IRC channel logs
2025-09-30.log
back to list of logs
<ColdSideOfPillow>I'm in the process of packaging the newer version, but it seems that they've completely reworked their codebase and now use a cargo-ninja hybrid model <ieure>ColdSideOfPillow, I must not care about versions. Caring about versions is the mindkiller. <ColdSideOfPillow>ieure: It matters in this case because I cannot import like 40% of my decks. <apteryx>ACTION updated gnucash on mesa-updates, for libxml2 compatibility <apteryx>mesa-updates: php 8 is failing its test suite <apteryx>any idea how can I use #$ in a let-syntax macro? <apteryx>It seems to be eager, and complains the 'input' binding passed to it may not exist. <mange>What do you want to do? #$ happens at read time, but the form it reads shouldn't be evaluated until later. <apteryx>I was gexpifying php, and trying to adjust the 'with' syntax to use this-package-input <apteryx>instead of (assoc-ref %build-inputs input) <apteryx>nice, php test suite time down to 60 s from 300 s, enabling parallelism <mange>Also, I forgot to mention: I played with that #$ bit and I couldn't find a way to get it working. The issue is that #$ is code staging, so the evaluation happens in different places. The #$ has to be syntactically visible (prior to expansion) to the #~ so it can be escaped properly when passing it to the builder. <apteryx>mange: I see; so it's just impractical here <madage>apteryx: msamr update on mesa-updates seems to be broken according to ci <madage>seems the hash was updated but not the revision/commit <apteryx>mesa-updates hackers: I just force-pushed with a fixed php package (and others). Please reset --hard or rebase on top. <apteryx>madage: clearly a mistake. reverting this one. <robin>yeah, it looks like gexp records ungexps at expand-time, with no fancy code-walking that would handle this. you could have something like (with-options (with _ _) ...) treating "with" as a keyword and expanding to (gexp (list (string-append ...) ...)) but not #~(with-options ...) <apteryx>ACTION is shaking out flakiness out of php's test suite <apteryx>does the new unprivileged means to run guix-daemon get rid of the requirement of having build users? <apteryx>they frequently run out, it's annoying. Perhaps I should create a hundred. <mange>Quoth the manual: "When using this option, you only need to create one user account, [...]". <mange>So it sounds like yes, it doesn't require build users. <apteryx>good. May be worth putting up with the other (unresolved yet?) problems it introduces. <apteryx>seems to work great. inkscape launches too, better than master ;-) <apteryx>unless someone has some serious issues to point me too, I'll merge mesa-updates very soon. <podiki>apteryx: i haven't seen that specifically for emacs, but random builds hanging sometimes just in general on ci (rarely, but i saw it a few times on all the mesa-updates runs) <untrusem>Chwoka2, I am using the same version but it works fine <untrusem>do you have locale issues in other programs as well? <Chwoka>shoot, additional detail i forgot to add: "guix install librewolf" did conclude with a hint hint: Consider setting the necessary environment variables by running: <Chwoka> GUIX_PROFILE="/home/guest/.guix-profile" <Chwoka>Alternately, see `guix package --search-paths -p "/home/guest/.guix-profile"'. <untrusem>I am not sure but you should set the env variables <eikcaz>Did shepherd have a memory leak bug recently? Trying to figure out how pid 1 shepherd is usig 6.9/7.6 gigs of system memory <untrusem>btw I am getting `No KVM` error while trying to make a vm in gnome-boxes. <Rutherther>Chwoka: unlikely. Symbol error means somehow wrong library is used, those aren't determined by the normal search paths in the profile. The only 'regular' thing I can think of causing this is having LD_LIBRARY_PATH set. It shouldn't be, my guess is you don't have it as it is a new installation. Then there is the possibility for an irregular thing - corruption. You can try guix gc --verify=contents,repair. But I am afraid it won't be that and then I am... <eikcaz>Rutherther: Thanks. Seems I pushed a buggy version to my entire swarm then. I see my laptop has 91GB out of 64GB allocated to shepherd. <Chwoka>yeah that's what's really raising my eyebrows is that this is like, fresh. i did guix pull, installed fish and kate, and then librewolf, and that's it <Chwoka>so i didn't do any setting of env variables or anything. those 4 commands are the complete history here <Rutherther>eikcaz: iirc it should be solved with guile-fibers 1.4.1, so you can check if your are on 1.3 or not <Chwoka>Rutherther: entered your command, still getting the same error. i'm kinda assuming this is some kind of error within librewolf or its dependent things now <Rutherther>Chwoka: so the command didnt print any errors? If not, it wouldn't change anything <Chwoka>nope, it didn't print any errors. <Rutherther>untrusem: right, I thought that was merged already <Chwoka>i'm trying out "guix environment librewolf" now <Chwoka>maybe that'll do... something. or nothing. or break shit, but it's just a vm and i already have a backup <Chwoka>ah yes. i see here it says shell supercedes environment. welp <Rutherther>Guix environment librewolf is going to give you development inputs of librewolf, don't see how that would help <Rutherther>With shell the equivalent would be guix shell -D librewolf <Chwoka>perhaps then i could install librewolf inside the environment with all its dependencies and see if it's like, somehow, a