IRC channel logs


back to list of logs

<jeko>someone knows how to make guix deploy verbose ?
<leoprikler>I assume you've already tried --verbosity?
<jeko>yep, with level 2 and 9 haha
<jeko>I think there is something wrong. I just got ;;; [2020/12/02 00:15:48.683027, 0] read_from_channel_port: [GSSH ERROR] Error reading from the channel: #<unknown channel (freed) 7f3ca3ef80a0>
<PotentialUser-20>how can I specify package outputs in config.scm? I want to install `rust:cargo` but if I write that in config.scm I get an error (using `guix install rust:cargo` works just fine)
<leoprikler>PotentialUser-20: you mean manifest.scm? (list package output)
<PotentialUser-20>sorry I'm very new to guix and don't know the terminology etc. yet. I mean the file at `/etc/config.scm`
<leoprikler>Ahh, okay, your OS config.scm
<mbakke>PotentialUser-20: (packages (append (list (list rust "cargo")) %base-packages))
<PotentialUser-20>Thanks it works now :)
<PotentialUser-20>btw do I have to also add `rust` seperately or is that automatically installed?
<mbakke>PotentialUser-20: I think you need to add 'rust' separately indeed
<PotentialUser-20>Ok thanks
<leoprikler>so (list rust (list rust "cargo")) it is :)
<mbakke>in general I recommend keeping the "system package" set minimal though, YMMV :-)
<leoprikler>Yep, better add yourself some (service profile-service-type ...) instead
<newUser>Hello, I just installed guix using the graphical interface few minutes ago, I have a question
<newUser>I just used "guix pulll" and "sudo guix system reconfigure /etc/config.scm" Now im just waiting for it finish, but I was reading a bit on managing packages
<newUser>The getting started pages says it will show this hint after installing some package "GUIX_PROFILE="$HOME/.guix-profile"
<newUser>But this file .guix-profile dont even exist in the my home at the moment
<leoprikler>because you only ran `guix pull` until now
<leoprikler>this message will pop up once you run `guix package` (or any of its aliases)
<leoprikler>it is mostly harmless, but it warns you that stuff might not be as expected when you keep using the shell in which you ran the install (rather than spawning a new one) without running the commands it proposes
<newUser>But I will need to add those lines setting GUIX_PROFILE in my .bash_profile or it is not needed when using Guix System?
<leoprikler>No, it is not needed on Guix System to add those lines to your .bash_profile
<leoprikler>Guix System already executes them in bash or zsh as part of /etc/profile
<newUser>Oh! so that hint is intended for people using guix as package manager out of Guix System, right?
<leoprikler>However, currently running shells (e.g. the one you just used to run the command) will not be up to date if this warning pops up.
<leoprikler>That, plus it reminds those on Guix system, that they have to refresh their environment variables every now and then.
<newUser>when using Guix System something like `. /etc/profile` will sufice?
<leoprikler>That's a bit overkill, but yeah.
<newUser>Thanks leoprikler
<save-lisp-or-die>Is it normal for the 'check' phase during a package install to be cpu intensive and take a while? Or is the process stuck?
<lfam>save-lisp-or-die: It completely depends on the package in question. We run the upstream test suite and some of them are quite expensive. Which package is it?
<lfam>If you'd like more detail, and willing to start that package build over, you can stop it and add '--verbosity=9' to the command. Then you will see the output of the compilation
<save-lisp-or-die>the gnu smalltalk package. `smalltalk`
<save-lisp-or-die>yeah I think I will thanks
<lfam>Ideally, you wouldn't have to build it yourself — there'd already be a "substitute". But Guix wants to build this package for me, too
*lfam tries it
<save-lisp-or-die>ah looks like there's a regression test suite
<save-lisp-or-die>that's where all the action is
<save-lisp-or-die>hmm.. maybe it *is* getting stuck during the build (if such a thing is possible). Just after one of the regression tests fail, nothing more seems to be happening and CPU kicks up. No further progress is reported.
<lfam>Hm, usually things should end when a test fails
<lfam>Well, it's dinnertime here, so someone else will have to take over helping. But I'll be back later
<save-lisp-or-die>no worries thanks much.
<save-lisp-or-die>(I was just exploring anyway)
<lfam>Searching our Git log for "smalltalk" it seems that package doesn't see much activity. I fear it might not be in a good state
<save-lisp-or-die>sounds plausible
<mbakke>indeed, 'guix weather smalltalk' does not look great
<save-lisp-or-die>what social/technical processes are used to keep packages healthy?
<save-lisp-or-die>(i.e. what could I do to help?)
<save-lisp-or-die>(I'll read the manuals)
<mbakke>save-lisp-or-die: that's a good question :-) for build failures, filing a bug report is a good first step.
<mbakke>more generally, we don't have enough people watching out for the red builds on the CI:
<save-lisp-or-die>ah cool. the ci is a good answer
<mbakke>and of course we always need more people submitting patches :-)
*raghavgururajan 's head spins
<kozo[m]>Hey all, how do I discover more about News for channel 'guix'
<kozo[m]> Local substitute servers discovery is now supported?
<kozo[m]>save-lisp-or-die: Hey, how did you make out with sdl2?
<save-lisp-or-die>well enough. I ended up manually loading sdl2's shared lib before loading the lisp system. basically as you suggested.
<save-lisp-or-die>I tried a few things and just did the stupidest, most direct one.
<save-lisp-or-die>🤷 whatever works :)
<mroh>one of my favourite woody allen movies ;)
<mbakke>kozo: there is a some information about it in (info "(guix) Invoking guix-daemon"), but arguably not a lot to go on
<mbakke>essentially you start guix-publish on remote nodes with the --advertise option, and guix-daemon with --discover, and your daemon will discover the other machines automatically
<mbakke>as long as they are on the same LAN
<kozo[m]>mbakke: That's a really nice addition. No more hardcoding them in.
<mbakke>agreed :)
<kozo[m]>Thank you
<kozo[m]>Are those options we can add to our service configurations?
<kozo[m]>Scrath that, I see it in the help
<apteryx>mbakke: in chromium, I just found out about chrome://flags, it's interesting
<apteryx>there are entries for WebRTC hardware video decoding/encoding, but of course 'not available on your platform' :-)
<apteryx>mbakke: the blacklisting of nouveau dates back to 2011... lol
<raghavgururajan>apteryx: Seems like it can only be enabled on Chrome OS and Android.
<raghavgururajan>Adreno GPUs?
<charles`>for creating a system image, where is gnu/system/install.scm
<charles`>supposed to be?
<apteryx>in the Guix source tree
<charles`>does that mean /gnu/store/blah1081273640-guix-system-source/gnu/system/install.scm?
