IRC channel logs


back to list of logs

<wheeler>Hope everyone's well and staying healthy. Packaging audacious for guix but running into some issues with audacious-plugins. A separate package but it insists on writing to the /gnu/store...-audacious-4.0/lib/audacious directory. It says permission denied. Very new to packaging but wanting to learn how to do the tightly coupled but separate package thing. Anybody know the right way to handle this?
<leoprikler>Probably some variant of substitute*
<wheeler>Know of any good packages I can look to for an example?
<leoprikler>not of the top of my head
<leoprikler>however, I located the culprit
<leoprikler>plugindir=`pkg-config audacious --variable=plugin_dir` in
<leoprikler>you should probably substitute that to plugindir=(assoc-ref outputs "out")/lib/audacious
<wheeler>Sounds good. I'll give it a whirl after dinner. Thanks for the help
<apteryx>sneek: later tell sirgazil emacs-elpy fixed on master
<sneek>Got it.
<brendyyn>My directory index is full. i think this is why the gnu store cant create symlinks in the store sometimes
<brendyyn>im searching but dont know if there is any way to increase it?
<brendyyn>for ext4
<apteryx>leoprikler: I implemented the eval trick, tested it well, and merged to master as afc6b1c0b635e3268795c0f766be408c5e9858e7.
<brendyyn>I was confused that Ack wasn't included. I guess no one wants it :d
<brendyyn>apparently i created a package for it in 2017
<lfam>brendyyn: You should search the mailing list archives of guix-devel and bug-guix. I think we have seen that problem before
<nckx>brendyyn: The only way to increase it is to wipe the file system and recreate it with a higher inode count. In that case, TBH, I'd recommend switching to btrfs while you're at it.
<brendyyn>it's a shock that a default ext4 system isnt enough for guix
<nckx>That's just the answer to your question. Doesn't mean you can't delete inodes somehow.
<lfam>I don't think that inodes and the directory index are the same
<lfam>Well, not sure
<brendyyn>i have plenty of inods
<lfam>I found the thread:
<brendyyn>this seems to be some md4 hash collision thing
<nckx>Oh, sorry. I didn't realise the dirindex could fill up.
<brendyyn>lfam: thats a private link
<nckx>Ah, that makes more sense.
<lfam>brendyyn: Alright, hold on
<lfam>I don't know how I am "logged in" there but that's another question
<brendyyn>i think its probabilistic that there is a collision. my system hasnt crashed so it seems it handles it somehow?
<nckx>brendyyn: You can disable dir_index entirely using tune2fs.
<lfam>I don't think it's a collision. It's a limit
<nckx>I don't know if that murders performance.
<lfam>Here's the first message, I'll post follow-ups:
<nckx>brendyyn: You can try other hash algoes too, e.g. tune2fs -E "hash_alg=tea" /dev/foo.
<lfam>This is the one where Mark figures out what's wrong:
<lfam>The solution was to increase the size of the virtual disk we were using
<lfam>Another classic Mark Weaver debugging story
<lfam>We should print a book someday
<brendyyn>he has a history of this does he?
<lfam>I'd guess that he has found and debugged more serious yet tricky bugs than anyone in Guix
<lfam>Let me try to find my favorite one
<lfam>This one: <>, or <>, "Store references in 8-byte chunks in compiled code"
<lfam>The timestamps of the messages show that he stayed up quite late to hunt down the bug
<lfam>We ended up having to patch GCC
<lfam>Like any good story, the opening line grabs you by the collar: "After running "guix gc", I've found that my fonts are broken."
<lfam>I'd read it on, the new interface sort of butchers a lot of the text
<Veera>Which pkg has git send-email
<brendyyn>Veera: git does, but it's in the send-email output
<brendyyn>guix package -i git:send-email
<rain>uugh. are NFS mounts broken too? :(
<rain>i can see that there is a package for it, but installing it doesn't let me mount the export from my ubuntu.
<apteryx>rain: sudo -E mount.nfs
<rain>oh. neat.
<apteryx>or install mount-utils as a system package then just use 'sudo mount ...'
<apteryx>err, should have read sudo -E mount ... (if you have installed mount-utils in your user profile only).
<apteryx>I think it should work, let me know.
<rain>well it doesn't fail immediately but it prints some ...warnings? errors? not sure.
<rain>let's see if it can handle a guix system init on the mount...
<apteryx>eh, what are you attempting? boot from NFS?
<apteryx>cause we don't have support for this yet in Guix, I'm afraid.
*apteryx is keenly interested in the matter, though.
<rain>no, install to another machine without wasting a DVD
<apteryx>ah, I see :-)
<rain>i only have an old Ubuntu Studio live DVD
<apteryx>no USB?
<rain>my pendrive is dead :/
<KE0VVT>rain: You can read DVD?
<apteryx>well, let me know how it goes :-)
<rain>well, it doesn't seem to mount. the command just hangs.
<rain>blech. i'll just install Ubuntu Studio and then install it from inside that.
<rain>thanks for the help tho
<apteryx>that should work too
<brendyyn>spk121: Are you the developer of guile-gi? I'm tryning to run the example with guix environment --ad-hoc -l guix.scm guile; guile-gi examples/browser.scm, but it says guile-gi doesn't exist. seems that instruction is wrong?. Below it the instructions for building from git dont work either; they say found development files for 3.0, but guile is 2.2
<guix-vits>query sneek
<guix-vits>Hi Guix.
<spk121>brendyyn: rekado did push a new guile-gi package a few days ago to Guix. Did you try that version?
<brendyyn>spk121: I haven't yet, but i thought it would be good if the instructions worked :P? if i wanted to build my own GUI program, should i just use the one in guix?
<brendyyn>hmm no such command guile-gi
<spk121>brendyyn: Well, it "works for me" as they say. But for Guix, because of how it needs to update so many environment variables and path, the Guix package is safer than building from master. The next rev 0.3.0 is just waiting on me to finish writing this manual.
<brendyyn>i tried it in --pure
<spk121>oh wait, where are you seeing instructions that say "guile-gi"? It is supposed to be 'tools/run-guile'
<brendyyn>yes in the readme
<brendyyn>since im a good hacker i do what the readme says xD
<spk121>ha. I fixed that mistake in the readme in only one of two places.
<brendyyn>ok so ill assume the git repo and 0.2 are incompatible
<spk121>I'm so close to getting 0.3.0 out the door: I just want the manual to make more sense.
<brendyyn>thanks. its great. id love to create my first gui program one day
<brendyyn>spk121: i think you have a bug in guix.scm. At the bottom, shouldn't it be (list guile-next guile-next-gi)
<brendyyn>is suspect you have a .guile-version file so you didnt hit this bug
<spk121>brendyyn: that makes a lot of sense, actually
<brendyyn>In procedure require: Typelib file for namespace 'Gtk', version '3.0' not found
<spk121>There it is looking for the file Gtk-3.0.typelib
<str1ngs>I think this case installing gtk+ and gobject-introspection resolve that.
<brendyyn>there is no gtk+ in the package definition, interestingly enough.
<str1ngs>you can test with guix enivronment --ad-hoc gobject-introspection guile guile-gi gtk+ . IIRC
<str1ngs>brendyyn: ^
<str1ngs>this is assuming you are writing something you that is going to run locally
<brendyyn>I think that programs that include a guix.scm in their repo, indicating guix support, should Just Work. it's great customer experience. when guix takes over the world it would be great if when someone sees a guix.scm in a projects root, they can think "sweet, i can just run guix environment -l guix.scm, and it will all magically work
<str1ngs>this does not apply to gobject-introspection
<brendyyn>what do you mean?
<str1ngs>it's not possible to know all the runtime dependancies a user might want to use with GI
<brendyyn>it should be possible to know what is needed to get the example scripts to run?
<str1ngs>guix enivronment --ad-hoc gobject-introspection guile guile-gi gtk+ webkitgtk . should probably work
<str1ngs>granted for the examples there might be away to improve on this spk121 might have some idea.
<str1ngs>but I'm not confident making gtk+ webkitgtk inputs is probably the best thing here
<str1ngs>or if that would just working without using propergation.
<brendyyn>doesnt seem to be making a difference.
<str1ngs>brendyyn: which example did you want to run?
<brendyyn>str1ngs: any of them
<brendyyn>there is browser.scm and editor.scm
<brendyyn>I see, looking at the examples, they doo require that stuff without loading it. i guess it makes sense. they require more than just guile-gi's deps
<brendyyn>now i get a completely different error
<brendyyn>In procedure module-lookup: Unbound variable: application:new
<str1ngs>brendyyn: with guix environment -l ./guix.scm --ad-hoc gtk+ gobject-introspection webkitgtk
<brendyyn>str1ngs: yeah i seem to have the deps now. i should have thought and read a little further before complaining
<brendyyn>now im just getting a scheme error. maybe the examples havn't been updated yet
<str1ngs>then try with ./tools/uninstalled-env guile examples/editor.scm
<str1ngs>but... caveat. when dealing with the source tree it's going to be less stable .
<str1ngs>they say using guile-gi from guix proper
<brendyyn>i dont think the examples are in guix proper.
<apteryx>bandali: I've started looking at building all of our packages with Emacs 27. Doesn't look too bad so far.
<str1ngs>brendyyn: I mean like so. guix environment --ad-hoc guile guile-gi gtk+ gobject-introspection webkitgtk -- guile ./examples/editor.scm
<obeafrae>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<obeafrae>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<obeafrae>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<brendyyn>oh dear, not spammers again
<Glefl>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<Glefl>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<esisly>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<Glefl>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<esisly>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<esisly>Dear chatters, please join ##ComputerTech123 - the best IRC channel. Regards, -ComputerTech
<str1ngs>spk121: did you have any luck looking into cannot register existing type 'GtkWidget' ?