conflict with the other things i already have installed? <Chwoka>running "guix shell librewolf" returns symbol lookup error: /gnu/store/5h2w4qi9hk1qzzgi1w83220ydslinr4s-glibc-2.33/lib/libpthread.so.0: undefined symbol: __libc_pthread_init, version GLIBC_PRIVATE <Rutherther>Do I understand correctly you are on guix system? <Chwoka>at a glance that looks to be the same error <Rutherther>And you sure the gc verify command finished and without errors, yes? <Chwoka>it went "reading the store, checking path existence, checking hashes..." and then put me back to prompt with no message <Chwoka>thanks for talking it out though <apteryx>looks like rpm built with cmake no longer has an easy way to configure its runtime localstatedir <untrusem>Chwoka, just mention what error you are getting <Chwoka>sure but mention... where? heck, on guix's codeberg? <apteryx>Chwoka: this sounds like some clash between your sysem glibc and the glibc your user applications are linked to <apteryx>re rpm, was able to set a runtime localstatedir by means of its #$output/etc/rpm/macros file <apteryx>(adding a file there with the right variable defs) <Chwoka>yes i'm on guix system. looking over my terminal log that contains everything from first pull to error it looks like this might indeed have something to do with multiple versions of glibc <Chwoka>guix pull brought me 2.83, librewolf install process also called on 2.83... and then "guix environment librewolf" actually pulled in 2.41-static? and the error indicates it's trying to pull on 2.33, which i can check the file structure and confirm is sure as shit there <apteryx>Chwoka: I'd recommend reconfiguring your system with 'guix system reconfigure ...' and rebooting it <Chwoka>no idea how or when 2.33 got installed. there's no messages about it in that terminal log <apteryx>it'll probably get rid of this kind of problem <apteryx>I'm not sure exactly what is at plays. It's the kind of situation I thought glibc was designed to handle well, but it seems it doesn't. <Chwoka>apteryx: i've never used reconfigure before. to my understanding i need to point it at a manifest.scm, yes? <apteryx>to your operating system declaration <apteryx>how did you install? using the graphical installer? <apteryx>If so, the file should be at /etc/config.scm, IIRC. <Chwoka>through the 1.4.0 system image qcow2 in gnome-boxes <apteryx>so 'sudo guix system reconfigure /etc/config.scm' should do it. <Chwoka>ahhhh i knew it was one of those two haha <eikcaz>apteryx: re shepherd memory leak, do you have a suggested work around? reboot on a timer maybe? <apteryx>I think civodul posted a good workaround on bug-guix, something replacing the builtin syslog of shepherd with the original syslog program. <apteryx>haven't tried it yet, but I should before my VPS is brought down again. <Chwoka>apteryx: i checked before i leaped: no "config.scm" in /etc/. according to "guix system describe" my config is actually located at /gnu/store/9n1vby8zb8kbjy5rfckaqifhdvlgy0i2-configuration.scm <apteryx>janneke may be of interest to you as well ^ <Chwoka>(looking at that file, it's got comments and everything telling me "this is a configuration for running guix in a vm") <apteryx>in case you'd like to be listed there too, you could send a PR editing the commercial/data.scm file and adding your logo, if you have one. <apteryx>you could copy this somewhere in your HOME and make it writable, for when you'll want to edit it. It's good to have it version-controlled as well. <Chwoka>alright, here goes the reconfigure <Chwoka>ah shoot that would have been a good idea. oh well, i have multiple kinds of rewind available to me <apteryx>yeah, if you don't plan on changing it yet, there's no rush <Chwoka>looks like the kind of thing i can fix myself but, funny to have errors in the configuration of the guix system system image <Chwoka>uest@gnu ~$ sudo guix system reconfigure /gnu/store/9n1vby8zb8kbjy5rfckaqifhdvlgy0i2-configuration.scm <Chwoka>hint: Did you forget `(use-modules (gnu packages nss))'? <Chwoka>oh, because they start with slashes <Chwoka>gnu/store/9n1vby8zb8kbjy5rfckaqifhdvlgy0i2-configuration.scm:110:26: warning: 'dhcp-client-service-type' is deprecated, use 'dhcpcd-service-type' instead <Chwoka>gnu/store/9n1vby8zb8kbjy5rfckaqifhdvlgy0i2-configuration.scm:79:46: error: nss-certs: unbound variable <Chwoka>helpful little error messages that tell me exactly what to fix at least <untrusem>nss-certs was moved to somewhere else, I don't remember, wait let me look up <untrusem>if you have (gnu package nss-certs) module change it to (gnu packages certs) <Chwoka>good catch but i saw it a couple seconds too late and have already run reconfigure on my modified-config.scm. hopefully (gnu packages nss-certs) is fine... <Chwoka>in the vm it kinda looks like it just froze but i don't know if that's what it's supposed to look like and i just gotta give it some time to think it over... <apteryx>Chwoka: hm, you don't need to list nss anymore, with current Guix it gets added to the base applications. <apteryx>it's just this template is as old as the image :-) <eikcaz>apteryx: I found one message on bug-guix that suggests to just remove shepherd-system-log-service-type entirely. Think I'll just do that. Beforem my VPS similarly goes down <apteryx>I'm ready to merge mesa-updates, unless someone has something in need of fixing. <apteryx>I've fixed everything CI reported as new, I think. <Chwoka>that last line hit me like a punchline. you're the one that told me to put it in dum dum... <Chwoka>oh