download failed: 404 ("Not Found")
<brendyn>for example guix refresh honsole
<brendyn>anyone else get this weirdness when exiting a guix shell? : guix shell: error: /gnu/store/d99ykvj3axzzidygsmdmzxah4lvxd6hw-bash-5.1.8/bin/bash: command not found
<nckx>What's even weirder is that ‘guix download ls-lR.bz2’ also works.
<KarlJoad`>Is there a way to send a channel specificationin a guix deploy definition?
<KarlJoad`>I have a channel defined on my local machine, but want the remote server to also have that channel available to it. Does the normal operating-system record allow such a thing to be shared?
<PotentialUser-78>Any idea why my fresh install gives a grub prompt before the normal grub screen?
<PotentialUser-78>Specifically, it says "GRUB C" where "C" is actually the fancy (spanish?) C with a squigly that makes a "k" sound.
<PotentialUser-78>I can boot normally if I input a key, but the point is that it should boot without interaction
<PotentialUser-78>I suspect it may be because I adapted a config that used full-disk encryption, so it "thinks" it should be prompting me for a password or something
<luchadoritos>Hello all! I've been having a rough time getting screen sharing on Wayland to work. I have a suspicion that it's because pipewire is using version 0.2.7 (calling $pipewire --version) but nothing in my profile uses that version, I do not know why it's even installed.
<luchadoritos>In my manifest I specifically install pipewore-0.3 and using Guix graph (reverse) I see that Weston and Telegram desktop use the old version of pipewire. Neither of those are in my manifest.
<PotentialUser-78>luchadoritos: try using which -a to see where those are comming from.  e.g. system install, guix-home, guix-profile, etc.
<PotentialUser-78>if you use guix package -m, then they shouldn't be comming from your guix-profile
<luchadoritos>Thank you for responding! Running which -a pipewire gives me $USER/.guix-home/profile/bin/pipewire.
<luchadoritos>This gives me some insight I'll see my home declaration again
<brendyn>If weston is updated to version 10 then it can use pipewire 0.3
<luchadoritos>I use Sway for my config looking at sway package declaration in wm.scm and in Guix graph I do not find Weston.
<luchadoritos>Trying to use guix gc -D says that pipewire-0.2.7 is alive. I hope it doesn't try to eat me lol
<PotentialUser-78>alright, I'm back and still get a grub prompt before the grub screen, preventing remote reboot.  I reconfigured ensuring that mapped-devices is an empty list.
<PotentialUser-78>luchadoritos: have you tried home reconfigure setting packages to an empty list?  I suggest using manifests XOR guix home for package installs, not both.
<luchadoritos>PotentialUser-78: I do not use Guix home packages because when you delete a package from guix home I noticed it doesn't actually remove it from the current profile. Thank you for your insight
<luchadoritos>I do not have it explicitly set as empty list though. I can do that. I thought default was empty list.
***Lightsword_ is now known as Lightsword
<luchadoritos>I think I got the right version. The shepherd service I was using was installing a different version of pipewire. I now have the right version but I still don't have screen sharing. I'm getting closer though, less errors in my TTY! Thank you everyone.
***dgcampea-2 is now known as dgcampea
***iyzsong-www is now known as iyzsong-w
<KarlJoad`>Is it possible to provide a channel specification inside of a file that provides an operating-system record for building, so the generated system uses that channel?
<KarlJoad`>I want to embed a channel definition into the machine's configuration for Cuirass. Then use guix deploy to deploy the machine definition to the machine, with the channel already specified for Cuirass.
<jpoiret>luchadoritos: are you using a wlroots-based compositor?
<jpoiret>i'm not sure screen sharing works in weston
<rekado>hi there
<sneek>Welcome back rekado, you have 3 messages!
<sneek>rekado, ngz says: why doesn't texlive importer use catalogue-version field, when available, to set the version of a package?
<sneek>rekado, apteryx says: haha! to reproduce our GRUB problem without any downtime on Berlin: sudo qemu-system-x86_64 -hdd /dev/sda -snapshot -nographic -m 2G
<sneek>rekado, ngz says: If you look at 280097b71f01608795fbf38ac6d87fd61fe16798, it seems there's a bug in `simple-texlive-package', as non-generated files should be handled in locations argument, but aren't.
<rekado>I’m using “guix environment --pure guix” with a fresh Guix on a feature branch, ./bootstrap && ./configure … && make clean-go && make -j
<luchadoritos>jpoiret: Yes, I was using Sway. I really don't know how Weston and Telegram got installed after using the pipe wire service from:
<rekado>but then when I use “./pre-inst-env guix build foo” it complains
<rekado>;;; WARNING: loading compiled file /home/rekado/dev/gx/branches/wip-r/guix/ui.go failed:
<rekado>;;; In procedure load-thunk-from-memory: incompatible bytecode version
<rekado>for every file
***user___ is now known as wonko-the-sane
<rekado>the .go files are for Guile 3.0.8, as far as I can tell.