***ChanServ sets mode: +o nckx
***nckx sets mode: +q $~a
***nckx changes topic to 'GNU Guix | register your nick to talk! | get it at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
***ChanServ sets mode: -o nckx
<nckx>Sorry about that, folks.
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | register your nick to talk! | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
***ChanServ sets mode: -o nckx
<brendyyn>str1ngs: i was using ./tools/run-guile
***ChanServ sets mode: +o nckx
<Sigyn>** Warning: if there is any bot in #guix which should be exempted from Sigyn, contact staffers before it gets caught **
***ChanServ sets mode: +o Sigyn
***ChanServ sets mode: -o nckx
<str1ngs>brendyyn: that looks better to use.
<brendyyn>im using the guile-next version too
<str1ngs>brendyyn: does ./tools/run-guile ./examples/browser.scm give you a GLib-GObject-WARNING **: 21:11:53.332: cannot register existing type 'GtkWidget'
<str1ngs>assuming you have a working environment of course
<brendyyn>i modified guix.scm at the bottom to this: (list guile-next guile-next-gi gtk+ gobject-introspection webkitgtk)
<brendyyn>and used guix evnironment -l guix.scm
<brendyyn>is that the same thing?
<str1ngs>if you want to use the in source guile-gi yes
<str1ngs>something like guix evnironment -l guix.scm --ad-hoc gtk+ gobject-introspection webkitgtk . should do the trick?
<brendyyn>ill try guix's
<str1ngs>for guix it would be more like guix environment --ad-hoc gtk+ gobject-introspection webkitgtk guile guile-gi. then guile ./examples/editor.scm
<brendyyn>amazing. the editor opens up.
<brendyyn>(editor.scm:25755): dconf-CRITICAL **: 15:18:26.081: unable to create file '/run/user/1000/dconf/user': Permission denied. dconf will not work properly.
<brendyyn>str1ngs: yes i get the gtkwidget warning
<str1ngs>strange. what DE do you use?
<brendyyn>does my system need some dconf service?
<str1ngs>hmm do you have dbus session?
<brendyyn>there is the default (dbus-service) yes
<str1ngs>okay it's not that then.
<str1ngs>I use gnome services so. probably I wont get that error. I can't really test sway would require a session restart here.
<spk121>brendyyn: I don't think it *needs* dconf. You won't get any visual customization that dconf or gsettings would have held
<str1ngs>dconf is generally is for configuration persistants so might not be as critical as the error implies. it the error terminal
<str1ngs>is the error terminal? I meant
<brendyyn>does the browser work for you?
<str1ngs>I'm getting GLib-GObject-WARNING **: 21:17:04.049: cannot register existing type 'GtkWidget' which is not a g-golf issue
<str1ngs>err guile-gi issue
<str1ngs>something related to gobject-introspection and guix. g-golf suffers from this sometimes as well.
<brendyyn> str1ngs:
<str1ngs>I've yet to figure out how best to fix it.
<spk121>str1ngs: you might get some information before it fails by adding the env G_MESSAGES_DEBUG=all
<str1ngs>I know multiple entries in GI_TYPELIB_PATH can compound the issue though
<brendyyn>GI_TYPELIB_PATH is something that may require wrapping.
<spk121>str1ngs: could it be that you've somehow pulled in both Gtk-2 and Gtk-3?
<str1ngs>spk121: when I asked in #introspection on gimpnet. they mentioned probably multiple libraries are being loaded.
<brendyyn>WARNING: (guile-user): `connect' imported from both (gi Gio-2.0) and (gi Gtk-3.0)
<brendyyn>maybe thats nothing
<str1ngs>spk121: try with. guix environment --ad-hoc gtk+ gobject-introspection webkitgtk guile guile-git . then guile ./examples/browser.scm
<spk121>brendyyn: not really a problem. GOOPS should make generics for both
<daviid>spk121: it won't atmatically
<daviid>user need to declare 'their wish to merge', including in the repl
<str1ngs>spk121: I've run into this with g-golf as well. it's does not do it all the time
<daviid>otherwise, the generic will be incomplete 'so to speak'
<daviid>str1ngs: in g-golf it should not be a proble, but if, send a bug report please, in #guile ...
<spk121>daviid: we add the grammar "(push-duplicate-handler! 'merge-generics)"
<spk121>to solve that
<str1ngs>daviid: this is related to GLib-GObject-WARNING **: 21:29:34.629: cannot register existing type 'GtkWidget'. which is not an issue with g-golf but GI in guix. guile-gi suffers from it as well.
<daviid>spk121: where is tha efined? and does this affect all your modules, i mean those modules that needs it of course
<daviid>but this is a guile problem we shoul move this talk to guile
<daviid>be right back, need to prepare something to eat :)
<spk121>daviid: ok, jumping to #guile
<str1ngs>spk121: can you please try try with. guix environment --ad-hoc gtk+ gobject-introspection webkitgtk guile guile-git . then guile ./examples/browser.scm when you have time. which should be on topic for guix since only guix is effected
<apteryx>oops, master is broken
<apteryx>error: linux-libre-arm-generic: unbound variable
<str1ngs>what did you do apteryx ?
<apteryx>hehe, I rebased ;-)
<spk121>str1ngs: Can't do it right now, I'm not on a guix box. Sorry. I'll try it soon.
<str1ngs>brendyyn: thanks for testing that.
<apteryx>commit 058e96c59f0d7700f8301a63154684da58acbad0 is the one not working here
*apteryx zzzz
<str1ngs>spk121: no worries, when you have time. if you have any ideas how to resolve it let me know.
<str1ngs>spk121: I think this happen when using guix on foreign-disto. might save you some time testing.
<raghavgururajan>Hello Guix!
<raghavgururajan>I am getting an error "No CMAKE_Java_COMPILER could be found.". I tried java-janino and javacc as input. Still not working. Am I missing something?
<raghavgururajan>As per the source code, I need "UseJava" module. Not sure which package has it.
<Veera>Hi guix
<Veera>Help needed regarding using guix.git repo
<Veera>What is the recommended procedure for using the guix.git repo for development
<Veera>Should I make a new branch develop patches from there
<Veera>Or should I do directly with master branch
<lfam>Veera: I recommend making a new branch
<lfam>It will be easier to keep track of your work
<lfam>For each project you work on, make a new branch
<Veera>If I make branch for each project, then I have to build guix for each branch?
<Veera>Or is there some other way
<guix-vits>lfam: adding package to guix need a new branch too, btw?
<lfam>Yes, Veera, but only rebuild the changes. You could also try git worktrees so you don't have to rebuild as much
<lfam>guix-vits: Nothing is required about the local development workflow. I'm just giving suggestions
<Veera>Okay I am not familiar with git; I do a simple test and see
<Veera>One thing I have to say building guix requires more than 2GB of ram if not using swap.
<Veera>I made it 3GB it built without failure.
<Veera>Is building guix with less ram and using swap would be light on swapping pages.
<Veera>The iso and vm download and install instructions, say that 512 or 1024MB for running in VM
<guix-vits>Veera: why do you building guix?
<Veera>I am running GUIX in VM and first install failed midway while doing guix building because of low ram.
<Veera>Not full os
<Veera>just guix.git
<guix-vits>Veera, lfam: why? aren't one can create a additional channel and test the package there, then send a patch?
<guix-vits>(i'm not did that, just wonder)
<Veera>Oh there is a easy way?
<guix-vits>i wonder too, never did anything like that.
<lfam>If you want to contribute your changes to Guix, you need to send a patch on our codebase. So you might as well do your work in our Git repo
<guix-vits>lfam: but building Guix from Git isn't needed for that, right?
<lfam>We'd prefer it if you tested your patch too...
<guix-vits>lfam: channel?
<lfam>If you send a patch that applies to guix.git, please test that your patch works in guix.git. You can't do that in a channel
<brendyyn>Veera: Are you building with make -j? maybe reducing the threads allowed would lower ram usage?
<guix-vits>so copying "(use modules" part of some .scm from git to an plain package-definition.scm and building that package on freshest pull isn't enough?
<brendyyn>it will work, you will build your package. but you cant just dump that file in an email, you need to decide how to integrate it into guix.git, and ensure you didn't break anything
<brendyyn>ive seen people completely break things in the past because they were a bit sneaky and didnt test something that was apparently trivial
<guix-vits>thanks people, then i'll build git too, when come up to something.
<Veera>brendyyn: I am just invoking make
<Veera>I will try doing make -j 2
<Veera>Because make -j 1 is slow
<Veera>To be safe i just called make
<brendyyn>how many threads do you have?
<Veera>qemu VM -smp 2
<Veera>And I see in my host os system monitor it using 2 cores
<brendyyn>oh, then maybe it already isn't using many :/
<Veera>2 core with 2 threads/core
<Veera>using all 4 threads
<brendyyn>Has Xfce become more popular than even KDE and GNOME?
<str1ngs>brendyyn: I use xfce for session management and i3 for window management
<brendyyn>str1ngs: I have heard of people using this mysterious hybrid approach
<str1ngs>it's not all that mysterious for me haha
<brendyyn>I think i tried with KDE and KDE was totally confused about how i'd just butchered it's innards
<str1ngs>when though gnome ever gets tiling support that's on par with i3 I'll switch :P
<str1ngs>or I should just switch to sway.. dunno
<brendyyn>I'm on sway. There are no tray icons atm
<brendyyn>at least in guix
<brendyyn>apparently we need some libappindicator things that i dont know where it even exists
<str1ngs>right I use xfce's panel for that
<guix-vits>str1ngs: will be a systemd-tiled connected to mutter via dbus
<str1ngs>haha you funny guy!
