IRC channel logs

2019-06-04.log

back to list of logs

<sebboh>At home, I don't have 32GB of ram at home, I meant to say. :) Yes, I switched to hyperV from virtualbox. I like that I can, if needed, run inside a hyperV guest something like a device emulator which requires virtualization stuff. Can't do that under virtualbox. Meanwhile I guess I sort of miss vbox's guest<-->host file sharing, though it was utterly broken with regard to permissions and handling of certain
<sebboh>errors. But I digress.
<emyles>bavier: I've just had a quick read of derivations which makes things clearer for me
<emyles>A while ago I saw similar strange behaviour when I changed the hello world example to use different source, but can't remember the details.
<emyles>Thanks for your help.
<bavier>emyles: np
<reepca>emyles: one important thing to note is that the hash-part (and thus ultimately store path) of sources is based solely on things like the name and contents. Importantly, it's not based on where the source comes from. And origin objects have hashes specified with them. So if you specify a different source for the source, but leave the hash unchanged, it will happily use the old source still in the store since according to the hash they
<reepca>should be identical.
<reepca>I'm not sure if that's what's happening here, but based on gc-ing some sources changing what happens it sounds sorta like it
<emyles>reepca: that sounds a lot like it. I was leaving the same hash there hoping to change it to whatever was suggested instead of finding out myself.
<reepca>emyles: haha, that's exactly the same lazy approach I took. You can just temporarily change one character of the hash to something random and it'll report the mismatch.
<dongcarl>Hey all, anywhere where I can see reproducibility statistics for Guix packages?
<daviid>fwiw, someone should update the guile-gnome package description and pickup the description that's in the README, the descrption that is there is almost 10y outdated ...
<str1ngs>daviid: I took a quick look at guile-gi . and janneke seems to have make some progress with guile-gnome and webkit.
<daviid>str1ngs: yes, indeed, we just spoke about this in #guile
<daviid>str1ngs: however, you saif you need d-bus api, that would rpresent an unbearable amount of work, to bind gnome gio for guile-gnome .. hence my suggestion to use guile-gi instead
<str1ngs>daviid: yes with newer version of GTK. I have a link that explains the DOM interaction
<str1ngs>specifically this is for interacting with the DOM. ideally when creating a web application for introspection
<daviid>str1ngs: and not to bind it manually yourself either, my 2c, but admittedly, guile-gi is new, and somewhat a bit 'fragile' still
<str1ngs>daviid: this is all good information.
<daviid>which is why janneke went back o using guile-gnome .. I beleive though, that the guile-gi's author will get back and solve the problem, let's see
<str1ngs>here is the reference to DOM lower page of. https://wiki.gnome.org/Projects/WebKitGtk/ProgrammingGuide/Cookbook#Dealing_with_DOM_Tree
<str1ngs>essentially you need to use extension with a dbus server and client
<str1ngs>Epiphany browser probably has the best working example that I've found so far.
<str1ngs>alternatively you use javascript. which is terrible :P
<str1ngs>older version for gtk webkit do not have this issue
<str1ngs>my thinking is to use sxml to autogenerate html. then wire javascript functions to scheme procedures. this should in theory create a two way interaction between a webpage and scheme.
<Sharlatan>Hi
<daviid>str1ngs: you really want to talk to janneke and mike (the guile-gi author), because they both are working on a way to use webkit2, they both are experienced (very experienced) guilers to ..., let them know you are trying to ... here is the email thread where they were talking about this: http://lists.gnu.org/archive/html/guix-devel/2019-06/msg00001.html
<str1ngs>daviid: thank you I will look check it out. I have talked to janneke a bit. my concern here is I'm going the wrong direction with nomad. or there is duplication of effort.
<daviid>str1ngs: shim-in that thread :), and let them know you are interested to get a working webkit2 and get some advice from them ...
<str1ngs>the segfault is probably thread related, that's a real issue with gtk
<str1ngs>daviid : and g-golf is going to be the higher level library in the future?
<daviid>str1ngs: I don't think gtk has 'thread issues', it is by def single threaded ... I think it is a bug in guile-gi, but read the other msg there, the idea is to coordinate you effort .. and not re-invent the wheel ...
<str1ngs>well it's a thread issue if you don't call something on the ui thread I meant.
<daviid>guile-gi and g-golf are diff in the way they approach the GI binding 'problem', but they should both work, at some point ... guile-gi is usable right now, g-golf eeds nother few months of work ...
<str1ngs>seems g-golf would be a logically evaluation of guile-gnome? and would help with a move to gtk3 I guess?
<str1ngs>evolution I meant
<daviid>I suppose you could say that, g-golf is highly inspired by guile-gnome indeed, and extenively uses goops, but they will be quite a few diff to ...
