<nckx>buenouanq: Guix doesn't handle hardware events, udev does. I'm not sure there's a Guixy way of writing udev rules (short of plonking PLAIN-FILEs into /etc). *nckx checks the Cookbook… nope. <rekado>had to rebuild it locally and then rsync it over <roptat>do you think we could automate this? <kernel-help>Hey roptat, nckx: Sorry I disappeared, I am on a significantly more powerful computer now and the hanging is still occurring here <roptat>can you check what it's doing with top or something? <roptat>here it wasn't actually hanging, but doing some work in the background (building a big package) <kernel-help>hmmm there seems to be a build process for python 2.7 <kernel-help>is there a way to turn up the verbosity for guix system build? <kernel-help>I wonder why the build output is being suppressed thouguh <lfam>I'm not sure specifically about `guix system build` but in general you should pass '--verbosity=9' to user-facing commands (as opposed to developer-focused commands like `guix build`) <lfam>I would assume that `guix system build` is verbose by default by I don't know. It's also not helpful to print build output when more than one thing is being built at a time <roptat>even with verbosity=9, I don't get any message about the kernel being built <kernel-help>I turned up the verbosity all the way. the output is still suppressed <roptat>it might be related to the fact it's an inferior ***sputny1 is now known as sputny
<roptat>so it seems to be the call to operating-system-derivation <roptat>I tried copying your os declaration to a guix repl and created a variable with it, named os <roptat>then ((operating-system-derivation os) (open-connection)) proceeded with building the kernel <roptat>instead of simply returning a derivation <cbaines>I use an inferior for the linux I use in my system definition <roptat>guix pull has a similar behavior now that I think about it <cbaines>I'd guess that the build should take place outside of the inferior though <cbaines>is there an example of the system definition (or use of the inferior) that exhibits the problem? <roptat>but there's really no problem per se, it's just doing work without telling anything <roptat>I don't expect guix to need to build a package in order to get a derivation (not build it) <cbaines>the config matches mine I think, I can't remember if I've seen this behaviour though <roptat>you wouldn't notice if you already have that kernel built <roptat>seems to be related with system-service-type <roptat>the build starts after fold-services <roptat>so in system-derivation, in (gnu services) <roptat>seems to be the call to mapm/accumulate-builds <roptat>now that sounds very familiar, I think we had the same kind of issues in (guix self) <roptat>I don't think I can track this down further, so I'll send a bug report with what I found <dissoc>i dont understand what use-service-modules or use-package-modules does in the config.scm file. where is the definition of this function? ***Guest42105 is now known as sturm
***jonsger1 is now known as jonsger
***catonano_ is now known as catonano
<roptat>dissoc, they are macros, basically (use-package-modules p ...) -> (use-modules (gnu packages p) ...) <wdkrnls>nckx: Re: 10 h ago on my printer issue - my printer is Samsung as you suspected and yes I did put splix into my cups extensions. Perhaps I also need to download it into my user profile as well? <wdkrnls> (list hplip-minimal cups-filters splix)) <dissoc>roptat: thanks. i think i looked everywhere but that. <cacsr>Will Guix support multiple C libraries in the future? For example, support both glibc and musl libc <xelxebar>cacsr: Yeah, I would love this too. I know Void Linux maintains binary package builds for musl. Ostensibly once could try the moral equivalent of s/glibc/musl/ on gnu/packages and see how far you could get. <Gooberpatrol66>I edited pulseaudio-service-type with (script-file "/home/nathan/.config/pulse/default.pa"))))) <Gooberpatrol66>In procedure stat: No such file or directory: "/home/nathan/.config/pulse/default.pa" <lfam>Maybe the default.pa file references something that does not exist <lfam>Does that file point to any other files? <lfam>Can you give any more info? Like, what did you try? Where did you change the service-type, etc <lfam>What command did you run? What was it doing when it crashed? <lfam>It's hard to help without more info <Gooberpatrol66>guix system: error: build of `/gnu/store/krlv0jxbpcrchhksbdwfwggdav0db3vb-system.drv' failed <Gooberpatrol66>bash-5.0$ bzip2 -dkc /var/log/guix/drvs/3y/bxvw8rmn4q7r2zlcrv28qimmk1619v-pulse.drv.bz2 <Gooberpatrol66>In procedure stat: No such file or directory: "/home/nathan/.config/pulse/default.pa" <lfam>I'm not familiar with the pulseaudio-service-type, but I'd guess the default.pa file needs to be managed by Guix <lfam>If it needs to access the file while building '/gnu/store/...