<mehlon>usually brand names in Chinese are adapted in a way that sounds like the original word and has some connection in meaning <mehlon>but someone has to decide which specific translation to use, since there's plenty of ways to do it <alextee[m]>ah i see, that's the impression i got too, because another person translated it differently, but i think they sound similar. <grubles>lfam: same error even with memory balancing disabled. the VM has 8 GB of memory too. hm... <grubles>i guess i'll try debian 10 instead of fedora 30 <rain>it looks like old parallel port printers need the lp kernel module. should i add it to the default modules or should i just add a note about it in the CUPS section of the docs? <ryanprior>joshuaBPMan: it is cool! Glad you're enjoying :) <joshuaBPMan>ryanprior: yeah, updating package definitions is a breeze apparently... <ryanprior>As long as no dependencies have been added it ought to be pretty trivial =D <joshuaBPMan>ryanprior: No dependencies. I had to update the meson build system. <joshuaBPMan>and right now meson is complaining that it cannot find openGL ES v2... <noobly>I have a weird issue in Icecat where it won't display numbers anywhere. I've disabled the LibreJS option but it still persist. Any idea? <pkill9>User-59087: this is a known issue i believe <rain>what the heck is up with guix offload test not finding / not being able to access a private key that is very clearly there and very clearly accessible? <rain>ah, nevermind, i think i misread the error <rain>now it's a different error <rain>looks like guix pull will keep trying to offload even with a failed config, instead of just moving on... <rdtc55>I have the same problem. It's infuriating because my password is way too long... Is there a way to use full-disk encryption on a T400 with libreboot without having to type the passphrase twice? ***catonano_ is now known as catonano
<grubles>hmm looks like gcc-mesboot-4.9.4.drv is building using a debian 10 appVM on qubes ***midnight is now known as midnightmagic
***midnightmagic is now known as midnight
<ryanprior>I'm proposing a GUADEC talk about packaging GTK apps in Guix. Does anybody want to co-present with me? <ryanprior>The deadline to edit my proposal is in two days, so if you're interested then seriously give me a shout. <raghavgururajan>ryanprior Regarding to help with your packaging, I am almost done with audacious. Once it's done, I can help you with the packaging some dependencies of mkcert. <ryanprior>Awesome raghavgururajan! I packaged Appstream and protonvpn-cli today. =D <raghavgururajan>ryanprior That is great. I was planning on packaging protonvpn-cli, glad you beat me to it. :-) <ryanprior>I just have to clean it up for submission :3 <apteryx>has anyone tried booting a degraded RAID with Guix System? <Kimapr>i'm not a RAID expert but you probably should expect problems on any degraded medium with any OS <Kimapr>and i still have problems with wi-fi (RTL8723BE), even though i installed the driver (rtlwifi-new) <apteryx>I'll add the degraded option in my GRUB menu until the drive replacement comes. I'll give it a try. If you don't see me around... struggles :-) <Kimapr>apteryx: i think you better make a backup <apteryx>yeah, IIUC, the initrd we make would need to support being passed some argument via the kernel command line given via GRUB. <apteryx>We don't honor anything except --root (the root device to boot from), so I won't have much of a chance there. <raghavgururajan>The "-Dc_link_args=-Wl,-rpath=" is used in meson-build-system. I am looking for what it's equivalent in gnu-build-system. <ryanprior>Gooberpatrol66: or for fancier output, guix package -I <pkg name> <ryanprior>I'm working on packaging some Meson, and I've got one that can see a library (granite, which I've also packaged) and another that can't see the library during its configure step. <ryanprior>One thing that's different is that the package that sees it delcares the dependency in its root `meson.build` file, while the one that has a problem declares that dependency in its `src/meson.build` file. <ryanprior>Is there a possibility that the root file is getting some kind of dependency path patching while other build files aren't? ***apteryx is now known as Guest30039
