IRC channel logs

2017-06-07.log

back to list of logs

***andreh1 is now known as andreh
<reepca>Hm, I'm getting a "unknown # object: #\\~" error on running guix pull
<reepca>... but apparently not when I run it from an emacs shell instead of from M-x guix-pull. Weird.
<eacces>hi all :)
<eacces>first week into this guix adventure, going pretty smoothly thus far. docs have been commendably good
<sturm>Is there anything like `guix build --target=mips64el-linux-gnu [package]` that allows you to build the entire OS? I'm picturing being able `guix system disk-image --target=...` to build an image for, say, a BeagleBone ARM board.
<sturm>(but `guix system` doesn't have a `--target` option)
<efraim>sturm: you could try --system=armhf-linux
<efraim>I think that system type is rightml
<efraim>s/rightml/right/
<sturm>efraim: wow, very cool. I'll give it a shot
<sturm>efraim: Guix downloaded a whole lot of sources, then said `guix system: error: build failed: a 'armhf-linux' is required to build ...-guile-bootstrap-2.0.drv, but I am 'x86_64-linux`
<efraim>sturm: some packages can't be cross compiled, you might need to wait until later in the day when more people are here and able to also help, i'm not too sure how to make it work otherwise
<sturm>efraim: thanks :)
<efraim>i think I have to tag ocaml@4.01 as not supporting aarch64
<efraim>might try `autoreconf -vfi' first
<reepca>upgrading vlc after a recent "guix pull" fails due to the package trying to download http://www.live555.com/liveMedia/public/live.2017.05.24.tar.gz when the site is only offering live.2017.06.04.tar.gz under that particular directory. I suspect it has less to do with the actual guix pull and more to do with the site maintainers deleting or moving that release. Anyone else have troubles trying to install livemedia-utils or anything that
<reepca>depends on it?
<snape>reepca: my fault
<snape>it's weird. Does that mean the link wasn't permanant?
<snape>well now it's obvious
<snape>I wonder where is the permanant link
<snape>Archlinux uses this one too
<snape>And the official website points to it
<snape>As a quickfix, I'll update the package version, but we definitely need to find a solution
<snape>(I think Hydra should have cached the link, it's pretty much unlucky it didn't have time for it)
<roelj>I get this error: "ERROR: no code for module (guix build utils)", when installing a package. (I can guix build <package> it just fine, but guix package -i <package> crashes with the error). What can I do about this?
<sneek>roelj, you have 1 message.
<sneek>roelj, dTal says: thanks for the shortcut to restart GNOME
<roelj>Can I give sneek a botsnack?
<rekado_>sneek: do you want a botsnack?
<sneek>:)
<rekado_>roelj: I think so!
<roelj>How do I do that?
<roelj>sneek: botsnack
<sneek>:)
<rekado_>roelj: I don’t know what to do about your (guix build utils) problem. Is GUIX_PACKAGE_PATH set?
<roelj>Yes..
<roelj>(guix build utils) still exists, so that should be fine..
<efraim>Roelj I had to 'make clean-go' to fix that
<alezost>reepca: "M-x guix-pull" has just finished without any errors for me. Did you get any backtrace in the REPL?
<roelj>efraim: Thanks! That seems to work
***jonsger1 is now known as jonsger
<jsierles>hey all! how can I get the guix daemon to listen on a port for remote rpc requests?
<roelj>jsierles: By running something like socat TCP4-LISTEN:<port> UNIX:/var/guix/daemon-socket/socket
<jsierles>roelj: ok. just thought there was something built-in given the docs suggest you can do it.
<jsierles>i'll try that.
<roelj>jsierles: Sorry, not that I know of. But maybe it can do something over SSH that I don't know about..
<jsierles>roelj: yeah. maybe ssh is the way to go anyway.
<roelj>jsierles: If you get that working, I'd love to hear about it
<jlicht>hey guix
<jlicht>catonano: Good to know you got something to work with freexl :-)!
<catonano>jlicht: thanks for your concern. It's appreciated. I actually adopted your suggestion, that's all. I couldn't properly read the thread that you indicated to me. I cold read it with new eyes only after giving up
<jlicht>catonano: sometimes you need to take a step back to properly solve a problem ;-)
<jsierles>is there a record of previous profiles used before running 'guix package -i'?
<jsierles>better question - if i use a manifest file with 'guix package', will i always get the same result, even if guix has been updated with 'guix pull'?
<rekado_>jsierles: “guix package -l”
<jsierles>ah, nice one.
<jsierles>where is that information actually stored?