<apteryx>there's more than one way to get the sources of guix; one is 'guix build --source guix', which will get the latest packaged guix sources; another way is to clone the Guix git repository from Savannah
<apteryx>I recommend the later, as it will make hacking on Guix (such as contributing package definitions) a breeze after you've set it up.
<charles`>so source tree is literally the source code for guix including the package definitions?
<ryanprior>charles`: yes, the package definitions are part of the Guix source.
***sturm is now known as Guest42291
<guix-vits>hello; it's ok for `adb` running in `guix environment` to work only if adb-daemon started as root? is `environment` cannot use suid stuff? thanks.
<guix-vits>no, i'm wrong. it's same on this box. whatever.
<guix-vits>i sorry for noise, on laptop all worked. i'll check difference later maybe.
<charles`>When I try to boot from the system image, I get mount error: permission denied. Any ideas?
<guix-vits>idk; system.scm?
<charles`>the stacktrace shows gnu/build/linux-boot.scm
<charles`>ice-9/boot-9.scm in procedure mount
<charles`>is sddm the only way to use wayland with gnome on guix?
<raghavgururajan>> charles`‎: is sddm the only way to use wayland with gnome on guix?
<raghavgururajan>Currently, yes. I am working on making gdm to work with wayland But gonna take a while.
<charles`>raghavgururajan Is there anyway I can help or is it a one person job?
<raghavgururajan>charles`: Thanks! The work is continuation of my outreachy internship. The branch is wip-desktop.
<raghavgururajan>charles`: There are several changes made on that branch. If you could pull, bootstrap and test via VM, it would be great.
<charles`>Is it just a matter of configuring the gdm package or is there more to it?
<raghavgururajan>More to it. The packages that gdm depends on, were also changed.
<raghavgururajan>Right now, the gdm crashes upon start. If you could find out what's causing the crash, it would be helpful.
<charles`>raghavgururajan I will do my best
<txgvnn>Hello Guix! I wonder how to effective manage the same command of packages.
<txgvnn>Example I install netcat and netcat-openbsd. Both package provide `nc` command
<raghavgururajan>charles`: Thanks!
***chrislck_ is now known as chrislck
<guix-vits>txgvnn: we can install them in separate profile, then make aliases in .bashrc: alias nc='~/guix-profile/bin/nc'; alias nc-bsd='~/guix-alt-profile/bin/nc' ?
<guix-vits>or .bashrc may search thru PATH, and make a variable for each dir, and see if any duplicates found, and make aliases automatically... idk how to do that. anyway, i'll be run every time bash invoked, probably not worth that for 1-2 duplicates.
<lle-bout>it seems vim on master fails to build: - failing test
<guix-vits>"c'est emacs desirrer toi"
<lle-bout>guix-vits: I'm too bad at it. I feel helpless.
<lle-bout>Reported upstream:
<charles`>lle-bout did you try evil mode?
<jackhill>I have some software that really wants to use python instead of python3. Is there an easy way to get that in a guix environment?
<lle-bout>charles`: vim is just my go-to shell-based editor, I'm otherwise more fond of fully graphical things like Visual Studio Code or IntelliJ IDEA
<lle-bout>charles`: I tried evil mode and all, spacemacs, doom emacs, it just doesnt feel right.
<lle-bout>I wish I knew though, but I don't. I feel like this could make me so much more productive writing Scheme but it doesnt sink in.
<guix-vits>jackhill: is `environment --ad-hoc python-2` not working?
<lle-bout>I honestly just need something that works and has great completion and discovery features. It doesnt look like Scheme is well supported anywhere but in Emacs and even then I find the Scheme completion and code-assist in general subpar compared to what it could be. I am all-in for something like Guile Studio similar to DrRacket.
<charles`>guix-vits would that environment still have python 3 if it was installed outside the environment?
<guix-vits>charles`: idk totally, but there also --pure flag if p2 not shadows out p3.
<lle-bout>I don't have the bandwidth to think about customizing my editor too heavily, I'm all about out of the box experiences.
<jackhill>guix-vits: I didn't look, but I have a script that wants to invoke python3 as python.
<charles`>lle-bout when you are using Emacs for scheme are you using geiser with company? they only take installing, no real config needed
<guix-vits>jackhill: fair enough, as p2 is obsolete (muhaahhahah)
<lle-bout>charles`: yes I tried those, guess I'll try again but I always ended up being frustrated and unproductive
<guix-vits>+1, but erc (irc client) is great.
<charles`>erc is the only irc client I can figure out
<jackhill>It also wants /bin/bash. I edited that, but I have a feeling that this script doesn't want ot be run on systems that are not the author's :)
<charles`>lle-bout I would be willing to help you tomorrow maybe
<guix-vits>jackhill: red alert, heh.
<lle-bout>charles`: That would be sweet!
<raghavgururajan>Folks! How to selectively *apply* particular part of a patch?
<HalPMe>Hello, when i use a "fuix search icecat" the return just shows the icecat 78.5-0 preview, how its possible to install the last stable release using guix?
<HalPMe>guix search...*
<raghavgururajan>I have a patch that modifies one file, creates one file and deletes one file. The apply of patch fails as the first couldn't apply. How do I selectively apply part of the patch that creates a file?
<raghavgururajan>HalPMe: [1] `guix pull` [2] `guix install icecat`. This will install the latest available version.
<HalPMe>I did this way, the last available version is 78.5.0-guix0-preview1
<HalPMe>but its somekind broked
<raghavgururajan>HalPMe: Broke how? Doesn't run or doesn't install?
<HalPMe>It install but when it run it dont display numbers inside the browser
<charles`>raghavgururajan I think I was able to bootstrap, how do I generate an iso to try on a vm?
<raghavgururajan>HalPMe: Numbers as in fonr?
<HalPMe>But in all cases I should install the last stable release
<HalPMe>which was 69.7
<raghavgururajan>HalPMe: You might need to install font-dejavu
<raghavgururajan>charles`, just a sec
<guix-vits>HalPMe: i remember icecat was once dropped from gentoo due to security issues (wasn't it?). i doubt that icecat team has power to maintain stable releases. can they?
<guix-vits>btw, it can be wise to install web browser in a separate profile.. not working? just roll it back.
<HalPMe> the last release here point to
<charles`>raghavgururajan I'll be at it again tomorrow
<raghavgururajan>charles`: [1] ./pre-inst-env guix system vm my-config.scm [2] /gnu/store/… -m 1024 -smp 2 -net user,model=virtio-net-pci
<raghavgururajan>charles`: Cool!
<raghavgururajan>charles`, Not sure about the version mentioned on that link. May be bandali can help you with that.
