<rekado>another reason why the pandoc closure is huge is due to the fact that the generated conf files hold lists of build time libraries and tools (such as glibc-static, gcc, etc) <rekado>I’m currently working on moving all the static stuff to a separate output, but there’s still lots of potential ***gf is now known as gf_
***gavlee_ is now known as gavlee
***sneek_ is now known as sneek
***sneek_ is now known as sneek
***jonsger1 is now known as jonsger
<Blackbeard>I upgraded parabola and grub and the kernel were updated and now I can't boot Guix <Blackbeard>I was unable to boot guix or parabola, but I was able to rescue parábola. But Guix I don't know how to fix it <bdju>.tell vagrantc I did get Sway working. I still have to get a gtk theme working and a few other things now, though. <bdju>sneek: later tell vagrantc I did get Sway working. I still have to get a gtk theme working and a few other things now, though. <fvr>guix reconfigure currently failys for me while building ruby-sass-spec-3.5.4, with this error <fvr>Throw to key `encoding-error' with args `("scm_to_stringn" "cannot convert wide string to output locale" 84 #f #f)'. <fvr>reconfigure gives a warning initially that it failed to install locale, even after installing glibc-locales as the hint suggests <lfam>fvr: Can you give the output of `guix describe`? <fvr>So is there a way to enter line breaks in here? Or do I have to use pastebin? <lfam>Thanks, I'm trying to reproduce now <lfam>Hm, I'm not sure. I didn't have to build that package. Instead I got a substitute. I wonder if the error is really in that package build? <lfam>Are you able to put the error messages in a paste? <lfam>Okay. I'm not sure exactly what's wrong but I think it's one of the rare cases where the locales not being set up right actually breaks something <lfam>On Guix System, you don't need to install locales at all. Instead, in config.scm, you use the locales field to configure them <lfam>When updating to new versions of glibc (this happened recently), it can happen that the system is using one version of glibc, but packages installed by users are using another version <fvr>My locale is `(locale "en_US.utf8")` in /etc/config.scm <lfam>Do you know if the system is more up to date than the user packages? Or vice versa? <fvr>I just installed this system yesterday and this is the only command I ran after guix pull <lfam>Reconfigure is the only command you ran after `guix pull`? <lfam>And did you run `guix pull` as your user or as root? <fvr>And the first time it failed, ran the command to install the locales and set the environment variable and repeat <fvr>as my user, non root <lfam>So I guess your command was something like `sudo -E guix system reconfigure /etc/config.scm`? <lfam>Is anybody able to help fvr work around this? They are on Guix System with locales mismatched between the system and their user, and it's breaking their ability to reconfigure the system <lfam>I have to go in a minute <lfam>fvr: Since you didn't update the root user yet, I would try logging in as root with `sudo --login` and trying to reconfigure again from there. And I recommend adding the locale-libcs thing too, until all the users and the system have updated <lfam>You could also rollback your user with `guix package -p ~/.config/guix/current --roll-back`, logging in again with e.g. `bash --login`, and then trying to reconfigure again. After you roll back and re-login, you'll be using the same locales as the system and things should work <fvr>I see, so login as root, guix pull, install locale, reconfigure and try to do it again in normal user <lfam>It's the root of the problem in this case <lfam>Hm... well, it's a bit of a corner case <lfam>I need to think for a minute while dinner cooks <fvr>Alright! I'm currently doing a rollback and reconfiguring in this user <lfam>Also rollback your packages, with `guix package --roll-back` <fvr>Okay, configuring in root I can try at a later time? <lfam>Normally, it's important to `guix pull` at least once before reconfiguring the first time <lfam>Let me give you a specific commit to pass like `guix pull --commit=asdfdsf` <lfam>Alright, I would use commit c81457a5883ea43950eb2ecdcbb58a5b144bcd11, which is the last commit before we updated glibc <lfam>You can pull that commit, add the locale-libcs thing, reconfigure, and then you should be good <lfam>Sorry this turned out to be so complicated. I think that I am either being dense or this is a weird corner case bug that we should try to deal with <lfam>Usually, if the locales aren't set up right, those warnings are just warnings. Annoying but harmless. But this is a case where the lack of proper UTF-8 locales actually broke something <lfam>I didn't realize that you'd left and sent a bunch of messages in the meanwhile <fvr>I am reading the logs <fvr>I am trying them out now <lfam>Alright, I have to go now. If you need more help, someone will pipe up here after a while <lfam>Or send mail to <help-guix@gnu.org> <fvr>After running