IRC channel logs

<bavier>has anyone here had success using guixsd with a radeon gpu?

<ng0>personal experience was that I got my old nvidia cards out and installed them again... radeon was terrible

<ng0>is anyone able to reach git.sv.gnu.org ?

<ng0>are they moving the git host right now?

<ng0>ah it works again

<bavier>ng0: my radeon card is "working", but no dri from mesa

<bavier>so games and video playback is painfully slow

<ng0>I'm not sure wether my cards are breaking or the computer I had them in is slowly breaking

<ng0>with the old nvidia cards I have, when I'm able to reach the desktop, at least video playback

<ng0>*card

<ng0>the emacs segfaults are annoying :/

<cehteh>that doesnt happen to me, but i had this weird problem yesterday

<cehteh>tested in another vm wich is memory constrained, doesnt happen there

<ng0>after 12 hours or so it happens

<cehteh>would be nice if guile-emacs would be userable/fast/stable/awesome

<ng0>though my emacs -nw session doesn't crash.. it's just the gtk(?) gui for now

<ng0>the -nw crashes not so often

<cehteh>ACTION just lets emacs-guix batch install a shitload of packages .. i hope it doesnt chrash meanwhile

<adfeno>ng0: I found another bug while using Emacs GUI (XEmacs?) but it's easy to solve.

<adfeno>It seems that, if one runs Emacs as daemon with `emacs --daemon`. Then uses `emacsclient [Some file or standard input signal]`, then Emacs tries to guess if it should start in graphical mode, but fails...

<adfeno>... however, it's not the failure that is the bug, because when it fails, it also seems to not try to correctly set the keyboard input.

<adfeno>So accents cannot be typed.

<ng0>How do I find out valid targets for a Qt package? I have all the dependencies figured out

<adfeno>... If I instead of `emacsclient [file/stdin]` do: emacsclient -nw [file/stdin] it works

<Apteryx>Is there a way to pause scrolling in xterm?

<adfeno>Perhaps Shift + PageUp?

<ng0>ah, it uses gyp

<ng0>I think the chromium package will help

<Apteryx>adfeno: Shift + PageUp scrolls up, but on new output (which is often when compiling), it goes back to the last line.

<adfeno>Oh...

<adfeno>Using GNU Screen, I can pause scrolling by going to the copy mode (Ctrl + [ ).

<adfeno>Then when done, I press q.

<cehteh>Apteryx: ctrl-q ctrl-q

<cehteh>Apteryx: ctrl-s (stop) ctrl-q (continue)

<adfeno>Which bot does we use to log messages?

<adfeno>People at #replicant are looking for good one :)

<ng0>an akward one

<ng0>one moment I can go into detail

<ng0>as far as I see it, the bot seems to be integrated into Drupal, at least it uses no plain text for the logs. it will be updated this year, but I might be wrong about the bot. One question is still opne on it.

<ng0>btw: elogind was added to Gentoo :)

<ng0>just yesterday

<mekeor>OMG, 0.12 implements LUKS-encrypted root?

<mekeor>you are awesome!

<OriansJ>now if only there was a way to convert a debian or a centos system using guix to a full guixsd

<mekeor>OriansJ +1, this should be added to the todo-list / roadmap :)

<Apteryx>Hi Guix! Can I understand "propagated-inputs" as inputs made visible to at the system level compared to "inputs" which are only made available to the package depending on those inputs?

<Apteryx>And "visible at the system level" would mean available for any software to find and use at the usual expected locations

<ng0_>I just re-discovered the stackexchange mode for emacs I packaged last year :)

<ng0_>and it works

<ng0_>good night o/

<albertoefg1>hello my friends :)

<albertoefg1>would any of you would be so kind to send me a screen shot of guixsd desktop (i know guix is not the desktop)

<albertoefg1>but any one, gnome, xcfe,

<OriansJ>albertoefg1: umm, some of use use things like xmonad and all you would see is whatever application we are using at the time

<albertoefg1>ohh

<albertoefg1>i found a few on guix site :)

<albertoefg1>i am talking to a few friends about guix

<albertoefg1>so they install it :)

<OriansJ>albertoefg1: have use the deploy script, try guix a little bit on the linux distro of their choice and let them decide for themselves if they wish to go with guixsd later

<OriansJ>And if they don't like it, the removal script will undo everything the install script did and remove all installed guix packages too.

<albertoefg1>OriansJ: yes thanks :)

<albertoefg1>i told them that

<albertoefg1>to install first guix not guixSD

<albertoefg1>:)

<Apteryx>Is it possible to hack on something low level things (python2, python-build-system) without getting embarrassed by non-stop rebuilding of all which depends on it?

<mekeor>ACTION is installing GuixSD

<mekeor>should i create a boot partition separate from root partition when fully encrypting my disk?

<cbaines>mekeor, I didn't create a /boot partition, but I did create a BIOS boot and EFI system partition

<mekeor>what's that?

<cbaines>mekeor, I'm not quite sure, I was mostly following the Gentoo paritioning guide https://wiki.gentoo.org/wiki/Handbook:AMD64/Installation/Disks#Default_partitioning_scheme

<mekeor>ah, nice, thank you :)

<cbaines>Apteryx, regarding hacking on low level things, I would guess so. You are not forced to rebuild everything if you make a change, so you should be able to choose what you rebuild.

<Apteryx>cbaines: OK. I think I got confused by having too many "work in progress" in master and then rebuilding guix. This interacted in ways I hadn't predicted and needed to rebuild lots of things. I'm working on one thing at a time in branches now.

<Apteryx>I'm trying to wrap my head around the way the Python build system works.

<Apteryx>Although working in branches causes the whole guix to be rebuilt. Heard someone propose git workspaces for that; I should read the manual a bit.

