<apteryx>the problem seems to be a too high number of connections to the database <apteryx>last errors occurred yesterday: May 12 00:06:10 localhost postgres[3308]: [8-1] 2022-05-11 22:06:10.705 GMT [3308] FATAL: remaining connection slots are reserved for non-replication superuser connections <apteryx>and when I restarted the service we could see many connections being shut down (see the tail of grep postgres /var/log/messages|less) <civodul>i suspect the "In progress" thing is not postgres-related though <civodul>lilyp: i'm ready to add the jobset on ci.guix, but if you prefer to do that another day, that's fine too <lilyp>I just messed up my git-fu is all <lilyp>currently at 77% of rebuilding guix so that I can push ^^" <apteryx>not sure if I should have restarted the evaluation or just resumed the builds; I've done the former for the master Cuirass specification <cwebber>sudo -E guix system reconfigure foo.scm <cwebber>because then it would use the actual guix in my profile I think? <cwebber>ice-9/boot-9.scm:1685:16: In procedure raise-exception: <cwebber>Git error: repository path '/home/cwebber/.cache/guix/checkouts/l74zueb3lgylhgxnuzx3d5fzraztxvzu2s4466wlqqvmz7hdct3a/' is not owned by current user <cwebber>so, is the right thing to do to use "sudo guix pull" these days <vagrantc>cwebber: that does not sound desirable ... <vagrantc>perhaps an unintended consequence of a "reasonable" change? *vagrantc wishes for a "guix bisect" command :) <cwebber>well, there's git bisect, and the guix repo :) <cwebber>it looks like the error is coming from guile-git <cwebber>but does "guix system reconfigure" really work using the user's profile guix? <cwebber>which the manual doesn't say, I have no idea how I ever came to that conclusion <vagrantc>cwebber: up until recently i used sudo -E, but more recently been doing "sudo guix system reconfigure config.scm" without the -E <cwebber>well I guess I'll just do that and hope for the best :) <vagrantc>it appears to take the user's profile ... but it's been some days since i've tried <vagrantc>cwebber: who owns the files in your .cache/guix/checkouts/... ? <vagrantc>i vaguely recall getting some problem similar to that when i accidentally ran guix pull as root and it mangled my user's .cache <vagrantc>e.g. created some files as root, maybe only in the .git subdir trees ... <cwebber>vagrantc: MysteriousSilver: well it seems like things went fine... <cwebber>it didn't start pulling down super ancient packages afaict <PotentialUser-57>hi, i just installed guix and every program installation fails, i did the "sudo guix system reconfigure /etc/config.scm" it also failed. By the way it keeps telling me to install something called "glibc-locales", if i install that will that fix it? if i can even install it <vagrantc>PotentialUser-57: "failed" is a bit unspecific ... maybe paste the error to paste.debian.net or something? <PotentialUser-57>it's really slow to build too, dunno if that is normal. it's not compiliing is it? <vagrantc>PotentialUser-57: have you intentionally disabled substitutes? <vagrantc>oh, nevermind, it is downloading substitutes <vagrantc>build of /gnu/store/3mc6ivs5x5snnrixsfdz9jmd7nymzg5c-webkitgtk-with-libsoup2-2.36.0.drv failed <vagrantc>View build log at '/var/log/guix/drvs/3m/c6ivs5x5snnrixsfdz9jmd7nymzg5c-webkitgtk-with-libsoup2-2.36.0.drv.bz2' <vagrantc>PotentialUser-57: it appears that ^^ failed to build, probably both for you and the substitute server. <vagrantc>and so the things that depend on that couldn't be built either... <vagrantc>you figure out why it's failing to build, and submit a patch :) <PotentialUser-57>yes a pull. Im following the manual. ran pull which succeeded, then the update command that failed <vagrantc>yes, well, some packages you're asking for are in a broken state, apparently <vagrantc>yeah, gnome has lots of dependencies, so it's very easy for it to be broken <vagrantc>PotentialUser-57: bzcat /path/to/log/foo.log.bz2 ? <vagrantc>PotentialUser-57: i usually just open it with emacs and it figures out how to display it <vagrantc>look for things that look like errors and give you a clue as to why it failed to build... <vagrantc>probably starting from the bottom and working your way up <vagrantc>how much free memory and/or diskspace does your system have? <PotentialUser-57>well its a fresh install of a 500gb disk, and ram is 8 or 16gb, dont remember which <vagrantc>should be plenty (although maybe not for icecat) <PotentialUser-57>gnome almost seemed to crash earlier btw, everything got really slow and then the browser crashed <vagrantc>PotentialUser-57: anyways, keep reading what lead up to the error a bit ... all that says it the compiler died for some reason <ryanprior[m]>I have 16g RAM and I never build icecat or ungoogled-chromium on this machine <ryanprior[m]>If I had to for some reason I'd spin up an offload machine <ryanprior[m]>but in practice what I do is wait for substitute availability <vagrantc>last build for master was from mid-april... <vagrantc>PotentialUser-57: you're not building off some other branch? <PotentialUser-57>I just grabbed it off of the official website, verified the keys, then installed hehe <vagrantc>i mean ... guix pull --branch=staging or something <PotentialUser-57>i tried to paste the crashlog to debian pastezone but it was too big lol <PotentialUser-57>ryanprior[m]: I'm so new I don't even know what that means exactly. Are you saying the packages servers are down? <vagrantc>it takes time to build substitutes, so if they haven't had a chance to build something yet, usually you end up building locally <ryanprior[m]>If you run `guix weather icecat` it'll tell you whether a substitute is available for icecat on your current version of guix <ryanprior[m]>If you wait, it will probably eventually become available <vagrantc>in guix, downloading from substitutes is considered an ... caching optimization, it is focused on building from source <ryanprior[m]>I often use `weather` and `update --dry-run` to find out what packages would be updated, and avoid updating packages that would require a local build. <ryanprior[m]>I wouldn't say that's true, it's extremely useful if you never build anything from source <ryanprior[m]>The ability to build anything from source is a prerequisite, and the way Guix makes it so easy is a huge value-add, but I wouldn't say that's the focus (from an end-user perspective) <vagrantc>if that were the case, there would be easier options to "only download substitutes and don't build anything locally" <vagrantc>but in the discussions i've seen on such features, that's how it's been described <ss2>oh, wow. Guix has such neat features. VT7 (gdm) will close eventually after not using while switching over to VT7 by mistake. <ryanprior[m]>I agree, that's a nasty functionality gap imo, we should make it easier to favor substitutes <PotentialUser-57>thanks for informing me vagrantc and ryan, i didnt know it works like that <vagrantc>PotentialUser-57: that was from the system reconfigure <PotentialUser-57>So what's the best way I can update and make this system usable? uninstall gnome and install something else and then it will work? apparently im on gnome 34.2 <vagrantc>it's kind of awesome that "guix build SOMEPACKAGE" basically behaves the same weather you're building from source or downloading a substitute ... it's just generally one takes longer and more resources :) <vagrantc>PotentialUser-57: if it's in the system profile, you just build a lighter system <PotentialUser-57>How do I go about doing that. I've only read the manual up to the point where you do the graphical install haha <vagrantc>if it's in your user profile (e.g. guix install icecat) ... you can ... guix upgrade --do-not-upgrade=icecat <PotentialUser-57>vagrantc: the upgrade failed before I had installed anything, other than selecting gnome in the graphical install menu i guess <PotentialUser-57>by the way is it safe to abort building a program with ctrl+C or will that mess it up <vagrantc>guix package -l ... doesn't have anything? <vagrantc>PotentialUser-57: to edit your system config, apparently you have it in /etc/config.scm ... you edit that file and remove the things you don't need ... <vagrantc>PotentialUser-57: ok, so you've got an empty user profile, that's normal for a new install <PotentialUser-57>i think the only thing in my /etc/config.scm is gnome-desktop-service-type <vagrantc>i haven't actually seen what kind of system config the installer produces ... but there are some lightweight desktop examples in the guix source code *vagrantc wonders if those are exposed in the manual or something <vagrantc>might work, but it might have some of the same things that fail <PotentialUser-57>i dont get what is failing on a fresh install anyways, is it cause of specific hardware? <vagrantc>i haven't really done a major overhaul on my configs for years, and barely understand guile, but maybe someone else could help more <vagrantc>PotentialUser-57: it is as simple as a the current version of the software you want to build does not build. <vagrantc>it's not like it's anything specific to your computer <PotentialUser-57>vagrantc: ok but do you know the command to rebuild? like i replace gnome with xfce in the scm file, then what command do i run? does the system become unusable? do I have to reboot? <vagrantc>"guix system reconfigure /path/to/someconfig.scm" ... sounds like in your case /etc/config.scm <vagrantc>PotentialUser-57: if all goes well, it should still work, but obviously, if all went well you wouldn't even be asking :) <vagrantc>the good thing is with guix it is usually easy to revert to a previous state, so you can generally feel free try things without fear of it breaking too badly <vagrantc>because you can get back to where you were <PotentialUser-57>do i need sudo to edit this file, it complains when i try to edit it with vi <PotentialUser-57>im still afraid to cancel this current building. i tried to install konversion irc client and right now its, i guess, compiling qtwayland haha <vagrantc>it's possible someone updated something recently that had too many rebuilds, or, you're just unlucky. <PotentialUser-57>vagrantc: thanks for all the help btw. im just gonna wait for this compile to finish then i'll try guix system reconfigre and see what happens :D <vagrantc>PotentialUser-57: either edit /etc/config.scm directly, or you can copy /etc/config.scm to your home directory and edit it from there and call "sudo guix system reconfigure config.scm" <PotentialUser-57>i edited with sudo. cancelled the current compilation cause i cant wait that long now hehe. so now im reconfiguring... should i really be doing this from within the DE? <PotentialUser-57>does it automatically pull down the latest version btw? Its just the insatll iso that has those old versions? <vagrantc>not sure, i use a much smaller, simpler environment ... sway <vagrantc>it boots to a console login, and i start sway from there <vagrantc>but i do spend most of my time at a few consoles <PotentialUser-57>vagrantc: i tried to play the video but this gnome web browser is not having haha <vagrantc>PotentialUser-57: gotta head out ... good luck! <apteryx>is someone using Guix System with LUKS encrypted drives? <PotentialUser-57>apteryx: i am right now but having all kinds of issues. the installer has it as an easy option that does it automatically btw <apteryx>the build farm has not been well in the last couple days <lilyp>cwebber: IIUC sudo looks up the command independantly of the user you su into <emacsomancer[m]><apteryx> "is someone using Guix System..." <- Yes, I just luks on Guix <apteryx>Do you specify your LUKS mapped device via their UUIDs? ***dgcampea-2 is now known as dgcampea
