IRC channel logs
2023-12-25.log
back to list of logs
<nckx>Hello, it's 00:01. Happy Christmas, all. <nckx>Happy Christmas all y'all. <alexey`>Hello there! Is there a chance to upload the `info' version GNU Guix manual? <nckxmas>I'm ? too. What is 'the info version' and what does uploading it look like? <nckxmas>Not HTML, clearly, I guess, so what? <nckxmas>I guess we could, but I'm not familiar with the use case of *.info files over HTTP. <alexey`>GNU-guix-manual-1.4.0.info.tar.gz something like that <simpleton>Merry Christmas. I am new to guix. Wanted to make the hello-world packaging tutorial. Entered "guix edit hello' and saw nano as editor. Now I'm here asking stupid questions. How do i set what "guix edit" is opening files with? <nckxmas>Sure. I'm just missing the immediate use case. curl | info? (Not a joke.) <nckxmas>simpleton: VISUAL=emacs guix edit, for example. <nckxmas>Or 'export VISUAL=emacs' for the entire session. <alexey`>No, I am talking about downloading to a local machine and studying the manual offline. <alexey`>It's more convenient to have and info version in GNU Emacs. <nckxmas>alexey`: I suggest opening a bug report (clearly marked as a feature request) via email to bug-guix at gnu dot org. <simpleton>nckxmas . Works! Thx. Setting permanently as default somewhere in my home via .shell.rc file? <nckxmas>Sure, if you're OK with setting a default editor for almost everything. <nckxmas>VISUAL overrides EDITOR. Both are a somethitg of a standard, much more than Guix honour them. <simpleton>nckxmas Oh....as first step i would prefer that my choice is default editor only for 'guix edit'. Is it possible to be so specific in GuixSD? <alexey`>nckxmas: Thanks, I wish I could, but, unfortunately, I do not have an email account. Once they started asking cell phone number for identification, I just stopped using it. <simpleton>if i unset "VISUAL" and only set "export EDITOR" then 'guix edit' throws error "sh: line 1: +92: command not found" <nckxmas>simpleton: No, we don't specifically check (say) GUIX_EDITOR or so. I'm not personally conviced we should; I'd tell user to write their own alias at that point, or to invoke 'VISUAL=foo guix edit' by hand. Sorry. <nckxmas>simpleton: You are explicitly setting EDITOR to the empty string, which is not equal to unsetting it, so Guix will invoke "". <nckxmas>With +92 (the line number) as its first argument. <simpleton>nckxmas NoNoNo! No soory. I thank you...was just wondering as i did not see basestuff like that easily findable somewhere.. <simpleton>nckxmas right again. 'unset VISUAL' and 'export EDITOR' works... <nckxmas>I'm not many things but right is occasionally one of them. <nckxmas>OK, I wasn't lying about going to sleep, but this is just too bizarre not to respond to: alexey`: Is there really no mail provider that doesn't ask for a telephone number? I don't have one, so oh dear. <simpleton>peanuts wow...i was like 2 seconds faster...muahahaha <simpleton>hmmm...i was faster than the bot(peanuts)...not so bad... <alexey`>nckxmas: Thank you ). I think, running my own mail server is, perhaps, the best option, but it involves other issues, which is difficult to resolve at the moment. <alexey`>simpleton: Thank you. It's just hard to trust `uncle google' :) <simpleton>alexey`not if you are a simpleton like me... :) <AwesomeAdam54321>alexey`: Does your ISP block port 25? It might be difficult to self-host an email server in that case <alexey`>simpleton: Ok, I am a human being. let's get down to the GNU Guix discussion. <alexey`>AwesomeAdam54321: We shouldn't violate the rules of this channel. This is my personal issue and has very little to do with the GNU Guix. Perhaps, things will change for the better, but at the moment they are what they are. But, anyway, thank you. <simpleton>alexey`you're welcome. I am so surprised by gnu guix. atm it feels like a bid odd. So far i only run it in a VM until i make a decision on it. BTW. Yandex and GMX seem not to take phone... <alexey`>Usually, many GNU packages include *.info version of its documentation. I was just a bit surprsied to miss this version on the GNU Guix documentation page. <Kabouik>Which package do I need to populate ~/.guix-profile/share/X11/locale/en_US.UTF-8/Compose? I have this file at this location on my other laptop, but not