IRC channel logs


back to list of logs

<CompanionCube>vagrantc: ZFS deduplication isn't exactly lightweight though. There's a reason why people talk about ZFS needing 1GB of RAM per TB.
<CompanionCube>(misapplication of figures can really become quite the enduring meme)
<nckx>vits-test: ‘the CI has successfully built php for aarch64’: these are the same machines.
<nckx>So it's great that it's available for download but it's not a separate data point.
***meow is now known as jess
<str1ngs>nckx: php is all good now, now off to fix jack2
<bdju>which use-module do I need for the libvirtd service? it says I'm missing one but the hint doesn't say what to add
<nckx>bdju: My safe guess is (gnu services virtualization).
<str1ngs>nckx: weather is not reporting webkitgtk being buit. though I have locally I'm just going to assume it's a long queue or sometihng.
<str1ngs>on aarch64
<bdju>nckx: nice, that appears to be working! thank you
<bdju>>libvirt.libvirtError: Failed to connect socket to '/var/run/libvirt/libvirt-sock': Permission denied
<nckx>(Unrelated: did you also add the virtlog service? An old gotcha was that it was basically required but not pulled in automatically, dunno if still true.)
<bdju>oh, no. this is my first time trying to use it and I didn't even know to add the libvirtd service until virt-manager gave an error that it wasn't running
<bdju>I'll look into that
<nckx>bdju: Sounds like you're not in the libvirt group (yet)?
<bdju>aha, lot of steps to this it seems. I am not in that group
<nckx>If that's missing from the manual please don't hesitate to add it.
<nckx>Or at least file a bug so it's not forgotten.
<bdju>I'm not sure if it's in there or not. struggling to search the page. I was already in the libvirt section, but my page search brought me back to the top to search for "group". anyway I just searched the manual for libvirt to find what to input for the service, I never looked at a sort of guide start to finish
<bdju>that aside, I now have the virtlog service going and am in the libvirt group but it still isn't working
<bdju>I don't need to run this with sudo do I?
<nckx>I don't actually use libvirt... but having to do so would be a bug.
<bdju>there's a very high chance I just don't know what I'm doing, unfortunately...
<bdju> here's the error I get in a little popup window
<bdju>on the bright side I'm only doing this for fun so it's fine if I can't figure it out right now. if anyone who uses this stuff comes on later, maybe they can help me
<nckx>Did you log out & back in, just to make sure your group change took effect? Especially with $modern things like polkit, it can be inscrutable magic.
<nckx>If ‘groups’ says you're in libvirt despite that error it's worth a try.
<bdju>I was gonna say I thought la unching from the shell where I updated woul be enough, but I see now in that shell I am not in that group. I probably need a new shell at least
<bdju>I'll open a new terminal and launch from that, and if that works then I'll log out and back in
<bdju>huh. new terminal shows the same. I haven't added a group in a long time, not sure if that's normal
<nckx>str1ngs: One of my Overdrives (both berlin build nodes) has built webkitgtk on 4546c0dd, but the problem is that berlin won't see or serve it if its worldview doesn't match.
<nckx>If berlin offloaded its build to a different node and that one failed, it won't retry.
<nckx>Or indeed still queued.
<nckx>str1ngs: Just to update my mental bug tracker: at least for PHP & WebKitGTK on aarch64, everything's good now, right? The JACK2 thing is unrelated?
<sunova>Ifam: Thank you. So I don't need to create seperate zfs datasets for a filesystem baselayout, everyting is simply a symlink to stuff inside the store. I can add some hooks to scm files to create zfs snapshots for each derivation inside store. Is that correct?
<sunova>I mean, zfs datasets. There is really no need for snapshots. Everything is hashed by guix if I get it
<sunova>It's really geniune
<nckx>joshuaBPMan: Have you used bcachefs with Guix?
<bdju>nckx: virt-manager does work after restarting my session! nice. thanks for the tip
<nckx>joshuaBPMan: I'd very much like not to be the only user 🙂
<nckx>bdju: That's great!