<Apteryx>(when switching branches)

<tumashu>How to delete generations?

<cbaines>If you mean system generations, currently there is not a command line option

<tumashu>yes, I mean system generations

<cbaines>But you can just delete the file in /var/guix/profiles

<cbaines>Also, I hear that the Emacs package for Guix provides an interface for deleting system generations

<tumashu>should I edit grub.conf by hand?

<cbaines>I would recommend against that

<cbaines>I'd suggest running guix system reconfigure

<tumashu>switch-generation then reconfigure seem to override current history system generation.

<cbaines>I'd expect reconfigure to make a new generation if that is what you mean

<cbaines>I don't quite understand what you mean by "override"?

<tumashu>if I switch-generation to #12, reboot, reconfigure seem to override origin #13

<tumashu>May be I am wrong:-)

<cbaines>I don't think reconfigure will override anything

<cbaines>It will add a new generation

<tumashu>I will test it again:-)

<tumashu>Any command which can add system package?

<tumashu>or just guix system reconfigure?

<cbaines>I think as the system packages are declared in the operating system definition, you have to modify that, and then run guix system reconfigure

<rekado_>tumashu: generations are linear, they are not arranged in a tree.

<rekado_>tumashu: this means that the *link* to generation 13 would be replaced with a link to the new generation

<rekado_>the old generation is still in the store but the link would be replaced

<rekado_>to actually remove things you need to run “guix gc”

<tumashu>This seem to be a tip removing broken generation:-)

<tumashu>By the way, how to deal with FlashPlayer in guixsd?

<rekado_>tumashu: didn’t you already ask this on guix-devel?

<rekado_>tumashu: were the answers not satisfactory?

<tumashu>Yes, but guix will not include *anythings* about Adobe Flash Player:-)

<rekado_>that’s right. We won’t add the Adobe Flash Player to Guix.

<tumashu>Maybe gnash is the only choice...

<tumashu>I will leave to test new reconfigure, bye.

<tumashu>warning: collision encountered: /gnu/store/j7242vqxk8vapd12qm8jbq905d2456wb-gtk-icon-themes/share/icons/hicolor/icon-theme.cache /gnu/store/vfrfjfa878cwcz7pkg899nm79p7053dz-hexchat-2.12.2/share/icons/hicolor/icon-theme.cache

<tumashu>what does the warn mean?

<tumashu>warning: arbitrarily choosing /gnu/store/j7242vqxk8vapd12qm8jbq905d2456wb-gtk-icon-themes/share/icons/hicolor/icon-theme.cache

<rekado_>tumashu: it means that two packages that are installed in the same profile provide a file of the same name.

<rekado_>tumashu: guix just picks the first

<rekado_>tumashu: in this case it’s harmless because we’re generating a new icon-theme.cache in a profile hook.

<rekado_>there’s a bug report somewhere about suppressing these warnings when they are known to be harmless

<tumashu>harmless is very good:-)

<jmi2k>Hi Guix, isn't openjdk packaged?

<buenouanq>icedtea I think is what you want

<jmi2k>Oh yes, I'm reading about it now. I thought it was only a web plugin, I don't know why

<jmi2k>Thanks buenouanq

<rekado_>StumpWM is nice. I wish I could talk to it with Emacs SLIME, but even without that it’s doing just what I need.

<rekado_>(it’s nicer than xmonad and I really don’t mind Common Lisp)

<buenouanq>rekado_: have you tried guile-wm? I've been intending to check it out myself for some time now.

<buenouanq>I'll get around to it at some point (;-___-)

<rekado_>buenouanq: I heard that it’s rather immature.

<rekado_>since I don’t have time to tinker much these days I went with StumpWM (just had a new release).

<Somelauw>What is the recommended method to get guix on Debian?

<rekado_>Somelauw: the easiest way is probably to use the binary installation method.

<rekado_>it’s described here: https://www.gnu.org/software/guix/manual/html_node/Binary-Installation.html

<Somelauw>thanks, my goal is to install it in such a way there is minimal conflict

<rekado_>Somelauw: guix only modifies files in these locations: /gnu, /var/guix, /etc/guix

<rekado_> /var/guix contains state (including profile links), /etc/guix contains the acl to determine whether to trust binaries from hydra.gnu.org (or elsewhere). /gnu contains the store, which holds any item build or installed by guix.

<jmd>I think that we should move /var/guix and /gnu/store to a common directory. One is useless without the other.

<jmd>... and if you modify one without the other, then both become useless.

<Somelauw>ok, but at some point i expect to do something like addpath "/var/guix/root/bin"

<jmd>That sounds like an odd thing to do.

<adfeno>Somelauw: Describe what addpath does?

<Somelauw>modify PATH environment variable, but I haven't finished reading the manual yet, so maybe that's not the idea at all

<adfeno>Somelauw: The ideal thing would be follow guix doc.

<adfeno>They do provide advices on how you should modify $PATH variable.

<adfeno>And believe me, "/var/guix/root/bin" isn't the right place to point to. :)

<iyzsong>yes, the information is in the manaul, as "For each user, ..." of "3.2 Invoking `guix package`".

<Somelauw>I think an improvement would be to make packages in gnu/store start with the package name and move their hash at the end

<Somelauw>not backward compatible, but cleaner in my opinion

<mekeor>ACTION agree w Somelauw

<EvilPython>Hello.

<EvilPython>How to set IP in Guix? I want to connect to Wi-Fi?

<rekado_>Somelauw: there’s little advantage to doing it this way.

<rekado_>Somelauw: a user doesn’t have to know the store path anyway.

<Somelauw>Can this be done for man pages: "for i in /var/guix/profiles/per-user/root/guix-profile/share/info/*; do ln -s $i ; done"?

<rekado_>Somelauw: having the hash first also allows sysadmins to more quickly disambiguate packages.