guix pull with that commit, guix is currently pointing to something in /gnu/store, where as the guix in ~/.config/guix/current/bin/guix is pointing to guix-command in /gnu/store. Both display the same versioin when I ask for --version <fvr>But as the hint they should both be pointing to the same thing <fvr>Okay, the guix is also a symlink to guix-command <fvr>but shouldn't it be pointing to the one in config/guix/current <fvr>lfam: It successfully reconfigured now. But it didn't have that ruby-spec derivation that was causing trouble previously <lfam>That's more or less expected. Since you were working from a different commit, the work done by reconfiguring would also be different <lfam>But in the future, it should also handle that thing without crashing, since it will have the correct locales in place <fvr>yeah thought so. So will I have the same trouble if I git pull from master now? <lfam>The locale-libcs field should handle it <lfam>It will make both the old and new versions of the locales available <fvr>Okay, Thanks for your help <fvr>Getting the same error again, lol <lfam>Either way for advice here or send a bug report to <bug-guix@gnu.org> <marusich>Just wanted to say, I love the News feature of Guix. The fact that you can use it in channels is really nice. So many times, I get a new version of software, and it's so hard to find a good description of the big changes. <marusich>I need a better machine. This x200 can barely keep up. <marusich>Every time I build stuff with Guix, while trying to do productive things, it ends up swapping, and then it just feels so slow. <marusich>It can take 500-1000 ms to just alt-tab... <marusich>Speaking of x200s and Guix, though, has anybody using an x200 here noticed problems switching monitors? Mine used to work just fine, but nowadays, when I plug in the VGA port, it never quite switches to use the external monitor. It just flickers and fails. I tried half a dozen distros, and only Fedora worked. I think it's becuase they use Wayland. <marusich>I'd use Guix System on this x200, if only I could use an external monitor. Using Xorg on Guix System, it flickers and fails :/ <apteryx>marusich: you might want to run the earlyoom daemon. It prevents swapping by killing memory hungry processes <apteryx>saves me hard reboots. I never have the patience to wait on the Linux OOM killer. <apteryx>did you max your x200 with 8 GiB of RAM? <marusich>apteryx, it has 8 GiB of RAM, yeah. I actually don't want the OOM killer to run; I'd rather have a slow system than a system with randomly dying processes. However, the slowness is a huge bummer. <marusich>I think perhaps I should try a lighter window manager. I use GNOME, and I'm sure it's resource heavy. <marusich>The really annoying thing is that once linux-libre starts swapping, it seems like it never cleans up the swap usage - "free -m" always shows swap being used after that. I wonder if I should just ditch my swap, but I don't want random OOM kills. <fvr>I just did a `guix install emacs` and it set EMACSLOADPATH=~/.config/guix/current/share/emacs/site-lisp:... but the lisp and site-lisp directories are actually at ~/.guix-profile/share/emacs/ <rekado_>fvr: did you set GUIX_PROFILE to ~/.config/guix/current? <rekado_>it should be set to ~/.guix-profile instead <rekado_>.config/guix/current is only used for ‘guix pull’ <fvr>Every time I ran some guix command, it was asking me to do that so I stuffed that into .bashrc <fvr>I am mistaken! It did ask me to set at ~/.guix-profile only <fvr>Wonder why I instead set it to ~/.config/guix/current <rekado_>in any case, please don’t set this in .bashrc because you’ll break ‘guix environment’, which spawns sub-shells. <rekado_>.bash_profile is a more appropriate location for sourcing etc/profile <fvr>Any GUIX_PROFILE? Or just the wrong one? <rekado_>the presence of the GUIX_PROFILE variable affects the behaviour of sourcing any profile’s etc/profile file <mroh>marusich: the kernel swaps pages that are not used. its ok that these are not loaded again, because it has more ram to use for e.g. caching, so you should not ditch your swap space. <marusich>OK. Maybe it's just my imagination, but I felt like I got better performance when I told Linux to turn swaps off and then back on...I might have just been deluding myself. <fvr>rekado_: you mentioned that .config/guix/current is only used for pull. Do I have to do anything for that? Because when I pull it gives me a hint like this http://paste.debian.net/1148380/ <rekado_>fvr: this is to ensure that the new ‘guix’ that ‘guix pull’ installs into .config/guix/current/bin appears first in your PATH <rekado_>note that it doesn’t ask you to *export* the GUIX_PROFILE variable <marusich>but rekado_, isn't it acceptable to do something like this to make the "guix" program available? 