IRC channel logs
2023-07-07.log
back to list of logs
<RavenJoad>Is it a known "thing" that sometimes Guix system tests (with marionette) will hang while trying to generate keys? <RavenJoad>Rather, while trying to gather enough entropy to generate the keys during boot. <GNUtoo>frog: is that supposed to take the host or the target package? <frog>GNUtoo: my guess is that it produces a path to the package built for the target platform. if that's not what you want, you can just use (gexp (ungexp-native packagehere)) <GNUtoo>For some reason the path it produce doesn't exist on the target <GNUtoo>And I've installed that package as well in the operating system definition <frog>thats beyond me. good luck. as a last resort, if the package is installed globally, you know its contents will be put into /run/current-system/profile/ <PotentialUser-31>I hate Gnome desktop, it is shame the ship mate 1.26 in last guix update <PotentialUser-31>I hate Gnome desktop, it is shame not ship mate 1.26 in last guix update <GNUtoo>It would require volounteers to update mate then <GNUtoo>frog: thanks a lot, it also taught me why I need to install packages in out/<some dirs> because otherwise it ends up there <GNUtoo>*everything ends up in that directory <PotentialUser-31>Also i really nervious that Guix will follows opensource-shames like debian does <gry>PotentialUser-31: what do you mean by that? <gry>PotentialUser-31: what is "follow opensource-shames like debian does"? can you please rephrase that? <cel7t>How does the Guix Hurd version work? Is it possible to run Guix with kernels other than Hurd or Linux (like BSD)? <gry>PotentialUser-31: not that I disagree, but I do not understand what you mean by it <gry>PotentialUser-31: I see. I think it should be unlikely, given guix is a gnu project <GNUtoo>cel7t: it's not that easy to do that, you cannot simply switch kernels like that. Like if I produce a binary for HURD, it doesn't run in GNU/Linux. <GNUtoo>cel7t: Many things are detected at compilation and the libc is the same though, so while not all package work on HURD, many probably do. <GNUtoo>cel7t: at least the packages that don't depend on linux features <GNUtoo>So I've to use a VM to test hurd stuff, Guix produce easy way to produce VM through guix configuration (childhurd) but I didn't try that yet <GNUtoo>I've only tried command line based hurd VM that I produce with guix system image, and that worked fine so far <PotentialUser-31>gry: I see a guix developer in a video were they talking, they put my nervious <GNUtoo>Also I tried hurd on real hardware directly but that didn't work due to SATA/ATA hdd naming issues and many missing drivers <GNUtoo>PotentialUser-31: nonfree software in Guix can exist by accident, but if it exist it's a bug and it need to be removed/replaced at some point. <GNUtoo>(like most software it's made by humans and humans make mistakes from time to time) <gry>PotentialUser-31: you can always fork into <custom name here> and keep it blob free. You'd be listed on the fsf endorsed distors list. <GNUtoo>It's more complicated, there is a tention between making users hardware work and staying fully free software. Nonfree firmwares could have been avoided long time ago, but the community at large started accepting them, so the solution I think is to have FSDG compliant distributions that refuse nonfree firmware and at the same time have people work on replacing nonfree firmwares with free software, <GNUtoo>the issue is that not many people work on replacing nonfree firmwares, though some are relatively easy to replace <cel7t>I hope it'll be possible to buy fully free hardware for cheaper soon <GNUtoo>So that doesn't make things easy with this tention <GNUtoo>So if it ships some nonfree firmwares it's a bug that needs to be fixed, and it can be reported, dealt with, etc <PotentialUser-31>The big problems is that there are already a huge fake-gnu/linux user everywhere, so they are pushing free operating system to opensource-blob oses, example in Purism PureO <GNUtoo>When they are found in FSDG distros, they are typically in less well known packages like some libraries for some hardware that use FPAGs that were OK before and that contain FPGA firmwares that depend on nonfree "compilers", or u-boot that added nonfree firmwares, etc <GNUtoo>AFAIK PureOS is FSDG compliant, they might have bugs though <GNUtoo>And note that Puri.sm hardware and PureOS is not the same thing <GNUtoo>You can install PureOS on an RYF laptop for instance <GNUtoo>(this is why there is multiple certifications: they are complementary) <PotentialUser-31>I really tired for what is happening in gnu, i have plans to move haiku. : ( Redhat, Ubuntu and already Libreboot are a same. <PotentialUser-31>It is shame that the FSF can not ship a raptor computer to trisquel team to release an iso of trisquel to raptop <GNUtoo>PotentialUser-31: Haiku probably contains