<raghavgururajan>Oh wait! I meant that message for HalPMe
<raghavgururajan>bandali ^
<txgvnn>guix-vits: Thanks. I want use only a profile
<txgvnn>I hope guix should have a alias alias as update-alternatives. Or something that I can alias nc (default) -> netcat, nc.gnu -> netcat, nc.bsd -> netcat-openbsd
<guix-vits>sneek: seen ani?
<guix-vits>sneek: die dick botsnack
<leoprikler>txgvnn: I think the manual has an example of setting emacs to be vi ;)
<leoprikler>[i.e. the vi command spawns emacs, not the other way round]
***apteryx_ is now known as apteryx
<kisaja[m]>never mind, i asked why kaddressbook doesn't have all features, it turns out it doesn't have a link to any kind of source code on official site
<leoprikler>Do shepherd patches belong to guix-patches or do they get their own ML?
<leoprikler>and the subject prefix is PATCH SHEPHERD?
<efraim>yeah, then everyone knows it's for shepherd
<leoprikler>45004 and rising
<leoprikler>A suggestion for 26877: Could we check in Guix whether fonts changed between the current and previous generation and suggest to run `fc-cache -rv` afterwards?
<efraim>or add to .bash_profile 'if [ $(which fc-cache) ]; then fc-cache -frv fi'
<efraim>or add to .bash_profile 'if [ $(which fc-cache 2>/dev/null) ]; then fc-cache -frv fi'
<efraim>or add to .bash_profile 'if [ $(which fc-cache 2>/dev/null) ]; then fc-cache -frv; fi'
<efraim>I think I'm done editing it:)
***holger is now known as Guest69514
<leoprikler>oof not too sure about running that on each login, but okay
<efraim>add an '&' to the end, then it's not blocking
<leoprikler>that's not the point, but yeah, a problem as well
<efraim>If I'm reading Debian's configs correctly it monitors 3 directories, and when they update it runs fc-cache -sfv in the background
<leoprikler>I don't think that's possible in Guix, you'd refresh the cache whenever .guix-profile changes.
<leoprikler>Which imo is a more reasonable approach than doing it on each login, but still a bit overkill
***sturm is now known as Guest36873
<kisaja[m]>can i get 'guix pull' and 'guix packagr -u' to skip 'network issue' and 'cant build this' it would be awesome if network issue would wait for me to fix internet and 'can't build' to skip that tree of packages, i don't even want to read log about some packages, for example anything from KDE, as i can live without it. have problems with kdenlive and audacity, tried removing them and starting 'packagr -u' again, and not a single
<kisaja[m]>build or download stays, everything starts from the ground up, i mean atleast could i have an option not to download same 20G of packages, newbie ofcourse, in a way i know an answer but still asking pros
<leoprikler>kisaja[m]: maybe not easily, but --do-not-upgrade=kde...
<jeko>Hey! How are you, Guixters?
<roptat>good news for ocaml users: I've contacted my colleague who was working on bootstrapping ocaml: we're going to finish it, the plan is to get it done by the end of the year/start of january
<roptat>we should be able to bootstrap ocaml4.07 with that, then we'll find a way to reuse that to build future versions
<OriansJ>sweet news roptat
<PotentialUser-52>I just issued an update to 1.2 and it seems to be rebuilding everything, none of the bigger packages come from a build server and so after about 20 hours it is still building things. What is the criteria for a package to be build by the build server ?
<PotentialUser-52> I guess it just cannot keep up with the commits .. there is also an awful lot of red there :|
<PotentialUser-52>Or I misunderstand something.
<mothacehe> PotentialUser-52: The CI is indeed lagging behind, but for the 1.2.0 the substitutes coverage should be acceptable >80%.
<mothacehe>Having everything built from source probably indicates an anomaly
<PotentialUser-52>Yea, it should, but it isn't :D
<mothacehe>Are you using Guix System or a foreign distro ?
<mothacehe>mmh strange
<PotentialUser-52>it actually just finished
<mothacehe>you can run "guix weather" to check substitutes coverage
<PotentialUser-52>it did not rebuild glibc so that is something :) it had to rebuild dovecot, all the browsers, libreoffice
<mothacehe>for instance "guix weather icecat" indicates there's an available substitute for me on 0216fc2 commit
<PotentialUser-52>i did not lock in the 1.2 commit specifically in channels.scm so perhaps thats my problem
<PotentialUser-52>thats pretty cool, good to know :)
<mothacehe>You can also use something like: "guix time-machine --branch=version-1.2.0 -- guix weather icecat" to look around
<PotentialUser-52>x86_64-linux: 46 (4.6%)
<PotentialUser-52>should have been more specific in channels.scm .. next time :)
<mothacehe>4.6% is the percentage of queued builds
<mothacehe>the interesting part is more: 100.0% substitutes available (1 out of 1)
<PotentialUser-52>aye 62% are available
<mothacehe>not great :(
<mothacehe>could you share the result of "guix describe"?
<PotentialUser-52>commit: e4c1818e1b614567b80e577f72a3c3cf3d44fc18
<PotentialUser-52>branch master
<mothacehe>oh yes, that's yesterday, pinning to an older commit should help
<PotentialUser-52>yea figured :) i did a pull and then reconfigure right away
<mothacehe>hopefully the CI situation will improve soon :)
<mothacehe>(I have 93.5% coverage on 0216fc2)
<PotentialUser-52>oh well .. it was a pebcac
<PotentialUser-52>i suppose it could be improved by publishing after a certain % treshold
<guix-vits>PotentialUser-52: problem exist between crossbow and comraden?
<PotentialUser-52>so that whatever guix pull is pulling is lagging behind a few days so the build server can catch up
<PotentialUser-52>yes exactly ! :D
<guix-vits>jeko: we're fine, our crossbows too.
<jeko>guix-vits: cool
<HalPMe>the default version of Linux Libre used on guix dont support netfilter?
<mothacehe>PotentialUser-52: yes that could be a way!
<rekado_>This command fails: guix time-machine --commit=a71d769d1e230463a043e12a10a557fb0c0aa8c4 -- package -p /tmp/r-test/.guix-profile -i r-minimal r-genomation
<rekado_>it says ‘error: %guix-register-program: unbound variable’
<rekado_>I suppose this means I cannot use the time-machine that far back
<rekado_>(this is like Primer)
<jeko>is there a parameter to create compressed disk-image using `guix system disk-image` ? Or do I have to compress it manually afterward?
<rekado_>I wonder if we could print a nicer error message in case ‘time-machine’ is used to travel too far into the past before the first time machine was turned on.
<mothacehe>jeko: you can produce qcow2 images which are compressed
<mothacehe>but there's nothing to produce a .raw.gz or so
<janneke>rekado_: somewhere march 2019, right?
