<leungbk>Would anyone mind reviewing patch #36381? I submitted it last month. ***jonsger1 is now known as jonsger
***jonsger1 is now known as jonsger
<rekado>my dev version of mumi now downloads the original emails from Debbugs and stores them for indexing with mu. <rekado>annoyingly, the Guile bindings for mu are a little primitive: after initialization I can’t get an updated view on the database. <rekado>so I’ll need to run something like “watch -n10 mu index” to update the database regularly <nckx>…and we're back to chewing gum. <rekado>and to make queries I’ll need to use a helper process, which is awkward… <rekado>so I’m thinking about writing Guile bindings for Xapian and using it directly. <rekado>but I don’t actually *want* to do that. There’s a whole line of yaks that wait for their turn. <rekado>even without mu there’s a noticeable improvement: we get the full messages with the new version, not some arbitrarily truncated message. (Example of a truncated message: /issue/36487#2) <apteryx>is there a limits.conf used by Guix, for the max number of file descriptors/inodes? <emacsomancer>(for Guix System), what's the best way of having one-liners run as root on boot? <jackhill>emacsomancer: One way would be to define a service that does that. I don't know if that's the best way, but might be a good place to start reading the manual or guix code for examples. <emacsomancer>jackhill: thanks. looks like perhaps a one-shot shepherd service could work, but it seems like overkill. <emacsomancer>how do I get the emacssql sqlite binary to compile in guix? it says it cannot find the C compiler. <efraim>emacsomancer You might need a configure/makeflag with "CC=gcc" <efraim>I assumed there was a package for it <efraim>I compile almost nothing by hand now <emacsomancer>efraim: but it's something that an emacs package is trying to compile <emacsomancer>I must have had this working before, but I don't remember what I did <efraim>I see. Do you have a link? I can take a look when I get back to my computer <emacsomancer>even org search is failing because it's missing (org search is fine until I add the pdf-tools and tablist -- from Guix itself) <efraim>it sounds like you would need gcc-toolchain installed, or the 'gcc' reference to be patched to refer to the gcc from the build phase, which assumably comes from gcc-toolchain <efraim>and not directly from the gcc package <emacsomancer>and I've installed guix's package for emacsql-sqlite, which seems like it should work <emacsomancer>in emacs, (executable-find "gcc") returns a path to gcc in my .guix-profile <emacsomancer>ok, I think I figured it out: I had to make sure that emacs-local emacsql-sqlite wasn't installed, but only guix's version <cyrozap>Hi, all, I'm trying to do something that I thought would be relatively straightforward, but appears not to be: I would like to install guix, on Debian 10, on a ppc64le system. <cyrozap>Trying to build from source results in "configure: error: `powerpc64le-linux' is not a supported platform." with a recommendation to check the manual, and short notes on "porting" and "bootstrapping" that I didn't really understand. <cyrozap>So, I guess my question is, "How do I build the bootstrap binaries without cross-compiling them?" <cyrozap>Because all the docs say on the matter is "... to port the distribution to a platform that is not yet supported, one must build those bootstrap binaries ..." and "... Guix can _cross compile_ those bootstrap binaries", which of course doesn't help if I'd rather not trust cross-compiled binaries given to me over the internet by a stranger. <kori>my manifest file is empty <emacsomancer>what is the equivalent of setting the path in ~/.xprofile ? <emacsomancer>I've tried adding: (setenv "PATH" "/home/emacsomancer/bin:$PATH") to my config.scm, but this did not have the desired effect <efraim>alpha3 should be available from bayfront <kori>what should i do if my manifest file is empty <kori>this was after a forced reboot <efrit>I am trying to install GNU Guix System 1.0.1 i686 <efrit>But when I use