<brendyyn>str1ngs: how does a program communicate to the xfce bar?
<str1ngs>I don't do that. I just use xfce plugins
<Veera>I use mate
<brendyyn>I suspect nobody is running enlightenment
<Veera>It good enough like old gnome 2
<brendyyn>I was using terminology, but it fails to run on wayland. now i have termite but there is no copy-paste with the mouse
*guix-vits runned out of enlightement
<guix-vits>brendyyn: what about selecting with a mouse and "Ctrl Shift c"?
<str1ngs>mostly though I just use emacs, nomad and a terminal
<brendyyn>there is no select with mouse i think
<brendyyn>well, not in this terminal with weechat
<str1ngs>I use circe with emacs for IRC
<brendyyn>it seems to but it in some other mode. i can actually click on channels to go to them. ncurses has become an actual interactive gui apparently
<bdju>brendyyn: hold shift when doing copy/paste with the mouse
<bdju>for the terminal parts anyway
<bdju>It makes sure you interact with the terminal itself instead of a program inside it. I have been doing it this way for a while because I use tmux which also has some mouse functions.
<bdju>and I'm also using termite on wayland but actually I don't think that part matterts
<bdju>termite also has a vim mode you can get into with ctrl-shift-space and then you can select stuff with v, copy with y, etc. esc leaves the mode again
<bdju>it may also be enough to disable mouse support in weechat if you can't deal with having to hold shift
<raghavgururajan>Hello Guix!
<raghavgururajan>So I am getting "error: Could not find suitable distribution for Requirement.parse('javalang>=0.10.1')".
<raghavgururajan>I tried openjdk, icedtea and classpath.
<raghavgururajan>Didn't work. :/
<bricewge>Hello Guix!
***apteryx is now known as Guest47456
***apteryx_ is now known as apteryx
<bricewge>Hey raghavgururajan
<guix-vits>Hi raghavgururajan. To fix this try to travel to IRL Java (and forget about the language). Hi bricewge.
<bricewge>You opened a year or so ago, are you still affected by this issue?
<bricewge>raghavgururajan: ^^^
<raghavgururajan>guix-vits I think I am onto something
<raghavgururajan>bricewge You can close it. Thanks!
<bricewge>Thank you :)
*efraim is using enlightenment
<raghavgururajan>So, are MIT license accepted?
<KE0VVT>Why wouldn't Expat be accepted, raghavgururajan?
<raghavgururajan>what's expat?
<raghavgururajan> I am not familiar with both mit and expat
<KE0VVT>raghavgururajan: That's the so-called "MIT" license one sees plastered all over GitHub.
<raghavgururajan>AH I see.
<kmicu>raghavgururajan: many versions are called “MIT” so check if it’s really Expat and not X11 or some other custom version.
<kmicu>(It doesn’t matter how we call it. The important thing is to link to the same license text. Guix could alias Expat as Mit to follow SPDX industry standard.)
<civodul>Hello Guix!
*raghavgururajan is banging his head on the wall
<rekado>raghavgururajan: that’s ill-advised :) Do you need help removing the bricks?
<raghavgururajan>So when I have "python-sphinx" as input, I get 'javasphinx' python module not found. So I packaged "javasphinx" and added as input. Now I get 'sphinx' python module not found.
<rekado>raghavgururajan: and both “python-sphinx” and “python-javasphinx” are added as inputs?
<rekado>raghavgururajan: one thing I like to do as a sanity check is to see if I can import the module in a REPL.
<raghavgururajan>rekado When both are as inputs, I get 'javasphinx' python module not found.
<rekado>could you show us your patch?
<raghavgururajan>Just a sec
<rekado>can you show me the full diff? I don’t see the definition of javasphinx
<raghavgururajan>Just a sec
<civodul>apteryx: (format "~s" (which "sh")) should be (format #f "~s" (which "sh")) or (object->string (which "sh"))
<civodul>the first form is deprecated
<civodul>which is no big deal so far but will be one day :-)
<raghavgururajan>rekado By the way, I was building "liblinphone", not "linphone-desktop".
*rekado is annoyed that the raw view of is not raw
*rekado edits HTML
<rekado>raghavgururajan: why do you use “inputs” in javasphinx instead of “propagated-inputs”?
<rekado>javasphinx needs these other python things at runtime
*rekado changes inputs to propagated-inputs and tries again
<rekado>./pre-inst-env guix environment --ad-hoc python-wrapper javasphinx
<rekado>then “python”
<rekado>then “import javasphinx”
<rekado>your variant fails, because javasphinx cannot find sphinx at runtime
<rekado>changing the inputs to propagated inputs makes it work
<raghavgururajan>So I have to change inputs to propagated-inputs?
<raghavgururajan>On which package-definition? javasphinx or liblinphone?
<rekado>I suggest testing whether it works to import a package in a REPL
<raghavgururajan>Thanks so much
<rekado>always try that first
<raghavgururajan>Sure :-)
<rekado>the fact that “import javasphinx” does not work in an enviromnent containing just Python and javasphinx means that the problem is with the javasphinx package, not with anything else
<rekado>when it told me that it cannot find sphinx I checked the definition of javasphinx
<leoprikler>apteryx: I see that this way even lexical code is wrapped in a gratuitous eval. Will this bite us in the long run?
<rekado>and that’s how I noticed that the inputs must be propagated, because they are needed at runtime.
<raghavgururajan>I understand it better now :-)
<Veera>I want to send xchat pkg as a patch tagged WONTFIX
<Veera>It has the xchat.scm, and two patches file
<Veera>How to send them to
<Veera>Should I make three patches
<Veera>Or one patch with all included
<leoprikler>If your xchat.scm depends on those patches, send it all at once.
<Veera>By three attachments?
<leoprikler>the two patches apply to xchat, do they not?
<Veera>Yes two are needed to build xchat
<Veera>(patches "1.patch" "2.patch"
<leoprikler>In that case, you give them descriptive names and store them in gnu/packages/patches
<leoprikler>You also use search-patches to find them.
<leoprikler>You add all the stuff to the relevant fields in gnu/
<Veera>Sorry I am using serach-patches
<leoprikler>Fair enough.
<Veera>I am Outreachy Contrib
<Veera>Yesterday we had chat here and they decided it should not be included in guix
<Veera>Gabor told me to send the patch as WONTFIX
<Veera>The Guix contribution page does not says much of how to make and format patches
<rekado>Veera: don’t worry about marking anything as WONTFIX. We will do this.
<rekado>Veera: have you read the Contributing section of the Guix manual?
<rekado>Veera: it explains how to submit patches.
<Veera>But how to make patches
<rekado>your case is a bit special, because it has been determined that xchat won’t be added to Guix, but for Outreachy you will need a recorded contribution.
<rekado>so just send your patch to and include in your message a summary of the discussion.
<Veera>Gabor said some link he would provide for that
<rekado>a link for the discussion?
<Veera>link for contributor page
<Veera>Should I send the three files plain as an attachment
<rekado>no, don’t send plain files
<Veera>Or use git format-patch
<rekado>always send patches formatted with “git format-patch”
<Veera>Three separate commits or a single commit with all the three
<rekado>when you add a package definition it should be one commit
<rekado>one commit = one patch
<rekado>the fact that your package definition uses patches is irrelevant here.
<rekado>those patches are added to Guix in the same commit
<rekado>the result is a single commit
<Veera>and for the needed(pkgs)patches
<rekado>when formatted as a patch with “git format-patch” the result is a single patch to send to guix-patches
<Veera>include in your message a summary of the discussion; how
<Veera>Do you mean the msg as body and the patch as attachment
<rekado>send a free-form message to in which you explain that the following message contains a patch for xchat, an unmaintained chat client.
<rekado>then wait for a confirmation email, which will give you a “bug” number and an address to which you can send follow up emails.
<rekado>(something like
<rekado>you then send your git formatted patch to that address.
<rekado>does that make sense?
<rekado>if this is your first message to it can take a little while before you get a response with the “bug” number and address.
<rekado>it can take even a couple of hours
<rekado>the first message is usually held in moderation for a while
<rekado>so don’t get confused by that
<rekado>it’s weird :)
<rekado>but this is the common workflow, especially for when you want to submit many related patches.
<rekado>instead of sending them all one by one to guix-patches you request a “bug” number, and then send the patches to that “bug” email address.
<rekado>you request a “bug” number by writing a so-called cover letter to
<Veera>In the start (20 days ago) I subscribed to guix-devel and
<rekado>subscribing to guix-patches is not strictly necessary — you’ll probably have received quite a lot of email through that list
<Veera>Yes that clear from submitting patches and tracking doc
<rekado>you don’t have to be a member to send to
<rekado>subscribing to that list is recommended for patch reviewers. For new contributors it might be overwhelming.
<Veera>Outreachy page gave them
<rekado>I see.
<rekado>I hope you’ve got a good email client that can filter emails so that your Inbox won’t overflow :)
<Veera>Thanks for help. I will submit them
<Veera>Well I subscribed the msgs are not more than 20
<Veera>a day
<rekado>ah, good.
<janneke>wip-hurd finally shows some success, but also "one or two" failures :-(
<bricewge>I have issues debugging my system configuration, it starts to be a bit complex and the error message aren't helpful at all.