'GUIX_PROFILE=$HOME/.config/guix/current; . $GUIX_PROFILE/etc/profile' <xelxebar>That sets GUIX_PROFILE in the environment of your running shell, which is different than *exporting* the variable. <fvr>hint: After setting `PATH', run `hash guix' to make sure your shell refers to `/home/fvr//.config/guix/current/bin/guix'. <fvr>What does this hint mean? What PATH should I set? <wxie>fvr: I think there are two kinds of profile. One for system wide profile, the other for current user. Correct me if I am wrong. <fvr>Yeah I didn't think it would be problematic because everytime I pulled it was showing that hint. But when I did an install it showed me a different path to set which I didn't realise <marusich>fvr, I think that as long as you "make the profile available" by sourcing its etc/profile file, you're fine. Just don't export GUIX_PROFILE in the process. <boeg>Any idea why i'm still having trouble with udevil? When I try to mount, it errors saying it was not run "suid root", but the following issue seems to suggest its fixed? https://issues.guix.gnu.org/40922 <boeg>also manually adding udevil as suid doesn't work for me <bdju>some icons in icecat are appearing blank for me. it's an interesting issue. first I noticed emoji didn't show up, so probably a font problem, but also some website favicons show up blank! <bdju>I just copied my guix manifest from an older machine which had an emoji font, so likely something different in my system config is the problem. I'm not using %desktop-services on this machine so I'm still lacking some basic stuff <bdju>my gtk themes don't show up in lxappearance and I have gnome-themes-standard installed, for example <bdju>so, in termite I set my browser to xdg-open. I *feel* like icecat is my default browser, but links clicked from termite open in a new qutebrowser window. is there a way to alter that priority somehow? *rekado_ disables dynamic linking for Haskell packages by default <bdju>Oh, it was way easier than I'd guessed... icecat was not my default browser, and changing that in the preferences fixed it instantly. that's one thing checked off the list <bdju>my timezone is wrong (panel showing wrong time), but it's set correctly in my system config. is there something I can run to make it sync up again? like an ntp thing maybe? <bdju>ah, found something in the manual worth exploring <bdju>I wish there were more examples, it just doesn't quite click with me what to do here. I've added (ntp-service-type), but apparently I have to give it some other info and even reading the manual I don't get it. <rekado_>so you need to do (service ntp-service-type) <bdju>oh, shoot. this sort of thing probably goes over my head because I don't really know scheme. <bdju>that got it not to error now. thank you <bdju>yes!! my clock is already correct now. <rekado_>service types are values that can be passed to ‘service’ <bdju>oh right, it's the older style without the -type on the end that doesn't need "service" before it <rekado_>by wrapping something in parentheses you’re trying to call it as a procedure <bdju>I've got a mix of them and I copied the wrong kind I guess <bdju>I'm not sure what a "procedure" is, by the way <rekado_>but a function is really something that doesn’t have side-effects, whereas procedures can have side-effects <bdju>is side-effects a concept in functional programming? I'm not familiar with that either, although I've done little more than a hello world in any language to be fair <rekado_>an example of a pure function is ‘sin’ <rekado_>it takes a number and returns a number <rekado_>it can’t also do some other work on the side that has an effect on the environment <rekado_>it can’t, for example, turn on your oven <rekado_>why does this distinction even matter? <rekado_>the return value of a pure function can be cached and you wouldn’t be able to tell <rekado_>when you compute sin(0.1) twice does it compute the value twice? <rekado_>or will it return the value from a lookup table? <bdju>interesting. that idea reminds me of the reproducible builds idea <rekado_>since it doesn’t have any ‘effects’ it doesn’t matter how often we actually compute the value <rekado_>you can’t do that with a procedure that has side effects like turning on or off the lights <rekado_>the property that pure functions and reproducible builds share is called ‘referential transparency’ <rekado_>that’s also why the model that Guix (and Nix before it) implement is called ‘functional package management’ <rekado_>we model a package build as a pure function that has inputs and outputs, and whose outputs only depend on its inputs. <rekado_>now, this is merely a model and reality is complicating things somewhat (what about the effects that hardware can have on the build?) <rekado_>but the goal here is referential transparency as well ***calher_ is now known as calher
<rekado_>and that’s why ‘guix build hello’ will eventually return just the location of the hello package output, no matter if it built hello locally, downloaded it from a substitute server, or looked the package up in its own local cache (i.e. /gnu/store) <bdju>So is "service" a procedure then? <bdju>and then the service itself is like an argument or something... <rekado_>yes, service is either a macro or a procedure <rekado_>everything in the first position after the opening parenthesis is either a macro or a procedure <rekado_>(a ‘macro’ is special syntax that can be expanded to other syntax) ***apteryx is now known as Guest93436