***apteryx_ is now known as apteryx
*raghavgururajan is getting screwed by meson <rekado>the equivalent to "-Dc_link_args=-Wl,-rpath=" is setting the LDFLAGS environment variable to “-Wl,-rpath=” <rekado>some build systems allow you to pass this as a configure flag <rekado>see the “hurd” package for example <rekado>I find themes in Emacs when using Guix very confusing. <rekado>the choices I have in M-x load-themes change after selecting a theme <efraim>emacs-vterm is missing a parenthesis it looks like <rekado>I’m trying to come up with a reproducer <sirmacik>hm, I think that sbcl-slime-swank package which was depreciated in favour of cl-slime-swank needs to be reverted <sirmacik>sbcl in general doesn't see it installed <jas4711>the manual says LVM is missing, but what does that mean? I never bothered to explore LVM usage (which is critical for my deployment of libvirt-based virtual machines) after reading that. Now that I tried 'guix package -i lvm2' and using LVM, it appears to work fine. what am i missing? <sirmacik>jas4711: docs might be quite dated on that <jas4711>sirmacik: maybe it refers to being able to specify LVM through config.scm? that is not critical for many uses of LVM though, so the docs is a bit misleading here <sirmacik>as I remember lvm2 + luks2 update were brought to repo few months ago so I can mount drive from my old arch installation. Probably there wasn't much work done in favour of configuring it through config file for running guix on it <sirmacik>oh, grub got luks2 support early this year \o/ <sirmacik>might be a good news for running guix on more secure encryption <jas4711>i think LVM support now just works, and the only thing missing is to add a no-op mapped-device-kind and update documentation. i just tried rebooting with LVM, and /dev/mapper contain them as expected. unlike LUKS or RAID, there is no operation needed to make LVM appear <jas4711>I'll prepare a documentation patch :) <jas4711>what is missing is maybe having root on LVM, but that's not essential for many LVM uses. the manual could be updated to say that (until someone tests it) <efraim>17 hours to build llvm-9 on powerpc <Kimapr[m]>you specify the filesystems directly in system config <raghavgururajan>Ah! I am looking for hot-pluggable devices. Xfe file-manager mounts/umounts based on entries in those directories. ***dingenskirchen1 is now known as dingenskirchen
<raghavgururajan>I have these code in the source: [1] #define FSTAB_PATH "/etc/fstab" [2] #define MTAB_PATH "/proc/mounts". <raghavgururajan>Can I replace these paths to something else, so that xfe's mount/umount operations work in guix? <raghavgururajan>Kimapr[m] I am aware of the option in system config. Here, I am trying to fix Xfe package. <dutchie>please do not use a channel with hundreds of people in to try out matrix's markdown support <Kimapr>how do i send messages starting with /? <Kimapr>i tried to send "/etc/fstab does exist and is generated in system config" but it just doesn't get sent <dutchie>start with a space will do the trick <dutchie>some clients will let you do "/ /message" <civodul>raghavgururajan: /etc/fstab and /proc/mounts are readable and contain valid info <civodul>what makes you think they're the cause of the Xfe problems you mention? <raghavgururajan>civodul I see. In Xfe, mount/umount options under 'Tools' menu are always greyed out. <raghavgururajan>civodul But when mount via udisks separately and come to Xfe, the umount option alone is un-greyed out. <Kimapr>how to test if a specific linux module is up and running? <Kimapr>wha- the module for my wifi card is shown in lsmod but neither nmtui nor ifconfig see it <efraim>raghavgururajan: is there a hardcoded location for mount/umount in XFE? *efraim needs to type slower <raghavgururajan>efraim I