***rekado_ is now known as rekado
<m4rk>icedove is the same rebrand made by Debian team? <EMax`0Mancer[m]>for a couple of things I've installed from Nix (dead-notification-centre, gargoyle), I get a weird error when trying to run: `/nix/store/ayrsyv7npr0lcbann4k9lxr19x813f0z-glibc-2.34-115/lib/libpthread.so.0: version `GLIBC_PRIVATE' not found (required by /gnu/store/5h2w4qi9hk1qzzgi1w83220ydslinr4s-glibc-2.33/lib/librt.so.1)` <nouun>I'm trying to package something and git-download doesn't recurse submodules? Is there something else I need to do to clone submodules? <vivien>I have a few git repositories cloned on my system, such as guix and others. I have set up a cron job to fetch them every hour. Today I discovered that the new git won’t do it if the directory is owned by someone else than root. <vivien>If you have a similar setup, check your mcron log. <vivien>This is very annoying, but I guess it’s that CVE they discovered on multi-user systems. <guixsd-n00b>so, I'm trying to patch Telegram desktop binary, and I get a weird error of runpath-validation <lilyp>vivien: solution: run mcron as user :) <dust_>Can anyone take a look at 53730, 53683 - 53692 series of new packages that I subimitted. Its of Haskell and its been months. <tex_milan>Hello, how to use guix home configuration to setup Inkscape extension? I'd like to call `git checkout url_to_some_extension` in ~/.config/Inkscape/extension folder... <lilyp>tex_milan: I think we ought to add a search path to inkscape, so that extensions can be packaged as normal packages <f1refly>how how can I define the kernel cmdline arguments? I searched the manual, but there's no mention of possible arguments in the bootloader section kernel options returns nothing but how to configure it with sysctl.. <PapuaHardyNet>Well, I have bypassed this issue by using the shell installer. <jpoiret>PapuaHardyNet: nice to see you got it working <jpoiret>hopefully, the latest installer (i think you're using the 1.3 one) handles this error better (or has it fixed even) <jpoiret>guixsd-n00b: telegram desktop seems to be GPLv3 and is c/c++ with cmake, it shouldn't be too hard to build it ourselves <jpoiret>by the way, runpath-validation checks that the needed libraries would be found by ld <jpoiret>so if that step fails, most likely you haven't patched the runpath to refer to the proper dependencies <PapuaHardyNet>Still haven't lol. I think I somehow killed my bootable USB too. Time for another try with another usb <lilyp>jpoiret: It may be C/C++, but the vendoring is strong ***tex_milan1 is now known as tex_milan
<ss2>Is there a way to have guix offload being more talkative? Connections are currently being refused on a host, but have log or anything to look out for. <PapuaHardyNet>guix system reconfigure (first install) throws a grub install error: "failed to get canonical path of 'none'." I tried using "/dev/sda" as target as "/dev/sda2" (partition with 1MB space) as target but this error kept occurring. <jpoiret>and, that means that the initial install succeeded (using `guix system init`), and now you're trying to reconfigure from the live system? <PapuaHardyNet>jpoiret: nope, I just created three partitions and mounted the first (the other is 1MB for boot and third is stub). Then i wrote a `config.scm` and tried a `guix system reconfigure config.scm` <jpoiret>oh no, for an initial install you should use `guix system init`` <jpoiret>well, that could explain that the bootable usb got messed up previously :p <jpoiret>guix system reconfigure is used to reconfigure a live system <jpoiret>so you ended up modifying parts of the bootable usb most likely <PapuaHardyNet>You are right, I actually missed that aprt mentioning the use of `system init` in the manual hahah <jpoiret>guix system init takes an additional parameter, the mount point of the future root that you want to install on <jpoiret>let's try that first and see if that solves your issue <PapuaHardyNet>jpoiret: thank you, the installation went smoothly after that! <PapuaHardyNet>BTW, I must write my LUKS password once in qwerty before GRUB starts and then in colemak after GRUB selection. How do I solve this? <PapuaHardyNet>okay, so `guix