***alMalsamo is now known as lumberjack13
<devmsv>Hi, does anyone know how to install google-compute-sdk? Can I just run the installer script? I guess it will not work on GuixSD
<luchadoritos>devmsv: In general, if you don't find a package definition most people make package definitions or use the Nix package manager ( )
<devmsv>Didn't now nix package manager was "official" to GuixSD
<efraim>the name 'GuixSD' is deprecated, it's been 'Guix System' for a couple of years now
<rekado>devmsv: there’s nothing official about nix on Guix System.
<devmsv>rekado: well the service is supported and in the official guix manual
<devmsv>efraim: sorry didn't now
<efraim>devmsv: no worries
***taiju_ is now known as taiju
<phf-1>Hello Guix! I run Guix on a foreign distro (Debian11) and `which ls' returns `/home/phf/.guix-profile/bin/ls' even if coreutils is not in the profile. I expected `/usr/bin/ls', why is that ?
<rekado>phf-1: perhaps you have a package that propagates coreutils? You can inspect the “manifest” file, which shows you what ends up in the profile and why.
<phf-1>rekado, ok thank you. I search in the doc how to << inspect the “manifest” file >>.
<rekado>phf-1: visit the ~/.guix-profile/manifest with an editor
<rekado>then search for coreutils
<phf-1>rekado, Ho... I see! Much better! I did not find this mentioned in the doc searching for "manifest" in guix.html
<phf-1>anyway, thank you.
<civodul>mothacehe, apteryx: hi! i've just reenabled armhf-linux for "master" on ci.guix
<civodul>turns out the Guile test failure was non-deterministic...
<mothacehe>hey civodul, nice only for the master specification?
<civodul>mothacehe: yes, in a effort to stabilize things towards the release
<civodul>hey long time no see, BTW :-)
<civodul>how's everything?
<mothacehe>terribly busy at work sadly, otherwise fine :)
<mothacehe>maybe we should also activate armhf for the "guix" specification?
<civodul>mothacehe: arf, i got the impression you were doing embedded stuff lately :-)
<civodul>i've now enabled armhf-linux for the "guix" spec as well
<attila_lendvai>hi civodul, let me know if there's anything i can do to help/ease the merging of the shepherd patches (#:rlimits and the shepherd-for-guix patch)
<civodul>attila_lendvai: hi! i'm sorry, i haven't taken the time to review these yet after the initial discussion
<civodul>i'll try to do that soon, at last
<attila_lendvai>my server (an older laptop) is running with them, and i haven't noticed anything unexpected
<KarlJoad`>Is it possible to provide a channel specification along with an operating-system record for building, so the generated system has that channel available? I want to add a channel definition into the machine's configuration for Cuirass and then guix deploy the definition to the machine, with the channel already specified for Cuirass.
<mothacehe>KarlJoad`: i don't think it is possible to override the default set of channels via the operating-system record yet
<mothacehe>i get that it might me handy to have by default guix pull operate on a different %default-channels list, but i don't get the link with Cuirass here
<KarlJoad`>Darn... I have a personal channel I want to build with Cuirass. I am specifying the Cuirass specification in the same file as I am writing the operating-system record.
<KarlJoad`>Something like `(define %cuirass-specification #~(list ...)) (cuirass-configuration (specifications %cuirass-specification))`
<mothacehe>yes defining a channel list is already possible at Cuirass level:, but i think we might need to have it globally at the operating-system level. Something to discuss. Looks like the unattended-upgrade-service-type already has a channels argument.
<KarlJoad`>From your article, there is no way to configure the Cuirass specification channel list from a file. You must manually do it?
<gnucode>morning guix!
<attila_lendvai>mothacehe, thanks for writing that up! i didn't understand this sidenote, though: "and even simplify it by running Cuirass directly on your main machine". isn't it run on your main desktop?
<KarlJoad`>Or rather, we cannot use guix deploy for this, but we can specify it in a file, but the file must be built on the CI machine, which must also have the channel manually added to the CI machine's channel list?
<mothacehe>KarlJoad`: specifications needs to be a gexp for now, you would like to be able to point to a local file instead?
<mothacehe>attila_lendvai: yw, i meant have cuirass directly running on the laptop instead of having cuirass running on a server and publishing substitutes with "guix publish"
<attila_lendvai>mothacehe, oh, ok. so cuirass is the substitute server, and guix publish is the operation to upload a built artifact to it, right? and it can be done either by the guix build daemon, or by invoking `guix publish`?