<str1ngs>okay thinks for the information. this is all very exciting :)
<str1ngs>I've been using C and even experimenting with QT and QML with nomad.
<str1ngs>but all this other work, looks like a more pure scheme way to deal with things
<str1ngs>but with all*
***emacsen_ is now known as emacsen
<bluekeys>Hi guix. Does anyone know why stumpwm is telling me that ql-quickload is undefined. Scheme and lisp newbie here.
<emsenn>Hi y'all, I'm attempting to create a package for a piece of software implemented in Golang, and I'm running into an error I can't move past, regarding permissions during the build phase of the package build. I've tried to outline things here: https://wordsmith.social/fgaqsat275 - Any help y'all could offer would be greatly appreciated.
<str1ngs>emsenn: you are on the right track. there is a go-build-system
<str1ngs>emsenn: how many go imports does this package have do you have an idea/
<str1ngs>?
<emsenn>I'm not sure how I would learn that. I know there's a go-build-system but the software uses make & make install as its build commands, so I thought using the gnu one would be correct. I'm quite ignorant about most things to do with Linux systems, so was probably wrong.
<emsenn>(I also don't know Golang, or... much about any of things I should know about before attempting to do this sort of thing, but I've got to start somewhere, I suppose.
<str1ngs>yes, that's understanable.
<str1ngs> http://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/ipfs.scm#n219
<str1ngs>is an example go binary package I found. keep your source field the same. but the rest of this package should help you
<str1ngs>also you do not need to reset the timestamps
<str1ngs>this is the only command package I could think of
<str1ngs>the problem you might have though is if writefreely has upstream imports they wont be found just by cloning.
<str1ngs>emsenn: honestly if this is just for personal use. its easier to install go with GOPATH and do go pull
<str1ngs>sorry I meant go get .
<str1ngs>emsenn: wait is uses make. thats odd
<emsenn>That link is helpful; thank you! I don't understand what you mean about reseting timestamps? I'd like to know how to make Guix packages for other things, too, so I'd rather do this Proper than just however works.
<str1ngs>on sec emsenn
<emsenn>I'm not in any rush; I'm happy to be getting such prompt help!
<str1ngs>the issue with go and guix is, that if the package has alot of go imports. you need to make a package declaration for them.
<str1ngs>so you might have a large dependency tree
<str1ngs>emsenn: I'm just checking this package in terms of the make issue
<str1ngs>that is generally not normal for a go program
<emsenn>Haha, I didn't think so but I've never built a Go program before today; I think the only two I've used are Hugo and this one.
<str1ngs>so go-build-system might not work. you might have been right to use gnu-build-system
<str1ngs>luckly for you I'm a golang contributor :P
<emsenn>That's what I thought, that being why I did it, but I think because it is golang under the Make, it's running into some issue with Guix's, uh... sandboxing? container-ing? Build purity, that concept.
<emsenn>That /is/ lucky!
<str1ngs>to do this the guix way you would have to make a package for each of this. https://github.com/writeas/writefreely/blob/develop/go.mod
<str1ngs>and then a package for each of there dependencies.
<emsenn>Oh my.
<str1ngs>the makefile basically calls go build. but it uses the new module system. so it pulls the dependencies
<str1ngs>and because $GOPATH is not right it breaks. so maybe with a TMP GOPATH this might work.
<str1ngs>but your kinda not making this functional at that point. though the go modules system is pretty strict
<emsenn>But the right and proper way would be to rewrite the Makefile inside the pacakge declaration, and then use go-build-system - and write any necessarily other packages, too. Am I following?
<emsenn>(Package declaration might be the wrong term, the stuff inside (define-public writefreely))
<str1ngs>you don't need the makefile to build this
<str1ngs>the makefile just has extra things. you could just do go get and it would build fine
<str1ngs>atleast that gets it working for you for now
<str1ngs>I would start there, and I'm not sure if Guix even supports the new go modules system.
<emsenn>Hmm, I think I follow. Thank you.
<str1ngs>I know that was not encouraging
<civodul>Hello Guix!
<janneke>Hello civodul!
<roptat>hi guix!
<civodul>hey roptat!
<janneke>hey roptat!
<civodul>gentlefolks!
<civodul>please grab substitutes from https://qualif.ci.guix.gnu.org and let me know how it goes!
<civodul>(there's no proper TLS certificate but hey, why bother?)
<jonsger>civodul: what is the diffence to ci.guix.gnu.org?
<g_bor[m]>hello guix!
<g_bor[m]>I am trying to create two channels, and one depends on the other. However it seems, that the channel that needs to find the module from the other channel does not find it. How can I make sure that the channel finds the module defined in the other channel?