***apteryx_ is now known as apteryx
<rekado_>in this case ‘service’ is actually a macro, which will expand to either (make-service the-type the-configuration) or a helper to include the default configuration for the given service type <rekado_>but it doesn’t really matter what it is ***sneek_ is now known as sneek
<bdju>or if anyone knows an equivalent to this: programs.fuse.userAllowOther = true <bdju>I see that various emacs-counsel-* things are packaged, but I don't see a plain "emacs-counsel" package. Is this just included with ivy or something? <bdju>bricewge: well the searchability is much worse... <bdju>How can I search the content for "fuse"? <rekado_>bdju: the manual has a curated index <rekado_>in your info reader hit ‘i’ and then ‘fuse’ <rekado_>unfortunately, it only lists ‘%fuse-control-file-system’ <bdju>well, I can manually edit /etc, but it'd just get overwritten I think. <rekado_>use an extension to the etc-service-type <bdju>I think I'm close, but it's complaining about my parens <bdju>I put one where I thought it might go and then got a backtrace instead of a useful-sounding-error. <bricewge>bdju: (simple-service 'myname etc-service-type myfile) <bdju>bricewge: sorry, I don't understand. do I literally type simple-service? and that looks like maybe it creates a file, but I also have to add a line inside it <bricewge>Yes, replace 'myname by 'fuse-etc (or something) andmyfile with then list of file you were using on line 44 <bricewge>bdju: And use a back-quote "`" instead of a simple quote"'" in '("fuse" <boeg>anybody here use udevil? I cant get it to work, says udevil should be run as setuid root, and hints me to do it manually, but even that doesnt work. There is this issue about it where it seems it has been fixed, but at least it doesnt work for me: https://issues.guix.info/40922 <bdju>bricewge: How close is this? (simple-service 'fuse-etc etc-service-type (list `("fuse" ,(plain-file "fuse.conf" "user_allow_other")))) <bdju>bricewge: (it doesn't work, but I'm very lost here) <rekado_>bdju: can you show us the rest of the config file? <rekado_>looks like you have a list where you shouldn’t *janneke merges another bootloader conflict on wip-hurd-vm <rekado_>you have (services (append (list … … %base-services))) <rekado_>your ‘append’ has only one argument: a nested list <rekado_>you probably meant to write (services (append (list … …) %base-services)) <rekado_>you try to append a list to a non-list <rekado_>‘services’ is a field that expects a plain list <rekado_>you have (append (list … %base-services) a-service) <rekado_>so that’s (append (list … (list …)) a-service) <rekado_>you want: (append (list … here-is-my-simple-service) %base-services) <bricewge>bdju: BTW the name of the file is "fuse.conf", not "fuse" <bdju>believe it or not, I know the name of the file, just not what any of this scheme I'm writing does <rekado_>bdju: you’re still putting %base-services inside of a list <rekado_>your append call still only takes one argument <rekado_>it should be two arguments, each a list <rekado_>one is your manually created list of services, the other is %base-services <bricewge>boeg: (simple-service 'udevil-setuid setuid-program-service-type (list (file-append udevil "/bin/udevil"))) <bricewge>This will create a setuid udevil in /run/setuid-programs. I'm not using it so I don't know if it need other things to work. <bricewge>boeg: raghavgururajan is using it IIRC maybe he can help you with it <bricewge>bdju: Yes counsel.el is included in emacs-ivy as they live in the same repo “tree $(guix build emacs-ivy)” <haha>guix system: error: glibc-bootstrap-system-2.2.5.patch: patch not found <haha>guix package: error: glibc-bootstrap-system-2.2.5.patch: patch not found <haha>Oh, my guix looks damaged, what should I do. <haha>guix system reconfigure /etc/config.scm <rekado_>are you running “guix system” as root? <rekado_>in that case: run “guix describe” as root <haha>I installed it using the Guix System installation CD that I built myself. <rekado_>different users –> likely different Guix variants <haha>guix describe: error: failed to determine origin <haha>hint: Perhaps this `guix' command was not obtained with `guix pull'? Its version string is 1.1.0-3.52b01cb. <rekado_>what happens when you run “guix pull” first? <haha>This is actually a brand new system and I haven't used it yet. <bdju>bricewge: thank you! I'm sorry I wasn't able to figure it out, I still don't even grasp the basics so imagining lists in lists and such wasn't going well <bdju>bricewge: and thanks also for the response to the counsel thing <haha>I will try guix pull first. <bdju>had to add a \n to the config line as well but now the fuse setting