-system.drv', it won't be able to find it, because there is no access to /home in the build environment <lfam>The environment in which things are built by Guix is limited <lfam>Also, my suggestion was only a guess. I don't know for sure that's the issue here <lfam>You could try adapting the example from the manual to use a different file in the pulseaudio package, or another package, and see if it still fails <lfam>The clearest measure is to strace the guix-daemon and see exactly what it's trying to do. "No such file or directory" can be a misleading error message on Linux <lfam>Get strace with `guix environment --ad-hoc strace`. Then, stop the guix-daemon, copy the full command that is used to start it, and run that command with strace: `strace -f /var/guix/profiles/per-user/root/current-guix/bin/guix-daemon --build-users-group=...` <lfam>I would add `2>&1 | tee ~/tmp/guix-daemon.log` <lfam>Then search that log for "/home/nathan/.config/pulse/default.pa" <lfam>You could share the log or at least the relevant parts <lfam>It will probably be really long <lfam>Copy it from `ps aux | grep guix-daemon` <lfam>I should also ask, what do you want to change about the defaults? Maybe there is a better to accomplish it <Gooberpatrol66>I tried disabling pulse but it didn't work so now I'm trying to add pulse-jack module so they can both run <Gooberpatrol66>I'm getting "error: cannot bind to socket '/var/guix/daemon-socket/socket': Address already in use" from strace command <Gooberpatrol66>strace -f /gnu/store/fvvpmrgnvr9jqxfn5m956xblisa8vzr4-guix-1.1.0-4.bdc801e/bin/guix-daemon --build-users-group guixbuild --max-silent-time 0 --timeout 0 --log-compression bzip2 --substitute-urls https://ci.guix.gnu.org 2>&1 | tee ~/tmp/guix-daemon.log <lfam>It means the guix-daemon is already running. You need to stop it <lfam>Honestly I don't think this direction of investigation is going to pan out <lfam>There are usually some people that use JACK on Guix here during the European day. I would try again then or ask on help-guix <lfam>I'm sure they can give you simple instructions to get it working <PotentialUser-10>Hello everyone! I'm having trouble trying to use guix on ubuntu 20.04. Could someone help me? <terpri_>PotentialUser-10, probably, what sort of problems are you having? <PotentialUser-10>Hi, terpi_. Actually, i'm having the same as in this issue https://issues.guix.info/41603 but doing the fix they advise (--authorizing) does nothing for me. The command seems to just not work as my terminal remains blank except for the usual 'hint' message. Using EOF helps leaving that freeze state and gives: guix archive: error: s-expression does <PotentialUser-10>I'm thinking about reinstalling guix, but don't know how to do it properly <terpri_>PotentialUser-10, how did you install guix in the first place? the official shell script for installing on "foreign" distros? (can't remember what it's called) <terpri_>er, or maybe guix-install.sh is supposed to do *all* of it, it's been a while since i installed atop a foreign distro (debian in my case) <terpri_>maybe we need an "uninstallation" section in the manual :) <terpri_>PotentialUser-10, probably not your problem, but are you using a non-default system or user umask? what does "sudo umask -S" say? <terpri_>(i recall some issues when i had a restrictive default umask set -- no reading for 'other' users besides the creator) <terpri_>yeah, that looks like the default, so not the issue <terpri_>PotentialUser-10, does anything work at all, e.g. sudo guix install hello? or it is only the "guix substitute: [...] substitutes may be unavailable problem" <rlp10>is there a small package that I can install as a test to make sure things are working? <terpri_>rlp10, "hello" (GNU hello) is the canonical one, i think <rlp10>> guix package: error: remounting /gnu/store writable: Invalid argument <rlp10>Looks like I have broken something :( <rekado_>apteryx: I initially liked Ivy but some of its quirks annoyed me so much that I went on the search for something … “better” <terpri_>guix uses its own copy of glibc, so it doesn't use ubuntu's locale files <terpri_>PotentialUser-10, and GUIX_LOCPATH is set for both user and root, in your shell