IRC channel logs
2025-10-04.log
back to list of logs
<NegateThis>Hi, all. I'm a little confused about spliting a guix config into multiple files. As a very simple test, I've split my config into two files: ~/my-config/system.scm and ~/my-config/packages.scm, where packages.scm simply has a list of packages. I've got `#:use-module (my-config packages)` inside of my system.scm (define-module (my-config system)). <NegateThis>When I run `guix system reconfigure ~/my-config/system.scm`, I get errors regarding there being no code for module (my-config packages). When I add the "-L $HOME/my-config" flag to my reconfigure call, though, I get errors about gnu, guix, etc. modules not being available <QUL>Hi, I just noticed that there is a fundraiser going on and found out there is a Guix Foundation. Why is there a need for a Guix Foundation I thought FSF was there to handle such things or did you needed an EU-based entety? <FuncProgLinux>I noticed that as well. I don't see it listed so I think the answer is "no" but do you guys accept cryptocurrency (BTC) as donation means? :) <QUL>FuncProgLinux: I saw that BitCoin is accepted through FSF on the link is in the blog post. <RavenJoad>NegateThis: You need the -L portion so that Guile knows to sarch "$HOME/my-config" for (my-config packages). This is not Guix-specific, it is just how Guile does their libraries. <RavenJoad>You get the errors after adding the -L (I suspect) because you have not imported the symbols used in packages.scm that you need. Guile looks at each module separately, so if you #:use-module or (use-modules ...) something in system.scm, none of those are seen by Guile when it is looking at packages.scm. <QUL>I _assume_ yes since this exact link is linked by the fundraiser blogpost. <singpolyma>Right. I just don't want to accidentally end up in FSF general funds or something <encryptedwhispe->Is there documentation for using (git-checkout (source (url "folder-name"))) ? I am trying to figure out how to change the branch it uses <gorilla>does anybody have thoughts on using guix deploy on foreign distros instead of ansible/chef/puppet? <gorilla>encryptedwhispe-: you've piqued my curiosity. git-checkout is a defined function? <RavenJoad>encryptedwhispe-: I don't know if that function exists. But git-reference accepts a commit field and passes that to Git. So you can just give a branch name there, but you really shouldn't, since branches change commit pointers. <NegateThis>RavenJoad: Thanks for the tips. Specifying -L as a full path instead of using an environment variable fixed my issue, as well as a bit of following the instructions here: <QUL>Hi, I just noticed that there is a fundraiser going on and found out there is a Guix Foundation. Why is there a need for a Guix Foundation I thought FSF was there to handle such things or did you needed an EU-based entety? <gorilla>how can I lookup documentation for specific guix modules? <craigbro>are here riscv substitute servers? I am having gnu findutils-boot0 fail on a test (test_rwlock) on my orangepirv2, and was hoping there were binaries available on a substitutor <gorilla>does guix deploy work on foreign distros? <mroh>gorilla: no, guix deploy pushes an operating system configuration to the target system; managed-host-environment-type is for machines that are already running guix system. Perhaps you could define your own type, but I guess it would only push some user profiles or so?! <RavenJoad>craigbro: CI builds them, but I'm not sure if they are available as substitutes. <bdju>My guix gc has been running for over 12 hours and it's up to about 300GiB it'll delete. I wonder how long this is gonna take. Things are pretty hot and laggy while it's going. <bdju>I worry if I canceled it now that it wouldn't actually delete any of this and it'd have been a waste of time, but I wanna be able to use my computer at some point. <bdju>Yeah. Might be my first GC on this install ever, or in years. <bdju>BTW, what exactly is it doing that is so computationally intensive? I assume it's not just rming the files. <RavenJoad>bdju: That feels a little long for just 300G, unless this is a spinning disk. If you want GC to go slower, you may want to use -C/--collect-garbage=MIN, to collect at least MIN bytes. <RavenJoad>bdju: It rm's, but also does some bookkeeping in the SQLite database. <bdju>I did have weird errors about the db and about being out of space (which was untrue as far as I could tell) before this. <RavenJoad>Yeah. Things get really weird when the disk gets full (even if it isn't). Try deleting things slowly. <bdju>Well, it's still going. I'm scared to stop it. <RavenJoad>I think the steps to make things atomic is move to a "recycle bin" then rm, which may consume more space. <bdju>I did `guix gc --delete-generations=1m` since it was in my shell history. <bdju>Some sort of overall progress bar or ETA would be nice. <bdju>I tried to run `du -sh /gnu/store` Figuring if I got the total size I'd know the maximum it could possibly delete as a sort of progress indicator, but somewhat predictably it's taking ages and seems stuck. I may not have the spare processing power to do that. <RavenJoad>I think the lag comes from the SSD literally just handling a boatload of I/O requests. If you cancel this gc, the system should return to normal usage. <RavenJoad>htop might show the I/O waiting time as another bar on the CPU usage graphs. <bdju>Load average: 10.08 9.30 9.34 <bdju>Oh, I found a meter I can enable for Disk IO. <bdju>Disk IO: 40.5% read: 11.1MiB/s write: 183MiB/s <bdju>That does sound kinda high. It's a