<civodul>jonsger: it provides lzipped substitutes in addition to gzipped substitutes
<civodul>should have mentioned it
<civodul>g_bor[m]: the "Declaring Channel Dependencies" section of the manual has something on this topic
<g_bor[m]>civodul: thanks, I am just reading that section :) I asked too early :)
<jonsger>ah, thanks for the clarification :)
<civodul>g_bor[m]: you said you experienced "password locking" issues, IIRC
<civodul>could you contribute to https://issues.guix.gnu.org/issue/35996 ?
<civodul>i really fail to understand under what circumstances that could happen
*civodul goes afk for a bit
<pkill9>does anyone know why this package isn't building with c++11 support and is failing? I added a config flag but it doesn't work: https://paste.debian.net/plain/1085981
<pkill9>it can be built with `guix build -f <that-file>`
<rekado>pkill9: does the build system override this value somewhere in CMakeLists.txt?
<efraim>Might be interesting to make snap a 'guix pack' target https://docs.snapcraft.io/t/pre-built-apps/6739
<pkill9>rekado: ah yes it does, and it also appends "CXX_FLAGS" to it which you can set, so I just used "CXX_FLAGS" instead of "CMAKE_CXX_FLAGS" and now it's compiling \o/ thanks
<pkill9>well, it appends it to some custom variables, not that particular one
<einarelen>Hi folks, I'm trying to build the guile-emacs package with -verbose=2. It has now been stuck in the step PROCESSING OKURI-NASI entries at 80% for more than 24 hours. Is this to be expected? I've tried to find answers to what's going on but most posts reference compiling guile-emacs on MS Windows. If it is common to take this much time, is it a step that can be skipped? Cheers!
<einarelen>I have guile-for-guile-emacs installed already if that makes a difference :)
<Gamayun_>What's the best way to apply a patch file to a local package?
<jonsger>efraim: could be a good GSoC or Outreachy project
<jonsger>I don't know how complicated or difficult snap is, but the project would have a very clear scope and target :)
<fishinthecalcula>Hello everybody, during the building of an image USB image launched with `guix system disk-image install.scm` I always get this error ( I quickly translated from italian the message could be slightly different)
<fishinthecalcula>guix system: error: canonicalize-path: This file doesn't exists "/gnu/store/cwlghngrh03igf8cfsp2mf49c2l9fnf5-guix-1.0.0-1.326dcbf/share/guile/site/2.2/gnu/installer/aux-files/logo.txt"
<fishinthecalcula>I was trying to create a custom USB image with some packages definition I created for testing purposes
<fishinthecalcula>How is that even possible? Shouldn't the hash in the name of the directory include that file? Why didn't Guix complain during installation?
<fishinthecalcula>Has anyone ever encountered this error?
<fishinthecalcula>Maybe I should also mention that I'm running Guix on Arch
<fishinthecalcula>and my $GUIX_PACKAGE_PATH is set to the directory of my personal channel
<efraim>jonsger: for a simple one it sounds like 'guix pack -f squashfs foo' with some metadata in a file and a list of files in /bin and /sbin
<efraim>doesn't take into account sandboxing or daemons or things like that though
<jonsger>efraim: even better, so you can start with a pretty simple thing and then add the complex stuff on the way...
<g_bor[m]>civodul: yes, I was having issues with passwords.
<rekado>fishinthecalcula: do you get the same error when GUIX_PACKAGE_PATH is unset?
<g_bor[m]>Also, I had a problem with my hdd, and I am now deploying my new development machine. Most probably the problem in my case was that the shadow file did not exist indeed at the time the activation script was running.
<g_bor[m]>One thing that is really problematic here, is that the root password is reset to empty. The other password resets are inconvinient, and can cause serious problems in certain situations, but the empty root password seems to be a security problem. It would be nice to have a policy configuration that tells guix not to regenrate the shadows file, even if it is missing.
<fishinthecalcula>> fishinthecalculator: do you get the same error when GUIX_PACKAGE_PATH is unset?
<fishinthecalcula>thank you for your answer, if the variable is unset Guix doesn't even load the modules in my channel even if it is consed in .config/guix/channels.scm
<fishinthecalcula>guix system: error: failed to load 'install.scm':
<fishinthecalcula>ice-9/boot-9.scm:2803:6: In procedure resolve-interface:
<fishinthecalcula>no code for module (mymodule packages linux)
<fishinthecalcula>this is the error with the variable unset
<g_bor[m]>What really happened here was that the broken hdd on the hypervisor resulted a problem in the guest, where certain files appeared/disappeard on the file system, and sometime created files appeared with a big delay. One example was the shepherd socket, consistently coming alive only after about a minute from boot. I could even log in before it appeared.
