IRC channel logs


back to list of logs

<notroot>After reconfiguring my system from bare-bones to desktop I ran guix pull and guix system reconfigure as the didn't finish due to premission problems but obviously I screwed up and should have run it as root...where do I go from here?
<bavier>rerun as root
<bavier>the 'guix system reconfigure ...' that is
<notroot>what do I do with all the stuff produced from running it under the normal user?
<bavier>the same store items will be used when run as root
<notroot>okay, thanks
<nckx>rekado_: thanks for that link.
<nckx>SSI is so cool...
<nckx>ACTION fondly remembers the OpenMOSIX ‘cluster’ in their childhood bedroom.
<notroot>I can't seem to get to my tty's from gnome or xfce...ctrl + alt + Fanything just sends me back to the graphical login
<taohansen>what are Guixers doing for X11 remoting? x2go isn't available. xpra?
<buenouanq>notroot: fresh install on hardware (not vm)?
<buenouanq>taohansen: ssh -Y
<notroot>buenouanq, yes
<bavier>taohansen: I think we have vnc?
<notroot>buenouanq, fresh install
<notroot>buenouanq, on hardware
<taohansen>both would work, i'm just looking for something a little more performant
<buenouanq>notroot: I'm worried we have some big problems right now with services on the 0.14 image, this might be related.
<buenouanq>things vm users aren't encountering
<buenouanq>I don't have another machine here to test things on though, and I can't risk byorking this one.
<bavier>taohansen: haven't heard of x2go before, would you recommend it?
<taohansen>bavier: it's what is generally recommended as an open-source low-latency solution. this would actually be my first time putting it into action
<bavier>taohansen: we also have rdesktop
<bavier>taohansen: ok, x2go might be fun to package
<sloth_walk>hello, I am trying to install guixSD but am encountering an error related to the bootloader when running my `guix system init` command
<sloth_walk>The error is: grub-efi-2.02/sbin/grub-install: error: failed to get canonical path of `none'.
<sloth_walk>the relevant config is pretty much a copy of the manual:
<sloth_walk>If anyone knows how I should go about debugging this issue please let me know
<buenouanq>sloth_walk: post your config.scm (or at last the bootloader and file-system parts of it)
<sloth_walk>buenouanq: thank you for responding, I posted it in that pastebin above, but I just figured out my issue
<sloth_walk>so I no longer need assistance
<buenouanq>what was the issue then?
<sturm>Rrekado_: thanks for the tip re. json_object_private.h for postgis. I think I understand what you mean. I see that in in json-c's, the json_object_private.h is specified as not to be output. Looks like PostGIS will build without json-c, so I might put this aside and come back when I have the build working.
<sneek>Welcome back sturm, you have 1 message.
<sneek>sturm, rekado_ says: The name “json_object_private.h” indicates that this is not part of the public API, which is why the build system does not install it. You may need to add the sources of json-c to the package to be able to access it.
<sturm>Heh, I guess the bot doesn't know when I'm back as I'm on Matrix
<taohansen>hey anyone have upower set to ignore lid close events and care to share their config?
<sturm>I'm try to patch a file on the fly, but while the patch has been applied in the /tmp/guix-build- postgis-2.4.2.drv-7 directory, it has not affected the configure. Does the patching happen after configure is run?
<sturm> I'm trying to change one of the PostGIS install directories to /share, because by default it uses the pg_config program to get /gnu/store/0haa85i5rhpxmmninqpkyn3rqax83887-postgresql-10.1/share
<sturm>ACTION is posting the above to guix-devel
<catonano>sneek later tell rekado is working ? I run guix weather yesterdy and again today and the percentage of coverage has moved a bit: 57.4%
<sneek>Got it.
<catonano>neek later tell rekado I meant hasN'T moved a bit: 57.4% yesterday and 57.44% today
<catonano>sneek later tell rekado I meant hasN'T moved a bit: 57.4% yesterday and 57.44% today
<sneek>Will do.
<rekado>civodul: we have the offload bug on berlin now
<sneek>Welcome back rekado, you have 2 messages.