the dd command as told to be in the reference manual to copy the ISO image on my USB key, it seems there are errors in the file system on the USB key <kori>why do you think that efrit ? <efrit>because when I connect the USB key on my computer again, there are errors popping up <efrit>“Error mounting /dev/sdb1 at /media/efrit/GUIX_IMAGE1: wrong fs type, bad option, bad superblock on /dev/sdb1, missing codepage or helper program, or other error” <kori>what command did you use to dd <efrit>and there are 3 mounting point on the file explorer I use <krusnus>your not supposed to mount the drive onto your system afteer flashing <efrit>dd if=guix-system-install-1.0.1.système.iso of=/dev/sdX; sync <kori>I assume you replaced of=/dev/sdX with of=/dev/sdb <kori>but yeah, what krusnus said <kori>youre not supposed to mount the flash drive, just boot from it <efrit>I mounted the USB key after I failed to boot from the USB key <kori>efrit: does your system support EFI boot? <efrit>on my laptop where I want to install Guix <kori>i couldnt boot the iso without having EFI boot enabled... but i was told this was not normal, so idk what to tell you <kori>maybe wait for someone like nckx <krusnus>did you select the drive as your primary drive in your bios? <efrit>there is a menu when to computer boot and I chose the to boot on USB key <efrit>but maybe my laptop has issues because to boot on the Trisquel installed on it I have to do some weird manipulations <efrit>just wait a bit I’ll try to boot on the USB on to another computer to see if it works there <krusnus>anyway, hey everyone! i recently installed GuixSD on my computer and one of the first things the guix manual tells you to do when you're done installing is to run 'guix pull' but when i do that i get the error "guix pull: error: Git error: the SSL certificate is invalid" i couldnt find any solutions when i searched for it so i thought i'd ask here <krusnus>and see if anyone has had the same problem and/or know how to solve it. thanks in advance! :) <kori>krusnus: check your date? <efrit>so I tried to boot on the USB key on another computer, and it fails too <roptat>I'm trying to use match inside a build phase, but I get an unbound variable error instead (on the content of the match) <krusnus>i changed the date and now it seems to work! thanks kori :) <roptat>nevermind, I was missing a #:module argument :) <roptat>so, I have a working maven-build-system (with binaries downloaded from maven), except for tests because surefire-booter hard codes a path to /bin/sh which we don't have in the build environment <roptat>and I was able to rewrite two packages (java-tukaani-xz and java-iq80-snappy, used by maven-install-plugin) so I can use them instead of their binary counterpart :) <roptat>I'll have to find a way to properly bootstrap the plugins, but it's going much more smoother than I though it would <roptat>I also think I can create a patch for our ant-build-system to install the jar and pom files to the right directory, for maven compatibility <efrit>So I retried by downloading the ISO again and using the dd command <efrit>not even booting on it after selecting the USB key in the boot menu <efrit>maybe there is an issue with the ISO image? *rekado builds qtwebkit on berlin <rekado>BTW: I’m now building one of the last openjdks on core-updates. My patches seem to work. <roptat>the idea is that java packages will be installed in lib/m2/ as if it were a part of the .m2/repository for maven <roptat>so the build system looks for jar and pom files in lib/m2, and symlink them into a maven repository in the build directory <roptat>with -Duser.home, I can point it to the repository I just built <roptat>they install the jar and pom files to the right directory <roptat>the second also has generic enough code that I will probably use for my patch to the ant-build-system ***pflanze_ is now known as pflanze
