IRC channel logs
2023-11-24.log
back to list of logs
<Kabouik>Are we having issues with the substitute servers? <Kabouik>I'm getting 504 errors on a guix install <Kabouik>lechner, jpoiret: sorry for the late answer, I missed your replies. he packages are nmail and nchat (both are TUIs, for emails and Telegram, respectively). Fine,, I'll only submit the release packages then, and keep the -master packages for my private channel (they use git-checkout jpoiret, indeed). <vivien>In the gnome-team to-do list, I see: – rename rest to rest-minimal and rest-next to rest, use rest everywhere possible and rest-minimal everywhere necessary; – update mm-common, mutter, pangomm@2.46, python-pyatspi, rygel, simple-scan, sushi, vte, yelp, yelp-tools, yelp-xsl; – add xdg-desktop-portal-gnome <vivien>Maybe we will also like to have gnome-tour, but I need a rust packager for that <Kolev>efraim, what issue does chromebook-alsa-lib have? <Kabouik>Package transformations like --with=branch=package=master only work for packages that have a (git-fetch …) though jpoiret, correct? Many have a (url-fetch …). <Kabouik>Oh, I see that I can ag1add a --with-git-url=package=giturl and then use --with-branch. Not something that can be automated very easily though if the URL has to be researched first, I guess my "master" packages based on (git-checkout …) iwll have a use in my private channel then. I was considering deleting them in favour of the released-based ones + transformations. <oscarvarto>LD_LIBRARY_PATH=$GUIX_ENVIRONMENT/lib ./WezTerm-nightly-Ubuntu20.04.AppImage --appimage-extract <oscarvarto>./WezTerm-nightly-Ubuntu20.04.AppImage: error while loading shared libraries: libz.so.1: cannot open shared object file: No such file or directory <hapster>when installing a package (from one of the unofficial channels), I am met with the error `guix install: Error: while setting up the build environment: getting attributes of path `/etc/services': No such file or directory`. I have been able to install other packages, but wonder whether this is a general problem unspecific to said package. Has someone here faced a similar issue? <hapster>I'll leave this one for now and later check the logs, thanks a bunch for any input :-* <jfauhn>Hi, I am running into issues in getting set up with user space services with herd. Where I can't boot into a tty after a reboot. Anyone else had a problem like that? <zimoun>civodul: About #67250 (builtin:git-download and guix-daemon version), does it lead to issues when offloading? If my laptop runs a new guix-daemon and it offloads to a powerful machine with older guix-daemon? <alethkit>Is there any specific trick to get nix.conf to accept substitute URLs? <civodul>zimoun: yes, offloading could lead to similar issues <civodul>alethkit: hi! what do you mean “nix.conf”? :-) <alethkit>as in what would go on the (nix-configuration (extra-config)) string <civodul>i guess the Nix manual has suggestions for that? <civodul>maybe we should link to it and/or include a typical snippet in our manual <alethkit>although I now seem to have an inode issue <alethkit>didn't realise that it's set at FS creation time <nikolar>i installed rust with guix on a foreign distro (ubuntu) <nikolar>and i get linker errors when i tried to build the simplest procejt <nikolar>error: linking with `/gnu/store/f95ljc7bism88x1vrrggnqxa8r9i4llb-gcc-11.3.0/bin/gcc` failed: exit status: 1 <jpoiret>nikolar: you probably need gcc-toolchain as well <nikolar>i relaunched the terminal and it seems to be fine <f1refly>I asked this before, what would be the best way to have 2000 be written to /sys/module/block/parameters/events_dfl_poll_msecs once at boot time? My system fails to detect insertion/removal of sdcards into the reader when i leave it on the default value 0 <old>f1refly: could a oneshot system service do the job? <f1refly>probably yes, but I wasn't sure how I could tell it to write to files <flypaper-ultimat>civodul:cool! don't they have the description of inputs and native-inputs swapped in the interface though? <flypaper-ultimat>the helper text in the 'native-inputs' says that its for running & building, and the helper text for `inputs' says its for building, but that should be the other way around, right? <yarl>I have a problem on my freshly reconfigured guix system. Networking