SATA SSD. <drewverlee>How do i tell herd to pass args through instead of trying to interpret them? e.g we have a shepherd action defined as (procedure #~ (lambda (a . b))...and when we call it `herd action-name a b c` where c is an option we want passed along, herd seems to grab c and think it's for it, spiting out: `herd: error: unknown option 'c'`. ideas or input welcome. How can i learn whats going on here? <RavenJoad>bdju: Sounds like it's working, albeit very slowly. <bdju>The drive is only 2TB and it's not like it's all taken up by the store, so... I guess the longest it could take is like 3-6 more days maybe. Which would be extremely awful, of course. <RavenJoad>You can always cancel it and try a smaller GC? <bdju>Is there any way to lock in my progress? Will it delete these 300GiB it's talking about if I cancel it? <bdju>df -h only shows 151GB free on / so it has clearly not deleted 300GB of stuff for real yet. <RavenJoad>You shouldn't need to lock anything in. Things get moved to the "recycle" spot and marked in the SQL database before the removal occurs I believe. <RavenJoad>Also, "guix gc" ends up double-counting removed stuff because of hard linking. <bdju>Well, my drive wasn't full before, so the 150GB is not from this, it had that much free space already yesterday. <bdju>So if I C-c out of this, what will it do? Will any more space be freed or not? Will it (this 300GiB) be freed on next gc even if I do a "small" one? <bdju>I just wanted to clean up the store before trying that verify thing again because it was taking too long. I was never out of space. I'm so far removed from the original problem now. Plus I was just trying to upgrade my packages. I don't understand how it all went so wrong. <bdju>I'll be in bed in a few hours and then it could run all night again, and then I'll likely be out of the house with friends on Saturday, so I could leave it all day. I just don't know if that's enough for it to actually finish. <RavenJoad>If you C-c, things should stop gracefully. No more space should be freed, unless Guile itself has gone haywire and a process remains alive somehow. Stuff should continue to be freed as you go GC-ing. Guix's GC is identical to heap GC. You can mark something and sweep it later. <RavenJoad>I've never seen a GC on an SSD take this long. <bdju>It just doesn't appear any space has actually been freed yet, so I figured it would do it all at once at the end. <bdju>Alright, I'll cancel it, I guess... <bdju>Can not save configuration to /home/brad/.config/htop/htoprc: No space left on device <bdju>Why am I getting stuff like this? I just don't understand. My / is nowhere near full! <bdju>I canceled the gc, I guess it had been 20 hours. I had thought to run it with `time` before it. <bdju>I can touch a new file in my homedir and I can delete it. I don't think my disk is full or readonly in reality. Something horrible went wrong when I canceled a guix build process yesterday. <RavenJoad>I did an 80G GC today and it took <3 minutes. So 20 hours is something very weird. <bdju>Speaking of something going horribly wrong, my CPU is still super hot after stopping the gc, I suspect there is a stray process somewhere still. <bdju>Should I just kill these processes? <bdju>Is there anything I can restart to make guix stop being crazy or do I need to reboot my whole machine? <bdju>This is similar to what happened when I canceled the chromium build yesterday. I don't think C-c is doing what it should for guix commands for some reason. <bdju>One of those I was able to kill normally and the other I had to kill -9. My CPU is still super hot, though. <RavenJoad>Yeah. Guix commands at the shell are actually socket requests to the daemon. (You saw that with the "guix-daemon 25526" process). So C-c may not always kill everything everytime. <bdju>Can --collect-garbage take a unit or do I need to figure out how many 0s to put? <bdju>I checked the man page and it was not helpful. <bdju>I guess I'll just close stuff and reboot. Nothing else seems to be working in the slightest. <bdju>Ugh.... `sudo reboot` is not even working. This is so screwed up. <bdju>Still stuck. Gonna hold the button in. <bdju>I didn't, it was taking significantly longer than it usually does. <bdju>I think it's all hosed. Maybe something in the hardware broke again. I'd just replaced the CPU fan in the last month or so, had an SSD in it fail once so maybe something similar happened again, some people have been telling me to run a memtest so maybe it's bad RAM. I guess I'm just gonna use another computer for a bit. <lsl88>I was an outreachy intern dec 2018-march 2019. trying to come back ☺️ <untrusem>so a program I am packaging which uses meson build system have a subproject, how to deal with that I have not package a meson application yet <untrusem>lsl88, awesome, how was your experience at outreachy? <mroh>untrusem: if the project uses git, you can try to fetch w/ all subprojects, see "guix edit chatty" as an example. Or you can try building w/o the subprojects (and maybe use other inputs instead), see termite as an example. <untrusem>btw this is the project I am trying to build, I had send a msg about this previously <mroh>untrusem: this project looks more like a python-build-system, no? the error in the build.log is from the other project you pastet, havent looked at that. <untrusem>I tried the flake.nix which itself uses python venv to install the project in the end, its so mixed up <untrusem>so I think, I would need to package the ignis-gvc one, unbundle the Gnome's libgnome-volume-control program and remove