IRC channel logs
2024-09-17.log
back to list of logs
<attila_lendvai>nckx, thanks! when i issue a `make check-system` in my checkout, then it doesn't even start testing due to the ganeti build failure. <mange>I think the #:select with #:use-module has to be a list? That is: #:use-module ((gnu services base) #:select (udev-service-type)) <apteryx>the traceback lacked that context :-/ <mange>Yeah, that error message was not helpful. I just got lucky that it was the first code change in your diff. :P <PotentialUser-78>Hello, I am using Renesas Synergy S7G2 controller with GUIX. I am using USB Mouse to handle the GUIX operations. It works fine normally. But when clicking a button if my screen changes, then the mouse cursor area of the screen which was clicked will show up in the new screen. It goes away after clicking again on the new screen outside the previous <PotentialUser-78>cursor frame. Can someone help me out if they have faced any such issues? <aswjrisp>PotentialUser-78: What desktop environment or window manager are you using? <aswjrisp>PotentialUser-78: What application are you using when this happens? <PotentialUser-78>I am using windows pc. I dont understand your question regarding application? Its a custom code for my application. <pjals>Oh god, I thought Azure GUIX users didn't exist. <Rutherther>PotentialUser-78 you will probably have to find a different channel. This channel is for GNU Guix, and it is something different. <jmes>I'm helping a friend install guix manually and it seems that guix system init didn't create /proc /dev /sys or install anything in /bin... We don't get prompted for LUKS passphrase at boot and cannot chroot because nothing exists in /bin. Not sure what went wrong here <Rutherther>jmes when I was first installing guix I also didnt get these folders created until the first boot that does the activation. I think init doesnt do an activation <Rutherther>jmes I even tried the bash from store, but the missing /etc files were still an issue. <jmes>Interesting, that makes sense. So if we can get the LUKS prompt in order to boot, then everything should be swell. <jmes>I just don't know why the LUKS prompt doesn't show at boot <Rutherther>jmes have you tried manually mounting from grub if at least that works? Ie. insmod luks, cryptomount with the partition <jmes>Not yet, but will try that now. <jmes>Thanks for the help by the way, Rutherther <apteryx>ACTION just learned newer joystick devices use the evdev API, older ones the joydev one. <jmes>Rutherther: grub rescue does not see the unecrypted boot or encrypted LUKS partition, just the "top level" device (hd0) in this case... <jmes>Makes me suspect the grub configuration is off, but it's auto-generated by guix system init with a config file that is near-default. <apteryx>hm, a 'nintendo switch pro' controller, with vendor == 057e and product ID == 2009 is not made available to unprivileged user, despite having installed the 89-joycond.rules udev rule file that has: 'ATTRS{id/vendor}=="057e", ATTRS{id/product}=="2009", ATTRS{name}=="*IMU*", TAG+="uaccess"' <jmes>Rutherther: Also, insmod luks and insmod luks2—when run manually—give an error about "unknown file system" <Rutherther>jmes yeah, you need the /boot partition. Are you using grub-bootloader or grub-efi-bootloader? <apteryx>any thought what could be going wrong? 'udevdadm monitor' only shows add/change/bind/remove/unbind actions happening <jmes>Rutherther ^ forgot to ping. <Rutherther>jmes hm, I dont really understand why grub would not see the unencrypted partition <jmes>Yeah, I'm a bit stumped. ESP flag is on, vfat FS, bootloader target is /boot/efi <jmes>Perhaps legacy mode is enabled on my friend's motherboard firmware or something. Will investigate <f1refly>I just noticed that my weechat is also unable to start python: ImportError: /gnu/store/ln6hxqjvz6m9gdd9s97pivlqck7hzs99-glibc-2.35/lib/libm.so.6: version `GLIBC_2.38' not found <apteryx>hm, I don't get why TAG+="uaccess" doesn't work on a device <apteryx>ACTION throws the towel for now and adds self to the 'input' group <futurile>morning civodul and apteryx-not-here-right-now <civodul>f1refly: re weechat, it could be that some of your packages were upgraded and end up using glibc 2.39, while others are still on 2.35 <futurile>Guix Social tonight - good talk by ekaitz on RISC-V bootstrap