<rekado>I pushed the modified mumi online. It might be a tad slower to render issue pages initially, but at least there are no truncated emails. <rekado>this also opens up the possibility of much faster search as we can do all of this locally. <efrit>should I write bug report about my issue? <efrit>since it seems it’s not easy to solve it? <efrit>thanks for the link, I’ll look into it in the future hopefully <jlicht>hey guix! Is there a programmamatic way to specify channels? I would like to create an install image where the `guix' command already has access to specific channels, without having to `guix pull' first <PotentialUser-36>Hi, I'm trying to install Guix System on a nonfree computer (unfortunately) - I know the consequences of doing so and I'm going to sell this machine eventually and buy a libre one, but for now I want to get used to Guix. The problem is running the system with linux-libre fails, probably because of nvme ssd disk and a highly proprietary motherboard <PotentialUser-36>- tested it on Guix 1.0.1 and Trisquel 8.0. I successfully added a nonfree channel from this repository https://gitlab.com/mbakke/guix-nonfree and run guix pull command, but while trying to install the system with 'guix system init /mnt/etc/config.scm /mnt' it tells me the variable linux-nonfree is unbound. I tried adding a module to the config fil <PotentialUser-36>e, but nothing I tried works, despite the fact nonfree definitions are present in the store. <apteryx>PotentialUser-36: before you go ahead attempting to pull random channels (and non-free) into your install, could you tell us what is the original problem with Guix? Any kernel log (/var/log/messages), or other hints about what is failing? <apteryx>my experience is that most PC work with linux-libre, but 1) wireless probably deosn't and 2) gpu might not if it's not intel <PotentialUser-36>apteryx: All I got now is a photo of what system writes while trying to boot, I could provide the log but I would have to install it again first. <apteryx>how does the boot abort? kernel panic? <PotentialUser-36>apteryx: Okay I'm back. Do you mind nonfree Javascript? I don't know a good file sharing site without javascript... <PotentialUser-36>sorry for the delay, I had to rewrite the link by hand form a mobile device <janneke>vagrantc: mescc-tools-seed comes in when you want to use mescc-tools in a bootstrap build, i.e., without using gcc, binutils. mescc-tools will build fine with a full toolchain. <vagrantc>janneke: i've successfully built mes with all dependencies from debian, by the way :) <vagrantc>janneke: needs a lot of fixes before i can upload mes itself... <vagrantc>janneke: should i try building mescc-tools-seed as well? :) <janneke>vagrantc: that's great! i'm just looking at the patches you needed (use bash, build changelog etc) in the hope that mes-0.20 would need no patching to build on debian <janneke>vagrantc: i don't think we need mescc-tools-seed until we start creating a real bootstrap path? <apteryx>PotentialUser-36: sorry, had to reboot <vagrantc>janneke: i guess i have my sights on a real bootstrap path :) <vagrantc>and as far as building guix in debian ... i'm wondering if we could... extract guile-gnutls out of gnutls sources and build it ... "out-of-tree" or something ... since the debian gnutls maintainer is silent on (re)enabling it <PotentialUser-36>apteryx: Is the link working? Can you see it? (don't know if this IRC stuff preserves message history) *vagrantc disappears for a bit <PotentialUser-36>Additional info: It works from a DVD disc and installs itself successfully to the SDD without any error, but then after rebooting it stops working like in the photo. <eric-guix>guix system disk-image --system=armhf-linux -e '((@ (gnu system install) os-with-u-boot) (@ (gnu system install) installation-os) "A20-OLinuXino-Lime2")' <eric-guix>But I received this error: |builder for `/gnu/store/5yxy7gfyhc5si02czsksz4k9jxgqvyjh-libtsm-0.0.0-1.f70e379.drv' failed with exit code 1 <eric-guix>build of /gnu/store/5yxy7gfyhc5si02czsksz4k9jxgqvyjh-libtsm-0.0.0-1.f70e379.drv failed <eric-guix>View build log at '/var/log/guix/drvs/5y/xy7gfyhc5si02czsksz4k9jxgqvyjh-libtsm-0.0.0-1.f70e379.drv.bz2'. <eric-guix>cannot build