<rekado_>EvilPython: are you using GuixSD?

<Somelauw>no, just guix

<EvilPython>rekado_: Yes.

<rekado_>EvilPython: is your wireless hardware supported by Linux libre?

<EvilPython>rekado_: I think it can work without firmware at all. Working on Trisquel.

<EvilPython>rekado_: *It is working on Trisquel.

<rekado_>EvilPython: good!

<rekado_>EvilPython: do you have wicd installed?

<EvilPython>I'm using GuixSD default USB image.

<EvilPython>It is not installed in default image.

<rekado_>oh, you haven’t installed it yet, right?

<EvilPython>No.

<EvilPython>I have not Wicd on GuixSD or Trisquel.

<mekeor>EvilPython: did you try this? https://www.gnu.org/software/guix/manual/html_node/Preparing-for-Installation.html#Networking

<EvilPython>How to use WPA2-PSK?

<mekeor>EvilPython: just create a wpa_supplicant.conf as described in the instructions

<EvilPython>dhclient isn't working. It cannot find IP.

<EvilPython>DHCP server is working.

<EvilPython>And Trisquel can use it.

<mekeor>EvilPython: did you connect to your wireless network using wpa_supplicant successfully?

<EvilPython>Yes.

<mekeor>whats the error message dhclient gives?

<EvilPython>I will try again.

<EvilPython>Something like "No DHCPOFFERS".

<Somelauw>I'm having trouble configuring the deamon: https://paste.debian.net/hidden/2dfe02a1/

<Somelauw>systemctl enable guix-daemon tells me Failed to execute operation: No such file or directory

<rekado_>did you drop the systemd unit in the right location?

<EvilPython>No DHCPOFFERS received.

<EvilPython>No working leases in persistent database — sleeping.

<adfeno>Somelauw: Remember that Upstart (init?) and systemd have different places to place service files.

<ng0>I need a sane way to rewrite: '''which' python2.7 > /dev/null && exec python2.7 "$0" "$@" || exec python "$0" "$@"''' for guix o.O

<mekeor>EvilPython: did you make sure connection (1) was established by wpa_supplicant and (2) you specified the correct interface for dhclient?

<ng0>this is mach build system and in a bash script

<ng0># The beginning of this script is both valid shell and valid python,

<ng0># such that the script starts with the shell and is reexecuted with

<ng0># the right python.

<ng0>my first attempt made an infinite repetition of python until the build-system bailed out

<EvilPython>mekeor: Yes.

<Somelauw>adfeno: i'm only using systemd?

<ng0>right now I have:

<Somelauw>rekado_: yes, see paste https://paste.debian.net/hidden/2dfe02a1/

<mekeor>no idea then, sorry, EvilPython. i guess, we'd need more info to help you, idk

<ng0>(substitute* "mach" (("'''which' python2.7 > /dev/null && exec python2.7 \\"$0\\" \\"$@\\" || exec python \\"$0\\" \\"$@\\"") (string-append (assoc-ref %build-inputs "python-2") "/bin/python2.7" " > /dev/null " "&& exec " (assoc-ref %build-inputs "python-2") "/bin/python2.7" " \\"$0\\" \\"$@\\" " "|| exec " (assoc-ref %build-inputs "python-2") "/bin/python" " \\"$0\\" \\"$@\\" "))

<ng0>and I know we don't have /dev/null

<ng0>but there MUST be an easier way than this mess

<ng0>and the original mess

<mekeor>Q: so, i'm still installing GuixSD: i'd like to encrypt my whole filesystem and i thought it's most easy to have a separate, not-encrypted boot partition, right? but how do i specify my boot partition in config.scm?

<adfeno>Somelauw: Let's see...

<cbaines>mekeor, if easy is just following the Guix documentation, then I'd suggest not having a seperate /boot

<adfeno>Try `systemctl enable` with the full file path.

<cbaines>mekeor, that is what I managed to succeed with

<rekado_>ng0: can’t you just replace the whole line with the path to the appropriate python interpreter?

<ng0>this is the original https://github.com/MoonchildProductions/Pale-Moon/blob/master/mach

<rekado_>ng0: because we know in advance which python is available, so the whole “which” and “||” thing isn’t needed.

<ng0>idk if it is python which later calls bash

<mekeor>cbaines: will it work without further configuration to have a fully-encrypted root-partition as described in the docs? so, the instructions in the docs (including the LUKS snippets) just work?

<rekado_>ng0: just remove the whole shell block and replace it with (string-append "#!" (which "python"))

<Somelauw>adfeno: # systemctl enable /etc/systemd/system/guix-daemon.service

<Somelauw>Failed to execute operation: Too many levels of symbolic links

<ng0>hm... I'll give it a try, I have the feeling Itried this in the past already

<rekado_>this shell hack is only to determine the path to python.

<rekado_>since it’s know at build time you can replace it all.

<rekado_>Somelauw: instead of symlinking the service file try copying it.

<rekado_>Somelauw: older versions of systemd don’t like symlinks.

<mekeor>cbaines: tbh, i don't even understand how it works that grub can be stored on an encrypted partition. like, how is it decrypted then?..

<cbaines>mekeor, I don't particularly understand either

<cbaines>I do have 2 other partitions though

<ng0>mekeor: iirc, hooks

<cbaines>one BIOS boot, and one EFI system

<Somelauw>there are some other symlinks, but the others are all directed at /lib/systemd/system

<mekeor>cbaines: if i undestood correctly, those two partitions are not neccessary for me (using MBR partition layout on a BIOS laptop)

<adfeno>Somelauw: Hm....

<Somelauw>adfeno: thanks, copying the file worked

<mekeor>cbaines: gentoo handbook says: "BIOS boot partition is needed when GPT partition layout is used with GRUB2, or when the MBR partition layout is used with GRUB2 when the first partition starts earlier than the 1 MB location on the disk."