<rekado_>it’s links in /var/guix/profiles
<jsierles>i see
<jsierles>and the default profile will always be used when running commands as root?
<rekado_>the root user’s profile will be used
<jsierles>when i ran 'guix package' just now, i saw: The following environment variable definitions may be needed:
<jsierles> export PATH="/var/guix/profiles/default/guix-profile/bin${PATH:+:}$PATH"
<jsierles>and it looks like in my current profile as root, i can't see what i just installed. i had to run the above command to see it
<rekado_>does the root user have a ~/.guix-profile link?
<jsierles>yup
<jsierles>.guix-profile -> /var/guix/profiles/per-user/root/guix-profile
<jsierles>i setup that link manually.
<jsierles>and set my path to: /root/.guix-profile/bin:$PATH
<jsierles>to get the initial guix environment.
<jsierles>since this remote mount seems to work, my intention now is to just use near-empty docker containers and manifests. not build images. seems this will work and be faster
<jsierles>rekado_: does the guix command line need to modify /var/guix?
<jsierles>wondering if i should mount /var/guix shared as well, or keep that local to the client container.
<jsierles>ok, i can confirm the client does need to be able to symlink there. so maybe sharing this volume is not the way to go
<epronk>Hello guix!
<epronk>janneke: Another attempt to run /gnu/store/abcd123-system/boot in lxd container. http://dpaste.com/0MR8ZAV
<jsierles>i see 'guix package -i julia' guix is building julia. why wouldn't use a binary cache in this case?
<efraim>If Julia hasn't been rebuilt since a package in its dependency graph changed then there isn't one yet
<janneke>hi epronk!
<epronk>Hi janneke: I'll post what I've learned so far to guix-devel.
<eacces>hmm, new to this, so might be a silly question: `$ guix build --rounds=2 --no-substitutes mercurial` returned
<eacces>`@ build-failed /gnu/store/79z9l63d4i7q5izk5wbv17y1cqrmf980-module-import-compiled.drv - 1 output ‘/gnu/store/8jy84rrarmwww8rdkiykhc6rppmbimsc-module-import-compiled’ of ‘/gnu/store/79z9l63d4i7q5izk5wbv17y1cqrmf980-module-import-compiled.drv’ differs from previous round
<janneke>epronk: great!
<eacces>is that unexpected? or expected? i'm not sure if reproducable builds are the norm yet
<davexunit>eacces: sounds like mercurial is not yet reproducible.
<davexunit>maybe you will be the one to fix that :)
<eacces>i'm keen to give it a go, though, isn't that a dependency long before we even get to mercurial?
<eacces>hold on, let me try out guix graph for the first time :D
<davexunit>ah, I guess so. that's a guile issue.
<rekado_>eacces: you get this when you’ve changed the code.
<rekado_>it’s a bit weird. I would build mercurial first and then run “--check”
<eacces>hmm, ok, thanks for the advice, will do that.
<sturm>Is it possible to cross compile a disk image? I'm thinking of trying to get GuixSD running on a BeagleBone Black (ARM). Tried `guix system disk-image --system=armhf-linux` but after downloading a whole lot of sources, it said `guix system: error: build failed: a 'armhf-linux' is required to build ...-guile-bootstrap-2.0.drv, but I am 'x86_64-linux'`
<rekado_>sturm: I don’t think GuixSD is available for armhf yet.
<sturm>rekado_: thanks, that would explain it then :)
<davexunit>it will be a glorious day when it is available...
<davexunit>I would love to have guixsd on my novena
<quiliro>hello!
<quiliro>i will install GuixSD on my laptop....It only has one usb port and i am using it for the network card...It does not have ethernet
<quiliro>It is using Debian
<quiliro>I resized the Debian partition but it would not be recognized after resizing. I had to enlarge it back
<quiliro>so i used the swap partition
<quiliro>i plan to eliminate the Debian partition after installing GuixSD on the old swap
<quiliro>or maybe copy the contents of that partition to the old Debian partition and use the old swap as a swap partition again
<quiliro>what do you think?
<quiliro>do you know how I can install GuixSD from Debian on the old swap partition now that I have formated it as ext4?
<eacces>how big is that swap partition?
<quiliro>eacces: let me see
<quiliro>1.8 GB
<quiliro>I have already partitioned it as ext4
<quiliro>and added the bare-bones.scm from data in the info manual in 0.13 in file /mnt/etc/bare-bones.scm
<yoyo_jelly>hello, is wayland the default gnome 3 session in guixsd? if not, is it supported?
<bavier`>yoyo_jelly: iirc guix's gnome has wayland support, but idk if it's default
<yoyo_jelly>thanks mate, does anyone else know if wayland is the default gnome 3 session?
