IRC channel logs
2023-12-08.log
back to list of logs
<pranshu>Hello, I am having an error whever I use `home-files-service-type'. Any file I use local-file to get the object of, I get a "invalid name error" <pranshu>can someone help me understand what that means <lh>trying to install emacs-transient from the melpa guix channel to override 29’s built-in version. but the built-in one is still loading <lh>checked load-path and /gnu/store/hash-emacs-pgtk-xwidgets-29.1/share/emacs/29.1/lisp is /before/ /gnu/store/hash-emacs-transient-version/share/emacs/site-lisp/transient-version. is that intentional? <lh>uh, in helpful the value has the transient guix package path, but *Pp Eval Output* doesn’t have it when evaling load-path directly. what in tarnation? <lh>never mind, my search in helpful was finding the result in initial-value. so something is removing it… <lh>and initial-value is a symbol property defined by Doom, so it’s a Doom issue. disregard <cmiller>sneek: Later tell pranshu: Does your file include a "."? I had the same issue with that. <xelxebar>Anyone here familiar with zig? Would be nice to have the latest 0.11 packaged up. <adanska>Hi isaneran! Doing anything Guix-adjacent today? <isaneran>Nah, not more than just running it as my distro <isaneran>I did do some tinkering the other day though <isaneran>mostly moving things I've installed through guix package into guix home file <adanska>same for me really. i might do some work on one of the tools im working on for my job, trying to learn how to use parser combinators to parse some strings in common lisp <adanska>using smug as a library, mainly cos its already packaged in guix <isaneran>Ah I should also mess around with parser combinators <adanska>i think im like one more day away from properly grokking them haha <adanska>just wondering if its just me: does gnome tweaks open for anyone? <adanska>its suddenly stopped working, and running it in the terminal doesnt show any debug output either <lilyp>I got a similar problem on my laptop with gfeeds. <lilyp>Pretty sure that some odd GLib/XDG thing is broken over there, but can't find out what. <adanska>i wonder when the regression happened. i've launched it before on this same install no issue <mfg>it's time-consuming but you could search for a previous system generation that worked and use the corresponding commit as a starting point for git bisect <mfg>assuming the problem isn't with some files in the home folder that are not changed by guix <adanska>that sounds like such a drag! but that may have to be what i do <adanska>hmm. failing with a 'Failed to register: Timeout was reached' <retropikzel>I'm making SDL2 program with C on guix and when I try to run it I get "could not initialize sdl2: dsp: No such audio device" would anybody have an idea what package to add to guix shell to make this error go away? <adanska>i added a few apps to autostart using gnome-tweaks, which stores some files in ~/.config/autostart <adanska>deleting this folder makes tweaks launch again <adanska>not sure why this is happening though... <mfg>adanska: Glad you found that out without having to bisect :D <mfg>adanska: did the version of gnome-tweaks change? maybe something has changed and the old file let's the program silently crash? <mfg>this happens for me all the time when using older blender versions... <adanska>no i dont think the version changed, i only added these autostarting apps a week ago, the last time i used tweaks. i think this is just a longstanding bug in tweaks <mfg>retropikzel: what packages do you have in that environment? <retropikzel>Quite alot for now because I'm not sure what would work so testing throwing everything into there <mfg>i don't see whats missing :( <retropikzel>mfg, I did some stuff and now it complains "-lasound" is missing. You know where that might be? <mfg>that's a compatibility layer between oss4 and alsa, it seems. I found that via google and at least debian has a package libasound2-dev i don't know the guix equivalent of that though <efraim>sneek: later tell civodul I'm building glibc-locales on powerpc-linux now <cnx>what's the difference between file-append and in-vicinity? <lilyp>file-append takes and produces a file-like object, in-vicinity works directly on files <abralek>I have this problem with a docker container I built with system image --image-type=docker. I added ssh-daemon there and configure ldap. However ldap does't work because ssh *in* the docker cannot find pam_ldap.so library. <abralek>I also have the very same configuration of Guix system itself which does work. and If I check sshd process with lsof in both environment docker doesn't have the required library and OS does <abralek>interesting, why is /run/current-system/profile/etc/ has different ld.so.cache? basically it links to some different packages <abralek>My main system ld.so.cache -> /gnu/store/y0zn6gnrp96pgjh16s5sci3582c6ggy5-dovecot-2.3.21/etc/ld.so.cache, server ld.so.cache -> /gnu/store/mp4s7s5by62zmy2l4mvaxn3gnhfjly13-tcpdump-4.99.4/etc/ld.so.cache <lilyp>That's interesting, but inconsequential. Each package pulls its own ld.so.cache, so you don't need them at the profile level. <gabber>pastor: i'll give it a review (which hopefully speeds up merging) <gabber>if you want/need it for your own purpose(s) you might want to consider spinning up your own channel to gap the waiting period(s) <gabber>what's the difference between (@ (guix licenses) lgpl2.1) and (@ (guix licenses) lgpl2.1+)? they both point to the same upstream URLs and only differ in the variable and license names <pastor>Thanks gabber! I already have it on a custom channel. <gabber>pastor: i just sent my review (: it's all good except for the license <lilyp>gabber: It's whether or not upstream permits you to reuse the package under a later version of the [AL]?GPL <gabber>lilyp: thanks! i thought this would be mentioned in the license file itself <cnx>how do i connect a guix system container to the internet? <gabber>how can i make a shared library available for a Python binary (which uses the ctypes FFI)? is patching the cdll.LoadLibrary() call the way to go or is there a more elegant/guixy way (which e.g. sets LD_LIBRARY_PATH in the wrapper script)? <pastor>gabber: should I resend a patch or since its such a minor change the person merging it will add the plus sign? <gabber>pastor: it's not just the plus sign, but also the letter `l' -> GPL2 vs LGPL2.1+ <gabber>i think the hacker doing the merge is capable of doing so but i think it would be nice of you to submit the most correct patches (but i have no idea how others feel about this) <lilyp>gabber: Hardcoding is typically the way to go. Don't set LD_LIBRARY_PATH. <pastor>gabber: okay. I will send the corrected patch <abralek>ok, seems like ssh cannot change its PID and groupid to nobody in docker <abralek>adding setuid/setgid capabilities doesn't help. ( <ieugen>hi, I am getting an error trying to launch guix shell: guix shell: error: failed to connect to `/var/guix/daemon-socket/socket': Connection refused <gabber>ieugen: are you running on Guix System or on a "foreign" distro? <ieugen>yes gabber, on Debian - but I did not have this issue before <ieugen>I was planning to try out guile-hoot <gabber>can you confirm the guix daemon is running (something like `sudo service guix-daemon status`) <ieugen>journalctl does not say anything either <gabber>i am not 100% sure but exited doesn't sound right - and might be the reason your guix client can't connect to the daemon <ieugen>yes, now I have to figure out how to start guile-hoot <kotet>Hi, I can't get eglot with clangd or ccls to play nicely with Guix. As the system headers are not in '/usr/include' but inside the Guix profile, the checkers don't recognize them as system headers and report bogus errors. Does anyone have experience with this? <gabber>my review for #66547 does not seem to have arrived at issues.g.g.o ? i've sent the email about 2½ hours ago <lilyp>Mumi is stuck the day before yesterday atm <attila_lendvai>why does shepherd support guile 2.2? can i assume a 3.x guile while patching shepherd? <lilyp>Shepherd is not exclusive to Guix. <attila_lendvai>cleaning up exception handling in shepherd quickly escalated to a point of breaking herd-shepherd comms protocol <attila_lendvai>i should probably write a mail to the list before i venture too deep into this <lilyp>It's also packaged for ALT Linux, SUSE Tumbleweed and (of course), the Hurd. <pastor>I was waiting to receive so I could answer back. It looks like it appears on debbugs though... <gabber>mumi is apparently stuck on the day before yesterday <gabber>pastor: my review didn't say more than what i said here (: <kotet>I have been able to get clangd to properly syntax check the current file, but still can't ignore the errors in Guix profile-stored system headers <pastor>gabber: Yeah. I've read it in the debbugs interface. I wanted to answer to your review to keep the thread look on the mailing list <gabber>pastor: you can still do that (it just doesn't currently show on issues.g.g.o) <graywolf>The build-aux/test-driver.scm is Guix original script? Or was it taken from somewhere else? <graywolf>I am trying to figure out how to credit it after copying it to my own project :) <Franciman>hi, today i had an issue with nscd, it kept resolving an hostname as 127.0.0.1, after a system upgrade it started working again <Franciman>when such things happen, how do i tell nscd to refresh its beliefs and correctly resolve names? <iK0u>can a graphical guix vm connect to the hosts wayland? <iK0u>by guix vm I mean guix system vm <nckx>Thanks lechner. Lookings, they will be took. <lechner>i have live in an english speaking country for 31 years and did not understand your second sentence <lechner>iK0u / i'm not sure. you may be looking at waypipe or vnc <nckx>failed to connect to debbugs.gnu.org (2001:470:142:5::43): Network is unreachable (101) <nckx>And the same for IPv4. Well that ain't good. <nckx><I have lived> Oh, that's because I was typing in nckx rather than English. I would have written the same on a realboard. <iK0u>lechner maybe exposing wayland socket would be simpler? <nckx>lechner: Thanks. Maybe our IP whitelisting got broken/lost/cancelled. <iK0u>maybe he meant "I will take a look" <lechner>2023-12-06 12:01:00 3798 Rsync debbugs data: rsync: [Receiver] failed to connect