<rekado>fishinthecalcula: did you run “guix pull” after adding your channel to .config/guix/channels.scm?
<fishinthecalcula>yes
<fishinthecalcula>this is my output for the command
<fishinthecalcula>export GUIX_PACKAGE_PATH=`pwd` && guix pull
*fishinthecalcula sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/hINIljliHMIFepwMogpNXffc >
<rekado>check if mymodule/packages/linux.scm exists under ~/.config/guix/current/
<fishinthecalcula>it does not
<rekado>then something’s wrong with your channel.
<fishinthecalcula>thank you very much
<fishinthecalcula>do you think the problem is in linux.scm or the install.scm?
<fishinthecalcula>give or take, obviously you have not read my code
<fishinthecalcula>i could link it but i'm pretty sure it's against policies of the GNU project
<rekado>fishinthecalcula: I can’t say. If you don’t even have the files installed after “guix pull” you need to check for something much more basic.
<rekado>for example: if your channel provides a module (mymodule packages linux) it must be defined in a file called mymodule/packages/linux.scm.
<rekado>this must be located in the root of the channel.
<fishinthecalcula>thank you again for your answers
<fishinthecalcula>i'll review my code
<ArtifTh>Hello. I have installed Guix on an existing system and I have issues with setting up a locale for Guix installed programs. I have installed glibc-locales and export GUIX_LOCPATH environment variable. Also I have tried setting up LC_ALL variable with the name of the locale. It seems that programs installed with Guix ignore this setting and also, when I run many guix commands, like installing packages I see warnings like warning: setlocale: LC_ALL: cannot
<ArtifTh>change locale (en_US.utf8), even though my locale is not en_US and no environment variable specify that locale. Can anybody give a direction where to look?
<roptat>ArtifTh, I think I've seen a similar issue recently (like maybe today ^^) on the mailing list
<roptat>are you using guix on a foreign distro with systemd?
<dutchie> http://lists.gnu.org/archive/html/help-guix/2019-06/msg00024.html
<ArtifTh>roptat: yes
<roptat>ArtifTh, so the mailing list thread seems to apply in your case
<ArtifTh>thanks, I'll give it a look
<roptat>if you want to use another locale, it seems that you can modify the environment in the systemd unit
<dutchie>i had that problem (indeed, the reply to that thread is me)
<roptat>otherwise, you should install glibc-utf8-locales as root so the daemon can find the default locale
<roptat>(which for some reason is not insatlled in glibc-locales...)
<ArtifTh>Installing glibc-utf8-locales as root seems to fix warnings
<ArtifTh>although changing LC_ALL in ssytemd unit to different locale does not seem to work
<ArtifTh>Ah, I had to specify .UTF-8 instead of .utf8
<DamienCassou>is there some support for running Flatpack applications in Guix? Part of the philosophy is close between both approaches, but the implementation is very different. Some closed-source applications I need are provided as Flatpacks
<davexunit>DamienCassou: there is no support for that
<rekado>DamienCassou: what kind of support do you need? You can install flatpak with Guix, no?
<DamienCassou>by no support you mean it's impossible to do or that I won't get help if I do?
<DamienCassou>that's exactly my question :-)
<DamienCassou>I want to know if I can install flatpack in Guix and install Flatpack apps
<rekado>DamienCassou: there is a “flatpak” package.
<DamienCassou>thank you
<efraim>I started packaging aerc since it looked like it had minimal dependencies and now I'm following go import errors
<bluekeys>hi guix
<civodul>hello bluekeys
<dutchie>how can I get the path to the build directory in an argument to a build system?
<dutchie>(context: i'm trying to define a go package that uses a non-standard build system, and it's erroring out because it can't write /homeless-shelter/.cache for its build cache. So I'm trying to set GOCACHE to somewhere it can write to)
<bluekeys>Guix, how do i get started with package development? I've got no scheme experience :(
<civodul>dutchie: you can do something like (setenv "HOME" (getcwd))
<civodul>bluekeys: there's a tutorial here: https://gnu.org/s/guix/blog/2018/a-packaging-tutorial-for-guix/
<jonsger>hm civodul was faster
<dutchie>civodul: cool, thanks
<bluekeys>civodul: sweet. Is there a list of software ear-marked for packaging, like a todo list?
<jonsger>dutchie: other then that you can see definitions of existing packages by running "guix edit PACKAGE", e.g. guix edit vim
<dutchie>jonsger: was that for bluekeys?