coming up! <f1refly>how could I fix that? I currently cannot guix upgrade my user profile because icedove-l10n fails to build :( <civodul>futurile: yay! nice program as always! <f1refly>is there another one you would recommend instead? <civodul>f1refly: argh; was weechat itself upgraded? <civodul>i don’t know much about graphical email clients <f1refly>I suppose I could try to upgrade only weechat <f1refly>"Hey! Builds initiated with `mach build $A_SPECIFIC_TARGET` may not always work, even if the code being built is correct. Consider doing a bare `mach build` instead." <f1refly>is their built system not deterministic? <f1refly>upgrading only weechat worked which is nice <f1refly>now it would be nice if my sudo would work again <civodul>f1refly: you’re supposed to run sudo from /run/setuid-programs (before) or /run/privileged/bin (starting from a month ago) <civodul>running $(guix build sudo)/bin/sudo cannot work because it’s not setuid-root <f1refly>/run/privileged/bin is also in $PATH, but shadowed by .guix-profile/bin <civodul>f1refly: not sure why, but make sure to remove sudo from your profile <f1refly>yes, sure. however that ended up in there <willcl-ark>Morning all, i've tried to add a custom channel (all config now removed), but `guix pull` gives a 502 error. Is there a chance that the servers are down, or is this likely a hangover from my previously messing with more channels? <willcl-ark>Ah I think I fixed. it was indeed a current channel issue. <Franciman>hi, is anybody still having sigfault for telegram desktop? <nckx>aswjrisp: Meh, ‘feature requests’ write about as much code as someone who doesn't write much code, but the comment I linked to directly links to the bug number (#27155) where this is mentioned as desirable by someone with commit access and some vague knowledge of Guix. All that is missing is someone who cares enough to add it (you? :-). <mrvdb->Is there a video recording of the jitsi meeting from Aug. 29? <apteryx>can a mapped-device depend on file-systems? <apteryx>on the root file system; it's an extra device for which I'd like to use a key file which will be stored in /root/secrets/my-secret.key <apteryx>looks like we already have luks-device-mapping-with-options for that <apteryx>which supports #:key-file e.g. /crypto.key. It's not clear if the dependency on the root file system is automatically added though. <apteryx>also, it seems that in theory using a motherboard-provided TPM module could host the keys... I'll have to read on that <kaij>having some issues with the root user. I don't fully understand where it's defined or how I'd declaratively change it's password. I've always used it ootb without any password, but now suddenly. <kaij>now, after a reconfigure, I enter root as username, and am the prompted for a password. upon submitting with enter, the password prompt just reappears (on the same line?) and can never login. eventually it just times out. <jmes>kaij: The root user should be in %base-user-accounts <apteryx>kaij: for a newly init'd guix system, the root password is empty <apteryx>after rebooting, you should change it with 'passwd' as root <kaij>jmes: the docs suggest that %base-user-accounts specifically don't include root? <jmes>Ah, then I'm probably wrong. <kaij>apteryx: It's not newly inited anymore, but I've always justed used it without a password (while just toying) but now I just can't log in. I also have other accounts that have a declarative password set, but can't login either. seems like I somehow broke the login without even touching it <jmes>ACTION regrets speaking with authority <ieure>kaij, I don't think you can declaratively manage user passwords, because Guix has no facility to store sensitive information securely. <ieure>I see. Not sure why the installer doesn't use that, then. <kaij>I'm decently confident that that works, but I've just been unable to get the login screen to actually function <kaij>I'm also unsure where the root account is specified, or whether I can override it somehow. not sure. I can't login with any other account either though. <ieure>That *is* less secure, though. The hashes usually aren't exposed -- /etc/shadow isn't world-readable like stuff in the store. <ieure>If you use that facility, anyone could grab the hashes and try to crack them. <divya>Hello, how do y'all setup GPG? I tried the setup