IRC channel logs
2022-11-13.log
back to list of logs
<vagrantc>admittedly, one is inscrutible, and one is almost understandable :) <awb999>I have been running gnome with gdm wayland. <awb999>something happened, and nwo when I do inxi -G it shows me that I am on x-windows. <awb999>could it have anything to do with this? <unmatched-paren>awb99: you sure you didn't accidentally switch to gnome from gnome (wayland) in the gdm session menu? <jab>I'm looking to upgrade to linux 6. I figured that I would wait until most people have upgraded to iron out any bugs, which there probably are not any. <awb999>unmatched-paren .. good point, you mean in the login dialog I accidentially switched the entry... <vagrantc>is it guix build --target= ... that is used for cross-compiling? <awb999>so I have logged out, and logged in again; it was selected Gnome (Not Gnome Xwin) <awb999>if I would have selected by mistake gnome xwin, <awb999>then it would have used after logout by default also gnome xwin I guess. <awb999>thanks unmatched-paren - I figure out something like that has happened... <awb999>or some app changed gnome to xwindows. <awb999>twopubsolar: I saw when I was in xwindows in gnome settings only x-windows. <awb999>nothing special to select to be able to switch between wayland and x windows. <twopubsolar[m]>couldn't use more than 2 fingers on the touchpad, saw X11 in settings, logged in and out without changing anything, and it was woring wayland again <awb999>does someone know how I can activate the kernel module br_netfilter ? <awb999>this is needed by docker to make a bridged network, <awb999>I looked into the git repo of guix, and see that netfilter gets compiled. <awb999>But I guess I have to set some kind of parameter... <awb999>lsmod |grep br_netfilter gives me back nothing. <awb999>modprobe does not need sudo it seems. <awb999>with modprobe, I get the name br_netfilter back <awb999>so I guess this means that this kernel module is available. <awb999>I called sudo modprobe br_netfilter <awb999>And now it appears in lsmod |grep br_netfilter <awb999>so I ugess with modprobe without -R I can load it... <awb999>now I only have to load the module on guix startup time... <awb999>so how do I load this kernel module at startup time? <awb999>I was looking in the wrong place .. in the os defintion in kernel-loadable-module <awb999>and there I could not add a string .. <awb999>to me your alias is just twopubsolar[m] <jts>hey y'all. I'm trying to try out Clojure and it seems there's a bit of complexity around the Clojure and Java ecosystem (shocking, I'm sure). Does anyone know a list of packages I can install to get, like... a basic core installation? <awb999>in there you see my clojure packages I install. <awb999>took me some time to figure out how the java stuff is installed.\ <awb999>I am using babashka from binary download as well, <awb999>this package is defined in this repo under awb99 packages babashka. <awb999>the gnu version of clojure is typically older, and is missing clojure-tools. <jts>this channel is for discussing gnu guix iirc ;) but thanks for the link I'll have a look <jts>ah, and it immediately answered my question: I need the jdk output. thanks! <user>guix openjdk package include javac? <jts>not seeing a binary for that in the result of `ls $(guix build openjdk)/bin` or `ls $(guix build javacc)/bin` but the former has a java binary and the latter has a javacc binary <awb999>I only have the packages in my list, <awb999>and which javac returns me something. <awb999>openjdk is currently java skd 18\ <jts>ah, yes, it's in the jdk output <user>So its "guix install opendjk18:jdk <jts>guix install openjdk:jdk i believe <jts>openjdk currently points to openjdk 18; if you want to specify that version specifically it would be like openjdk@18. not sure if that comes before or after output selection though <mirai>how do I debug a service definition that hangs shepherd after boot? <lechner>mirai: is anything useful in the logs after the next successful boot? <mirai>lechner: it just hangs after either "generating machine-id or ssh-keygen" <lechner>mirai: maybe dropbear works better for you, or perhaps you can generate the server keys manually <apteryx>can guile be used as a replacement to apache to serve basic, static web pages? <apteryx>(do we have a service for that in Guix?) <apteryx>mirai: if it's your own definition, it's best to write a system test for it <apteryx>instead of locking up your