worked and my sshfs mount went through <haha>guix pull: error: glibc-bootstrap-system-2.2.5.patch: patch not found <haha>It looks terrible, and guix pull also failed. <bricewge>bdju: No problem. There are some resources at to get more familiar with Guile. <haha>I seem to find a temporary solution, although it is not safe. I manually copied this file to the guix directory, and then it worked. <haha>I think there should be no such problem after the system is updated. <janneke>okay, where has this torbutton gone? *bricewge got spooke by the CVE checker once again <sturm1>has anyone had gnome-boxes working for virtual machines? I'm getting an error about missing virtualization support but I suspect it might not be me <bricewge>Ben Sturmfels: I think cbaines is using it <mbakke>sturm1: does 'grep vmx /proc/cpuinfo' return anything? <bricewge>Ben Sturmfels: I gave it a try, and it seems to work: Igot manjaro started <civodul>reepca: thanks for the guile-fibers patch! <sturm1>thanks mbakke yes I have vmx listed - could be something to do with coreboot. <sturm1>Thanks liberdiko, good to know it works <rekado_>civodul: I just saw that doc/build.scm uses the fat texlive package *bricewge don't know anything about how gnome-boxes work <rekado_>civodul: I think I found the package that was missing, so we can use texlive-union now <rekado_>I wanted to add info.js to our generated manuals; that’s when I noticed <rekado_>haskell update: building without static libs works and packages no longer keep references to build tools <rekado_>I still have one minor problem: doctest no longer finds any modules <rekado_>I’m disabling it for now, just to see if ghc-pandoc can be built and how much smaller it is ***calher_ is now known as calher
<PotentialUser-74>Hi guix. Fresh install. Gnome to heavy do installed evilwm but it's not shown up in the chooser. what am I missing or not doing correctly <mbakke>PotentialUser-74: did you add evilwm to the packages list of your OS configuration? <PotentialUser-74>Also, I did a previous install without gnome. Just enlightenment and openbox but the chooser option did not appear... hence the reinstall with Gnome <mbakke>PotentialUser-74: you need to edit /etc/config.scm and add evilwm to the (packages ...) field and reconfigure for it to show up in the greeter <mbakke>just installing to a user profile won't make a difference <PotentialUser-74>I just ran guix install evilwm and thought it would show up... Ah Thanks. I'll take a look <abralek>It would be awesome to have everything in guile, no? ) <civodul>guile-wm just needs a little bit of love <civodul>it's a very nice piece of code but Someone needs to address the tiny remaining issues <civodul>though there seems to be an impedent mismatch: it exposes key bindings and features that Info/Emacs users are fond of <civodul>except that Info/Emacs users are unlikely to use info.js much, by definition <abralek>civodul: Indeed, can I find those issues somewhere? I am currently can run it within a gnome-session. It requires some dbus activate/deactivate messages to be send. Would it make sense to make and provide an additional session in GDM? Or maybe you have some other thoughts? <fvr>Something has happened to gnome, on booting it's displaying a white screen with the message " A problem has occurred and the system can't recover, contact system administrator". I am using TTY2 now, but good news I was able to pull and reconfigure system from master without the earlier errors due to locale <fvr>Does anyone know how to change the screen refresh rate or resolution from terminal (TTY2). I have gnome as the windowing system installed on my system <abralek>fvr: xrandr should be enough, but some DISPLAY vars needs to be available <fvr>I don't have xrandr available, are there other ways? <abralek>Hm, don't know but ~/.config/monitors.xml <sturm1>liberdiko: thanks, gnome-boxes is working for me now with your config and group libvert <sturm1>liberdiko: btw, is `ix` in guix? I couldn't see it <kamil_>Does someone know how to configure geoclue service using the scheme procedure "geoclue-application name [#:allowed? #t] [#:system? #f] [#:users '()]"? I'm supposed to allow redshift to read my location <mbakke>kamil_: if I read the guix manual correctly, geoclue-application only works for applications that ship .desktop files, which redshift does not currently seem to do <mbakke>kamil_: does manjaro ship a .desktop file for redshift? <kamil_>mbakke, well, if to trust that Manjaro's redshift is similar to Arch Linux's, then yes.. >.< <sturm1>When I run choqok (social media app) it tells me "Cannot open KDE Wallet manager". Is installing "kwallet" as a system package likely what I need? <kamil_>mbakke, this is what I have in my "services" (geoclue-application "redshift" [newline] #:allowed? #t [newline] #:system? #f [newline] #:users '("1000")), and this is what I'm getting: "wrong type argument in position 1 (expecting struct): "[redshift ]\nallowed=true\nsystem=false\nusers=1000\n" <mbakke>kamil_: hmm, seems like the geoclue service and documentation needs some fixing :-/ <sturm1>Woohoo, installing kwallet in my system config has fixed both owncloud-client and choqok! <mbakke>it would be great to update the geoclue service API to be similar to other services :/ <kamil_>mbakke, thank you. funnily enough, I did something similar, and guix reconfigured the system just fine. If it doesn't work, I'll try to copy your suggestion <mbakke>I suspect the geoclue service actually needs that .desktop file though ***jonsger1 is now known as jonsger