wait i see -- i had a typo there <eikcaz>oh lord, the ai scrappers brought all of gnu.org down? <eikcaz>Chwoka, I can't seem to ping it. <Chwoka>it opens fine in my browser (outside of my vm) <eikcaz>I figured this would be an easy fix with Anubis, but I noticed in an FSF newsletter that they consider Anubis malware? Don't quite get their stance, but curious what a better solution would be <eikcaz>Chwoka: did you refresh? Browsers like to cache static pages like gnu.org. I can't pull it up or even ping it, and other sites load fine for me <apteryx>I'm told in #gnu that it works for others <Chwoka>i double-checked by pulling up pages i'm 100% certain i'd never looked at before <Chwoka>maybe a netsplit type situation, to use irc terminology <eikcaz>ah. guess my ip is blocked. I'm on Starlink's CGNAT. <apteryx>Chwoka I'd start with a configuration file that works before making any edit, it'll be easier :-) <Chwoka>apteryx: yeah i was beginning to get the feeling that this was gonna work up to me rolling my own config.scm... i don't feel i'm ready for that quite yet but alas <apteryx>Chwoka: other tip, to learn how to use each interface/service: 'info guix' then you use the magic 'i' to access something indexed. Most of the API should be indexed. <apteryx>dhcpcd-service-type should probably have a default configuration <Chwoka>you want me to try to reconfigure based on this? <apteryx>if you are about to push to master, hold it for 2 minutes please :-) <apteryx>hm, rebasing on master invalidated a bunch of stuff. oh well. <apteryx>alright, the mesa-updates branch is merged! kudos to everyone who helped drive it home. <efraim>I remember what it was, I couldn't figure out running the tests with the 'unittests backend and skipping a test and the other tests actually running <trev>untrusem: final last words <trev>i guix pulled today and now i have no keyboard inputs <untrusem>resuming guix pull, someone had to sacrfice :P <efraim>can I add a home-environment to a user in my OS config or did I dream that it was possible? <untrusem>there is home-service-type system service <efraim>ok, I see the guix-home-service-type. I guess it's a service and not part of the user itself <efraim>I'll see if I can add it nicely to my offload machines <mart-e>Hello! I have just installed guix on my pc. Fetching the binaries on ftp.gnu.org was quite slow (took me around an hour at ~20KB/s) but now my first 'guix pull' is awefully slow. I am still at 1% after 15min fetching from git.savannah.gnu.org. Is it an issue on my side or same for everybody? Should I be using a mirror instead? <trev>change to the codeberg mart-e <nutcase>efraim: khard again (see #1456) fails to build? Do you know the reason? <efraim>I got it to build in the end, I ended up using pytest to run the tests and then had to skip 2 additional tests that also failed on 0.19.1 when using pytest <mart-e>Thanks trev, I guess I need to change the channel. Will try that. <nutcase>efraim: thanks for the fix. It's now also solved for future khard versions or will it reoccur? <efraim>no clue, hopefully it'll just work with the next version also but time will tell <DarthTRump>is there a way to install nonguix I need some seperate drivers <nutcase>efraim: sure, stupid question. I just wondered, why building khard failed again with the same message as in 1456. <efraim>nutcase: I'm not sure why it failed for you in 1456 but I know there's been some changes in the python- and pyproject- build systems so that might've triggered something <DarthTRump>attila_lendvai I mean rather like I dont want that someone installs apps with .sh code <attila_lendvai>DarthTRump, nonguix is a guix channel that you add. .sh installs wouldn't even work easily on guix due to the /gnu/store stuff. <untrusem>ok ffmpeg-4.4.6 don't have substitute yet, I am not gonna build that on my t480 <untrusem>this is during the guix pull after the merge of mesa-updates branch <jakef>i cannot wait to be able to use guix-science-nonfree again <efraim>hmm, having some trouble with vulkan and qutebrowser <efraim>12:19:59 WARNING: Failed to load vulkan: Cannot load library vulkan: vulkan: cannot open shared object file: No such file or directory <jakef>Deltafire: yep im giving it a try. first fail i've hit is perl-xml-libxslt <Deltafire>wouldn't be surprised if that's somehow related to the libxml2 update <Deltafire>wow, 2.8gig of substitutes for system reconfig <untrusem>now webkitgtk is trying to build, I am sure it would take hell lot of time on my machine <simendsjo>Pulled with the merged mesa branch. "The following derivations will be built:" .. 339 derivations. Wish me luck! <untrusem>the substitute build but not yet available <untrusem>best of luck simendsjo, may the machine be with you :P <untrusem>I have already pulled it with merged mesa branch and reconfigured my home with it <attila_lendvai>yay for the mesa-updates merge! it was a precondition to finishing the cleanup of the libxml2 issues, right? <untrusem>yeah, I don't see the libxml warning for ffmpeg which I was getting before <attila_lendvai>i'm primarily concerned about my custom xkb keymap disappearing from gnome. that's what keeps me on an older commit prior to the breakage. <untrusem>ok, I am still getting libxml warning for gnome-boxes, which also gives `No kvm` error while trying to make a vm <simendsjo>Cannot build f5vpn because it depends on libxml2.so.2. Can libxml2.so.16 be used instead? <tesseract>hello everyone. i installed guix with the script. it was very slow. it was downloaded