<castilma>only that after a reboot mount still gives me ...data=ordered.. <str1ngs>I don't know about data option, but I use options for setting a subvol, and it works. <thomassgn>I'm on a fresh guixsd install with linux-libre-4.14.3 and when I run guix system reconfigure or guix package commands I get 'error: open-file: No such file or directory: "/gnu/store/...-linux-libre-headers-4.4.47.drv"' Why is it looking for these headers, and more importantly - how do I solve that? <civodul>thomassgn: that's on the installed system or in the installation image? <thomassgn>it's in the installed system after booting and loging in <civodul>what does "guix gc --list-live | wc -l" return? <thomassgn>"guix gc: error: build failed: program `/gnu/store/pii5cimi72lj5l7793h54g5sg0sr2apl-guix-0.14.0/libexec/guix/list-runtime-roots' failed with exit code 1" <civodul>hmm does this command fail systematically? <thomassgn>3 times in a row --list-live fails the same way, if that was what you meant with systematicly <castilma>looks like a corrupted guix and linux-libre.drv. bad hard drive? <thomassgn>haha, I think I understand. I did a fresh reinstall on an old disk. <thomassgn>running guix gc --verify removes a bunch of stuff from a db I was unaware of right now :P <thomassgn>maybe I should redo the reinstall but remove more than the store ... <thomassgn>going to redo the install, but remove /var/guix also... <thomassgn>civodul: not sure if you remember, but you helped troubleshoot this machine some time ago (a month or so) because of a duplicate coreutils in profile. In the end it looked like I had a bad/corrupted disk. I found it to be a bit strange - because everything else worked fine on the system. I discovered when I reinstalled now that I got the same error while on the live disk, and found I had coreutils in my OS <thomassgn>declarations packages field. Commented that out and haven't seen the error. <castilma>when I want to start the guix daemon _service_, i get :guix system: error: exception caught while executing 'start' on service 'guix-daemon': <castilma>ERROR: In procedure delete-file: Is a directory <castilma>how can i get more info? I mean, it should it least print the filename it wants to delete... <castilma>(I had multiple errors in the fs. maybe that changed some things I need to correct manually now) <thomassgn>I've had coreutils there since I started using guix, and the problem appeared in the beginning of september. I'm not sure I should write a bug or something on it, seeing as the error looks nothing like an error in config... ? <civodul>thomassgn: if you're unsure about your hard disk, you should probably make sure that it's healthy first <civodul>otherwise you could have all sorts of problems <castilma>does user-homes start on your systems, guys? <castilma>fixed my problem: /var/log/guix-daemon.log was a directory... <castilma>don't know, how that happened. however, shepherd or it's services need to be more verbose! <str1ngs>castilma: I think user-homes just create user directories. it only meant to run once. <castilma>arother problem: ssh-daemon fails when started on system boot. but launching it manually works... <pkill9>where does guix get all the scheme files/package definitions (not sure what they're called)? <pkill9>as in, where does it search for them when you run `guix package -i something`? <clacke[m]>after guix pull the latest definitions are in ~/.config/guix/latest <clacke[m]>I'm very hazy on where the guix client itself ends and where the pulled guix begins. There's more than the package definitions there, but one still needs to upgrade the guix package from time to time. <efraim>lumina can't find poppler-qt5.h, it lives at poppler-out/include/poppler/qt5/poppler-qt5.h and the #:include is <poppler/qt5/poppler-qt5.h>, am I missing something? <rekado_>roptat: I’m going to continue the review of your Java patches today. <roptat>by the way, I'm stuck with gradle :/ <roptat>I'm trying to build gradle-docs, which contains api-mapping.txt and default-imports.txt that are required at runtime <roptat>but the way they are built is a bit obscure. They use a custom gradle task that require a full gradle (including gradle-docs) to be run <ofosos>In a build recipe, how can I get the /gnu/store/...foobar path where the output goes? <efraim>If you search through the source for 'output' you should get a bunch of examples ***Steap_ is now known as Steap