to debbugs.gnu.org (209.51.188.43): Connection refused (111) <nckx>Your host doesn't try both? <gabber>is it possible to declare in a package (to be deployed on aarch64-linux-gnu) to use python 32bit (aka for system armhf-linux)? <nckx>lechner: My battery's about to die but I'll discuss this with the admins later. <lechner>nckx / i'd be happy to raise it with ian if you like <abralek>Has anyone built a Docker container that emulates the FHS? <lechner>nckx / i remembered now: i have to use IPv4 because my ISP supposedly does not assign static IPv6 prefixes, based on the peering dispute. (i never tested the stability of my prefix.) <lechner>nckx / and just in case that didn't make any sense, my ISP operates out of HE2 in Fremont <lechner>nckx / so i couldn't give FSF an IPv6 to whitelist <pastor>is there a way to get a 'message-id' in debbugs? <lilyp>debbugs is built on mail – fetch the mail, you have its ID <lilyp>if you mean "search by message ID", I'm pretty sure frontends understand those <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, efraim says: I'm building glibc-locales on powerpc-linux now <lechner>Hi, I find it super painful to use 'guix shell' in Eshell. Is there a better way? <abralek>ok, I found that nscd that runs inside the docker with LD_LIBRARY_PATH for its name-services gets things like this /gnu/store/sp2c47k890yi61b5nryd58jhc7q72mgh-nss-pam-ldapd-0.9.12/lib/libnss_ldap.so.2: error: symbol lookup error: undefined symbol: _nss_ldap_endsgent (fatal) <abralek>And because of that getent doesn't receive anything back <lechner>pastor / thanks, i used to use EAT separately but a recent update prevents it from starting if the module is already running as part of an Eshell process. Do you have readline support in 'guix shell'? <abralek>dammit, I get the same on the main system( <lechner>abralek / did you upgrade without rebooting? <pastor>lechner: If you mean auto complete and command history, yes I do. I can do everything you can in the shell (bash in my case). <lechner>pastor / you literally just run 'guix shell' in Eshell? Which shell do you get? <pastor>lilyp: I don't know how to find where debbugs is storing the local message to see it raw. <lechner>pastor / you can get the mbox, i think, but it's cumbersome <pastor>lechner: a bash shell provided by `guix shell` with the correct guix environment <abralek>lechner: no, I have just double checked another server <pastor>lechner: I see. What I did is use `gnus-summary-reply' to grab the `message-id' from there <lechner>pastor / you use guix system and guix home? <abralek>What I am basically doing is ps waux | grep nscd, check pid, its environment path and config. and then run it manually as root with LD_DEBUG=libs <pastor>lechner: I use guix system but I do not use guix home. I manage everything on my own through guix profiles and `.profile`. <pastor>lechner: do you want to check out my `.dotfiles`? <lechner>pastor / yes, please---although my bash config is very basic <lechner>abralek / maybe that also applies to containers that are out of sync with the base system <pastor>lechner: take a look at `.profile`, `.guix-environment` and `.bashrc` <lechner>i have literally used it since the day it came out. maybe it gained some new config options and needs <pastor>lechner: It is probably important to enable something since in my guix shell I get this prompt: <pastor>pastor@bordercollie ~/.dotfiles [env] <pastor>λ echo $EAT_SHELL_INTEGRATION_DIR <pastor>lechner: this is inside the `guix shell` <pastor>lechner: does it work for you If you set that variable? <lechner>pastor / i am not sure it's setting it, but rather using it to load something? <pastor>lechner: Yeah. I believe is loading something <lechner>pastor / i'm reconfiguring my guix home, which due to my custom eudev will take about three hours (after a guix pull) <pastor>lechner: maybe is the terminfo that it compiles <lechner>i can probably shortcut if you can give me your output for echo $EAT_SHELL_INTEGRATION_DIR <pastor>lechner: a few messages above you have it <abralek>lechner: I see, well according to ldd nslcd nscd and getent they glibc is the same ( <lechner>i only see the actual prompt line with the lambda, but no output <pastor>lechner: I see. IRC confused the / from the path to be a command. One sec <pastor>λ lechner: echo $EAT_SHELL_INTEGRATION_DIR\n/home/pastor/.emacs.d/straight/build/eat/integration <lechner>okay, great! i don't use straight so will look <lechner>abralek / can you find the missing symbol with 'nm'? an online search for it came up empty <pastor>lechner: maybe take a look at the variable `eat-term-terminfo-directory' <lechner>pastor / you do not use the official emacs-eat package from guix either, right? <pastor>lechner: I'm telling straight to grab that file specifically since we need it to compile the terminfo database <lechner>pastor / that was true magic! now it's working. I found the integration folder but just looked at it when TAB started to complete <lechner>maddening, in some unimportant way... <lechner>pastor / thanks so much for your patience and help today! <pastor>lechner: did you enabled `eat-eshell-mode'? <pastor>lechner: I can toggle