pull` doesn't work for me after I reboot into the installed system because I don't have internet connection, and I can't add internet connection without `wpa_supplicant` which isn't installed... but I do have a LAN port. I'll try that <rekado>apteryx: /dev/sd{r,q,t,s} are four paths to the 100TB volume on the SAN <pashencija[m]>I'm trying to get aarch64 rootfs on aarch64 host with guix system init <pashencija[m]>- 'build' phasebuilder for `/gnu/store/p68swbk18vikfl45nvvzhcpahnan6cq7-gcc-cross-sans-libc-aarch64-linux-gnu-10.3.0.drv' failed with exit code 1 <pashencija[m]>I can run `guix system init` successfully with exactly the same config and a cross-compiler on x86 host <pashencija[m]> * I'm trying to get aarch64 rootfs on aarch64 host with guix system init <pashencija[m]>I can run `guix system init` successfully with exactly the same config and a cross-compiler on x86 host <roptat>cross-builds are different from native builds, so if one succeeds the other can still fail <roptat>but this looks like a very important package, so I wonder how we could have let it be broken... which version are you using? <roptat>oh, wait, you're building an aarch64 cross-compiler on aarch64? that doesn't sound right <roptat>also your gist seems to be incomplete <roptat>could you paste just the last say 50 lines of the build log? <pashencija[m]>This file has been truncated, but you can view the full file. <roptat>what does your system config look like? <superkamiguru>also tried adding hid_apple.fnmode=2 to the kernel arguments but that didn't work either. <roptat>superkamiguru, sounds like the right place to ask, but I'm the wrong person to answer, sorry :/ <pashencija[m]>roptat: Nothing special. Just basic system + a few custom receipts to support my hardware <roptat>how do you run "guix system init" exactly (more arguments?)? <pashencija[m]>guix system init --target=aarch64-linux-gnu --skip-checks ./test.scm ./rootfs <roptat>then drop --target=aarch64-linux-gnu <roptat>without a --target, it builds for the native architecture, which is aarch64 in your case <superkamiguru>roptat: no problem. Just thought I would throw it out there. It seems the link I provided is for loading modules that don't auto load, which the hid_apple one does. <pashencija[m]>Doen't it detect target architecture is the same as host? O_o <roptat>nope, if you ask for a cross-compilation it'll cross-compile <roptat>superkamiguru, in these cases I try to understand how it's done on traditional distros and try to figure out if there's something in Guix that implements that <superkamiguru>Yeah, usually you add "options hid_apple fnmode=2" to /etc/modprobe.d/hid_apple.conf and then run the kernel init command. <roptat>does that set the kernel command line, maybe? like in grub? <roptat>maybe you can try to set these options in your bootloader <bjc>you can see if those options are set in your running kernel by looking /sys/modules/hid_apple/parameters <superkamiguru>I tried adding hid_apple.fnmode=2 to the kernel command line but that didn't fix it either <bjc>i believe you should be able to ‘cat /sys/modules/hid_apple/parameters/fn_mode’ to get its current value <superkamiguru>I checked the value and it's still reporting 1 after setting hid_apple.fnmode=2 in the kernel command line. <bjc>2 may not be valid for some reason, or maybe the option isn't being found <bjc>you can ‘modprobe -r hid_apple’ to remove it, then try an explicit ‘modprobe hid_apple fnmode=2’ <superkamiguru>If I edit /sys/module/hid_apple/parameters/fnmode and change it to 2 the keys swap correctly <bjc>i'm not super-sure the latter command line is correct, but it's in the ballprak <bjc>ok, so the issue is that the parameter isn't being found on boot <bjc>have you made a file in /etc/modprobe.d with the text ‘options hid_apple fnmode=2’? <bjc>if so, then try to just ‘modprobe -r hid_apple && mod_probe hid_apple’ to see if its being picked up like it should <bjc>oh, right, of course you can't edit /etc stuff <superkamiguru>On reboot I have a hid_apple.conf in my etc/modprobe.d with "options hid_apple fnmode=2" <superkamiguru>But it doesn't load it for some reason. I imagine it is because I dont have the ' (kernel-loadable-modules (list ddcci-driver-linux))' section mentioned in that link <bjc>is that file still there? and if so, and you manually unload the module and reload it, does the parameter stick? <bjc>i can't say for sure, but i believe if the module loads at all it should be reading that file and setting options <superkamiguru>Yeah if I follow the steps in the link and add the conf definition and services the /etc/modprobe.d/hid_apple.conf persists on reboot. If I unload the hid_apple module and reload it with the commands you provided, the keys swap correctly <superkamiguru>Just seems like its not loading the /etc/modprobe.d/hid_apple.conf file <bjc>maybe it's an initrd thing? <roptat>maybe the /etc file is created too late? <bjc>have you tried adding ‘hid_apple’ to the (initrd-modules) list? <superkamiguru>Like the services that define it? should I maybe add that above the gnome desktop service? <roptat>I don't think the order in which you declare services matters <bjc>it shouldn't matter, no <bjc>if guix is creating the /etc link at boot, then it might be being created too late, but i have no idea how that stuff works <superkamiguru>According to that link that procedure is for loading driver modules that don't auto load at start, which the hid_apple driver does. So I wonder if there is another way you are supposed to edit autoloaded modules <roptat>I would have guessed kernel-arguments <bjc>it's a long shot, but i was hoping that by adding ‘hid_apple’ to the initrd-modules, it'd also pull in the modprobe.d config, but i don't think it actually does <superkamiguru>I think kernel arguments might work, if I can figure out what package provides the hid_apple driver. Like in that link it references ddci-driver-linux for providing the driver. <bjc>fwiw, %base-initrd-modules doesn't include ‘hid_apple’, so it's being loaded post-boot <bjc>well, that or it's being compiled into the kernel <superkamiguru>would it be referenced by an identifier other than hid_apple possibly? <bjc>no, i don't think so <superkamiguru>I found this old mail list for guix from 2016 that mentioned the /run/modprobe.d/ folder is scanned like the /etc/modprobe.d/ folder on debian. The person created a /run/modprobe.d/{file_name}.conf file and said it worked on reboot, but that didn't work for me. <jpoiret>bjc: if it's compiled in, you wouldn't be able to modprobe it right? <bjc>i think you're right <jpoiret>it's possible that the module is just loaded automagically at initrd time, before /etc/ exists <bjc>it's not copied into the initrd, though <bjc>hmm. unless it's a dependency of one of the listed modules <bjc>it's in %base-initrd-modules <bjc>so yeah, it's being loaded at initrd time, and there's no config for it set up on the ramdisk <bjc>there's also no facility to do that in guix <bjc>you can work around it with a one-shot shepherd service <bjc>all it'd have to do is ‘echo 2 > /sys/modules/hid_apple/parameters/fnmode’ <superkamiguru>Would you happen to happen to have a good starting place for someone who hasn't messed with shepherd services? <bjc>i don't. i've barely looked at shepherd, but maybe someone else here is more knowledgable <roptat>pashencija[m], looks like it got further at least :) <roptat>I think you're not using substitutes, is that on purpose? <faust>i gust run command guix refresh -u emacs-next and got error guix refresh -u emacs-next <roptat>also, are you on guix 1.3.0 (stable) or did you run guix pull recently? <roptat>guix refresh is for guix developpers, did you mean "guix package -u emacs-next"? <faust>i run guix package -u emacs-next but it doest update to 28.1 <bjc>if you already have emacs installed, it's probably not pulling emacs-next from your path first <roptat>yeah, looks like that's the latest version in guix <bjc>i'd suggest removing emacs, then installing emacs-next <roptat>so, "guix refresh -u emacs-next" would update the package in the guix repository, but if you run it like that, it'll try to update the read-only copy in the store. It's supposed to be used by developpers when they're working on the repo, like "./pre-inst-env guix refresh -u emacs-next" <bjc>oh, sorry. i misread. i thought you were on 27 <roptat>if you want to update to the latest emacs version, you could try the package transformation option: "guix install emacs-next --with-latest=emacs-next" <ulfvonbelow>is there an easy way to add a kernel command line parameter in my config.scm? <roptat>--with-latest will figure out which is the latest release and use the recipe for 28.0.50 with it <roptat>ulfvonbelow, look for "kernel-arguments" in the manual :) <roptat>that's a field of the operating-system record <ulfvonbelow>ah, thanks, I was C-s'ing for 'kernel command', and it missed 'command-line of the kernel' <faust>roptat thaks! guix install emacs-next --with-latest=emacs-next works! <djeis>Is it absolutely nuts to have a volatile-root guix system whose root fs is actually a bind-mount of a computed-file (which would, presumably, be copied in to the initramfs)? <superkamiguru>So I need to create a service that runs a command once. Not sure if it should be a system service or a user shepherd service <djeis>Does it need root power, and when do you want it to run? <superkamiguru>I tried creating a service definition in ~/.config/shepherd/init.scm but I don't think it runs because it needs root <djeis>Yeah, if you need root power it would almost certianly need to be a system shepherd service. <superkamiguru>I am trying to get it to run once on boot. Basically using it to replace a /etc/modprobe.d/ configuration change <superkamiguru>Just need to run the command 'echo 2 > /sys/module/hid_apple/parameters/fnmode' once to swap my fn keys <djeis>And this is a full guix system box? <superkamiguru>I was working with a couple people on here earlier to try and change the parameter with a conf defined in the config.scm file, but something about the driver being loaded in initrd modules meant it wasn't possible to configure it on boot <djeis>Add a simple-service targetting the shepherd-root-service-type to your system config, the value should be a list with your shepherd-service def. Lemme hunt down the relevant part of the docs... <superkamiguru>Ok. I have gone through so many of the service docs now I am not even sure what applies haha <superkamiguru>I saw an example of someone elses config during my searching where they defined a simple service <superkamiguru>Just wasn't sure how to modify it to do what I was trying to do <djeis>Essentially it's (simple-service 'pick-a-name shepherd-root-service-type (list (shepherd-service ...))) <djeis>You're making an inline guix system service that extends the shepherd-root-service-type with some additional sheperd-services. <djeis>Well, probably only one additional shepherd-service, in your case. <djeis>Are you familiar with the gexp stuff that comes with defining a shepherd-service in a guix config? <superkamiguru>No. I am honestly just hacking away at a copied shepherd service config inside the example simple service you sent haha <djeis>For your specific case, it's probably enough just to wrap the start expression in #~ <djeis>Ah, no, not quite: the start needs to be a guile scheme function, not just a string. However, the provide a convenience constructor for a start that just runs a command. <djeis>I believe it's make-system-constructor... <djeis>make-forkexec-constructor would do, but you need to be a little more careful with it and it wouldn't support all of the shell syntax you're using in that command. <djeis>make-system-constructor takes a shell command and doesn't consider the service as "started" until the command exits. <djeis>That's stripped down from an example in my own config. <djeis>(which, for the moment, is private) <djeis>Ah, no, you don't want a stop kill destructor b/c there is no long running daemon associated with this shepherd service. <djeis>That's why I have one-shot? #t in my example. <ss2>has anyone got a recent checkout with offloading still working? <superkamiguru>hmmm now I am getting shepherd-root-service-type: unbound variable <djeis>Ah, you need to use (gnu services shepherd) <superkamiguru>I knew I was missing something. Thank you again haha. I just want my media and fn keys swapped! <superkamiguru>months have gone into this. You don't understand what this means <guixsd-n00b>I've cloned guix repo and every time I get a 'unbound variable' y try 'find <path-to-guix> -iname <unbound-symbol>' it works more often than not <guixsd-n00b>you can clone it shallow, if you care about hd space <guixsd-n00b>guix can tell sometimes where a symbol is defined, maybe it can be made it work for services (same with 'guix edit') <djeis>superkamiguru: Glad to hear it! <roptat>PotentialUser-50, guix.gnu.org is on a different server and it seems the network on which it is, is rejecting russian IPs, nothing we can do... ***guixsd-n00b is now known as warpedliver