init file? (~/.profile for bash iirc) <terpri_>to dirs that actually exist? ($HOME/.guix-profile/lib/locale for both user and root's $HOMEs) <terpri_>well, just check with ls -d (sudo'd for root) <rlp10>Ugh, sorry to trouble you folks. When I install the hello package, it tells me that the connection is refused to /var/guix/daemon-socket/socket. If I run guix-daemon it says that it can't bind becaues the address is already in use. What am I doing wrong? I can't understand if it is running or not. <terpri_>rlp10, ps aux | grep guix-daemon should show the daemon running as a root process, if it's running <terpri_>(there might be a built-in command for it, but that's how i check) <rlp10>hmm, so it's not running. if i run it I get 'address already in use' and if I run it as root, it can't find the command. how do I run it? <terpri_>rlp10, are you running guix system, or on top of a foreign distro? <rlp10>i thought i'd try to get to know it first (i'm using debian) <PotentialUser-10>terpi_, so both dirs exist, I've reinstalled locales, appended both .profile with export GUIX_LOCPATH=$HOME/.guix-profile/lib/locale and still the same <terpri_>PotentialUser-10, and set it in currently running shells / logged out and back in? <terpri_>rlp10, sudo systemctl start guix-daemon.service should start it, i think <sneek>Welcome back civodul, you have 1 message! <rlp10>terpri_: That works, yay. Thank you for your help. Sorry I'm such a newb. <PotentialUser-10>terpi_, I'm beginning to remember, that once I was lacking time to wait some command to finish (it was too late in the evening) and having in mind that guix is so 'bullet-proof' just interrupted it. May be that's the reason? <terpri_>rlp10, np, glad it worked :) (a bit surprising that the installer doesn't do it for you) <PotentialUser-10>terpi_, moreover, from some point I can no longer use info guix and man guix <terpri_>PotentialUser-10, generally it's safe to just kill guix commands, ime <terpri_>PotentialUser-10, but, you can check the store with "guix gc --verify=contents" <terpri_>that checks the hashes of files installed under /gnu/store to make sure there's no data corruption <rlp10>I've installed the hello package :) I can run it manually be doing so from its /gnu/store folder, but it isn't in my PATH. How do folks normally set that up? <PotentialUser-10>terpi_, this line seems to say everything: 'guile: warning: failed to install locale' <terpri_>rlp10, sourcing the guix profile with ". ~/.guix-profile/etc/profile" in ~/.profile (for bash) will set up $PATH and other environment variables for you <terpri_>(prepending rather than overwriting) <rlp10>terpri_: Works great, thank you <xelxebar>Anyone else having issues with guix lint erroring out about problems downloading release info from github? <rlp10>OK, now comes the real test - I'm installing emacs :) <terpri_>PotentialUser-10, that's a bit odd...maybe double-check that $GUIX_LOCPATH is set sensibly, and run again with "guix install --debug --verbosity=2"? <terpri_>(possibly only one of --debug and --verbosity=2 is relevant, also, so you could try twice with one of the options if it's too noisy) <PotentialUser-10>terpi_, here's the last line in my .profile: export GUIX_LOCPATH=$HOME/.guix-profile/lib/locale and the same under root <cacsr>xelxebar: If it is just a simple replacement of keywords, it may not work. There are also differences between the linker and nscd and locale. At present, I have not found a replacement for nscd under musl. <cacsr>xelxebar: I plan to use two or more libc in Guix at the same time. Now it seems that a non-FHS system like Guix can do it. <terpri_>also i just remembered that GUIX_LOCPATH needs to be set in the systemd service file for the daemon to see it, but that was fixed quite a while ago... /etc/systemd/system/guix-daemon.service should have an Environment='GUIX_LOCPATH=...' line <terpri_>PotentialUser-10, do you have substitutes (~= binary package downloads enabled)? it should be fairly fast if so <PotentialUser-10>terpi_, let's continue some other time, I'm no longer able to go on. Thank you so much! <terpri_>i think the install script asks about enabling substitutes, not sure how to check whether they're enabled though <terpri_>PotentialUser-10, sure, and no problem :) <terpri_>PotentialUser-10, generally it's a good idea to resolve whatever you're warned/hinted about, unless intentionally doing something unusual <terpri_>however, the locale warning for example shouldn't prevent you from *using* guix (in english anyway), installing