the subprocject gvc from ignis and provide ignis-gvc as a input to it <untrusem>mroh, can you give pointers to unbundling those, I have not that yet <mroh>untrusem: you already unbundled that volume-control thing, no? But it doesnt build. I am not sure, but I would try the other route and check out the original project with (recursive? t). Then, I guess you need to split the build-phase in two, one meson-build for the now checkouted subproject and one python phase. But this is just another (happy) hack to make it work. In the long run, you are right, unbundling is better. For that make the... <mroh>... subproject build and then, I guess you might have to build the root project with --skip-modules gvc (or so). you are on the right track, keep on hacking ;) <untrusem>ok thanks, that subproject uses another subproject so then I would need to make a package for that too, lol its a cycle <untrusem>anyway thanks mroh, I know I would eventually do this :) <mroh>subprojects are evil, but hey, the universe was build to hack it, so... ;) <Chwoka>for "guix system build" i'm getting "wrong number of arguments for command build" but the reference manual doesn't uh, tell me anything about what those arguments should be <Chwoka>yeah that would make sense. i figured unspecified it would just use, the current system i was on to save me the trouble of hunting down the configuration.scm again <Chwoka>but that's okay i'll just do that <characteristic>“Build the operating system described in FILE” seems obvious to me <Chwoka> Build the derivation of the operating system, which includes all the configuration files and programs needed to boot and run the system. This action does not actually install anything." <Chwoka>you can see how i got the idea that it would simply take the operating system i was currently on from that phrasing <Chwoka>what are you pulling your quotes and synopsis from? <characteristic>guix-system(1) is pretty sparse, it only has edit, reconfigure, roll-back and describe <Chwoka>i think build is... not the right tool anyway. (i'll circle back into here when i think i have more of a grasp on what i want to do step-by-step) <Chwoka>actually i might as well just recap in case my overall plan is totally cockamamie and needs to be shutdown <Chwoka>basically the vm system image i downloaded from the gnu guix ftp is somehow a bit corrupt, it seems -- barely usable, keeps crashing at increasing frequency, throws up weird errors, etc etc i went over this a couple days ago <Chwoka>but downloading a new vm system image from the site hasn't worked because that takes like 12 hours and my attempts at downloading it have failed like three times. network hiccups, laptop going to sleep, etc <Chwoka>so i figured that it would be faster and, since local, more reliable to run "guix system image -t qcow2" on a configuration.scm that i copied out of the VM's configuration.scm <apteryx>kestrelwx: hi! I'm looking at your jami update :-) <apteryx>was wondering if the problem you found with JAMICORE_AS_SUBDIR had been reported upstream? <apteryx>OK! Is this something you could do? I like to put an upstream issue URL in patches we carry <apteryx>did you check if there were updates to jami/daemon/contrib ? Things like dhtnet and opendht are regularly updated. <kestrelwx>I'll see what needs to be updated and open issues for FFMpeg and the JAMICORE thing. <kestrelwx>Should I report the emoji picker? Or that a packaging problem here? <ColdSideOfPillo`>I'm trying to package something (in a personal channel) using `cmake-build-system', but the build keeps failing during the `configure` phase. <Chwoka>"guix system: warning: multiple 'nss-certs' packages found; 'nss-certs' is now included by default in '%base-packages'; ensure it is not explicitly listed in the operating system 'packages' field" <Chwoka>should i edit the configuration.scm i pulled out of the vm or just ignore what this warning is telling me <Rutherther>Chwoka: better to get rid of warnings, so I would follow it. Having a lot of warnings can make the important warnings less visible <Rutherther>ColdSideOfPillow: so what's the issue? Why don't you supply the version as it's asking you to? <Rutherther>as with anything you define with cmake, you pass it to the cmake args. Do you know cmake? maybe good to read its manual at least partially <Rutherther>ColdSideOfPillow: that doesn't apply at all. There is no git repo when the version is being obtained during guix build... <ColdSideOfPillow>Rutherther: Yes, I know the workaround doesn't apply, but I'm saying that they knew about it. <apteryx>having a conditional code path that depends on an environment variable in a manifest.scm file is a bad idea, right, because of caching? <apteryx>it'd cache the original state that depended on the env. var. at this initial time, and later wouldn't react to a change of it. <apteryx>kestrelwx: I'm not sure, if you install some emoji font like noto-emoji and it doesn't work, I think it's a bug, or worth checking <kestrelwx>The Emojis appear fine, it's the picker that can't appear. <apteryx>I was trying to check now, and couldn't figure out how to add emojis ^^' <kestrelwx>Me and another person experience this with Sway on Guix. <apteryx>probably related to Wayland more than Sway, if anything <apteryx>looks like they're targetting 6.8.3 or newer, not too far <kestrelwx>When you hover over a message a menu appears, if you click on the hamburger, you get to place emoji reacts and then the default 3 appear but the "more" menu emits a type error into the console. <kestrelwx>default [1759580736929][WRN][7f38a3857f00]: qrc:/commoncomponents/ShowMoreMenu.qml:102: TypeError: Property 'open' of object EmojiPicker_QMLTYPE_419(0x7f387c262650) is not a function <apteryx>kestrelwx: ah, it's because it relies on webengine <apteryx> var component = WITH_WEBENGINE ? Qt.createComponent("qrc:/webengine/emojipicker/EmojiPicker.qml") : Qt.createComponent("qrc:/nowebengine/EmojiPicker.qml"); <apteryx>well, it doesn't, but the component differs, and apparently the second one lacks an 'open' method? <untrusem>GC Warning: Out of Memory! Heap size: 18 MiB. Returning NULL! <kestrelwx>apteryx: For JAMICORE... specifically the typo in the CMakeLists.txt? <untrusem>this happens when I try to do `guix build python-pyqtwebengine` <apteryx>kestrelwx: re emojipicker, the nowebengine version looks like an empty stub, so I guess it's a WIP <apteryx>could be worth reporting an issue to signal there are users building with WITH_WEBENGINE=OFF ;-) <kestrelwx>Makes sense that it works at all in other issues. <apteryx>did the emoji picker worked in prior versions? <kestrelwx>No, it's broken in June too. But on their tracker people do get to see it. <apteryx>they're probably using a build with qtwebengine <untrusem>is that guix gc warning harmless as it builds the program? <kestrelwx>apteryx: They have stuff on master undone by mostly unrelated commits somehow, haha. <QUL>Hi, I just noticed that there is a fundraiser going on and found out there is a Guix Foundation. Why is there a need for a Guix Foundation I thought FSF was there to handle such things or did you needed an EU-based entety? <craigbro>because the FSF doesn't pay for their non-fsf controlled infrastructure, is my understanding <QUL>craigbro: Bot attacks? <craigbro>QUL: amongst other things, but perhaps the biggest issue I see (as a observer and user) isthat misfit between what they wanted to do with the contribution and distribution process, as well as the need for guix specific build and distro sources. <craigbro>hence the move to codeberg, and the need to raise their target amount to pay for the guix specific infra <craigbro>I don't think any diss or split or break with the FSF is implied by the existance of the Guix Foundation <QUL>craigbro: Wdym by that? Who wanted what? I understand that the Guix Foundation does very needed things I just don't get why was the descision made to create a sepparte foundation rather than expand FSF build resources seems like more work to me. <kestrelwx>I'd bet more work is getting anything moving at FSF. <craigbro>guix maintainers/team/contributors wanted to move to a "forge" <craigbro>see the mailing list archives for the discussion of the proposal and the voting etc.. <QUL>craigbro: I mean yeah thats the switch to codeberg this year. An awesome choice, but what has it to do with creating a separte Foundation 8 years ago? <craigbro>owning build servers, as it states on the info page of the foundation <ekaitz>i don't think moving to Codeberg has anything to do with the Guix Foundation <ekaitz>and when we talk about infrastructure cost, it doesn't count Codeberg, because I believe we don't pay them money <untrusem>off-topic from this convo, but really liked your interview with prot ekaitz :) <ekaitz>untrusem: thank you mate, I appreciate that <craigbro>same here, BTW, especially the point made about supporting users with less resources <ekaitz>back to the foundation: anyone can make a foundation which has helping guix as a goal <ekaitz>the FSF doesn't have the monopoly over guix or anything <ekaitz>and they don't intend to do so, either <untrusem>I believe the views on his videos will be higher we we libre peole were not using our weird youtube clients in terminal, emacs and other alternative frontend <ekaitz>QUL: back to your question, the FSF is like a bank for Guix and other projects <craigbro>ekaitz: Indeed, looking at the guix foundation ledger I see no contribution to Codeberg in 2024, only payments for guix days, and hosting of build servers in Aquilnet and Hentzner <ekaitz>they collect money for the projects and then pay for the bills the projects have <ekaitz>but the FSF is not a "free money" institution, the money they collect for the project is specifically sent to the project <untrusem>but I remember in the document where the move to codeberg was proposed, something related to atleast supporting codeberg <ekaitz>they don't assign money that was donated to the FSF in general, i believe <ekaitz>untrusem: I proposed supporting codeberg, with money and/or participation <ekaitz>but they are also a fundation that is commited to help to projects like ours <ekaitz>so I think we'll be fine for the moment, and if we have some spare money, or development time, or people that want to join Codeberg to have some voice in their decision process, we could do that later <ekaitz>as far as I know codeberg is doing well for the moment <ekaitz>way better than savannah was doing, honestly <untrusem>btw, people have been praising the new banner on the website that is done without any js, all thanks to luix felipe. <untrusem>how do I link to a specific comment from lobsters <ekaitz>clicking in the comment number or something I believe <craigbro>QUL: my apologies for confusing the issue of support for codeberg, it was a misremembering of mailing list conversation <untrusem>off-topi but this warning i harmless right <untrusem>GC Warning: Out of Memory! Heap size: 25 MiB. Returning NULL! <kestrelwx>apteryx: Should I submit DHT packages as separate commits? <apteryx>if they don't cause breakage, you could bump their version in separate