<m4t3>Is it possible to have Hurd as Main kernel and Linux as sub-kernel? And what are the benefits of such design? <hnhx[m]>Hey! Does anyone know how could I use my own configs for dwm and st instead of the default one in the repos? <unmatched-paren>hnhx[m]: you'll need to write a package that inherits dwm/st and adds a patch <unmatched-paren>hnhx[m]: can i see your /etc/config.scm please? i'll remove gdm from it for you and you should be able to see how it's done <hnhx[m]>do you need the whole config file or only the services part? <unmatched-paren>re config.h: i think it's possible somehow? you'd need to read the file at package-load time, then write it into config.h at build-time... idk <unmatched-paren>(sorry for the condescending-sounding question, just need to make sure) <hnhx[m]>cant I just cast gdm to the void and completely remove it *unmatched-paren can't remember whether they had this problem themselves... <hnhx[m]>Imma be honest I don't like guix, I don't like the concept of it, the only reason why I use guix is because it not outdated like all the other FSF approved distros <unmatched-paren>it's weird at first but once you get to know it it's far more powerful than a traditional distro <hnhx[m]>unmatched-paren: I don't like that I need to manage my whole system from a config file, like I cant really tinker with my system, I have to do everything in that config. I used NixOS in the past and I disliked it for the same reason. <unmatched-paren>(e.g. i think many of the resources for guile/scheme are pretty inpenetrable, and the guix manual is more of a reference) <unmatched-paren>hnhx[m]: that's literally the entire point :P everything managed in one place configured with the same language, which also allows you to roll everything back <hnhx[m]>Like I just want to compile something from source (e.g.: dwm or st) and I need to create a custom scm file just for that. <hnhx[m]>unmatched-paren: ye as I said I don't like this concept <nckx>pashencija[m]: I just approved a gmail address which I guess was yours. Sorry for the delay and welcome! <hnhx[m]>but Parabola is heavily outdated even though its "rolling release" <hnhx[m]>icecat havent been updated since 2021 <hnhx[m]>and all the other fsf approved distros are LTS <hnhx[m]>so ye thats why I eventually went with guix <unmatched-paren>writing scheme files isn't hard once you learn how. the problem is that it's hard to learn right now (I struggled at the start) <hnhx[m]>ye I'm starting to get the hang of it, its just that I wish FSF approved distros would get more attention. <unmatched-paren>the manual doesn't have a tutorial or anything, just e.g. a list of commands, a list of what you can do in a (package), what services there are, etc <hnhx[m]>I have a repo on github with my custom config <hnhx[m]>a single repo for both of these suckless utils <hnhx[m]>could I get that to install via a scm file? <unmatched-paren>hnhx[m]: probably, but you'd need to modify everything and it's probably not worth the trouble. since you are the only one who'll need this [...] <unmatched-paren>you'll probably want to just set up a local dir containing the $SUCKLESSTOOL plus modifications <hnhx[m]>unmatched-paren: so according to this <hnhx[m]>should I just put my custom dwm folder to a tar file <hnhx[m]>and I could install it via guix package --install-from-file? <unmatched-paren>you'll be able to use `(local-file)` to install from the directory instead of a remote source <unmatched-paren>my messages aren't going through very quickly, so it looks like i'm not replying, sorry <unmatched-paren>hnhx[m]: in this manifest.scm, you should `(use-modules)` the following modules: <PotentialUser-57>goes anyone know when the subsitution servers will work again? i kinda dont have time for these build errors right now <unmatched-paren>(guix packages) for `(package)`, (guix gexp) for (local-file), and (gnu packages suckless) for st and dwn <hnhx[m]>unmatched-paren: thats why I decided to make my own matrix homeserver, public homeservers can be very slow at times, very annoying. <hnhx[m]>thought you are using the bridge as well, my bad <unmatched-paren>(define <the suckless tool>-my-patches (package/inherit <the suckless tool>)) <PotentialUser-57>does anyone know why webkitgtk-with-libsoup2 keeps failing to build for me <hnhx[m]>PotentialUser-57: It failed on me as well, turns out I ran out of RAM. Try decreasing the amount of cores to use. <unmatched-paren>then at the bottom of the file write `<the suckless tool>-my-patches` which will make this file return that package when it's evaluated <hnhx[m]>I have 16GB of RAM and 16 cores. I decreased the cores to 7 and it built without issues. <unmatched-paren>hnhx[m]: now we'll want to fill in the inherited package's fields: (name "<the suckless tool>-my-patches") <hnhx[m]>where should I put the define thing? <unmatched-paren>and (origin (local-file (dirname (current-filename)) #:recursive? #t)) <ZhuAisi[m]>`guix weather gcc` reports 1000 queued builds, 999 for powerpc, 1 for x86_64 <unmatched-paren>this file isn't