programs, etc. <terpri_>i'm updating guix on my debian laptop now, so maybe it will refresh my memory of guix usage on foreign distros for your next debugging session :) <xelxebar>cacsr: Right. The words "moral equivalent" probably hide a whole gnu-ton of work involved, but it's interesting to note that guile builds under musl. <civodul>cuirass/base.scm:473:19: In procedure regexp-exec: Wrong type argument in position 2 (expecting string): #f <mothacehe>civodul: yes i have a patch, updating Cuirass! <terpri_>gnu-ton isn't an SI unit, but if it's equivalent to an newton, that's merely the force exerted by an Apple at earth gravity ;) <mothacehe>the issue is that we a build succeed i did assume that its derivation is in the Builds table, which is wrong. <aadcg>I reconfigured my system so now linux-libre went from 5.4.43 to 5.4.49. My usb wi-fi doesn't work anymore though (it's a TP-WN821N). Is anyone having the same issue? <alextee[m]>did something change with chromium? it started not being able to load some pages <aadcg>just to be more clear, I have the wifi chip AR7010 and it doesn't work anymore with linux-libre 5.4.49. Anyone sharing this issue? <clodeindustrie1>everything indicate it should be working, but it's not, I get "no default controller" from bluetoothctl <aadcg>which linux-libre version are you running? <terpri_>clodeindustrie1, do you have the corresponding firmware blobs installed, since you're using non-libre linux? i can msg you instructions for installing/updating them if not <clodeindustrie1>I followed the instructions to load the blobs for my wifi (which works great) and my bluetooth but that would great if you could send me your so I can compare <terpri_>clodeindustrie1, np, i'll send in a few minutes after i update my own nonfree kernel/firmware to check my mental "runbook" <mbakke>alextee: can you link a page that fails to load? <alextee[m]>mbakke: if you log into your mastodon/twitter page for example it breaks <mbakke>alextee: weird, I created a mastodon account and can't reproduce, I'm also typing this through riot.im <mbakke>ah, but I'm not using the very latest version, maybe e6db8912fdec8b33e39c03b5e3b3268df12648ab broke something <alextee[m]>i think i had this problem in the previous version too <alextee[m]>i just upgraded my guix to see if it's fixed but it's still there <bdju>how do I get M-x run-python to work in emacs? it can't find python. it probably has to look for python3 instead, but I'm not sure how to make it check <aadcg>install python-wrapper with guix install python-wrapper <rlp10>So, I have emacs running :) Where does it look for its home directory (~/.emacs)? Is there a guix location with separate dot files or does it just use my normal home directory. <mbakke>alextee: can you reproduce it with a clean user profile? i.e. 'chromium --user-data-dir=/tmp/test' <alextee[m]>mbakke: i think it started happening because i had to force-reset my pc and it corrupted some chromium files <alextee[m]>i deleted ~/.config/chromium and everything is fine now <mbakke>alextee: glad it worked, terrible that it happened :-) <civodul>aadcg: oh, something broke, thanks for the heads-up <civodul>i'll take a look if nobody beats me at it <Formbi>is there something wrong with the linux-libre package? <travankor>alextee[m]: i've had to force-reset a few times and have not had that happen with btrfs <travankor>... and some people say btrfs is "fundamentally broken" ;) <kmicu>travankor: you are not lucky. I can make Emacs broken and crashy by not reading its manual and using experimental features then complaining that Emacs is fundamentally broken. <kmicu>I can mount /gnu/store as writable, mess it up and start complaining how broken Guix is xD <alextee[m]>travankor: so you're saying it's a disk IO issue? <alextee[m]>i thought my system ran out of memory or something. it usually happens during my gdb sessions <Formbi>but Guix doesn't have only LTS stuff… <travankor>kmicu: unrelated, but I'm trying the emacs-wayland port and seems like it works nicely :D <travankor>hopefully, the wayland branch is merged by the end of the year <kmicu>Good news. Now you can complement it with some nice pull request adding Wayland services into Guix System, yey. <kmicu>Formbi: d566f533b1 gnu: linux-libre@5.7: Update to 5.7.6. <kmicu>Frequently pulling is the best workout to get big Guix biceps. <kmicu>(and the latest kernel version) <aadcg>kmicu: ahahahah I loved that joke <leoprikler>I think people mean "linux-libre" as in the linux-libre variable, which points to