IRC channel logs


back to list of logs

<rain1>did they give the donation as actual money or in handshake-bitcoins ?
***jonsger1 is now known as jonsger
<dustyweb>congrats on that huge donation
<apteryx>hmm... #:imported-modules vs #:modules?
<apteryx>amazing news about the new funds! :-) congrats!
***rekado_ is now known as rekado
<rekado>I think I fixed the problem with mozjs / gjs.
<rekado>this would probably also fix the gnome-shell segfault.
<wigust->Nice, thank you rekado!
<civodul>Hello Guix!
*vagrantc waves
*rekado reconfigures for another GNOME test
<civodul>rekado: BTW let me know when you plan to reboot berlin
<rekado>in a couple of hours, probably
<rekado>I’m a little worried about the store items referred to in the kernel arguments
<rekado>e.g. the boot script
<rekado>I suppose they also need to be available on the root disk, in addition to the kernel and the initrd.
<civodul>rekado: i don't think so, because the initrd will mount the real root fs before it tries to access those
<janneke>Hello Guix!
<civodul>hey janneke!
*civodul .oO( how much time in my life went into watching build logs? )
<civodul>that's a bit scary when you think about it
<brendyyn>is it mesmerising?
<civodul>kind of!
<civodul>with some experience you can recognize the build system and maybe the package being built
<civodul>just by staring at the log for a few seconds
<brendyyn>civodul: So how do you feel now, after thinking Guix would be a fun project you'd spend 2-4 years on?
<brendyyn>(I think you said something like that)
<rk4>is it ask civodul question time :D
<brendyyn>I've actually being curious about polling developer morale
<rk4>is it a free time thing? or paid? i just noticed your Code Staging in GNU Guix paper has Inria on it
<rekado>hmm, gnome-shell no longer crashes, but I’m still unable to start up the desktop with “gnome-session”
<rekado>but I managed to start the new gnome-shell directly
<rekado>(the mouse pointer moves much faster now)
<civodul>brendyyn: heh, i feel good
<civodul>it's true that the thing has taken me more time than i thought ;-)
<lprndn>rekado: If at some point, you need someone for testing, I would gladly help.
<civodul>rk4: currently part of my work at Inria has to do with Guix in HPC:
<rekado>I can only start the gnome-shell in wayland mode
<rekado>but hey, it no longer crashes!
<civodul>it's a clear sign that Wayland is the future
<rekado>looks like we need to add broadway and run broadwayd before starting gnome-session.
<rekado>it tries to connect to ~/.cache/broadway1.socket
<rekado>ah, it’s part of gtk
<lprndn>I need to set an env variable to the system fonts. How would you do that in guix (in a package definition)?
<rekado>so, I can now run gnome-session in X11 mode after setting GDK_BACKEND=x11 and setting DISPLAY=:0.0
<g_bor>hello guix!
<g_bor>rekado: is this on 3.30 already?
<rekado>I can no longer change the brightness with the new gnome
<rekado>no! This is 3.28 after realising that I needed to build mozjs with “--disable-jemalloc”
<rekado>I’m half done with 3.30, but thought it would be good to check if that mozjs configure flag really was the problem.
<rekado>I’ll push to wip-gnome-upgrades again tonight
<rekado>it entails fewer rebuilds than the upgrade to 3.30, so maybe we could merge it sooner.
<g_bor>rekado: understood
<civodul>rekado: great that you got this far
<rekado>I think we’ll need to make a few minor changes to the gnome service.
<rekado>I wonder how this wayland stuff is supposed to work; you can’t have the xorg-server running while using it, I think.
<lsl88>Hi Guix! First outreachy day :-D
<civodul>heya lsl88! great!
<g_bor>lsl88: Hello!
<g_bor>Good to see you!
<g_bor>I'm trying to get the Hungarian translation into shape. I have about 40% untranslated.
<civodul>g_bor: of the manual?
<civodul>i'd like to release tomorrow, it'd be great if you could upload to the TP what you have by then
<g_bor>actually the manual is still not in good shape.
<g_bor>I will try to send what I have, translation of guix is lagging here, we have only about 5-6 person doing translations all together :)
<g_bor>I assume I can do a quite full translation, maybe except the packages by late spring.
<g_bor>Also there are some communications problems recently, I was trying to reach the Hungarian community leader several times in the last month, but they are not very responsive.