on the new one despite using more or less the same config.scm, so that must come from a package. glibc-locales is installed. <Kabouik>Got my answer with realpath on the other laptop: it's libx11! <Kabouik>That is weird, because `libx11` is included in the (packags (append (specifications->packages (list …)))) of the config.scm of the new laptop, not sure why it was not installed automatically after a system reconfigure. <roran>podiki: Did you say you weren't using ansel/vkdt very often anymore? <podiki>roran: I only ever briefly tried both; vkdt is super fast but I haven't tried to figure out the interface and not sure it has everything for me yet; ansel could be good just haven't seen what the changes are recently <roran>NODE GRAPH PODIKI, NODE GRAPH <roran>I agree, however. vkdt doesn't have hot pickle removal, and so my dark shots are usually wanting. <roran>Everything that I questioned, refreshingly, on where it was, was put into a spot where I already knew it should be after I learned it. His interface design makes a lot more sense, and it's probably because him and I grew up in the same period of product design <podiki>I do want to try both more, just been behind on getting to photo editing for a while <podiki>Getting both into guix would be good steps to doing that too :) <Kabouik>Yay I got my Ryzenadj package to work, will try to make some patches during the holidays (I kinda forgot the process, have a few other patches to submit that stayed on my private channel out of laziness). <hwpplayer1>I don't have a ethernet port so how can I install Guix Libre distribution ? <hwpplayer1>I don't have mobile network connection also for usb tethering <lilyp>Do you have a removable hard disk? If so, you could install Guix onto that disk from another running Linux system (assuming it has internet connection) <hwpplayer1>I only have one disk which is inside the machine <hwpplayer1>maybe I can find some tool to connect ethernet port via usb port <lilyp>there might be some ethernet to usb adapters out there, perhaps even ryf ones <tissevert>I'm afraid guix is broken since commit f6817e71dff7d0d9fdb55db8b85a1d3d04e2bf5a <tissevert>which removes ed.scm, unfortunately still used in gnu/packages/base.scm (and in 5 other locations: algebra, cook, patchutils, version-control and lisp) <tissevert>am I the only one who can't pull anymore? I get "no code for module (gnu packages ed)" <lilyp>Oof, that's why one ought to try such things before pushing <tissevert>I know right? well, the commit is dated from december the 23rd, but do yo you know how long ago it was pushed? I'm kinda curious/worried how long it's been broken ^^' <tsmish>tissevert: committer field says 2023-12-25 17:44:47 +0100 <tissevert>also, another commit from the same person removed nano.scm <tissevert>and version-control, web-browsers and gnu/system.scm <tissevert>so I take it you folks are already working on a fix? <lilyp>Have reverted ed.scm locally. Efraim, what are you up to? <efraim>I can do a follow-up commit and fix it <tissevert>tsmish: how did you look that up? from git log I can only see the author's date and not the committer's <tissevert>great news, then I can go back to my Xmas washing-up :) <tissevert>oh yeah! great, I didn't know it was visible there, that's most convenient <tissevert>so that's a relief, our beloved package manager has only been broken for 2 hours <efraim>actually I'm more worried about circles with ed <lilyp>Should I submit my original patches to QA real quick or are you fine alone? :) <efraim>I'm about to go back AFK, go ahead and send them to QA :) <efraim>although I pushed the two patches removing the extra imports and reverting ed <lilyp>Is 'make as-derivation' still the right check? <efraim>I just ran git clean -dfx and am running make now, I'll check 'make as-derivation' in a few minutes <lilyp>Okay, then I'll hold back my patches since I didn't catch the 5k bomb anyway. <efraim>I think I've hit ed accidentally before <efraim>lilyp: make as-derivation built just fine <simpleton>Hi, Merry Christmas folks. Is there a known issue producing this error message? "guix pull: error: unsupported manifest format" ??? <tissevert>hi simpleton, not that I know of but guix pull has been broken for a couple hours and from what I understand a fix is on its way so