linux-libre-5.4 when they ask stuff like that, because that's the default kernel in the config <mbakke>kmicu, Formbi: the 'linux-libre' package is stuck on 5.4 because no one has volunteered to update the config for newer versions <bdju>it would be nice if the guix package search on the website had a search instead of all those pages to click through. maybe something more like what nixos has on their site <aadcg>bdju: do you use emacs? I search for packages with M-x guix p and then you can search with regex, etc <aadcg>to use that you need emacs-guix installed <bdju>aadcg: I use emacs a little, and I normally use "guix search", but the web ui is nice for linking to people to show them what version a package is on with guix <Formbi>kmicu: I pulled yesterday, I think <travankor>bdju: the hpc guix website has a better search function <bdju>travankor: ah, okay. I'll take a look <bdju>wow! you're right. that is fantastic <bdju>yep, just found it. I'll bookmark that and try to use it from now on <aadcg>I think there are another way to search for them... but I forgot where <mothacehe>I have the impression that when restarted, Cuirass does not resume old builds. Frustrating to see berlin mostly idle with all those queued jobs! <kmicu>Formbi: I’m sorry here, as pointed out by leoprikler, I totally forgot about linux-libre alias. (I tend to est such stuff explicitly, especially for kernel). <kmicu>So Guix build farm builds d566f533b1 only for fun? <leoprikler>I think gnu.org only allows static content, hence no search. One more reason for guix.info :D <aadcg>leoprikler: I volunteer to update the linux-libre config. <terpri_>the gnu.org website is still maintained using CVS (😱), iirc <civodul>(not necessarily a bad thing in itself, except for CVS) <civodul>of course guix.gnu.org is not on CVS, it uses Guix System and high tech like Haunt <janneke>hmm, --target missing on guix environment *civodul writes a goggles-service-type <mothacehe>janneke: having 'guix environment --ad-hoc --target ...' could be nice but it would require most of the time that qemu-binfmt is setup on the machine I guess <janneke>mothacehe: just looking for a better alternative to: "guix build --target... foo --keep-failed" ^C; source environment-variables <rekado_>civodul: thanks for the goggles-service-type! <rekado_>civodul: can we let that service restart automatically? <rekado_>guile-xapian has an undiagnosed crash that forced me to run goggles in a loop <civodul>rekado_: it has (respawn? #t) by default i think <civodul>so don't worry about fd leaks, memory leaks, and crashes: it'll respawn automatically :-) <civodul>do we need to run "goggles index" periodically? ***dddddd__ is now known as dddddd
<civodul>rekado_: so a bi-daily mcron job that runs "goggles index"? *civodul realizes /var/empty was maybe not a good choice for $HOME <stikonas>is the store location (/gnu/store) on foreign distro hardcoded? Or can I change it (for new installs)? <NieDzejkob>you could change it but you would be forfeiting the possibility of using substitutes, so you'd have to compile everything yourself <cacsr>Maybe you can use proot or fakechroot. to fake a /gnu/store <cacsr>This is just an idea, I actually did not practice it. <cacsr>It looks like nix seems to use proot on nix-on-droid *kmicu waits for guix‑on‑replicant <cacsr>I installed Guix on my Android phone, but it requires a lot of manual operations, such as manually creating /etc/group and /etc/shadow <stikonas>NieDzejkob: ok, thanks, maybe I'll try to do bind mounts to be able to use substitutes... I do have Gentoo machine at home though <rekado_>civodul: the goggles code says (define %log-xapian-db "/var/cache/logs.xapian/") <stikonas>oh no, symlinks are explicitely forbidden <stikonas>rekado_: guix itself: guix pull: error: the path `/gnu' is a symlink; this is not allowed for the store and its parent directories <civodul>rekado_: oh ok, i'll add it to the container <civodul>BTW, the xapian test suite is still running on bayfront <aadcg>cacsr: how hard would be to have a "guix-on-droid"? <lispmacs[work]>hi, am trying to work through the shepherd user services blog post <lispmacs[work]>to run a user service, do I have to start a user instance of shepherd, or does the system one somehow detect to use my local init.scm? <nojr>how come the user manual is not available offline in GIMP <mbakke>civodul: the hash algo additions (and daemon improvements) look great to my untrained eye :-) <mbakke>lispmacs[work]: you have to start a user instance of shepherd yourself <mbakke>civodul: I don't think xapian tests are supposed to take hours <lispmacs[work]>mbakke: if I don't want to manually start shepherd each day, what service should I use to start shepherd? <ajgrf[m]>lispmacs[work]: I just put a desktop file in ~/.config/autostart, so shepherd starts when my DE starts. <civodul>it's frustrating that it's still building Xapian (restarted it with --cores=2 in case it helps) <nckx>civodul: I see you made a Goggles service. Does that mean you got guile-xapian to build? I tried running ‘goggles index’ the past few days but never got past the test suite. <nckx>The logs. search index is 2 months out of date. <mwelt>Hi all. Short question regarding config.scm. If I want to change a service configuration, do I need to remove the xyz-service-type from the service list and readd it with a new configuration, or is there a way similiar to `set-xorg-configuration` for other service types too? <mwelt>... provided that the service in question is already referenced in %desktop-services and such <wdkrnls>I'm not sure I understand what you are asking, but in my config.scm I am using (modify-services ...) to make changes to a service specified in a list of services. <wdkrnls> (vpn-plugins (list network-manager-openvpn))))) <mwelt>wdkrnls: looks promising. Did not find modify-services in the docu, maybe I looked in the wrong places <wdkrnls>See the "System Services" section of the Guix Info manual for the "official details" :) <wdkrnls>Its in "Using the Configuration System" (section 8). <amber_lambda>Hi! I'm an arch user who is looking at Guix SD, but I've read in a couple places that Guix SD doesn't support LVM2. These posts were from a couple years ago - has GUIX SD since added LVM2 support? <lfam>Not yet amber_lambda :/ Help wanted :) <lfam>To clarify, we have LVM packaged, but booting from LVM isn't yet supported via the tools used by Guix System to manage the OS <lfam>It would be great to get support for it, although I think some of the use cases are being superseded by features of btrfs <lfam>There's no incompability. It's just that nobody has done the work to make it work with our tooling <lfam>Guix System is configured by editing a configuration file and then running a command to effect those changes. So somebody needs to write some code to support it via that workflow <amber_lambda>Ah, I see. If I had the guile skills and the knowhow of how both LVM2 work and how the Guix config system work, I'd be happy to help <lfam>Guix also works great as a package manager on top of another distro. I use it on Debian extensively and it works well without interfering with the Debian stuff. It's a nice way to test the waters <lfam>Cool! Feel free to ask for advice here if you need it <lfam>There are several X200 users here <vagrantc_>i can't remember who, but someone had some patches for lvm support with guix that needed a little refresh and was planning on sending "soon" <lfam>Sometimes that little refresh ends up being harder than writing the patches to begin with <vagrantc_>it was a missing feature that slowed my willingness to try guix a bit ... eventually just used my old laptop for guix <vagrantc_>and then started going crazy with arm machines ... which is a... challenging way to use guix :) <lfam>It's interesting to me that you were put off by a missing feature and then decided to tackle a huge project :) <vagrantc_>got a little disheartened last time i tried my Debian packages of guix ... i really have a hard time debugging the test suite failures and it seems like new ones keep cropping up every time i fix or disable or workaround a few <vagrantc_>was really hoping to get guix 1.1.0 into Debian... <civodul>nckx: it works on my laptop but it times out on bayfront <civodul>during "Running test: newfreelistblock1" :-/ <civodul>i'm trying with --max-silent-time=36000 <civodul>vagrantc_: sorry for the frustration :-/ <civodul>i'm happy to give a hand if/when you want to give it another try <civodul>it would be really great to complete that effort <civodul>it might be that there are test failures that show up in an FHS context specifically <civodul>which is something that's much less tested <NieDzejkob>What do people use LVM for? I've heard of the use case of snapshots before a system upgrade, but... y'know <txt-file>NieDzejkob: I use it to have variable partitioning. With msdos partition table or gpt it is very difficult to resize partitions. <lfam>You can use it to migrate disk without downtime <lfam>Thinks like that. Basically an abstraction of physical storage <lfam>I mean, things like that <txt-file>and if you use dm-crypt you (more or less) have to use it if you want to have "partitions" inside the crypto container <wdkrnls>mcron can't seem to read ~/.config/cron. <mwelt>is there any thing like easy suspend or hibernate in guix? <wdkrnls>(I was trying to schedule my mail to get updated every few minutes) <wdkrnls>@mwelt: I have been running guix on my laptop with suspend no problems. I just shut the lid and open it again later. <mwelt>wdkrnls: cool! But I would like to suspend my tower PC :) No lid to shut <mwelt>just need some kind of command <mwelt>I suppose it's some shepherd thing <wdkrnls>Hmm... now I'm thinking it's actually controlled through elogind-service... but I forgot. <lfam>You can use sysfs to manage power states <wdkrnls>I remember I wanted once to disable suspend on lid close. <lfam>Check the section Basic sysfs Interfaces for System Suspend and Hibernation <lfam>That's basically what e.g. `systemctl hibernate` is doing <lfam>Apparently you can use elogind as well. Check our manual section Desktop Services <nckx>mwelt: No, just the standard echo mem > /sys/power/state. <nckx>Hibernation also work but I realllly need to push the resume patches. Is next on my list but it's still buggy here. <nckx>So don't hibernate (echo disk) just yet. <cbaines>civodul, bayfront out of space earlier today, and only has ~60GiB free at the moment, I guess that might be causing issues building things <vagrantc_>civodul: yeah, you've been helpful with debugging the test suite failures on debian ... i've had a lot of other things going on and it was just a bit too much to keep going on it :) <ajgrf[m]>What's the best way to capture the output of a command from inside a package definition? Like `foo="$(echo captured)"` but in guile. <vagrantc_>civodul: had it down to 5 failures at one point and got too hopeful :) <lfam>ajgrf: I think you could use Scheme ports <nckx>As soon as I said that I got a spew of backlog from the lag fairy, all credit to lfam. <ajgrf[m]>lfam: OK, thanks. Looks like call-with-output-string might be the easiest options. <rekado_>nckx: sorry about the index; I forgot to start the index job when I restarted goggles. <pinoaffe>hi guix! I found two broken packages: package lxrandr is broken, it can't find xrandr when installed, to fix it xrandr should either be made a propagated input, or lxrandr should be patched to refer to the xrandr in the store <nckx>rekado_: No sorries needed. If anyone noticed, I didn't hear any complaints. Gave me an excuse to read goggles. <pinoaffe>and another one: deluge can't find libtorrent when running, even though it is an input of deluge, here deluge prolly needs to be patched too <nckx>But you should do the latter. *nckx will be popping in & out while debugging hibernation. <civodul>vagrantc_: feel free to post test-suite.log to bug-guix, i'll take a look <nckx>pinoaffe: Thanks for the reports! Don't hesitate to send bug reports for things like that (patches welcome but not required). I'm installing deluge & lxrandr now. <civodul>cbaines: still, 53G should be enough to build xapian, no? <cbaines>maybe, although I think the machine might get slower and slower as there's less and less disk space <civodul>the load is high, but i don't think that's related <cbaines>it might just be anecdotal, I'm not sure <nckx>pinoaffe: How may I credit you in the fixes? ‘Reported by pinoaffe in #guix’? <nckx>pinoaffe: Deluge uses libtorrent to provide the optional ‘thin client’ mode and clearly explains what to do if it's missing (install it), that's fine. <nckx>Er, other way 'round of course. <pinoaffe>nckx: installing libtorrent does not make the notification go away, and `guix environment --ad-hoc libtorrent deluge -- deluge` also results in the same error message <pinoaffe>furthermore, libtorrent is an input of deluge and this didn't use to happen until a couple of months ago, so I think this is a bug/regression <pinoaffe>nckx: no need to credit me, but if you want to 'Reported by pinoaffe in #guix’ would be fine <nckx>pinoaffe: Oh, okay. lxrandr should be fixed on master. <janneke>hmm, noone ever cross-compiled gdb eh? *janneke "almost" has a gdb for the hurd <civodul>so the xapian tests are (unsurprisingly) very i/o intensive <civodul>i've temporarily stopped cuirass to let it make progress <wdkrnls>I just reconfigured my system over ssh. Is there some way to restart it without restarting the computer? I'm trying to debug why I can't get my printer working and my significant other is using the