<mothacehe>you can run "guix system --list-image-types" for the complete list
<janneke>the farthest back it can travel
<rekado_>janneke: yes, something like that
*rekado_ tries 152cbc3ca588be8a0db8bd1c9a9746068542e449
<jeko>mothacehe: hi ! Ok thank you qemu master :p
<mothacehe>jeko: adding a "compressed-raw" format would make sense though
<HalPMe>when i install iptables or nftables
<HalPMe>both dont present the utility to setup firewall rules
<jeko>mothacehe: as far as I am concerned, I am fine with the qcow2 😊️
<HalPMe>Some tips on how is possible to setup a firewall on Guix?
<jeko>mothacehe: Did you tried `guix deploy` to digitalOcean with for your blog?
<jeko>mothacehe: Oh I re-read the article and I see you are not really interested into the actual process.
<mothacehe>jeko: yes that's because I didn't like how deploying to DigitalOcean worked (installing Ubuntu then Guix System on top).
<mothacehe>I prefer to deploy a Guix System disk-image directly
<mothacehe>and possibly update it using deploy over SSH
<mothacehe>deploy for DigitalOcean could also be tweaked to do what I proposed in the blog under the hood
<jeko>mothacehe: yeah I like the idea too…
<jeko>mothacehe: you mean uploading a custom image then creating a droplet using it ?
<mothacehe>yes exactly
<mothacehe>that would be somehow cleaner I guess
<mothacehe>but more expensive are custom images are not free :p
<jeko>yeah I just created a DO space to upload a bigger disk-image haha
<jeko>I think it can be done once then get rid of the space
<jeko>however I am not sure I will go for a 25Go… haha
<dannym>Is anyone using guix system disk-image on armhf-linux successfully?
<roptat>I have an armhf machine, but it's been running guix system reconfigure for 2 days now, so I don't really want to try
<roptat>are you having a specific issue?
<dannym>qemu-minimal fails tests
<roptat>oh, we should fix that :/
<dannym>I have the build directory, and it said that it dumped test logs somewhere, but not where
<dannym>Definitely should fix it, especially since it's not possible to remove qemu-minimal from the closure of guix system disk-image (I tried).
<roptat>it should be in /var/log/guix/drvs/
<roptat>but there are many logs in there
<dannym>No, I mean the build of qemu didn't print the logs on screen but in an extra file, which it didn't tell me the name of
<roptat>guix should have told you where it was : Vous trouverez le journal de compilation dans « /var/log/guix/drvs/... »
<dannym>(First qemu-minimal gets pulled in by grub-efi, but only if qemu-minimal's supported systems contain the current system)
<roptat>right after the red line
<dannym>(But why is grub-efi used in the first place? That's not an EFI system)
<roptat>isn't the disk-image an efi system?
<dannym>No, armhf-linux does not support efi (most of the time)
<dannym>They use u-boot
<dannym>And I did specify:
<dannym> (bootloader (bootloader-configuration
<dannym> (bootloader u-boot-novena-bootloader)
<dannym> (target "/dev/mmcblk0")))
<dannym>the build of qemu = the actual makefiles
<dannym>I don't mean guix
<roptat>er... don't you build qemu with guix?
<dannym>I mean that qemu makefiles dump the test logs into extra files in the working copy of the build
<dannym>That guix does not see
<roptat>ah I see
<dannym>I have this working copy because I built with "-K".
<roptat>you can re-run the build with -K to keep the files around
<dannym>I have it
<roptat>ah good :)
<dannym>But what is the name of the file?
<dannym>In the working copy
<dannym>I have those ".log":
<roptat>maybe it's not a .log file
<roptat>have to go, see you later
<dannym>Thanks :)
***chrislck_ is now known as chrislck
<simonsouth>dannym: Not sure if it's relevant but I'm also working on debugging a test failure in qemu-minimal on aarch64 right now.
<simonsouth>Thought it was because I was missing the virtio modules in my kernel, but adding those hasn't fixed the problem.
<simonsouth>One of the tests is failing with: kvm_arm_vcpu_init failed: Invalid argument
<simonsouth>Wonder if you'll find the same when you look at your logs.
<lispmacs>had anybody here played the bsd-games trek game? the "restart" command seems implies that there is a way to save your game in progress, but I can't figure out the command to save the game
<jeko>Do I have to add nginx in the packages list if I use nginx-service-type in my guix system declaration ?
<jeko>or is it implicitly installed ?
<roptat>jeko, it's not installed on the command line, but it's enough to declare it in your list of services for the shepherd to be able to run it
<dannym>MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} QTEST_QEMU_BINARY=alpha-softmmu/qemu-system-alpha QTEST_QEMU_IMG=qemu-img tests/qtest/boot-serial-test -m=quick -k --tap < /dev/null | ./scripts/ --test-name="boot-serial-test"
<dannym>Broken pipe
<dannym>tests/qtest/libqtest.c:175: kill_qemu() detected QEMU death from signal 4 (Illegal instruction) (core dumped)
<dannym>ERROR boot-serial-test - too few tests run (expected 1, got 0)
<dannym>make: *** [/tmp/guix-build-qemu-minimal-5.1.0.drv-0/qemu-5.1.0/tests/Makefile.include:650: check-qtest-alpha] Error 1
<dannym>make: *** Waiting for unfinished jobs....
<dannym>PASS 7 qom-test /aarch64/qom/musca-b1
<simonsouth>dannym: "alpha-softmmu/qemu-system-alpha": Shouldn't that be "arm-softmmu/qemu-system-arm"?
<jeko>roptat: hi! sorry I don't understand if I need (packages (cons* screen git nss-certs >>nginx<< %base-packages)) in my system declaration
<simonsouth>Thought you said you were building for armhf.
<dannym>I am
<roptat>jeko, if you only want to have the server running, no you don't
<dannym>It's the qemu target list
<dannym>Per default qemu builds all the targets
<simonsouth>Oh, right. qemu-minimal does not.
<roptat>jeko, it's only useful if you want to call nginx manually from the CLI
<jeko>roptat: ok cool! I thinks that's all I want. Thankx!
<dannym>simonsouth: It doesn't? That's from qemu-minimal O_o
<simonsouth>dannym: Ah...
<simonsouth>It should be building for you only for arm, I believe.
