IRC channel logs
2023-06-11.log
back to list of logs
<VesselWave>Guest28: Could you send me channel file system (directories) and files there, I'll try to inspect it <rekado>some time ago my laptop stopped booting correctly. Upon power on I enter the EFI shell and I have to manually switch drives and execute /EFI/BOOT/BOOTX64.EFI. <rekado>it’s one of the many obstacles that keeps other people from using my laptop without my permission, but it’s also a bit of an obstacle for me. <rekado>my system is configured to use grub-efi-removable-bootloader <rekado>should the location of that file be recorded in some EFI variable to make EFI execute it automatically? <graywolf>rekado: I do not think so. The removable is designed for USBs and stuff afaik, so really nowhere to record. <lilyp>rekado: the relevant invoke is this: <mirai>I'm not, I've really no idea what is being asked here. AFAIK the macro does not change the datum and is simply plumbed into (apply …) without any checks <graywolf>But is feels bit weird. I also use removable, and I just set the disk as default and it boots fine. Do you have other (nonremovable) EFI loaders installed? <lilyp>mirai: you're talking in invisible implementation details here <lilyp>how are the users to guess that using only (define-configuration ...) as a hint? <lilyp>even if just for documentation's sake, it is worth asserting that this is how things *are supposed to* work <rekado>graywolf: no, it just stopped working at some point after I got an error at the end of “guix reconfigure” <lilyp>was that a grub error by chance? <rekado>it’s been a long time since then, but it could very well have been that, yes <rekado>I see there’s an EFI variable PlatformRecovery0000, which mentions EFI\BOOT\BOOTX64.EFI, but I don’t know what any of this means. <rekado>came with nothing installed at all <mirai>lilyp: I think I see your point now <mirai>I'll send the extra cases in the next round to give some time for others to chime in as well <VesselWave>Guest28: thanks, it will take time but I will fix it <csepp>jpoiret: yup, it works now. still not sure why it was breaking guile and guile-gcrypt. <Guest28>VesselWave okay, if I create a new directory it works. Interesting. Thanks for helping out <VesselWave>Guest28: It's pitty that raspberrypi-filmware is nonfree <Guest28>VesselWave Ah, I know why it was not working. I missed some imports and therefore it couldn't run. But this error message was totally misleading me with saying that it can't even find the definition. Should this reported as bug? <Guest28>VesselWave yes, I am ware of that. AFAIK there are no non-free SBC. There is always some hardware that needs proprietary firmware. Although if you know one, don't hesitate to tell me <Guest28>VesselWave unless I missunderstood you, and you meant that it is pitty that RPis are not fulle open <VesselWave>Guest28: I've had this error with guix system reconfigure. It's not obvious at first time. If you won't find this issue here https://issues.guix.gnu.org. Send bug report to bug-guix@gnu.org <VesselWave>Guest28: I meant it, but remember: nonfree software is offtopic here <Guest28>VesselWave well, I was aware of that but in the Emacs manual I read that for debugging facts are important. So I thought if I change something I could lead to false information <Guest28>The question was related to Guile Scheme and my intention was not to hurt someone, if I did, I apologize. I didn't meant to <Guest28>You can add loadpaths with -L, you can set any path even though they don't even exist on the filesystem. Should this be reported as bug? <Guest28>I did a type specifying the path and it just runs it. <VesselWave>BTW, you can set username /nick. Or even set password, so no one can use your nick <ChocolettePalett>I think I have ruined my GNU/Guix... (but I guess I figured out how to fix it) <ChocolettePalett>Anyway, a GNU/Guix-riddle: assuming you have descended to zeroth generation using ``guix pull --roll-back`` and now your guix isn't in ``~/.config/guix/current/bin``, what would you do? <Guest37>I'm having some issues with lightdm. I'm getting an error message when attempting to login telling me it "failed to start session". I checked out the log files and think it's related to the greeter attempting to load the default session, but the only xsession file my system has is xfce.desktop. The logs seem to show that lightdm is loading one <Guest37>configuratoin file (which has some of my tweaks in it), but lightdm --show-config is showing a default config file in the store. Anyone have any debugging tips? <HiltonChain[m]><anemofilia> "Why there is a empty /run/..." <- It's included in %elogind-file-systems. <anemofilia>I ain't using elogind, so tought I could, and as for now, I am having no problems because of it <anemofilia>Because thinkfan CmakeLists.txt says that /run/thinkfan.pid is used as pid-file only if systemd is found <anemofilia>Otherwise it should use /var/run/thinkfan.pid (which is what I