derivation `/gnu/store/s3ll1mdn7vg1aib75fx4g19iz94v1igr-kmscon-0.0.0-1.01dd0a2.drv': 1 dependencies couldn't be built <eric2>I wrote earlier, someone answer me? <eric2>My real goal is learn how can I make a ISO for an Olinuxino64 whit the final intention to try it on a Olimex Teres I (a DIY laptop). <eric2>but I need to know What is wronge with that <nckx>eric2: That should be possible. Were you eric23 earlier? If so, what was your question? I only remember the term-auto thing but that's harmless. <brendyyn>eric2: btw i just meant that noone had replied while you were disconnected, not that i was denying your question, incase i sent the wrong impression <nckx>eric2: Could you provide a bit more info (for example, the command you tried to run + its output → paste.debian.net; we also have a wgetpaste package if you crave the pipe)? ‘Didn't work’ isn't much to go on. <nckx>Disclaimer: I don't know the first thing about non-UEFI ARM, I can only help you get started with Guix. <nckx>eric2: That's why we tell you in the channel topic to use a pastebin like the one I mentioned above. <nckx>So your device is a "A20-OLinuXino-Lime2"? <nckx>I know enough about ARM to know that the command won't work on any other board. (Almost) every board needs explicit support. <eric2>No, actualy I tried only this command to switch later <eric2>this is the line of error: build of /gnu/store/902q2hs6wdhb9q6phyxf4j95fbb1vda2-u-boot-a20-olinuxino-lime2-2019.04.drv failed <nckx>We'll need much more than that. *nckx AFK for a few minutes to learn some knitting tips. <vagrantc>should probably update u-boot to 2019.07 anyways ... <vagrantc>though i won't have a chance to work on it till mid-august, probably <vagrantc>had a few surprises when updating u-boot for debian <eric2>Thanks, I have one other question. How can I find all the supported platform or soc like "Olinuxino" <nckx>eric2: From the page you linked above, under the command you ran: ‘A20-OLinuXino-Lime2 is the name of the board. If you specify an invalid board, a list of possible boards will be printed.’ <nckx>I think that's the only way for now. <janneke>hmm, something fishy with one of my substitute servers <nckx>That and ‘[core-updates] Due to flawed bootstrap tarball, full rebuild needed’ just landing in my box promise a fun week ahead. <nckx>(Although I'm not certain you'd agree with that diagnosis.) *nckx 's substitute server just panicked hard, tf. *janneke 's substitute server's guix-daemon needed a restart... <kori>icecat is failing to display several pages properly <nckx>[Non-sarcastically] Is it really? <kori>numbers don't show up in several pages <nckx>Ah, that's a Guix + IceCat + fonts issue. The good news is it's not just you, and it's easily fixed, the bad news is I don't remember how. <kori>sometimes things show up, sometimes they don't *kori is looking into using regular firefox <nckx>kori: ‘I confirm that OP's solution of `guix package -i font-gnu-freefont-ttf' fixes this problem.’ <nckx>Fair warning: I don't see why that wouldn't affect regular FF equally, unless it just happens to have changed font handling in a way that this problem isn't hit. <kori>I installed Inter (my preferred font) and that hadn't fixed it <kori>guix can recommend things post-install, right? <kori>tell the user to install some fonts? <nckx>Well, if it's IceCat/FF crying when it's only favourite font isn't available instead of falling back to one that is, I think it is a bug, just not in Guix. <nckx>kori: It should just use the fonts that are installed. <nckx>‘Boohoo you don't have Comic Sans so I can't render the letter ‘a’!’ is not acceptable behaviour. This is similar 😛 *nckx is demotivated by digging around FF the past two days to get it to do anything that isn't hard-coded, so ignore them. <kori>if freefont fixes the issue <nckx>Well, sure. Welcome to Firefox. <kori>forcing icecat to use inter works... <nckx>I agreed above that FF/IC should just use system fonts instead of demanding that you install its favourite. You don't neet to convince me. <nckx>And you seem to have managed to convince IC 🙂 <kori>for some definition of convinced... <kori>i need to learn how to package software <kori>of (free) software I wanna package <nckx>Hi Marlin[m]! I was distracted. <nckx>Marlin[m]: That probably require propagating the font, and I think that's evil and against user freedumbs, but I won't fight it if everyone disagrees. *nckx propagated something the other week; still feels dirrty. <kori>it should at the very least recommend the user install a font <kori>rather than... not doing that <kori>doesn't need to be a specific font either <kori>just a post-install notification <kori>"hey, install a font. or else." <nckx>Those are all very hypothetical features that do not exist. <kori>what about when i installed vis, and it told me to update my $PATH? <nckx>Post-install notifications are fodder for the bitbucket anyway. If you rely on them, you're compensating for missing features. </pinion> <nckx>kori: That's Guix's search-paths feature. It's not a generic ‘oh, by the way…’ message printer, which is a good thing, it's a lot more powerful, but it only handles search paths (variables). <Marlin[m]>i also have a small gentoo chroot if i need to do stuff with unpackaged binaries <nckx>Wait a minute. It's actually a case of ‘do one thing & do it well’ done well. <Marlin[m]>gentoo's ACCEPT_LICENSE="-* @FREE" is handy, it blacklists all nonfree stuff <nckx>That's built into Guix 😇 <Marlin[m]>ya, i'll package the stuff i use, but i wanna be able to run it before i package too <kori>nckx: THERE MUST BE A SOLUTIon *nckx hands kori some coffee. <kori>noooooooo dont hand me coffee ill get even More headclenchy *nckx shrugs, drinks kori's coffee. <kori>always always use xfce4 as a stopgap on new distros *nckx has never used xfce (the DE) on Guix. Is it in a good state? <kori>sometimes a man's gotta ride the bull, though, am I right? later skater <nckx>I use thunar, sometimes, that's it. *Marlin[m] hands tea over <Marlin[m]>What is the one receiving the most attention? <nckx>xfce4-whiskermenu-plugin? <kori>considering it already took the liberty of deviating from the default no-xfce-config... <nckx>(Not saying it should be optional, just pointing out that it is packaged…) <kori>it should be configured more to be more like The Usual Desktop Metaphor (tm) (imo) <kori>i.e. panel on the bottom, maybe whisker menu, maybe a nice wallpaper with some gnus <kori>but tbh as I said I'm only using it as a stopgap <nckx>I suspect the deviation is less of a conscious choice than a side-effect of the general state of DE packaging in Guix, though. <kori>I wanted to get X working asap rocky <nckx>If it were a deliberate theme it would probably have a menu. <kori>I need to get Xlambda working <kori>learn how to use standalone X <kori>why is X installed setuid? <nckx>Rootless requires code/work which nobody has done yet. That's often the answer to ‘why iz stupid?? are u all stupid??’ on Guix. We are not, in fact, stupid. <kori>are you saying I can do that work? <nckx>Well, I am, but the rest is pretty competent. <kori>and it will in fact get accepted? <kori>im gonna read the [expletive manual] then <nckx>Unless it's very badly done, but even improving a patch is better than having none. <kori>because I definitely want to get involved with guix development <nckx>Just don't post any Pepe memes and we'll get along swimmingly. <nckx>Marlin1114: It was only added ‘Sat Mar 23 11:49:52 2019 +0000’. <asterope>is it required for guix to have initrd? I'm planning on just compiling in all that's needed, but maybe guix depends on something in its generated initrd <kori>nckx: oh I tend to avoid pepe on account of the fact it's been coopted by [POLITICAL RANT DELETED]... <kori>asterope: well, shepherd is in guile... <kori>but why can't just guile be pid1? <kori>because shepherd mounts the filesystems too <kori>answered my own question <Marlin1114>i have quite a bit of packages i started making and didn't finish <Marlin1114>nckx, did you guys get any progress with lutris since the time i was packaging it? <nckx>Marlin1114: I