<simonsouth>See gnu/packages/virtualization.scm:337
<dannym>from the log:
<dannym>target list aarch64-softmmu alpha-softmmu arm-softmmu avr-softmmu cris-softmmu hppa-softmmu i386-softmmu lm32-softmmu m68k-softmmu microblazeel-softmmu m
<dannym>icroblaze-softmmu mips64el-softmmu mips64-softmmu mipsel-softmmu mips-softmmu moxie-softmmu nios2-softmmu or1k-softmmu ppc64-softmmu ppc-softmmu riscv32-softmm
<dannym>u riscv64-softmmu rx-softmmu s390x-softmmu sh4eb-softmmu sh4-softmmu sparc64-softmmu sparc-softmmu tricore-softmmu unicore32-softmmu x86_64-softmmu xtensaeb-so
<dannym>ftmmu xtensa-softmmu aarch64_be-linux-user aarch64-linux-user alpha-linux-user armeb-linux-user arm-linux-user cris-linux-user hppa-linux-user i386-linux-user
<dannym>m68k-linux-user microblazeel-linux-user microblaze-linux-user mips64el-linux-user mips64-linux-user mipsel-linux-user mips-linux-user mipsn32el-linux-user mips
<dannym>n32-linux-user nios2-linux-user or1k-linux-user ppc64abi32-linux-user ppc64le-linux-user ppc64-linux-user ppc-linux-user riscv32-linux-user riscv64-linux-user
<dannym>s390x-linux-user sh4eb-linux-user sh4-linux-user sparc32plus-linux-user sparc64-linux-user sparc-linux-user tilegx-linux-user x86_64-linux-user xtensaeb-linux-
<dannym>log of qemu-minimal
<dannym>It's because the match in gnu/packages/virtualization.scm:337 should be "armhf", not "arm".
<simonsouth>dannym: Ah, that makes sense.
<dannym>But that means where were no disk images for armhf built since May 2020, which is when this was changed :(
<dannym>I'm pretty sure that most of the others are wrong, too
<dannym>Because the old (correct) code in e39b857d7461e32cf737c97f47b522f9acc6f877 used string-prefix
<dannym>efraim: ??
<dannym>The commit that introduced the bugs is a7e6ec18d7b
<dannym>and doesn't use string-prefix but exact match--without there being more matches
<dannym>That is very unlikely to be correct
<dannym>Also, the match is on (or (%current-target-system) (%current-system)) but the former is a triple and the latter a nix system. But there's no gnu-triplet->nix-system anywhere to be seen.
<efraim>I did what now? oh, did I mess that up :(
<bavier[m]>anyone here familiar with how gcc creates it's release tarballs?
<efraim>dannym: yeah, arm was probably supposed to catch armhf, which is also why powerpc64 was before powerpc, and powerpc64 to catch powerpc64le. looks like I meant string-prefix.
<dannym>It was supposed to catch arm, armhf-linux, arm-linux-gnueabihf, armv5tel-linux-gnueabi and so on
<dannym>and so on
<dannym>There's also a helper function target-arm32? and that uses, predictably, string-prefix
<dannym>other example: arm-none-eabi
<dannym>And for powerpc:
<dannym>./gnu/packages/cross-base.scm: ,@(if (equal? "powerpc64le-linux-gnu" target)
<dannym>./gnu/packages/image.scm: ((string-prefix? "powerpc" target)
<dannym>./gnu/packages/libffi.scm: ,@(if (string-prefix? "powerpc-" (or (%current-target-system)
<dannym>./gnu/packages/libffi.scm: (if (string-prefix? "powerpc-" (or (%current-target-system)
<dannym>./gnu/packages/linux.scm: "Some systems (notably powerpc-linux) require a special target for kernel
<dannym>./gnu/packages/patches/libffi-3.3-powerpc-fixes.patch:src/powerpc/ffi_powerpc.h:65:9: error: '__int128' is not supported on this target
<dannym>./gnu/packages/tls.scm: ((string-prefix? "powerpc64le" target)
<dannym>./gnu/packages/tls.scm: ((string-prefix? "powerpc64" target)
<dannym>./gnu/packages/tls.scm: ((string-prefix? "powerpc" target)
<dannym>./gnu/packages/valgrind.scm: `(,@(if (string-prefix? "powerpc" (or (%current-target-system)
<dannym>./NEWS:*** New cross-compilation targets: aarch64-linux-gnu, powerpc-linux-gnu
<dannym>./tests/packages.scm: (parameterize ((%current-target-system "powerpc64le-linux-gnu")
<dannym>And for mips,
<dannym>./doc/guix.zh_CN.texi: #:target "mips64el-linux-gnu")
<dannym>./gnu/ci.scm: (string-contains target "64"))) ;x86_64, mips64el, aarch64, etc.
<dannym>./gnu/packages/emulators.scm:MIPS target, targeting mips-linux-gnu or mips-elf.")
<dannym>./gnu/packages/gcc.scm: (cond ((string-match "^mips64el.*gnuabin?64$" target)
<dannym>./gnu/packages/maths.scm: ;; Each architecture has its own make target, and there is none for mips.
<dannym>./gnu/packages/tls.scm: ((string-prefix? "mips64el" target)
<dannym>./gnu/packages/video.scm: '("--target=mips3-linux"))
<dannym>./gnu/packages/virtualization.scm: '(list (string-append "--target-list=mips-softmmu,mipsel-softmmu,"
<dannym>./gnu/packages/virtualization.scm: '(list "--target-list=mips-softmmu,mipsel-softmmu"))
<dannym>./gnu/system.scm: ((string-prefix? "mips" target) "vmlinuz")
<dannym>./po/doc/guix-manual.zh_CN.po:msgid "Cross-build for @var{triplet}, which must be a valid GNU triplet, such as @code{\"mips64el-linux-gnu\"} (@pxref{Specifying target triplets, GNU configuration triplets,, autoconf, Autoconf})."
<dannym>./po/doc/guix-manual.zh_CN.po:msgid "@var{target} must be a valid GNU triplet denoting the target hardware and operating system, such as @code{\"mips64el-linux-gnu\"} (@pxref{Specifying Target Triplets,,, autoconf, Autoconf})."
<dannym>./po/doc/guix-manual.zh_CN.po:" #:target \"mips64el-linux-gnu\")\n"
<dannym>./tests/gexp.scm: (let* ((target "mips64el-linux")
<dannym>./tests/gexp.scm: (mlet* %store-monad ((target -> "mips64el-linux")
<roptat>can you not spam please?
<dannym>./tests/gexp.scm: (target "mips64el-linux")
<dannym>./tests/ build coreutils --target=mips64el-linux-gnu --dry-run --no-substitutes
<roptat>you should use a paste service, like
<efraim>I can test out changing it to 'string-prefix?' but it takes a while to test the various combinations
<dannym>efraim: Sure, no problem. I worked around it anyway...
<leoprikler>bavier[m]: is it not make dist?