<efraim>janneke: which commit did you build your bootstrap binaries from? I wanted to try piggybacking on it for powerpc-32bit
<sneek>efraim, you have 2 messages!
<sneek>efraim, lfam says: Something like this: `ffmpeg -pattern_type glob -i "render/*.png" -c:v librav1e output-av1.mp4`
<sneek>efraim, lfam says: Of course you need rav1e from the wip-rav1e branch for that. I need to finish that work but struggling to find the focus
<efraim>sneek: botsnack
<Kimapr>anyone knows how to set up multiseat on guix system with one GPU, two keyboard/mouse pairs and two displays?
<bricewge>How do you debug your configuration? With message like “guix system: warning: invalid argument list” or “guix/ui.scm:1876:12: In procedure run-guix-command: In procedure find: Wrong type argument in position 2 (expecting list): #f” for example.
<Kimapr>multi-login is unnecesary
<Kimapr>i can already make two pointers with xinput
<leoprikler>but how would you lock the pointers to one screen?
<Kimapr>that's unneccesary for my purposes too
<Kimapr>and i didn't
<Kimapr>awesomewm doesn't support two pointers correctly
<Kimapr>anyone knows a window manager that does support multipointer?
<janneke>efraim: i have mentioned that in the commit message of: 74e4caa572 gnu: bootstrap: Add support for the Hurd.
<efraim>janneke: thanks
<janneke>it was the previous commit of that commit...however, we rebased a couple of times
<janneke>the resulting hashes /should/ be identical, i think, but we may need to rebuild them and rewrite that commit once again before we merge
<lprndn>Hello guix!
<bricewge>Hello lprndn
<Kimapr>is there any multipointer wm/de in guix repo?
<janneke>is there a nice shorthand for using a shell's environment in an emacs compilation buffer? i am using this two-step kludge:
<janneke>echo bash -c "'source $GUIX_ENVIRONMENT/etc/profile; make'"
<janneke>=> bash -c 'source /gnu/store/4r2jajivkwrj2a5y90pcjr0kmwks6asz-profile/etc/profile; make'
<janneke>and then use that as my make command ...
<civodul>janneke: i don't have anything better to offer
<peanutbutterandc>Hey there!
<peanutbutterandc>Long time no see
<iyzsong>Yeah, hello :-)
<peanutbutterandc>iyzsong, Hey :)
<guix-vits>Hi peanutbutterandc.
<peanutbutterandc>guix-vits, Hello :)
*guix-vits got new ${PS1}
<peanutbutterandc>Does anybody here use haunt? The static website generator written in guile?
<guix-vits>janneke ^^^
<peanutbutterandc>is janneke online?
<peanutbutterandc>I need a little bit of help with my demo haunt website in gitlab:
<guix-vits>just was 20 min back
<peanutbutterandc>When I try to click on a link, it throws a 404:
<peanutbutterandc>While, in theory, I could set builder #:prefix to hello-haunt and move index.html from that dir to a directory above, I was wondering if there was a more elegant solution
<scmguru>Good morning, all.
<guix-vits>peanutbutterandc: when i click "Hello,..." i asked to log in to GitLab.
<guix-vits>Hi scmguru.
<peanutbutterandc>Also, has anyone come across this: ? guile-s-expressions-compile-to-JS
<peanutbutterandc>guix-vits, curioser and curioser
<guix-vits>peanutbutterandc: from w3m;
<peanutbutterandc>It was throwing a 404 for me
<janneke>civodul: good to know, i have avoided looking at direnv stuff ...
<peanutbutterandc>guix-vits, Oh wow. *just looked up `guix show w3m`*
<guix-vits>peanutbutterandc: from nomad (WebKit2GTK) the same.
<civodul>janneke: oh yes, i think cbaines has a nice setup with direnv
<civodul>never tried it but it looks cool
<civodul>howdy iyzsong!
<peanutbutterandc>guix-vits, Yes, sir. I tried that from firefox too (private browsing) and I get the same result.
<iyzsong>civodul: yeah, just got work done, and now relaxing...
<civodul>good, enjoy :-)
<iyzsong>thx, feel a bit guilty about lacking time to hack with guix, but my connection to is still too slow...
<civodul>bah, too bad
<civodul>i remember discussions to set up an rsync mirror
<civodul>not sure what the outcome was
<iyzsong>um, i didn't follow it. maybe a nginx proxy cache is enough? I have my VPS in japan, and it sometimes provide 200KiB/s, and other time almost 1KiB/s. In the end, I think it's just my ISP's speed limiting so bad..
<civodul>would you be able to rsync berlin's 'guix publish' cache in China?
<peanutbutterandc>sorry I had a power outage. Did I miss anything?
<guix-vits>peanutbutterandc: no;
<peanutbutterandc>guix-vits, I was just checking it haha
<peanutbutterandc>janneke, Are you still here? I would like to get some assistance with haunt, if you have the time, please
<iyzsong>I haven't tried rsync from berlin. and i don't have enough resource (only a small 1U/1G/20G VPS) now..
<peanutbutterandc>Also, have I shared this thing here: ? It is a guile-compiles-to-js thing. Looks super neat!
<janneke>peanutbutterandc: the haunt output looks fine, have a look at you webserver root
<iyzsong>peanutbutterandc: to set 'baseuri' for haunt, maybe it's "(blog #:prefix "your.domain/PATH/")" in haunt.scm?
<peanutbutterandc>janneke, gitlab takes everything from 'public' directory. And the url is
<peanutbutterandc>iyzsong, I will try that now. Thank you
<peanutbutterandc>iyzsong, Just ran a local build. It gives creates a 'my.domain/path/' directory and puts everything therein. Which isn't exactly what I want. I think I just want the index.html to be outside of any sub-dirs and the posts to be under a directory name /hello-haunt...
<peanutbutterandc>which I could do just by manually `mv`-ing stuffs around, I figure. I was only wondering if there's a better way to do it
<iyzsong>okay, then i guess you have to looking into the code and hack with it :-)
<peanutbutterandc>iyzsong, That is the problem. I am not yet that much of a guile expert. But yes, perhaps I'll have to go with my manual `mv`-ing for now
<peanutbutterandc>I am completely new to web-designing
<andydarcyjewell>hey guix, I've been trying to remove the "configure" step from my package definition manifest, following the "Extended example" section in
<peanutbutterandc>iyzsong, is janneke the designer of guix website? It looks so cool! And guile's website is one of my favorites ever! Clean and elegant
<andydarcyjewell>I've done what it seems to suggest, but I'm now getting an error message: error: (arguments #:phases (modify-phases %standard-phases (delete (quote configure)))): invalid field specifier
<brendyyn>andydarcyjewell: i think you want a backtick `
<janneke>peanutbutterandc: no, i'm certainly not
<andydarcyjewell>brendyyn: Thanks, but I think I'm using one in the manifest code: (modify-phases %standard-phases (delete 'configure)))
<brendyyn>i mean before #:phase
<brendyyn>`(#phases ... )
<andydarcyjewell>brendyyn: OH
<andydarcyjewell>I thought "that's not in the example!", but it is, though it looks like a bit of dust on the screen!
<andydarcyjewell>brendyyn: Thanks, will try that.
<guix-vits>peanutbutterandc: -- but uses Haunt for sure.
<peanutbutterandc>janneke, I see. :)
<peanutbutterandc>guix-vits, Hmm... yes, that is another one that I did find. In this first phase, I'm just trying to see how viable a website in haunt will be if one does not have dedicated hosting. And I've got the elephant in already (CI/CD is done) but the tail is stuck (the link issue), as they say around my part of the world.
<andydarcyjewell>brendyyn: Yep, that was it, thanks. Now onto the next error! I'll inevitably be back later
<peanutbutterandc>Ah! That is how I know Mr. janneke from! I was so sure I'd read his name a couple times before! He even got guix to run on his pinebook! Neato, Mr. janneke!
<peanutbutterandc>(from joyofsource!)
<rekado>iyzsong, civodul: the only remaining problems with the rsync daemon are: “guix publish” sets restrictive permissions on some files; the rsync daemon should be started by shepherd.
<peanutbutterandc>janneke, Whoa! You are the 'founder' of lilypond too!?????? Seriously?????
<peanutbutterandc>And the mutopia project and everything!
<brendyyn>There was a legend living amongst us all this time??
<peanutbutterandc>I had a feeling I was talking to somebody super important.... whoa. Cool!
<peanutbutterandc>brendyyn, IKR!?
<peanutbutterandc>so much super uber smart people hanging out around here. And I'm troubling them with 'my site isn't working'
<brendyyn>I mean, besides the Great Mark DeWeaver of Bugs.
<brendyyn>I haven't heard of mutopia
<brendyyn>I guess I learnt with tabs because I was too rock n roll to learn sheet music
<peanutbutterandc>It looks neat. Kinda' like IMSLP, I guess. But I am not that into classical music, yet. But I do use lilypond. It's been of great help to me. I use it to transcribe ukulele tabs with personal annotations (RH and LH fingerings, etc.)
<peanutbutterandc>brendyyn, You might like TuxGuitar then!
<brendyyn>are they still stuck on guile 1.8? i heard they had some hard times
<peanutbutterandc>brendyyn, Yes, it seems so...
<peanutbutterandc>I wonder how many legends I bump into every time I am here... without even knowing.