<sneek>rekado, catonano says: is working ? I run guix weather yesterdy and again today and the percentage of coverage has moved a bit: 57.4%
<sneek>rekado, catonano says: I meant hasN'T moved a bit: 57.4% yesterday and 57.44% today
<rekado>guix offload: error: corrupt input while restoring archive from #<input: channel (open) 272ade0>
<rekado>civodul: do I need to upgrade all build nodes to fix this?
<civodul>let's see
<civodul>should be fixed
<civodul>mv ~/.config/guix/latest{,.bak}
<buenouanq>attaching an external harddrive freezes everything
<buenouanq>detaching unfreezes
<buenouanq>any ideas?
<marusich>During boot, or while the system is running?
<buenouanq>while running
<marusich>Hm, then I'm not sure. Are you running GNOME?
<marusich>Does it happen in another desktop environment, like Xfce?
<buenouanq>feels like another symptom of the service problems
<buenouanq>I haven't tried that yet and said machine isn't here.
<marusich>If it doesn't, then it's probably GNOME-specific. I think GNOME uses some special daemons to watch for devices; I don't know much about how it's built.
<buenouanq>was it you that was checking out the dbus thing with/for me?
<marusich>I was looking into SSH not starting
<marusich>I wasn't able to reproduce it in a VM
<marusich>And recently, my desktop which used to have the problem, is now not failing to start sshd.
<buenouanq>I worry there are things going undetected becaues everything gets tested in vms...
<marusich>I'm happy it's easy to test in VMs.
<marusich>But yeah, it's a bummer when you're the only one suffering from a problem that seems hardware specific :(
<buenouanq>I actually just sent an email out asking if anyone has any old computers they're not using I could use to continually test this stuff myself.
<buenouanq>I can't at the moment risk byorking the machine I'm on which is at the moment unafflicted.
<buenouanq>And my macbook has specific guixsd problems of it"s own.. so that's no help.
<buenouanq>Like, have you ever heard of this:
<civodul>rekado: is berlin using the latest guix-maintenance.git?
<civodul>guix-daemon is still running with "--log-compression bzip2" for instance, instead of gzip
<buenouanq>nearly every other boot, the touchpad maps left/right to the cursor"s up/down and there is no moving it left/right, you can move it up and down in a line
<buenouanq>reboot fixes
<buenouanq>and it works as expected
<buenouanq>very bizarre, and I've no idea where to even begin trying to figure it out...
<civodul>rekado: also, it's guix-0.14.0-4.c04ffad instead of 0.14.0-6.0dcf675
<marusich>buenouanq, that's bizarre, and my guess would be that Apple's hardware is not playing well somehow with the drivers in linux-libre? but yeah, very bizarre.
<buenouanq>why would it work half the time then, how would a reboot predictably change/fix it?
<marusich>The only thing I could think to do is to check the output of dmesg (e.g. in /var/log/messages)
<buenouanq>and I never had this on any other distro, only guixsd
<marusich>I don't know. Hardware can fail in very bizarre ways, I do know that.
<marusich>I don't know enough about how mice specifically work with linux-libre to be able to provide a helpful response, I'm afraid
<rekado>civodul: yes, it’s using latest maintenance.
<rekado>ACTION has to go
<civodul>rekado: ok
<rekado>ah: mv ~/.config/guix/latest{,.bak} !
<rekado>sorry about that
<buenouanq>marusich: how about this:
<buenouanq>writing the website"s current verrified 0.14 iso image to a flash drive results in the following fdisk error:
<buenouanq>GPT PMBR size mismatch (2319631 != 3891199) will be corrected by w(rite).
<buenouanq>this flashdrive then freezes/crashes loaders like refind
<marusich>using dd, I guess?
<buenouanq>again, something no one testing with VMs would discover.
<marusich>Does this happen with other flash drives?
<buenouanq>yes and no
<buenouanq>yes and yes
<buenouanq>have tried multiple drives, same thing
<marusich>did you plug them into the same USB port?
<buenouanq>yes, as well as others
<marusich>Does this happen with different USB ports?
<marusich>does it NOT happen in any cases?
