IRC channel logs
2024-02-24.log
back to list of logs
<podiki>xsetroot i do in my stump config specifically. other things in xinitrc or zprofile or....etc. <Kingsy>podiki: well tbh it raises mroe questions this delcaritive config, what about picom? thats something that needs to be executed, I usually do it in xinitrc. I am guessing thats the same? <podiki>autostarting stuff can be done in so many places...xdg/autostart, with a utility like dex, in your wm config, ... <podiki>i try to put things where i feel like they go together, like stuff for a graphical session in a wm config; though it depends on what environment it needs sometimes and what is set up already <Kingsy>screw it. I am going to go back to xinit, and just try and fix the fonts and other things. I actually cant bare slim.. I am loosing sleep! hhaha :D <podiki>i did picom in a (when _initializing_ form in my stump config, as well as calling "dex -ae stump" to get xdg autostart files <Kingsy>I usually just set it in my xinitrc <podiki>main thing is make sure your XDG_ env is set up correctly, that should be most things like fonts, icons, etc. <Kingsy>oh I see the env vars now tbh. slim might have set them. Icould just copy and paste em <PotentialUser73>Is anyone here able to manually add a subject to an issue in the issue tracker? I'm in the process of submitting patches that I develop at my day job and since my employer's security permissions on email don't allow me to use git send-email I was manually sending plaintext messages with patch files attached and just saw on the tracker I apparently <podiki>lilyp: when i copy the same as glib I get a "cairo: unbound variable" when trying to build cairo... "define cairo" comes before cairo-with-documentation though. i'm missing something simple <PotentialUser73>Alternatively I can just resend the patch with a subject if that's easier <podiki>PotentialUser73: you can look up debbugs control commands, should be possible, i think "retitle" is the one <PotentialUser73>podiki I just looked at the docs. So do I just send an email as if I were commenting on the issue but in the body write the command "retitle bugnumber new-title"? <PotentialUser73>Sorry if that's a dumb question, I've not really done much with debbugs before <podiki>There's a different address to send control messages to I believe <podiki>control@debbugs maybe? I don't have it handy <podiki>debbugs had issues recently but i believe should be back <podiki>hmm it is texlive-bin's inputs that has cairo listed and is now not found... <PotentialUser73>I see in the debbugs browser that the retitle request went through properly. I assume it will still take a bit to propagate to the issue tracker, which is fine. <podiki>mumi is every 15 minutes or so i think <podiki>welcome! and thanks for the patches <Kingsy>podiki: you xperienced anything weird with alacritty on guix not detecting the config file? haha it seems everything I do is broken! :D <podiki>it really sounds like XDG_ isn't set up for you <Kingsy>podiki: I mean that is probably true, but how do I set it up correctly? <podiki>probably in your shell profile, as that i believe will be loaded on login before the xsession/wm? <PotentialUser73>podiki: Actually, one other question. I had a patch I had submitted a couple weeks back and was unaware that my employer's email would put a big confidentiality notice in the email body since it was being sent outside of our organization. I responded to the issue from a personal email explaining that that patch is not encumbered by any <PotentialUser73>confidentiality requests, but I can certainly resubmit it from a personal email that won't add that notice if you all need me to do so. Is there anything else I should do at this time? <Kingsy>podiki: you mean .profile? I don't have anything custom in this regard at all. <podiki>i can't speak to that (if those are even enforceable? who knows), you could ask on guix-devel or maybe someone here will know during a busier time <podiki>Kingsy: where do you have packages? in system config and default user profile? or multiple profiles? <Kingsy>system config and home config, I havent istlled anything with guix install <PotentialUser73>podiki: No worries, I'm just trying to not create any additional confusion/trouble for the maintainers when submitting patches. <podiki>well, make sure you follow whatever the manual says for guix home then, that should setup your shell and env <podiki>PotentialUser73: appreciated! i think that you already responded directly to that patch should be fine for now. someone can ask for more ifneeded <podiki>Kingsy: it should do it automatically if you manage your shell with guix home, if i remember. but if you don't then you need to source some stuff in your shell (.profile or similar) <podiki>sneek: later ask lilyp I'm running into an odd issue. with define cairo and setting it hidden and adding define-public cairo-with-documentation (remove hidden property), I get that cairo is unbound variable in other package's inputs. I feel like i'm missing something simple but i don't see what's wrong <podiki>Kingsy: good luck! i'm off for now. others use guix home more extensively and can help i'm sure. <lilyp>note that glib needs to be exported to prevent circles from