monitor for the computer I'm trying to fix. <wdkrnls>I was thinking I could just run 'herd restart' as root, but since ssh is one of those services, I figure maybe that would not be a good idea. <civodul>wdkrnls: you can "sudo herd restart S" for each service S that you want to upgrade <civodul>that'll stop the running instance and start the new version <wdkrnls>Really the only change I made was add my non-root user to the lpadmin group. Maybe that requires no service changes? <wdkrnls>Well, I guess it still makes sense to restart CUPS. <nckx>wdkrnls: By the way, it turns out (from grepping my print server logs) that I've had the same problem as you. I don't remember how I fixed it :-/ <civodul>wdkrnls: in that case, you might need to start a new shell for that user, or, if you're in graphical session, log out and log in <civodul>you can run "id" in a terminal to see if the lpadmin shows up ***vagrantc_ is now known as vagrantc
<wdkrnls>nckx: you have a split-driven printer then? <nckx>Yes. M2026W, and I was able to print with it. Haven't used it in 2020 but I still have it (just not here, now). <nckx>I think I added splix for it. <nckx>d [10/Dec/2019:02:25:28 +0100] Samsung-M2026W: File \"/gnu/store/x8rmh754pqyv3jg39p2bk67bm1hpb1dh-cups-server-bin/lib/cups/filter/rastertoqpdl\" permissions OK (040555/uid=0/gid=0). <wdkrnls>I actually got it to print from piping an old org file to lpr <wdkrnls>So, I don't think anything is wrong with cups or splix. <wdkrnls>I'm thinking the problem might be a missing input to system-config-printer? It doesn't know about splix. <wdkrnls>I used it to setup my printer and I was basing all my concerns on it's inability to print a test page. <nckx>It shouldn't have to! It should talk to CUPS, which does. (…of course should doesn't mean does). <nckx>wdkrnls: Can you print a test page from CUPS? <nckx>wdkrnls: By the way, when you say s-c-p complains about missing /gnu/store/.../cups/filter/rastertoqpdl, what's the non-hash part of ‘...’? <wdkrnls>nckx: I gave more details on help-guix. <wdkrnls>I realized that it was looking for rastertoqpdl in cups, instead of in splix. <nckx>I read all your mails, but need more convincing 😉 Are you sure? See my log line above; cups-server-bin != cups. <nckx>It's a union directory that should contain all extensions which is exactly what you want. <nckx>If you mentioned /gnu/store/…-cups-2.3.x explicitly, I missed it. <nckx>Hm? I don't see any attachments… let me check again… <wdkrnls>Printer 'SCX-4200' requires the '/gnu/store/rbwxfbg89125qb0f17vj5zh1nw03gfi6-cups-2.3.3/lib/cups/filter/rastertoqpdl' program but it is not currently installed. Please install it before using this printer. <wdkrnls>(but that is what it says underneath "Missing driver") <nckx>wdkrnls: Does /var/log/cups/error_log contain that message (or any error involving rastertoqpdl)? <nckx>I've never used s-c-p, sorry. I don't know if it can be trusted not to be too clever. <wdkrnls>cat error_log | grep -i installed returns nothing - so I don't think it's there. <wdkrnls>however, there is one entry for a rastertoqpdl <wdkrnls>... but I think that is from when I successfully printed my org file. <wdkrnls>I [26/Jun/2020:16:55:41 -0400] [Job 1] Started filter /gnu/store/id1b2573mppw9yhwa7nmyf49862cpsl2-cups-server-bin/lib/cups/filter/rastertoqpdl (PID 4456) <wdkrnls>I will try printing a test page from the CUPS page once I figure out how to over the local network. <nckx>wdkrnls: Yep, that's the right path (server-bin) and confirms that (at least a part of) CUPS proper does know its correct location. <nckx>Then it's under the Maintenance drop-down. <wdkrnls>I was trying to figure out how to do that from a different computer. I guess I'm not sure how to "tunnel". <nckx>I'd use something like ssh -L <local port>:<remote host>:631 you@<remote host> but don't mean to send you down a rabbit hole. <nckx>Then localhost:<local port> in your local browser should connect to the remote CUPS. <wdkrnls>Any suggestions on what <localport> should be? <nckx>:-) Anything over 1000 will work. <wdkrnls>ssh doesn't seem to like what I'm typing (ssh -L 8631:me@desktop:631) - it's poping up 'usage' <nckx>wdkrnls: Drop the me@ for that part, only use it for the 2nd argument. <nckx>Wait, actually it can be ‘localhost’ — I think? <wdkrnls>Okay, that ran :) Now, just trying to figure out how to get it in the browser. <wdkrnls>No worries. In a few hours I can reconnect the monitor and give it a shot. <nckx>wdkrnls: ‘ssh -L 8631:localhost:631 my.print.server’ works here.