saw these lines in code: [1] #define FSTAB_PATH "/etc/fstab" [2] #define MTAB_PATH "/proc/mounts". <efraim>right, i mean does it hardcode /usr/bin/mount or something <raghavgururajan> [1] #define DEFAULT_MOUNTCMD "mount" [2] #define DEFAULT_UMOUNTCMD "umount". <efraim>I always feel like it's 'acceptable' to assume Guix System for GUI applications, the alternative almost seems to be searching through $PATH for a mount command that is setuid <efraim>if it's not expensive to compile try substituting /run/setuid-programs/mount for mount and see if that changes the behavior <raghavgururajan>But I doubt it, as the command `mount` will be the path of `which mount` of the shell. <jvshahid`>I followed the instructions in section 'Contributing/The Perfect Setup', but when I execute `guix-devel-use-module' (or C-c . u), my CPU fans start spinning for a few minutes. I usually wait like 5 minutes and then kill the process. Is this expected behavior? Am I missing optimization flags? <Kimapr>the driver (wifi-card) seems to be working but it isn't used by anything :( ***jvshahid` is now known as jvshahid
<raghavgururajan>efraim I fixed some other issues in Xfe. (#40980). Would you be able to push it? Thanks! <Kimapr>^ this looks weird, how do i unblock Wireless LAN? <raghavgururajan>Kimapr Config looks good to me. You do not have any Desktop Environment installed. How are you connecting to WiFi? <Kimapr>i'm not connecting to wifi, i want to run a wifi access point or something like that, i'm already connected through ethernet <raghavgururajan>I see. Is nmtui an application of some sort? `guix search nmtui` gives me nothing. <raghavgururajan>Kimapr I would suggest you temporarily install XFCE and try the network-manager applet. Just to check if nm is working correctly. <Kimapr>nmtui is part of network-manager package (or at least it seems to) <kraai>Hi. I'm trying to package GHC 8.8.3, which can build with GHC 8.4 or later. Is it better to build it with 8.4 or 8.6? <raghavgururajan>Kimapr Okay, that explains. Your WiFi card is not supported by libre-driver in linux-libre kernel. <Kimapr>> 0a:00.0 Network controller: Realtek Semiconductor Co., Ltd. RTL8723BE PCIe Wireless Network Adapter <Kimapr>and RTL8723BE is listed in h-node <Kimapr>i made a package definition for driver found from there (based on another driver's definition) and installed it as linux-module <janneke>raghavgururajan: about the sudo-removal patch for ktsuss: the "sudo" version (current master) seems to for me, look: <janneke>$(./pre-inst-env guix build ktsuss)/bin/ktsuss /run/current-system/profile/bin/id => [popup] <janneke>uid=0(root) gid=0(root) groups=0(root) <Jonsger-phone>Is there a way to get a console in gdm without function keys? <janneke>raghavgururajan: i tried emacs, works too :-) <raghavgururajan>janneke Hmm. Even I tried and it works inside env.Not sure why it doesn't works with spacefm. <bdax>hello, I have a laptop with integrated intel graphics, and I'm wondering if it's compatible with linux-libre. It's specs are: <bdax>Graphics: Device-1: Intel HD Graphics 520 driver: i915 v: kernel <bdax> Display: tty server: X.Org 1.20.4 driver: i915 resolution: 1920x1080~6 <bdax> OpenGL: renderer: Mesa DRI Intel HD Graphics 520 (Skylake GT2) v: 4.5 <raghavgururajan>janneke Nah, it is not working with spacefm. I tried again with `./pre-inst-env guix environment --ad-hoc spacefm`, in current master. <raghavgururajan>janneke May regardless of spacefm, shall we use su instead of sudo? It was suppose to be gui of su anyway. ;-) <rekado>rust 1.19 fails to build on i686-linux (while trying to build icedove) <janneke>raghavgururajan: i have no strong opiniouns on that...however i /think/ that Guix tends to like SUDO more than SU (see documentaiton) <janneke>raghavgururajan: ...and i am not too happy with a commit that says "ktsuss's sudo is broken", when it actually works... <janneke>...and it could still be that spacefm is broken? <efraim>IIRC we're ignoring the rust i686 build errors and hoping they go away