breaking the build <sneek>Welcome back lilyp, you have 1 message! <sneek>lilyp, podiki says: I'm running into an odd issue. with define cairo and setting it hidden and adding define-public cairo-with-documentation (remove hidden property), I get that cairo is unbound variable in other package's inputs. I feel like i'm missing something simple but i don't see what's wrong <lilyp>I suppose the same is true for cairo then <podiki>well it just says unbound, but is that since it is just "define" and not "define-public"? i didn't think that was an issue. <podiki>i'll try exporting, but thought it might be a caching thing so now i'm rebuilding my checkout... <podiki>lilyp: i think you are right, still letting guix build but it doesn't warn about cairo in compiling package modules now. that's two I owe you! <Kingsy>podiki: thanks man! apprecaite all your help and responses <lilyp>you need to hide it with the hidden? property, but still export it <Kingsy>can someone do me a favour and pastebin the output of env | grep -i xdg <- please <Kingsy>oh no I think there is a problem with my xdg vars and I want to compare <lilyp>those have especially personal values – I'd recommend pasting them with $HOME replaced for example :) <Kingsy>personal values? they are paths.. <oriansj>Kingsy: I am assuming you are having the values set in .xinitrc ? or did you set them to be populated by .xsession ? <Kingsy>oriansj: I havent specifically set any xdg values anywhere. but the problem is my applications are not able to access .config files. so there is something wrong here. its like I am missing a step or soemthing. but I do have the variable -> XDG_CONFIG_HOME=/home/chris/.config <reedm>nckx: thanks for the response yesterday. bcachefs isn't in /proc/filesystems for me. i tried to load the kernel module via the "kernel-module-loader-service-type" but that didn't work. i forget the exact dmesg error i got from that, but i could reproduce it, if necessary <reedm>should I be doing this all through the config? i.e should i just be declaring /dev/sdb as bcachefs through the filesystems section of the operating system declaration? <reedm>ive got to run again. but ill check the logs if anyone responds. thanks!! <Kingsy>right, its dbus that is broken, is someone aware of a devel channel or something where I can get some detailed advice on dbus? <lilyp>How exactly is your dbus broken – did it not start, or…? <Kingsy>if I install dbus and run it manually via xinit then it runs but the DBUS_SESSION_BUS_ADDRESS is pointing to a directory that doesnt exist <Kingsy>meh this is really not worth it. I am going to just take a step away from guix I think. I have been battling for a week now. declaritive config is amazing. but there are alot of things that just are notfor me I suppose. one being the lack of freedom to run the OS as you want without going through hell <lilyp>perhaps you are making this slightly more complicated than it needs to be <Kingsy>lilyp: I don't understand how thats possible. I installed the OS. added stumpwm in the system config, xfce4-terminal is broken, I want to run my WM via startx, I cant without heavy deviation <lilyp>the typical advice is to start with a simple system and go from there – or to take what someone else has already configured <Kingsy>like I said, I don't know how much simpler it can be, I had about 30 lines in teh system config <lilyp>taking literally any login manager is simpler than manually tinkering with startx <Kingsy>lilyp: as I said, I did that. the system was broken, xfce4-terminal wouldnt save configs. <lilyp>you seem to have this problem with other terminals as well. Is your $HOME even writable at this point? <lilyp>anyhow, it's getting late. in case you got your config wrong, you might want to share it with a paste service; you can also try asking on guix-help <lilyp>the xfce4-terminal thing doesn't sound like a known bug to me – plus, since it's not just one terminal, and folks are known to use alacritty (not too sure about who uses xfce, but I'd also guess a fair amount), I'm pretty sure we'd know if they were actually broken out of the box <Kingsy>its ok, guix just isnt for me. thanks anyway <podiki>hrm. is there some issue if you inherit a package and use gexps in modify-phases but inherited package did not use gexps? shouldn't matter right? <podiki>ah, original package had no phases and didn't specify a default <spiderbit>Is it possible or better said what is the easiest way to compile on a strong machine a build for another machine? <spiderbit>on my thinkpad x220 icecat or firefox to build takes very long no interest in doing that there if I can avoid it... <podiki>well assuming you don't want to just use the substitute servers, i think guix offload is what you want <spiderbit>I would want to and I have them setup but somehow a few days old pull <spiderbit>maybe because I switched on that machine to gnome-team branch <spiderbit>well I think guix publish makes more sense setup looks a bit easier and I can have only run other build machine while building... <spiderbit>because I have a old i5 (x220) vs a ryzen 2700(x) I think the ryzen should be done 10 times faster <podiki>ACTION starts to build icecat with newer cairo... <echolib>Would really like guix with cinamon