<kamil_>mbakke, You're right. Guix's redshift is missing the file despite the fact that it should be included by the upstream since they distribute those. Is it possible that Guix strips them during build stage? Anyway, thanks. I'm at the same point where I was with (geoclue-service) alone, but if there was a .desktop file included, I bet it'd be working just fine now. <boeg>Should one not be able to install firefox addons in icecat? <cbaines>No, I'd expect addons for Firefox to work with Icecat <mbakke>kamil_: I updated redshift on the master branch to include a .desktop file. <boeg>does anyone here have "Firefox Multi-Account Containers" working in icecat? It says the addon looks corrupt. I can easily install other addons. <bdju>I'm having some ssh issues where some machines still ask for a password after copying my key over from the guix machine. I made a key with a non-default name for the first time. any ideas? heard something about an ssh agent. when I try ssh-add commands it says can't open a connection to my authentication agent <cbaines>bdju, if you tell SSH to use the key that it should be using, does the connection work without a password? <bdju>cbaines: nope. I configured the key to use in my .ssh/config file already <cbaines>I'd check what SSH says if you specify -v or -vvv for more output <bdju>it looks like it's going through default-named keys, not finding any, and falling back to password... I don't really understand <cbaines>maybe try using -i on the command line to specify the private part of the key you want to use <bdju>oh, yeah, I just tried that and it worked <bdju>oh! I don't have this machine in my config... silly me <bdju>thanks for the help cbaines <civodul>sneek: later tell abralek re Guile-WM, i don't know if the issues are documented, but you can give it a try and see how well it works for you! <kamil_>mbakke, I can't help but smile :> Does it matter that I use redshift-wayland instead of redshift{-gtk}? <bdju>can I use the xdg config dir for my emacs config on the version of emacs in guix? <bdju>I can't remember what version the change was in <bdju>sounds like it's maybe 27 or 28 and we have 26.3 in the repo... <bdju>ah, but there's emacs-next which is on 27 <atw>"Emacs now uses the XDG convention for init files." in the 27.1 startup changes, so that's exclusive to emacs-next <bdju>nice. I'm gonna install emacs-next then I think <civodul>like ~/.config/emacs/init.scm instead of ~/.emacs? <civodul>sounds surreal, similar to when "top" got colors :-) <civodul>at this rate the kill/yank terminology will soon be replaced by copy/paste <pinoaffe>would an emacs server service for guix make sense? <nckx>pinoaffe: A system service? Not to me. It's a user thing. <bhartrihari>Hello, how can I execute binaries in /home/username/ without installing it via guix? <pinoaffe>bhartrihari: open a shell and type `. [path to binary] ` and then `<enter>` <pinoaffe>nckx: would be great if there was a way to define and run "user services" though :) <bhartrihari>It seems I always get the error `No such file or directory` <nckx>pinoaffe: I understand why ‘run foo as user’ system service is tempting (because shepherd user services are … spartan) but it's probably not the right place. <nckx>pinoaffe: You're aware that shepherd *can* run as something very similar to ‘systemd --user’ (or whatever), though, right? <nckx>It's just not very well-documented & tooled but it works fine. <bhartrihari>The file exists, and is executable (755) but still I get errors like `The file ./file does not exist or could not be executed`. pinoaffe <pinoaffe>nckx: I'm not aware what systemd --user does, so no :) <nckx>bhartrihari: You're using Guix System? <nckx>bhartrihari: That's because the executable wants to load /lib/ld-foo.so which doesn't exist there. <nckx> /lib/ld-linux-x86-64.so.2 (or similar) to be precise. <nckx>bhartrihari: There are two work-arounds <nckx>one is to modify your binary using patchelf <nckx>the other is to create a /lib/ld-linux-x86-64.so.2 symlink to /gnu/store/…//lib/ld-linux-x86-64.so.2 in your system.scm <pinoaffe>nckx: ah, that's neat, I'll take a look at it <nckx>your executable probably needs more than just ld.so, though, so read up on patchelf first. <pkill9>nckx: how do you specify users to run shepherd services as? <nckx>The real solution is to build from source 😉 <nckx>pkill9: You don't, you write ~/.config/shepherd/… and then just start ‘shepherd’ when you log in. <nckx>I do so from my .xsession. <pkill9>I was reading that as running root services as a non-root user <nckx>pkill9: That's a good question, I've never had to write a service for a daemon that didn't support ‘foobard -u <user>’ or similar. <nckx>(I think they stole it from me so it's OK.) <boeg>Any idea how to get screen lockers to work when I suspend? Suspend works, but screen lockers are not activated. I have tried with both (screen-locker-service slock) and (screen-locker-service xlockmore "xlock") and both, but they are not activated. I have to manually activate slock and then close my laptops lid to suspend <NieDzejkob>boeg: screen-locker-service is responsible for setting up pam and setuid to make the screen lockers work <NieDzejkob>you can either configure something in your desktop environment or use a tool like xss-lock <boeg>NieDzejkob: ah okay, i'll check out xss-lock <NieDzejkob>I then bind a shortcut to `xset s activate', so that when I force locking, xss-lock knows about it and doesn't conflict <boeg>NieDzejkob: good to know <boeg>NieDzejkob: do you just run xss-lock from your .xsession or similar, or do you hook it up with loginctl some way or? <roelj>For me GDM seems to crash immediately (I see a "Oh no! Something has gone wrong." screen). How would I go about debugging that? <pushcx>Perhaps there are logs in in ~/.xsession-errors? <anadon>I guess I should get around to fixing my patch requests from last weekend. <mbakke>kamil_: oh no, I fixed the wrong redshift? :-) <boeg>NieDzejkob: Can you explain what you mean regarding the `xset s activate` and force locking? <boeg>NieDzejkob: because I don't understand. Do you mean if you manually lock like just call the locker program but without suspending? <kamil_>mbakke, all three of them: -gtk, -wayland and regular one need the fix, I think ^^ <boeg>NieDzejkob: nevermind, makes sense now <bdju>Just got a new USB WLAN adapter and plugged it in. dmesg shows it being detected, but no wireless options appear in nmtui. Is there another step I should take? <bdju>"sudo herd restart networking" didn't quite do it either <bdju>adapter doesn't show under "ip a" or "nmcli device show" <boeg>with "guix install" i can install "transmission:gui" to get transmission with gui; how do I denote that in my package list in my config? It says it cannot find the package <boeg>seems like its called an output the ":gui" part <mbakke>boeg: you can also use the specification->package syntax in the packages list, which is more intuitive, i.e. (map specification->package '("transmission:gui" "foo" "bar")) <boeg>mbakke: that's actually what i have <mbakke>derp, you need 'specification->package+output' in that case, but then you need to deal with multiple return values ... <rekado_>new closure size for ghc-pandoc: 2176.8 MiB <rekado_>I still see references to GCC, perl, binutils, so we haven’t reached the end yet <davexunit>anyone else have a broken gdm when reconfiguring their system from master? <rekado_>ryanprior: this is the size of ghc-pandoc and all things it references <rekado_>davexunit: this is a recurrent theme when upgrading, I’m afraid. <davexunit>I see a few murmurings about it on the list but no real solution afaict <rekado_>davexunit: gdm may have accumulated state that you can get rid of <ryanprior>So if it references a lot of things then that's not a problem right? But it probably shouldn't reference GCC, that seems weird. <rekado_>there’s /var/lib/gdm, and state in the user’s home directory <rekado_>davexunit: ~/.local/share/gnome-shell/ for example <davexunit>mbakke: well at first the computer would completely lock up after trying to start gdm. couldn't even type in a console. <davexunit>mbakke: then I deleted /var/lib/gdm and I got to a white screen with a sad computer image that says "Oh no! Something has gone wrong." <rekado_>perl and gcc come from ghc itself. I’ll see if I can remove those references there. <davexunit>I'm in a slightly weird situation because I'm reviving a thinkpad x220 that has broken for about a year. <rekado_>bricewge: BTW I thought I finished my work on lightdm, but I can’t get it to work in a VM since my changes. <rekado_>bricewge: I haven’t given up on it yet, but if you’re interested in playing with it I could send you my diff <mbakke>davexunit: I've seen several messages on IRC about the "oh no" issue in the last day or two, I think a recent change borked GDM good. No bug reports yet though. <davexunit>it's a little disheartening because my previous generation is soooo outdated. <davexunit>I've been a long, somewhat frustrating journey to revive my dead thinkpad. I've replaced the motherboard, flashed many coreboot builds and finally got one that seems to do everything correctly, and soldered on a mod chip to replace the stock screen with a 1080p one. <mbakke>davexunit: if you have the time and some motivation left, can you try to pull a guix from a few days back to verify whether it is a recent regression? <rekado_>davexunit: replaced the motherboard with a custom one or just another X220 board? <mbakke>davexunit: do you see anything in the log files about why it might have crashed? <davexunit>rekado_: another x220 board, but I was able to source one with a faster processor. <davexunit>mbakke: /var/log/gdm/greeter.log has stuff but no smoking gun that I can see. <davexunit>removing ~/.local/share/gnome-shell did not help. <bricewge>rekado_ Please share it, I'll give it a try <davexunit>mbakke: I can try to bisect and find the regression. <davexunit>just wish I had a more recent good build. I'll just take a guess at a commit and see. <davexunit>any core-updates merges recently? I don't want to go back too far and have to build a ton of stuff. <davexunit>ah okay, looks like there was one 2 weeks ago <davexunit>well I'll start just after that and see if gdm works. <kkebreau>davexunit: I had the same error message. Looking into my /var/log/gdm/greeter.log, I suspect it's an issue with a recent libnma update. <kkebreau>I might have had a similar issue on the wip-gnome3.36 branch. <bricewge>kkebreau What make you think it's from libnma, can you share the log? <davexunit>kkebreau: well hopefully the regression was introduced after the latest core-updates merge. I'll find out shortly... <bricewge>davexunit: If it's libnma just try a commit from 2 days ago and it should work <davexunit>kkebreau: what makes you suspect it is libnma? <davexunit>I was just looking and didn't see it mentioned at first, but then I found something <davexunit>I can't easily share my log due to the state my system is in <kkebreau>"(.gnome-shell-real:779): ... JS ERROR: Error: Requiring NMA, version none: Typelib file for namespace 'NMA' (any version) not found" <kkebreau>I'll post it, it's on a different machine. <kkebreau>It's only the last 700 lines, since the whole log is too big for the site. <kkebreau>Line 169 is what made me think it was libnma. <kkebreau>There's also some error that seems to be caused by GJS by itself. <kkebreau>Maybe caused by the failure to find the NMA typelib file? ***kkebreau is now known as kkebreau__
***kkebreau_ is now known as kkebreau
<bricewge>I pushed and wrote the libnma patchset yesterday <bricewge>libnma has been removed from network-manager-applet <bricewge>What are the system tests that exercise Gnome? <davexunit>bricewge: not sure because I can't get to the gnome desktop <lfam>Probably some of the desktop tests would suffice for testing GNOME <kkebreau__>The only test I see under in gnu/tests/desktop.scm is one for elogind. <bricewge>Argh, I missed that gnome-shell was using network-manager-applet <bricewge>You could try to add libnma to the inpuits of gnome-shell <mbakke>gnu/tests/install.scm seems to pull in %desktop-services and gnome-desktop-service-type <mbakke>specifically %test-gui-installed-desktop-os-encrypted and %test-gui-installed-os <mbakke>so make check-system TESTS=gui-installed-os-encrypted might work <bricewge>mbakke: This one is already broken unfortunatly <lfam>Yes, we discussed it in patch review <mbakke>unrelated: I think #:glib-or-gtk? is unnecessary for libnma <bricewge>mbakke: I removed that arguments in c25ca6b79ea73a71eacbfa28a21ebdc5433e3c2a <bricewge>Building gnome-shell with or without libnma there are no complains nor acknowledgment of it's use. <mbakke>oh, I'm on staging and did not notice, sorry :P <bricewge>kkebreau__, davexunit: Does one of you have a guix repo setup to try adding libnma to the inpuits of gnome-shell? <bricewge>I'm quite sure it's the issue because that's the only package which is using netowrk-manager-applet that I forgot to update. <kkebreau>We'll see. If that's the case, GNOME 3.36 might work, too. <kkebreau>The same mistake is present in my own wip-gnome3.36 branch, lol. <kkebreau__>Both network-manager-applet and libnma are necessary inputs to gnome-shell. <bricewge>I should have been more careful when looking for package depending on network-manager-applet <mbakke>I keep a gnome.scm around for 'guix system vm' purposes when making changes to the GNOME ecosystem. <bricewge>mbakke: That's a good idea can you share it? <bricewge>kkebreau__, davexunit: I have pushed the fix to master. <davexunit>bricewge: kkebreau__: reconfiguring from commit 9af90aa works <kamil_>mbakke, I don't mean to come across as one in a privileged position, which I'm not in by any means, but can I expect a patch for redshift{-wayland} from you? It'd be nice to have geoclue to work, but it is not necessarily the end of the world ^^ <azvede>hi all! is this channel guix the OS or guix the packagemanager? or both? <azvede>cool! after a long time thinking about it, I've been diving heavily into FOSS stuff since February... trying to look for a daily driver OS after reviewing a lot and guix seems nice. <azvede>does the package manager have fairly high penetration of the same sorts of packages/projects that one might find in apt / bsd's pkg / snap / etc ? <azvede>(please forgive if this is a dumb question) 