<buenouanq>the 0.13.0 iso image
<buenouanq>and all previous
<marusich>I don't think I've tried the 0.14.0 image, myself
<marusich>does this prevent the image from being used?
<buenouanq>yes, will not boot
<buenouanq>at least on my macbook using refind
<buenouanq>I have successfully installed it on normal computers, but then we have all these service problems...
<civodul>rekado: there are changes i made in berlin.scm but not berlin-new.scm, so i'm factorizing all this
<civodul>rekado: i'm reconfiguring berlin, but it has to build linux-libre
<ng0>how do you feel about inclusion of 2 scriptpacks ("variants" / "extensions") of epic5? it's not like epic5 is super active and can seem at times like it is abandoned development, but it still works. so do these scriptpacks. I've tested them in one of my repos.
<ng0>talking about amnesiac and darkstar
<buenouanq>the dark star orchestra
<buenouanq>west coast tour next month, hope to see you all there~
<ng0>DarkStar is also a nice ambient/black metal band. and a good sci-fi parody movie, one of the first
<marusich>buenouanq, I'm sorry I don't have more advice to give at this time. You are brave to install GuixSD on a Mac laptop. Maybe as more people do that, you can work with them to figure out what the Mac-specific problems are, and if they can be overcome.
<marusich>As for testing things like the installation image on real hardware, I'm pretty sure some people do it before the releases. You are, also, welcome to do that and provide feedback. It would be very helpful.
<ng0>Indeed, it needs people like you, trailblazing on new devices and running into tons of problems :)
<buenouanq>oh man ng0, the 1974 one? it's been on my list forever - maybe now i'll finally watch it
<buenouanq>I just need to be more serious about real bug reports and emails.
<buenouanq>doesn't matter how many problems I discover if I only ever mention them casually in here
<ng0>don't get the Bluray, I heard it's one of the terrible 'bluray all teh things!11' where it's worse than original VHS or DVD
<efraim>buenouanq: I'm catching up on chat logs, your comment about Apple touchpad from 2.5 hours ago happens to me too, I haven't figured out how to fix it
<buenouanq>efraim: oh good, always nice when i'm not alone in something
<buenouanq>efraim: if you ever figure it out or need me to try/confirm something for you, let me know
<buenouanq>no idea how many[few] of us have actually put guixsd on mac hardware
<rekado>civodul: we can replace berlin.scm with berlin-new.scm
<rekado>the old server that required berlin.scm is not coming back.
<pkill9>is there a webpage that shows the queue for the build server's builds?
<civodul>rekado: ok
<civodul>rekado: can we reboot berlin?
<civodul>if not i can fiddle with shepherd services
<efraim>so the aarch64 box is on berlin?
<civodul>well, if everything goes well
<civodul>rekado: i upgraded guix-daemon/publish/cuirass "by hand" (herd eval root ...)
<efraim>guix weather on berlin currently says '.0% substitutes available (0 out of 6604)'
<civodul>efraim: publish was turned off for a moment
<civodul>so you may want to try again :-)
<efraim>i was checking for aarch64-linux anyway
<rekado>civodul: yes, berlin can be rebooted, but it may take a while to boot because I think there are firmware updates pending for this server.
<rekado>gotta go to the data centre now, so I’ll be offline for a while.
<civodul>rekado: np, i went for the no-reboot option :-)
<pkill9>how do i mark a package such that the garbage collector doesn't delete it, without installing it to my user profile?
<Apteryx>Q: Is .profile sourced before .xinitrc?
<Apteryx>(not terribly on topic, I know)
<efraim>no irrelevant, I'm still wondering what I may be missing since my .profile doesn't seem to be loaded
<jlicht>hi guix!
<Apteryx>FWIW on GuixSD I use .bash_profile, which is loaded. It's a bash specific equivalent to .profile, IIUC.