I don't know but it might be causing that mistake? <rekado>simpleton: this means that you’re using an older version of Guix on a newer profile <rekado>the profile you’re operating on has a file called “manifest” <simpleton>it happens so fats i doubt there si any interaction with a online entity when issuing `guix pull` <simpleton>the manifest version at " ~/.guix-profile/manifest" states version 4 <Jaeme>It seems that my patch to remove ed.scm actually broke Guix. <rekado>simpleton: that’s the wrong profile for “guix pull” <rekado>simpleton: “guix pull” operates on ~/.config/guix/current <rekado>simpleton: make sure that “type guix” points to the right thing <tissevert>Jaeme here present commited a delete of ed.scm, without editing all the files depending on it <tissevert>same goes with nano.scm, required in installer.scm and system.scm <tissevert>true, sorry for that imprecision, Jaeme ordered that delete, and Mathieu Othacehe trusted him to do so <tsmish>I'm surprised hydra didn't scream loudly about that change, considering it broke 'guix pull'. <Jaeme>tissevert: I'll accept that :) <tissevert>efraim has already commited a patch but unfortunately the original change should have touched 5k+ packages <Jaeme>I will like to extend my formal apology to all the Guixers who had a rude awakening this Xmas morning. <tissevert>tsmish: I don't mean to be disagreable but I'm surprised no one tried to simply build guix with a `make` <tissevert>Jaeme: no rude awakening, I rested all day and had a wonderful Xmas, and just noticed when I thought I'd do something useful of my day and update my machines <simpleton>`~/.config/guix/current/manifest`is empty. Sorry that i take some time but i'm new to guix <tissevert>and anyway the commit was pushed only at 16:44 UTC so it hasn't been around long <simpleton>Jaeme , i guess shit happens . Relax. I need to understand what is causing this. I need to understnd guix <tissevert>in a sense, it's even a good thing that it happened on a day when statistically lots of people have better things to do than a guix pull :) <simpleton>if i do a reconfigure it says ```error: aborting reconfiguration because commit aeb494322ca9dec4a4d66a7d063239c8536bd538 of channel 'guix' is not a descendant of 892ec53b7eedf998afa149b9cdd169f0cab8f41d``` <Kabouik>How do I set /sys/power/disk to [reboot] in Guix? I am trying to fix my hibernate not working, and someone recommended trying that. <tsmish>Kabouik: does "echo reboot | sudo tee /sys/power/disk" not work? <Kabouik>Could it be a kernel argument such as `"disk_default=reboot"`? I use `"mem_sleep_default=s2idle"` for /sys/power/mem_sleep and that seems to work. <Kabouik>Oh, I can edit those directly in Guix tsmish? <Kabouik>It did work indeed, thanks, but I thought these files were not persistent in Guix <tsmish>Kabouik: I mean, that is the way theese files usually work. Note however that it will last only until reboot. <tsmish>Kabouik: you could try something like "sleep 3; echo disk | sudo tee /sys/power/state". I think I remember computers waking up from pressing enter to submit a command. <tsmish>Kabouik: I think checking dmesg might also give you some info. I never had to solve problems with hybernate myself. <Kabouik>The thing is suspend (s2idle, S0) works fine, even if I go into sleep by pressing enter after a command `loginctl suspend` <tsmish>Kabouik: this command should make computer hibernate. <Kabouik>Will try the command you recommended in a few minutes (I'm compiling something) <tsmish>Kabouik: resume=/dev/mapper/cryptroot? How is that supposed to work? Also I don't think you can avoid looking at the entire dmesg, there could be some messages at the end about failing to hybernate that don't use the word 'suspend'. <tsmish>It should still save the state to disk and halt. I don't think it will wake up from an encrypted partition, but it should still halt. <Kabouik>tsmish is resume=/dev/mapper/cryptroot wrong? I have a very similar config on another machine running Guix system, and it does hibernate succesfully: https://0x0.st/Hg5T.txt <Kabouik>I got the same issue as in the video when using sleep 3; echo disk | sudo tee /sys/power/state <Kabouik>On the other machine with that cnofiguration, and luks encryption + hibernate to a swapfile, I just have to type my passphrase