<gnucode>well it looks like I am finally installing guix on my optiplex 7020! <gnucode>I also find it pretty funny that I am finally running into partition size issues. Looks like legacy boot only support partitions up to 2TB. I have a 4TB HDD. <kitzman>hm, what could lead to '.drv' derivations being empty files in the store? before erasing the store and re-creating it ***LispyLights is now known as Aurora_v_kosmose
<clever>kitzman: ive observed that sometimes with ext2/3/4, after an improper shutdown <clever>only the metadata is journaled, so the file contents are at disk, but the fs can repair itself automatically <kitzman>i don't think it's fs related, it got changed recently when i tried to update a package and multiple things broke <cmack>I am trying to get xdg-desktop-portal-wlr with sway and pipewire... does anyone have this working or know of a config that does? <KarlJoad>I want to build a special version of stumpwm that includes contrib <KarlJoad><KarlJoad> modules in the package, similar to how gcc-toolchain <KarlJoad><KarlJoad> stuff to make gcc, binutils, and glibc understand each <KarlJoad><KarlJoad> gcc-toolchain be a good idea to look at for inspiration? <KarlJoad>Jeez... This keeps getting worse... My copy-paste-foo is worsening with age... <yuu[m]>i have this version `1.3.0-1.771b8661` <muradm>normally i don't want to update (@ (gnu packages crates-io) rust-libc) right? :) <muradm>i'm preparing a rust program to submit which depends on rust-ctrlc which when updated asks to update rust-libc <muradm>should i prefer bumping rust-libc or live with lint warning that rust-ctrlc could be updated ... <apteryx>KarlJoad: a 'contrib' output could be dandy <gnucode>Well let's see if rcirc is a little better that circe. Time will tell. <muradm>update to seatd-service-type 56690 <muradm>this last one is worth to mention in news also.. <gnucode>muradm: how long have you been using seatd? <gnucode>muradm: I imagine the experience is more or less the same. I would like to give it a try. <gnucode>hmmm apparently rcirc doesn't give me notifications that someone has sent me an irc message. <gnucode>muradm: does 56690 refer to a guix commit ? <muradm>after this i will add some gui/tui greeters <gnucode>muradm: tui and gui greeters like alternatives to gdm ? <muradm>gnucode: em.. kinda, greeters are for greetd-service-type <muradm>so, yeah basically alternative to gdm, although at very infant stages <rekado_>apoorv569[m]: add the module to the (use-modules …) clause at the top of your configuration file. <rekado_>dirtcastle: this means you’re using an older Guix on a newer profile. <rekado_>new profiles use manifest version 4, which is only supported by the version of Guix that introduced it. <jpoiret>the workaround is to `guix time-machine -- pull` dirtcastle <roptat[m]>or, using a recent guix on a very old profile, since support for some older versions was dropped recently, I think <jpoiret>although, what does `command -v guix` say? you shouldn't end up with this issue if you haven't installed the whole system recently <rekado_>apoorv569[m]: yes, I think so. You can also add ‘spice’ to the use-services-modules expression <apoorv569[m]>Adding `spice` to `use-services-modules` also doesn't help. <rekado_>apoorv569[m]: perhaps you need (use-modules (gnu)) at the very top <unmatched-paren>sneek: later tell muradm: You're using seatd+gtkgreet, right? Does gtkgreet hang without starting when you boot on kernel 5.18.2? I get the same with tuigreet, and I presume gtkgreet would be similar. <unmatched-paren>sneek: later tell muradm: Btw, I encountered some... issues with the packaging of tuigreet. Turns out it depends indirectly on a crate 'rust-embed', which depends on multiple web frameworks. As you might expect, packaging all those web frameworks is quite exhausting :) <dirtcastle>I think it hasn't been a month since I installed it <roptat[m]>dirtcastle, looks like the error says you need to use (spice-vdagent-service) with parenthesis <dirtcastle>I'm not using a virtual machine. I'm using it as a package manager on foreign distro <efraim>looks like grub-efi32 FTBFS on aarch64-linux <dirtcastle>efraim: is that msg for me? aarch is arch linux for arm right? i use artix linux which is arch linux but with openrc as init system. <efraim>dirtcastle: no, that was me shouting into the void <roptat[m]>dirtcastle, that's what I understand when reading the message. Also, you can't reconfigure on a foreign distro <unmatched-paren>Well, you can, but at the end of it you won't be on a foreign distro anymore. <dirtcastle>hmmm. how to fix the error now. can't use any of the guix packages. <unmatched-paren>Resuming my tuigreet packaging marathon, I have found that I will need to package a new