<efraim>i was thinking of removing .profile's "source .bashrc" and adding to .bashrc "source .profile"
<efraim>not sure if it'd break anything
<jlicht>efraim: I'm not sure if ssh'ing into your machine will still work properly then
<efraim>and that's why I haven't done anything yet :)
<jlicht>with my settings, it probably would break in spectacular ways :-)
<ng0>on the rare chance that the people who wrote emails to me today and yesterday: I'm going to be 'out of office' wrt emails for the next 7 days. switching email clients + (primarily) taking time off to learn. chris marusich wanted to know if I worked on the namazu module since I posted it: I don't think so. If you can not find it in or or
<ng0> I've done no further work.
<ng0>*to me today are here
<ng0>I have some terrible old patches that I'm still working on though. so thanks for asking :)
<ng0>my GUIX_PACKAGE_PATH is getting so long now that I'm really looking forward to the implementation of the rfc for channels :)
<pkill9>what rfc is that ng0?
<iyzsong>hello weekend :-)
<ng0>see guix-devel
<civodul>heheh, howdy iyzsong! :-)
<snape>pkill9: ng0 was talking about this:
<pkill9>cool i thought so, i foudn that thread soon after
<pkill9>is that the same concept of channels that Nix uses?
<snape>I don't know
<quiliro>my system is unbootable
<quiliro>i made it bootable by inseting efi_gop and efi_uga in grub to make messages bootable in efi...then i could boot
<quiliro>i saved the proceedure in my personal log
<quiliro>but it is in the unbootable machine
<quiliro>'insmod efi_gop' and 'insmod efi_uga'
<quiliro>should be in the grub command line and then it boots
<quiliro>i guess that is it
<quiliro>then i get a guile prompt (bournish shell)
<quiliro>and i just type ,b
<quiliro>and then it completes the boot process
<quiliro>is thios comand good?: 'sudo -i guix pull && sudo -i guix system reconfigure /path/config.scm --fallback'
<str1ngs>quiliro: that's fine
<wigust>quiliro: I think you don't need "-i" flag in sudo
<snape>pkill9: you can use 'guix build --root'
<snape>or manually symlink the file to /var/guix/gcroots/ I guess
<snape>also, pkill9: the webpage for the build server's builds is
<quiliro>wigust: str1ngs recommended to use it for using root's guix
<quiliro>str1ngs: do you still think it is necessary?
<quiliro>about: sudo -i
<snape>I personally use sudo -E
<pkill9>nice thanks
<snape>but there is an Emacs command that makes it humanly readable
<wigust>quiliro: SUDO(8) points to SUDOERS(5) about environment variables setted by '-i', which I don't see useful be setted. Also it will read root's .profile and .login which are not useful, too. Could even not work for non-GuixSD systems, I guess.
<mubarak>Hello Guix!
<wigust>Hello mubarak
<mubarak>When installing guixsd-install-0.14 by running this command(guix system init /mnt/etc/config.scm /mnt) does it going to compile the kernel?
<mubarak>because their are alot of information about that laptop hardware like(hardware name, cpu flags cpu code) I remember last year I did gentoo install and i add some of these information to gentoo config file
<mubarak>the problem is that it keep repeating for more than two hours
<mubarak>Hello wigust :D
<mubarak>how you doing?
<civodul>mubarak: no it's not going to compile the kernel
<mubarak>30 minutes ago I tried to cancel it(CTRL+C) but it work again and I can't stop it or go to tty3
<mubarak>i think i should fore the laptop to shutdown and try installing amd64 iso. my laptop support amd64 but i want to use i686
<mubarak>Thank you civodul for the info :-)
<quiliro>ran 'sudo -i guix pull' and 'sudo guix pull'
<quiliro>neither worked
<quiliro>'sudo -E guix pull' is working
<quiliro>previously i ran:
<quiliro>'guix pull'
<quiliro>before 'sudo -E guix pull', do i need to run also:
<quiliro>guix package -u'
<wigust>quiliro: If you don't mean 'sudo guix package -u' then 'guix package -u' upgrades user's packages and has nothing with root's packages.
<catonano>ah, increased the coverage to 57.8
<quiliro>wigust: so i do not need 'guix package -u'?
<quiliro>'sudo -E guix pull && sudo guix system reconfigure /path/config.scm --fallback'
<quiliro>is that correct?
<rekado>catonano: yeah, there was a problem, which I had not detected. We’re still lacking monitoring for the build farm.