<hulten>I am using GuixSD 0.14.0 on real hardware. Can I add a second operating system to the system? <hulten>That is, in grub, but by using the GuixSD configuration, so an extra (operating-system ...) declaration? <hulten>OriansJ, should I do this in the config.scm, or should I edit the GrUB configuration directly? <OriansJ>hulten: depends if you want guix keep your changes to Grub or if you want to manually add them every time you have guix reconfigure <hulten>As I am not so into scheme syntax, would this be (operating-system ...) (operation-system ...) as two separate declarations? <shiranaihito>i'm trying to install GuixSD with a system configuration, but it contains an nginx definition too, and i get an error that says: "ssl-certificate in the nginx configuration for the server with name "(www.example.com)" does not exist" <shiranaihito>so i guess i should start with some kind of bare bones system config and then move on to a more "fully featured" one once GuixSD is installed? <shiranaihito>.. or is there some kind of standard workaround for this kind of stuff? <hulten>shiranaihito, I am quite new to Guix(SD), but I would say doing a more minimal install, and later on adding configuration and then 'guix system reconfigure your-scm-file' should be a good aproach. <shiranaihito>it would be nice if the installation image included some kind of minimal python btw - then you could presumably use Salt Stack for partitioning disks and stuff <OriansJ>shiranaihito: you don't need Salt Stack when you have guix <shiranaihito>OriansJ yes, but i don't "have Guix" like that yet when installing GuixSD :) <OriansJ>umm shiranaihito you mean to a virtual machine or to an alternate operating system? <shiranaihito>so before i can use Guix in a way where i wouldn't need Salt Stack, i need to install the whole thing, etc <OriansJ>well on a virtual machine, you would be loading the guixsd Image which has guix by default and can simply just apply your guix config <shiranaihito>right, but i need to partition disks and set up file systems (and transfer over the config file) and so on, before i can do that <OriansJ>However if you are referring to the creation of the virtual disks, that can be done in the host system. <shiranaihito>i ran guix system init with a config file, and it's been just printing that message over and over again <shiranaihito>btw, the installation says "Some deprecated features have been used" <shiranaihito>.. and i doubt it's *me* using those "deprecated features" :p <shiranaihito>is this normal btw: "warning: collision encountered: /gnu/store/54k4zlx3azmhqydlilysirpqmhvcw8gr-inetutils-1.9.4"? <shiranaihito>any ideas about this: "grub-install: error: /gnu/store/[..]-grub-2.02/lib/grub/x86_64-efi/modinfo.sh doesn't exist. Please specify --target or --directory" - it seems i do have a "target" for the bootloader in the config: "(bootloader (bootloader-configuration <shiranaihito>OriansJ well, thanks :). but i think what i have has been copy pasted from some docs <shiranaihito>maybe i modified and messed it up somehow, but i don't know that i did <OriansJ>you can also copy the template examples that come with the guixsd install image <OriansJ>shiranaihito: it sets a name, it is not required <hulten>I thought it meant that you want to use label names instead of UUIDs or /dev/... <OriansJ>hulten: it is about setting device label names if I remember correctly <hulten>The apostroph I'm curious about. Is that Scheme syntax? <OriansJ>hulten: yes, it is short for (quote label) <rekado_>hulten: you can only have a single operating-system declaration in your configuration file. <rekado_>Guix won’t install two systems like that. <hulten>rekado_, I tried with two operating-system declarations, and guix did not complain. I got extra entries in GrUB when booting but the alternative OS did not boot. <hulten>shiranaihito, that looks fine (assuming two partitions with ext4 filesystems are labeled root-disk and data-disk). <shiranaihito>.. and it's alright to keep re-running "guix system init /root/config.scm /mnt"? <hulten>shiranaihito, I guess so, but once you booted your system, replace init → reconfigure. I think you can alway reconfigure later on in a working system. <hulten>You can put your config file anywhere. Just be sure you will have it after rebooting your system. I have put it at /etc/ of the root of my installed system. <shiranaihito>well, this seems relevant: "/gnu/store/xqfpwxbghvi3h85pfji6kdzw7db2sgxz-grub-2.02/lib/grub/x86_64-efi/modinfo.sh doesn't exist". (from the middle of the error message) <shiranaihito>hulten i have the config file on my desktop computer, so it's fine if it's "lost" in the installation :) <hulten>Good, but I don't know how to solve your error. <shiranaihito>"modinfo.sh doesn't exist" .. if that's the problem, it doesn't seem like something i caused <hulten>I don't know why modinfo.sh is missing. <hulten>Maybe you can check if there are multiple profiles in /var/guix/profiles/per-user/root/ , and if so then see if there is a modinfo.sh in one of those, and then use that profile. <pkill9>if i run `guix pull` on my user account, is it safe to then copy the symlink created in ~/.config/guix called 'latest' to my /root/.config/guix/latest ? <pkill9>otherwise it tries to compile guix all over again rather than just symlinking to the one already created in the store <shiranaihito>there was nothing under /var/guix/profiles/per-user/root/ <hulten>Hmm, I'm looking at an installed instance of GuixSD. I don't know if that compares to install-time configuration. I hope someone else here can