host system :-) <apteryx>sneek later tell civodul yes it sounds like the same bug! thanks for sharing the workaround, even crude <apteryx>rekado: re SAN, that'd explain things! <apteryx>would multipath really safeguard against this kind of problem? <apteryx>ACTION just packaged c-intro-and-ref if anyone is keen on learning on C from the comfort of their info reader <ghostarchist[m]>This is a static site generator written and configured in guile, doesn't look like it needs apache and not sure about a service for it in guix <yarl>I am trying to use emacs + geiser to play with guix sources but I am confused. I don't know really how to do to have the equivalent of guix repl. <yarl>For example, I would like to hack on (guix import texlive). I wan't to do that because I tried to import a latex package and it failed with "failed to download description...". I would like to know how to hack this myself. <yarl>I tried to open the file that contains the module (guix import texlive) and use `geiser-load-current-buffer` but it fails with "no code for module (gcrypt hash)". And it seems that's not always the case. <yarl>Now if I try to load the current buffer it says no repl is running, but it is. <yarl>Ok, "no code for module (gcrypt hash)" error again. Is it known? <yarl>I opened guix.scm then `geiser-mode-switch-to-repl-and-enter` then I did `,use (guix import texlive)` then `,in (guix import texlive) (tlpdb)`. after a while, I got a &package-input-error. <yarl>I have a few questions. What populates /run/current-system/profile/lib/guile/3.0/site-ccache with .go files? <yarl>Is it possible to use emacs + geiser to directly try guix code, even the code involving the store? <yarl>(as what I tried with (tlpdb)) <yarl>How are you guix developers hacking on guix? <yarl>I read the guix manual and the geiser manual but I don't find these particular uses very clearly explained. <yarl>Am I alone in this case? <euandreh>yarl: what populates $PROFILE/lib/guile/3.0/site-ccache is the guix package <euandreh>when doing a "guix system reconfigure", compiled guile packages get installed there <euandreh>to get rid of those, you can uninstall the relevant guile packages <yarl>euandreh: Ok, I asked that because when geiser loads code, it complains about source files newer that compiled files. <euandreh>you can use geiser to try guix code directly <yarl>I don't know, maybe I misconfigured something. I can execute some code, for example in (guix import texlive) I can run the simple function "guix-name". <yarl>But, for example, as I explained, (tlpdb) fails. <yarl>I also set up geiser-guile-load-path to a git worktree pointing to the commit of my running guix (as per `guix describe`) I assumed this is important. <yarl>If you have any insight, I'll take it :) <unwox>hi guix! is it possible to configure autologin for greetd + agreety combination with existing system services? <yarl>Aside, wouldn't it be nice that the guix package (or a separate package) installs the guix source somewhere so we could point geiser-guile-load-path to this? <yarl>Or maybe that's stupid, I don't know. <unmatched-paren>unwox: iirc there's a way to configure autologin for greetd but our greetd-service-type doesn't support it <unwox>ok i'll take a look at source code a bit later then. thanks! <yarl>unmatched-paren: Did I miss your last message? <yarl>Ok, I set up geiser-guile-load-path to "~/.config/guix/current/share/guile/site/3.0". I opened ~/.config/guix/current/share/guile/site/3.0/guix.scm". still "source file newer than compiled". I carry on. <unmatched-paren>yarl: ah, you also need to add ~/.config/guix/current/lib/guile/3.0/site-ccache to guile's load-compiled-path <yarl>unmatched-paren: Hmm. One moment, I am a bit confused. so, as euandreh said, the guix package populates $PROFILE/lib/guile/3.0/site-ccache. What's taking care of ~/.config/guix/current/share/guile/site/3.0 ? <yarl>unmatched-paren: I just checked, it's not installed either in root or user profiles, home-configuration or /etc/config.scm. <yarl>Still, $PROFILE/lib/guile/3.0/site-ccache is there. <unmatched-paren>Because that contains any compiled Guile modules, not just Guix modules. <yarl>unmatched-paren: Hmm, ok <yarl>Let me set guile's load-compiled-path. <yarl>unmatched-paren: Wait, sorry but how do I " add ~/.config/guix/current/lib/guile/3.0/site-ccache to guile's load-compiled-path"... <unmatched-paren>yarl: i'm not sure with geiser specifically, but there's the