<bavier[m]>leoprikler: no :/
<dannym>roptat: Your last statement right there is why slack exists and is not "just IRC" ;)
<dannym>qemu-minimal with the s;arm;armhf; just built fine on armhf, and now it's building grub-efi for no good reason
<leoprikler>dunno, you can configure at least some irc clients to use a paste service when doing huge-ass pastes instead of spamming line for line
<leoprikler>Polari offers at >= 5 lines
<leoprikler>(value found through reverse engineering, not by looking at the code)
<dannym>leoprikler: What problem does that solve? If you then search in Polari later, or in the irc logs of the channel, does that include those >= 5 lines?
<dannym>Also, what problem does that solve in the first place ? (People don't have dial-up internet anymore, so it's not because of the traffic)
<leoprikler>Well, no, but do you need those >= 5 lines archived for all eternity?
<leoprikler>Chances are no.
<dannym>The ones above ? Most definitely. In fact, I'd add a comment with those lines to the fixed source code
<dannym>Otherwise someone else will break it in exactly the same way again
<dannym>(which is completely understandable)
<leoprikler>I'm pretty sure one can find an easier digestible comment.
<kisaja[m]>wait, polari can search in chats?
<kisaja[m]>never mind i thought it was a matrix client
<dannym>(Now, guix system disk-image is building qtbase uselessly, because u-boot-tools@2020.10 -> sdl2@2.0.12 -> fcitx@ -> extra-cmake-modules@5.70.0 -> qtbase@5.14.2. fcitx is an input method thing--which we definitely don't need for u-boot-tools...)
<guix-vits>sneek: later tell HalPMe try use `nft -f rules.conf` or `iptables-restore`. things like `nft add` not working for me either, but loading rules from files do work.
<leoprikler>kisaja[m]: No, it can't. It's replacing stuff you paste in the input window. Should have been clear about that.
<guix-vits>lispmacs: wow, is it was merged!?
<lfam>I wonder, does anyone have a Guix package of the Glimpse image editor (fork / re-name of GIMP)?
<leoprikler>not atm
<leoprikler>is it more work than (package/inherit gimp (source [origin for glimpse])), though?
<lfam>I'm not sure
<lfam>They do claim to make many more changes than just changing the name
<lfam>I'll try it
*guix-vits ye, it was.
<civodul>lfam: my understanding is that it's a fork, not just a rename
<lfam>Right, that's my limited understanding as well
<lfam>For one thing, it builds with meson
<leoprikler>In that case it will most likely require a decent amount of work.
<lfam>s/gnu-build-system/meson-build-system, hopefully
<guix-vits>lispmacs: "If a filename is stated, a log of the game is written ... If the "-a" flag ... that file is appended" maybe that is a savefile? try trek -a file and restart? (idk)
<efraim>cond is first-match, right?
<efraim>looks like it
<dannym>efraim: Yes
<oogel>anyones mpd stop working recently? or can tell me where to find the error log for services?
<dannym>oogel: /var/log/messages, or if the mpd service specified #:log-file then that. I checked, they do the latter. /var/run/mpd/*/log (O_o)
<pineapples>Anyone up for fixing our flatpak package recipe? I've found a few issues, and have figured out what inputs are required to fix them, but I'd like to hear your opinion before I rush to submit a bug report
<leoprikler>oogel: a bug has already been reported afaik
<leoprikler>iirc you should be able to start it manually, but shepherd disables it because it fails too much during startup
<oogel>ah i see, thanks!
<lfam>I was mistaken. Glimpse uses gnu-build-system. It should be easy to package
<nckx>Mornin' Guixoids.
<nckx>pineapples: Perhaps in the form of... a patch? 😊
<pineapples>nckx: The thing is, I don't know whether those inputs should be propagated or not. If not, then we'll have to patch paths within Flatpak's shell script helpers, whereas propagating the missing inputs just works™ at the expense of cluttering our users' profiles
<pineapples>The shell script helpers in question can be found in /gnu/store/xxx-flatpak-1.8.2/share/flatpak/triggers/ for what it's worth
<nckx>Thanks. I'm not familiar with Flatpak's but shell scripts are generally easy to patch.
<lfam>I'm optimistic that EFI oops in linux-libre 5.9.11 is fixed in 5.9.12
<lfam>I see some reversions and other commits
<pineapples>Ifam: I'm still wondering as to why we unmount the efivars filesystem to begin with
<lfam>"Someone" will need to do the work to learn why and if it can be changed :)
<lfam>I'm just guiding the kernel packages
<nckx>pineapples: I'm still confused about what's so bad about unmounting it. The [next] bug could just as well affect only systems that don't... Is it fundamentally wrong?
<pineapples>nckx: I follow this watered-down, simple logic: if systemd doesn't do something, meaning most of the GNU/Linux distributions, then not doing it can't be bad, can it?
*nckx coughs.
*pineapples clears throat
*nckx shuffles around awkwardly.
<pineapples>Heh. To be frank, I'm just a mere user.
<pineapples>I have no idea about this stuff
<leoprikler>That logic still seems kinda backward to me.
<leoprikler>If a bug is discovered in systemd (directly or indirectly), it is a pretty bad one, that affects most Linux distros.
<leoprikler>It also affecting ones that run shepherd is not a good thing.
<pineapples>leoprikler: Good point
<leoprikler>That being said, it might be worth discussing what the efivars filesystem does and whether we want something like that in Guix or not.
<pineapples>Question: do I have to include a copyright notice to patches sent to the mailing list?
<nckx>leoprikler: ‘ Tools such as efibootmgr rely [...] on the new /sys/firmware/efi/efivars API. The latter needs to be mounted.’
<leoprikler>The answer to that is a clear and resounding "maybe".
<pineapples>I'm willing to cook up a patch to add xdg-utils to Alacritty's propagated-inputs
<leoprikler>If it's an insanely small patch, then definitely no.
<leoprikler>If it's a larger one, then the reviewer may still add you.
<leoprikler>And if you do it on your own on a patch, that is waiting on the ML for a long enough time, that line will cause some merge conflicts :)
<leoprikler>[been there, done that]
<kisaja[m]>sounds appealing
<nckx>But they are the best merge conflicts because they are trivial to fix so nobody minds.
<charles`>raghavgururajan how do I build the iso for testing?
<leoprikler>perhaps trivial, but I still hate git for barfing them into my face
<nckx>...well that's a graphic metaphor for deleting <<<< >>>> 🤭
<bdju>looks like alacritty never got updated :(
***ba is now known as bandali
<gusterhack>Hello I am looking for the ssh-askpass package how can I install it
<bdju>I see there's an lxqt-openssh-askpass package
<bdju>you could `guix install lxqt-openssh-askpass` or add it to your config.scm file or a manifest.scm file
<leoprikler>that's one option
<leoprikler>gnome actually uses its own askpass as part of gnome-keyring, that's another
<gusterhack>I installed the package "lxqt-openssh-askpass". I need it to connect via ssh to virt-manager but despite that virt-manager tells me that it cannot find ssh-askpass.