<g_bor>civodul: I belive that I cannot make this translation for this release, so please go ahead.
<g_bor>what is nix/scripts/guix-authenticate?
<g_bor>I have a bit of a problem. I can't make on staging.
<g_bor>I got no variable bound to glibc-for-target in module (gnu packages base).
<g_bor>Any idea?
<g_bor>this is in gnu/system/locale.scm:124
<iyzsong>g_bor: i got it too. run 'make clean-go; make' to rebuild all files will fix it.
<g_bor>ok, this looks good.
<g_bor>Sometime I just forget this :)
<kmicu>“$100,000 for GNU Guix and GuixSD” nice.
*kmicu wonders whether Guix/FSF converted Bitcoins to real money
<civodul>g_bor: yeah, like iyzsong wrote; it's due to the fact that glibc-for-target was removed in core-updates, which we just merged
<civodul>kmicu: i think they did :-) see also
<kmicu>Hah, great! “The least we could do to show our appreciation was to add a package definition for the Handshake resolver daemon, which we just did.”
<roptat>hi guix!
<rekado>is there a simple way to serialize an <operating-system> value and write it to a file?
<rekado>here’s why I want that:
<rekado>I generate an operating system on a server, build it, and then push it to dozens of nodes.
<rekado>I then reconfigure the nodes remotely.
<g_bor>rekado: we were thinking along similar lines with roptat a few day before.
<civodul>rekado: not possible!
<rekado>previously I tried to use guile-ssh for the last step.
<civodul>i've been meaning to implement that for ages
<rekado>now I wonder if maybe I could use an inferior remote guix.
<rekado>I guess I could reconfigure “manually” by working directly on the tranferred operating system store item / derivation.
<g_bor>My current workflow is just distribute the os config as a scheme file, and build locally on the nodes...
<civodul>rekado: you could do: GUIX_DAEMON_SOCKET=ssh://host guix system build ...
<civodul>but that's not quite sufficient since you need to perform the effectful step once you've built the thing
<civodul>i guess for now you pretty much have to do it manually, though that's scriptable
<rekado>yes. I’ve already transferred the whole system to the remote.
<civodul>like you could send the config.scm to each machine, run guix pull --commit=XYZ on each machine
<civodul>and then guix system reconfigure
<civodul>that way you'll have the same thing everywhere
<rekado>well, I’ve got slightly different configs; different machine names and network settings, etc.
<civodul>yeah well, you could do (gethostname) and things like that :-)
<rekado>I’m currently building guix and the generated system, and then send it to the target with open-ssh-session / send-files
<civodul>or 'guix copy'?
<civodul>(that's what it does anyway)
<rekado>yeah, it’s a Guile script, so I tried to just use the Scheme API directly
<civodul>even nicer
<_tibbe>Hi, what is needed in the operating-system declaration to let fuse use extensions like fuse-exfat?
<lsl88>Hi again! I have two questions. 1) is there a way of scrolling up a tty? not a console. Sometimes I do | less but it is not good at all. 2) I know you don't like vBox, I was installing GuixSD over qemu. I faced strange issues, like having /dev/sda already partitioned, with Microsoft stuff. This did not happen at all with my retroPC and over vBox. I haven't used qemu in the past. And I am using the same .iso. I have searched over the inter
<lsl88>et some issues, but it is strange.
<g_bor>lsl88: how do you start qemu?
<g_bor>for scrolling usually shift+pgup/pgdown works
<g_bor>I have seen something interesting, I got a warning load-linux-module/fd possibly unbound in gnu/build/linux-modules.scm:261
<g_bor>is that ok?
<lsl88>g_bor: qemu-system-x86_64 -m 1024 -smp 1 \ -net user -net nic,model=virtio -boot menu=on \ -drive file=guixsd-install- \ -drive file=guixsd.img
<g_bor>lsl88: and what is guixsd.img? How is it created? Is that your /dev/sda?
<lsl88>g_bor: and I create the qemu-img create -f qcow2 guixsd.img 50G first. I deleted the partitions, created them with cfdisk
<lsl88>g_bor: sorry, I was writing that part. And the .iso is the one I used in the past: guixsd-install-0.15.0.x86_64-linux.iso
<lsl88>g_bor: also, I will better paste some weird things I get
<civodul>_tibbe: i believe you just need to install "fuse-exfat"
<lsl88>g_bor: I am delaying my paste because I believe I messed sth up with guix pull after guix package -i qemu complained about not guix pulling. I was checking my last guix generation and its date is not match when I guix pulled.