GUILE_LOAD_COMPILED_PATH variable <yarl>unmatched-paren: Ok, did it in ~/.guile, restarting emacs. <yarl>unmatched-paren: Ok, I don't have "source file newer..." now. But something is rather strange. If I enter (guix import texlive) and try to use the function "guix-name", the number of arguments does match with the declaration of the function. <yarl>i.e. (guix-name "something") says "Wrong number of arguments to #<procedure guix-name (prefix name)>" <yarl>But in "~/.config/guix/current/share/guile/site/3.0/guix/import/texlive.scm" : (define (guix-name name) ...) <yarl>unmatched-paren: Do not be sorry, thank you for taking the time, at least I got clues thanks to you. I need a break anyway. <brendyn>yarl, there is a procedure in (guix import utils) that your are using instead most likely <yarl>brendyn: I see. this is strange because I changed to module (guix import texlive). <yarl>I thought I would be able to use code in this module. <yarl>unmatched-paren: I just tried and this does not work :( <yarl>unmatched-paren: The other guix-name is still shadowing it. <yarl>That's strange because before I add .config/blablabla to load path I could call guix-name. <anschlussy>i must drag up all of the old gods, all of the world, and purify my mind with the pursuit of guix <anschlussy>i will read the manual after this youtube video, i will <yarl>I'm gonna eat (maybe take a nap...) then try to clean my config since the old load paths are still there. <unmatched-paren>anschlussy: chapters 5-13 and 22 are the ones i think you should read <itd>unmatched-paren: Would you mind telling me the title of chapter 5-13? (I can't seem to find it.) <itd>paul_j: Thanks. I misread the range 5 to 13 as index 5-13. :) <anschlussy>funny, as a little bit of a story i tell myself: recently, i've been more and more drawn to lisp, and even before i knew that the common method of delineating was hyphens, i switched from spaces/underscores to hyphens in my folder and filenames; kind of a lovecraftian moment, where the cthulhu of lisp was invading my dreams <a12l>When setting up a Guix home config from scratch should I plan for using multiple files, or is one `home.scm` file enough? <mirai_>apteryx: what's this "system test" and how do I write one? Are there any existing examples? <mirai>yes, it's a service definition that I'm writing here for upstreaming <a12l>I get a backtrace when I run `guix home list-generations`, and I think it's caused from that I don't have any generations. A message saying explicitly that you don't have any home generations currently would be preferable (I think). <a12l>I've looked at the issue tracker, but not sure if I'm missing something (don't know what the "standard way" is to name issues in the Guix community). <bost>Hi. Where is log file with the output of `guix pull`, `guix upgrade`, etc? <sterni>do I have it correct that you don't have GNAT packaged? currently looking for a better way to bootstrap it in nixpkgs <nckx>mirai: See gnu/tests. System tests run a full Guix System (VM). <nckx>sterni: Correct. We're unaware of any way to bootstrap Ada at this time. <sterni>nckx: I figured as much; sadly also the situation is getting kind of bad even if you accept a binary bootstrap seed (as we do in these situations) <nckx>sterni: How does Nix do it? You said 'better', but maybe I seek to much behind that. <sterni>before that we used a gnat 6 bootstrap seed published by Gentoo, but seems that they switched to using AdaCore's gnat-gpl fork over the gcc gnat proper <sterni>but their binaries are a huge pain to package in a non fhs/sandboxed build iirc <mfg>Is itpossible to use a video device in an `guix shell -C' environemnt? if so, how do i expose it? <apteryx>mirai: there are many examples under gnu/tests/ in the source <pkill9>twopubsolar[m]: possibly network issues,iget broken pipes with network issues <mfg>Ok, got a bit further. Now my question is: Which package provides the dri driver for amdgpu? mesa seems to be missing it. Or is this left out intentionally? <apteryx>do we have offline docs for the linux kernel? <apteryx>I guess these docs are under doc/ in the source tree of linux; I'll see if we can build them <apteryx>with info going to "out" (if not overly large) and html to "doc" <apteryx>ghostarchist[m]: I know about haunt, the guix web site uses it, but it doesn't use guile to host it (it serves it via nginx) <PotentialUser-30>Hi, crafted a variation of gst-plugins-good (it's build-system meson), but after compilation a check failed:?? (99/111 elements_souphttpsrc3 FAIL exit status 14) <apteryx>the guix way would be to investigate the failure, and if it's not caused by something on our side (such as failure to patch FHS stuff), report it upstream, then disable that single test in a phase in the guix package definition, with a link to upstream issue. <PotentialUser-30>I already tried to "disable that single test in a phase..." (i copied stuff?? gstreamer.scm) but didn't right.?? Is there a good tutorial / howto about "hacking phases" in meson? <zamfofex>Hello, Guix! Has anyone here been able to use ???