<catonano>rekado: yes I was reading the chat log right now. Thanks !
<quiliro>but i get a warning that the system is 14 days old unless i run:
<quiliro>'sudo -E guix system reconfigure /path/config.scm --fallback'
<quiliro>141 days old
<bavier>neat, aarch64 substitutes on
<wigust>quiliro: Because sudo's '-E' flag preserves invoked user's environment. So you change '/home/quiliro/.config/guix/latest' not '/root/.config/guix/latest'.
<wigust>quiliro: See 'sudo -E printenv'
<wigust>quiliro: and 'sudo printenv'.
<wigust>quiliro: 'sudo guix pull && sudo guix system reconfigure /path/config.scm --fallback' is enough.
<wigust>quiliro: You failed to 'guix pull' with just 'sudo', because you probably already did this with 'sudo -E', which did your '/home/quiliro/.config/guix/latest' belongs to root user.
<wigust>quiliro: Just "sudo unlink '/home/quiliro/.config/guix/latest'" and pull again.
<quiliro>wigust: ok, i did
<quiliro>but now i get error with 'sudo guix pull'
<quiliro>i am doing 'guix pull' now
<wigust>quiliro: Ah, I mean you failed 'guix pull' without 'sudo' because you probably already did …
<wigust>quiliro: What is the error?
<quiliro>wigust: i do not remember..i am running 'guix package -u' now
<quiliro>perhaps that will repair the error
<quiliro>'guix pull' worked well
<ng0>is someone working on a recursive Go importer? gitea's vendor directory makes me sad..
<ng0>I'm not doing manual work for 71 - 100+ bundled dependencies if it can be avoided
<quiliro>so... in the future i will always run: 'guix pull && guix package -u && sudo guix pull && sudo guix system reconfigure config.scm --fallback' toavoid any error when i want to update the whole system
<ng0>It's one of these packages where I thought Nix must be cheating, while all they do is just keep the vendor inclusions.
<bavier>ng0: Nix doesn't do as much to unbundle as we do
<ng0> party hard and bring your party.json hat
<ng0>yeah, I noticed
<civodul>ng0: didn't lfam have some initial Go importer?
<ng0>a Go importer exists
<ng0>i think
<lfam>I didn't make a Go importer yet, but Go itself handles dependency resolution, so it should not be too hard to instrument Go for that
<ng0>I could just throw the whole list at it, which is what I did with the still non-functional rust-branch I have
<ng0>is that too small for a GSoC, completing the rust build-system?
<ng0>or maybe too time intensive
<ng0>though I think having the previous GSoC work continued would be far better :)
<civodul>lfam: oh ok, thanks for clarifying
<wigust>quiliro: Glad it works for you. But to be clear, commands with sudo don't care about previous two. You could have any number of users, which have their version of guix (~/.config/guix/latest) and every need to 'guix pull' and 'guix package' independently.
<wigust>quiliro: So administrator with sudo privileges will not break users stuff.
<pkill9>TIL there's a web interface for the Hydra build queue
<pkill9>hmm I notice it uses the name 'Nix' instead of 'Guix', for example under the tab 'Details' here
<bavier>pkill9: yeah, the Hydra software is borrowed from the Nix project, and afaik the instance we run isn't modified much if at all
<quiliro>wigust: somehow i broke sudo guix pull
<quiliro>perhaps it had to do with sudo -i guix pull
<pkill9>If i use `guix build --with-source=vala-0.26.0.tar.xz vala`, how do i then create an environment with that created derivation? Running `guix environment --container vala@0.26.0` just returns 'guix environment: error: vala: package not found for version 0.26.0'
<quiliro>i am running guix package -u now
<quiliro>it is taking a long time
<quiliro>then i will run sudo guix pull
<quiliro>because i cannot run it now
<wigust>quiliro: I don't know, didn't use this flag before for this purpose. Only to make sudo interactive to get a root shell session. But as I read the sudo and sudoers man pages, it shouldn't break guix. I believe you could unlink '/root/.config/guix/latest' at any time if you break something.