in the Manual and it gives me the 'No Pinentry' error. When I do have `pinentry` and `pinentry-emacs` installed system-wide. I also have the `home-gpg-agent-service-type` in my home configuration. <divya>(service home-gpg-agent-service-type <divya> (home-gpg-agent-configuration <divya> (file-append pinentry-emacs "/bin/pinentry-emacs")) <divya> (extra-content "allow-emacs-pinentry"))) <divya>Without this, I also tried doing it manually with a hand-written `gpg-agent.conf` that also didn't work. <civodul>divya: i have the same, just without ‘extra-content’, and it works for me <divya>Let me try both of them, wait. <Lumie>The env var GPG_TTY is also set to tty, but I'm not sure if it makes a difference <divya>Nope, I just removed the `extra-content` and added the environment variable but it still shows No Pinentry error <divya> ✘ ~/.dotfiles/.config/guix master $ which pinentry <divya>/run/current-system/profile/bin/pinentry <Lumie>That's because it needs to be "/bin/pinentry" set in pinentry-program <jmes>divya: Did you also try "/bin/pinentry" instead of "/bin/pinentry-emacs"? <jmes>Yeah, what Lumie said is what I'm thinking. <Lumie>Hence the packages I mentioned <divya>jmes: I did change it, copied verhabtim from Lumie's config and still gives the same error. <divya>I have pinentry and pinentry-emacs installed systemwide. <Lumie>pinentry-tty is what you also need, I don't know why but without that combination it seems to fail, at least for me <divya>Okay, putting that in system config and let me try a `guix system reconfigure` and try again. <jmes>divya: Just to be sure, you tried (file-append pinentry "/bin/pinentry")? This is what I have and it works. <divya>Yes, jmes, this is what I have as of now: <divya>(service home-gpg-agent-service-type <divya> (home-gpg-agent-configuration <divya> (file-append pinentry "/bin/pinentry")) <Rutherther>Lumie better to relog after home reconfigure, or at least to restart gpg agent. Did you do that? <Lumie>Also, if I may suggest: use a pastebin instead of spam <divya>Rutherther: How can I restart gpg-agent? Is there a herd service for that? <jmes>Yes, herd restart gpg-agent <divya>the system reconfigure is still happening <Lumie>Sorry, seems I'm impatient today. 😅 <aswjrisp>nckx: You are pointing out a big benefit of free software. Being able to make contributions that fix problems. In this case I think I would probably need to learn more about using guile. As an expediant I will manage the daemons with shell scripts. <divya>Thank you jmes , Lumie it worked :D <divya>`pinentry-tty` was the guilty, it was needed. <Lumie>I wonder if pinentry-tty is even needed if you just omit the GPG_TTY environment variable <Lumie>You would think that pinentry was suficient <jmes>divya: np, glad it's working for you <Lumie>jmes: do you have the tty addition of pinentry, or does it work for you without it? <jmes>For the record, I don't have pinentry-tty nor GPG_TTY set and it works <Lumie>This is what happens when in the beginning you simply steal configs from multiple sources <divya>Yeah, but I didn't have tty anywhere though... <PuercoPop>When sending a patch, does sending it to guix-devel in addition to guix-patches increases the chance that it will be acted upon? <ekaitz>PuercoPop: but you could ping some people you know here and we might try to prioritize it <PuercoPop>ekaitz: so I just share the link to the issue here? Are you also a Spanish speaker? <civodul>sneek: later tell i fixed two test failures; ‘tests/crate.scm’ remains, but requires a bit more expertise; any idea? <civodul>sneek: later tell apteryx i fixed two test failures; ‘tests/crate.scm’ remains, but requires a bit more expertise; any idea? <civodul>efraim: or maybe you would know why that one fails? <ekaitz>PuercoPop: oh! i'm from the place where your surname is coming from <fnat>futurile: I may pop out for a bite, are you around to check audio/video/Jitsi in ~25 mins? <kaij>is there any good way of debugging my login issues? like can I somehow make the init process drop into a logged-in shell directly and find out why it's not working? I'm super confused what's going on <Rutherther>kaij what login issue specifically? Login into what? <kaij>the system. I described it a bit earlier. but basically after my latest system upgrade I cannot login anymore. I have never set the root password, but