when we move to bootstrapping with 1.29 <rekado>maybe the last commit broke things 61fc72d3ac949d0c65690ee25bb1daf01f39bb42 <raghavgururajan>janneke Hmm. So what do we do now? Are patches for spacefm, udevil and ktsuss good to be pushed? <janneke>and i'm a bit hesitant...many commits need fixups later; so <janneke>now i get a commit that says ktsuss' sudo is broken, i test it, it works... <roelj>Anyone getting “ice-9/eval.scm:293:34: no binding `zip' to hide in module (gnu packages compression)” when compiling the current master branch? <rekado>there was a problem with the commit before 4ac3996036 <janneke>maybe tomorrow you have spacefm fixed, and want ktsuss' sudo back (because it saves typing passwords all the time) <rekado>there’s still a problem with gnu/packages/backup.scm, though <rekado>more wrong with commit 61fc72d3ac :-/ <janneke>raghavgururajan: i very much appreciate all your efforts, but i think you may need to slow down a bit <janneke>raghavgururajan: sure, you did very well! <janneke>but possibly a bit too hastily (maby it's just bad luck, we all make mistakes!) <roelj>rekado: The preceding line to my error is “error: failed to load 'guix/scripts/pack.scm':”. Should I attempt to dig deeper and send a patch? Or are you already on it? :) *rekado pushed another fix for commit 61fc72d3ac <rekado>(this would have been avoided with a test build. Looks like an incomplete copy paste job.) <NieDzejkob><NieDzejkob> I'll push a fix for /backup.scm </NieDzejkob> nevermind <roelj>rekado: Is “#:hide (zip)” in guix/scripts/pack.scm correct? <janneke>raghavgururajan: and maybe you think of a fix tomorrow for spacefm->ktsudo->sudo? :-) <rekado>roelj: usually this kind of error has terribly misleading error messages <rekado>there’s nothing wrong with hiding zip <raghavgururajan>janneke Just a sec. I am pretty sure I had error using sudo in ktsuss. I even discussed it here with reepca. <roelj>rekado: Okay. Then I know the problem must be somewhere with me. :) <raghavgururajan>janneke I installed ktsuss in my profile and did `ktsuss xfe` in terminal. After authentication, the application opens but not with superuser mode. <raghavgururajan>Could you try installing ktsuss and open an application? Like `ktsuss foo`? <janneke>raghavgururajan: and that's the newer version, that uses /run/setuid-programs/sudo, right? <janneke>like i said, works for me: ktsuss emacs <raghavgururajan>janneke Okay, I'll dig into this later. Since spacefm is the only application that uses ktsuss right now, shall we proceed with current patches? wdyt? <raghavgururajan>May be we can change "in guix system" to "in spacefm" in commit-message of ktsuss? <efraim>TIL Debian skips running the test suite on mesa <raghavgururajan>janneke Please ignore #40753 and #40901 for now. #40922 is good for sure. *janneke goes afk for a bit <apteryx>I think I may have found a good reason to let users pass root options at the CLI (GRUB -> Linux kernel cmdline -> picked up by our initrd): degraded arrays. I have a degraded array and I'm afraid to reboot because I'm sure I'll get stuck with no way to pass the "degraded" option in GRUB. I have the luxury that my system is still running, so I can reconfigure with that option in my operating system <apteryx>configuration, but it could have been otherwise. <apteryx>perhaps a good time to revisit those root file system on a Btrfs submodule patches (some older version of those patches even had that feature :-) <raghavgururajan>And since #40753 is already closed, I have sent revised patch as #40982. *civodul is drowning in pending patches :-/ <NieDzejkob>alextee[m]: I don't know if you've noticed, but zrythm failed on CI, even though it worked locally for me. Could you investigate? <NieDzejkob>(if you can reproduce this, but perhaps not consistently, maybe you need to disable #:parallel-tests?) <apteryx>civodul: don't be sorry :-) but with retrospect and with my current RAID failure, I think it'd bring nice value that hour initrd honor 'rootflags' <NieDzejkob>interestingly, i686 