<adfeno>Hm....

<adfeno>It seems systemd's support resources say that `systemd enable` **doesnot** fllow symlinks.

<cbaines>mekeor, I'm using a GPT layout, so yes, reading that, you might not need that

<mekeor>alright, thanks cbaines, i'll just try it :)

<adfeno>... So, our assumption to `ln -s` the systemd service to the systemd's expected service files location doesn't hold well.

<adfeno>I'll try investigating if there's a way to make `systemctl enable` follow such symlinks.

<cbaines>adfeno, I'm just reading https://bugzilla.redhat.com/show_bug.cgi?id=955379

<cbaines>Sounds like it isn't possible with some versions, and is with newer releases

<adfeno>Oh... So Somelauw Has to update his.

<rekado_>Somelauw or someone?

<rekado_>:)

<adfeno>Or, refrain from running as a systemd service,( by doin a cron job instead????)

<rekado_>oh, I read this as “this” (i.e. our docs)

<adfeno>Because, as far as I know, our service file is sometimes changed according to updates and upgrades.

<adfeno>ACTION starts to see why people hate systemd so much.

<adfeno>This is why I'm going to... Shephrd. :)

<adfeno>... Once I start using GuixSD that is.

<ng0>there's an additional problem I have, for more than one application (pike, palemoon): files with /bin/sh shebang get created and I'm looking for the easiest way to fix this. This happens in a phase where I can#t substitute anything, and our substituter chokes on both applications because of some patterns in the files.

<ng0>my current approach is to replace the /bin/sh with CONFIG_SHELL, but idk which syntax I have to use... ${CONFIL_SHELL} did not work

<ng0>*CONFIG_SHELL

<ng0>in bash I mean

<cbaines>If you can't patch the files themselves, would it be possible to patch the file that creates them with that shebang?

<ng0>that is what I'm triyng

<ng0>and substitute chokes on that file, and I'm asking for the correct way to do this in bash

<rekado_>you don’t have to substitute with CONFIG_SHELL

<rekado_>that’s unnecessary indirection.

<rekado_>if you’re going to substitute anyway you might as well use the final path to the shell.

<ng0>what do you suggest I add then, in inline /bin/sh which is not in the first line?

<ng0>but I can't guess it

<ng0>and I can't apply substitute

<rekado_>?

<rekado_>I don’t understand the questions.

<ng0>I have to use a patch

<rekado_>why?

<rekado_>what does “substitute chokes on that file” mean?

<ng0>I can't give the output right now, but for pike there is a pattern in the file whic hcreates the /bin/sh file before the occurences of the /bin/sh which throws the substitute off

<ng0>I'm not doing pike right now though

<ng0>and for the mach build system, a configure file is created in a running phase

<rekado_>“which throws the substitute off” — how so?

<ng0>sigh

<ng0>okay I will run it

<rekado_>ng0: the configure is probably created from a template, no? In that case you could change the template.

<ng0>my plan wasn't to get into the mess which is pike for another weeks..

<rekado_>then don’t.

<ng0>yes that's what I'Äm trying to do

<ng0>I try to change the template

<ng0>but this is going on for weeks and months now, and I have tried several approaches, the plain patch method is the last way I see for pike

<ng0>for palemoon (mach) I still have to find the file which actually creates configure

<rekado_>have you tried tracing it?

<rekado_>e.g. strace -f …

<rekado_>ACTION has to go

<ng0>that would require that I'm able to build pike without the build system, which I am not.

<ng0>or am I wrong?

<ng0>for pike, you can check out the lpc/pike branch from https://pagure.io/guix-dev.git and see it fail. Most likely at some other points as I'm getting more and more frustrated with it

<Somelauw>Well, not yet, systemctl enable worked after copying, but systemctl start not.

<Somelauw>What's the l permission?

<cbaines>I may have missed at bit of the conversation

<cbaines>What is the context about the "l permission" Somelauw ?

<Somelauw>cbaines: oh, it means symlink

<cbaines>Ok

<cbaines>So are you having problems getting the guix-daemon to start?

<Somelauw>cbaines: i copied guix-deamon, but "sudo systemctl start guix-daemon" still doesn't work The permissions of this file are -r--r--r--

<Somelauw>like other daemon files

<cbaines>What does systemctl status guix-daemon.service say?

<Somelauw>well, guix package: error: failed to connect to `/var/guix/daemon-socket/socket': Connection refused

<ng0>rekado_: sorry. I know I did not provide much details. I had some progress last week with both applications, so maybe some break from them will help.

<Somelauw>cbaines: https://paste.debian.net/hidden/400403f1/

<Somelauw>so i needed to do restart

<Somelauw>and now it says "build failed: the group `guixbuild' specified in `build-users-group' does not exist"

<Somelauw>which i haven't created yet

<cbaines>Ok, sounds like you are making progress :)

<Somelauw>this stuff needs a good makefile install or something, but I'll carry on

<ng0>okay, I think with "choking" I meant decoding error. I might be wrong about this, but could be..

<ng0>pike is... special.

<ng0>a special pain

<Somelauw>jay, 'guix package -i hello' was doing something

<Somelauw>that involves downloading multiple stuff from mirror.hydra.gnu.org

<cbaines>Somelauw, have you enabled substitutes yet?

<cbaines>otherwise you it might take a while to bootstrap

<cbaines>(of course, you don't have to if you want to build pretty much everything from source)

<Somelauw>yes, i only get this warning on each command: warning: failed to install locale: Invalid argument

<cbaines>Somelauw, locales are always fun

<Somelauw>any locale is fine that uses English, starts days of the weeks on monday, uses yyyy-mm-dd for dates, uses SI units and uses . as decimal separator

<ng0>so none

<ng0>yyyy-mm-dd is in no locale other than en_DK

<Somelauw>than anything but american date notation is fine

<ng0>oh, locale.. I wasn#t following what you do but there's a section in the documentation about the locale

<Somelauw>So how can I run the hello package I just installed?