<davexunit>it's not
<yoyo_jelly>ok thanks
<bavier`>I'm considering doing some work on build-aux/compile-all.scm to get it to integrate with make's job-server for parallel builds
<bavier`>currently compile-all.scm compiles with all cores available on your system, regardless of what you pass with -j
<bavier`>which is a bit counterintuitive
<rekado_>bavier`: that would be great!
<bavier`>rekado_: I was hoping it'd be appreciated :)
<quiliro>bavier`: all your work is greatly appreciated!
<quiliro>would someone please guide me with the installation i propose above?
<quiliro>i have a desktop machine i could use as a substitute server...mirroring hydra
<quiliro>i could then install guixsd offline
<quiliro>but i cannot install GuixSD on this laptop because it only has one usb port and i have to choose to use it for the wifi card or the usb installer....
<quiliro>i was told i could install GuixSD on a partition from a running GNU OS
<quiliro>I am currently running Debian GNU with Guix on it....and i have a 1.8GB partition with ext4 which i can install GuixSD on
<jsierles>is it possible to generate a manifest from an existing profile?
<apteryx[m]>Any sysrq trick to recover from a trashing to swap state? I got into that by building Guix apparently.
<apteryx[m]>I tried Sysrq-f, which triggers the OOM killer, bit it refused to kill guile as guile is apparently the init process :)
<apteryx[m]>Oh well. Powered off.
<jsierles>is there a way to generate a docker image with 'guix pack -f docker' that will just mount a global /gnu/store? i guess it doesn't need packages, but needs to at least work to boot up in docker
<janneke>how do i know where `git pull' will bring me / has brought me? i'd like to create alignment between `guix pull' and `running guix from git' setups
<efraim>perhaps ./pre-inst-env guix --version will give you something to work with
<janneke>efraim: perhaps...? i'm just wondering how i can find out which branch/commit guix-latest.tar.gz is built from
<efraim>From master
<bavier`>janneke: you can do it the other way and sync your 'guix pull' to your git checkout
<janneke>efraim: OK, and how often is that done?
<efraim>Autogenerated for each time someone runs guix pull unless it already exists, saved forever unless they fixed that
<janneke>bavier`: that could work
<bavier`>janneke: the trick I've seen is to pass a url to 'guix pull' that pull the tarball of the commit you're interested in
<janneke>bavier`: ah, that's nice
<janneke>efraim: great, so every commit of master can be pulled!
<janneke>bavier` efraim: is there a way retrieve the git commit hash from a guix pull?
<bavier`>janneke: I don't think so
<janneke>bavier`: OK
<janneke>it would also be nice to produce the tarbal myself, and have others pull --url that?
<quiliro>how can i install guixsd from debian to another partition?
<quiliro>it is in an efi machine
<janneke>quiliro: the obvious answer would be
<janneke>guix system init /mnt/etc/config.scm /mnt
<quiliro>no herd start cow-store /mnt/
<quiliro>'
<quiliro>?
<janneke>eh...you're not running GuixSD?
<quiliro>no
<janneke>afaik, debian will sync before unmounting /mnt ...
<quiliro>Debian
<janneke>but it can't hurd to do something like `sync; sync; sync; umount /mnt; reboot' ?
<janneke>after system init
<janneke>*hurt
<janneke>:-D
<quiliro>it is running now!
<quiliro>:-)
<quiliro>with this error:
<quiliro>warning: failed to install locale: Invalid argument
<quiliro>but continues
<quiliro>janneke: how can i boot it in efi?
<janneke>quiliro: you need to set GUIX_LOCPATH, see `2.6.1 Locales' to avoid that warning on foreign distros
<quiliro>Disklabel type: gpt
<janneke>quiliro: efi: oops...i'd hope/expect that the Grub install will notice that and take care of it
<quiliro>is reported by fdisk -l
<janneke>does the manual mention special things for EFI?
<demotri>quiliro: Why don't you use just an USB hub to have WiFi + memory-stick?
<quiliro>demotri: i used 2 and they would not recognize the flash memory
<demotri>Ah, OK. Too bad :-(
<janneke>ACTION heard many bath-mouthing about efi, but when i got this laptop with efi and i saw that it is FAT, i decided i don't want that in by boot proces and i removed it
<quiliro>janneke: you removed efi?
<quiliro>or you just removed the partition?
<quiliro>how did it boot without efi?
<quiliro>the partitioning is gpt
<janneke>quiliro: yup, tell bios not to use efi
<quiliro>how?
<quiliro>there is no bios in this machine that i know of
<janneke>quiliro: repartition disk
<janneke>no bios?