used by any others so it doesn't need to export anything <unmatched-paren>(dirname (current-filename)) gets the directory that the scm file is in <nckx>CI is, at least, ‘experiencing issues’ and delays and general not-goodness. <PotentialUser-57>hnhx[m] so what's the full command? Sorry but I'm here from pretty much a fresh install so I havent read the whole manual yet, only up to that point. trying to update the system <hnhx[m]>PotentialUser-57: do you want to reconfigure your system? <PotentialUser-57>hnhx[m] yes and it keeps failing on installing gnome control panel or whatever, with the webkitgt-with-libsoup2 package <hnhx[m]>in that case: `sudo guix system reconfigure /etc/config.scm --cores=7` *unmatched-paren wonders if they could start writing up an "Introduction to Guix" that's a bit more approachable than the manual... <PotentialUser-57>it's 4 core intel, 8 with hyperthreading i guess. i dont remember if this machine has 8 or 16 gigs of ram to be honest <hnhx[m]>unmatched-paren: should I just use current-filename? <unmatched-paren>(as in, how to write packages, how to do x and y in build recipes, etc) <unmatched-paren>hnhx[m]: no, that would make the entire source code of your package just the scheme file <ZhuAisi[m]>IIRC, webkit will consume tons of memories while linking <hnhx[m]>PotentialUser-57: The Gentoo guide recommend that for every core you use you should have 2GB of RAM. <unmatched-paren>(this scheme file is in the checked-out patched <the suckless tool> source code, remember) <hnhx[m]>so when you compile with 4 cores you must have 8GB of free RAM <hnhx[m]>unmatched-paren: so (dirname (cdwm.scm))? <unmatched-paren>hnhx[m]: no, (dirname (current-filename)), (current-filename) returns the filename of the current scheme file <PotentialUser-57>ok running it, now, probably going to take a while haha. Anyways what does "warning: cannot determine provenance for current system" mean anyways <unmatched-paren>PotentialUser-42: that happens if you haven't reconfigured before afaik <hnhx[m]>PotentialUser-57: hope it will build without issues this time <unmatched-paren>hnhx[m]: now try `guix shell -f manifest.scm` and run `<the suckless tool>` inside the shell <hnhx[m]>/home/hnhx/Code/suckless/dwm/dwm.scm:7:2: error: (package (inherit dwm) (name "dwm-my-patches") (origin (local-file (dirname (current-filename)) #:recursive? #t)) (replacement (and=> (package-replacement dwm) loop))): extraneous field initializers (origin) <unmatched-paren>(origin) is what you put inside (source) when you want to fetch remote source code <hnhx[m]>should I remove dwm from my config.scm file? <unmatched-paren>then yes, you should remove it and use `guix package -f manifest.scm` to install it to your profile, unless you want to use guix home <unmatched-paren>btw, this whole thing is a bit of a hack, so if you want to do things Properly, you'll probably want to either: <unmatched-paren>(1) use `substitute*` inside the build recipe to patch config.h automatically instead of replacing the source <unmatched-paren>(2) upload a git repo and use (origin) for it instead of (local-file) <unmatched-paren>but if you're happy with imperative package installation, just use it <unmatched-paren>(2) as one of the fields in `origin` add `(method git-fetch)` and import `guix git-download` in `use-modules` <hnhx[m]>how would I go with the substitute method< <unmatched-paren>(4) run `guix hash -rx` in your patched repo and replace HASH with the result <hnhx[m]>unmatched-paren: it starts but it doesnt work now with my custom dwm xp <hnhx[m]>I had to switch to another tty and just do startx <hnhx[m]>idk why display managers are a thing grrr <unmatched-paren>hnhx[m]: please show me the contents of `/var/guix/profiles/system/profile/manifest` <unmatched-paren>AceTheMercenary[: i believe that it's not fully packaged yet, and won't be for some time <hnhx[m]>unmatched-paren: do you want the full file? <hnhx[m]>hnhx@guix ~/Code/suckless/dwm$ grep -i gdm /var/guix/profiles/system/profile/manifest <unmatched-paren>ok, i have no idea what's going on... i'll see if someone else does; nckx: are you currently using gdm? <hnhx[m]>unmatched-paren: no, as I said it starts but when I login it doesnt work so I just switched to another tty and did startx there <unmatched-paren>hnhx[m]: you'd need to add the appropriate `use-modules` and return the package object too <unmatched-paren>btw, you can have multiple (("REGEX") "REPLACEMENT") in a single substitute* <hnhx[m]>and this will move the package to guix-system, right? <hnhx[m]>I dont want to have any packages as guix-user <unmatched-paren>well, you can then move that package definition to `/etc/config.scm` <hnhx[m]>could I just move the one that you sent me previously as well< <hnhx[m]>* could I just move the one that you sent me previously as well <unmatched-paren>btw, is your goal to (a) have everything in the configuration file, or (b) to have everything installed in the system <hnhx[m]>well both, I want to have everything in config.scm and I want all the packages to be under guix-system and not guix-user *unmatched-paren will upload his ~/conf to sr.ht and paste a link <unmatched-paren>reasons for (a) are obvious (declarative, reproducible, etc) but i don't see why you'd want (b) <hnhx[m]>because Im the only user so I dont really see the reason for having packages as guix-user <hnhx[m]>my priority rn would be removing gdm though qwq <unmatched-paren>btw, what they're saying in that reddit thread about "no dm being unsupported" is completely false <hnhx[m]>ye ik since I got startx to work lol <hnhx[m]>also