commits yes <QUL>craigbro: Yeah no problem, so is anyone here a member of the Guix Foundation cause I sence a lot confusion. I do not want to be annoying but im just very curious about the descisíon. <craigbro>ekaitz: oh, do now that you are here... riscv subsituters? I am trying to build on my orangepirv2, and hitting a test_rw fail in findutils boot0 and was hoping to bin substitute thru it <apteryx>untrusem: strange, I'venever seen that here I think <QUL>ekaitz: untrusem mentioned you were on some podcast. Can somebody maybe send me a link i would like to check it out <ekaitz>craigbro: I just worked on the bootstrapping, and it's not even merged yet so imagine how knowledgeable I am about that hehe <apteryx>untrusem: does your machine have little ram or something? <craigbro>ah, so there are no public substituters for riscv? <ekaitz>craigbro: in general, riscv substitutes are a pain in the balls, because we do need machines to build them, and we don't have that many <craigbro>well, I have two orangepirv2s that are available... <ekaitz>in fact they don't even exist that many riscv machines that can do all the building work <apteryx>should be more than enough. some arch other than x86_64 perhaps? <ekaitz>craigbro: efraim has all the knowledge about riscv substitutes you might need, and he's probably happy to have some more hardware working on the builds <untrusem>guix describe shows this commit apteryx. <apteryx>untrusem: interesting. I don't know why it would happen; but that's not normal. <apteryx>You could report an issue, someone might have an idea of what is going on, or reproduce. <untrusem>wait a sec, I am sending a paste, with guix describe <ekaitz>apteryx: I saw the merge in guix-artwork and so on, thanks for doing that (and sorry for the branch name LOL, i didn't do that on purpose) <craigbro>ah, mayve I'll spin up a riscv qemu and make a local substitutor, provided qemu doesn't hang on the same test <ekaitz>craigbro: qemu also does some weird things sometimes, good luck! <apteryx>maybe some channel could be at play, I'd try to reproduce without channels just to see <untrusem>I don't think guix gc verify would do something in this case <ekaitz>QUL: (yes, that's me, and yes my english is that bad) <QUL>ekaitz: I really don't care about english proficency. Nothing can ruin a talk about bootstrapping! <untrusem>apteryx, I made a guix binary with the just the guix channel with that particular commits, still same warning <ente`>[ 1.200314] init[1]: segfault at 3fff00 ip 00000000004d4953 sp 00007ffc63740ef0 error 4 in guile[d4953,401000+201000] likely on CPU 2 (core 1, socket 0) <ente`>[ 1.200373] Code: 8b 05 d1 ec 34 00 48 83 c4 28 c3 0f 1f 40 00 48 8b 05 c1 ec 34 00 48 2d 00 01 00 00 48 89 05 b4 ec 34 00 48 8b 05 ad ec 34 00 <0f> b6 38 e8 65 a5 ff ff 48 8b 44 24 08 48 8b 15 99 ec 34 00 48 05 <untrusem>if you see this after grub screen, its common, if you use encrypted disk <ente`>don't think I use an encrypted diskk <ekaitz>ente`: i have the same segfault, too, and I don't encrypt <ente`>well, I'll just accept it for now <ente`>since my system seems to work otherwise <ente`>just had this show up for the first time after the update <Chwoka>okay guix in the vm either went to sleep or crashed and now i'm at a login prompt screen but the account doesn't have a password and just putting in nothing and hitting "enter" does nothing <ente`>anyone know how I enable cups-browsed? (cups' "show printers shared by other systems" feature) <untrusem>Chwoka:, the default user is root without password <mroh>interesting, I've seen that error in a cuirass build log file. <Chwoka>yeah sorry when i say login prompt i don't mean terminal i mean some GUI type deal <ente`>I'm not finding anything about guix and cups-browsed <ente`>don't make me go to the windows computer to print :( <kestrelwx>Chwoka: Can you switch to a different tty? Ctrl-Alt-F2:F6 <Chwoka>kestrelwx: that switches my tty outside of the vm haha <kestrelwx>Ctrl-Alt-G for to toggle QEMU grabbing input exclusively, I think. <Chwoka>yeah that does not do the trick for this cmd unfortunately <Chwoka>i've never used this feature before -- is it basically workspaces but above the window-manager level? <Chwoka>if so, could be a handy way for me to try out a bunch of different desktop environments <kestrelwx>Yeah, you can have couple Wayland sessions that way. <kestrelwx>I think with X it's a bit involved but probably can do it that way. <Chwoka>doesn't help me with my current roadblock unfortunately <kestrelwx>`Ctrl-Alt-2` to get into QEMU console and then `sendkey Ctrl-Alt-F2` there. <Chwoka>not sure how to progress from there to get back into, if not the desktop environment i was in, at least a desktop environment, but for now i'll just get back to the thing i left my computer doing when i walked away, guix pull <untrusem>apteryx, I tried with just guix proper channel too, same warning <untrusem>currently running guix gc --verify="contents,repair" <untrusem>damn, I am getting download speeds of 23KiB/s <kestrelwx>apteryx: OpenDHT version bump maybe needs newer GnuTLS, though maybe it's the next test failing, I'll check tomorrow or Monday. <Noisytoot>I want to package ectool but I'm blocked by this <untrusem>apteryx, I am not getting the warning at guix gc and guix pull, but in package, build, home commands <QUL>Hi again, I am sorry to spam with this question but I atleast try space it out and make sure not to interupt any ongoing discussion. Does anybody knows why was Guix Foundation setup cause I thought the functions that it acomplishes were FSFs