<brendyyn>peanutbutterandc: guitar pro was way better tho
<peanutbutterandc>Thank you, Mr. janneke for lilypond and all the things that you're doing. I, for one, am most certainly blessed by your works. :)
<peanutbutterandc>brendyyn, Tuxguitar is GPL3, though. Freedom is worth the inconvenience. :D
<peanutbutterandc>Edit: LGPLv2.1 (from wikipedia)
<brendyyn>i was happy with plain text tabs until all these stupid websites started become advertisement ridden garbage with these proprietary formats
<janneke>peanutbutterandc: you're welcome :-)
<peanutbutterandc>janneke, wow... it is an honor, sir! You rock! :)
<peanutbutterandc>brendyyn, whoa!!!!!!!! all these super cool people around here!!!!!!!!!!!!!!! But you should so totally try out TuxGuitar!
<peanutbutterandc>brendyyn, There's this beta website: and they've got 'stairway to heaven' and everything!
<rekado>peanutbutterandc: “But I am not that into classical music, yet.” <— the “yet” is the right attitude!
<peanutbutterandc>rekado, You do not happen to be Mr. Ricardo Wurmus, do you? Oh... yeah. I know when I'm no yet smart enough to appreciate the beauty in things. Like LISP. And even guile and scheme and guix and all the crazy sorcery that the wizards around here do it. :)
<rekado>peanutbutterandc: yes, one of my other names is Ricardo Wurmus.
<peanutbutterandc>I only hope to be smart enough to do magic with guile and stuff... like all the crazy smart hackers around here. And hopefully I'll learn to appreciate classical music too....
<peanutbutterandc>rekado, Whoa! You're one of the maintainers!!!!!
<rekado>all the Guix co-maintainers are here: civodul, nckx, apteryx, mbakke, and rekado
<peanutbutterandc>It's kinda' intimidating to be back in this channel after so long and being in the presense of legends.... I'm fanboying right now.
*guix-vits (pen sounds) thier leader are...
<peanutbutterandc>Mr. nckx is the coolest. He's helped me out a lot. Mr. Courtes, I know, was the one who started guix. Super cool, too. I haven't been well-acquaintaited with the rest... sadly
<brendyyn>peanutbutterandc: What are you working on with Guix?
<peanutbutterandc>brendyyn, I use guix as my primary package manager. I am actually working with haunt and not guix. Also, I'm just a normal user (hobbyist). Not a professional. Hence the fanboying. :)
<brendyyn>peanutbutterandc: It's ok I've spent 3 years working up from Novice to Beginner here.
<peanutbutterandc>The reason I love guix is because now I don't have to worry about screwing up my machine when I install software. I can clean everything up back to the way that it was. It is so very super cool. It has put fun back into computing for me. Guix and Guile.
<peanutbutterandc>brendyyn, I see. I hope I get to your level of expertise someday, too.
<rekado>there’s no need to be intimidated. We all struggle with remembering how to use xrandr, and we are all puzzled by the finer points of GDM…
<brendyyn>peanutbutterandc: well today i started trying to translate ed25519 cryptography to guile so its forcing me to learn for sure.
<janneke>^^^ what rekado says!
<peanutbutterandc>rekado, I literally have a printout of 'GNU Emacs Survival Card' next to me. I am that level of n00b. But yes, I am so glad that the maintainers and all the crazy smart people here are so very helpful. The other day, I was talking to a gentleman who was developing a bootstrapping thingy that, in theory, could restore all of our present day technology in 7 days (or so), even in case of a total technological apocalypse. Everyone is so very smart
<peanutbutterandc>here! :)
<brendyyn>peanutbutterandc: Emacs is simply not designed that well. Magit is an example of an Emacs tool that is actually good. You just need to press 'h' to start learning to use it.
<peanutbutterandc>janneke, I realize that you, sir, too, have something to do with the M2 project. GNU Mes and them are related, if I am not mistaken.
<brendyyn>peanutbutterandc: Things like helm help you with popups that show your bindings
<peanutbutterandc>Such crazy smart wizards hanging out here. And all willing to help us n00bs. We're truly living in a great time.
<peanutbutterandc>brendyyn, Hmm... how can I go about installing magit using guix? (I use guix as my emacs package manager too. emacs-yaml-mode, etc.)
<janneke>peanutbutterandc: M2 is made by OriansJ; Mes just make happy use of it
<brendyyn>peanutbutterandc: Magit appears to be in guix. I assume you would (require ...) it and then it launches with M-x magit-status.
<brendyyn>peanutbutterandc: I actually use Spacemacs which is a config with loads ot pre configured stuff.
<brendyyn>it comes with it automatically
<peanutbutterandc>janneke, Like I said, you all are super smart and super modest. And not to mention lilypond. Such beautiful scores! And such great program! I use it to transcribe my ukulele sheets with annotations.
<guix-vits>peanutbutterandc: what is your instrument?
<peanutbutterandc>janneke, And because it is Free and Open Source, I know that even my kids and grand-kids (I'm still in my early 20s) will be able to use it. Your work has benifitted me so much. And lilypond prints such neat scores!
<peanutbutterandc>guix-vits, I am not that great with anything really. But I've been trying to learn the ukulele. Here are a few of my transcriptions:
<peanutbutterandc>brendyyn, Hmm... I always get overwhelmed by how much things there are to be done with emacs (and family). But I'll try to look those up as well. Thank you very much.
<brendyyn>peanutbutterandc: thats why i use spacemacs. it doesnt give you something perfect, but its all setup and integrated. if you open a python file, it will automatically download emacs packages for working with python.
<peanutbutterandc>brendyyn, Oh wow. That does sound very helpful. I will take a note of that as well, then.
<rekado>peanutbutterandc: thanks for sharing your transcriptions! These will be useful for the two Ukulele players in my apartment.
<janneke>peanutbutterandc: thank you -- we all have our heros and we do what we can
<peanutbutterandc>rekado, I will consider that my small token of gratitude to you, sir. I would also recommend the CC-BY licensed (thereby Open Culture) books by Mr. Al Wood ( for the learners. I hope they do great! (P. S: My repo uses git-lfs to track tuxguitar (.tg) files ) :)
<peanutbutterandc>janneke, I would suppose so, sir. I am extremely grateful for your work. Thank you very much once again.
<peanutbutterandc>Okay... so the workaround seems to work: but I am surprised as to the URL of the link. I wasn't really expecting that. This makes me think even more lowly of my 'hack'/work-around. If any haunt user comes around, and has a better suggestion, please do let me know.
<rekado>peanutbutterandc: you can add different builders to place files wherever you want them.
<rekado>peanutbutterandc: see;a=blob;f=haunt.scm;h=461d39b5f9bcd7760e1a858a19e721463c1822d5;hb=HEAD#l260
<rekado>you can use “make-page” to place things at arbitrary locations. That’s what “pin-blog-post” does.
<sirgazil>I decided to use slim and mate instead (too much manual configuration with sway).
<sneek>Welcome back sirgazil, you have 5 messages!
<sneek>sirgazil, apteryx says: emacs-elpy fixed on master
<sneek>sirgazil, guix-vits says: for input configuration in sway, please see
<sneek>sirgazil, guix-vits says:
<sneek>sirgazil, guix-vits says: `find /gnu/store/ -name base.lst`
<sneek>sirgazil, guix-vits says: and `grep "grp:.*toggle"` on it, for example (to use in sway's config)
<sirgazil>apteryx: Thanks a mil!
<sirgazil>guix-vits: Ah, then I already have keyboard configuration in sway. Thanks :)
*sirgazil found several little annoying bugs while getting started with MATE.
<peanutbutterandc>rekado, Oh wow. That means I can probably (pin-blog-ost as index.html) too. I will try doing that then. Thank you very much! :)
<rekado>sirgazil: thanks for your good bug reports! I appreciate you taking the time to test things and report problems you find.
<anadon>Good morning all
<guix-vits>Hi anadon
<anadon>I need to get back to testing the package I made, then submitting the PR. If only my work's software development was as matured as this.
<civodul>the GNOME Shell bug is particularly serious
<civodul>not sure how we can go about debugging it
<civodul>perhaps a bisect or something as a start?
<rekado>civodul: would it make sense to first upgrade to a later version of Gnome?
<rekado>we’re at 3.32 but 3.36 is the latest
<civodul>rekado: that sounds quite a lot of work, no?
<civodul>i mean it would be ideal, but what's the time scale?
<civodul>and who's volunteering? :-)
<civodul>strangely, my partner's laptop has GNOME and doesn't seem to experience that problem
<rekado>civodul: I’ve also been using Gnome for a few months and I haven’t seen any extreme memory leaks.
<rekado>civodul: I was hoping that upgrading Gnome would become easier the sooner it happened.
<rekado>the previous upgrades were painful *because* we had let so much time pass.
***wxie1 is now known as wxie
<rekado>is the problem even seen in Gnome? Or is it just when using gdm without Gnome?
<rekado>in any case we should enable debugging in gdm (see and inspect the logs
***lx0 is now known as lxo
*rekado just got kicked and banned for spamming freenode o_O
<jonsger>we have a wip-gnome3.34 branch on savannah
<janneke>rekado: wow?! you said a magic word?
<rekado>happened right after my last message. Very weird.
<civodul>rekado: i vaguely remember that recent GNOME relies on systemd some more
<civodul>not sure if it's .36
<bandali>two links i’d bookmarked a few months ago:
<lprndn>if help is needed for Gnome, I might be able to take some time to help ;)
<civodul>bandali: ah yes, that's the blog post i had in mind
<bandali>ha :-)
<brendyyn>Did someone say blog post? yes please.