<g_bor>I've got to go now, see you later!
<_tibbe>civodul: I tried that but it did not work. I will try again and verify that it does not work
<_tibbe>no that definitly does not work for me: Neither with mount, gio mount or Thunars mount
<civodul>_tibbe: what i do is "sudo mount.exfat ...", and that works
<civodul>however i don't know what it would take to make it work with GVFS & co.
<civodul>perhaps we should somehow make exfat known to udisks or whatever is taking care of that?
<_tibbe>civodul: You're right. That works. That is kind of difficult: Fuse itself is in no way configurable. There is no kind of service or firmware from which we can detect which extensions are installed and supported
<_tibbe>It just gets installed with the kernel... There is nothing to track what extensions the system has to offer
<g_bor>hello guix!
<rekado>I’m not getting substitutes for openssh-7.9p1.tar.gz (nor for the binary)
<rekado>but this appears to be available on berlin.
<rekado>same for pyaml-17.7.2.tar.gz
*rekado needs to leave now
<civodul>oh it seems that Cuirass was somehow stuck trying to fetch the core-updates branch which no longer exists
<civodul>rekado: the /var/cache bind-mount was missing, hence these errors
<civodul>should be fixed now
<civodul>this should really be in the config
<civodul>(nginx still has 404s and 502s in cache, that should vanish within a few minutes...)
<OrangeShark>madage: sorry about that. I removed it so config.scm was not distributed in the tarball. I should have just specified in the to not distribute it
<lsl88>hi again! I am still trying to install GuixSD with qemu. I am facing different issues: wget works fine. It is the image I already had. But if I go to: (it exactly the same in both sites) I first run wget
<lsl88>inux.tar.xz and get "No such directory ‘guix'". So I just added gnu in between (wget and it is found but the last message is: ==> RETR guixsd-vm-image-0.15.0.x86_64-linux.tar.xz ... No such file ‘guixsd-vm-image-0.15.0.x86_64-linux.tar.xz’. Any ideas?
<cbaines>lsl88, I just tried the last command you mentioned, and it worked for me
<cbaines>(so I ended up downloading guixsd-install-0.15.0.x86_64-linux.iso.xz)
<cbaines>are you still having issues?
<g_bor>hello guix!
<g_bor>lsl88: it seems to me that you are confusing the vm installation method and the installation from usb method.
<g_bor>in the installing into the vm it is recommended to download a vm image, and just fire up a vm from that.
<g_bor>You can install from the .iso into a vm, but that way you have to do it exatly as you would do it on bare metal
<g_bor>does that make sense?
<lsl88>g_bor: I see, but if I want to install it with guix system, maybe I am not understanding it well, but how can I download the prebuilt image from here?’ (the one that I could not download)?
<lsl88>g_bor: or should I use guix system vm and the .scm file is one of the that is described? I don't understand very well the difference between vm and vm-image and which one I should use. I believe it worked on vBox because I set it, as you said, like a live cd/DVD
<cbaines>lsl88, I think one of the key differences between vm and vm-image, is that the vm command will share the host store with the vm, while the vm-image command will copy store items in to the image it's creating. Does that make sense?
<g_bor>lsl88: the image you could not download is a fully baked one, you just give that one to qemu as the only disk, and you are good to go.
<g_bor>you can create one from a working guix using guix system vm-image
<g_bor>a guix system vm is not self-contained, it shares the store with the host, and the store is read only from the guest.
<g_bor>(i.e. it does not have a guix daemon)
<davidl>if I set the mount point in the bare-bones.scm file to /mnt instead of /, Im getting a wrong type argument error, which is probably not what I should get.
<lfam>What is returned by `guix system vm` is a short shell script, which launches a QEMU instance using files found in the host machine's /gnu/store. These machines are immutable — the file system is read-only
<lfam>I recommend reading the script to understand what's going on
<lfam>Whereas vm-image returns a QEMU image file that must be copied out of /gnu/store in order to use.
<lfam>That's more like a full system, as g_bor said
<lfam>Does anyone have any idea about the problems with libboost_system since the core-updates merge?