help you with this missing modinfo.sh. <shiranaihito>so if the official bare bones example doesn't work, i guess it's not a configuration issue <shiranaihito>not sure what else could be wrong either though - i've followed the instructions from the docs, with partitions and stuff <civodul>shiranaihito: are you doing a UEFI install? <hulten>pkill9, I don't know the details, but packages are not recompiled when adding them through guix as different users. <hulten>For instance, I just installed tmux under root, and it was already installed under another user. In both guix-profile directories bin/tmux are symlinks to the same file in /gnu/store/. <hulten>From the documentation I understood that every menu-entry added in menu-entries is additional to the default current GuixSD and submenu for old generations. <hulten>So may I then assume that other configuration like file-systems and users applies only to the current GuixSD generation, and not to the extra menu-entries? <mb[m]1>civodul: Any idea what's going on with Hydra? <mb[m]1>Are all build machines kernel 3.2.0 or higher? <efraim>I'm pretty sure the x86_64 ones are <efraim>The arm ones should have to be, that was back at like glibc 2.20 for them <g_bor>I have a package with a lot of inputs, and I would like to have one with one of the inputs replaced. <g_bor>I've inherited from the original, but how should I specify inputs in this situation? <efraim>,@(delete foo (package-inputs bar)) I think <efraim>The python packages might have a model you can use <mb[m]1>librsvg 2.40.20 is the last version that does not require rust <pkill9>hulten: yeah it does that for me, `guix pull` seems to behave differently though, it tried to compile guix instead of just symlinking to the already compiled version which I had created by running `guix pull` as my non-root user <shiranaihito>civodul i did set something to "UEFI" .. should it be "Legacy BIOS" instead? <pkill9>shiranaihito: i get that too, i think it's updating different files but the url is too long to show <pkill9>but they're from the same domain <pkill9>that's what i think, not sure though <shiranaihito>pkill9 but it's supposed to download a *list* of substitutes.. why wouldn't the list contain all the subs it needs <shiranaihito>or will there be a list of subs for .. every package/lib involved? <pkill9>tbh i don't know exactly what it's updating, whether it's a list or something else <shiranaihito>also, i don't think the official installer should be producing warnings like: "Some deprecated features have been used" <castilma>shiranaihito: its a known issue. you can find more info on it on the mailing list(guix-devel I think). it is a target for the next release to fix that. <castilma>but the 'deprecated' thing was mentioned there, too <shiranaihito>so is there something i can do to fuck up my VM's settings to make that error happen? <shiranaihito>i just changed "firmware type" from "UEFI" to "Legacy BIOS" <shiranaihito>.. or is there something i can do to fuck things up while following the official installation instructions, with network and partitioning stuff, before the instructions get to "guix system init"? <efraim>You need a small 1 MB partition at the front of the disk for the BIOS <castilma>it is 'the old' way of booting a system. the first machine code that loads some code(grub in this case) on a special place on the hdd. <castilma>you can use it. but you need to make an aditional partition with Type-UUID: 0FC63DAF-8483-4772-8E79-3D69D8477DE4 <castilma>Type-UUID: 21686148-6449-6E6F-744E-656564454649. you get that in fdisk with >t >partnumber >4 <castilma>or, if you like to live dangerously. just ignore the error and run grub-install --force /dev/sdX after every reconfigure <shiranaihito>castilma so.. do you think i should go for UEFI right away then? it seems without it, there's a limit of max 2TB per disk.. but it's not like that would be a problem any time soon :p <castilma>are you installing on a vm or real machine? <shiranaihito>castilma VM first, but i'm scraping together a "production environment" along the way <shiranaihito>castilma and ofc, i'll be running it on a VPS to begin with, etc <shiranaihito>btw, it would be nice if GuixSD's installation instructions mentioned the need for a BIOS partition, for us noobs :p <castilma>but if your real system doesn't support bios boot, only uefi, you need to use that ofc. <shiranaihito>castilma mmm, yeah. well, i guess i'll stick with BIOS for now <castilma>see 6.1.4.3 paragraph 3 mentions the bios boot partition <shiranaihito>castilma i'm going to bed now :) i'll continue later, thanks guys! :) <hulten>I get an scm-error misc-error when doing `git pull`. Is there an issue? <ofosos>hulten: b8396f96bfeadfa63e7ad2afc2ab5a37f37f5f81 ? <ofosos>civodul: I think aforementioned commit breaks guix <pkill9>there's an error in the package for surfraw <ofosos>pkill9: a teacher of mine once said: it's easy you sit down, think hard and then write down the solution. <pkill9>it looks simple to