version of rust-windows, which depends on five other rust-windows packages. <unmatched-paren>I've only used a foreign distro briefly, so I can't advise you here, but probably not. <jpoiret>did you try `guix time-machine -- pull`? <jpoiret>so you haven't sourced the ~/.config/guix/current profile <jpoiret>and you haven't been using the pulled guixes <jpoiret>only the distro-installed one, which isn't really kept up to date <dirtcastle>I did that multiple times. i have no idea. I almost do that daily <jpoiret>you'll need to do `export GUIX_PROFILE=~/.config/guix/current; . $GUIX_PROFILE/etc/profile; unset GUIX_PROFILE` <dirtcastle>jpoiret: that fixed the problem. should I add it to xprofile or bash_profile or smtg? <dirtcastle>I didn't know .profile even existed. if I login, this has to run right? <jpoiret>again, this depends on how you log-in <dirtcastle>I login normally? like when I boot in, it asks username and password first <jpoiret>if it's through agetty and then a login shell, then .profile is good <jpoiret>remember though that if .bash_profile or .bash_login exist, they'll be picked up instead of that one <unmatched-paren>agetty is the program that runs when you switch to an empty tty, iirc <unmatched-paren>there are other getty implementations (agreetty, mingetty) but agetty is the most common ***toluene7 is now known as toluene
<rekado_>apoorv569[m]: you used spice-vdagent-service, which is a procedure. You should use the spice-vdagent-service-type instead <sneek>Welcome back muradm, you have 2 messages! <sneek>muradm, unmatched-paren says: You're using seatd+gtkgreet, right? Does gtkgreet hang without starting when you boot on kernel 5.18.2? I get the same with tuigreet, and I presume gtkgreet would be similar. <sneek>muradm, unmatched-paren says: Btw, I encountered some... issues with the packaging of tuigreet. Turns out it depends indirectly on a crate 'rust-embed', which depends on multiple web frameworks. As you might expect, packaging all those web frameworks is quite exhausting :) <unmatched-paren>Specifically, the greetd-started sway hangs. But manually started sway doesn't. <antipode>Found a potential to the SDDM doesn't load ~/.profile issue. <muradm>unmatched-paren: hi, regarding your first issue, yes i'm aware, it is nothing to do with kernel. #56690 and few subsequent fixes will fix that. root cause is that greeter can't access seatd.sock after update. <muradm>for now i do durty workaround: login from some spare tty and do "sudo usermod -G users -a greeter && sudo herd enable term-tty1 && sudo herd start term-tty1" <muradm>fix the tty you configured, mine is 1 <muradm>don't reboot frequently :D have a goal to beat your self in uptime :D <apoorv569[m]>rekado_: Yes it worked. Thanks. Let's see if I get proper screen resolution after reboot. <RudraveerMandal[><nckx> "Right. I'm familiar with the..." <- I was using that for a while now <RudraveerMandal[>Sometimes I needed to adjust the size when compiling software but except that it works well <nckx>I have a tmpfs /tmp set to 15 GiB and it doesn't cause me trouble (but I don't build browsers :) <muradm>unmatched-paren: if you apply 56690 and 56699 and then lightly tune config, that "hang" issue <unmatched-paren>I should have tuigreet working in a few days, so i'll send that in soonish. <muradm>on the way to airport i realized that they can be fixed independently :) and now i'm happay that could manage it before take off :) <muradm>unmatched-paren: once 56690 and 56699 are merged, we can coordinate on greeters if you wish <muradm>having consistent configs should be appreciated <muradm>let's see, i suppose it will require some tiny framework where swapping greeters should be easy <muradm>cool, my pending issues list is growing on guix :D <antipode>Does someone know what the bug number if for 'TLS-fout in procedure ‘handshake’: Fout in de 'pull'-functie’? <unmatched-paren>RudraveerMandal[: Probably network-manager-service-type or something along those lines <roptat[m]>antipode, it's probably transient, try again? <antipode>roptat[m]: I know, but error messaging is improper so I'd like to add a little more information to the already existing bug report <unmatched-paren>Anything 'TLS error' is probably actually a network disconnection error. <antipode>Possibly, but then it should just say that it's a disconnection error. <antipode>Do you want pulse, or do you want alsa? IIUC, there are separate systems albeit with compatibility code. <antipode>Do you want bluetooth? If not, no bluez. If yes, I don't know. <unmatched-paren>RudraveerMandal[: Pipewire support in Guix is wip, but i think some people have gotten it to work <antipode>I thought you wanted to use pipewire, not pulseaudio? <antipode>(or does pipewire