fails to start at boot. <yarl>I can then `herd start networking` without problem. <yarl>Where can I find or configure something to see verbose logs about what happened please? <yarl>It's a very basic 'dhcp-client-service-type' <lechner>yarl / will you please pastebin your config? <lechner>sneek / later ask mirai / Hi, is there a way to deprecate or supersede service configuration fields? <lechner>yarl / what was the state of your guix the last time you reconfigured, please? <yarl>lechner: What do you mean? <lechner>i think there were some changes in the ordering and execution flow in the shepherd services. is there anything in /var/log/messages? <yarl>lechner: only "Service networking could not be started." "Service networking failed to start." <yarl>lechner: Can I pass options to, I don't know, the kernel, to have a more verbose /var/log/messages? <yarl>or having shepherd being more verbose? <lechner>yarl / looking at g/s/networking.scm, i would add user-processes to shepherd-requirement in your dhcp-client-configuration <lechner>yarl / i also believe it should be in the defaults, rather than '() <civodul>yarl: are you using static-networking-service-type? <yarl>civodul: troyfigiel tylerm unmatched-paren unwired unwox user3456 V vhns viaken <yarl> vivien vlorentz voroskoi vup warren whereiseveryone wingo_ wose xd1le <yarl> Xe xelxebar xgqt xigua xMopx yang yarl yelninei yvan-sraka zamfofex <yarl> zimoun zoglesby zxtx zyd <yarl>16:27 -ChanServ- [#guix] Be kind to everyone: <lechner>yarl / i posted your config for civodul and mentioned dhcp <lechner>Hi, guix deploy announces builds but does not show their logs. is there a way to tail them in real time if i have the drv location? <lechner>i'm bootstrapping so it takes super long <lechner>actually, it's guix pull, not deploy <yarl>Yes lechner I'm being reconfiguring with this (without "s" at requirement) <lechner>sorry, just trying to help but i am fishing <yarl>lechner: You are helping a lot! <lechner>yarl / you may want to switch to a higher-order service, such as connman <yarl>lechner: Works perfectly with connman. <yarl>Still, that's annoying not knowing why it doesn't work with dhcp-client-service. <lechner>yarl / mirai may know the answer but i don't <yarl>While I am playing with config.scm. My system is currently running on one single drive, I got another one inside not used with about the same capacity, I would like to make a raid0. Is that possible with reconfigure? I guess it's a bit "dangerous", I won't have roll-backs. Any advice on doing this? I saw 'mapped-device' <civodul>yarl: i’d expect dhclient error messages in /var/log/messages; is there no such thing? <yarl>civodul: it just says it can't start. <civodul>yarl: and nothing above that “failed to start” message from shepherd? <civodul>we really to look at everything between “Starting service networking” and “failed to start” <lechner>yarl / what kind of network card is it, please? <yarl>Still, it's strange that I can `herd start networking` successfully if it's an hardware problem, don't you think? (I only say this if civodul did not see) <lechner>yarl / you could potentially boot a reduced system without dhcp-client-service-type and then run dchpclient yourself. it's a pretty simple program if you know your way around command-line interface config ('ip' or 'ifconfig' of old) <yarl>lechner: Why would I do this? :) <lechner>i am not sure. you wanted to investigate <civodul>yarl: i agree, you could just run “sudo dhclient -v” from the terminal and see what happens <yarl>civodul lechner: ah I see. <yarl>let me reconfigure, reboot and test. <lechner>you could start by turning off the interface and then running the command civodul suggested <yarl>“sudo dhclient -v” worked. <lechner>yarl / well, that's more or less the same as herd restart networking. the challenge is at boot <yarl>The strange ACPI messages appear just below "Starting service console-font-tty6...". <lechner>yeah, i can't tell what device they are from but ACPI generates a lot of errors. most implementations are faulty <yarl>What about adding some delay before starting dhcp-client-service, is that possible? <yarl>I don't know, maybe the network card need some time? <yarl>Actually, it might not be that stupid since "Starting service networking..." appears