<nckx>joshuaBPMan: Or without Guix for that matter.
<joshuaBPMan>nckx: I have not.
<joshuaBPMan>I've heard that some others have used bcachefs.
<str1ngs>nckx: it's unrelated. jack2 is blocking gst-plugins-good on aarch64 which is a dependency for nomad. it's less critical then webkitgtk
***stefanc_diff_ is now known as stefanc_diff
<bdju>can anyone recommend a dark QT theme and tell me what program I can use to set a QT theme?
<vits-test>sneek: later tell lfam: '..iOS user who "lost" something..' except, of coure, freedom.
<sneek>Will do.
<vits-test>sneek: botsnack
<brendyyn>Does anyone else find with icedove and icecat, when one clicks "reply via email to <name>" on the mailing list and selects icedove, it switches to icedove but fails to open a new reply email?
<sneek>Welcome back brendyyn, you have 1 message!
<sneek>brendyyn, rekado_ says: babel will most definitely not be easier to package. It’s written in JavaScript and has lots of unpackaged dependencies.
<peanutbutterandc>Okay... I'm trying to package something up, but multiple `guix download`-s show that the .tar.gz file as having different hashes
<peanutbutterandc>what should I do?
<marusich>It would be a good idea to inspect the files to see why they differ, and then decide which one is correct.
<sneek>Welcome back marusich, you have 1 message!
<sneek>marusich, lle-bout says: for cuirass, when I use as specfile - guix-master and wip-marusich-be1 error: - wip-lle-bout-be1 error: - it makes no sense.
<peanutbutterandc>marusich, Hmmm.... but I probably am just using a wrong mirror:// url? Here is what I am using: "mirror://sourceforge/projects/qt5ct-1.1.tar.gz" for
<raghavgururajan>nckx: You around?
<peanutbutterandc>Yup.... every single hash is different.....
<peanutbutterandc>I mean every single download, the hash is different. Probably the mirror thingy
<marusich>peanutbutterandc, it sounds perhaps like the URL is wrong and you are getting an error message, maybe one which includes a timestamp
<marusich>While downloading it, guix probably prints the URL it tried. Can you try that yourself using 'curl' or 'wget' or a web browser and see what happens?
<peanutbutterandc>marusich, It gives me this: "", after being redirected a while.... and the hash of the thing is different everytime. "guix download"
<marusich>So, does not download a tarball for me
<marusich>Try it out. You'll get a webpage instead.
<marusich>Perhaps the way in which you specified the mirror URL is incorrect. You might find it helpful to double check how others are doing it for sourceforce mirror URLs.
<peanutbutterandc>I don't know... sourceforge keeps on sending me on a loop
<peanutbutterandc>take for instance libquicktime's package definition: that is the closest to my package, and yet, it doesn't work for me
<str1ngs>woah! has rainbow deliminators
<raghavgururajan>str1ngs: For service-types accountsservice, udisks, geoclue and lirc, the default-value field is missing. Do you know what it should be?
<raghavgururajan>(default-value foo) or (default-value (foo-configuration))?
<str1ngs>geoclue-service is a procedure
<bdju>I have no sound in qutebrowser. anyone ran into that or know why that'd be?
<str1ngs>it takes keywords.
<str1ngs>raghavgururajan: ^
<str1ngs>raghavgururajan: guix system search geo
<raghavgururajan>str1ngs, There is also geoclue-service-type
<str1ngs>I think that's used within the geoclue-service proc
<str1ngs>if you read the arguments for geoclue-service it has the default values
<raghavgururajan>Ah I see.
<leoprikler>raghavgururajan: anything new on gdm?
<raghavgururajan>leoprikler: My wrapping my head around that.
<raghavgururajan>leoprikler: I suspect, the problem is not with gdm at all, but with gnome-session.
<raghavgururajan>If you could compare current pack-def of gnome-session in master and wip-desktop, you would notice something.
<raghavgururajan>In wip-desktop, it is new version, which came with meson options to disable systemd. So I did.