<jonsger>mbakke: I don't think it hurts to write why you revert a commit
<lfam>It's true, but we all sometimes forget or fall short of our goals
<nckx>gusterhack: You need to create a symlink to your chosen *-*ssh-askpass project, called ‘ssh-askpath’, in a directory in/added to your $PATH.
<nckx>*askpass, grmbl.
<jonsger>lfam: you mean my statement?
<nckx>We could create foo-askpass-wrapper packages for each one like we do for python...
<lfam>Yes, regarding the lack of explanatory comments
<lfam>jonsger ^
<jonsger>lfam: if you revert multiple commits at least the first really should have a reasoning
<lfam>We know
<lfam>I'm sure it's about this message:
<nckx>I'm sure they predate that message.
<nckx>They caused too many rebuilds.
<lfam>I see
<jonsger>lfam: no it's not about "faulty pushed" patches, its about reverting them without writing down the reason...
<nckx>Like jonsger I resorted to the IRC logs to find out why I got spammed with reversion notifications 😉
<lfam>jonsger: What I'm trying to say is that mbakke knows that one should explain why they revert patches
<jonsger>lfam: I'm sure he know :)
<lfam>In any case, I agree with Mark, too. But sometimes I run out of energy to ask for more changes to the patches, and it starts to feel adversarial
<lfam>It is interesting to me that JPEG2000 is such a heavy package. As a format, it's only used in the video / film production industry. But I know that "real-life" use cases don't match dependency graphs
<mdevos>The issue tracker returns a 500 Internal Server Error when accessing with IceCat. Anyone knows what's up with that?
<mdevos>other issues seem unaffected
<lfam>mdevos: Dunno, but I recommend using <>. That's the "real" bug tracker website anyways
<lfam>The domain shouldn't be used at all going forward. is supposed to be used but is much less reliable than
<mdevos>fam: ok, I'll keep that in min
<lfam>We want to make the first-class site, but it's not quite there yet. The debbugs of is not designed to be used as the backend of a webpage so has trouble getting good information out of it
<dissoc>are there hooks you can add for guix deploy? say you wanted to restart a nginx or something like that
<mdevos>lfam: the ‘real’bug tracker has the same issue btw
<euandreh>yeah, probably a problem with mumi, not debbugs
<lfam>Yes, I know. The problem is deeper than '' vs '', but we aren't continuing with the domain in the future
<mdevos>I now see 44964 has been merged into 44914. Perhaps mumi has an issue with merges? (only speculation)
<euandreh>dissoc: you can do that in the (operating-system ...) definition
<euandreh>dissoc: let me give you a snippet
<jeko>My first time ! "guix deploy: déploiement de ynm1606939399 réussi" (french for success haha)
<euandreh>make a (simple-service ) that runs after a system activation, and restart nginx there
<euandreh>the nginx restart from the cookbook does it without stopping and starting nginx
<euandreh>jeko: nice
<euandreh>jeko: is the os definition public somewhere?
<euandreh>now I'm curious
<dissoc>jeko: i recently just started to use deploy too. it is great
<dissoc>euandreh: thank you. i will try this
<jeko>oh my *.*
<jeko>euandreh: give me 3sec…
<mdevos>in case anyone here is investigation recent test failures in gnutls-dane (issue #44914): gnutls-dane reproducibly fails to build each time for me (no race issues or anything), I get the exact same test failure as #44914. gnutls (without dane support) builds perfectly though
<rekado_>mdevos: looks like you found a bug
<mdevos>rekado: in mumi or gnutls?
<jeko>euandreh: is it accessible ?
<rekado_>mumi. Something about the email address not being a pair, so parsing must have failed.
<rekado_>possibly even a bug in guile-email
<jeko>(without the /edit part haha)
<rekado_>I still recommend using, by the way.
<euandreh>jeko: yes :)
<euandreh>ty, 👀
<mdevos>rekado: sounds like a solid lead
<lfam>Yes, please use That's the only way we can improve it :) But remember that you can fall back to if necessary
<dissoc>euandreh: your config is also useful to help me set up cgit. thank you again
<jeko>euandreh: using the droplet declaration doesn't work actually
<euandreh>dissoc: cgit is working, but TLS isn't yet
<euandreh>jeko: the ynm-droplet-declaration.scm file?
<jeko>euandreh: yep. `guix deploy ynm-droplet-declaration.scm` is broken right know. I think there is an issue with guix, I opened a bug about it yesterday
<civodul>lfam: re, we should set redirects in the nginx config
<lfam>That's a good idea
<lfam>"We" should do that ;)
<lle-bout>1003 patches pending - nightmare
<new-user>hey guys, I have a question about guix import. I want to install `cargo-edit` and used the command `guix import crate cargo-edit` to get the scheme code but now I don't know how to proceed.
<euandreh>new-user: do you get an error?
<new-user>No the code got generated just fine I just don't know what to do next to install the package
<euandreh>hmm, ok. You can put it in a file and install from it
<nckx>civodul, lfam: Oh, please don't edit the config on berlin, I'm doing so right now ☺
<euandreh>i think it will be:
<lfam>nckx: Don't worry, I can't
<nckx>Reconfigure just finished, looks good.
<luis-felipe>Hi. I jsut added GNU Guix 1.2.0 stuff to the store:
<euandreh>guix import crate cargo-edit > cargo-edit.scm; guix environment -f cargo-edit.scm
<luis-felipe>The first T-shirt:
<euandreh>new-user: the 'guix environment' will only load on the current shell, not install globally
<euandreh>To install system-wide, you can 'guix package -f cargo-edit.scm'
<new-user>When I run the guix environment I get an error: `f: unrecognized option`
<lle-bout>nckx: hey, is there any trick to creating more future-proof patches? For example, patches against python-xyz.scm like in depend on the last package in the file staying the same, which means that if someone adds a new python package in python-xyz.scm all my patches break.
<euandreh>new-user: oops, with guix environment the flag is -l
<euandreh>'guix environment -l cargo-edit.scm'
<euandreh>luis-felipe: oh, that's just fantastic
<new-user>Now I get another error: `rust-cargo-edit.scm:5:0: error: package: unbound variable`
<new-user>Ah wait I fixed it, I had to add some `use-modules` statements at the top
<euandreh>new-user: yep, I was just typing that
<new-user>But there is still an error: `license:asl2.0: unbound variable`
<euandreh>same thing, but now using a prefix. See the "Extended example" on the cookbook:
<euandreh>new-user: the second line on the first snippet, yeah?
<new-user>No it's the name of the license, here is the full error: `rust-cargo-edit.scm:10:4: error: license:asl2.0: unbound variable`
<bavier[m]>luis-felipe: very cool.