<warpedliver>Hi! how can I configure the timezone for a container, with guix shell/environment ? <lilyp>warpedliver: have you tried adding tz-data to the container? <lilyp>the environment variable to use is TZDIR btw <warpedliver>yes... mmm it didn't work. Added tzdata and preserved TZDIR env <warpedliver>can anyone confirm that this doesn't work? 'guix shell -C coreutils bash tzdata -E TZDIR --expose=${TZDIR} -- date' <roptat>right, doesn't work for me either <warpedliver>(plp away from greenwich meridian would be preferable xD) <roptat>we probably don't understand how date knows about the current timezone <warpedliver>or any other application, I was using date, for debugging; i'm trying to use an app inside a container <warpedliver>but date get's messed up (and can put accents so maybe keyboard config also is messed up, not shure) <warpedliver>time to go 4 a walk, thanks for the help, see ya later <djeis>I appear to have gotten myself into a very odd corner case involving guix system and shared store virtual machines: I get "too many levels of symbolic links" errors all over the place, in ways that don't really make any sense to me. <djeis>Happens during the boot of the virtual machine, and if I boot to guile I can reproduce the errors just poking around the file system. <djeis>Seems to happen whenever I try to read a symlink from the shared store. <djeis>Even just trying to readlink a symlink in the store triggers it. <lilyp>sneek later tell warpedliver TZ=$(cat /etc/timezone) guix shell -C coreutils bash tzdata -E TZDIR --expose=${TZDIR} -E TZ -- date <f1refly>how can I edit the kernel cmdline? I checked the reference manual for the bootloader and searched around a bit, but I didn't find any references. <ss2>f1refly: (kernel-arguments (cons* whateveryouwant=bar %default-kernel-arguments)) <ss2>place that into your (operating-system ...) <ss2>sry, it should be: (kernel-arguments (cons* "whateveryouwant=bar" <ss2> %default-kernel-arguments)). They need to be strings. <char[m]>Is there a way to build guix without documentation? It is giving me strang errors, and take too long. <djeis>Ah, it was because the store wasn't mounted in passthrough mode. *civodul is listening to it <civodul>my Saturday evening was hijacked in a pleasant way :-) ***guixsd-n00b is now known as warpedliver
<warpedliver>btw, 'guix shell -C coreutils bash tzdata -E TZDIR --expose=${TZDIR} --expose=/etc/localtime -- date' works, it needs /etc/localtime, not only TZDIR <sneek>warpedliver, you have 1 message! <sneek>warpedliver, lilyp says: TZ=$(cat /etc/timezone) guix shell -C coreutils bash tzdata -E TZDIR --expose=${TZDIR} -E TZ -- date <maximed>antioxidant-build-system update: 'agate' builds. Inputs aren't propagated anymore (at least on the package level -- there is a form of propagation inside the build code). More robust handling of ‘native’ libraries. Support for crate renaming. More implied features. <maximed>(i.e., hexyl, sniffglue, agate and all their dependencies) <maximed>Should be in sufficiently good shape now to be tested by other people <maximed>Though you'll have to add missing dependencies to packages, sometimes resolve version conflicts, enable features ... <fishinthecalcula>Good evening guixers, is there any way on the Guix System to disable a substitute server without reconfiguring the system? I have a malformed substitute server in my Guix daemon, hence guix fails to substitute all packages. the only way i found is to reconfigure with substitutes disabled but it's taking quite a while... <podiki[m]>I got the right time with -E TZ (but I may have that set manually on my system) 