average 15KB/s. after installation i am using "git pull" but it can't download from git.savannah.gnu.org/git/guix.git <klm`>apteryx: ref our conversation on non-free blobs in integrated GPUs: I just bought a used Intel i9-13900K to use with my GNU Guix system but was very disappointed to find that the GPU on that needs proprietary firmware blobs to work. <klm`>I intend to buy a Radeon HD 6850 and hope that works. The GPU jungle is _very_ difficult to navigate. Please let me know if someone knows that to be working with GNU Guix or not :-) <klm`>tesseract: hi, I'm just encountering that too on my new system <klm`>tesseract: you can specify a new url for the guix repository, let me dig up what I used. hang tight <tesseract>and i don't know how to change the repo though <klm`>you don't have to change anything, you can specify it in your guix pull command <klm`>I'd love to have someone else comment on this way of doing it. That commit hash points to a commit from june this year or something. I find that if I pull the latest guix, I have to build many of the packages myself because the remote build pipeline hasn't caught up yet. <klm`>So I specify older commits. <kestrelwx>I think any modern GPU from the big 3 you will need blobs. <klm`>tesseract: Glad to hear it! Guix has officially moved to codeberg.org, so doing this should be uncontroversion I think - but I don't know why the documentation around it isn't clearer. Specially since savannah is pretty much completely unuable (at least when I've tried) <kestrelwx>I'm on an early September commit because of the 'libxml2' issues, but otherwise I tend to stay on master. <kestrelwx>tesseract: Yes, it's mentioned on the blog post on the official site. <identity>klm`: by the time guix pull finishes, the build farm has already built all the packages i use… <klm`>oh, interesting. how do you manage to stay behind when you `guix pull`? by specifying the commit like in the command above? <klm`>identity: hmm maybe I'm using odd packages then. I have to rebuild ungoogled-chromium even on that commit above. But that build fails - so I probably need to find a better commit ... <identity>klm`: no, i just have a slow laptop and a slow internet connection do not use many packages that i could not build on it (save for, say, a browser) <klm`>so you don't use a browser? <identity>no, i do, much to my dismay, but it is generally built by the time i can get my hands on it <identity>i mostly have emacs and small-ish programs installed apart from that <klm`>which one, if I may ask? <erru>hello o/ i have a systemd question: im developing software designed to run on an RHEL system. it has a systemd dependency to notify systemd of successfull startup / failure. i would like to build said software on guix, since i like the OS. how to i get the headers and the library so i can build and link my executable dynamically? do i need to build systemd from source for that? <klm`>oh, interesting. I find icecat hard to use, and ungoogled-chromium won't build (but maybe it's fixed by now) <identity>besides, before big branch merges the substitutes are made available <tesseract>guys it gives me error in the end "guix pull: error: `/gnu/store/5kj8lyybjrdl7xd09g9vzkz8sklqsy-guix-1.4.0/bin/guix substitute' died unexpectedly" <identity>tesseract: probably a connection error, just retry <identity>ugh, dealing with old guix versions is pain <identity>tesseract: does your distro not provide an icecat/iceweasel/$firefork package? <tesseract>i am using debian 13, so it provides firefox <identity>love re-downloading 1gb of data because ‘unzip’ got rebuilt <simendsjo>Should I report failing packages, or is the CI listing everything? I'm not really a friend with the cuirass GUI, so I'm having some problems navigating. <klm`>my new GNU Guix installation crashes right on start: segault on init[1] (likely on core 20) it says. How do I debug? It puts me in a Guile shell, but the keyboard isn't working (which is _very_ strange isn't it?) <klm`>My kernel arguments are the same as when booting from the official GNU GUix image (USB stick) <identity>simendsjo: builds that fail on master are surely worth a report <nutcase>I think, I remember there was a cli option to exclude a package from a guix system reconfigure. Did I dream that or does such an option exist? <simendsjo>identity: But to me, it looks like it has been failing for a long time even before merging to master? So it was decided that breaking the package was ok? Again, I don't really know cuirass. <identity>nutcase: i doubt that, but you can just comment the package out <tesseract>now it says: "guix upgrade: warning: Consider running 'guix pull' followed by 'guix package -u' to get up-to-date packages and security updates." <klm`>kestrelwx: but I don't know how to proceed. Should I just try a different --commit=... hash in guix pull? It takes _forever_ to reconfigure my system that way. Is there a way to use the guix from my installed system, but booting from the GNU Guix 1.4 USB stick? <kestrelwx>If you could boot previously switch into that. <klm`>I don't have any working boots (happend after "guix system init ...) <kestrelwx>tesseract: No, I think you pulled into a July commit so it warns you about an older version. <tesseract>i checked "guix --version" and it tells me 1.4.0 <klm`>tesseract: That would be the same commit I'm struggling with a segfault on now <kestrelwx>It's 1.4.0 for 3 years now could see the commit you're on with 'guix describe'. <tesseract>it says this. "commit: 8e2f32cee982d42a79e53fc1e9aa7b8ff0514714" <kestrelwx>tesseract: Did you pull as your own user? You shouldn't run 'sudo guix pull'. <craigbro>tesseract: I don't think that's the correct repo url <craigbro>hmm, that url seems to redirect to the correct one tho <kestrelwx>I got 'cc87e77' which is the latest commit pulling from the mirror link. <tesseract>i wish there was official compiled icecat tarball <klm`>kestrelwx: it looks related, but my machine doesn't continue. it just hangs :-( <klm`>but something is strange with my installation: /etc/ is empty! there should at lease be some /etc/group files and things there after a successful `guix system init ...