job. If anyone here is a Guix Foundation member or has an E-Mail adress or any other means I can reach them I would be very thankful. <untrusem>apteryx, I rebooted and I am not getting these gc warnings <encryptedwhispe->Hi again folks. I am running into a strange issue where I have automake, autoconf, pkg-config defined as native-inputs, but if I make a pure guix shell autoconf cannot be found (as well as make, etc). Any idea why this would happen? Sometimes it does not happen and everything works. I'm using a pure shell because a guile module isn't properly found without making it pure. <Deltafire>QUL: pretty sure your question was answered already.. to pay for the guix infrastructure, build servers etc <QUL>Deltafire: This part is clear to what I do not understand is why was Guix Foundation set-up when with my understanding this was the job of the FSF. <Rutherther>encryptedwhispe-: how does your guix shell command look like and what package has those native-inputs? <ekaitz>QUL: I already mentioned a little bit about this: the FSF does many things, but supporting Guix is just one of them. The Guix Foundation only does one thing: Guix Guix Guix <encryptedwhispe->Rutherther, guix shell -f guix.scm -C [or --pure]. The package is one that is defined in the folder I'm in. <untrusem>sneek: later tell apteryx: the guix gc warning appeared again when I did `guix home reconfigure` <ekaitz>QUL: And FSF's job is not exclusive, you could make a new foundation tomorrow for it <Rutherther>encryptedwhispe-: so the guix.scm is returning that package? Are you sure that is the case? <ekaitz>QUL: what is what is unclear to you, specifically? what part makes you uncomfortable? <encryptedwhispe->I get autoreconf: command not found, Rutherther if I try to execute autoreconf in the shell <encryptedwhispe->Rutherther, it is also worth noting that I am running on a foreign distro. <trev>encryptedwhispe-: what if you put them in propagated-inputs instead? <QUL>ekaitz: Oh ok so now I think I am starting to understand. I thought that FSF and GNU were divided into projects and were people were there to support these projects, but from your message it seems that it more of everybody does many things at once and GNU Guix as a distro and a Package manager needs dedicated effort. Did I got it right? <ekaitz>QUL: yes, but it's more than that. The FSF is in the US, and the Guix Foundation is in EU (France). <ekaitz>and there are way more reasons than that one <QUL>ekaitz: Oh that was one of my first ideas to have a EU Based Foundation. Probably to easaly hire EU-based devs right? <Rutherther>encryptedwhispe-: I disagree, I don't think it's worth noting you're on a foreign distro, Guix shell should work the same way on Guix system and on foreign distros. As for your issue, guix shell -f will give you the package itself, not its inputs, you would have to do guix shell -D -f guix.scm to get the inputs. This is a distinction between using "guix shell" with implicit guix.scm and explicitely with -f. <ekaitz>QUL: I don't expect the Guix Foundation to hire anyone in the short term... The FSF doesn't hire that many devs either... <Rutherther>encryptedwhispe-: I would use a manifest with package->development-inputs rather than guix.scm, it makes it clearer on what's happening <encryptedwhispe->Rutherther, I have also tried with --development to no avail. Experimenting with propagated-inputs though. <ekaitz>QUL: I think you believe these project have way more funds than they actually have hehe... <Rutherther>encryptedwhispe-: did you put the development flag BEFORE the -f flag? <kestrelwx>`guix shell -D -f guix.scm --pure` works just fine for me. <encryptedwhispe->Rutherther, Ah, interesting. It seems the -C has to be at the end, and the -D needs to be before -f <QUL>ekaitz: I don't think it is the case. I just mentioned hiring as one aspect I can think of that would be easier with an EU-based Foundation. But why is there a need for an EU based foundation? If it is not to large of an answer to be typed here of course. <encryptedwhispe->I was lumping the -C with the -D Rutherther. Thanks, this may have fixed my issue. <kestrelwx>You can have -CDf guix.scm if you really want. <kestrelwx>The -D option affects the next entry specifically. <FuncProgLinux>now I know what was wrong with a personal go project i was working on with guix shell as a development environment ._. <ekaitz>QUL: some people are not comfortable with sending money to the US. But also interacting with the US every single time is a slow process... Some people want to find alternative ways to help Guix in their own ways... There are many reasons to have new foundations. Also, some people are not happy about how the governance works in the FSF. <dgr>how to best reinstall guix after an upgrade to Debian 13 (trixie)? <QUL>ekaitz: Oh ok makes sense. I thought the reasons were more legalistic and technical lol tbh. But it does not matter thanks for claryfing it. <ekaitz>QUL: well, making a legal entity is mostly about legalistic things, isn't it? <ekaitz>if it was just for technical reasons, an uninformal assembly of friends would be more than enough <QUL>ekaitz: Yeah but I was reffering for the reasons for their existance. With legalistic I was reffering to something like: Tax reduction for donations comming from the EU, applying for EU Grants and etc. <FuncProgLinux>dgr: wasn't guix scheduled to be removed from the debian repos? :( <FuncProgLinux>I also want to donate to guix directly but I get chills when stripe wants my name <ekaitz>FuncProgLinux: 1) yes, debian removed Guix, but