internally pulseaudio? I wouldn't expect it to, but I don't know) <rekado_>what do you think about writing a blog post with examples for the Guix blog? <antipode>teddd: IIUC, that would compute the size of the derivation. <antipode>(output of profile derivation = the profile) <antipode>If you are using "guix install", try "guix size $(readlink ~/.guix-profile)" <nckx>(Only 7 months; you're almost caught up.) <phf-1>Or maybe the opposite: making it such that in the container we have /usr/bin equal to $GUIX_PROFILE/bin <antipode>This computed-file creates the directory #$output "/bin" <antipode>After that, it calls 'patch-shebang' or such on bin/etcetera <antipode>No good results for "Icedove disable emojification" <nckx>Resistance is futile. You will be emojified. <nckx>I forgot why --expose=/foo/$(guix build coreutils)/bin=/usr/bin doesn't work. <antipode>phf-1: Someone has been looking into FHS containers, so you don't need to patch shebangs. <nckx>According to a quick Thunderbi grep, you're alone today. porcupirate and tex_milan do. <antipode>"Dispay emoticons as graphics" -> false in settings <nckx>antipode: Every single one of those is a :s <nckx>antipode: So any conversion only happening on your side, no need to keep sending :s to the channel. <nckx>Does Guix's GNOME have a virtual keyboard? <drakonis>they should've rebranded to phoenix with 102 though <antipode>Technically I'm typing from Icedove, not Thunderbird, not that it matters. <drakonis>it does matter a little, given how much thunderbird has changed from an update to another <antipode>I thought Icedove was just a rebranding of Thunderbird? <antipode>Currently a not very recent LTS because we choose for LTS, but in principle it seems like we could just update it to latest. <nckx>I didn't mean to correct ‘LTS’, had already typed it when you sent yours. <drakonis>though icecat is not a simple rebranding <nckx>I don't use Thunderdove, but at least with Firefox/IceCat the rate of change is so ridiculous that the differences are quite noticable, ignoring GNUzilla changes. <drakonis>besides riding the ESR train, it also takes certain liberties <antipode>Yep, icecat changes some stuff (telemetry, etc.) <nckx>Sure, but Mozilla churn contributes the absolute lion's share of the visible changes. <drakonis>some of those changes breaks normal browsing, sadly. <nckx>Which reminds me: I can't log into GitLab because they use a particularly virulent strain of the Cloudflare malware. Any specific setting I should look at? I disabled a few obvious ones like resistfingerprinting, but no dices. <antipode>nckx: The antiox jobset is stuck at an "old" commit <antipode>I can continue without antiox for a while, but I though you might want to know. <nckx>Might be due to berlin being quite poorly and unSSH'able. <antipode>The ci.guix.gnu.org jobset for testing the 'antioxidant' build system & transformations <antipode>Now with support for running tests! (which dropped it from 98% -> about 30% or such) <nckx>Does not seem any better? <antipode>unmatched-paren: It doesn't, except for macros. <unmatched-paren>instead of just shoving the source code in the output after running the tests. <antipode>Shared libraries would be nice, but doesn't interact well with monomorphisation, etc., so with some exceptions shared libs won't do. <antipode>unmatched-paren: Doesn't interact well with "guix gc --references" <antipode>space savings when not doing .a (at least, for non-Rust things, might be different with how Rust works) <antipode>(TBC, grafts work, but you can't graft a static library by another static library meaningfully) <nckx>unmatched-paren: I was thinking of grafts. <nckx>Just hygiene in general. <nckx>What's great about berlin is that when I'm busily working over SSH, it still kills my session after a few minutes. <nckx>However, 360 zombie SSHd processes for days: no problem. <nckx>antipode: Short of poking at some checkboxes in the Cuirass Web UI in the hopes to trick it into evaluating something, there is nothing I can do. <nckx>Somebody needs to poke a button. <antipode>guix build -M $CPU_OR_THREAD_OR_WHATEVER_THE_TERMINOLOGY_IS_FOR_WHAT_APPEARS_IN_HTOP fails because "too few build users". Luxeproblemen. <roptat[m]>changing uses agent used to work, but not since a year or two. haven't been able to log in since then <nckx>antipode: Ignore if you know this, but you can add more of those. <nckx>drakonis: Thanks. At first glance (long thread is long) I don't see anything I haven't tried (resistfingerprinting, timing stuff, …) but I'll read it at my leisure later. <antipode>nckx: More processors to fill the extra build users I added :p? <nckx>Uh, yeah, on AliExpress. <drakonis>i think running a newer firefox version gives you less problems <nckx>drakonis: But I am running 🦊, not 🐈. <nckx>But with a lot of similar tweaks applied. No excuse, because GitLab.com really is the only site I'd use that gives me any trouble. <nckx>Even other Cloudflare victims work, they just ramped up their ‘omg halp cyburthreatz’ slider without knowing what it does I gues. <nckx>It was a good impetus to self-host. :) <KarlJoad>apteryx: a 'contrib' output could be dandy <- I am starting with this in my own channel, but may upstream it. What I would like is for the user to specify the contrib packages as inputs. Kind of how Nix does python environments. <panosalevro>Hi all, the Free Software Directory meeting has just started over at the #fsf <kitzman>a bit ashamed of asking this - but is the only way to fix a broken store is to remove it and re-install my profile? <kitzman>particularly - guix created some empty derivations <kitzman>or maybe i'll luck out and pull daily until it tries to create other derivations <unmatched-paren>or `guix build --repair BROKEN-ITEM` as root if you know which items are broken <kitzman>a qt application was failing, so i tried installing qtwayland in the profile. that broke first. then qtbase, then openal <nckx>‘find /gnu/store -maxdepth 1 -name \*.drv -size 0’ should print all of 'em. <nckx>Which you can probably $() into the command (untested). <kitzman>thank you for the tips. a re-reading of the manual should always come in handy. i'm trying now to 'build --repair' them, the gc --verify=repair seemed that it fixed some parts <nckx>I'm not sure whether ‘--verify=repair’ (without ‘,contents’) checks file sizes. I thought it checked only existence. <nckx>It might check sizes though; that's relatively cheap if it's already stat-ing them. <kitzman>yes, it only removed the manually deleted from the database. re-downloading i think might solve things ^^ yay <roptat>kitzman, empty store items can sometimes indicate fs corruption, I'd suggest running fsck at some point :) <antipode>It's not necessarily file system corruption. <antipode>The daemon doesn't do any fsync or sync (except for the sqlite daemon). <antipode>So likely it's a problem with the daemon (not keeping /gnu/store and the database consistent in a power-failure safe manner), not a disk or fs problem. <nckx>Which part? Are you starting cow-store? <ArneBab>Is there a clean way to start eclipse on Guix? <nckx>RudraveerMandal[: So, the purpose of cow-store is to redirect all writes to /gnu/store to your mounted file system instead of to the RAM drive. I guess it's possible that you're running out of RAM even for other things (like ~/.config/guix, and /var/guix, I guess?) but I've never seen that. Do you have very little RAM? ***Dynom_ is now known as Guest5031
<lilyp>somewhat random question, but does anyone know how to get the path separator (i.e. ":" in POSIX) in C(++)? <dlowe>nothing built-in until C++17 <dlowe>the usual strategy is to #ifdef WIN32 <lilyp>If you mean std::filesystem::path::separator, that's / or \ <lilyp>well, actually ::preferred_separator <samplet>It says it sets a ‘PATH_SEPARATOR’ variable. <lilyp>I was planning to use meson, but right now I'm invoking GCC manually <nckx>It's hard to cite a negative, but I don't think there's one for C. C++ is so vast, who can say. <atka>podiki[m]: I just ran across a picture of Guix on the deck credited to you.. small world :) <podiki[m]>I only tried with a simple live image and it worked, but didn't try much <podiki[m]>couldn't rotate the screen, might be because of just using linux-libre kernel (amd...) <atka>podiki[m]: that's cool though, thanks for the info <atka>If I ever get one, I'll throw guix on it <podiki[m]>I should go back and test more, but I was just running off a usb stick (you can choose boot device in bios) <atka>podiki[m]: do you have a guix live image you created/ <podiki[m]>I just used the sample templates in the guix source, I guess what is used by the installer as a starting point <podiki[m]>but you can just use any ol' system configuration to create an image <garjola>I am installing the guix system on a workstation. This is my first time installing the distribution. I have used the package manager on Debian before. <garjola>The installation goes well until it gets to "building database for manual pages" and seems to get stuck there when it reaches 100%. <garjola>Is this normal? I don't know how long am I supposed to wait. Or how I can check if something is wrong. <bost>Hi. I just run `make check` against the cd3c4ad8fc resulting in XFAIL: 3, FAIL: 9. Should I send a bug report? Is anybody interested? <rekado_>nckx: the gitlab problem disappeared for me when I changed the user agent string. <lilyp>Does our gcc-toolchain miss std::format? *civodul feels a bit stupid to ask 