`, right? <erru>does anyone have a fine manifest.scm for building systemd? i get an undefined reference error for libcrypt.so.1 when trying to reference __snprintf@GLIBC_PRIVATE <kestrelwx>klm`: Well, that segfault wouldn't be the problem. <klm`>kestrelwx: ah, of course... <klm`>I will run guix system reconfigure ... and carefully check the logs <tesseract>klm`: hell yeah! i installed icecat and it runs :D <tesseract>do you guys know where icecat keeps profiles now? <klm`>but just to be clear, after `guix system init ...` is done, there should be some files under /mnt/etc/ right? like /mnt/etc/group. Or am I missing something? <pinoaffe>hi folks! my PR to add digikam was merged, so naturally I pulled and tried to run it from master, but for whatever reason I'm getting the following error: CommandLine Error: Option 'debug-counter' registered more than once! LLVM ERROR: inconsistency in registered CommandLine options <pinoaffe>does anyone know what might be causing that? <civodul>pinoaffe: hi! could that be related to the recent Mesa update? (Mesa uses LLVM) <erru>i was able to resolve my issue by installing libxcrypt <pinoaffe>civodul: it didn't work with the broken libxml either, so it's difficult to figure out what broke this <pinoaffe>though maybe the llvm broke it before libxml broke it, I dunno <klm`>to answer my own question: no, it's ok for /etc/ to be empty. everything goes into /var/guix/system/profile/etc and friends. looking elsewhere for the cause of the problem <civodul>warning: 'python-setuptools-next' is deprecated, use 'python-setuptools' instead <civodul>could it be that we deprecated things that are still being used internally? :-) <csantosb>Dumb question of the day, is there a good reason to `(delete-file "autogen.sh")` with `gnu-build-system` ? <csantosb>To my understanding, there is not (see nvc in electronics) <cbaines>guix pull --list-generations seems to have some network dependency which fails for me: <cbaines>guix pull: error: Git error: SSL error: syscall failure: Resource temporarily unavailable <klm`>i fixed it, now i can boot my new machine. I had a wrong entry in (mapped-devices ...). I recorded a video of the boot process and read all messages carefully until I found "pre-mount actions failed". <klm`>for next time, what's a better way to debug early boot like this? I was dropped to a guile repl but the keyboard didn't work. <kestrelwx>klm`: You can see these messages in '# dmesg' output, or they are also logged to '/var/log/messages' '.../messages.$NUM.zst'. <erru>in the guix reference material in 2.4.1 there is an explanation for how to make better use of disk space otherwise swallowed up by glibc-locales. i don't understand that section though. can someone explain it to me? the section provides a definition for a package using some locales. what can i do with it after i added these lines into an .scm file? <kestrelwx>You can see the compressed messages with zstcat/zstless or just use Emacs. <kestrelwx>erru: You would then use the resulting package in your profile. <tesseract>it worked on my VM, but it doesn't fetch that on host lol <identity>tesseract: 803964399cd64afab17599d49e9c6212553e937e is the latest commit, cc87e772bb68ad6c12376ef35745f58937bbaa57 is an older commit <tesseract>identity: is it because of internet connection? <ieure>tesseract, You aren't going to get help by saying "doing this gives an error," you need to paste the error into a pastebin and put the URL here. <tesseract>guys, wait a sec. i am redoing it. will you give error <panosalevro>Apart from here, is there another platform to post this where this could be more visible? A mailing list? <gabber>yelninei: ? what is holding you back? <gabber>yelninei: maybe compile natively for the target architecture? <ieure>tesseract, What Guix browser are you using that doesn't already have uBO? <gabber>yelninei: e.g. `guix build libxml2 --system=ARCH-linux` ? <tesseract>ieure: there is no preinstalled ublock origin addon. only some other addons <ieure>tesseract, You can install ublock-origin-icecat guix package in the same profile as icecat. <yelninei>gabber: this is for my childhurd so my system depends on that <gabber>yelninei: so... `guix build libxml2 --system=i586-gnu` ? <gabber>or does that not work when the childhurd doesn't work? <gabber>it seems to work for me (and something IS wrong with my childhurd) <yelninei>gabber: the fix is to also add python to inputs (or disable the python extension when cross compiling) <yelninei>the first one would be the correct one but that is a world rebuild, disabling the python extension just for cross compiling would be easy <gabber>yelninei: i am not sure but at some point my childhurd refused to let me log in and i haven't taken the time (yet) to get to the core of the issue. there *should* be one of my keys able to get me in there but it keeps asking me for a password (which i think i never set). so i guess i want to just reset it? <yelninei>wait, adding the input does not rebuild the native package so that would work as well <gabber>yelninei: world rebuild for the i586-gnu platform? <FuncProgLinux>Curious question, would it be too difficult to get docker-podman at guix? <yelninei>no only a rebuild for the cross packages which are broken anyway. I guess guix deduplicates the native-inputs and inputs so it does not change anything <ieure>FuncProgLinux, Your question is unclear. <FuncProgLinux>ieure: I skim through issues almost daily and I see that the team is supporting both docker and podman. I use podman at my job and find it does like 99.99% the same things as docker. Maybe this is a bad idea but the docker-podman package is (almost) aliasing docker to podman (even compose) but it exposes the socket to the known docker location I think /var/docker/docker.sock (?) <ieure>FuncProgLinux, Isn't podman already pretty much drop-in? s/docker/podman/ and most stuff works, yeah? <ieure>FuncProgLinux, So what does docker-podman get you? <FuncProgLinux>Not trying to boss anyone around. Just came to my mind that, if podman is already supported in Guix and such support is better than the docker one where compose is still the python version. Wouldn't it free some weight on the maintainers? <FuncProgLinux>ieure: It auto-aliases "docker" for you and exposes the podman socket to the known docker socket location. Because podman exposes it in another route by default. <gabber>yelninei: sounds like either fix would be great to have (: <ieure>FuncProgLinux, Okay. Well, it sounds like it also needs a service or something also, if it's doing sockety stuff. Guix is volunteer-driven, if you feel strongly that it should have this, you should send a PR. <FuncProgLinux>ieure: Yeah I thought so, I think shepherd does support sockets right? I still have to learn much about the Guix APIs but I do have a TODO list for things I wish to contribute :) <ieure>FuncProgLinux, I don't think shepherd is involved, unless docker-podman needs to run its own daemon. <FuncProgLinux>Another thing that's really good in podman is that, if you got a compose environment running you can just "export" a kubernetes pod file and load it either on a podman installation without compose or into kubernetes itself. <tesseract>ieure: yeah it looks like it works. but i installed from repo. it's fine i guess <ieure>tesseract, I think Mozzarella changed locations and IceCat needs to be updated to reflect that. <FuncProgLinux>ieure: Mmmm looking at the source it seems it already comes with podman ._. or at least most distros label it as a "subpackage" of podman <tesseract>ieure: i think the ublock version is older than the repo <hugohugo>podman is not 100% a drop-in replacement for docker, but for me it was good enough. Better actually. The main difference I noticed is who is the owner of the files in the guest when you mount a directory on the host in the guest. <FuncProgLinux>hugohugo: ah, been there. Lost like 4 hours trying to figure out why some logs weren't preserved on disk and it turns out it was a JS library called winston that had issues with such setup <hugohugo>In a docker container, the files have the same userid as on the host (which might or might not map to any user in the guest). On podman the files are all owned by root in the guest. (In podman it is fine to be root in the container, because the containers run as the local user.) <hugohugo>Both podman and docker can probably be configured to behave like the other (not sure), but it is not a drop-in. (And I prefer the podman default.) <FuncProgLinux>hugohugo: Mark all the volumes as ":z" on the compose file lol <hugohugo>But my actual goal is to get rid of all that and use Guix all the way trough, for systems, for containers, and for environments. With Guix there is effectively no difference between those things. But not yet there though <FuncProgLinux>I would love too as well, but I can think of some services that would take weeks to get right :s <yelninei>gabber: this works for me guix/guix#3161 (i stopped at cross curl). The x86_64-linux derivation is unchanged <FuncProgLinux>I tried to build tutanota from source (GPL3) last week...big mistake when something is Rust + Typescript :s <tesseract>guys, i can't watch videos on nitter with icecat. why? <tesseract>"JavaScript error: , line 0: AbortError: The fetching process for the media resource was aborted by the user agent at the user's reques" <ieure>tesseract, IceCat refuses to run non-GPL'd JavaScript by default. <ieure>tesseract, The addon is bundled. <dthompson>if you're gonna use icecat, immediately disable librejs. <dthompson>so frustrating that they still include that useless extension <ieure>I'm generally in favor of Free Software, but refusing to run non-GPL'd JavaScript is too much for me. <untrusem>doesn't ublock origin suffice the adblocking needs <dthompson>it's not specifically non-gpl'd js. it's any js that doesn't implement librejs's particular metadata system to associate licenses with js files. <ieure>kestrelwx, default-deny NoScript. <tesseract>ieure: do you know if i can enable it in about:config <ieure>tesseract, It's an addon, you disable it the normal way. <untrusem>isn't ublock origin + noscript overkill? <tesseract>ieure: i disabled all except "Privacy Redirect" and ublock addon <ieure>tesseract, Or you can use LibreWolf, I maintain that Guix package and daily drive it. It's IMO the only browser on the market worth