<quiliro>sure i can remove that partition, how do i boot then?
<janneke>i'd love to go libreboot and no bios but...
<janneke>yeah, you need to check bios/ask google about your machine if and how that's possible
<quiliro>janneke: no libreboot for macbook air a1304
<janneke>quiliro: oh, a mac; that's why no bios. i bought one in 1998, it had openfirmware -- i have no idea how macs boot these days
<quiliro>i would never buy a mac
<quiliro>it was given away because it did not work
<quiliro>janneke: i wonder if there is a good quality hardware without the user subjugation
<quiliro>available
<janneke>yeah, i wanted to support powerpc vs intel, hacked on linuxppc for a bit
<janneke>we all know how that went
<quiliro>oh...i could never make it work at that time
<quiliro>yup
<quiliro>locale command did not work
<quiliro>but init is working
<quiliro>it broke saying that it had no root priviledges for /mnt so i ran it with sudo
<quiliro>and it broke again and saying to add --fallback on python download
<quiliro>so i added --fallback
<quiliro>but since it now downloaded python correctly, i stopped with ctl c
<quiliro>and started again without --fallback
<quiliro>but had the same problem as before so i started with --fallback again
<quiliro>weird!
<reepca>alezost: "ERROR: In procedure primitive-load:
<reepca>ERROR: In procedure scm_lreadr: /gnu/store/0c79sfycdzsyv74s02b6hmzfyx8y37m8-guix-source/build-aux/build-self.scm:148:7: Unknown # object: #\\~"
<reepca>
<reepca>I checked the mentioned file - there's a gexp where it's complaining
<reepca>,bt just gives "In ice-9/boot-9.scm"
<alezost>reepca: oops, sorry, I have no idea why it failed there :-(
<reepca>me neither. I think I'll restart emacs and try again and see what happens.
<reepca>M-x guix-pull is working after the restart... strange.
<reepca>baby graphics card arrived a day early! Time to reboot and hope I get my screens back! Wish me luck!
<quiliro>i should sudo guix pull
<quiliro>and then sudo guix system install /mnt/etc/bare-bones.scm /mnt
<quiliro>?
<reepca>Well, apparently the R5 230 doesn't work with linux-libre after all. At least, not the one I got. Dangit h-node...
<quiliro>reepca: which one?
<reepca>this one https://www.newegg.com/Product/Product.aspx?Item=N82E16814125841&cm_re=radeon_r5_230-_-14-125-841-_-Product
<quiliro>/gnu/store/417f7dl5dq1249fnnp0mzwj7xyszyqyp-grub-2.02/sbin/grub-install: error: /gnu/store/417f7dl5dq1249fnnp0mzwj7xyszyqyp-grub-2.02/lib/grub/x86_64-efi/modinfo.sh doesn't exist. Please specify --target or --directory.
<quiliro>guix system: error: failed to install bootloader on device /gnu/store/0x7sln6aay56yc8pii4yz9gpj5zshiz9-bootloader-installer '/dev/sda'
<slyfox>ACTION keeps getting 504 for https://hydra.gnu.org/project/gnu most of the time
<ng0>the memory leak we have, is there a formal bug open for this, or was it jus temails so far? A friend just ran into this and there was nothing obvious in the bug list, I just told him before about this issue that might appear
<jonsger>sneek: later tell reepca I think this is the last series who work with linux-libre https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units#Radeon_X1000_Series I'm not sure but that's the last series who don't show up really here https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/tree/radeon
<sneek>Got it.
<janneke>sneek: later tell OriansJ: i have removed 45/51 lambda's from mes's .o output -- hex3 now passes 80% of the test suite. I also have experimental hex2 output. Problem with hex2 is i need several flavours of labels: function-local goto labels, absolute address labels, relative address labels (and 1, 2, 4 bit labels -- that bit is easy and resolved). Usually assembly handles this, but with hex...I could do with some help here going
<janneke>forward.
<sneek>Okay.
<janneke>sneek: botsnack
<sneek>:)
<janneke>ACTION -> zZzz
<quiliro>i could not boot on guixsd
<quiliro>i could only boot on debian
<quiliro>i could not even get the rEFInd menu
<quiliro>i reinstalled rEFInd from Debian (which i am using now)
<quiliro>but it will not recognize guixsd
<quiliro>everything (for bare-bones.scm) is installed except grub
<quiliro>how can i repair this via booting from the guixsd install usb?
<snape`>I've never been able to get 'guix build --check some-package' to work. I must miss something obvious.
<snape`>oh, with --no-grafts, it works