<quiliro>i get an error...i think i posted it in help-guix
<bavier>pkill9: if you do the 'guix build' command with the -d option, you'll get the derivation file. I think you can then give that derivation path to 'guix environment'
<bavier>pkill9: do you mean to use '--ad-hoc'? otherwise the environment would be the same
<pkill9>the same as what?
<quiliro>that is the sudo guix pull error
<quiliro>but with -i
<pkill9>bavier: it's not taking the derivation path, hmm
<wigust>bavier: here was a discussion about passing derivation to 'guix environment'. As I remember it didn't work and not implemented currently.
<bavier>wigust: ah, ok
<wigust>quiliro: Did you try what Chris Marusich suggested?
<wigust>quiliro: Remove '/root/.cache/guile/ccache/'.
<quiliro>wigust: should i stop guix package -u, or do it on another tereminal
<quiliro>unlink '/root/.config/guix/latest' ?
<ng0>does someone run their own substitutes server in a private LAN and got some practical advice before I hack up too much repetition?
<wigust>quiliro: You don't need to stop, because 'guix package -u' and any 'guix' commands do nothing to the system, only to user under which you run it.
<quiliro>wigust: oh..i see... the mailing list
<ng0>it's mostly berlin as a base, but local.. and the local part would then just be http://IPhere/ for the substitute-server to address, right
<ng0>with all the berlin-specific fields adjusted (I've done this before)
<quiliro>sudo unlink '/root/.cache/guile/ccache/'
<pkill9>bavier: what does the --ad-hoc flag do? I read the manual but didn't understand it
<wigust>quiliro: Also to make it 'reconfigure' and 'package -u', you could pull and wait about a week for substitute server build things for you.
<bavier>pkill9: rather than placing the inputs of the named package in the environment, it puts the package itself
<pkill9>ah ok
<ng0>hm private https.. under some folder I had my personal CA. should revive that and integrate it into Guix.
<pkill9>bavier: if i use that then i don't hav ebasic commands like 'ls' and 'make', do I need to add 'coreutils' to the packages to include in the environment?
<bavier>pkill9: yes
<quiliro>wigust: good idea
***jorge is now known as Guest66265
<quiliro>wigust: so i do 'guix pull' now and in a week i can do 'guix package -u' and then immediately 'sudo guix pull' and 'sudo guix system reconfigure config.scm'?
<Guest66265>\\nick Jorge
<wigust>quiliro: You want to to do 'guix pull' 'sudo guix pull' at the same time and then wait
<wigust>quiliro: Or you could pull a commit by hash from one week ago, see guix pull docs.
<quiliro>sorry caps
<quiliro>pressed caps instead of tab
<ng0>or by branch if you feel lucky
<wigust>quiliro: pull from all your users like quiliro and root (sudo guix pull). Then wait a week.
<Guest66265>Hi. I am installing and setting up Guix. This morning I have invoked ~guix package -i glibc-utf-8-locales~ (note I misspelled the package name, for there is an extra dash) and it took hours (it seemed to build many packages, including many different versions of GCC), only to complain the package name had not been found. This afternoon I invoked ~guix package -i glibc-utf8-locales~ (spelled correctly) and it has been running for h
<bavier>Guest66265: it sounds like you don't have substitutes authorized
<bavier>in which case a lot of building is to be expected
<quiliro>wigust: I cannot sudo guix pull
<Guest66265>In fact, for more than three hours it is testing gettext- It is running a process named test-lock. And hits is a Core i7-6700 with four cores.
<quiliro>it gets an error
<bavier>Guest66265: oh, test-lock can hang
<quiliro>wigust: 'sudo unlink '/root/.cache/guile/ccache/' gives error
<bavier>Guest66265: are you installing from a git checkout?
<quiliro>wigust: unable to link, it is a directory
<wigust>quiliro: no unlink, do 'rm -rf'
<Guest66265>No, from the 0.14 tarball
<Guest66265>This morning I invoked ~guix archive --authorize < ~root/.guix-profile/share/guix/ ~. I can see it in Bash history. I seem to recall there was no error message, but my memory is weak.
<Guest66265>I invoked it as root
<quiliro>wigust: and then?