<cbaines>So, installing will have ensured that it is present in the store, and added it to your profile

<cbaines>It probably also printed out some environment variables

<jmd>Somelauw: type "hello"

<Somelauw>So I have this line at the bottom of my .zshrc export PATH="/root/.guix-profile/bin:/root/.guix-profile/sbin${PATH:+:}$PATH"

<Somelauw>hello does nothing

<Somelauw>command not found

<Somelauw>maybe i have missed a step somewhere

<cbaines>Did you run guix package -i hello as root?

<cbaines>and have you created the .guix-profile symlink (I can't remember if you have to do this manually)

<Somelauw>yes, i skipped all of 2.6 Application Setup

<mekeor>for the record: i get "error: access denied", "error: no such cryptodisk found" when i try to boot guixSD. anyway, gotta go now, see you later <3

<Somelauw>"hello" works now as root

<cbaines>You might know this, but in Guix, you have per user profiles

<cbaines>so if root installs something, its in root's profile

<cbaines>but it does not change anything for other users of the system

<cbaines>they must install hello as well if they want to use it

<Somelauw>oh right, but they can share the same binary

<cbaines>Yep

<Somelauw>Can I skip the name service switch section?

<cbaines>If you are using a foreign distro, the manual says its strongly recommended

<cbaines>I'm not very knowledgeable of that though

<jmd>How can I find out on which device / resides?

<cbaines>jmd, if you run mount, you can look and see what device is mounted on /

<jmd>on a unionfs apparently, but that's not a device. :(

<ng0>when I stop trying to unbundle and use some autoconf before real configure, I get some progress. doesn't make me happy, but it's okay, i will unbundle later

<rekado_>Somelauw: the manual lists 6 simple steps to get this set up. It also says to set up the users before starting the service.

<rekado_>ng0: guix environment might help you get into an environment that is similar enough to the build environment. There you could strace the build scripts to see what files are read.

<ng0>hm... ok. I don't use environment enough. it's useful

<rekado_>it sure is :)

<ng0>that's a funny configure. all it has is #!/gnu/store/2qxzpfc8hzd1n91xjwyyrq70wl5f5f0i-bash-4.4.0/bin/sh

<ng0>i hope at the end of this exercise I understand the mach build system a bit better and have more knowledge I don't really wanted and don't really need

<alezost>rekado_: I hack my stumpwm from Emacs using SLIME/swank, but I use stumpwm from a git checkout, not from a guix package. FYI Guix does not have a package for stumpwm with slime, there is some 'sbcl-stumpwm-with-slynk' though (slynk is an alternative for slime AFAIK)

<rekado_>alezost: oh, does stumpwm needs to explicitly support SLIME?

<paroneayea>I need to figure out how to enable stumpwm again

<paroneayea>I ran into the problems others ran into about not being able to start it, but it looks like people solved it on the list

<paroneayea>maybe I should do that later today.

<paroneayea>however, the SLIME thing kind of held me back

<paroneayea>because they haven't fixed the security issues

<rekado_>alezost: thanks for the hint with slynk. I’ll try that.

<paroneayea> https://github.com/slime/slime/issues/286 (same problem guile had)

<rekado_>paroneayea: in my ~/.xsession I just have “exec stumpwm” at the bottom.

<paroneayea>rekado_: oh really? that easy?

<alezost>rekado_: I think there should be 'sbcl-stumpwm-with-slime' package, otherwise slime will not be available

<rekado_>should we also have an emacs-sly package to go along with the slynk variant?

<efraim>civodul: i'm going to try regenerating my aarch64 bootstrap binaries before I try too much more with the current ones, should I make them based on master or from core-updates?

<alezost>rekado_: TBH I don't know, I didn't look at Guix's stumpwm, slynk and slime packages :-)

<buenouanq>guile-wm should recieve more attention and support ( ._.)

<alezost>janneke: re guix-prettify: it wasn't touched for a long time; what error did you have?

<paroneayea>buenouanq: lots of people would like that, but someone needs to step up and do it.

<ng0>speaking of WMs which do not work, I fixed detection of dwm and it works

<janneke>alezost: i have no clue...i had (require 'guix-prettify) in my init.el

<alezost>janneke: oh, maybe you didn't know that emacs stuff moved away from guix, so after "guix system reconfigure" you just don't have guix-prettiy.el and other files anymore

<janneke>and after upgrading emacs it bombed-out

<ng0>janneke: for me it started earlier, in november emacs started crashing

<janneke>alezost: i don't really understand yet, but that sounds like the problem

<janneke>i certainly did guix system reconfigure

<alezost>janneke: if you didn't install "emacs-guix" package, then that's it: you don't have any "guix*.-el" files anymore

<janneke>ng0: yes, i didn't upgrade for quite a while

<alezost>janneke: you can do "guix package -i emacs-guix" if you wish

<janneke>ng0: i am using an old emacs right now -- last week was a nightmare

<janneke>alezost: ah, great!

<ng0>I wanted to express, it doesn't have to be the removal of emacs guix

<janneke>emacs started crashing, my touchpad's palm detection was gone so i kept getting weird jumps in emacs, 2-finger scrolling goes the wrong way, gsettings list-recursively is almost empty...

<alezost>btw recently I also had several emacs crashes; it always happened when I was working in org-mode or markdown-mode

<alezost>janneke: I don't recommend to use (require 'foo) because it fails if 'foo' is not available (as you could see). Rather use something like this: (when (require 'guix-prettify nil t) (global-guix-prettify-mode))

<alezost>this will never fail

<janneke>alezost: thanks

<paroneayea>rekado_: well lookie there

<paroneayea>yeah, it was an easy fix!