didn't continue working on it. Er. Hm. Did I say I would? <nckx>Marlin1114: My take-away was that you'd work on it, try to fix errors you understood, and ask questions when you didn't. <efraim>still can't get ghc-ncurses to build :( <Dynamicmetaflow>Hello! I was wondering if someone could assist me. I'm learning how to create packages and I'm attempting to package gnome-boxes (most of it's dependencies are already in guix except one). I believe I already created a package for libgovirt but for gnome-boxes runningo into an error related to bash-minimal and ./configure, I think it's a path issue. <nckx>what does that mean i am not from the intarnet ***Marlin1114 is now known as Marlin1113
<nckx>Oh, I guess you own it then. Unless it already existed. I really hope not. <nckx>Dynamicmetaflow: Could be. I can take a look if you paste the output to paste.debian.net or use wgetpaste (packaged in Guix). <nckx>Heads-up: there's also a glib-or-gtk-build-system with some extra G* magic. It's probably not related to this particular problem, but who knows. <Dynamicmetaflow>/gnu/store/q19l04vd2za80mk1845pz7r8cz29qk43-bash-minimal-4.4.23/bin/bash: ./configure: No such file or directory <Dynamicmetaflow>That's what I think the error is related to although not sure how to address it <Dynamicmetaflow>So i pull guix from source and created an environment, the steps from part one in the video <nckx>Dynamicmetaflow: That reads like there's no ./configure script to begin with. You can pass -K to the build command, this will keep the (failed) build directory in /tmp (printed as a Note: at the end), you can inspect it. <nckx>*If* that's the problem, adding auto{make,conf} to native-inputs is *usually* enough to let the gnu-build-system's bootstrap phase magically create ./configure. <nckx>Dynamicmetaflow: How do you build that linked file, anyway? <nckx>Yes. I've always hacked on the git repo directly and ‘guix build -f’ requires module imports you don't have. Just curious. <nckx>Not related to your problem. <Dynamicmetaflow>pull guix from source, ran some commands for clearning paths, ran make, created a git branch for gnome-boxes, <Dynamicmetaflow>and the package that i'm workin on added it to gnome.scm which has all the modules etc <Dynamicmetaflow>don't know if the way i did it is "correct" or not but I've gotten further creating package this way that I have before <nckx>Ah, so this is a fragment of gnome.scm? <Dynamicmetaflow>It's also been helpful to observe how other packages in gnome.scm have been created <nckx>That's fine, I just had to add a tedious amount of use-modules to get it to compile on its own. No prob. It's ‘building’ now. <Dynamicmetaflow>I added gnome-boxes and libgovirt to gnome.scm and i'm hacking away there with guix pulled from source in an environment <nckx>I've added the autofriends (why propaget them in libgovirt?) and let's see… <nckx>Spooked by propaget. *ate <nckx>./configure: No such file or directory <nckx>Oh, I should have, y'know, looked. <nckx>It's a meson package! I know this. <nckx>meson-build-system? Let's try. <Dynamicmetaflow>I was experimenting with the other ones too gnu build systek and glib or gtk <nckx>(gnu-build-system is only for ./configure && make && make install autotools [and compatible] packages.) <nckx>Using meson-build-system produces… new & exciting vomit! ‘No such file or directory: 'glib-compile-resources': 'glib-compile-resources'’ <rekado>janneke: I’d like to use one of our big servers to rebuild the bootstrap binaries to provide verification. <Dynamicmetaflow>FileNotFoundError: [Errno 2] No such file or directory: 'glib-compile-resources': 'glib-compile-resources' <rekado>janneke: do I just need to apply your patches to current core-updates? <nckx>Which is in glib:bin (this you just have to know). <nckx>Dynamicmetaflow: So you can add ("glib" ,glib "bin") to native-inputs, yep. <nckx>(If I'm being to basic let me know.) <nckx>You'll get another pretty self-explanatory error then. <Dynamicmetaflow>I was reading a little of the manual and studying other packages, I