<jonsger>eh yes dutchie :P
<jonsger>bluekeys: https://libreplanet.org/wiki/Group:Guix/Wishlist
<dutchie>efraim: lol i just realised i'm doing exactly the same as you
<roptat>I'm starting to get used to paredit in vim
<str1ngs>civodul: turns I can use both qtwebkit and qtwebengine as backends. so I switched focus to qtwebkit for now .
<civodul>ok!
<str1ngs>actually it was pretty trivial to switch, and eventually I think I can do it at runtime
<str1ngs>and I should have some patches for qtwebkit soon :P
<str1ngs>I'll guess I'll leave the qtwebeninge patches on the bug tracker encase someone else wants to manage the chromium sources?
<bluekeys>jonsger: Ty
<bluekeys>guix edit opens a new emacs instance. Can it be configured to open in an existing emacs instance?
<str1ngs>bluekeys: you can set EDITOR to do that
<str1ngs>maybe something like EDITOR="emacsclient -nc "
<str1ngs>or EDITOR="emacsclient" if you would like it to block
<str1ngs>this requires you to enable the emacs server socket though
<str1ngs>(server-start) is usually enough to do that in your emacs config
<bluekeys>str1ngs: ty, where is the best place to set EDITOR? config.scm?
<str1ngs>~/.bashrc assuming you use bash
<str1ngs>also the flags -n to use existing emacs and -nc to uses existing emacs with a new frame
<str1ngs>bluekeys: EDITOR is simply a environment variable. I should have mentioned that sorry
<bluekeys>str1ngs: no need to apologise, this is super informative. I'm gonna try and edit my bashrc with a -n after running (server-start) in the scratch. Wish me luck!
<str1ngs>bluekeys: no problem . probably emacsclient -n is the behavior you want here.
<str1ngs>also make sure to export it. ie export EDITOR="emacsclient -n "
<str1ngs>bluekeys: umm I should have metioned this will effect other programs like git commit and anything that uses EDITOR
<str1ngs>which might be okay, but good to know
<str1ngs>if you want this only for guix edit . you can do . $ EDITOR=emacsclient guix edit bash
<efraim>dutchie: here's what I have so far https://bpaste.net/show/d69d6edc9087
<dutchie>that's probably better than what i did; i was trying to tweak gnu-build-system to build it using the make && make install instructions from the README
<efraim>I started with that too, and then I figured I ran into the homelessdir error and figured it would be easier to start from the go-build-system
<dutchie>ah right, makes sense. what do you mean by "following go import errors"?
<efraim>widgets/compose.go:11:2: cannot find package "github.com/emersion/go-message" in any of:
<efraim>and the like
<dutchie>I assume you are aware you can just look in the go.mod file to see all the dependencies?
<efraim>This is my first time looking at go programs
<quiliro>hello
<quiliro>whats up?
<quiliro>installing guix with version 1.0.1
<Marlin1113>hi guix
<str1ngs>hello, quiliro
<Marlin1113>anybody knows what this means?
<Marlin1113>Migrating profile generations to '/var/guix/profiles/per-user/marlon'...
<quiliro>video resolution is not as good as was with 0.16 or previous
<Marlin1113>guix pull: error: symlink: File exists: "/var/guix/profiles/per-user/marlon/current-guix"
<Marlin1113>i get it when trying to do guix pull
<str1ngs>Marlin1113: that means that symlink exists already. you can probably move that to a back up. and it should recreate it properly on guix pull
<str1ngs>Marlin1113: ls -l /var/guix/profiles/per-user/$USER/ . should give you an indecator of where that link should point to. my guess is the link is broken
<Marlin1113>weird
<Marlin1113>wonder how that happened, i didn't really mess much with stuff lately
<str1ngs>I'm not sure if this will orphan any profiles in the store that can not be gc'd after.
<str1ngs>but it should resolve your issue, none the less
<Marlin1113>yeah, i moved it somewhere else and did pull, seems to work
<str1ngs>was the link broken as I suspected?
<Marlin[m]>Yea i guess
<str1ngs>then keeping the backup wont help much :P
<str1ngs>guix pull -l might help list those generations
<str1ngs>actually know that I think about it. if the links are broken then they do not exist in the store anyways. so you don't need to worry about an orphaned profile
<quiliro>how can i get a better resolution?
<str1ngs>quiliro: in X11?'
<quiliro>on the installer
<quiliro>usb
<str1ngs>I guess the installer in ncurses so in a console?
<Marlin1113>hey
<Marlin1113>how can i make a manifest of the packages on the backup so that i can reinstatall them?
<Marlin1113>i haven't used the manifest thing yet
<str1ngs>Marlin1113: users installed packages I assume?
<Marlin1113>yes
<str1ngs>or "user" I should say
<str1ngs>it's probably best to start with a manifest
<Marlin1113>how do i do that stuff?