<paroneayea>gah

<paroneayea>I don't seem to have multi-monitors working right in stumpwm anymore though...

<rekado_>they say you need xdpyinfo for things to work right

<rekado_>I can’t test this because I only have the laptop screen.

<paroneayea>oh right

<paroneayea>perfect :)

<paroneayea>now I'm back to stumpwm land

<paroneayea>woohoo!

<rekado_>yay!

<rekado_>are you using a pretty font? I found the default on GuixSD to be unreadable.

<paroneayea>rekado_: whatever this font is, it's not pretty :)

<rekado_>I’m using (set-font "-adobe-courier-medium-r-normal--14-100-100-100-m-90-iso8859-1")

<rekado_>after doing this:

<rekado_>(run-shell-command "xset +fp ~/.guix-profile/share/fonts/X11/100dpi/")

<paroneayea>rekado_: I'll try that, thanks

<paroneayea>I thought about maybe using inconsolata

<rekado_>would this look good without antialiasing?

<paroneayea>probably not :)

<paroneayea>looks like stumpwm changed some stuff up

<paroneayea>some of my old hacks don't work

<dadinn>hi all

<adfeno>Hi dadinn :)

<dadinn>a question regarding the GuixSD image: I am trying to boot it in a KVM image but getting "No bootable device." message all the time. Version 0.11 and 0.12 both. Any ideas?

<dadinn>I extracted the xz archive, but the file inside does not have .iso extension

<dadinn>not sure if it is supposed to be used as boot image

<lfam>dadinn: It's not an ISO. It's a bootable disk image, and you can run it in QEMU

<lfam>We want to be able to generate ISOs, too, but so far the work is not complete.

<lfam>There's recent discussion of this in the guix-devel archives. Help wanted :)

<dadinn>lfam: I see, do you have a url to the discussion? Just to put it in my radar

<lfam>dadinn: It starts here, on help-guix: https://lists.gnu.org/archive/html/help-guix/2016-11/msg00073.html

<lfam>dadinn: I believe the hoster serveraptor.com will accept a QEMU qcow2 image and host that for you. I don't know any other VPS service that offers this feature.

<lfam>If true, that wouldn't require any more work. We already have tools for generating QEMU VM images built in to Guix

<janneke>lfam: anything more recent than nov 2016?

<lfam>janneke: I'm not sure... ask paroneayea ;)

<paroneayea>huh what huh? :)

<lfam>Hey, how's the GuixSD ISO project going?

<paroneayea>oh

<paroneayea>I've been working on 8sync, not guix deployment stuff :)

<jmd>lfam: What!?!!

<janneke>ACTION reads *help wanted*, looks, sees 2 months ago, meh...

<janneke>:-/

<jmd>Does the International Standards Organisation want to make Guix into a published standard?

<paroneayea>lfam: ".iso" images :)

<paroneayea>er

<paroneayea>jmd:

<paroneayea>joliet whatever stuff, ie cdrom images :)

<jmd>There is no operating system but GNU, and Linux is one of its kernels.

<lfam>Heh

<jmd>Guix was created in the image of ISO.

<paroneayea> https://github.com/emacs-jp/replace-colorthemes I guess I ought to package this sometime soon

<paroneayea>now that I've reverted to using snow again from my ancient color-theme.el copy

<jmd>Talking of snow. I wanted to package xsnow, but I don't think the licence qualifies as free. Any insights?

<civodul>efraim: master would be fine (and fewer packages to build on your own)

<janneke>silly question: i sometimes manage to get stray bright yellow hilights in emacs (touchpad / palm detection?) and i don't know how to get rid of them

<Somelauw>i guess i can now install anything i want through guix

<Somelauw>What do i need the guix-daemon for?

<Somelauw>apt-get doesn't need a daemon?

<lfam>Somelauw: The guix-daemon is what builds things and accesses /gnu/store: https://www.gnu.org/software/guix/manual/guix.html#Invoking-guix_002ddaemon

<lfam>It lets unprivileged users get the advantages of building in a chroot (requires root) and provides transactional access to /gnu/store

<adfeno>lfam: A quick question: Is it possible to, instead of starting guix-daemon through systemd or Upstart: start it from cron job?

<dadinn>lfam: how can I sign up for the mailgroup?

<adfeno>(root's cron job of course)

<lfam>You can start the guix-daemon any way you can run a program

<lfam>dadinn: There are a few mailing lists. This page helps you find the right one: https://www.gnu.org/software/guix/contribute/

<adfeno>Some hours ago, I remember Somelauw having issues with systemctl enable.

<lfam>Somelauw: Did you get that sorted out?

<lfam>I saw your questions and started a discussion on our mailing list about how to avoid them

<adfeno>Because his version of `systemctl enable` didn't (or still doesn't?) support following symbolic links.

<lfam>Basically, we decided to take advantage of some new features in systemd, but it seems that many distros don't include this version of systemd yet.

<Somelauw>adfeno: yes, i had to copy the file and i believe run something like system-reload or the like

<adfeno>ACTION wishes he could send his patch to games.scm (update abbaye) and also try updating recipe to manaplus (games.scm also)... But mail service provider's SMTP server is broken. :(

<lfam>Somelauw: Did you edit the ExecStart value of the file?

<Somelauw>probably not

<Somelauw>because i don't know what that is

<Somelauw>i didn't add +x

<adfeno>lfam and Somelauw: Hm... I wonder if our systemd service file has the paths hardcoded.