<euandreh>but it's just that: the (guix licenses) module is being used with a 'license:' prefix
<euandreh>that is the '#:use-module ((guix licenses) #:prefix license:)" part
<nckx>lle-bout: Don't add packages to the end, as you say it causes needless merge conflicts. Try to add them somewhere inside the file, if possible where it makes some alphabetical sense.
<lle-bout>nckx: heh okay
<nckx>Nothing's future-proof but it helps.
<nckx>civodul: So... I made the changes before I saw your suggestion and redirected ‘’ as well, since it's the same machine as ci. OK?
<lfam>Awesome luis-felipe :)
<civodul>luis-felipe: yay, very nice!
<civodul>well done :-)
<lfam>I love the drawing
<civodul>nckx: oh excellent, perfect!
<civodul>nckx: you're talking about HTTP(S) redirects, right?
<lfam>luis-felipe: I'm curious about the store. Is it easy to offer the hoodie on black or gray fabric? I know the poem might be tricky...
<nckx>civodul: Yes. It was long overdue.
<civodul>yup, awesome
<luis-felipe>lfam: For black or darker cloth colors I'd need to do a another collection. I plan to do that this week along with printing on accesories (backpacks, for example)
<lfam>Cool! I'll purchase it when available, along with a white tshirt
<civodul>luis-felipe: feel free to send the link on guix-devel/help-guix as well if you want!
<luis-felipe>Oh, ok, thanks.
<lfam>I think I just figured out the idea behind the drawing :)
<luis-felipe>I haven't publish anything elsewhere, I wanted to let people know here first.
<luis-felipe>lfam: What is it? :)
<lfam>"Out of the cloud"
<lfam>Ascending out of the clouds, speeding towards our destination
<lfam>And the people in the clouds are watching with great interest!
<luis-felipe>Ha, not really. It is simpler: The Guix ascending through golden clouds, which fly alongside the plane with a smile on their faces, like dolphins and ships.
<lfam>I put too much thought into it
<lfam>It's a really nice drawing.
<luis-felipe>The clouds of "The Cloud" would be angry because they let some people get away.
<luis-felipe>lfam: Thanks, glad to hear that :)
<lfam>I was thinking something like this:
<lfam>Not that Guix is really an anti-cloud project or anything. But there is some synergy
<nckx>Does the grid overlay mean anything?
<luis-felipe>nckx: This was supposed to look like a painting on a mural. Those tiles and the lines that make the drawing look more dirty are supposed to look like a wall in downtown somewhere.
<nckx>luis-felipe: That makes sense. Thanks!
<ces>Hey, just switched from archlinux, and i can't figure out how to change the available de/wm's in gdm. Also, am i supposed to do this in the /etc/ file?
<ngz>ooooh improved rust importer landed on master!
<brown121407>ces, yeah, if there's a service available for a DE, add it to your services. Also if you install a WM, it should be seen by your display manager and made available from its menu, but most display managers, if not all, also execute ~/.xsession after login so you can start your WM from there.
<brown121407>Where "add it to your services" means the "services" list in your operating-system in config.scm
<ces>brown121407, thanks! I normally use xmonad, and i have installed it but it does not show up in gdm. Also, is there some way to get documentation on a specific package, such as wether it has a service?
<ces>The .xsessions sound useful, but i would prefer a way to avoid it starting with a DE
<lfam>I notice we seem to be missing the substitute for the freedo image
<lfam>Hm... it downloads okay with curl
<mbakke>jonsger: I'm aware and apologise for not explaining the reverts. I was frustrated and too busy studying which of the commits in the series actually caused rebuilds to follow due diligence. I hope the courts will consider my continuous complaining in this channel as mitigating circumstances :-)
<jonsger>mbakke: it's okay :)
<mbakke>ces: you can try 'guix system search package' to see if there's a service for it
<nckx>$ guix show-irc-logs-leading-up-to <commit>
<mbakke>yay for the Rust importer, really great work by Martin and Hartmut
<mbakke>now, if we could get rid of '#:cargo-inputs' and '#:cargo-build-inputs' I'd be a very happy camper :P
<nckx>Sweet Guix dreamz. o/
<mbakke>rekado_: have you looked into Pipewire at all?
<mbakke>(I just read the JACK email on guix-devel)
<mbakke>we could probably use the underutilized Zabbix setup on Berlin to plot size changes over time
<mbakke>at work they have an automated and declarative Zabbix configuration using 'zabbix-auto-config':
<mbakke>I realize I haven't read guix-devel in a long time due to not making a shortcut in my updated email UI :P
<Sharlatan>Hi folks
<Sharlatan>I try to follow this thread to resolve issue with emacs tramp to Guix,
<lfam>How is that going, Sharlatan
<ngz>Hmmm where should I put Monolith package definition, in web.scm (but I have to import Rust-related modules) or in rust-apps.scm?
<ngz>Let's go for web.scm.
<lfam>Sounds good ngz
<pineapples>mbakke: May I have the link to the discussion about Pipewire in question?
<mbakke>pineapples: there have been no discussion about Pipewire, just JACK 1 vs JACK 2 ... I'm just polluting it with semi-related topics, as is my specialty :P
<mbakke>I'm going to try converting my system to be "fully Pipewired" one of these days though..
<lfam>Pipewire is promising. Handling both audio and video makes it doubly-useful
<pineapples>I'm eagerly looking forward to Pipewire's debut on Guix :')
<Sharlatan>lfam, good night time with Guix :)
<Sharlatan>collecting dependancies for pgloader...
<lfam>Handling A/V streams within an OS is a newly mainstream use case, since life has fully migrated online :/
<mbakke>lfam: indeed, and the protocol is designed to work natively with containerization technologies, and to be ABI compatible with both Pulseaudio and JACK.
<lfam>There are all kinds of half-baked solutions being popularized right now
<mbakke>really promising stuff :)
<mbakke>Pulseaudio has been popularized for some time already ;-)
<lfam>Okay :) I'm talking about for A and V both
<mbakke>ah yes
<lfam>If GNU/Linux can deliver a solid solution, it will gain a new class of professional users
<lfam>Especially if it can properly integrate remote sources and "sinks"
<lfam>Whoever makes a simple GUI for SRT AV streams can make a lot of money
<lfam>Right now people are paying $15 per month for awful quality Zoom streams to solve that use case
<lfam>And staying up all night trying to trick zoom into looking better
<lfam>Or, they spend several thousands of dollars on hardware solutions
<lfam>Meanwhile all the software involved is free
<lfam>And you also have hacky things like OBS-VirtualCam just to move a video stream between applications on the same OS
<mbakke>Great points. I do think Pipewire has potential to deal with all those cases.
<lfam>There's a lot of opportunities now for free software developers to get paying customers in these areas