<KarlJoad`>mothacehe: Not necessarily a local file. I want to be able to specify channels to build that may not already be added on the remote machine. I want to add the channels to the machine's definition, then be able to `guix deploy` it over to the remote machine. The Cuirass specification can then use the channel provided in the machine's definition.
<attila_lendvai>mothacehe, i also welcome if you extend the post, and point me to it to re-read it again
<mothacehe>KarlJoad`: as i said you cannot for now override the default set of channels defined in (guix channels). You can however specify a set of channels you want Cuirass to build and reconfigure/deploy your machine to perform that task (see the blog post and our build farm configuration for examples:
<mothacehe>attila_lendvai: i do not plan to extend the post but if you have specific issues i'd be glad to help
<attila_lendvai>mothacehe, thanks, noted (including the links)! maybe i'll set this up in the future.
<apteryx>civodul: re armhf-linux great, thank you!
<iskam>Guix website is blocking me
<civodul>hi iskam! there are known issues accessing from Russia lately, perhaps that's related?
<iskam>Probably, I'm russian
<civodul>rekado, apteryx: can i restart hydra-guix-101? its childhurd service fails to start and i'm clueless
<iskam> isn't in the repository of sites blocked in russia though
<iskam>At least as far as I could tell...
<civodul>iskam: see for a possible workaround
<civodul>we don't know why is inaccessible from Russia and haven't yet been able to set up a mirror of the web site
<iskam>In the forum it says theres a lot of port scanners here
<civodul>right, so that could be a reason, but we can't tell for sure
<rekado>civodul: you can restart it.
<civodul>rekado: done, thanks!
<civodul>that may sound like a desperate move... and it kind of is a desperate move
<civodul>in other news, berlin still has a short and now deprecated .onion address
<civodul>that name was never really publicized i think, so we should be able to generate a new one
<phf-1>Dear Guix, how to get auto-completion for `git' for example?
<apteryx>hmm, we're getting quite a few 'guix pull' related bug (networking-related?) reports, e.g. #54278
<civodul>yeah, i noticed and this is pretty annoying
<civodul>that's a poor UX
<apteryx>it's nice to know that it fails often, but the output doesn't provide much of a clue as to what went wrong
<civodul>yeah, i looked at the backtrace that some reported and it's unhelpful
<civodul>looks as though the daemon had hung up or something
<apteryx>I'm guessing most issues happen when pulling from 1.3.0 though
<apteryx>at least in the bug I pointed above the host version is reported as 1.3.0
<wonko-the-sane>civodul: v3 .onions yes please!
<sneek>wonko-the-sane, you have 1 message!
<sneek>wonko-the-sane, nckx says: Maybe, if you mean correctly! Like with ‘guix build -f’, a file evaluates to whichever came last (if you write the world's coolest operating-system definition followed by ‘2’, the file will evaluate to 2, and you'll get an error).
<maximed>if an onion is set up for the website, could the Onion-Location header be set (
<maximed>If I guess correctly, that makes Tor Browser automatically redirect to the onion, making the onion address a little more discoverable
<maximed>Looks like <> is out-of-date, it mentions Freenode instead of Libera Chat
<apteryx>civodul: as long as you don't reconfigure/reboot berlin itself, it should be OK I think
<civodul>maximed: yes, we should do that, but the nginx config is kinda hairy
<civodul>would you like to give it a try? :-)
<gnucode>I wonder how hard it would be to package veloren for guix. It's a rust based (so probably hard) minecraft like exploratory world/game.
<gnucode>I personally am not a gamer, but I think FOSS gaming is pretty cool and that seems like a good example of one.
<lfam>OpenSSL security update
<lfam>Building now
<Guest2948>Hi, in which languages ​​can we use the gnu guix system?
<lfam>OpenSSL update patches sent
<ryanprior[m]>The list of languages is here:
<jonsger>is spam in HTTP requests on web servers a new thing?
<lfam>What kind of spam?
<nckx>I've seen a few skids claim ‘credit’ for trying to GET /wp-login.php. But consumer-oriented spam? Never seen that.
<lfam>I think it's more informational messages from scanning bots
<nckx>maximed: Pay no attention to the date on the Freenode → Libera commit I just pushed.
<nckx>Are the videos auto-regenerated or must I do another thing?