<alezost>janneke: what you describe reminds me about (setq visible-bell t) but it's probably not that as it flashes when you press C-g

<lfam>Somelauw: You should edit the ExecStart value. You need to replace the part that begins with '/gnu/store...' with '/var/guix/profiles/per-user/root/guix-profile/bin/guix-daemon'. Leave the rest of the line in place

<Somelauw>lfam: why and what file?

<adfeno>lfam and Somelauw: ↑ that's what I was talking about .:)

<janneke>alezost: thank...got a very fast answer from wasamasa on #emacs

<janneke>it's called secondary region and started by meta-mouse-1

<janneke>prolly a touchpad/palm detection thing

<Somelauw>adfeno: Do I need to change the service file and why?

<adfeno>Essentially, since he copied the file, he has to watch out everytime he does `guix package -u` (upgrades), because these can change the store location of the "guix" package.

<lfam>The file tells systemd to execute a particular build of the guix-daemon, because it's assumed that the file will live in root's profile (~root/.guix-profile) and be updated when Guix is updated. Since you copied it, the file won't be updated to refer to new versions of the guix-daemon.

<lfam>This means that your daemon could be garbage collected later.

<lfam>The path I'm suggesting you use will have the equivalent effect: refer to the current version of the guix-daemon in root's profile.

<Somelauw>i can prevent GC by editing it?

<alezost>janneke: oh, I see, btw you can change that face with "M-x customize-face secondary-selection"

<lfam>I recommend you read the manual section 3.5 Invoking guix gc.

<janneke>alezost: ah...well i first need to read about it and find out if i have a use for it

<janneke>;-/

<janneke>having a friendlier default face for it could be nice...it's *very* in your face

<Somelauw>adfeno: GUIX_PROFILE="$HOME/.guix-profile" \\

<Somelauw>ExecStart=/gnu/store/gj2i5xs7nm80mwrb139pa8iqfccw08zz-guix-0.12.0/bin/guix-daemon

<Somelauw>ExecStart=/var/guix/profiles/per-user/root/guix-profile/bin/guix-daemon

<lfam>Somelauw: That line should still have '--build-users-group=guixbuild' at the end

<Somelauw>okay, it's there

<Somelauw>i'll also restarts the guix-daemon service

<Somelauw>systemctl restart guix-daemon.service && systemctl daemon-reload

<lfam>Somelauw: You have to reload before restarting

<lfam>Reloading tells systemd to re-read the service file.

<Somelauw>ok

<Somelauw>i have re-restarted

<rekado_>janneke: it’s light gray in my theme. (solarized)

<janneke>rekado_: that's probably nicer. i don't think i selected a theme

<ng0>I'm making progress, for whatever reason, with palemoon again.

<ng0>the bigger task will be unbundling for real.

<ng0>or better, check what I forgot the first time

<ng0>browsers be like, suddenly it works and you don't know why it took months to get there

<Apteryx>Is there any recipe to follow for creating patches to be applied in a package definition?

<cbaines>If its a simple change, there are the substitute and substitute* utilities

<Apteryx>cbaines: Ah, right. I think a substitute could do.

<Apteryx>This will be simpler.

<ng0>Uhm... do we have a note somewhere already that we don't use tarballs from melpa? If not can we add it somewhere or to the elpa melpa importer? I regulary forget it

<quiliro>hello

<quiliro>i wonder why desktop.scm does not have all of gnome

<quiliro>and how to install all of gnome with guix package -i

<cbaines>quiliro, I don't think "all" of Gnome has been packaged

<cbaines>and some of the packages don't work as well as they could do

<rekado_>we have a “gnome” package.

<rekado_>it’s a meta-package installing all the gnome things.

<rekado_>tb

<rekado_>oops

<quiliro>thanks rekado_

<quiliro>nice to wee you here

<Apteryx>cbaines: I'm replacing an already present patch, so in the end it's simpler to make another patch, even if the change is relatively small.

<efraim>well, i made new aarch64 bootstrap-binaries and put them up, now to see how it works out

<Apteryx>Which switches should I pass to diff to generate the usual format? Or does it not matter?

<rekado_>diff -u

<Apteryx>rekado_: Thanks. And the convention seems to be that the A (original) file is named "<actual_name>.orig" ?

<ng0>depends

<ng0>or it's in an entire new folder

<ng0>that's how I do it

<ng0>diff -u orig-folder/path/file.c b.orig-folder/path/file.c

<ng0>or the other way around idk, i'm distracted

<Apteryx>I guess patch doesn't care about file A in the patch name.

<Apteryx>It applies whatever change is defined there to file B.

<lfam>It's easiest to use Git, with `git diff` or `git format-patch`

<ng0>but that only works in a git controlled folder.. or what was this patch question about?

<rekado_>ng0: you can easily put a directory under version control. I sometimes do this when I see that the changes aren’t going to be simple.

<ng0>ah yeah

<ng0>I just copy folders

<ng0>so there's ~/re-src/$name/b.$name-version and without b.

<ng0>git is probably easier

<Somelauw>Is there zsh completion for guix?

<rekado_>depends on the case. I sometimes work with two directories as well.

<cbaines>Weird question, has anyone encountered guix loosing patches? I'm building guix from a git repository checkout at a specific revision, and it looses one specific patch, so I get the following error when I try to use it: guix environment: error: libxt-guix-search-paths.patch: patch not found

<rekado_>I have this patch in master.

<quiliro>gnome-calculator will not work

<quiliro>GLib-GIO-ERROR **: Settings schema 'org.gnome.system.proxy' is not installed

<cbaines>The patch is present in the repository, and I've traced through the code putting it in to the store, and its not excluded

<cbaines>but if I look at the generated package in the store, its missing just that patch

<rekado_>cbaines: it could be missing in the list of patches in gnu/local.mk

<lfam>I frequently use `git init && git add -A && git commit -m "Initial import"` while creating patches in unpacked tarballs

<rekado_>cbaines: this was fixed in 5713993f3549feb3b4ecb55f486df68ad79d6c66

<civodul>Somelauw: there's a zsh completion file, yes

<Somelauw>civodul: where is that file to be found?