before "e1000e: Intel blabla Network Driver" <yarl>then "dhclient: No broadcast interfaces found - exiting" <yarl>Haha! There's dhclient "-w" option <yarl>lechner: I don't know, I think I got some precious informations now but what can I do with these? <lechner>i would drop the -w and reconfigure from Git (or use a custom guix pull, depending on your requirements) <voroskoi>I have mixed feelings about the recent blog post on packaging. IMO guix is not short on packages, but people to review them <voroskoi>the issue tracker is full of patches without any answer <yarl>lechner: I'll look. Have to go away for now. Thank you a lot. <lechner>voroskoi / packages do not have to be accepted into Guix in order to be useful. i have a small custom channel and also run a custom copy of Guix for that reason <voroskoi>lechner: sure, You can do that, but if everyone does so it is quite a waste of resources <voroskoi>and I am not aware of a semi-official channel like aur for arch for example <voroskoi>where the community can share packages with each other <lechner>voroskoi / i'm not sure i agree. i'd rather maintain my own channel than comment on others. i have a kind of directory people could potentially submit their packages to but it's not quite ready <NoahMetz>Does anyone have experience with getting `nslcd` to authenticate tty logins and SSH? I've been banging my head against this for a few days and have it narrowed down to a library issue, but not sure how to proceed from here. <NoahMetz>What works is running `LD_LIBRARY_PATH=/run/current-system/profile/lib su - ${ldap_username}` and `LD_LIBRARY_PATH=/run/current-system/profile/lib /run/current-system/profile/sbin/sshd -D -p2222 -f ${sshd_store_path}`. <NoahMetz>Running `strace su - ${ldap_username}` and `strace /run/current-system/profile/sbin/sshd -D -p2222 -f ${sshd_store_path}` shows that they're both trying to load `libnss_ldap.so.2` from the `glibc` store path(which it doesn't exist in obviously). <NoahMetz>I've tried adding libnss-pam-ldapd as an input to the openssh package and using that as the `openssh` in the `openssh-configuration` for the service(plus installed on the system for the cli test), but strace shows the same `ENOENT` issue trying to find the libnss_ldap.so.2 library in the glibc profile path. <NoahMetz>Any help appreciated(or a redirect towards a better channel to talk about this). <elevenkb>Has anyone tried to connect to a computer running the installation image over TRAMP? <oriansj>elevenkb: it isn't the most effective way to deal with remote installation; as you could just have wget download the desired configuration file and apply it. <Franciman>is there a way to specify a pattern in ~/.config/guix/shell-authorized-directories? Or do I have to manually write each and every directory? <bdju>is anyone working on fixing gajim-omemo? I saw issue 67367 on the mailing list and have been scared to upgrade gajim now. but if I hold it back some other python stuff has conflicts <oscarvarto>Hi, I'm trying to run/execute an AppImage. This "almost" works: <oscarvarto>guix shell --container --network --emulate-fhs \ <oscarvarto> pkg-config zlib libgccjit libxcb libx11 fontconfig wayland libxkbcommon \ <oscarvarto> xcb-util xcb-util-image xcb-util-keysyms xcb-util-wm \ <oscarvarto> xkeyboard-config xdg-desktop-portal xdg-desktop-portal-gtk \ <vagrantc_>wow. loving the whole qa process! just tried applying patches for 67261 ... and even though QA has no information ... brodeaux does at least have substitutes <imadnyc>Hey, I'm trying to run guix in a vm. I want to enable the qemu guest services, so I ran "guix system search qemu", and found "qemu-guest-agent". I ran "guix system edit qemu-guest-agent", and found that the name of the service is "qemu-guest-agent-service-type". Then, I added that to the default config.scm, in the form of "(service qemu-guest-agent-service-type)". However, when I try and run "sudo guix system reconfigure... <imadnyc>... config.scm", but i get the error message "qemu-guest-agent-service-type: unbound variable". What should I do to fix the problem? <cbaines>vagrantc_, I think QA does have some information on #67261, maybe you got an error page? <vagrantc_>at least the vast majority of builds were substitutable ... makes it easier to focus on the broken parts :)