lot of nonfree firmwares <PotentialUser-31>I know that Haiku already ship a lot of blob, but at least it has less fragmentation that gnu/linux-opensources oses <GNUtoo>Though this is #guix so maybe #libreplanet or some more general IRC channel would be more apropriate to find about the most free distributions out there <GNUtoo>ACTION never understood why "fragmentation" was a real issue <GNUtoo>Finding how to best collaborate between different group with different goals should rather be the way to go <PotentialUser-31>now Gnome 3 looks a freaking clone of IOS, more worse they depends of systemd <GNUtoo>Guix has gnome3 and doesn't have systemd <GNUtoo>Though it requires people working to make that possible as usual <PotentialUser-31>i know but extra work to get gnome from non systemd and many things does does not work well <GNUtoo>ACTION is unsure if gnome is really the issue here, it's probably more that systemd is not that modular so you can't pick the login system without the init system for instance <mirai>RavenJoad: from my experience that is usually a malformed service hanging shepherd <RavenJoad>mirai: That makes sense. After 2 hours, there should definitely be enough entropy on this desktop, so it had to be something in the VM's definition. <PotentialUser-31>sorry guys i nostalgy now for what is happening, raptor computer theys started like a free software things, now theys are selling a lot of firmwareblobs in userland by selling freaking expensive amd gpu for raptor. FSF never endorsed the blackbird PCB. So in resume i guess that FSF and GNU it almost dead. IMHO <mirai>PotentialUser-31: You're conflating several unrelated things at once <GNUtoo>Note that GNU started as programs that run on top of nonfree OS. Since there the free software community came a long way. It's a struggle for users freedom and things can get better or worse but the struggle continues. <GNUtoo>Also the RYF certification is somewhat recent and things were way worse before that <GNUtoo>Though at the same time the attack on users freedom also increased <PotentialUser-31>GNUtoo: Though at the same time the attack on users freedom also increased DO YOU KNOW FOR WHAT? <GNUtoo>In the 80's nonfree software didn't attack users as much as it does now for instance <GNUtoo>Now there are DRM, anti-features, surveillance, etc <GNUtoo>new restrictions are being invented (like BIOS that refuse to boot the computer if you change the WiFi card), etc <GNUtoo>So you kind of need more free software in more places as well <GNUtoo>like it's more crucial than before <PotentialUser-31>yeap, but i guess that redhat ubuntu and etc theys are control the gnu/linux to other address <GNUtoo>I wound't put all the distributions in the same bag <GNUtoo>The thing is also that RYF requires the hardware to work, so you can also use fully free software on hardware with features that don't work and workaround <GNUtoo>Example: ARM computers with builtin WiFi that doesn't work, and an usb dongle or no wifi <GNUtoo>ACTION will go to sleep now (/me was waiting for a package to finish compiling and it's almost done) <PotentialUser-31>yeap, however all my computer is free software, like Librem 14, Thinkpad x60, t60, x200, raptor blackbird, so i using pure free software in my computers like trisquel aramo, BUT i do not want to use SystemD anymore, then i want to move to GUIX, but guix still not ship mate 1.26 that is already old, i have more the one year tracking guix to ship <PotentialUser-31>So in resume FSF need to put more manpower to the future of GNU other wise gnu will dispel. <PotentialUser-31>of course if gnu dispel the freedom for the operating system will gone and all will be mess <oriansj>anyone know a simple and clean way to have /var/guix/profiles/system/profile/libexec/iwd run on boot as a standalone shepherd process? <RavenJoad>oriansj: Honestly, the easiest thing to my knowledge, would probably be to define such a service yourself. <RavenJoad>If I have a program used in a service that accepts a configuration file as a command-line arg, does it may any sense to put the config file in /etc/ using etc-service-type? Or just leave the file in the store and use the command-line to point to it? <ulfvonbelow>anyone else find that sbcl-bordeaux-threads hangs forever? <ulfvonbelow>looks like it's the CONDITION-VARIABLE test that's broken. <frog>actually, sometimes it does 31 for me too. it varies. <frog>or i guess "checks" could mean "asserts" and some of them only run if certain race conditions are hit? its SHOULD-HAVE-THREAD-INTERACTION <sneek>civodul, you have 1 message! <civodul>mirai: hi! i tried with 3.0 and IIRC it would still occasionally crash <civodul>so we'd need to check and investigate <ulfvonbelow>SHOULD-HAVE-THREAD-INTERACTION only shows up as the last one to be run when it's deadlocked because there isn't a newline to force a buffer flush after