twice again when waking up. It's clunky but at least I can find my programs where I left them. <Kabouik>But on that one, the screen turns off but it doesn't completely enter hibernate mode (I can hear the fan still spinning), and I assume it somehow aborts/fails, and the screen turns back on. <Kabouik>I have tried disabling Modern standby in the BIOS, to no avail (it would break the regular S0 suspend, not working anymore, and hibernate would turn off the screen and never stop spinning the fan, without any way to re-wake the computer). <peanuts>"~mlaparie/guix-config#1: Hibernate on LUKS encrypted Guix system <tsmish_>Kabouik: ok, that will work, my bad, wasn't aware about resume_offset. I suggest posting full dmesg logs. <tsmish_>TIL you can resume from about anything as long as initrd puts required numbers in a special file. But you are not even getting to resuming, so that was moot point anyway. <Kabouik>Just a stupid question before I try again and check dmesg. My computer has 64GB of RAM, of which I think 16 (?) might be assigned to the iGPU VRAM. My swapfile is only 20GB, so it's smaller than my remaining RAM. Could that be the issue, even if I still have 44GB of unused RAM when trying to hibernate? <lilyp>It will be an issue if you need to write out more RAM than you have swap space for <lilyp>Let's say 8G of regular RAM and all 16G of the VRAM are used, then you're out of swap space. <Kabouik>But you're confirming that only the amount of RAM+VRAM actually used will matter, so I don't need to make a 64GB swapfile which will almost never be necessary unless I try to hibernate with 4 VM running. <lilyp>I assume that only used RAM will be written, but note that already used swap space also counts towards the limit <noobly>guix install vmware-open-vm-tools --> unknown package <lilyp>Run `guix describe', there's a faint chance you have a bad commit. <Guest14>noobly: Maybe you are on an old version that does not yet have the package. Try guix pull and run the command again <lilyp>Yeah, release commits aren't supposed to be used for more than running the installer :) <noobly>Guest14: that was my thought too, pulling now. but it was a fresh install (installed this morning), so I thought everything would be fresh. <noobly>older install image though, didn't know that would matter <noobly>"building /gnu/store/wb1cw7rjcs703cv1k2f62sibn34ivrfr-guix-packages-base.drv..." has been stuck at 0% for 30 minutes or so after guix pulling, second attempt. any ideas? <Kabouik>So I'm trying to recreate a bigger swapfile in case it would help with my hibernation issue, and I notice that `sudo btrfs property set /swap/swapfile compression none` errors out (invalid argument). Could that be the cause of the issue? <simpleton>regarding my 'guix pull' issue... anything that i can do to fix it manually? <noobly>After guix pull, I get stuck at 0% wihle building "building /gnu/store/wb1cw7rjcs703cv1k2f62sibn34ivrfr-guix-packages-base.drv...", the first thing I build, for at least 30min. Is this expectd behavior? <noobly>I have not yet seen it grow past 0% as I rebooted my VM containing Guix the first time, and my entire PC the second time. <Guest14>noobly: No it should not, unless it runs on very slow hardware <simpleton>`guix pull`can be broken for some because of some changes. See post at Jaeme 19:57:36 . But i am not sure it will affect everybody and i need to understand why it hit me and others. Stll a newbie <noobly>Guest14: Any ideas for a solution or debugging? My desktop is old, but not _that_ slow. <rekado>noobly: how much memory do you have? Less than 2GB is not great for compiling all of Guix. <rekado>noobly: I recommend getting an older version of Guix that is available on ci.guix.gnu.org <tsmish>simpleton: what is your issue with guix pull? "unsupported manifest format"? <simpleton>tsmish Jaime referred to some issue due to a commit. Is this the one? id=f6817e71dff7d0d9fdb55db8b85a1d3d04e2bf5a <tsmish>simpleton: no, that should be unrelated. <tsmish>simpleton: do you have ~/.config/guix/current/bin in your $PATH? <tsmish>alternatively, does '~/.config/guix/current/bin/guix pull' work <simpleton>tsmish no...not in the path `PATH=/root/.guix-home/profile/bin:/run/setuid-programs:/root/.guix-profile/bin:/root/.guix-profile/sbin:/run/current-system/profile/bin:/run/current-system/profile/sbin` <rekado>you’re