<jonsger>the poblem when I worked on 3.34 were major updates to glib, gtk etc. which resulted in heavy rebuilds
<apteryx>would someone have an example about adding a keyfile to decrypt a LUKS drive at boot?
<rekado>jonsger: same for 3.28, 3.30, and 3.32 unfortunately. Always world-rebuilding changes via glib :-/
<nckx>Hullo Guix.
***ChanServ sets mode: +o nckx
***Sigyn was kicked by nckx (Sigyn)
***ChanServ sets mode: -o nckx
<anadon>Following , when I get to `$ ./pre-inst-env guix build hello` it fails with tons of warnings about `Unbound variable: git fetch`. Why and how do I fix this?
<nckx>That's for kicking rekado.
<jonsger>maybe we should focus on xfce and/or mate. They seem to be more handlebar
<bgardner>Good morning guix! I'm trying to better understand how my disk space is being so I can repartition and plan better. I have a 30G / partition that crossed 80% usage, so I reviewed my packages and decided that instead of openjdk, I should be using openjdk v11. So I issued 'guix package -r openjdk:jdk -i openjdk@11:jdk', which ran fine. Then I ran a pretty aggressive gc: 'guix gc -d 1w'. My expectation
<janneke>nckx: :-) it gets weirder, for me
<bgardner>was that this would reduce disk space or be the same, but in fact consumption grew by 2%. Why would this be?
<nckx>janneke: What do you mean?
<rekado>bgardner: have you removed your past profile generations?
<bricewge>Do I need to create a configuration record-type for every new service? Even if it only take a list of strings?
<rekado>bgardner: as long as they exist the items they refer to won’t be cleaned up
<rekado>bricewge: yes, please.
<rekado>bricewge: they don’t have to be complicated.
<bgardner>rekado: Since I don't know what that means I'd guess no, let me read the docs and get back to you
<janneke>nckx: i don't know why or who it was that kicked rekado; but that you can/do kick them ... this is all beyond me :-)
<mbakke>for keeping up with GNOME, we mainly need more ARM build power, so that we can afford the huge rebuild cycles... Also humanpower for actually putting in the work.
<bricewge>rekado: Ok. Just curious, what's the reason behind this?
<nckx>janneke: Don't be confused. Sigyn is Freenode anti-spam bot. There was a small spam wave brewing y'day night when I was about to sign off, so I invited her in.
<mbakke>we should merge that GNOME 3.34 branch to core-updates, though 3.36 is already old by the time that is merged :/
<mbakke>I already tested it last year, worked fine...
***ChanServ sets mode: +o nckx
<rekado>bgardner: “guix package -l” will show you all past generations of your default profile
***nckx sets mode: -q $~a
***nckx changes topic to 'GNU Guix | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
***ChanServ sets mode: -o nckx
<rekado>bgardner: you can delete all past generations with “guix package -d”
<janneke>nckx: ah, now it starts making more sense, thanks -- the spambot is a bit uptight, prolly
<bgardner>rekado: No past generations, just the (current) one.
***ihoiho is now known as vincelevi
<anadon>How do I get details for a failed test group from `make check`? "tests/channels.scm" failed.
<bavier`>anadon: check *.log
<bavier`>i.e. tests/channels.log
<anadon>OK, first patch submitted. :D
***wxie1 is now known as wxie
<bgardner>rekado: Do we have any rough guidelines for how big of a partition to give to guix? Maybe I just haven't been giving enough, I keep running out of space during 'guix upgrade'
<NieDzejkob>you do run guix gc every so often, right?
<bgardner>NieDzejkob: I have it on an mcron job for all my servers, definitely. Checking it manually it says nothing to do.
<rekado>bgardner: “guix pull” also has its own profile at ~/.config/guix/current — check if there are past generations.
<bgardner>rekado: As root or regular, or both?
<rekado>and there are system generations added when reconfiguring
<rekado>for all user accounts that use Guix on this machine.
<sirgazil>When I run "git push" in my git repositories in MATE, I get this error:
<sirgazil>MATE (or whatever it should use for that) never asked for my password.
<bgardner>rekado: Ye gods, that says there are a bunch under there. So regular maintenance should touch these as well?
<bgardner>rekado: By which I mean, I should be checking that periodically as well?
<jonsger>sirgazil: to which repo you wanna push?
<sirgazil>jonsger: one of mine. It works alright in sway (where I get a prompt to enter the password for my ssh key), but in MATE I just get the error and I'm never asked for my password.
<guix-vits>sirgazil: check your machine ports, maybe it's evel Russian хацкер
<guix-vits>trying to prevent you from switching to MATE?
<sirgazil>Is anyone working right now on MATE 1.24?
<mbakke>rekado: do you remember what the "target-" inputs of GHC@8.4 are for? I don't see how they differ from the "normal" GCC, binutils and ld-wrapper inputs.
<sirgazil>I don't know if I should report the bugs I found in MATE 1.22 or wait for a MATE upgrade...
<mbakke>sirgazil: not to my knowledge
*sirgazil is reporting MATE bugs to his tracker, though.
<sirgazil>mbakke: Ok.
<mbakke>holy maloney, removing the extraneous ghc@8.4 inputs seems to have fixed the build on core-updates
<Kimapr>is this: present in guix repos?
<Kimapr>i can't find it
<mbakke>rekado: I suppose those inputs are there so that GHC refers to the correct programs when cross-compiled? But then they should not be used for building GHC itself.
<mbakke>Kimapr: does not seem like it
<Kimapr>i want to use two windows with two keyboards and mice independently
<brendyyn>Kimapr: It's X multiseat. I think kmscons was developed with this in mind.
<mbakke>Kimapr: I suspect you can do that with 'sway'; though packaging multicursor-wm should be a good first contribution ;-)
<brendyyn>That's for X11. Perhaps on wayland there are better ways? I know with wacom tablets they have multiple for mice and pens
<brendyyn>it would be awesome if someone explored this stuff and got it working with guix
<Kimapr>i use x11. i have 0 experience with wayland WMs
<brendyyn>well, multiseat is for multiple desktops
<brendyyn>if you want two on the same desktop its a different matter i guess
<Kimapr>i would do multicursor if multiseat is too difficult and vice-versa
<brendyyn>dunno whats best you will have to explore. linux was never designed for this stuff initall you see, so its a bit of a hack
<rekado>mbakke: I have no recollection of working on this, sorry.
<mbakke>rekado: no worries! Seems like it was just a thinko, I'll push my fix once the test suite had completed. :-)
<jonsger>sirgazil: I thought about updating mate to 1.24, but never started
<efraim>I generally keep on top of enlightenment, still haven't worked out Wayland support
<rekado>mbakke: thanks!
<Kimapr>okay so i just started sway and figured out how to run apps
<andydarcyjewell>Hey Guix, I'm back with more daft questions! I've got Factor to build with, but now I need it to download a boot image and then "bootstrap" it. The first involves downloading a file, so I'm wondering if that should go into a separate package, and be depended upon by this package?
<andydarcyjewell>The bootstrap step involves running the new factor binary with a flag to continue the build of the Factor modules.
<andydarcyjewell>So for that, I think I need to run this bootstrap command from within this package.
<andydarcyjewell>Am I thinking along the right lines, or is there a more Guix canonical way to do it?
<mbakke>andydarcyjewell: I don't really understand the question, but you can add the boot image (and any other "origin") as inputs and refer to them in the build process
<bricewge>Would it be a bad idea to submit a patch to support modifying services based on their name and not only their type?
<bricewge>To make something like this possible:
<andydarcyjewell>hey mbakke: the make process requires the "boot image" for the Factor system to be in the build dir, run make, and then invoke the bootstrap process, which pre-compiles the Factor ecosystem (i.e. factor modules)
<bricewge>It would allow modifying services defined with simple-service.
<andydarcyjewell>mbakke: so in the "(source (origin ... ))" call, I currently have just one "(uri ...)" call and a "(sha256 ...)" call; do I just add another pair like that for the boot image download?
<nagamalli>I have completed writing package description and 2 of my patches were
<nagamalli>merged successfully with help of Danny.
<nagamalli>I am looking for Next steps for my Outreachy Application Process
<mbakke>andydarcyjewell: You'll need to add the boot image as a native-input, and do something like (copy-recursively (assoc-ref inputs "boot-image") build-directory)
<mbakke>andydarcyjewell: see e.g. how GHC unpacks its external test suite
<andydarcyjewell>mbakke: ok, off to look at GHC
<andydarcyjewell>mbakke: thank you.
<mbakke>andydarcyjewell: have fun & good luck! :-)
<andydarcyjewell>mbakke: you just know I'm going to be back...
<mbakke>andydarcyjewell: GHC might not be the best example as it does not use the standard (modify-phases ...) syntax, but I'm sure you'll figure it out ;-) there is also 'git-manpages' which is an input to the git package.
<andydarcyjewell>mbakke: yeah, GHC does seem to be pretty *massive*
<guix-vits>nagamalli: Hi, you maybe need to check later for someone aware about Outreachy.
<g_bor[m]>nagamalli: hello!
<nagamalli>Hi,Hope u know about my status,What should be my next task?
<NieDzejkob>LPT: check out staging and core-updates branches with `git work-tree` so that you don't have to recompile everything each time your switch branches
<g_bor[m]>Yes, I know a little bit. Have you already recorded your contributions?