--sysroot=??? with Clang? I feel like the patche applied to it in Guix prevent it from being usable. <apteryx>zamfofex: never tried; I just submitted clang@15 to guix-patches yesterday, built with DYNLIB, not sure if that could help, but I guess not <zamfofex>I see. It would be neat if that were a use???case tested, given that cross???compilation is a significant feature of Clang. <zamfofex>Whenever I tried using ???clang --target=wasm32-wasi --sysroot=.../wasi-sysroot???, I never got it to work. <zamfofex>It seemed to be pulling in system headers anyway. <apteryx>can't 'mg' navigate to a line number? <zamfofex>Also, on a similar note: How does cross???compilation work in Guix? Because if I remember correctly, GCC isn???t built with multilib support. <apteryx>a GCC-based cross-compiling toolchain is built and used <zamfofex>Is there any way to build such a toolchain for e.g. installing it to one???s profile? I remember when I wanted to compile files to 32???bit x86, I had to install a GCC that also ran in 32???bit. Which was fine, but it makes me wonder whether it???d have been possible to run a version of GCC for x86???64 that can compile to 32???bit x86. <apteryx>I don't think so; you'd have to write a profile which invokes the procedure to derive the cross-gcc-toolchain package <apteryx>(so probably possible, but no readily available from the command line, as far as I can tell) <apteryx>hopefully someone who knows more about this can tip in <apteryx>is TRAMP broken in Emacs for someone else? <apteryx>C-x C-f /ssh:my-host: RET => File not found and directory write-protected <zamfofex>apteryx: Ah, that???s fair! On a related note: That GCC is unable to compile to Wasm feels very unfortunate. I know that someone who goes by ???pipcet??? once started work on porting glibc, coreutils, binutils, etc. to Wasm, and they also took the time to add a Wasm backend to GCC. <zamfofex>Now, the whole thing was tied together using Perl scripts and this weird JavaScipt + C preprocessor hybrid, so it was kinda hacky. But I wish someone knowledgeable would polish up their work for merging upstream, at least for GCC. <zamfofex>When I say ???someone knowledgeable??? I just mean ???more knowledgeable than I am???. Of course, pipcet is ver knowledgeable themself. ???? <lechner>nckx: Hi and sorry I never got back to you on XDG_RUNTIME_DIR. Has it been a problem outside of Guix Home? <lechner>apteryx / Hi, I see the same but have otherwise not used TRAMP before. <rekado>connection to debbugs times out, so we can???t sync issues.guix.gnu.org <gnucode>I just found out that guix had packaged zig! That's awesome. I wonder when someone will package river wm. <a12l>I want a permalink to GUIX's `.profile` file. Currently .profile -> /gnu/store/dann7r1095xll0kji5yl0ql07096rc8j-shell-profile, but I've been trying to figure out how to do look up what symbolic links exist to link to a path. <nckx>a12l: What is a 'permalink'? <crono>Does anyone get a gexp symbol error for emacs-guix compeletion or guix shell commands? <crono>The error 'Unknown # object "~#"'. Im trying to see how to load gexp in the Guix Repl automatically. Would I just put use-modules in my .guile? <a12l>nckx: Something similar to `/var/guix/profiles/per-user/a12l/guix-home`, a link that points to the current version of a component in the store. <nckx>a12l: Why not link to .profile though? <nckx>There is no (other) way to make Guix update a symlink for you, and no (sane) way to ask for all symlinks to x, although the latter might not be what you meant above. <a12l>nckx: I'm unsure what you mean; but my problem is that the link `~/.profile` is removed everytime I reboot my system. <nckx>crono: I searched issues.guix first but found no report. <a12l>I have scripts that create symbolic links that I specify, but then I need to know where the `~/.profile` symbol link need to point to. <nckx>I am also unsure what you mean, so we are matched there. <a12l>The TL;DR. is that to lessen the amount of state held by the computer. <crono>nckx: I havent filed one, and I havent seen anyone else that has this problem. Its just the invoking M-x guix-command or the guix completion has a gexp symbol error, like the module wasnt loaded. I should just be able to load gexp module. <nckx>a12l: A ~/.profile isn't created or managed by Guix, but it is one of the services offered by ???guix home???, so I guess the supported (if not fastest) way to recreate it is ???guix home reconfigure???. <nckx>crono: So silici???us' problem is different? <nckx>Anyway, I can reproduce it, it's not you. <nckx>I've never used M-x guix-command is all. <crono>could you post the link I must not have seen it. <nckx>It's in the scrollback. I switched clients, so it'll take a sec for me to find it again. <nckx>logs.guix.gnu.org/guix/2022-11-08.log#204945 <nckx>That's what I get and what it sounds like you get, so I assume it's the same. I just wasn't at a computer with emacs when I first linked to it above. <crono>nckx: Yah im trying to get more of an emacs workflow for most things. After Geiser Debug workflow ive been addicted <nckx>sneek: Later tell crono: Would you have the time to report this bug to bug-guix at gnu dot org? It's beyond my abilities, since I assume it's some reader thing ((guix gexp) is exported just fine by (guix scripts deploy)), which are beyond my abilities. <nckx>sneek: ???never mind. Botsnack. <crono>nckx: sorry emacs keeps crashing for some reason. I sure can, this is a worthy bug for my first bug report :). <sneek>crono, nckx says: Would you have the time to report this bug to bug-guix at gnu dot org? It's beyond my abilities, since I assume it's some reader thing ((guix gexp) is exported just fine by (guix scripts deploy)), which are beyond my abilities. <nckx>Sorry for the noise. And the typo. <a12l>nckx: The problem is that the .profile only begin read at log in, and if it's missing I need to run `guix home reconfigure ...` and then logout and log in again. <nckx>OK. So you can create symlinks before log-in, but not run arbitrary commands? <nckx>I guess that rules out getting ???clever??? with ???guix home build???. <nckx>However, while I'm not the resident Guix Home expert, this really does sound unsupported. <nckx>If you can't run ???guix home reconfigure??? at the right time (before log-in), you *must* carry some state over reboots somehow. So why not just carry the end result??? <itd>unmatched-paren: how is emacs launched? and why would emacs's output be in that file? <itd>unmatched-paren: Thanks. Does the problem ("emacs behaves strange") persist when launching emacs manually, e.g., in a terminal? <unmatched-paren>the service is almost certainly not the problem, as it works fine with my previous require-based config <itd>unmatched-paren: Then set the load path as the service would before starting emacs? Even while working correctly, the service adds complexity. And since something is amiss, I'd try to simplify. <unmatched-paren>I found the mistake; I'd written ``rainbow-delimiters'' instead of ``rainbow-delimiters-mode''. <itd>Interesting root cause. Happy to hear it's working now. :) <raghavgururajan>sneek, later tell gnucode: Regarding mail service, feel free to use GuixRus channel as test bed. <a12l>nckx: I can only keep state (i.e. files) in selected directories. To save .profile would also mean that I need to keep the state directly under `~/home`, which one of the most common places to keep state. <nckx>a12l: Could you fiddle with ???HOME=/some/saved/directory guix home reconfigure ??????? The activation scripts (and they are scripts, there's no /gnu/store/???-my-home directory with all the symlinks) honour $HOME. <nckx>sneek: later tell mothacehe: I've updated & restart cuirass-r-w on both POWER nodes. I'll init your password tomorrow. <nckx>sneek: later tell mothacehe: ???which is Monday :) <sneek>Welcome back gnucode, you have 1 message! <sneek>gnucode, raghavgururajan says: Regarding mail service, feel free to use GuixRus channel as test bed. <gnucode>I was actually thinking about doing that! <jorge[m]>Hi, Gpaint for some reason I can't download and it's not in Guix. Can you help me. <pkill9>jorge[m]: what specifically do you need that gpaint provides? <jorge[m]>just copy, paste, delete, change color is something easy and fast.