IRC channel logs

<PotentialUser-24>does anyone know how to edit an existing udev rule? how should i define it in my system config?

<PotentialUser-24>sorry i disconnected there

<nutcase>Hi, is it only me having "ssh: Could not resolve hostname codeberg.org: Name or service not known" ?

<nutcase>dig codeberg.org gives me an IP

<Sambalala>try to conect to the ip if its alowed maybe something with the dns

<Sambalala>or check if its the same ip aver ssh and dig

<nutcase>Sambalala: I'd have to reconfigure my git remotes. That's not what I want to do. I could wait until tomorrow, if I knew it goes away

<nutcase>SSH does not have an IP, it seems

<nutcase>SSH does not have an IP as in "debug3: resolve_host: lookup codeberg.org:22 ssh: Could not resolve hostname codeberg.org: Name or service not known"

<nutcase>30 minutes ago, it worked

<Sambalala>try ssh -vvv -o ConnectTimeout=10 -p XXXX user@domain

<nutcase>same

<nutcase>ssh can't resolve the domain name. dig can resolve the domain name

<nutcase>ssh can connect to codeberg via IP but not via domain name

<nutcase>it's only with codeberg. Other hosts work

<Sambalala>try getent ahosts codeberg.orgAND getent hosts codeberg.org, then grep '^hosts:' /etc/nsswitch.conf

<nutcase>I did a reboot and now it works again. Strange

<Sambalala>you should all years min. reboot ones! ;)

<nutcase>I was heavily confused, since I just started using emacs-agitjo and ran into this issue. Of course, I thought I'm using agitjo wrongly, but the issue was not an agitjo issue but a DNS issue.

<nutcase>it feels so wrong to reboot in order to solve a problem and to succeed with that strategy :-(

<Sambalala>its better to reboot then to repair than or more min.

<Sambalala>10 or 30 min.

<Sambalala>at reboot you do not have to change something

<Sambalala>its better to wrote a iphone away as to try to install fdroid on it or?

<nutcase>probably

<PotentialUser-24>is there documentation for writing udev rules? specifically existing ones? i need to edit the TAGS for a device i have to MODE instead of uaccesss

<nutcase>PotentialUser-24: you know https://guix.gnu.org/manual/devel/en/html_node/Base-Services.html#index-udev_002dservice_002dtype ?

<nutcase>and https://guix.gnu.org/manual/devel/en/html_node/Base-Services.html#index-udev_002drule

<PotentialUser-24>nutcase yes. if i make a udev service for the software i have, wouldnt that rule erase what was already there?

<nutcase>PotentialUser-24: sorry, I don't know that.

<PotentialUser-24>nutcase hmm...thanks for the links though

<PotentialUser-24>nutcase since im using MODE, is `0660` is a good permission to use? i need it for my user group

<nutcase>PotentialUser-24: I can't help with that, sorry. I just showed you documentation for writing udev rules, but I don't have and knowledge about that.

<nutcase>s/and/any

<dcunit3d>PotentialUser-24: the udev rules "compose" by concatenating rules. lots of services do that.

<dcunit3d>when you can, you should always approach the specification of services by leveraging that

<dcunit3d>PotentialUser-24: you can find other channels, programs, services by searching https://search.guix.moe and https://toys.whereis.social

<dcunit3d>switchy: that's awesome. thanks for letting me know. so the whole swapfile device is also encrypted? are you booting from a vfat /boot or /boot/efi?

<PotentialUser-24>dcunit3d say if my current rule is defined like this

<PotentialUser-24>`(string-append "ACTION==\"add\", SUBSYSTEM==\"misc\","

<PotentialUser-24> TAG+=\"MODE", MODE=\"0660\")` is this the correct format? i need to specifcally edit the `uinput` kernel.

<dcunit3d>it needs to be one line, i think. i would need to check. sometimes, like with android-adb-rules you can add the package, then the guix-daemon knows how to take that and find/produce the udev rules. i need to check the specifics of that though

<dcunit3d>like here with libfido2 https://codeberg.org/gofranz/panther/src/branch/master/px/system/panther.scm#L26-L35

<PotentialUser-24>dcunit3d oh ok. so i dont have to do anything special to the snippet i posted? that would be enough?

<dcunit3d>PotentialUser-24: the examples with android-adb-rules are pretty good when you're first figuring it out https://guix.gnu.org/manual/devel/en/guix.html#index-file_002d_003eudev_002drule

<PotentialUser-24>dcunit3d oh thanks! ahh, i wish i read the documentation more closely.

<dcunit3d>yeh i had some issues with the specifics when i tried to move the udev rules under a service extension of another service. https://guix.gnu.org/manual/devel/en/guix.html#index-openssh_002dservice_002dtype

<dcunit3d>but it makes managing custom udev scripts pretty simple.

<dcunit3d>let me know if you have other questions about it. i haven't messed with udev in a long time. i'm working on getting some guix servers set up in my homelap. i had run a guix i3 desktop on my laptop for awhile

<PotentialUser-24>dcunit3d i do have another question. i assume `file->udev-rule` lets udev know that the rule must be edited? what happens if i just had the snippet i posted in the config? will the rule that im trying to target erase what was defined by the package?

<PotentialUser-24>this is what i want to achieve just for the sake of posterity:https://wiki.archlinux.org/title/Udev#Components_of_udev_rules . im looking at the allowing regular users to use devices section.