<civodul>in the source tree it's under etc/completion, and then it gets installed somewhere where zsh should find it AIUI

<cbaines>rekado_, ah, great, I'll try updating then

<cbaines>quiliro, those errors are common with Gnome apps

<cbaines>at least in the version of Guix I'm running, the gnome-calculator works on GuixSD under Gnome

<paroneayea>wow, I'm feeling much happier having moved from xfce -> stumpwm.

<paroneayea>ACTION almost wrote xkcd -> stumpwm

<paroneayea>speaking of xkcd, https://xkcd.com/1785/ may be a little bit too true for those of us running as libre of hardware as we can :)

<civodul>ah ha :-)

<Apteryx>Hm, if I use substites* in phases, the sources won't be touched when using 'guix build -S', but when using patches in the origin field the sources will appear patched already. Which is better practice?

<Apteryx>paroneayea: I've been using ratpoison for 2 months; uses 5 MB of ram. I want to try stumpwm at some point, since it's written by the same people.

<Apteryx>Coming from Ubuntu (Unity) my old system is very well refreshed ;)

<Apteryx>rekado_: Right, a git init could do it. I'm not sure how smart is our patcher, safest would probably be to version the full folder (Python).

<Apteryx>Or even one folder up, it seems.

<Apteryx>Anyway, I'll figure it out, thanks :)

<cehteh>anything changed in the offloading thing recently? i get ssh auth errors, but logging in manually works

<jmi2k>How can I mount a FAT filesystem by its UUID in my config? It's not a real UUID, but it's what points to my disk in /dev/disk/by-uuid. Referencing that path seems to fail.

<cehteh>what does 'blkid' show?

<cehteh>mount UUID=.... /mnt

<jmi2k>cehteh: I'm talking about the system configuration file

<cehteh>yes that should work there too

<jmi2k>cehteh: blkid show the UUID is B75A-021B

<cehteh>well i guess a bit but something along (title 'uuid) ... need to read the doc

<jmi2k>cehteh: that's what I use for the root fs, but it doesn't work with that (maybe because it's not a proper UUID).

<cehteh> https://www.gnu.org/software/guix/manual/html_node/File-Systems.html#File-Systems

<cehteh>could be

<cehteh>no idea then

<cehteh>why does it fail? does the code make some validity check on that which is overcautionous?

<cehteh>does the mount UUID= way work?

<jmi2k>cehteh: yes, and thanks for that, TIL

<jmi2k>cehteh: here is what happens http://paste.lisp.org/+77S3

<cehteh>meh .. i am not a schemer

<jmi2k>I'm starting to feel comfortable with it, but I had to learn it from scratch to start doing something useful in GuixSD :P

<jmi2k>Well, at least, thanks for the mount trick

<cehteh>just wanted to figure out at which point it fails

<cehteh>weeh . .something is broken here, i installed 'lsh' its listed as installed but its not in path

<cehteh>even after reboot

<jmi2k>cehteh: System-wide or in your profile?

<cehteh>roots profile .. and iirc its installed system wide as well, have to check

<cehteh>no only lsh-service system wide

<jmi2k>Ok, try this in a terminal: guix environment --ad-hoc lsh

<cehteh>just remove and reinstall .. lets see what happens

<cehteh>if that fails i try

<jmi2k>If running lsh inside the shell generated by that command succeeds, everything it's fine and you can install it in your profile or define it in your config file.

<cehteh>i installed it before and it worked

<cehteh>but some reboot later it is listed as installed but not in path

<cehteh>that environment command worked

<cehteh>now after reinstalling it works again, wtf

<cehteh>i try to get offloading to work but it fails with access denied

<jmi2k>No idea, never touched that, sorry. Good luck!

<cehteh>i'll try

<mekeor>remember me saying i can't boot guixSD? it was because the keyboard layout at the LUKS prompt was not set to my preference. anyway: i still can't boot guixSD but this time it's not guixSD's fault: it's because of my laptop, i guess some hardware problem. maybe you have an idea? http://superuser.com/q/1167705/648469

<cehteh>mekeor: try some other linux live distro from usb flash or cdrom

<cehteh>if that works investigate what hardware you have there

<cehteh>guixsd is a bit picky because it uses the linux-libre kernel without any non free blobs

<mekeor>i used debian non-free before. my laptop often was beeping, too, while booting. but this time, with guixSD, the problem is that it takes too long too load the OS because i have to type in my LUKS-pw twice

<mekeor>cehteh: it's not guixSD's fault. it's not because of linux-libre. the beeping-and-then-halting-problem also occurs if i just stay in BIOS

<cehteh>then some hardware problem?

<mekeor>yeah, i guess so too

<mekeor>but i have no clue

<cehteh>dust blocking the vents? memory bar oxidized, doesnt give good contact, remove/reinsert it, same for hdd/ssd

<cehteh>you may try to stip the hardware down, remove anything not needed (only one memory bar, no wifi card, only one or no drive)

<cehteh>then work towards until you find the bug

<cehteh>also some hardware vendors release lists what the beepcodes at start mean

<adfeno>Hm.... Wait, I'm trying to see video.

<adfeno>Must use youtube-dl -o - [link] | mpv -

<adfeno>Because YouTube requires non-free JS.

<cehteh>sounds like overheating? can you monitor the heat form the bios?

<mekeor>i'll check, give me a sec

<cehteh>older laptops often have dust blocking the fanholes

<cehteh>or heatsink making bad contact because the thermal paste got brittle

<mekeor>i can't check termperature in BIOS

IRC channel logs

2017-01-15.log