likely using an older Guix to pull <noobly>rekado: I think 8gb total system-wide, but the VM is limited. I checked that earlier and it seemed like it was the issue. <noobly>rekado: was that 'old guix' comment directed at me too? <noobly>I think my image is from a month or two ago <rekado>noobly: generally, we build these derivations on ci.guix.gnu.org, so when building this takes too long it may make sense to wait for ci.guix.gnu.org to build it first so all you need to do is download it. <simpleton>rekado `guix (GNU Guix) 1.4.0-16.aeb4943`, `guix is hashed (/run/current-system/profile/bin/guix)` <rekado>do you have ~/.config/guix/current/bin/guix? Or have you never run “guix pull” before? <tsmish>simpleton: rekado knows way more than I do, but you can probably create a different profile with modern guix by running something like 'guix pull -p /tmp/test'. At least I think it will work, because manifests should be created by 'guix pull'. <noobly>rekado: no this is the first run of guix pull on this system (on vmware) <simpleton>tsmish added to path '.config/guix/current/bin/guix` command `~/.config/guix/current/bin/guix pull` works but also throws error "unsupported manifest format <tsmish>simpleton: you also said that ~/.config/guix/current/manifest was empty? <simpleton>tsmish this is correct and still the case. the file `~/.config/guix/current/manifest`is empty <rekado>simpleton: then try as tsmish wrote and pull to a different profile <rekado>should work because it’s a new profile, so there’s no manifest file to get in the way <tsmish>simpleton: "guix pull -p /tmp/test" <tsmish>that will pull to a new profile in /tmp/test <tsmish>simpleton: it looks like you had a filesystem error resulting in manifest file getting truncated, which lead to guix getting confused and saying that it doesn't understand manifest file. <Kabouik>Well there go my attempts at increasing the size of my swapfile: https://0x0.st/HgRo.jpg (and unfortunately older system configurations fail with the same error), I truly borked my Guix install <tsmish>simpleton: Ok, you should probably replace root current-guix profile with a new version and I have no idea how to do that. <simpleton>tsmish its still doing the `guix pull -p /tmp/test' . You are probably right and i should replace `root current-guix` and yes, i have no real plan <simpleton>tsmish regarding truncation of file. Machine is running in VM and VM got destroyed when Host-system powered down <tsmish>simpleton: something like "ln -sf $(readlink -f /tmp/test) /var/guix/profiles/per-user/root/current-guix" should work if I my understanding of gcroots is correnct. But beware, I never done this, could be horribly incorrect. <simpleton>how about copying the /tmp/test/manifest to ~/.config/guix/current/manifest which is empty <tsmish>~/.config/guix/current/manifest is immutable <tsmish>since it ~/.config/guix/current is actually in /gnu/store <simpleton> guix system reconfigure seems to work again.... <Kabouik>If anyone thinks my system is recoverable, I'm interested. What I did was recreate a /swap/awapfile of 64GB instead of 20GB, and edit my config.scm to change (file-system (mount-point "/") (options "subvol=swap,nospace-cache,compression=zstd"))[nto (file-system (mount-point "/") (options "subvol=swap,nodatacow,nospace-cache)). Reconfigure went well, but reboot brings the above picture. <tsmish>Kabouik: is your root on /swap, or only a swapfile <tsmish>because bootloader thinks your root is in /swap <nckxmas>Kabouik: Bit late but: Linux can't hibernate with over 50% of RAM used (it will try its best to drop caches &c. first), so at least on paper allocating more than 32 gigs of swap 'for your RAM' is pointless. Furthermore, in practice, the image is also LZO-compressed and it tends to compress quite well unless your workload is particularly pathological. <Kabouik>So I probably did not need to increase my swapfile to fix hibernation, I see. And yet I did and it broke. I am not sure if what broke is the change in my (file-system) or the /swap/swapfile recreation. <Kabouik>Thanks simpleton, I will try the rescue <tsmish>Kabouik: if you are using grub, you can enter 'editing entry' (I think it's "e" key) mode and correct locations of kernel and initrd. 