<str1ngs>guix package -l should give you a list of installed packages. from there create a manifest file.
<Marlin1113>hmm, no way to get the list of packages on my backup?
<str1ngs>on https://www.gnu.org/software/guix/manual/en/html_node/Invoking-guix-package.html . read the --manifest section
<quiliro>str1ngs: console installer and ncurses installer both are at 640*320
<Marlin1113>thanks, i was looking for it
<str1ngs>Marlin1113: basically if you convert the packages found with guix package -l . you can create a manifest. but I guess what I'm trying to say is if you start with a manifest. there is no need to back then
<str1ngs>you can just add to your manifest when you want to install something. and now your use packages are declarative
<str1ngs>user*
<str1ngs>quiliro: what video card do you have?
<Marlin1113>thanks str1ngs
<str1ngs>no worries
<str1ngs>Marlin1113: once you convert your existing package list to a manifest. this will be easier to manage
<quiliro>01:00.0 VGA compatible controller: VIA Technologies, Inc. CN896/VN896/P4M900 [Chrome 9 HC] (rev 01)
<str1ngs>is this a virtual machine quiliro ?
<quiliro>no
<str1ngs>I don't know much about this card sorry.
<str1ngs>is the installer unusable with this resolution?
<quiliro>xf86-video-openchrome ?
<str1ngs>that will only work for X11, not in console
<Marlin[m]><str1ngs "Marlin1113: once you convert you"> done, many thanks
<quiliro>str1ngs: yes...some things are not visible on the graphic installer
<str1ngs>no problem Marlin[m]
<quiliro>i getan error about pata
<quiliro>on installing
<str1ngs>quiliro: can you do lsmod | grep viafb
<str1ngs>or lsmod | grep via
<str1ngs>I'm trying to figure out what console driver should be used with via. I would think it should just use the right one
<quiliro>lsmod | grep viafb
<quiliro>viafb 98304 2 via_camera
<quiliro>i2c_algo_bit 16384 1 viafb
<str1ngs>hmm it seems it is using viafb
<quiliro>str1ngs: do you think it might just be a uqestion of resolution?
<str1ngs>possibly if monitor does not report VBE info properly
<str1ngs>or maybe viafb is causing issues
<str1ngs>you could try . add modprobe.blacklist=viafb to the kernel arguements of the installer
<quiliro>previous versions did not have issues with this exact hardware
<str1ngs>that would stop viafb from being loaded. but that just a huge guess on my part
<quiliro>str1ngs: will try that...thanks
<str1ngs>previous versions of guix and the installer?
<quiliro>any other suggestion before rebooting
<quiliro>?
<str1ngs>quiliro: oh also do you have this problems with other liveCD's ?
<quiliro>live usb
<quiliro>did not try cd
<str1ngs>what I mean is say another distro
<str1ngs>like you have seen higher resolutions in console before?
<quiliro>did not try another distro on this hardware
<quiliro>i did see better res on previous guix installation
<str1ngs>okay that does help
<quiliro>but from some time to now i had guix with bad resolution
<str1ngs>maybe try to black list viafb. though that could cause a black screen. but atleast we know it's using that driver then.
<quiliro>ok....b back then...thanks
<str1ngs>also is this the only video card you have?
<quiliro>yes
<str1ngs>any in CPU video graphics or anything like that?
<quiliro>is it '. add modprobe.blacklist=viafb'?
<quiliro>str1ngs: onboard graphics
<str1ngs>yes that should work. to kernel arguements
<quiliro>old machine
<str1ngs>I figured it was something like that.
<quiliro>including .?
<str1ngs>no . sorry
<quiliro>ok
<quiliro>what if i unload the module now
<str1ngs>should beable to do that from the grub menu.
<quiliro>without rebooting?
<str1ngs>the module is in use, will be hard to do that
<quiliro>with rmmod?
<quiliro>ok
<quiliro>b back then
<str1ngs>kk
<quiliro>in a few minutes
<bluekeys>hi guix, my first attempt at building a package for trigger-rally game has failed :) How do i open a bz2 file?
<dutchie>bzless is the easiest way in my experience
<dutchie>but bunzip is the general decompressor for bzip files
<dutchie>oh, it's bunzip2
<bluekeys>dutchie: what's bzless?
<dutchie>it's part of the bzip2 package; it decompresses the file and opens it in less(1)
<dutchie>just run it in a terminal on the file
<bluekeys>cool, that's neat ty
<Tirifto>Hello all!
<Tirifto>So I got fonts working on Icecat started with GNOME Shell, using a fontconfig configuration file I found on the internet!