fix, i'm curious how it gets fixed. The error is: <pkill9>surfraw ERROR: couldn't find global config in /gnu/store/api0anihz9qa8clab7i35mgsx340swnk-surfraw-2.2.9/etc/xdg/surfraw/conf or $XDG_CONFIG_DIRS <pkill9>looking at it the conf file is there <pkill9>no, the conf file is in my profile's /etc/xdg/surfraw/conf <pkill9>it's looking for it in the surfraw store <rekado_>pkill9: it says it will also look for the file in XDG_CONFIG_DIRS (e.g. ~/.config/surfraw). Maybe copy it there. <pkill9>i don't really have much use for the surfraw program, i'm more interested in the error <pkill9>i don't like that it doesn't work out of the box when there's no reason for it not to <rekado_>maybe it should install a default config file then. <rekado_>in Guix programmes should not expect to find anything in /etc <pkill9>/gnu/store/api0anihz9qa8clab7i35mgsx340swnk-surfraw-2.2.9/etc/xdg/surfraw/conf <ofosos>/home/mark/.guix-profile/bin/surfraw: line 163: perl: command not found <ofosos>pkill9: do you want to create a patch for that? <pkill9>i was just wondering if it's the fault of Guix or the fault of the surfraw package <pkill9>and if it could be solved by changing Guix, so that similar errors don't occur and less patching is needed <ofosos>If you look at the `surfraw' shell script, there's the config lookup beginning line 47. I would assume the configure script gets passed --prefix and surfraw uses that to build the path to the global config <ofosos>Also, see line 163, if your profile does not contain a perl binary it will flatly fail, since it doesn't include the path to the perl store item that is passed as an input <rekado_>it really should not look for the config at /etc; if the file at /gnu/store/…/etc/ exists, then this should be diagnosed, not changed. <rekado_>and it should be patched to retain a reference to perl. <ofosos>pkill9: do you mind if I go ahead and try to find a patch? <civodul>ofosos: re 'guix pull', oops; i'll push a fix shortly <ofosos>So the idea is: copy the config file that is already built into the output directory and patch the surfraw file to the perl interpreter found in inputs <pkill9>what was the issue with `guix pull` civodul? <civodul>pkill9: the commit that ofosos mentioned added a hard dependency on gdbm, but 'guix pull' would try to build it without that dependency <civodul>that fix was to make that a soft dependency, as we do in similar situations <ofosos>pkill9: so from the source tarball it looks like the perl call is hardcoded and can't be modified by some `./configure --with-perl=` magic <ofosos>The config file get's written by configure and obviously configures the available browsers at build time, but... there are no browsers at build time :( <civodul>it should be changed to pick a browser at run time <ofosos>Interesting, there's also something happening in the git repo <ofosos>civodul, rekado_: perl interpreter selection: move into configure script and upstream? <rekado_>ofosos: I’d just patch the script that calls perl. <ofosos>I did that, the last release is from 2013, but upstream has had activity this year and we might profit from stuff that happened on git. I hope they do a release soon <rekado_>sneek: forget I did that, the last release <rekado_>I have a problem with a module cycle again. <ofosos>I need to run `autoreconf' during a package build, I added an 'autoreconf phase after 'unpack, it looks like I need to add autoconf as a native input. Does that sound correct? <ofosos>rekado_: I didn't make too much of a fuss, I simple let autoconf look for 'perl' and inserted that path into the shell script <ofosos>rekado_: I requested the status of upstreams release schedule and the idea is that they want to release before years end <davidpgil[m]>I just installed guix package manager on trisquel gnu/linux <davidpgil[m]>then i installed a program via "guix package -i program-name" <davidpgil[m]>typing "which program-name" doesn't show where the program is <davidpgil[m]>"guix package --list-installed" confirms the program is installed <wigust_>davidpgil[m]: "Source etc/profile to augment PATH …" <pkill9>i would first look in ~/.guix-profile/bin <pkill9>do you see the program name in there? <wigust_>davidpgil[m]: "printenv PATH" will show your PATH. /home/<YOUR_USER>/.guix-profile/bin must be there. <pkill9>change to ~/.guix-profile/bin and type ./program-name <pkill9>though for future you'll wanna add that folder to the $PATH by sourcing ~/.guix-profile/etc/profile <pkill9>in your shell's rc file, or profile file <pkill9>so you can just type 'program-name' <davidpgil[m]>ok thanks for the help at least i know where the program is <pkill9>what happens if you run `$HOME/.guix-profile/bin/program-name`? <pkill9>yeah, you could put `source ~/.guix-profile/etc/profile` at the top of ~/.bashrc if your user's shell is bash <pkill9>how do you update the local package definitions (the scheme files)? <pkill9>how do I tell `guix import pypi <package>` what version of python to get the pypi package definition for?