<lfam> <> and <>, at least
<lfam>My first reading suggests it may be that Boost needs to be built with C++14
<lfam>But, that would be a really annoying solution, and I'm not sure it's right
<lsl88> I see, so if I cannot download that ftp guix-sd-vm-image, then I could use guix system vm-image with one of the .scm files described here: ie desktop.scm?
<lsl88>(and change it with what I need)
<lfam>lsl88: Exactly
<lsl88>cbaines: sorry, I did not see your last message. What did you run exactly?
<lsl88>lfam: thanks I will try it :) Hope this time works
<cbaines>jsut the wget command: wget
<lfam>lsl88: The file used to create the VM image that we offer for download is at 'gnu/system/examples/vm-image.tmpl'. However, I don't recommend using that unless you are going to use it on a hosted VPS service
<lfam>Just use something that fits your needs more directly
<g_bor>hello guix!
<lsl88>cbaines: yes, that compressed iso file worked, and I used it both for bare metal installation and the instalation in vBox. The problem was using it with qemu. And the ne that did not work was the other one
<lfam>Hi g_bor, welcome back
<lfam>Hm, QEMU is the virtualization solution that we "officially" support. It should work if you follow the instructions
<lsl88>lfam: yes, I know, that is why i wanted to have it with qemu. But really strange things happen when I follow this: Like having /dev/sda already partitioned with Microsoft stuff, and it gets really weird, I try to use cfdisk to delete the partitions and it starts complaining about floppy disks, the distro is populated with a lot
<lsl88>f things, that I did not see over bare metal or vBox, and so on. And I don't know how to copy and paste it from quemu to show it in paste debian
<davidl>lsl88: I just figured out how to do that yesterday: you can add -virtfs local,path=/home/<user>/vm_share_dir,mount_tag=host0,security_model=none,id=host0 to the qemu command line options. Then from vmguest, mount it with mount -t 9p -o trans=virtio,version=9p2000.L host0 /mnt. You then have shared directory that you can write files to from VM at /mnt.
<g_bor>lsl88: can you help me trying to reproduce what you have?
<davidl>lsl88: from there on you can open the /mnt/mytext.txt from host and copy paste to paste.debian
<g_bor>I guess it goes like this: qemu-img create hda.img, wget the guix iso installer.
<lfam>Does anyone have a build log for the pre-core-updates Boost 1.66.0?
<rekado>civodul: sorry about the bind mount. I thought I had added it to the config already.
<rekado>I’ll do this now.
<g_bor>lsl88: now I know what is wrong.
<g_bor>As sda you are seeing the installer image.
<g_bor>you should partition sdb or set up the devices in such a way, so that sda is really the hard disk, and the usb is somewhere else.
<civodul>lfam: found one!
<lfam>Awesome, thanks civodul
<civodul>rekado: np! i was a bit worried initially but then i found quickly enough
<lfam>There are a few issues with applications being unable to link with libboost_system since the core-updates merge which are fixed by building Boost with C++14
<lfam>As a fix it seems rather heavy-handed though
<civodul>it might lead to a new set of problems
<apteryx>hello! I'm finishing to package Openboardview, and struggling with a small detail; I want to use the cmake-build-system, but include also the two extra phases provided by the glib-or-gtk-build-system. I've found an example in the cups package, but I can't seem to get it working with my package: Any ideas welcome!
<lfam>apteryx: Can you add a paste of the error messages you get?
<apteryx>oh, I added that important bit at the end of the previous paste:; sorry for lefting it out.
<civodul>rekado: i think it makes little sense to include the full list of new packages and upgrades in NEWS now, WDYT?
*vagrantc can't wait to see what #guix looks like from a european timezone
<civodul>heheh :-)
<apteryx>lfam: I used the idea of the "#:imported-modules" bit from the hp-minimal package in cups.scm
<g_bor>I am not sure what is going on with qemu either, I can reproduce a problem with cfdisk, complaining about can't acces floppy then crashing.
<g_bor>I can use parted without problems
<rekado>civodul: well, there are a lot. Since “guix pull” gives users the same information I’m not convinced we should include that very large list for each release announcement.
<g_bor>oops my bad, it is not crashign, just the screen gets cluttered with output
<civodul>rekado: right, i'll change update-NEWS to not do that
<g_bor>the normal way to avoid the flopp problem seems to be using nodefaults
<g_bor>see you tomorrow!