desktop. I use this DE for years, mainly thanks to its magnifier accessibility, and its settings around. Do you know about the work on having it ? <lilyp>Our cinnamon packages are fairly recent (5.6.2 vs. upstream 6.0.0) <lilyp>I sadly can't find any prior art w.r.t. cinnamon-service-type, but you could supposedly repurpose gnome-desktop-service <lilyp>In particular, over at gnome-team you might get away with instantiating a gnome-shell-service where the "shell" and "utilities" fields are replaced with useful values for cinnamon (i.e. cinnamon instead of gnome, nemo instead of nautilus, etc.) <bdju>will the yt-dlp package be updated sometime soon? I tried to update to a newer commit myself but got a build failure, something about a version conflict with websockets 11.0.3 and needing 12.0 or newer maybe <lilyp>python-websockets causes <400 rebuilds – you might update it to a newer commit as well and see whether it works <vivien>Dear guix, my excitement seeing all these juicy updates on gnome-team is tarnished by a depressing amount of “cannot build missing derivation” errors on CI. Is it a problem with the state of the branch? Can we do something to fix it? <bdju>hm despite updating python-websockets to latest release commit and then trying to update yt-dlp to latest release commit again, I got the same error in the build log implying I still had 11.0.3. <lilyp>Is there a good reason as to why python-scikit-build, when compiled locally for x86-64, should ever fail to find gfortran? <lilyp>or, well, to compile stuff with it <Guest35>hi I'm having some problems with guix os and java after installing the latest openjdk I still can't find any binaries for javac, should I install any other packages? <lilyp>use openjdk[@version]:jdk on the command line to get that or `(,openjdk "jdk") in scheme <Guest35>I'm still new to guix what does the : do in openjdk[@version]:jdk <lilyp>it separates package name from output name, openjdk:jdk means "take the jdk from openjdk" <lilyp>there are some well-known outputs such as "lib" for packages that provide both binaries and libraries or "doc" for packages with huge documentation <Guest35>I see, these names are also mentioned in the output field of openjdk, and the quasiquote syntax that you mentioned should be in my guix home profile if I want to do it in scheme right? <lilyp>or your manifest.scm – no need to go full guix home for that <Guest35>oh ok one more thing the about the package outputs, "lib" "docs" or "jdk" are these directories found in the root of the package in /gnu/store when it's build or are they special strings that are parsed and mapped to the correct files somewhere else <lilyp>they are built as extra folders under /gnu/store with their own hash <lilyp>for instance you have /gnu/store/xxx-openjdk and /gnu/store/yyy-openjdk-jdk (simplified) <Guest35>I see, thanks again that was a huge help <Kabouik>podiki: sorry I fell asleep in front of my computer and libusb issue yesterday, I didn't thank you for your help aand suggestions. I haven't solved it yet. Since this is urgent (already late on the project), I may have to use another distribution at least temporarily, before fixing my machine later. <civodul>cbaines: hey! nginx is now serving the wrong certificate for foundation.guix.info; any idea how that can happen or how to debug it? <civodul>i don’t think anything changed since last month <civodul>cbaines: did you reconfigure it on Thursday? <civodul>the config file looks completely bogus compared to what we have in Git: diff -u /gnu/store/z802b6mbzf6gvyhzxxnhh2ammza735f5-configuration.scm hydra/bayfront.scm <thewrinklyninja>Having issues with guix os install. I'm based in Aus but on every single try I always get errors that the substitute server is not responding in a timely manner and the install bails. Any advice? <civodul>ACTION reconfigures bayfront back to where it was <oriansj>then the worst thing that happens is you build the programs from source code <oriansj>and if for some reason those builds fail; then you have a bug to report <Zambyte>Does anyone have recommendations for making small config changes quickly? I am just trying to update my home-openssh-configuration, but now it's pulling over a GB of packages :\ <jpoiret>Zambyte: that's likely because you guix pulled in between, no? <jpoiret>you can roll-back your guix pull with `guix pull --roll-back` <Zambyte>That is likely it yeah, but it's a problem I run into often for some reason. Wondering if there is an easy way to use guix time-machine or something to avoid upgrading packages until later <Zambyte>jpoiret: I will have to try the guix pull --roll-back in the future, thanks for the tip. It took 40 minutes to update my ssh config ;( <PotentialUser-18>Greetings. Noob here, following the Reference Manual but with some questions: <PotentialUser-18>I have python3 installed, and type (from the Reference Manual) "guix shell python python-numpy". I see guix downloading python-3.10.7 (I already had installed) plus "python-minimal-3.10.7". It seems redundant to me. If it is not, could you please tell me why? <oleander>Hello everyone, are substitutes in Guix System disabled by running `# guix-daemon --no-substitutes`? How can I see whether they are enabled or not? <civodul>is someone