using. <dthompson>getting firefox from nonguix is the easiest way to get a browser that works <ieure>LibreWolf is hugely better than Firefox IMO. <untrusem>I mean I liked the interface of umatrix rather than ublock origin <ieure>They remove all the garbage MozCorp keeps jamming into Firefox. <FuncProgLinux>ieure: I also use Librewolf! thank you for maintaining it :) <untrusem>damn, you must really like icecat tesseract <ieure>FuncProgLinux, Thanks for saying! You're very welcome. <dthompson>what's the lag between firefox releases and librewolf releases? <dthompson>that's typically the thing that makes all derivative browser variants not usable for me. too long between updates. <ieure>dthompson, Depends on the release, point releases usually within a day or so, sometimes a few days for a major release. <FuncProgLinux>Icedove is also good. Though idk why the IRC thinks I'm in thunderbird lol <kestrelwx>IceCat is on ESR so for the long time it couldn't run hoot. <ieure>I agree, IceCat has had multiple periods of being years out of date and riddled with security issues. <Chwoka>ha just last night i was having a hell of a time installing librewolf. (but it seems like more of a glibc problem or a pathing problem or something, reconfiguring) <ieure>dthompson, Yeah, they've been excellent at tracking upstream. My only gripe is that they only have rapid releases, I'd much rather maintain a slower-paced ESR. <untrusem>I haven't used it but people I follow like daviwil, you know if your are a crafter :P use them <FuncProgLinux>🧙♂️ so much knowledge in the chat I could begin start taking notes from the wiser people <untrusem>alse guix concensus site by noe is also using haunt <Chwoka>seconding that -- i'm just learning to scheme and such but my #2 goal after "configure guix and make my system look Pretty" is "make a personal website blog with scheme" and haunt looks perfect for that <dthompson>I need to release 0.4.0 sometime soon but haven't had time <untrusem>yeah, noe I should just steal your site code :P <noe>Don’t look too close, its got some pretty nasty hacks 😂️ <FuncProgLinux>Looking at the example haunt config, I see the only thing that's keeping me from learning other scheme variants, I love the guile keyword arguments <kestrelwx>FuncProgLinux: It has a Scheme spirit trapped inside. <untrusem>won't icecat have a different fingerprint as it has many extensions preenable, even if you disable them <untrusem>gnuzilla.gnu.org don't loads for me, gnu.org in down ig <FuncProgLinux>Reading the 005-GCD o.o I have to hurry with the mate stuff then <untrusem>btw what is the advantage of librewolf, I assume icecat don't phone homes to mozilla and other stuff right <untrusem>maybe it is not as hardened as librewolf <FuncProgLinux>I just didn't understand if Guix will turn into a point-released distribution with one release per-year or if two channels (stable/rolling) are kept <Chwoka>for me it's that i pretty immediately ran into some configuration options icecat simply would not allow me to use <Rutherther>FuncProgLinux: Guix will still be rolling and there will be yearly tags(releases) <Rutherther>FuncProgLinux: do you have issues on codeberg for what you're doing with mate? <Chwoka>throwing up "oh this system is managed by an administrator who won't let you change this about:config option" thing -- presumably i could just text-edit the file itself to do it but sheesh <ieure>untrusem, LibreWolf tracks rapid releases, has hardened defaults, and disables Firefox misfeatures. And comes with uBlock Origin bundled. <Chwoka>if i wanna use firefox's misfeature sync, that's my perogative :P <FuncProgLinux>Rutherther: I only submitted a PR but I don't want to nag anyone into reviewing it, since I've noticed the issues & PR's regarding python have some maintainers busy <ieure>Chwoka, You can use sync with LibreWolf. <Chwoka>yeah hence why i was trying to sync <FuncProgLinux>Also the things missing from MATE while being from the core MATE team, are extras or updates. The only "usability" issue left is pluma being incomplete, but other than that no real hurry. I'm tracking everything under #2997 <Chwoka>anyway my guix vm install looks Well And Truly Fucked after that reconfigure. increasingly looks like corruption because it's throwing up a "file system check on /dev/vda1 failed" error <untrusem>you could make an iso from your current vm, and make another vm so that you don't have to configure everything again <Chwoka>why would i make an iso from the vm that's busted? <Chwoka>i BARELY did any configuration so it's no big deal to start over <Rutherther>Chwoka: that's not a corruption. The configuration is just wrong, you need to use /dev/vda2 for root <Rutherther>it cannot run e2fsck on /dev/vda1, because /dev/vda1 is FAT <Rutherther>this is at least the case if you're using the QEMU image from configuration made from the guix channel <Rutherther>qcow2 is format used by qemu and I was refering to that as the QEMU image, yes <Rutherther>and you made it with guix system image -t qcow2? <Chwoka>nope i downloaded it from the guix manual through my browser outside of the vm <Rutherther>yesh, that was created by giux system image -t qcow2 <Rutherther>you currently need to manually adjust it to vda2 in the configuration <Chwoka>but we changed it around from its default because some things seemed screwy <untrusem>what is the text format that we use in package description? <tesseract>ieure: i used gnome web briefly. version 49 is usable <htgoebel>untrusem: texti. search for '@command{' for