and armhf also failed, while aarch64 worked <alextee[m]>NieDzejkob: heh, i dont think my code is expecting more than 16 threads :-) probably array overflow <NieDzejkob>the failures look like you should try ASAN and/or valgrind :P <alextee[m]>i do have a flag to limit that, i should probably add it to the tests, but i will fix the array lengths too now <alextee[m]>NieDzejkob: out of curiosity, how many CPUs does the CI server have? <NieDzejkob>** Message: 09:55:46.550: Number of CPU cores found: 64 <jonsger_>its different. The new ones are all AMD Epycs. I think some of them are only 2x 8cores <bdax>I know it's not ideal, but is it possible to install non-free firmware (iwlwifi) on a guix system, until I eventually get a decent wifi card <grubles>ok i made some progress getting guix working in a qubes appvm <grubles>running in a debian 10 appvm seems to resolve whatever issue i was having running in a fedora 30 vm. but now i'm getting a hash mismatch for icu4c-64_2-src.tgz. *ecbrown experiments with zfs send /gnu | ssh zfs receive *ecbrown concludes that it looks like sending a lot of data over ssh <pkill9>what does this `guix copy --from=... /gnu/store/...` error mean?: guix copy: error: implementation cannot deal with > 32-bit integers <NieDzejkob>I think you should be able to get a backtrace with guix repl -> ,use(guix scripts copy) -> (guix-copy "--from=..." "/gnu/store/...") <efraim>ecbrown: is that using zfs from guix? <ryanprior>grubles: cool, are you working on a GuixSD TemplateVM or installing Qubes in a Debian AppVM? <ryanprior>ecbrown: would it help to `guix pack` instead of relying on `zfs send`? (I'm not familiar with the latter.) <grubles>ryanprior: installing guix in a debian appvm <zimoun>Does it exist a function to expand '~' in Guile? Because the long options fail. For example, "guix pull --channels=~/path/to/file.scm" <ryanprior>zimoun: ~ I think the shell has to handle tilde expansion. Would it work for you to do `--channels=$HOME/path/to/file.scm`? <zimoun>ryanprior: yes it works. But '~' and '~username' should be expanded. I mean, it is what I expect. :-) Emacs provides "expand-file-name", does Guile have similar? <bavier>I think there are some shadow utilities that will do that <bavier>bash, at least, will not expand ~ if it's not preceded by whitespace, it seems <jvshahid>Asking the same question I asked earlier. I followed the instructions in section 'Contributing/The Perfect Setup', but when I execute `guix-devel-use-module' (or C-c . u), my CPU fans start spinning for a few minutes. I usually wait like 5 minutes and then kill the process. Is this expected behavior? Am I missing optimization flags? <NieDzejkob>jvshahid: I'd guess that you have to run `make' in your checkout <NieDzejkob>(this is because loading all the code in the project from source takes a while) <ecbrown>efraim: no, ubuntu 20.04 lts with zfs on root <ecbrown>i see there is a zfs package: might check it out ***MatthewAllan93_ is now known as MatthewAllan
***MatthewAllan is now known as MatthewAllan93
<pkill9>where is discussion on paramterized packages? ***dddddd__ is now known as dddddd
<bavier>pkill9: some discussion has been happening on the mailing lists <rawmat>hello, i'm new and i installed guix on vmware <rawmat>i've problem setting the console resolution <rawmat>cannot modify /etc/config.scm to set the resolution <bavier>rawmat: you can set the grub gfxmode parameter, is that what you need? <rawmat>i see in the manual speaking about grub-theme <bavier>rawmat: see the example there for overriding the default resolution <rawmat>and trying to set 1920X1080 like the example in the manual <rawmat>when i try guix system reconfigure it gives invalid field <mbakke>janneke: commit 541122957c68a15f8657775e62448b1cfafd7cb8 forgot to mention the new native-inputs, which confused me quite a bit when resolving a merge conflict <mbakke>also it would be much better if it had gone through master <bavier>rawmat: oh, I see, that section has not seen much love; it doesn't seem to have been updated since the bootloader-configuration changes <rawmat>i inspected the grub generated file and it's set to gfxmode auto <bavier>rawmat: ignore the "grub-configuration" bit in the example; the "theme" field and everything under it should work though <bavier>i.e. 