you can still install it with our script 2) you can ask the guix foundation for an IBAN if i'm not mistaken, or you could use the FSF to donate to Guix (they accept some other payment processors, including bitcoin) <dgr>FuncProgLinux: yes, the package is gone and I try to reinstall with the shell script, got it to overwrite but is there more to it? <bdju>I left my T440p off overnight and now hooked it back up again and it was able to boot. It's not saying the disk is full anymore, free space looks the same as before. <FuncProgLinux>ekaitz: I'll see those options, stripe is being a crybaby with the captchas lol <trev>FuncProgLinux: you could use guix to bootstrap bitcoin and then send a donation using your own node <trev>(just kidding, but it's true) <boringrgb>hello everyone, I am new to Guix and very excited to learn it. I am experiencing extremelly long waiting times for a guix pull operation though, and I am wondering if there's a way to activate some caching, or perhaps point at a mirror nearby? <luca>Are you using savannah or codeberg <boringrgb>what I see in the terminal is savannah, I have't changed any defaults. do you suggest I change my channel to pint to codeberg? <luca>other than switching to codeberg there's not that much to do. guix pull is slow for everyone for the first time <boringrgb>ohwow, @luca, changed it and this is about 4x faster... thanks :) <luca>Personally I'm a firefox4life kinda guy. ad blocking and tree style tabs are my jam <tesseract>the only thing keep me using firefox is about:config <luca>From what I've looked around nothing comes close to ublock origin. So I'll stick to firefox for the time being <gcarlos>hi guix, i'm facing a problem: when I have both gcc-toolchain and cross-gcc-i686-w64-mingw32-toolchain on the same env, I can't compile thing with the default gcc (the one from gcc-toolchain), as it gives me the error: "/gnu/store/ka2lskkzmgp04mznfj471p6as5j4rxin-profile/lib/crt1.o: file not recognized: file format not recognized" <gcarlos>that crt1.o is a PE file, used only for mingw cross compiling <futurile>QUL: it's basically two things (a) FSF is in USA, and many of Guix's infrastructure/participants are in Europe, this makes it difficult for the FSF to handle things (b) FSF has their own priorities, it tends to mean that it's faster for us to handle things through our own Foundation <gcarlos>but the gcc is trying to use it and idk why <futurile>QUL: there's definitely things we could do with a EU-based non-profit that couldn't be done with something else - for example apply for grants - but realistically the total size of the foundation is tiny - 40 members, and we're trying to raise a small amount of money - so there's probably no particular advantage other than it's easy for us to pay for things, and it's totally dedicated to Guix <nutcase>how useful is the donation of computing power (metal or virtual)? Does integration of particles mean more costs than benefits? <jhackdee>[Network Announcement] AMERICA NEEDS YOU - Secure our borders, get a signing bonus, enter the federal workforce and subvert the swamp from within, no degree required. https://join.ice.gov <futurile>nutcase: I depends is the unsatisfactory answer. So bourdeaux is mostly hosted in Chris's house, and then others also have machines that are part of it. I think it would be best to get away from that as it's finicky and we have reliability issues. But, it's what we have and I can't really see it changing. <futurile>nutcase: I definitely think the overall resourcing of the build farm is a big problem. We need two to cryptographically assure ourselves is my understanding. But the total amount of horse power we have is too small which is part of the reason we struggle to integrate branches quickly (they take days to build) <futurile>nutcase: that's not a universal opinion, I believe Chris thinks that's not the root cause for example and they have the most insight. I'm just one of those people who's willing to throw a lot of machines at a problem in some cases <QUL>futurile: Thank you for explaining it to me!! Yeah I saw the blogpost thats how I found out about Guix Foundation in the first place!! Thank you again! <futurile>QUL: no worries, someone pointed out to me that we never introduced Guix Foundation on the Guix site. Probably should have done that <sigh>. Anyway, if you're thinking of donating, donate through whatever method is good for you - donations to FSF or Guix Foundation all count ;-) <nutcase>futurile: wdym by "cryptographically assure ourselves"? You mean to make sure that the hosts really build what they actually should build and nothing else? <futurile>nutcase: we build twice and compare .... <now wanders into territory not sure about> <futurile>nutcase: so each build farm builds, and that gives us two packages which should be the same signature right <nutcase>futurile: ok, that matches my assumption <futurile>nutcase: I realised I was unsure how the build farms compare, so I don't know about that part <QUL>futurile: Already did! Will setup a recouring donation next week! Can not wait to see GNU Guix popularity match it's amazingness lol <nutcase>futurile: to not build trust around just one machine+admin <futurile>nutcase: if you have hardware that you'd be interested in using in the build farm then Andreas is a good person to talk to <futurile>nutcase: they host some themselves, and have organised various hardware items from people <futurile>nutcase: that sounds very 'future', would be cool! <nutcase>futurile: you are talking about Andreas E., right? <bdju>I got a GC to