<apteryx>oops: sha256 hash mismatch for /gnu/store/v4lx2zi5zgzzfkicsvzpvc5jkz3g6pjl-rustc-1.34.1-src.tar.gz:/sha256 hash mismatch for /gnu/store/v4lx2zi5zgzzfkicsvzpvc5jkz3g6pjl-rustc-1.34.1-src.tar.gz:
<apteryx>on master
<apteryx>while attempting to build icecat
<apteryx>actual hash: 1951plm9h61xaf1k3y7m0f1vsjqvcznq282za2qfrbjwwj4v3q0c
<nagamalli>g_bor:No,Not yet , I Have done two patches.
<nagamalli>I'm not sure how to proceed next. Is there something I could do next. Currently I'm at home due to Corona virus lockdown. I could spend Time if some continuous guidance Is there.
<raghavgururajan>So, I added "python2" as input along with module (gnu packages python). But I get "error: python2: unbound variable".
<raghavgururajan>Oh, variable is python-2.7. Gotcha!
<civodul>apteryx: oh, modified in place?
<apteryx>looks like that :-(
<apteryx>I haven't diffed yet
<apteryx>this is on Guix 7ff639510096ff762b9cced5fba6db254a961af9
<munksgaard>I just did a fresh install of GuixSD and after adding the nonguix channel ( I've run `guix pull` and now `sudo guix reconfigure /etc/config.scm` which fails in a substition of nss-certs-3.5, something about a broken pipe. Does anyone know what could be wrong? I'm brand new when it comes to guix
<civodul>munksgaard: could you paste the complete output?
<munksgaard>civodul: Yes, just a sec
<munksgaard>I guess I can try the --fallback option?
<munksgaard>Also, the part about /root/.config/guix/current not being found confuses me a bit. I though guix was supposed to run through my account when I run it like that? (according to the note here: )
<roptat>I'm getting closer to the maven-build-system, the resources and compiler and jar plugins did their job successfully, the surefire plugin still has some troubles running tests, and the install plugin has problems provisioning a dependency
<roptat>but I passed the build phase successfully, I have built a *hello world* with maven :D
<guix-vits>munksgaard: i'm not a developer. Yes, supposed to be run as non-root. `guix pull` `sudo guix reconfigure /etc/config.scm` `guix package -u`
<g_bor[m]>nagamalli: I would do the administration, so record your contributions. Then you could write a timeline draft for the internship if you have not done so.
<munksgaard>guix-vits: The --fallback flag seemed to fix things. I still don't know what's going on, but now I'll get to actually reading the manual...
<civodul>munksgaard: could you use ( rejects Tor users)
<guix-vits>munksgaard: sometimes the substitutes on the fresh `pull` aren't ready yet; try `guix weather` to see what is ready.
<munksgaard>civodul: Of course:
<munksgaard>Didn't know about the tor blocking, sorry!
<bricewge>munksgaard: You got hit by
<bricewge>Your log say you should have installed glibc-locales
<munksgaard>bricewge: Ah, I see. Thank you! I thought I'd wait until I'd at least completed Chapter 3.7, but I guess it couldn't wait :)
<nagamalli>g_bor:Ok. will that be sufficient for my application process. Sorry i am new to this hence lot of questions
<civodul>bricewge was fast :-)
<munksgaard>He sure was :D
<munksgaard>I'm still at little baffled that I need to manually go and install glibc-locales. This is a completely fresh install on a new machine. But perhaps my addition of the nonguix channel was the culprit.
<bricewge>No, a lot of first comers have issue with locales. People are trying to fix it. So it's probably not the unholy channel.
<munksgaard>I see. Thank you bricewge :-)
<munksgaard>and guix-vits, civodul as well :-)
<g_bor[m]>nagamalli: Yes, it is a formal requirement. Any other contribution is welcome, but having these administration is important.
<g_bor[m]>I would suggest checking the easy bug list for documentation related bugs.
<g_bor[m]>I have to go now, see you later.
<webstrand>I'm interested in trying Guix, but due to a lack of media, I can't use the installer ISO. Is there something like Gentoo's stage3 tarball that I can use instead?
<NieDzejkob>you could install Guix-the-package-manager on your current distro and `guix system init` on another partition out of there
<jonsger>webstrand: or run it in a VM
<webstrand>I'll see if I can get it installed
<webstrand>*the package manager, that is
<jonsger>webstrand: which distro are you using?
<webstrand>currently, gentoo
<troydm>how do I check that ssh service is running/
<troydm>it seems that I have openssh-service-type in config.scm
<troydm>but I don't see any sshd running in system
<troydm>also I see guile: failed to install locale
<troydm>every time I run guix command
<mroh>webstrand: there is also a guix overlay for gentoo that might help to get a guix binary.
<roptat>troydm, the locale warning is just a warning
<roptat>troydm, you can see what services are running on your system with herd: "herd status" will give you the list of services
<roptat>ah! `Cannot run program "/bin/sh"`
<roptat>now I get why I can't run tests
<jonsger>troydm: this "failed to install locale" is just a warning. you can ignore it
<NieDzejkob>make sure your $GUIX_LOCPATH is set
<NieDzejkob>also, if the warning is prefixed by substitute:, you might want to check all that in the context of guix-daemon
<NieDzejkob>so look into how your init is starting it
<NieDzejkob>(you might need the locale package installed in root's profile)
<amartens[m]>> troydm: also I see guile: failed to install locale
<amartens[m]>I had a false locale specified in my config.scm ... I got rid of it by correcting it and rebuilding the system.
<amartens[m]>changing it from de_DE.utf8 to de_DE.UTF-8 fixed it for me
<lfam>troydm: There is a bug where sometimes the OpenSSH daemon fails to start:
<nckx>That's weird, utf8 is correct.
<amartens[m]>hmm okay, maybe it was something else then ^^
<lfam>troydm: It might be related to the IPv6 interface not coming up quickly enough for the sshd
<amartens[m]>but I don't have the warning anymore
<nckx>amartens[m]: I don't disbelieve you. Locale handling *is* weird and broken. I just mean it's a bug and not normal.
<amartens[m]>Of course, I understand :)
<apteryx>civodul: hmm, I could successfully build the rust@1.34 source upon a second try, after freeing some space. This makes me wonder if the ENOSPC issue was only partially fixed... Will keep an eye or further occurence, as this system has a really small drive (55 GiB).
<apteryx>err, 55 GB
<troydm>lfam: well I've changed locale to correct one in config.scm, reconfigured the system and now it started without issues
<lfam>Great! I'm not sure it was related to locales but it's good to hear that it's working
*jonsger updates mate to 1.24
<troydm>it seems that I can't get term-auto service to start using herd
<troydm>how can I get startup log?
<troydm>something like journalctl
<roptat>I don't think term-auto is supposed to run
<roptat>logs are in /var/log
<roptat>the shepherd doesn't capture outputs from services
<troydm>roptat: where are service descriptions stored?
<webstrand>in guix philosophy, are working packages ever removed? For example, in gentoo repositories package maintainers regularly clear out old versions of packages that they no longer want to maintain.
<lfam>webstrand: We don't remove packages unless they are broken or there is some other good reason. There are not any hard rules though
<alextee[m]>uhm, is it acceptable to delete/replace packages? i moved "zlfo" (just merged) to a "zplugins" repository that includes other software (for maintainability reasons), so would it be ok to send a patch to add "zplugins" that deletes zlfo? or should i keep the old repo too for legacy reasons?
<nckx>alextee[m]: You can deprecate packages. Then Guix will tell users using the old foo to install bar instead.
<alextee[m]>oh cool
<alextee[m]>and then after a while deprecated packages get deleted?
<alextee[m]>(asking because i really want to purge the old repo after some time :D)
<civodul>troydm: "guix system search ." will show you all the existing services and their source code location
<civodul>though term-auto doesn't appear as such there, hmm
<nckx>alextee[m]: Funny you asked, I'd just created a branch here to delete old (>1y) deprecated packages.
<nckx>Then forgot about it.
<alextee[m]>oh, sounds nice to automate the deletion
<nckx>webstrand: We also don't keep older versions of packages in Guix ‘just in case’. Only there's a good reason (some human or package requires the older version). I know some distributions do.
<nckx>alextee[m]: Eh, deleting a few lines once a year isn't really worth it IMO.
<webstrand>nckx: So I shouldn't rely on a valid system configuration to continue to be valid a couple years down the road?
<nckx>webstrand: Not if you want to use it with a new Guix. That's not how a rolling release works in general. Upstreams break/change things.
<troydm>not sure what this term-auto is but it's not starting
<webstrand>I wouldn't expect it to work with newer packages, the maintenance burden would be too high.
<troydm>and I can't find service description file related to it
<nckx>But unless you hard-code ‘foo-3.1’ somewhere or a package gets renamed, your system won't break outright.
<nckx>troydm: Why do you need term-auto?
<troydm>nckx: nah, just wondering why on a cleanly installed system, only term-auto service is not started and other services are running
<troydm>nckx: was wondering what that service does and what it's about
<lfam>I think the terminal services are "one-shot". That is, they don't keep running, but just execute and exit
<lfam>Are you having problems with those services?
<troydm>lfam: nah, just wanted to see the description of serive out of curiosity
<lfam>Check the code in 'gnu/services/base.scm'
<lfam>I notice they don't dislay in the "one-shot" category of `herd status`. Maybe we can fix this
<nckx>troydm: I don't think it's supposed to succeed on most systems.
<nckx>It's a fall-back or fail-safe or serial console thing IIRC.
<lfam>Right, it's a fallback of agetty
<lfam>Do we still use mingetty as the default?
<nckx>There is no ‘term-auto’ service (it's similar to a systemd term@auto instance). It just means ‘agetty without explicit terminal’.
<lfam>When I added agetty is was for somewhat esoteric use cases but I was not paying attention for a while
<nckx>lfam: Code said agetty but I merely skimmed it.