<raghavgururajan>Now it looks like, it uses nothing (null) for session tracking. Neither systemd nor elogind.
*vits-test reboot->new-kernel->thanks,<str1ngs>.
<efraim>raghavgururajan: the newer version of efl combines the systemd and elogind options and --disable-systemd also disables elogind. maybe the same for gnome-session?
<raghavgururajan>efraim: Nah! No mention of elogind in gnome-session's source code.
<raghavgururajan>I mean for 3.36.0
<rekado>raghavgururajan: this wouldn’t surprise me. We’re patching the gnome-session source code to replace libsystemd-login with libelogind, and similarly for other cases.
<leoprikler>raghavgururajan: as rekado pointed out, you should enable systemd support for elogind
<leoprikler>and patch the sources of course
<leoprikler>Regarding 69e7da457dd6a8bc3092729a1393575a58b61117: Is there a good reason to do this? GST_PLUGIN_SYSTEM_PATH is set according to your profile, so if you install gstreamer plus the plugins you want, you get playback in epiphany
<leoprikler>this patch forces the use of bad and ugly plugins, when that's not necessarily what someone would want
<rekado>the whole point of having separate gstreamer plugin packages is so that users can pick the set that they want (or none at all)
<rekado>so forcing any particular set is not a good idea
<rekado>I’m trying to package a Rust thing and got this error: “error[E0658]: use of unstable library feature 'map_get_key_value'”
<rekado>do I need to use a different version of Rust?
<efraim>you could try using a newer version, but everything depending on it will likely need the newer version too
<efraim>what package is it?
<rekado>I need it for swc
<efraim>if it fails during build I'd just skip build for now and see if you can ignore it
<efraim>if you do need a later rust version it's like the other language ones #:rust ,rust-1.42 or whatever
<rekado>okay, thanks
<brendyyn>rekado: great to see you having a go at swc. it looked kinda daunting. does it require adding many rust packages to guix?
<brendyyn>if you didn't know already, there is an update to the cargo importer that does the versioning automatically here you could make use of and test
<matijja>Hello Guix!
<mroh>Hey matijja!
<brendyyn>looks like map_get_key_value was stabalised in rust 1.40.
<matijja>mroh: Hi!
<matijja>I did upgrade my Archlinux, but still getting following error when invoking guix command (
<matijja>I get this error only with my user. If I run guix with sudo it works fine.
<matijja>Any clue, where to start?
<vits-test>matijja: Hello. Do You have a nscd running on Arch?
<vits-test>, 2.6.2 "Name Service Switch"
<matijja>vits-test: nscd service is disabled
<vits-test>matijja: try enable --now it?
<matijja>I started it with SystemD, it creates /var/run/nscd/socket but the error remains.
<zjgkkn>Hi! I've a 24 core old server that plan to use as substitute server for laptop. Is there a way to organize loop with pull and build steps for certain packages?
<leoprikler>zjgkkn: mcron?
<leoprikler>or do you rather want something like
<leoprikler>you probably want to look at cuirass in that case
<vits-test>matijja: so, `guix` works for root, didn't work for user ("guile: symbol lookup error: /usr/lib/ undefined symbol:"), the nscd daemon running. Is guix working for a new user?
<vits-test>"useradd .. test"
<zjgkkn>leoprikler: Yes, this is what i need (and missed in the manual).
<matijja>vits-test: It works for new users.
<vits-test>matijja: Cool, so setup is correct.
<vits-test>matijja: Try remove a symlink ~/.guix-profile, and pull again?
<vits-test>The items in store will be not affected.
<matijja>I don't have ~/.guix-profile folder.
<matijja>I delete it few days ago when I tried to reinstall whole Guix.
<pinoaffe>hi guix! I've been using for a while now, I'm currently trying to install it on an additional machine, decided to try out the graphical installer but it consistently loops back to selecting locale and the like after I select partitioning details - is this a known bug?
<sneek>Welcome back pinoaffe, you have 1 message!