<Tirifto>I guess I should experiment with it to pinpoint which part was necessary and make a post to the lists, in case it would be of any use.
<bluekeys>Tirifto: I'd be interested to see that when you're done.
<Tirifto>Copy & paste, merit-wise. :P
<bluekeys>Librarians are paid good money for a reason
<Tirifto>xP
<Tirifto>I also found out that GNOME Shell apparently crashes when I try to declare $XDG_DATA_DIRS, so I opened a ticket there.
<bluekeys>Is there a way to skip the configure step in the gnu-build-system
<str1ngs>bluekeys: I think you can modify phases and (delete 'configure)
<bluekeys>str1ngs: got it, I can see the example that links to your advice in the packaging tutorial now. ty
<str1ngs>okay that's good. I've only seen it in passing. the example is probably a better source
<raingloom>hey, there is a warning on the website of Singularity that it's moving to https://www.sylabs.io/docs/
<raingloom>it is linked to in the documentation for guix pack
<raingloom>i'd update it but idk where the source for that is hosted
<PotentialUser-22>anybody out there? I installed the Guix os on windows hyper-V (windows 10 pro). I installed with launching the GUI upon login (after the install) but I'm seeing some error messages 7 no GNOME or other gui is sertup.
<bavier>raingloom: it's in the 'docs' directory of the master repo
<lfam>Howdy everyone
<bavier>hey lfam
<JarthB>Hi People
<kmicu>( ^_^)/
<JarthB>Trying out a few linux distros as my choice for the next five years, Guix is real candidate
<JarthB>however, when i get the installation, it states it will format the harddisk, not a partition, the entire disk ...
<JarthB>kmicu: hi
<kmicu>JarthB: I didn’t use Installer so no idea about formatting specifics in it. Please idle here a little bit and others should help for sure.
<JarthB>kmicu: thanks, what did you do then to install ? did you follow a process ?
<Marlin1113>hi guix
<Marlin1113>how can i make a custom service? i'd like to run emacs as a daemon on boot
<lfam>JarthB: Those of us who installed in the past followed the instructions in the manual: https://www.gnu.org/software/guix/manual/en/html_node/System-Installation.html#System-Installation
<lfam>JarthB: The "installer" is new and many of us haven't used it yet
<JarthB>lfam: aah, ow, ouch
<JarthB>lfam: i'd recommend guix adopts the debian installer or something, if it works at all with guix
<JarthB>i use the awesome powers of qemu to install from a running system onto disk
*lfam afk
<JarthB>i don't get why it is not advertised in installation manuals, it works like a charm
<str1ngs>probably because you can just use the guix binaries to install from a running system. no qemu required
<JarthB>str1ngs: haha, okay, i'll check it again then
<str1ngs>basically you do a binary install. followed by a manual declaration install
<str1ngs>it won't use the installer. but the installer is really only intending to for new users
<str1ngs>intended*
<JarthB>str1ngs: not sure if we speak of the same though
<str1ngs>how so?
***ym555_ is now known as ym555
<JarthB>str1ngs: the steps i read about are awfully complicated
<JarthB>str1ngs: or involved rebooting from USB or ISO
<JarthB>I'm writing down what i mean here https://bit.ly/2ZcIr2R
<str1ngs>maybe I misunderstood. then I got the impression you wanted to install GuixSD from within a existing host system
<JarthB>str1ngs: either way, i'm first installing into a VM, i don't want to destroy an entire HDD without being certain of what the installer does
<str1ngs>that's understandable . it seems you are testing the installer by using a VM. if that is the case I agree I misunderstood what you meant meant by host system.
<JarthB>str1ngs: no prob, this is forum is amazingly civil
<lsl88>hi Guix!
<lsl88>I have installed guix on top of Fedora. I am facing a strange issue. After running the installation script, I am getting error: failed to connect to `/var/guix/daemon-socket/socket': No such file or directory. Ang guix-daemon does not exist :s
<emyles>hello
<str1ngs>lsl88: you used guix-install.sh ?
<lsl88>str1ngs: yes
<lsl88>i have installed guix tons of times, it is the first one I am facing this issue
<str1ngs>and pgrep -a guix lists a guix-daemon?
<str1ngs>better yet. how about systemctl status guix-daemon
<str1ngs>I'm assuming you are use the systemd service?
<lsl88>str1ngs: I am getting guix-daemon.service: Failed to execute command: Permission denied
<lsl88>and yes, I am running systemd
*kmicu recalls that rekado uses Guix on Fedora.
<jonsger>lsl88: maybe that is SElinux blocking that request. I'm not sure if this is running on Fedora per default...
<lsl88>oh you are sight
<lsl88>right* i will check it
<str1ngs>lsl88: can you run the daemon by hand?