<dcunit3d>so file->udev-rule is a procedure (scheme function thing) that expects a "file-like object" (a thing guix knows how to turn into a file) and it returns a udev-rule

<dcunit3d>i think i've described that accurately

<dcunit3d>so you would instead use `(udev-rule "99-file-name-in-created-system.rules" the-udev-string)`

<dcunit3d>and then pass that scheme value into `(udev-rules-service 'test-udev-rule udev-rule-value)`

<dcunit3d>file->udev-rule is more used internally, i think

<PotentialUser-24>dcunit3d i understand that part. so the full define snippet should be:

<PotentialUser-24>`(define %my-rule

<PotentialUser-24> (udev-rule "99-my-rule.rules" my-rule

<PotentialUser-24> (string-append ...)`

<PotentialUser-24>then add the udev-rules-service?

<dcunit3d>)) yes.

<dcunit3d>and later, how you do that depends on how you want to share code between systems.

<dcunit3d>but ultimately the udev-rules-service just needs to be cons'd or appended to the list of services you provide to the operating-system record

<dcunit3d>so that could include multiple invocations of udev-rules-service

<switchy>dcunit3d: yeah, I have a fully encrypted root, except for a vfat /boot/efi

<dcunit3d>nice. i just chose to disable swap lol

<switchy>I'm just sick of my laptop dying in suspend, so next step is hopefully that elogind's suspend-then-hibernate just works

<jresich>hi guix

<dcunit3d>yeh that was a problem i was dealing with for awhile switchy. i just adapted around it at the time bc i wasn't sure what to do

<dcunit3d>i think i had disabled hibernate (no swap anyways) and only suspended, but battery life was the main issue actually

<jresich>how do I "stop" using guix home? I've fully switched over to using the system home service instead of the 'guix home' command but 'guix home describe' indicates i still have a profile pinned from over a year ago

<dcunit3d>i've always been a bit confused about that. i had the same issue when setting up nixos bc i had planned on not using home manager

<dcunit3d>jresich: run `guix gc --list-roots` and then check through `/var/guix/profiles` to see what they actually are. idk if you can safely decouple them.

<dcunit3d>without determining what the cause of `guix home describe` output is

<dcunit3d>for me, i wanted the opposite: to decouple home updates from system updates. i wasn't sure what would happen if i started with a home configuration in system and then needed to decouple it later

<dcunit3d>or like if i set it up the other way (using `guix home`) and then did `guix system reconfigre` (either with or without a home configuration) like what would happen and what transition pathways were available

<dcunit3d>also, code sharing is a bit complicated for any of the possible combinations, since you usually need /some/ guix system components aware of logic in your home configuration AND vice versa

<jresich>dcunit3d: i think it works with guix having three 'profile lineages', each for the guix `system`, `home`, and `package` commands, then the selected profiles from each of these lineages are overlaid

<jresich>it's tricky because each of those profiles might not necessarily be built from the same version of guix

<jresich>hence why eventually i went with just doing everything in the system profile, and using the system home service from there

<dcunit3d>yeh see, i'd rather not have to worry about all the details. there's a ton of functionality that i want from `guix home`, but there's that extra complexity, a huge potential slowdown in updates and the inability to simply update files

<dcunit3d>but i think `sudo guix system reconfigure` uses the same profile as `guix pull` right?

<dcunit3d>i meant **the same channel spec in ~/.config/guix/channels.scm

<dcunit3d>and the guix in ~/.config/guix/current/bin/guix

<dcunit3d>bc i need to run `sudo --login guix pull` to update the root user's guix

<dcunit3d>how can i be sure that guix-daemon will use substitutes...

<dcunit3d>nevermind i see here that it is... strangely, i didn't see the traffic show while running tcpdump on the router. i think maybe because the router didn't accept those frames which were directly sent on the vlan

<dcunit3d>ah ok cool. it also logs to the systemd unit

<folaht> https://forum.systemcrafters.net/t/folaht-s-first-attempt-to-build-a-cargo-package-on-guix-system-the-one-with-windows-dependencies/2073/7 <- Can anyone help me with this?

<folaht>Hmmmm... I think I have an idea

<czan>Anyone feel like reviewing+merging a pretty minor change that would let me remove some lines from my systems that I deploy with "guix deploy"? https://codeberg.org/guix/guix/pulls/6866

<wehlutyk>Hello Guix

<wehlutyk>DNS lookup on my Guix system fails since end of June

<wehlutyk> https://paste.debian.net/hidden/bf8d26ba

<wehlutyk>(appeared between June 19 and June 22)

<wehlutyk>any hints?

<sham1>Are you using NetworkManager? If so, restart that service

<wehlutyk>yes, using it, I'll try, thanks

<wehlutyk>sham1, thanks! I came across 35162b8e195ce7f2eeb1e7cc925c2d19e57cb1e6 about this and it worked

<look>folaht: https://paste.debian.net/hidden/0fc50517

<folaht>look, can you explain the minalloc part?

<folaht>mimalloc

<look>folaht, https://github.com/WithAutonomi/ant-node/blob/93b35798a56c78be58ec6da9b5fd25cae76007f6/Cargo.toml#L26

<look>its a direct system lib dependency

<look>additionally, you could, in theory, make all the tests pass, but its just too much trouble for a build container with no network access, so its ok to simply disable the failing ones

<folaht>If it's a problem with the makers of that app, I'll add them.

<folaht>I'm just glad to be able to make this package available.

<folaht>Or even a package available.

<folaht>Is there a reason for going git-reference?

<folaht>I thought it'd be nice to avoid github.

<look>There are some reasons sometimes, because crates-io does additional dependency stripping and the resulting lock will often be more concise to the strictly required dependencies, but it makes updating the package harder because it takes a while for new releases to show up on crates-io, and often the crates-io dependencies don't match exactly upstreams, which can cause problems

<look>there are some other more complex reasons for why I prefer using the git source directly, but overall using git-fetch makes updating the package easier

<look>sometimes its the opposite though, really depends on the app tbh

<yarl>Hi

<saeed>nvm

IRC channel logs

2026-07-11.log