<lfam>ryanprior[m]: From my own server, one of many such messages: "NetSystemsResearch studies the availability of various services across the internet. Our website is
<lfam>So many companies build their own index of the web, and it's good if they leave a note in our logs to explain who they are
<ryanprior[m]>Is it sent over and over (in a "spammy" way) or just occasionally? Sounds like it could be courteous, in case sysadmins wonder "who's doing this scanning"
<nckx>It's sent with every request. Changing the UA for each request would be… worse.
<lfam>It only seems to occur once per scan, but they may scan frequently
<lfam>Anyways, it's hardly spam IMO, since I don't just read the logs for fun
<jonsger>lfam: rather build logs :)
<nckx>It all depends an what jonsger means by spam. The above is common, but it's arguably ‘being nice’ (even if the message is sometimes misleading or the company is hardly nice), not spam. They aren't trying to get you to try their cool service *by* sending you HTTP requests.
*nckx also curious for an example!
<lfam>The vast majority of "non-human" traffic is things like "GET bitcoin", anyways
<nckx>Worth a try.
<nckx>200 GOT bitcoin
<nckx>uh oh.
<jonsger>nckx: yeah, thats what I called "spam"...
<nckx>Wait till you see what spam browser makers are sending you, it dwarfs this stuff.
<nckx>I will *not* try Mozilla 5.0, thank you very much.
<jonsger>but its still kind of an advertisement, as I have heard now of some Palo Alto subsidiary...
<nckx>And plenty of mirror owners and forges have now heard of "GNU Guile", which Guix sends out with every source request. If everything is spam then nothing is spam.
<nckx>I appreciate it, and you can set nginx not to log whatever fields you find annoying if you want.
<nckx>The nice ones include an ‘opt out’ link which I wouldn't touch with a longish pole. Has anyone here ever tried one of those?
<jonsger>topic change: how to I verify if a python package really needs an input as propaged-input? Can I trust `guix import pypi`
***jackhill_ is now known as jackhill
<ryanprior[m]>My technique is: try moving everything to "inputs" and see if it still works. Putting everything in propagated-inputs only makes sense for libraries IMO
<jonsger>ah okay, its a library so I guess its okay :)
<Noisytoot>Why does the cdrtools package install a man page for printf(3)?
<nckx>Because it implements one.
<nckx>Probably shouldn't, and it probably shouldn't be installed as part of the main package, but libschily is a NIH-fest worthy of downloading the source out of curiosity :)
***iyzsong- is now known as iyzsong
<atka>hi quix, I'm testing a server configuration and I'm not quite clear on how to set up the ssh keys. I have a test server which I currently have password based ssh access, I have my laptop which I'm sshing from, and I have generated my test key on said laptop ( I'd like to have my keys in the config so they are "set up" already when config is deployed.
<nckx>jgart: What's the reason for python- packages in check.scm rather than python-check.scm? I know you're not the first to put one there, but the reasoning is not clear to me.
<atka>this line: `(("janedoe" ,(local-file "")), seems like the .pub key would need to be on the server already?
<nckx>atka: Hi! What's not clear?
<nckx>Ignore that; lag.
<nckx>Yes, the public key needs to be on the server.
<nckx>It has to know which key to allow.
<atka>ok, where should it be placed then? not using traditional ssh-copy-id correct? Is it possible to have the key in the config.scm for the server? say I want to install a server remotely and on reboot automatically have password access disable and only allow the key from the config?
<nckx>I've never used ssh-copy-id on traditional distroes, always scp. On Guix Systems I put my pubkeys in /etc/guix, together with my system.scm and its dependencies.
<nckx>There is no right place.
<nckx>It's possible to have the key in the configuration file in that you could, ehm, programatically write it to a temporary file and then pass that file's name to authorized-keys…
<atka>ok, oh my server I've been using ~/.config/guix/systems/server.scm, if I put my key in there it would be found?
<nckx>…plain-file, that's the one.
<nckx>I'm not sure what the semantics of (local-file "./relative-name") are. You'll just have to test.
<nckx>Whether you use local-file or plain-file, the effect is the same, your server will boot with the keys authorised.
<atka>yeah I don't know scheme well, so I was/am confused how it deduces the path to the key from ,(local-file)
<nckx>(authorized-keys `(("nckx" ,(local-file "")))) ; is on one of my machines, so it works.
<nckx>Where the key is in the same directory as the system.scm.
<atka>nckx: thanks for the help, I will experiment with it, part of the fun :)
<nckx>Good luck! Just FYI local-file/plain-file are very much Guix things (related to G-expressions), not general Guile procedures.
<atka>ok, thanks for info, I will play around with both and see what works best for my use case.
*the_tubular Me when I guix pull today and see 190 new commits
*the_tubular realizing they are all R packages
<the_tubular>*Insert Meme here*