CONDITION-VARIABLE. <ulfvonbelow>see 010d579f460519691660e6b6e3ab2fcb30871b76 in bordeaux-threads git <ulfvonbelow>replying to frog re: sbcl-bordeaux-threads tests hanging <ulfvonbelow>anyway, we should either update it or patch out that test - it's always a bit annoying to have to try building something 5 or 6 times before it finally works, especially when you forgot to add a --keep-going, you have a bunch of long-running parallel jobs going, and the issues show up only under heavy load <f1refly>I'm trying to upgrade my user packages, and yet again the download of texlive-textmf is stuck on approximately 20kib/s, requiring multiple weeks to download the complete package. Is there anything I can do to speed up the download? Is it possible to specify another mirror for this packages source? <civodul>now, i wonder why you get this low a bandwidth <rekado>f1refly: where are you connecting from? <rekado>(for comparison: from an AWS EC2 instance in eu-central-1 we get around 50MB/s) <f1refly>when reading the instructions i didn't mind having 4GiB of texlive on my system at all. I thought it'd probably be easier for me to have everything there in case I end up in public transport trying to write a document <oriansj>anyone know why this: https://paste.debian.net/1285210/ results in this error: In procedure service-kind: Wrong type argument: #<<shepherd-service> documentation: "iwd service" provision: (iwd) requirement: <ulfvonbelow>literal genie answer: because a <shepherd-service> is not a <service>. <ulfvonbelow>actually useful answer: <shepherd-service> is the object type used to extend shepherd-root-service-type. You want something like (simple-service 'iwd-shepherd-service shepherd-root-service-type (list MY-SHEPHERD-SERVICE)) <ulfvonbelow>since guix services and shepherd services are distinct things. <civodul>ACTION renews disarchive.guix.gnu.org TLS cert, laments that we still have this certbot service bug <mirai>oriansj: there's no iwd service yet™ <janneke>ACTION starts to wonder what hurd-team is doing already ;) <civodul>i thought it was around the corner, but maybe i'll have to intervene? :-) <janneke>i'm not sure what's holding up jpoiret, i'm sure they're busy...but yeah <jpoiret>yeah, my bad, I haven't really had too much energy to commit recently, mostly been reading mail or answering easy questions. I worked a bit on the patches I have in my queue this week, but not hurd-team yet. <jpoiret>I can try to make room for it this week-end <janneke>jpoiret: there's no hurry, don't feel pressed -- but it would be nice! <janneke>ACTION hopes that simply squashing the squash! commits and removing the commit+revert commits should enable us to push the lower ~30 commits <janneke>up to and including gnu: gnumach: Disable builtin networking. <civodul>then there are a few long patch series that build on top of that <janneke>civodul: yeah, that's the first series by jpoiret (that has seen some changes, but the bottom of hurd-team should be OK) <janneke>(modulo the squash! and collapsing commit+reverts) <jpoiret>janneke, civodul: I'll push my queue later today and try to merge 63641 after that. Am I right in thinking that I can just take the commits from hurd-team? <janneke>jpoiret: yes sure, if you the squash! and collapse the commit+revert <janneke>ACTION kept separate squash commits for easy review of their modifications, by jpoiret <rekado>I don’t know how to tell if the tex-team branch affects it <distopico>Hi Guixers, QQ,what is the alternative for `journalctl -k -b -1` in GuixSD? to see previous boot dmesg or see previous boots logs? <nckx>distopico: No exact equivalent, sorry. Guix System doesn't have any fancy proprietary logging system. Most things just log to /var/log/messages (including dmesg kernel messages). Other things log elsewhere in /var/log. Some things don't log at all! The kernel logs a ‘vmunix: Linux version’ line to /var/log/messages which can be used as a (re)boot marker. <nckx>Basically, how things used to be before systemd. <podiki[m]>question: i'll propose changing godot to godot-lts (current 3.x release) and adding godot@4; should this be two (proposed) commits or done in the same one? <distopico>nckx: journalctl propietary? I guess no so far as I know, thank you for the information, I'll check there <bjc>systemd is (mostly) lgpl2.1+ <podiki[m]>on godot: only one dependent and godot-lts won't inherit from godot (too complicated changes in definition) <nckx>I'm pretty sure journalctl's file format is entirely proprietary? <nckx>Also, ‘[n]ote that the actual implementation in the systemd codebase is the only ultimately authoritative description of the format, so if this document and the code disagree, the code is right’ of course. <nckx>podiki[m]: I'd add godot@4 then rename godot@3 to godot-lts@3 in a second commit, but this is just a random second opinion of taste. <podiki[m]>nckx: so add new variable