examples <untrusem>I wanna link to another package in a package description <Chwoka>rutherther: unfortunately i did not save a copy of the default config outside of the VM where i can quickly show it to you, though i can just use the vm snapshot feature to roll it back to a working condition in a sec <Rutherther>you do not have to, you should be able to just boot into the first generation <Chwoka2>resetting the snapshot (in the future i'd like to know how i could dig myself out of a "bournish@[guile-user]>" hole but i'm already in too deep i think) <untrusem>is it .texti or .texi, I am getting latex cheatsheet if I search `texi format cheatsheet` <Chwoka2>i'm gonna go outside and water some plants, be right back <Rutherther>but if you just sent the /run/current-system/configuration.scm, I am well aware on how it looks, it's also visible in the guix channel <tesseract>how can i check if a package installed or not? <tesseract>i am doing "guix search whatever" but it doesn't tell me if it is installed or not <Rutherther>tesseract: depends where... for the user profile you install to with guix package / guix install, it's "guix package --list-installed", for system it's "guix system describe --list-installed", for home it's same as with system, but home instead of system <tesseract>Rutherther: what happens if i try to install a package that already installed before? <Deltafire>since updating, icedove will no longer open .eml files <Rutherther>tesseract: it will update it, or keep the same if not changed <untrusem>btw what would take more space, using git worktrees or making branches? <Rutherther>worktrees, because you have the whole tree checked out multiple times <tesseract>guys, my icecat can't even play invidious media <tesseract>it tells "No compatible source was found for this media." <untrusem>btw suppose you made a pull req using agit workflow and your git directory got delete accidentally, then how would you force push to the same pull req, in the case of using a fork and pushing a branch to it and then making a pull req, atleast you have your worked saved, right? <Rutherther>tesseract: it's possible there is a bug after mesa-updates merge that has been done today <Rutherther>untrusem: the work is of course still saved even with agit workflow as it's on codeberg. You can just pull the ref of the PR, then you force push to the same topic <tesseract>Rutherther: do i need to do "guix pull" before "guix upgrade" everytime? <untrusem>ohh I am not familiar with ref and all, :P <Rutherther>tesseract: need to... I don't know... you guix pull to get new, updated package definitions. So if you want to pull to newest then yes <tesseract>Rutherther: so, current version of icecat is 140.3.1, let's say they released 140.4, do i need to do "guix pull" beforehand? <Rutherther>"git fetch origin refs/pull/XXXX/head", then the PR is at FETCH_HEAD, so you can check it out, ie. "git checkout -b my-pr-topic FETCH_HEAD" <Rutherther>tesseract: no, that was an answer for untrusem, forgot to tag <untrusem>thanks, I will translate it to magit workflow later <Rutherther>untrusem: note that you can change the .git/config to automatically fetch such revisions and get them under origin/pull/XXXX/head for example, that would be "fetch = +refs/pull/*:refs/remotes/origin/pull/*" "fetch = +refs/heads/*:refs/remotes/origin/*" <untrusem>ok its getting little confusing to me, I will read the git manual on this stuff <trev>untrusem: from my understanding, you still have the branch on your remote (since the PR is created). Just need to pull down from the remote <untrusem>yeah that what I get but just these +refs/pull/*:refs/remotes stuff is new for me <eikcaz>Is there a way to set keyboard layout from config for gdm+gnome+wayland? Earlier I set it from (gdm-configuration (xorg-configuration ...)), but now that only works if I manually select xorg gnome at login <ieure>eikcaz, Some keyboard layout stuff is broken due to libxml2. <ieure>eikcaz, I think it should be fixed once mesa-updates merges. <eikcaz>Ah right, I saw that here last night, right after I pull'd. Thanks, I'll pull again! <simendsjo>Hmm, isn't the "cgroup" group special? I'm inheriting an operating-system configuration, and get the following on reconfigure: `guix system: error: supplementary group 'cgroup' of user 'foo' is undeclared` <simendsjo>Ok, I notice many things break here. cgroup and libvirt groups, but uncommenting these gives `'file-system-/boot/efi' requires 'udev', which is not provided by any service`. I operating-system inherit doesn't work like I thought. <simendsjo>Ah, found it. I used `operating-system-services` which gave me an error, and I thought maybe the systems was appended and removed appending in the base services. I should have used `operating-system-user-services`. <nutcase>after mesa-updates got merged, I see some (leaf) packages not building anymore: obs, wf-recorder, texlive-biber (and nextcloud-client, already reported). The individual reasons seem unrelated. Is this known, or should I report via codeberg issues? <Deltafire>eikcaz: (keyboard-layout ...) should be all you need, this sets it in several places <Deltafire>hmm.. i'm finding multiple issues with icedove since updating <simendsjo>Memo to self: don't be an early adopter the next time a branch merges ;) <ieure>simendsjo, I still can't `guix home reconfigure' because the python branch that merged last week or so broke stuff in my profile :/ <ieure>Current system generation is an improvement, though. Caps is ctrl again.