work now, told it to free 10GB, took 20 minutes and it said it freed 263GB. In real world space I went from 152GB free on / to 289G free on /. <bdju>Maybe I'll try a big one again now. <futurile>bdju: what underlying filesystem are you using? <bdju>Trying a bigger guix gc again now and in a matter of minutes it's saying it's deleting upwards of 300GB, definitely way faster than when I tried yesterday. <bdju>Is the [xxx GiB] text before deleting more how much data it's gone through than amount to delete? It's still not deleting in real time AFAICT, like checking df -h while it's going. <bdju>It's up to 850GiB there but my free space hasn't gone up since the small guix gc. <Rutherther>it first moves folders to /gnu/store/trash and removes at the end <Chwoka>on the other end of the spectrum from the garbage collection problem, my guix pull requests on the brand-spankin-new vm install are failing with "exit code 1", and earlier "note: build failure may have been caused by a lack of free disk space" and even-earlier before i was running as root "could not write to bla bla bla networking.scm [first thing on the to-do list: No space left on device" <bdju>I was getting weird out-of-space issues while totally not being out of space a day or two ago. :/ <bdju>I wonder if it's a new bug? <Rutherther>Chwoka: are you on foreign distro? Is your /tmp a tmpfs? <Chwoka>i'm running a qcow2 in a gnome-boxes inside linux mint. i do not know how to find out if my /tmp is a tmpfs <Chwoka>i've got tmpfs on /run/user/1000, /run/user/0, and /dev/shm but no /tmp <futurile>you're looking on your main host right - no noexec or anything like that set for /tmp <Chwoka>i'm not sure how to tell of otthat? i ran mount on root@gnu <futurile>so on your main host, check in your /etc/fstab to see how it's mounting /tmp <bdju>How long is the deleting of /gnu/store/trash supposed to take? Been at that for over 10 minutes now. It had gotten to 1.05TiB before that. <Chwoka>when you say main host do you mean within the guix vm or outside of my vm? <umanwizard>Please wait while gathering entropy to generate the key pair; <Rutherther>bdju: impossible to say, depends on speed of your disk, fs etc.... it has to recursively delete everything <Chwoka>futurile: "UUID=38af4c98-f6b1-2062-aa06-b17338af4c98 / ext4 defaults" is literally the only thing in the fstab in the vm <bdju>Deltafire: I tried with /dev/mapper/cryptroot as mount point but it didn't work. <bdju>Ah wait, do I just want /... <umanwizard>perhaps you have to run guix archive --generate-key as root <bdju>Not seeing a meaningful difference from regular df -h to be honest, don't think compression was the culprit. <Chwoka>4.0 gib of memory; boxes is not having me allocate space for the thing the file just seems to get larger and larger (indeed i have other attempted guix vms that take up way more space and i did nothing to allow it or set boundaries) <futurile>Chwoka: hmm OK, I don't know boxes, so I manually create an image <Chwoka>yeah i also manually created this image i was just running it in gnome-boxes <Chwoka>idk if this is a guix problem or a boxes problem <Rutherther>that the space is not preallocated doesn't mean that it doesn't have a limit to grow to <Rutherther>and even if the qcow2 didn't have a limit somehow, the fs has a limit <Chwoka>well we can rule out the latter, i checked the file system of my Actual machine and there is plenty of space in every partition <Rutherther>wdym in your actual machine? are you not checking stuff in the VM you're using guix in? <Chwoka>because i have no clue where this is coming from, if it's external to the image or internal to the guix vm <Chwoka>yeah in the vm something is super up when i check with df in terminal just a sec <Chwoka>did a gc and got that down to 4913012 and it still says 100% and 0% available. maybe i need to do... something different with partitions and mounting in my config.scm? <tesseract>let's say you broke the guix system when doing "guix pull". can you still rollback? <tesseract>wait a sec. how can i update to a newer guix system then? <Rutherther>you first pull to get a newer guix channel revision and then run guix system reconfigure with your configuration <tesseract>o.o i just do "guix pull" and after than "hash guix". but am i still using the same guix system that was installed with the official script? <Chwoka>ahhh shit okay that's what i fucked up! <Chwoka>when making a system image there's a flag to set its maximum size and i didn't do that so the maximum size was just "the size it was when it was made" <Chwoka>this is a stupid mistake but i'm glad i figured it out myself :) <FuncProgLinux>Maybe this is for another channel but has anyone successfully combined tempel with evil-mode? <umanwizard>Or are you using guix installed on another distro? <umanwizard>If you are actually using guix system, the full OS, then yess you need to do a `sudo guix system reconfigure` to upgrade your OS. If you're using Guix on top of another distro, you just need to do `guix pull` to upgrade yoru user installation of guix, and `sudo guix pull` to upgrade root's version of guix / the daemon <umanwizard>Since you talk about installing Guix with a script, I suspect you are actually talking about the latter <aitor>I've downloaded the 1.4.0.x86_64-linux iso image and I'm burning it to usb stick right now <aitor>I would like to learn about your packaging system and attempt to contribute in some way... <aitor>I come from devuan, I'm the developer of gnuinos <aitor>Is there a way to get the list of packages in the installed system? 