<str1ngs>as jonsger mentioned I'm wonder if this is selinux related
<lsl88>str1ngs: thank you! It was selinux ¬¬, I disabled it, restarted the daemon and now it works :)
<str1ngs>no problem
*kmicu sends thank you note to jonsger
<jonsger>lsl88: maybe you can find out the relevant parts of the SElinux log (related to guix-daemon) and send it to guix-devel@gnu.org
<lsl88>jonsger: do you think it is useful for the community?
<g_bor[m]>hello guix!
<lsl88>hi g_bor[m] :)
<g_bor[m]>iirc there is a selinux policy for the daemon. Is that in effect?
<g_bor[m]>rekado tired to create one, and I belive it is for fedora...
<jonsger>yes, lsl88 the feedback is welcome. rekado created the policy for guix-daemon, but its far from perfect I guess :)
<g_bor[m]> https://www.gnu.org/software/guix/manual/en/html_node/SELinux-Support.html
<jonsger>I tried creating an AppArmor profile, but the nature of it being file based makes it a little "difficult" with /gnu/store and friends...
<g_bor[m]>and yes, if it does not work out of the box, then the log is very welcome :)
<lsl88>oh great! I will take a look at it and send it then :)
*vagrantc blinks hard at the idea of trying to apply apparmor to guix-daemon
<lsl88>in fact it is the first time I am running guix over Red Hat based distros, did not remember about selinux, and I am trying different distros because I am doing some research over another issue
<g_bor[m]>jonsger: regarding the apparmor yes, in this case it is not very easy. Do you think we could programmatically generate one, based on the current situation?
<g_bor[m]>Would it worth the hassle?
<JarthB>just finished a very simple guide on Qemu wizardry (not mine) https://bit.ly/2ZcIr2R
<JarthB>it may be useful for considering to include into guix as an installation approach
<jonsger>g_bor[m]: that's how far I went https://gitlab.com/jonsger/Guix/commit/db1c43c7fe6153d88ac8ed265c7c7c03e49c7c3e But this approach wont led to something good
<jonsger>how do you think we can create on "programmatically"?
<recj>hi, when trying to boot i get the following https://youtu.be/2yq5QfNqtPk
<recj>no idea why gdm is behsving this way... how can i best diagnose the problem?
<JarthB>recj: woah, fascinating
<recj>lol
<JarthB>recj: does not look like gdm at all to me
<JarthB>recj: then again, pretty new to guix
<recj>but that is the process with all the sessions
<recj>i cant open i3 or anything either
<JarthB>recj: my uneducated impression is there is an issue with shepherd configuratoin
<JarthB>runlevel mismatch or something
<recj>i see
<JarthB>recj: check here ? https://www.gnu.org/software/shepherd/manual/shepherd.html#Invoking-herd
<recj>i just installed
<recj>so makes sense
<JarthB>ok
<JarthB>recj: consider https://paste.rs/web to share such information, a video is hard to diagnose
<JarthB>recj: definitely something runlevel, just now i saw you get root: first then the boot chaos starts
<recj>yes thank you
<JarthB>recj: look at 0:15
<JarthB>there is mention of credential errors
<recj>i currently have no internet on the computer unfortunately because of proprietary wifi drivers and no ethernet right now
<recj>interesting ok
<JarthB>recj: download the firmware onto a stick and copy to the firmware folder ?
<JarthB>recj: also, network cable to the rescue ?
<JarthB>recj: check here for firmware https://www.intel.com/content/www/us/en/support/articles/000005511/network-and-i-o/wireless-networking.html
<recj>great yup
<recj>i can just download it using wget once i find it right?
<recj>and cp to firmware folder
<JarthB>recj: ehr, maybe not, i could have expected but was unaware of https://en.wikipedia.org/wiki/Linux-libre
<JarthB>recj: this is inside guix
<JarthB>have to go, cya people
<recj>yeah i found a nonfree kernel external package (not sure if we can talk about/ endorse obtaining nonfree software via guix here) but it didn't work when i tried to reinit, will try again when i get to ethernet though
<rekado>I only have one workstation in the office running Fedora with Guix
<rekado>It’s still on Fedora because of problems with Active Directory integration when using Guix System.
<rekado>if SELinux blocks things after installing the policy it would be most helpful to a) switch to permissive mode, and b) use sealert to report any reported problems
<rekado>otherwise I fear we may never know for sure if the policy I wrote long ago actually works properly.
<lsl88>rekado: ok, I will switch to permissive, I directly disabled it
<rekado>have you installed the policy?
<ItsMarlin>Hi guix
<ItsMarlin>how can i add a new xsession?