<bavier>Guest66265: running 'guix pull' might help
<Guest66265>The environment of the daemon currently lacks ftp_proxy, and I am at my workplace and proxy is required. But it does have http_proxy and https_proxy. Is it possible the substitutes are not being downloaded because FTP is blocked, even though HTTP is functional?
<Guest66265>What do I do with the current build?
<wigust>quiliro: then try to 'sudo guix pull'
<bavier>Guest66265: substitutes are downloaded over http, so that should be fine
<Guest66265>test-lock has just accumulated a full day of CPU time. I guess it is accumulating the time for all eight virtual cores (four cores with Hyperthreading). What should I do?
<bavier>Guest66265: kill it, it'll never finish
<quiliro>sudo rm -fr /root/.config/guile/ccache
<quiliro>gave no error
<quiliro>wigust: sudo guix pull did not work
<wigust>quiliro: What's the error?
<quiliro>(dispatch-exception _ _ _)
<ng0>okay, that was easy (setting up the private substitutes server)
<ng0>1.3% substitutes already. cool.
<Guest66265>I killed it. Also I restarted the daemon, so its environment should now have both http_proxy and ftp_proxy.
<Guest66265>What should I do now?
<wigust>quiliro: Copy and paste it here please
<Guest66265>bavier, should I now run guix pull and then invoke ~guix package install glibc-utf8-locales~ again?
<bavier>Guest66265: sure
<quiliro>wigust: sent to the help-guix list, but it is not in the archive yet
<ng0>efraim: how do you deal with building the everything in guix? something like guix package -A | cut -f1,2 | sed -e 's/\\t/@/' | shuf | parallel --jobs 1 guix build .... I'd add --keep-going, but it seems like the command itself has some issue, because it keeps getting stuck at random packages like.... parallel: Error: No more processes: cannot run a single job. Something is wrong at r-grohmm@1.12.0.
<ng0>building the everything... haha. I meant to write 'everything'.
<Guest66265>guix pull is compiling. Is that expected, even though this morning I invoked ~guix archive --authorize < ~root/.guix-profile/share/guix/ as root?
<bavier>Guest66265: yes, it may need to build some supporting packages
<ng0>or maybe someone else could help.. this line used to work, or I saved it wrong in the script
<bavier>ng0: our parallel package needs 'perl' in the environment to function properly
<ng0>ah! okay, this system has a very empty profile
<bavier>I've talked to Ole about that problem, but we haven't come up with a good solution yet
<ng0>I hear fans spinning up... I think it worked with adding perl :)
<ng0>usually I have perl, so I don't notice problems when it's missing
<bavier>ng0: good luck
<ng0>actually using cuirass would be better, but I have to read into it to make it build my world-view (custom packages + guix master)
<quiliro>wigust: here's the message
<ng0>how much disk-space does one round of master cost right now? I have a couple of TB here
<bavier>ng0: I'm guess a few tens of GB
<ng0>I guess I'll find out
<ng0>800 GB free on the disk where /g/s is located
<ng0>theoretically I could use this when not at home to offload and get substitutes from, but guix doesn't play very well with torify
<ng0>or the proxy value guix could set
<quiliro>would building a substitute server which (i would get online once a week) double as an offline repo for an unconnected network?
<ng0>once a week is a bit too little for building all of guix.. you could build a subset of it.
<quiliro>.........then one week every month
<ng0>I'll note the time when I started and tell you when it's finished on this very slow 1600 ADSL connection with spinning disks
<amz3>1600 ADSL?
<quiliro>1600 what
<quiliro>1 can work at 1 megaByte /s but guix works at 180 kbytes /s sownload speed
<ng0>amz3: hm?
<quiliro>bonan vesperon amikoj!
<amz3>I barely go above 100Mega bytes / s
<amz3>Kilo bytes i mean
<Guest66265>I think git pull had already compiled openssl, and now it is compiling it again.
<ng0>amz3: 1600 is just a fancy way to express 200kb/s upload (if you are lucky) and 1.6MB/s download (but actually around ~700kb/s), the "new" routers in the street suffer from the stone age old wiring in this house and router behind router behind router in the streets..