xinit wouldnt be in the repos then <hnhx[m]><unmatched-paren> "(is there even a /var/gdm?)" <- nope <hnhx[m]>At this point Im just going to make a shell script that constantly kills gdm or idk lmao <unmatched-paren>there should be a box you can tick in the installer to say you don't want gdm <hnhx[m]>could I get away with removign desktop services? <horizoninnovatio>You can edit the config.scm before installing. There's a section I read somewhere about removing/replacing gdm with slim <unmatched-paren>you'll probably also need pulseaudio, alsa, dbus, wpa-supplicant, network-manager, fontconfig, ... <hnhx[m]>like I do startx and its a black screen for like 5 seconds <hnhx[m]>meanwhile on parabola starx was instant *unmatched-paren is wondering if it's because the system guix is older than the fix for the stat storm? <hnhx[m]>I have been using this xinitrc for like a year <hnhx[m]>btw do you happen to know how could I switch to zsh? <unmatched-paren>let's set up guix home for zsh and move dwm-my-patches there. maybe that'll fix it (maybe???) <hnhx[m]>also, do you happen to know what would be the GUIX equilivent of this: <hnhx[m]>I have been struggling to set this up qwq <hnhx[m]>unmatched-paren: I will just stick with bash for now ig <unmatched-paren>"Note: It is highly recommended that you manage your shell or shells with Guix Home, because it will make sure that all the necessary scripts are sourced by the shell configuration file. Otherwise you will need to do it manually. (see Configuring the Shell). <janneke>do we have ssh-agent, git-daemon, znc etc. homes-services yet? <unmatched-paren>there doesn't seem to be home-services for znc, git-daemon, or ssh-agent <janneke>i tried home-services once, but it's really no use for me without ssh-agent and znc; ssh-agent would be the very first service to add, right? <unmatched-paren>not sure what you mean by the first service to add? you mean the one you'd add first if you were implementing all three? *janneke wrote an ssh-agent service example as a user service for shepherd, just before home-services came out ;) <janneke>i mean isn't that the single service everyone really needs? <janneke>or everyone has already scripted that in their login scripts, i suppose *unmatched-paren now wants it :P <janneke>it's real easy, there's an example in the shepherd docs ;) <hnhx[m]>(simple-service 'etc-modprobe-service etc-service-type <hnhx[m]> `(("modprobe.d/vfio.conf" ,(plain-file "vfio.conf" "options vfio-pci ids=10de:13c2,10de:0fbb")))) <hnhx[m]>unmatched-paren I solved the latency with startx <rekado_>civodul, apteryx I don’t think there should be many connections from cuirass to postgres. But I do see that WITH-DATABASE in src/cuirass/database.scm opens a connection but never closes it. <rekado_>in fact, nothing ever calls db-close (defined in src/cuirass/database.scm) ***littlebo1eep is now known as littlebobeep
<hnhx[m]> (kernel-arguments (cons* "nomodeset vfio-pci.ids=10de:25e2,10de:2291" %default-kernel-arguments)) <hnhx[m]>please don't tell me I have to switch back to parabola just because of this <hnhx[m]><unmatched-paren> "hnhx: what is that supposed to..." <- It allows my NVIDIA GPU to get passed to a KVM <littlebobeep>Do you really call virtualized guests OSes like they are "a KVM"? I thought KVM was the hypervisor code in Linux, but not sure if it refers to guests as well <hnhx[m]>I think a "kernel virtual machine" makes sense <hnhx[m]>instead of loading vfio via initrd I did this: <hnhx[m]>I'm not sure why it doesnt work via initrd <hnhx[m]>since on parabola I do it via initrd <hnhx[m]>I mean the VFIO module loads in when I do it via initrd <mekeor[m]>is there a ticket for making home-services compatible with system-services? <mekeor[m]>i.e. so that any/some system-services can be used as home-services <lilyp>aiming for any, currently more likely some <lilyp>I don't know about a particular ticket, but it's a known concern <hnhx[m]>does anyone know how could I autologin on tty1? <hnhx[m]>but when I reboot, I cant login to tty1 at all, it says error in the service <PotentialUser-57>ok so I finally got the update to compile but it gave me some message about "in procedure raise exception" and something about match error? did the update fail? or how do i load it? I rebooted but its still all the same old software <PotentialUser-57>no i did it on a console cause otherwise the system would freeze, and i rebooted afterwards <unmatched-paren>you won't be able to find it in there, the error message tells you.... <PotentialUser-57>i can tell u the last line i got, i took a very blurry photo, wait 1 min <PotentialUser-57>Throw to key 'match-error' wotj args ("match" "no matching pattern" #<package.polkit@0.12Q gnu/packages/polkit.scm:54 7f6650alf20> <unmatched-paren>looks like someone may have messed up a change in gnu/packages/polkit.scm <PotentialUser-57>ok i'll try... i hope it works this time. this has been the most complicated process in any linux distro i've tried:D but thats probably only becase the substition/package server is down or whatever <civodul>rekado_: i think with-database creates a worker thread and opens a postgres connection, and then that connection is reused throughout the process lifetime <unmatched-paren>PotentialUser-42: i don't think it's down right now, just misbehaving <PotentialUser-57>i meant complicated "update process" not process in general:D . Just misbehaving ok I see <munksgaard>PurpleSym: Thank you for your e-mail/work regarding cabal/hackage imports. I've only done a cursory reading so far but it looks nice. I'll try it out locally