godot-4; then rename godot to godot-lts and godot-4 to godot? or just leave as godot-4? <nckx>Meh, one commit is fine. <nckx>I think that'll be the clearest diff (?) which is the point. <bjc>nckx: what do you mean by “proprietary”? looking at the source it seems like the journal parts are all under gpl 2.1+ <podiki[m]>(i'm sure one could do some inherit from new version and slice and dice all the changes back to godot@3's definition, but sounds like as much code as the actual package definition) <nckx>bjc: ‘Unique to and controlled by’, I guess. I was not referring to licencing. <nckx>I was going to ask if one could lock down a file format but I don't think I want to know. <bjc>i don't think you can lock the format, but you definitely can lock down the algorithms that speak it <bjc>at least, in the us. not sure about the eu <podiki[m]>sort of the situation with things like mp3 no? <nckx>Patenting algorithms required to read/write. <nckx>Anyway, I'll endeavour once again to try to remember not to use the needlessly ambiguous ‘proprietary’ around you commie free software types. <civodul>(sorry i only have short time slots at the keyboard...) <jpoiret>aha no worries, I don't have too many either <jpoiret>research has been a bit too exciting/time-consuming and I have other real life obligations these days unfortunately :( <juliana[m]>woohoo! my gamedev friends have been getting Very excited about Godot 4 and I've been meaning to try it; this is a good little kick in the pants reminder to do just that <podiki[m]>i haven't tried much of godot before (more tempted now) but seems really nice from my quick tests <podiki[m]>CI shows just a few aarch64 builds left for this branch <Guest28>How can I add an entry to /etc/hosts? <Guest28>podiki[m] perfect, thank you very much <Guest28>Shepherd 0.10.1 uses always 1 core that is pinned at 100% on my Raspberry Pi 4 with Libre Generic Linux. Since the shepherd is at 100% I can't run sudo herd status, reboot or shutdown. As I searched I didn't find any bug reports related to the high CPU usage <Guest28>My question would be basically if I should report it, since I am not sure with the hardware and non-free software is off-topic. <jpoiret>janneke: I should be looking at commits up to 27e57517817ba7cd5cf39291c74193060187e748 rigth? <Guest28>nckx: alright, seems like it is not yet reported. I am going to report it. Do you may know some useful debugging information I could include in the report? I don't know much about GNU Shepherd <nckx>Nope, sorry. But since you say ‘always’, it shouldn't be hard to reproduce if someone asks 😉 <Guest28>hmm could be, since I can't cross compile ntp my system time was at 1970. but I had the issue also with correct date <Guest28>though i am not 100% sure anymore, need to do some testing <jetomit>I can reproduce it by setting the system clock forward by a day, which will hang shepherd for about five seconds <jetomit>or two days for ten seconds and so on <jpoiret>nutcase[m]: you need xdg-desktop-portal{,-wlr,-gtk} along with pipewire set up <jpoiret>and if you're using icecat/firefox, you need to manually add pipewire's /lib to LD_LIBRARY_PATH <jpoiret>(at least that was what was needed last time I tried) <jpoiret>also now I think you also need to configure a screen picker for xdg-desktop-portal-wlr <jpoiret>yep just tested works for me with all that, good luck :) <nutcase[m]>grim -g "$(slurp)" - | swappy -f - allows me to select a region and take a screenshot from it <juliana[m]>Hey y'all, a friend of mine submitted a patch to fix a package they use and I offered to advocate it for them. I've already provided some review but if other folks (especially with commit) wanna take a look it's here: https://issues.guix.gnu.org/64521 <nutcase[m]>jpoiret: Would you mind to share your configuration? <nckx>juliana[m]: Do you know what the idea was behind the gstvar/pixvar stuff? <nckx>(Since as a friend you probably know their style :) <mirai>the documentations are unclear whether or not IWD should be started by its own service or if it should be autolaunched by NM <mirai>so its hard to say what the actual fix should be… but do let us know how iwd managed by shepherd turns out <mirai>if it works fine then perhaps its not too difficult to write the define-configuration for it <oriansj>mirai: you don't need network manager at all with iwd <Guest28>any1 recommendations for encrypting a directory which main purpose is to store personal documents like invoices and so on? <oriansj>and it appears that service works just fine (if you are comfortable with dhclient or ip commands) [so minor tweaks needed if you want it to automatically setup the ip on connect) <RavenJoad>What is Guix's preferred way of handling system config files used by services started by shepherd? Place the config in /etc/? Or leave the file in the store and have the program point at the config's store path?