still working on upgrading Emacs? <lilyp>civodul: It's already upgraded on emacs-team and I recently worked out a solution for #67260 <lilyp>Importantly, this fixes grafts for native-compilation. <lilyp>We still don't have build system support for specifying which emacs to use – bummer, I know – but we could do an early merge from emacs-team with 67260 if that's wanted <PotentialUser-18>If I type "guix shell python python-numpy", I see guix re-downloading python-3.10.7 (that I already had) as well as well as "python-minimal-3.10.7". Does anyone know why? <lilyp>python-minimal is used to build numpy, whereas python is the one you requested via shell <lilyp>as for why it's redownloaded instead of used as-is, there might be many reasons, but most often it's a recent pull or gc <PotentialUser-18>Thank you for replying, lilyp. Wouldn't the "proper" python be fit to compile? Or guix does not provide substitutes / alternatives? <PotentialUser-18>Now that I think about this a bit longer, perhaps it is a matter of ensuring reproducible results. Possibly the same reason python3 says it was compiled in 1970. <pkill9>guix is packaged in alpine, nice <lilyp>PotentialUser-18: there's also bootstrapping issues, but tbh I don't know why the python/python-minimal split is made <PotentialUser-18>What are the /var/guix/profiles/per-user/$USER/current-guix-N-link profiles? <civodul>PotentialUser-18: hi! these are profiles created by ‘guix pull’ <PotentialUser-18>Oh. I see. I was a bit confused because, under "share" I see references to emacs and zsh (that I installed) and fish (the shell? I did not install that!). <civodul>you can try ‘guix pull -l’ to see more info <lilyp>Not pushed yet, no. You'd have to apply it locally <the_tubular>Gotcha, I'll wait a bit, seems like a good read for newcomers <podiki>lilyp: thanks again on cairo yesterday. built fine here, built icecat as a test and it ran fine too (with new libdrm and mesa as well) <apteryx>civodul: a rationale for the pkg-config -> pkgconf should soon reach your inbox :-) <picnoir>Hey Guix. I sent a message to the guix-devel@gnu.org ML this morning (8:26 UTC). It seems like it did not reach the list yet. I was not subscribed to the list when I sent it, could be related. Should I wait more or should I assume the message has been silently dropped and re-send it? <apteryx>picnoir: initial messages sent to the GNU mailing lists are moderated <nckx>picnoir: If your message doesn't hit the list within the next few minutes, it wasn't received. <nckx>A subscription is not required to post. <nckx>I discarded one duplicate message. <nckx>Re my 'next few minutes': it might actually take a bit longer in some cases, and the Web archives are updated only several times an hour. <picnoir>ah, apologies if I was the source of the duplicated message, it was unintentional, I thought I sent it only once. <picnoir>Alright, I'll see if it lands tomorrow. <nckx>I don't know which message was yours, but it was over 2 days, so probably not. Regardless, no apology needed :-) <podiki>i got a guix-devel message from a "picnoir" (about nscd), so if that is the one it did go through to subscribers <gnucode>I kind of wonder how hard it would be to package hyperland...probably pretty hard. <gnucode>actually maybe not. It's written in C++ <gnucode>I thought it was written in Rust for some reason. <jackhill>gnucode: it wasn't me, but I was just reading about it in my mail and then saw your question :) <ieure>So, when I run `guix style', that formats my package in a way that makes `guix lint' complain, mostly "line 1234 is way too long." Should I be deferring to `guix style' and ignoring the lint warnings, or correcting the `guix lint' warnings and not running `guix style'? <ieure>`guix style' also leaves whitespace on an otherwise-blank line. <ieure>Guessing I should follow `guix lint', but it's irksome that the Guix tools don't agree with each other. <podiki>gnucode: several minds on hyprland these days :) <podiki>guix style is not perfect, i would make sure guix lint is good (or mostly, it can complain on something you can do nothing about like an input name not matching) <podiki>in great news for hyprland, i now have emacs keybindings everywhere so i can stay in hyprland <lilyp>ieure: Defer to `guix lint'. There were some style improvements suggested on the ML, but they sadly haven't landed yet <spiderbit>I try to use guix offload but get ther error message GSH ERROR The file does not exist or permission denied: /root/ssh/id_rsa <spiderbit>What I don't understand is why it searches for rsa when I just created new keys and by default they are not rsa <spiderbit>I did not do the guix archive --authorize yet, because it doesn't state in the manual where I find the "master-public-key.txt" <spiderbit>so it probably has to do with that, but still I find that error message very strange <spiderbit>but I added that content already to the machines.scm file so it would be strange that it needs it again... <spiderbit>this should generate it but it's not described where I would find that file then... because it's not in the CWD <spiderbit>The generated key pair is typically stored under /etc/guix <spiderbit>ok did use the guix --archive authorize on both machines but still the same error