still get prompted for a password. if I try logging into a user with a password, I just get asked for a password again (weirdly on the same line). <kaij>eventually the login process just times out (regardless of what I do). haven't managed to get in <kaij>I'm getting some warnings about shepherd and unsupervised fds but not sure how that's related, if at all. <Rutherther>kaij: but what does into the system mean? display manager? tty? something else? <kaij>it's a VM on my server, so just the "main" tty. <Rutherther>times out, as in you get another prompt for username and password? <kaij>I don't remember the exact text, I can bring it up in a minute. it's just the standard login timed out after 60 seconds etc. text, as if no password was entered. <Rutherther>hm, I have never seen something like that, so I don't really have any pointers for debugging :( maybe just rechecking pam config, and it's just a hunch, I don't even know if wrong pam config could cause this <kaij>yeah the problem is I can't really do anything to diagnose, just rollback and try again... <ekaitz>fnat futurile i'm ready for the sound check <fnat>ekaitz: Bear with me a min. <fnat>ekaitz: Ok, I'm online now. <fnat>futurile: You also around already? <kaij>Rutherther: just in case you're interested, I checked my git diff and found I had added a line to a shepherd service, with a startup command that was obvious nonsense, basically "TODO: add stuff here". apparently that made everything go crazy. interesting. <kaij>Rutherther: shepherd system <Guest4>should the guile package set GUILE_EXTENSIONS_PATH so that it can find foreign libs? <PuercoPop>ekaitz: are you from Pais Vasco? Or do you even know a more precise location for my surname? I've been told it was from there. <ekaitz>PuercoPop: yes i am from the pais vasco <ekaitz>PuercoPop: it means "smith"-"house" <jaft_r>For anyone who uses Flatpak on Guix, is there anything particular one had to do beyond "guix install flatpak"? <PuercoPop>jaft_r: There is a variable you need to have in your profile, at least for the app to be picked up by launchers <PuercoPop>ekaitz: Nice, I thought it was related to wheat or barley. So 'casa de herrero' but the first part means smit and the -chea house? <ekaitz>PuercoPop: it's Ola -> smith + Etxea -> house <ekaitz>and written in spanish it becomes a CH <ekaitz>(sorry for the offtopic but it's an interesting topic, we must admit) <ekaitz>PuercoPop: also for your issue, maybe janneke is interested on it <ekaitz>he's one of the authors of lilypond, and also a guix hacker <Rutherther>aswjrisp: yesterday you asked how to disable autostart of a service, I think this is a general way: https://paste.debian.net/1329650/, there is an example usage for openssh-service-type. Note that I have not tested this in actual operating-system, only printed in repl the expected computed value. <Rutherther>it will probably not work when the service is not a leaf service <poddo3>Hey can someone help me with some silly stuff <poddo3>I'm trying to run a binary I downloaded, <freakingpenguin>Is there a standard way to add+set a locale in a guix shell container? Seems like manually setting LANG+GUIX_LOCPATH in the container is the only option. <poddo3>I assume its like, individual libraries being linked to packages that require them, but I have installe lbzip2 bzip2 <poddo3>I didn't know guix could even do that <poddo3>I noticed that on re-run it doesnt re-grab the libraries I supplied, where do they get configured? <ieure>poddo3, How did you "supply" the libraries? <ieure>I don't understand your question. <poddo3>guix shell --container --network --emulate-fhs coreutils make autoconf <ieure>You mean it doesn't download them a second time? <ieure>That's normal, intended, and good. <ieure>Anything Guix downloads goes into the store -- /gnu/store <poddo3>I guess I'm wondering, is the directory getting configured for the shell, or just at the command run? is the store grabbing these things and adding them to the users store? idk <ieure>Things are downloaded at the time they're needed, if they need to be downloaded. <poddo3>wait whats the difference between running guix install and reconfiguring through the system then <ieure>The first time you ran `guix shell', it needed to; the second time, it didn't. <ieure>The libraries were inside the container both times.