<Guest66265>Also, I cannot change my nickname here. I am invoking ~/nick Jorge~. Apparently it does nothing. There is no output. I am accessing via
<ng0>eduroam at university is the fastest I get so far.. upgrade at home would cost too much
<natsu>quiliro: The error log is almost the same as at top of mailing list thread.
<amz3>ng0: 700kb/s... my dream...
<natsu>oh, why did wigust rename himself to natsu
<amz3>ng0: old age wiring, broken community, etc... lead me to this situation... I subscribed to fiber 3 months ago, still no fiber...
<Guest66265>bavier: guix pull failed.
<Guest66265>guix pull: error: Git error: failed to connect to Connection timed out
<Guest66265>I then went to /tmp, created a test dir, and invoked ~git clone to test conectivity. It sarted, then I interrupted it. Is it possible that git inside guix is not using my proxy?
<bavier>amz3: some places have fixed wireless with good speeds
<ng0>amz3: they tried to scam us into fibre here (door selling contractors).. theoretically we have it in this city, in practice nope. city next to us is working on ~80% city coverage of gigabit access (at least advertising it)
<natsu>quiliro: Hm, if you have a working guix on your user, maybe just brute force way update it for root with 'sudo mv -v /root/.config/guix/latest /tmp/root-guix-latest-backup && sudo ln -v -s $(readlink /home/quiliro/.config/guix/latest) /root/.config/guix/latest'?
<ng0>speaking of them, no contract sellers showed up in over 2 months here, that's a new record of not slamming doors into peoples faces or handing out 'go now and never come back' notes.
<natsu>quiliro: Then you should be able to 'reconfigure' without 'system too old' error.
<Guest66265>lxo: Alexande Oliva? Eu participo do UltraGNU
<ng0>too offtopic, too much Java and too much coffee today, sorry.
<jorgemorais>Hi I was here as Guest????.
<jorgemorais>I invoked guix pull and it failed with network timeout during the very git.
<jorgemorais>Is it possible git inside guix is not using my corporate proxy?
<jorgemorais>Outside guix, I have just cloned from
<ng0>oh... oh... oh :) I was able to a torify guix weather request :)
<ng0>maybe I can manage to get offloading working aswell
<ng0>tor: 28.8 requ/second, not-tor: 5217.9 requ/seond :)
<bavier>jorgemorais: it's possible
<jorgemorais>How do I configure a proxy for git to use inside the guix build container? I have been searching the web and the manual and found no answers.
<ng0>it's tricky, it's what got me stuck on tor offloading about a year ago
<bavier>jorgemorais: it might not be too hard (I don't personally have experience), since the git that's running is not within a container iirc
<bavier>jorgemorais: guix uses the guile-git module to clone, there might be some documentation there about proxies
<bavier>it might even honor the normal git proxy environment variables
<ng0>(my issue is different though.. 3.3.4 suggest that I should look into the code and fix up what I want)
<atw>the time is wrong on my installation. I have tried running the ntp service with #:allow-large-adjustment? #t and I have a long-running ntpd -nqx (should exit after setting time, has not exited and it's been a few days). I also tried something with hwclock...I forget what. What else can I try?
<sneek>atw, you have 1 message.
<sneek>atw, Apteryx says: for the record, I ended up replacing LightDM by SLiM on my Ubuntu install, and got what I want under 10 minutes... :)
<atw>nice Apteryx
<ng0>atw: as long as it still works with servers out there: tlsdate
<ng0>herd stop openntpd or ntp .. then run guix environment --ad-hoc tlsdate -- sudo tlsdate -l -V ... and check what the command does before you run it, in case I typed it wrong from memory .. and run hwclock aswell
<ng0>by default it contacts, but you can use any other server
<jorgemorais>I have failed so far. I will ask for help in the Guix mailing list.
<atw>ng0: thanks! looks like I want hwclock --systohc ?
<ng0>I'm considering picking up the work on the tlsdated service again, but with tls3 approaching and tlsdate appearing like it is (currently) not maintained, I'm not sure if it's worth the time.