tomorrow, I hope. <munksgaard>(Also, I hope that it's okay that I contacted you here) <PurpleSym>munksgaard: Thank you and you can always ping me here. <civodul>lilyp: if you're around, you're welcome to rebase + push on top of 'master' <civodul>PotentialUser-57: how did you encounter the bug? was it with "./pre-inst-env guix system reconfigure" or something else? <civodul>the bug is stateful, so mostly likely if you re-run the same command after the crash, it'll work <PotentialUser-57>civodul: just doing a clean install from the official guix iso, then running the update commands <civodul>PotentialUser-57: you ran "guix pull" (successfully), and then it's "guix system reconfigure" that failed, do i get it right? <PotentialUser-57>took like 5 attempts to get it to compile at first too but managed to solve that one with the help of some people in here hehe <PotentialUser-57>although, over the course of those failures to compile i did install some other programs like netsurf, tried and failed to build firefox, and maybe something else i dont remember <PotentialUser-57>yeah im not hating but it's a little bad to have such failures directly from a clean install, it makes me think of ubuntu and that is never good :O <PotentialUser-57>civodul: I got the same error, despite running sudo guix system reconfigure ............. twice <PotentialUser-57>Actually the last couple of lines look a little bit different this time, after the stuff about ice: <unmatched-paren>PotentialUser-42: looks like your /etc/config.scm's partitions section might be wrong or something? <PotentialUser-57>ok i rebooted and now gnome seems to be updated despite those error messages earlier... probably why i couldnt launch gnome apps just now before rebooting <PotentialUser-57>ok here is my /etc/config.scm . Only edit I've done is change "(target "/boot/efi")" to "(targets "/boot/efi") because guix complained that "target was deprecated everytime i did something. Also removed my full name from this paste. http://paste.debian.net/1240876/ <PotentialUser-57>should i change it back? guix complained to me everytime i tried to install everything, and it stopped complaining after i changed it <KarlJoad>Is there a "best practices" way to share configurations between machines in Guix? <PotentialUser-57>how can i remove all lxde packages? I run guix remove lxde-desktop-service-type but it says it's not found in profile. i had it in my config.scm but not anymore <mekeor[m]>PotentialUser-57: what does "guix package -I" show? i.e. upper case i <KarlJoad`>I remember someone telling me where Guix keeps kernel configs after building, but I forgot where. Anyone know where the kernel compile-config is stored after building is done? <mekeor[m]>PotentialUser-57: you can't "guix remove" is an alias for "guix package -r", i.e. it's for removing packages, not services :) <mekeor[m]>PotentialUser-57: as i said, please use an upper-case i, not a lower-case L <PotentialUser-57>guix package: error: invalid argument: Missing required argument after `-L' <PotentialUser-57>glibc-locales 2.33 out /gnu/store/nrr24nvf6548if5wdpvxhlvjif3x9jjp-glibc-locales-2.33 <PotentialUser-57>netsurf 3.10 out /gnu/store/skn0y903xf3sgkwb71bamglik5pm7sqr-netsurf-3.10 <KarlJoad`>Can someone point me to a configuration repo that makes heavy use of sharing between operating-system definitions? I want to unify my users list, my packages list, etc. to a single file. <mekeor[m]>PotentialUser-57: so you only have two packages installed into your user profile. glibc-locales and sqr-netsurf. lxde is not installed. <PotentialUser-57>mekeor[m]: I installed it by changing my config.scm, gnome-desktop-service-type to lxde-desktop-service-type. Then cancelled midway because compiling was taking forever so now I have some lxde programs lol <PotentialUser-57>might be a stupid question but my does my home folder stay persistent even if i change everything else? like if I do a "guix package --roll back" or change my config.scm to a different DE, home stays unscathed? <KarlJoad`>PotentialUser-57: For the most part, yes. In the worst case, files are _added_ to the home directory, never removed. <PotentialUser-57>Karljoad: I see that makes sense. No data loss but might be a good idea to do some manual cleanup if you change the system completely, I gather <KarlJoad`>Yes. If you change DE, and they do not use the same config files, then new ones might automatically be generated by the DE. But you should never lose files through a system reconfigure. <KarlJoad`>How can I specify make as a package for my home-environment? I always end up with an unbound variable error because of a missing `use-modules' form. But, I have the (gnu packages base) module present. <f1refly>is there any utility i can use to pack rar archives? <ZhuAisi[m]>Maybe we should provide a better code editor in livecd <lilyp>because rar is a proprietary format which is only partially supported by free software like libarchive <ZhuAisi[m]>Nano or Zile doesn't support bracket matching, it sucks for who new to Guix to edit the config file <f1refly>because rar is a requirement for a specific use case I have, it has to work on someone elses computer and the only accepted format is rar unfortunately <ZhuAisi[m]>I see a lot of newcomers in another Guix chat group and they often encounter strange problems that are usually caused by bracket mismatches <lilyp>f1refly: as stated above, you might get lucky with libarchive, but it depends on the rar file <lilyp>note that if you're using the default setup you should already have file-roller as a GUI that uses libarchive