wanted) <HiltonChain[m]>I think that's decided at build time. And according to the package definition, "/run" is used. <razlix77[m]>guix system: error: error parsing derivation /gnu/store/rdzdv6w1l2l85s2jq92w5mm8hwpj1g44-elogind-dbus-service-wrapper.drv': expected string Derive([' <razlix77[m]>is it safe to use the master branch of the guix channel? <jonsger>razlix77[m]: you should use the master branch. The 1.4.0 release doesn't receive any security updates... <jpoiret>razlix77[m]: this is likely to be due to some corruption at the file system level <jpoiret>in any case, you can try to repair this with `guix gc --verify=repair,contents` <razlix77[m]>a bit late for that :) already started a new install <razlix77[m]>but fwiw I had the nongnu channel added as I need the firmware blobs <jpoiret>.drv files being empty is a more low-level bug <jpoiret>i don't think there's any way you could've unknowingly caused this but who knows <lilyp>these kinds of issues usually pop up when you get a forced shutdown before guix has written stuff to disk <lilyp>so unless you provoke a rainstorm during guix pull, it's basically force majeure <jpoiret>but isn't adding new files to the store atomic? notwithstanding adding them to the db <jpoiret>I haven't read too much of the internals of the daemon but are you suggesting that a new empty file is created and then later on either written to or simply replaced? <f3n1x>hello guixers ! when guix-home in use... say for instance that i require a new python-package. Are both : "a) 'guix package -i python-package' b) add 'python-package' to my home-config.scm file "- valid options ? thanks, thanks, thanks <lilyp>when using guix-home, you ought to edit your home-config as main resource <lilyp>note that doodling with random python packages is what guix shell is for <ekaitz>hi people! anyone managed to compile a java app with javafx in guix? <pistrie>hi everyone. I asked this question a couple days ago as well, but even after the comments and reading some of the documentation it is still unclear to me. So the whole reason I want to try Guix is the generations, and the declarative approach to managing your system and its packages. So as far as I know there's two types of configurations; the <pistrie>system config and the home config. In the system config I would put packages that I want root to have access to as well (think neovim), and in my home config I would put packages that are specific to me. <pistrie>In /etc there is a `config.scm` file where I think I'm supposed to declare which packages I want to have installed. I see `(packages (append (list (specification->package "nss-certs")) %base-packages))`, but Im not very familiar with scheme syntax so Im not sure how I would add more packages here. Furthermore, what do I do to install packages in my <pistrie>Lastly, what is the difference with this config file approach and the `guix install` approach? Isnt `guix install` imperative and therefore against the whole point of system reconfigures? <VesselWave>pistrie: To add packages to your system config just add another string to specification->package e.g. (specification->package "nss-certs" "neovim") or see how I do it in my config gitlab.com/VesselWave/dotfiles/-/blob/main/.config/guix/systems/base-system.scm#L76 <VesselWave>pistrie: Haven't really used guix recently, so I hope someone will correct me if I am wrong. To install package you just add it to config and reconfigure as in system config, nice example: https://guix.gnu.org/en/blog/2022/keeping-ones-home-tidy . And I think when you run guix install and use guix home, it will automatically add packages to your config. <spacecadet[m]>Morning guix, trying to define a package with a git-fetch origin that uses git modules, I added the (recursive? #t) option under the (git-reference ...) but it's not doing a recursive clone of the repo. Cross-referenced other packages that use git modules but I can't find any hints, am I missing something? <spacecadet[m]>I am declaring (recursive? #t), but it still doesn't clone the submodules. I could package the submodules independently if that's more guixy. <pkill9_>nothing, just using guix to make money, heh <lilyp>well, you could host a VPS server, but that's already done :) <pistrie>step 1: install guix. step 2: get a job with BYOD <lilyp>you could also try selling guix as a workflow to $company who uses $nasty-build-system at the moment <elevenkb>but hopefully one wouldn't just simply have to reimplement $nasty-build-system as scheme code :p. <pistrie>why do I need to add (use-modules (gnu packages vim)) to config.scm when installing neovim? <VesselWave>pistrie: because scheme is a programing language, I guess. It's a common thing, you always import or include something <spacecadet[m]>The package definition for neovim is in gnu/packages/vim.scm, which is a guile module that can be accessed with that use-modules form <VesselWave>pistrie: BTW, you can see how package definition looks like with `guix edit neovim` it will show /gnu/store/..../gnu/packages/vim.scm <pistrie>ah right that makes sense. does that mean that if I were to define a bunch of packages in config.scm that I would have quite a lenghty list of (use-modules) as well? <pistrie>alright. how do most people on guix install packages? am i being unnecessarily difficult by not using `guix install`, or is the consensus that a git-versioned config file is pretty handy? <lilyp>the consensus is that the versioned config file is handy *and* that precise imports are good, actually <pistrie>great, thank you all for being so helpful :) <lw>VesselWave: long time, no see 🤓 <lw>VesselWave: I'm reading 7.6 Creating a Channel. I get the ideia, now I will acquire the knowledge to set it up <oriansj>Some days I think guix would be better off if everything was in channels and that there would be no default channel <oriansj>that way if someone wants a stable channel, they could pick that. Or if they wanted the bleeding edge, they could pick that too. <lw>VesselWave: did you sent to pedrogbranquinho@gmail.com? <lw>It seems to not be in my inbox <lw>Let me see other folders <ieugen[m]><oriansj> "Some days I think guix would..." <- I think that as well. I would use stable for servers and such. <oriansj>ieugen[m]: it would also allow the operating system to be on the stable channel and the users to choose to be on the bleeding edge if they wanted. <lw>VesselWave: I replied you :) <spacecadet[m]>Bumping my question from earlier, trying to build a package with git sub-modules, (git-reference ... (recursive? #t)) doesn't seem to work <pkill9_>.sneek later tell pistrie you can also use (specification->package "package name"), then you don't need to import modules as you aren't using the package variables <lilyp>now that's an extremely cursed package then <Guest28>guix pull returns invalid certificate. what could be the cause for this? I have nss-certs installed system wide. the issue is with guix running on a rpi <VesselWave>lw: Have you set my address as not spam? I am not sure you saw my reply <ToxicAegis>Hey all, I have a few questions about an issue I'm having, is this the correct place to ask? If not, I apologize and would appreciate if I could be pointed in the right direction, thanks. <ToxicAegis>Ok, so I have a Pinebook Pro. I have a Wi-Fi USB dongle which uses the ath9k kernel module. As far as I can tell, this module is not included in the default linux-libre-arm64-generic kernel. Is there an easy way I can tell it to include the driver/module in my config for building images? <Guest28>If the RK3399 chip works without non-free firmware, does that mean something like an Asus Tinker Board 2 would work without problems on GNU Guix System or would it still require proprietary code for wireless? <mekeor>ToxicAegis: afaik, linux-libre only comes with ath9k_htc. if you need ath9k instead, you need to look for non-official, community-maintained guix channels. but proprietary software is off-topic here <mekeor>Guest28: check out what wifi driver is needed for the wifi-card of that asus tinker board 2. alternatively, you can always use a wifi-dongle that supports ath9k_htc <mekeor>Guest28: also, guix system can also be installed with community-maintained guix-channels that offer non-free drivers. it'd arguably still be guix system. anyway, proprietary software is off-topic here. <ToxicAegis>Oh, it is actually the ath9k_htc module I need afaik. Is there a way to verify that the driver is included? The USB works on my desktop, but when I generate the image and put it on a USB for my Pinebook Pro, there's no wireless devices. Perhaps there is something else I'm missing? <mekeor>ToxicAegis: or just "sudo modprobe ath9k_htc" <mekeor>the latter will attempt to load the kernel module <ToxicAegis>lsmod didn't seem to provide any output on the Pinebook, I'm not sure if this is an issue with it being ARM? Attempting to modprobe the module it says no such module is found. <Guest28>if someone can recommend sbcs, I wouln't mind either <apteryx>was there a trick for sourcing a failed build's environment-variables file in a way to extend rather than override the current environment? <mirai>apteryx: what are you trying to do? <mirai>'extending' the environment can mean: set variables in $file not in current-env; append values to certain env vars (with some divination for the delimiter) <mekeor>ToxicAegis: but you could try to put ath9k-htc-firmware into that field explicitely <mekeor>ToxicAegis: you use guix system, right? <mirai>there's also the thing where the same var might have relevant but completely distinct datum <ToxicAegis>I have done this as well to be sure, and I have verified that the firmware files are present on the image, which is what led me to assume the driver wasn't being included in the kernel. <mekeor>ToxicAegis: i'm a kernel noob, so i don't know what "being included in the kernel" actually means, and i don't know where guix installs those firmwares to.