<sneek>pinoaffe, apteryx says: your autossh-service works well! Thank you for writing it. It was my dynamic DNS that had stopped working. :-)
<pinoaffe>apteryx: thanks!
<vits-test>matijja: what about ~/.config/guix/current ?
<raghavgururajan>leoprikler, rekado: Thanks!
***jonsger1 is now known as jonsger
<matijja>vits-test: Same. I don't have ~/.config/guix
<vits-test>matijja: `which guix`?
<matijja>vits-test: /usr/local/bin/guix -> /var/guix/profiles/per-user/root/current-guix/bin/guix -> /gnu/store/5jcl8636hps1pm8vqfkjjk3rmknv9rhk-guix-command
<vits-test>matijja: try /usr/bin/guix pull?
<matijja>I don't have /usr/bin/guix ...
<vits-test>oh, /usr/local/bin/guix. Which is `which guix`.
<matijja>Same error.
<vits-test>matijja: and for root/another user `which guix` is?
<matijja>For root is: /root/.config/guix/current/bin/guix -> /gnu/store/5jcl8636hps1pm8vqfkjjk3rmknv9rhk-guix-command
<matijja>And for test is again: /usr/local/bin/guix
<roptat>you should use "type guix" rather than "which", because type takes hashing into account
<roptat>(but in this case it doesn't seem to be important, justa general remark :))
*vits-test eww..
*raghavgururajan is might pissed at the message "Oh no! Something has gone wrong." message in GNOME
<matijja>vits-test: I found it. I have `export LD_LIBRARY_PATH=/usr/lib/:/usr/local/lib/` somewhere in /etc/profile.d dir, for some reason.
<matijja>vits-test: Thank you for your help.
<leoprikler>Lucky am I for I have only ever seen that once
<raingloom>coqide is borked. it's looking for but the executable's name is coqide. not sure if that's related to the way it's configured on guix.
<vits-test>matijja: Thanks. "LD_.." noted.
<alextee[m]>does faust fail to build for everyone?
<lukashevich>Hello everyone! :) I was wondering if I could ask some short Lisp-related questions here once in a while. Or, maybe someone knows of some mailing lists or something. Thank you!
***jonsger1 is now known as jonsger
<bavier[m]1>lukashevich: short scheme-related questions might be fine here, but #guile might be better for wider discussion.
<lukashevich>Oh, thank you, bavier! :)
<raingloom>alextee[m]: i tried building it, looks like it's playing fast and loose with some function pointer conversions and GCC doesn't like that. -fpermissive would allow it to build, according to the error message.
<alextee[m]>raingloom: do you just append that to CFLAGS?
<joshuaBPMan>Hello #guix! You lovely genius-es you!
***jonsger1 is now known as jonsger
<joshuaBPMan>odd question: I've got let's encrypt set up on two websites. with the let's encrypted service. I'm tempted to set up auto-redirect to https as well as setting the strict transport security header on both HTTP and this going to cause let's encrypt difficulty to verify my does let's encrypt need to view my site in http
<joshuaBPMan>mode in order to update a cert?.
<joshuaBPMan>Oh, also I just realized that you only need to set Strict-Transport-Security HTTP header on the https version of your site, The client ignores that header when communicating in http. :)
<str1ngs>joshuaBPMan lets encrypt only varies the DNS record matches the ip address you are creating the cert for.
<str1ngs>joshuaBPMan: it could be a problem if you let your certs expire and how you use certbot though.
<str1ngs>just don't let them expire :)
<joshuaBPMan>str1ngs thanks! I guess the certbot service that guix system provides, will make sure my certs don't expire. It is pretty slick how ya'll set that up. :)
<dissoc3>is there a way to save a package build output if the build succeeds? i know you can pass -K for a failure but I want to keep if succeeds
<dissoc3>i guess i can return a #f of the last stage
<vagrantc>guix build --log-file
<vagrantc>i think
<vagrantc>dissoc3: ^^
<vagrantc>something like that