in the background <jfred>Hmm... I have a microSD card with an ext4 filesystem on it that I can't seem to mount from my Guix laptop for some reason. I can mount it on one of my other laptops running NixOS, but on Guix I just get "mount: /mnt: wrong fs type, bad option, bad superblock on /dev/sdb1, missing codepage or helper program, or other error." <jfred>Can't say I've encountered this before on a device with an ext4 fs... any ideas? <jfred>oh, it would have been formatted case-insensitive, maybe the guix kernel doesn't support that? <jfred>yep, I think that's it - don't mind me, rubber-ducking myself :) <plattfot>Hi, has anyone encountered `emacs-minimal-28.1-builder:1:2852: Unknown # object: "#<"`? Just got that when I did a guix pull and updated my emacs profile <the_tubular>Hey lilyp, just saw the new emacs package, thank you for it :) <the_tubular>But does that mean that emacs-next is deprecated or it's just a lack of time ? <plattfot>itd: Any idea what the issue could be? I assume it has something do to with the move to emacs-28.1 <zacchae[m]>Anyone have problems with network manager GUIs in exwm or other non Gnome DE (missing icons, crashing)? I suspect it is because I removed the GDM service <itd>plattfot: Not really. Issue seems to be around line 49 in https://paste.debian.net/hidden/baff4ff8/ (that's a paste of the file mentioned in the error); looking at recent guix.git commits I'm intrigued by the "new package style". I'd like to learn more about that. <emacsomancer[m]><plattfot> "Hi, has anyone encountered `..." <- the flatwhatson emacs-nativecomp-28.1 package seems to be having the same issue <KarlJoad`>Is there a "best practices" for how to structure packages for multiple machines across multiple files? <plattfot>emacsomancer[m]: hmm, maybe that's what's causing this as I'm using emacs-pgtk-native-comp from that channel *plattfot doing a poor mans git bisect with `guix pull --commit=…` <zacchae[m]>KarlJoad`: I think it depends on how different machines are, but there is no standard. I have all my machines use the same config, except for the "host-name" line in operating system. Then all differences are made based on logic with the host-name <KarlJoad`>zacchae[m]: I'm just trying to share package lists, users, and services. The rest of the configuration can be specified manually. There are only 2-3 computers. <zacchae[m]>my situation is similar. Probably what you want then is to have a standard config, then a config for each machine that inherits from your base OS definition <KarlJoad`>Given that all of this is tracked in a git repo, I am planning on first moving packages and the like out to variables first, then inheriting base OS info through a function call. <KarlJoad`>Would it make any sense to make the various definitions part of a channel? Or would modules be enough in this case? <plattfot>Looks like something is broken with commit: "64557bc695 gnu: emacs: Use new package style." Works fine with the previous one, but with that I hit the error. I'm going to create an issue, if there isn't one already. <zacchae[m]>KarlJoad`: I would guess not. You can have your config files in a directory and add them to your load path when calling guix with -L I believe. (Though, if you have .scm files in a directory, that might be a valid channel no?) <KarlJoad`>Looking at my actual channel, the .scm files in a directory does make a channel, yes. I guess my question would be, is there a benefit to having the configurations in a channel? <itd>plattfot: Thanks for the link. :) <lilyp>So I am very sorry for breaking third party channels unilaterally, but having Emacs 28 in Guix is a first step forward towards having that channels feature's available for everyone <lilyp>We do still need native-comp everywhere tho <KarlJoad`>How can I use a local definition of a channel (I made some edits to a channel locally) over the `guix pull`ed version? <plattfot>KarlJoad`: guix pull -L <path/to/local/channel> might work <bjc>lilyp: does the guix version of 28.1 include nativecomp? <plattfot>lilyp: No worries, thank you for your awesome work! The issue I have seems to be unrelated to third party channels though. I just remove the flatwhatson and my own channel and it still fails on commit 64557bc695. I'm on a foreign distro, not sure if that makes a difference. <mroh>I can confirm that commit 64557bc695 fails to build with `emacs-minimal-28.1-builder:1:2852: Unknown # object: "#<"`. <KarlJoad`>So... My package list is in my channel now, but is there a reason to put system descriptions in a channel? CI with Cuirass, maybe? <plattfot>and I just notice I missed to remove a trailing : in the "guix describe --format=channels" *facepalm* <lilyp>bjc: sadly not, imho that'd also require adding native-comp to emacs-build-system <the_tubular>lilyp, I have a very basic understanding of the emacs package, but wasn't native-compilation already enabled in a package ? <luishgh> hi guix, has anyone here got screen sharing working on sway with Guix System? For me, ungoogled chromium always crashes when I start sharing and in icecat nothing happens. I've already got audio with pipewire working and xdg-desktop-portal and xdg-desktop-portal-wlr installed. <djeis>Native-comp's in the flatwatson channel, yea. <luishgh>forgot to mention, screen recording in obs works after installing obs-wlrobs <user_oreloznog_>...and Vagrant Cascadian for the lyrics and Luis Felipe for the illustaration <lilyp>the_tubular: basically yes, you only need to add the configure flag which you can easily do via transformations <lilyp>but that's exactly why for the