<drakonis>pinkieval: oh hello
<lfam>Look like mingetty is stilled used for the main consoles but agetty is part of the base servidces
<pinkieval>drakonis: hello
<mehlon>we're all here!
<nckx>Yeah: (or tty "auto") ... (or tty (default-serial-port))
<nckx>auto = serial console.
<lfam>So it's a way to automatically have a non-virtual terminal when that is possible
<lfam>Maybe we can fix the UI. It's not great to have stopped services by default
<nckx>Just don't break my Overdrives.
<lfam>Sounds like you just volunteered to test
<lfam>Or were you busy this week
<nckx>Sick & Busy.
<nckx>(My new band name.))))))))))))))))))
<nckx>Whoah there kernel.
<nckx>It's been doing that a lot this week, sorry.
<mehlon>ah no problem friend))))
<nckx>lfam: Double-check my reading, but if it is, adding ‘serial’ to the name would at least clarify things. I don't mind failed services for hardware obviously missing. I prefer it to fake success. Ideal would be for Shepherd to have an ‘irrelevant for you, never mind’ status.
<nckx>s/is/is correct/
<nckx>lfam: I'm available to test, sure 🙂
<mehlon>maybe we should have a bot that automatically does the regex substitution
<nckx>I'm sure there's a channel that has such a thing.
<nckx>We just have... sneek, the very special specialbot.
<nckx>where's sneek
<mehlon>it's almost as good as having it as an actual feature in the protocol!
<mehlon>sneek: later tell sneek: sneek you need to come back
<mehlon>sneek? sneek?
<lfam>Sneek was probably removed by freenode after yesterday's spam attack
<nckx>Oh no, what's this? Something just fell out of my pocket! Well, I'll just leave this BOTSNACK here for some lucky finder.
<nckx>lfam: Interesting hypothesis, it would be the first time.
<drakonis>there was a spam attack?
<lfam>They came through asking for us to let them know about any bots we wanted to save
<nckx>drakonis: Naah. 3 bots, then I locked down the channel & went sleepytimes.
<nckx>I /invited Sigyn but I don't think she was even needed after all.
<nckx>lfam: I asked about sneek before, they said it wasn't a problem.
<lfam>I didn't say anything last night because I didn't see how they could know that sneek was a bot
<lfam>I figured they might spammy bots
<lfam>They meant
<mehlon>wait, sneek was a bot?
<nckx>I don't think sneek is the kind of bot they mean (think ‘[O_O] COMMIT BOT’) but I asked just in case.
<drakonis>yes yes
<drakonis>i'm aware
<drakonis>nixos's commit bot uses notices tho
<mehlon>[OwO] COMMIT BOT
<drakonis>i've drifted away a bit in the past few weeks due to some things
<drakonis>so i haven't kept up with the latest stuff
<nckx>2020-03-23.log:[21:24:13] *** Quits: sneek ( (Ping timeout: 264 seconds)
<drakonis>is there any fun news?
<nckx>sneek just got tired of your shit brah.
<drakonis>ah there's a jvm importer proposal
<nckx>I don't really understand how ‘uses the JVM’ is a class of things that can be imported, but then I don't use it.
<mehlon>that's so maven!
<drakonis>they have a consistent build system i think
<nckx>Query: maven
<nckx>that shit that can't be bootstrapped
<nckx>end query.
<drakonis>gradle is advertised as a jvm build system
<nckx>That's all I know and sneek isn't here to help me.
<drakonis>ant, maven and gradle
<roptat>nckx, actually I'm pretty close to having a maven-build-system
<drakonis>its required for jvm based software builds it seems
<mehlon>well that gave me an idea for a Java-based OS called Jix
<drakonis>port guix to clojure
<mehlon>not to be confused with Jessyx which is NodeJS based
<roptat>i'm struggling a bit with surefire, but the rest is working: I can build and install packages (but not test them yet)
<nckx>roptat: From source?
<roptat>I had to write a small java parser to be able to build maven plugins
<roptat>it's probably not correct, but good enough that I can run the resources-, compiler-, jar-, enforcer- and installer-plugin
<roptat>yeah, took me a year or so, but I'm almost done
<roptat>I'll need to do some cleanup, and lots of reordering of builds in guix itself, so don't expect patches just yet
<roptat>my work is here:
<roptat>the java parser:
***nckx is now known as impressed_bot
***impressed_bot is now known as nckx
<drakonis>you wrote a whole parser just for that huh
<roptat>I completely reworked the dependencies to build maven as soon as possible (dropping some build dependencies and test dependencies, except junit), while installing everything in lib/m2, using the maven layout. I added a java parser and a plugin.xml generator to build the plugin.xml required to run maven plugins (this is usually done by the maven-plugin-plugin)
<drakonis>that's always cool
<roptat>it's far from complete
<roptat>it can only understand top-level definitions
<roptat>but yeah, it was necessary ^^"
<drakonis>a impressive guix cultural feature eh
<roptat>I'm pretty happy with what I did, but surefire is still resisting me
<roptat>that thing tries to run a separate thread, I already fixed it trying to run /bin/sh, but now the new thread seems to be having some classpath issues
<roptat>it can't find a class that's a requirement for surefire, so it must be there...
<roptat>oh, also I wrote something to parse the pom.xml and rewrite dependency and plugin versions, because otherwise maven would need so much more packages...
<roptat>(at least the jar is in the maven repository, so there's no reason why it can't find it)
<roptat>but all it tells me is NoClassDefFoundError
<webstrand>If I don't want to accidentally wipe my existing bootloader, is there a way I can disable the bootloader in the system-config?
<roptat>(bootloader #f) I think
<mehlon>well I think you can just delete the whole bootloader part from your config
<webstrand>mehlon: wouldn't that just use the defaults?
<webstrand>roptat: I'
<mehlon>but! that means it won't get updated, so you just have to do --no-bootloader every time you reconfigure if you want to keep that
<webstrand>I'll try that
<mehlon>and I'm not sure... do whatever roptat says I think
<webstrand>I'm using an encrypted root, so I can't actually use guixsd's kernel or bootloader
<roptat>why not? guix system's bootloader is able to decrypt an encrypted root
<roptat>that's what I use
<roptat>btw, it's Guix System, not GuixSD anymore ;)
<roptat>two things: I can't seem to make maven more verbose, it won't give me debug outputs (even with -X, it doesn't, and suggests running with -X ...), and I don't know how to make it keep the "generated manifest-only jar" for surefire...
<roptat>I thought it could be because I use slf4j-simple (maven doesn't like it), but even with maven-slf4j-provider, it doesn't change anything (except more colors I guess)
<roptat>also, if I manage to fix that, I wonder if I should use maven-3.6 for that bootstrap purpose, or simply maven-3.0 (since it's already a dependency for the plugins), then use the maven-build-system to properly (re)build maven
<webstrand>everything I've read online suggested it wasn't possible, since the kernels don't get copied to /boot.
<webstrand>Is there an existing guide? I don't recall any section in the manuals, but it's been a few months since I last checked
<roptat>it's possible, grub can decrypt your root fs, then load the kernel from it
<roptat>I think you only have to decleare that your rootfs is encrypted, the grub will be configured correctly
<roptat>let me find the manual page...
<nckx>webstrand: What Guix doesn't support is an *unencrypted* boot when your root is LUKS.
<webstrand>Oh, I remember something like that.
<nckx>Your bootloader record doesn't change. If you set up your file-systems and mapped-devices correctly GRUB will do the right thing.
<webstrand>I'll take a look, thanks!
<nckx>webstrand: I don't see it mentioned above so beware: you'll be entering you passphrase twice every boot if you use this method. Guix doesn't support any of the work-arounds required to avoid that.
<roptat>haha! I was able to make it wait for a debugger. During that time I could get the generated manifest-only jar, it seems that it's missing a lot of things
<webstrand>nckx: ah, because there's no way for grub to pass the password to the kernel. That makes sense. It'll also prevent me from unlocking remotely via ssh
<NieDzejkob>I should finally get to doing that >.<
<nckx>webstrand: Yep. Would require (most likely) embedding the key into the initramfs.
<webstrand>does Guix have the capability of generating an initrd of any kind?
<nckx>I didn't know there were distributions that came with an SSH initramfs hook.
<nckx>webstrand: Guix can't boot without one. It doesn't have a user-friendly customisation API though.
<webstrand>dracut it awesome in that regard
<webstrand>its so easy to add things into the generated initrd
<civodul>i think it's easy too, but it's certainly different
<nckx>I tried dracut long ago and it was horribly bad. Rolled my own dropbear/busybox system. That was very early dracut, I hope they've improved.
<webstrand>well, I was using genkernel up until recently, so my perception may be distorted
<nckx>webstrand: Oh, there's no sh in the initrd, beware.
<nckx>Guile all the things.
<roptat>mh... ok the pom for surefire-booter uses surefire-api *except* its maven-shared-utils dependency, which is what is missing inside the new forked process
<roptat>because of that, the generated jar doesn't use maven-shared-utils either, hence the failure
<sirgazil>More than 5 hours of work on a reconfigured system with slim and GNOME, and no memory leak.
<roptat>so it's only gdm?
<sirgazil>It seems.
*sirgazil goes to perform a prison workout
<kkebreau>Is gdm acting up?
<roptat>ok, my surefire package is wrong, the version on central "shades" some of its dependencies, including maven-shared-utils
<roptat>let's try using jarjar for that
<mbakke>roptat: incredible work on maven-build-system :-)