'grub-configuration' should be 'bootloader-configuration', and the indentation fixed *bavier on puny netbook right now, otherwise could test faster <rawmat>i tried, but it still gives "invalied field specifier" <bavier>rawmat: would you mind pasting your config.scm somewhere? <bavier>lfam: maybe, didn't check. I think I'm looking at the 1.1.0 manual <bavier>and those bootloader changes were far before that <lfam>Can you say which section it is? I can take a look now <bavier>Last few paragraphs of "Bootloader Configuration" <bavier>under 'grub-theme', "For example, to override..." <bdju>try using paste.debian.net/ <bavier>rawmat: "theme" is a field of bootloader-configuration, so should be moved in one level of parens <lfam>I'm testing the correction for the manual now. Sorry about the confusion rawmat... <rawmat>im new to guix and what i see for now seems very good <nojr>Where can I find more information on node-build-system, I could find a few paragraphs in the guix manual <nojr>This variable is exported by ‘(guix build-system node)’. ***Server sets mode: +cnt
<lfam>nojr: Besides the manual, the code will have to serve as documentation, unless there is a blog post or presentation online <ruffni>is openlibm the math.h library in guix? *reepca`` grumbles about spotty connection turning him from reepca to reepca``````` and causing him to miss pings ***reepca`` is now known as reepca
<NieDzejkob>ruffni: math.h should be shipped by glibc, and therefore available after installing gcc-toolchain <ruffni>i'm trying to write a package definition (with scons-build-system). so the package needs gcc-toolchain as an input? <ruffni>or is it already part of the scons-build-system? <NieDzejkob>(I usually run the build without any inputs and then watch it fail :P) <lfam>That's the packaging canonical workflow :) <lfam>Or, the canonical packaging workflow <lfam>I think that Scons is a bit obscure, so there may be some issue with how the package in question is using it <ruffni>so the SConstruct file searches for the libraries and tells me library "m" (math.h) is not.... <lfam>You might take a look and see where it is searching <ruffni>it does a conf.CheckLibWithHeader(). conf = Configure(env) <ruffni>so i might just give it the path with a prefixed var? <lfam>ruffni: I would look at other packages using scons to see if they have to do something similar <ruffni>there's no 'ack' in guix? how do i find packages built with the scons-build-system? <ryanprior>In a Guix package definition what do you do if the program wants there to be a "cc" executable? *civodul finds it amazing just how much activity there is in this repo <civodul>ryanprior: you tell it to use "gcc" instead, for example by setting CC=gcc (environment variable) <ryanprior>That works for some things, where it respects the environment. But in this case there's literally parts of the build process that just call "cc" cold <ryanprior>I patched a few instances of that but that got me to wondering "maybe there's a pattern for this?" <bavier>ryanprior: I think there was one package where I created a 'cc' link in a temporary directory and added that to PATH <bavier>as long as the package doesn't try to call '/usr/local/bin/cc' or something like that... <ryanprior>That sounds slightly janky but not jankier than doing a bunch of substitute " cc " " gcc " <lfam>The jankiness is proportional the janky problem it aims to solve ;) <ryanprior>civodul: the activity is because Guix is solving real problems and helping people :3 <mbakke>janneke: I merged commit ce0f1b1eefbb6d7d6339c0919cc051f9f5cf9b8d to core-updates which undid your a7dbb95ca5189382e3bea9de5c97b929da79fc32, so you may need to do a new update *mbakke fades back into obscurity