main channel I don't think that doing that is enough <lilyp>we also have to think of all the emacs-* packages like magit etc. <KarlJoad`>unmatched-paren: I was hoping for a more temporary solution, like specifying the channel's path in the build command. <lilyp>okay, I think I got the error and it is an incredibly stupid one <KarlJoad`>Does Guix Home have support for writing X11 configuration files, like .Xsession? <lilyp>Guix Home can write arbitrary files if you go low-level enough <plattfot>KarlJoad`: Have you tried `guix build -L /path/to/channel <package>`? That what I'm using when testing out changes to packages in my channel before commiting. <lilyp>To everyone interested in the emacs drama, I'm currently building the fix for it. <zacchae[m]>KarlJoad`: to my understanding, all X11 is still only cover by guix system <zacchae[m]>Which is now giving me problems because i updated my guix system and that caused my guix home env to break... <the_tubular>FIx for what exactly ? I didn't follow the whole conversation lilyp <plattfot>KarlJoad`: You can use home-files-service-type combined with local-file or plain-file to copy/write files with guix home. <lilyp>so, if the build succeeds, it should be pushed soon <lilyp>if not, it might take some while because I'm afk now <mekeor[m]><lilyp> "To everyone interested in the..." <- whats the emacs drama? <rekado_>civodul: I’m probably missing something, but it looks like with-database uses make-worker-thread-channel, which unconditionally calls db-open. <rekado_>I suppose it doesn’t matter because with-database is only used by cuirass scripts, and presumably the database connection is closed when the process exits. <rekado_>cuirass-evaluate is spawned repeatedly, for example, and if postgres never cleans up abandoned connections left behind by dead processes I could imagine it running out of open slots eventually. <doncatnip>KarlJoad`: i let guix home handle my X org conf by passing a local xorg.conf to xinit in a custom startx script <doncatnip>KarlJoad`: you can then use a simple service to write or copy that xorg.conf to your home directory <civodul>rekado_: most likely postgres notices when a client dies and thus frees associated resources <civodul>so each Cuirass process opens one connection, and it gets closed when it terminates <civodul>lilyp: hey! emacs-dash fails to build <civodul>looks like i didn't pay enough attention to the ci.guix dashboard <civodul>"dash.el:615:8: Error: docstring wider than 80 characters" <civodul>lilyp: (inputs (search-auxiliary-file "emacs/guix-emacs.el") ...) has no effect <civodul>emacs-xwidgets and others that inherit from 'emacs' are broken too <civodul>i wonder if it would have been safer to revert and fix up on a branch <rekado_>how do you connect to postgres on berlin? How does db authentication work for cuirass? <rekado_>(I’d like to poke around in the db stats) <rekado_>ah, just need to become the cuirass user <f1refly>how can i skip checks in ruby-build-system? I'm trying to use a rubygem in a personal project, but the gems creator didn't include tests so that step fails. <mroh>civodul: I think emacs-dash needs a similar 'disable-byte-compile-error-on-warn phase/hack as in proof-general... Patch incoming ;) <rekado_>FYI according to pg_stat_activity we have 166 server processes at the moment, most of them idle. <rekado_>postgresql-service-type is configured with max_connections at 300. <rekado_>pg_stat_database says cuirass has 161 numbackends and 5 deadlocks. <ulfvonbelow>I'm trying to reconfigure my system with 'sudo -E guix system reconfigure /etc/config.scm', but I get 'Git error: repository path '/home/<me>/.cache/guix/checkouts/l74zueb3lgylhgxnuzx3d5fzraztxvzu2s4466wlqqvmz7hdct3a/' is not owned by current user' <KarlJoad`>ulfvonbelow: I got that before too. I was having other issues too, so I just reinstalled, but let me know if you find a better solution for that. <plattfot> I got a tiny function emacs-header-parse in my channel which I find useful when writing in guix.scm for emacs packages. I would like to push this upstream. But before I do, is this something that would be useful for people? <mekeor[m]><ulfvonbelow> "I'm trying to reconfigure my..." <- why do you use sudo -E why not just sudo? <ulfvonbelow>for now I managed to work around by setting 'git config --global --add safe.directory ...' for all the repositories it complained about <cwebber>civodul: hrm. have you tried using "guix deploy" since your fix? I'm getting the very error you reported and apparently fixed in bug#54812 ... but I'm confused why nobody else would have reported having the same issue. <cwebber>unless guix deploy isn't that widely used currently, heh <cwebber>ulfvonbelow: heh I bumped into that yesterday, was told -E wasn't needed anymore (if it ever was?)... I've always used sudo -E, don't know why :) <cwebber>but yeah that issue appears to be with guile-git <mekeor[m]>ulfvonbelow: did you try: sudo GUIX_PACKAGE_PATH=/foo/bar guix system reconfigure /etc/config.scm <ulfvonbelow>I tried 'sudo --preserve-env=GUIX_PACKAGE_PATH guix system reconfigure /etc/config.scm' <ulfvonbelow>and that gave `Git error: repository path '/home/reepca/.cache/guix/checkouts/dfwc22oq6jzs2attt3mirpdnqyzjmhoqkhu7xrh2vdub5zhwsoeq/' is not owned by current user' <mekeor[m]>iiuc, sudo -E guix will create root-owned files in your personal users ~/.cache dir <ulfvonbelow>'ls -l' ~/.cache/guix reports all of them as owned by me <rekado_>civodul: and now we’re at 185 connections. <mekeor[m]>ulfvonbelow: hmm, so "find ~/.cache/guix -user root" doesnt show anything? 