it and it works as expected <lechner>pastor / have you ever used guix environment in hope of staying inside Eshell? <abralek>lechner: that symbol is in libnss_ldap.so.2 that is not the only one, basically all the symbols that nscd needs from nss-pam-ldapd (libnss_ldap.so.2) are shown as fatal same for libnss_mdns.so.2 <pastor>lechner: no. I can use `guix shell` just fine within `eshell` :). Although before having eat I was using vterm <lechner>pastor / but guix shell switches you to Bash, right? <lechner>pastor / i used to use the similar buffer-env.el but those things also get loaded when I access the folder with Magit! <lechner>pastor / dunno, but i often change a lot of basic environment variables like LD_LIBRARY_PATH. it seemed like an overreach <sarg>hey guix, my system doesn't boot after latest reconfigure. It mounts file systems and then does nothing. Any tips on how to increase verbosity of the boot process? I think one of the activation scripts misbehaves, though I have no idea which. Also, I haven't done much changes, it was just kernel version bump and removal of the docker service. <lechner>sarg / what are your current kernel boot options? you can see them with 'e' on the relevant entry <sarg>nckx, it seems that channel logs are not indexed, the search returns only matches from some old days <sarg>lechner yeah, I have quiet. Is it honored by guile's boot script? <lechner>sarg / okay, that I don't know. is your issue before or after the pivot_root in the initrd? <sarg>lechner: yeah, removing quiet produces quite a big output. It seems that elogind filesystem can't be mounted <sarg>though there is no scrollback in the tty nowadays and the boot log isn't preserved <pastor>lechner: this looks very interesting. I guess we cannot use it to substitute `wrap-program' since most executables are not interpreted and do not have shebang to modify <sarg>okay, mystery solved. I played with lxd-service-type and made a typo in file system type. This fs couldn't be mounted and shepherd couldn't reach the root-file-system target <sarg>removing the patch from my guix checkout and reconfiguring fixed boot <sarg>I'm playing with a MIPS SBC. The cpu has MIPS1 ISA and I want to try building software for it with guix. So I've added a new platform `mips-linux` and I'm trying now `guix build hello --target=mips-linux-gnu`. It builds expectedly a cross toolchain and then the `hello` derivation. What I don't get is how do I instruct MIPS1 ISA for building `hello`? It should be CFLAGS=-mips1, so should I create a child package with this flag added? <ieure>Hmmm. So I have a package for LibreWolf, which builds using their source tarball. But since LW is a fork of Firefox, they have a build process to *make* the source tarball, which involves fetching the Firefox source tarball and applying patches to it. <ieure>Just for my own curiosity / to learn things better, I want to move *that* build process into Guix -- the creation of the source tarball ahead of building it. <ieure>How would I build something that has *two* sources -- the Firefox tarball and the Git repo containing the patches + build process to transform it into the LibreWolf source tarball? <ieure>Actually, I believe it's more than two sources -- the LW stuff is in multiple repos, and I believe their build process clones/downloads from several of those. The main LW source repo is pretty much just the orchestration of all the other parts, each of which is its own source. <ieure>Can I do something like define a package for each of the sources (but not have any build process) and use the source of that as an input to another package? <ieure>Or would I need to have multiple packages which get the various source artifacts and combine them into one thing? Like one package that just repacks the Firefox source tarball, then a second package with a source of the LibreWolf settings repo and takes the output of the FF source as input, and builds a combined source tarball. <ieure>And chain together all the stuff I need that way? <sarg>you can have your patches repo defined as an anonymous package in `inputs` <ieure>sarg, That sounds very promising, is there some prior art I can look at which does that? <ieure>Okay, thank you. I'd definitely have been confused without that, since it adds the source in a build step, not in the package inputs. <sarg>also https:/git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/base.scm?h=06f25a9a85be1bbe7a709e58ce41c1a834e5f1ae#n1625 <sarg>inputs take, quoting the manual, either a package, origin, or other “file-like object” <ieure>Gotcha. Thank you very much, this is very helpful. I appreciate it. <sarg>hmm, how do I refer to the cross toolchain for some platform? I want to open a guix shell with it <civodul>hey! i installed passff-icecat, it shows up in about:addons, but there’s no passff icon in the toolbar <apteryx>is the rsync job not working for mumi:? <nckx>(It was discussed here earlier today.) <nckx>ACTION hopes the FSF sysadmins work week-ends. <nckx>apteryx: Oh, it's running now, should be synced soon. Their (gnu.org) rsync daemon dieded and had to be manually restarted. <nckx>Ping me if it doesn't sync in a few minutes.