realized that gnome-boxes has almost all of it's depdencies and I wanted to package it for guix <Dynamicmetaflow>was trying to get virt-manager working and it was a pain, i tried gnome-boxes flatpak package and that worked for the most part <Dynamicmetaflow>so i thought ok gnome-boxes worked for me so let's try and package it <nckx>(For that error, I'd add gettext-minimal [also native]; if later it turns out we can't get away with -minimal we can always add more, but it's unlikely.) <nckx>Apparently virt-manager 5.5.0 is broken right now, although it works for me ☹ <nckx>ERROR: Invalid version of dependency, need 'libosinfo-1.0' ['>= 1.4.0'] found '1.0.0'. <nckx>Bugger. We might have all the deps, but it seems one's too old. <nckx>(Hint about the one you're probably seeing now is above.) <Dynamicmetaflow>Dependency libosinfo-1.0 found: NO found '1.0.0' but need: '>= 1.4.0' <nckx>Dynamicmetaflow: Please use a pastebin for multi-line pastes in future. <Dynamicmetaflow>I will, my apologies. Was feeling an adrenaline rush debugging this. <rekado>Dynamicmetaflow: looks like that should be the next step. <rekado>maybe you can just upgrade the version we already have. <nckx>Dynamicmetaflow: Indeed. You'll have to dive into the build, since simply bumping 1.0.0 to 1.4.0 won't cut it. <Dynamicmetaflow>rekado: Yeah, I'm going to look at libos now and see what it looks like, I imagine maybe it's changing the url to point to a recent one? <nckx>(mkstemp! "test/Makefile.in.XXXXXX" #<undefined>) → In procedure mkstemp!: No such file or directory <nckx>Maybe they've switched to Meson or CMake or another hip build system du jour. <mbakke>rekado: The 16G RAM nodes behind Berlin still use --cores=8 (the default), right? <rekado>mbakke: yes, I haven’t reconfigured them. <nckx>Dynamicmetaflow: Changing the version and sha256 is enough to download the new sources. Then you'll have to get them to build. <Dynamicmetaflow>Well at least we are one step closer! and one step closer to my first contribution, hopefully *croses fingers <rekado>mbakke: waiting for a “guix deploy” fix <mbakke>I'm trying to build ungoogled-chromium in parallel with `jumbo_file_merge_limit=8" on .141. <mbakke>Will be interesting to see if it works with 2GB RAM/CPU core :) <nckx>Hm, weird, they didn't switch build systems & at first glance tests/Makefile.in exists. <davidl>I just did a guix pull for root user and now Im failing a mozjs build when reconfiguring. Getting {standard input}: Assembler messages: <davidl>{standard input}:53610: Warning: end of file in string; '"' inserted *nckx is building 1.5.0 with the offending phase simply removed. Maybe they fixed their test suite anyway. *nckx rebuilds 6 all dependents — which include virt-manager, heh — if that works you won't even have to write your own. <nckx>Dynamicmetaflow: The libosinfo package in (gnu packages virtualization) has a 'disable-broken-test phase. I commented it out. <nckx>I had to double-check the name. Nice. <nckx>It is borken no longer (although I want to verify that before I push an update). <Dynamicmetaflow>if i were to comment out the borken-test would it not prompt me to do so? <Dynamicmetaflow>src/meson.build:122:0: ERROR: Dependency "libusb-1.0" not found, tried pkgconfig <nckx>libosinfo 1.5.0's on master. <Dynamicmetaflow>libvirt-glib also needs updating from what I can tell to build gnome-boxes <Dynamicmetaflow>I changed the version to 2.0.0 and seems to be building gnome-boxes again <nckx>That's good. I'm searching my build farm for libvirt-gconfig but it spins rust so will be slow. Also, my guess is we don't have it. *nckx has to go running now. <Dynamicmetaflow>I was getting a vomit earlier that it found 1.0 but needed 2.0 and determined that by updating libvirt-glib to 2.0 it updated gconfig <Dynamicmetaflow>Thanks for all the help! I appreciate it! hopefully by the time you come back i can get gnome-boxes running <Dynamicmetaflow>and hopefully soon submit this so it can be on master, my first contribution so yay! <Dynamicmetaflow>Hey all, was wondering if someone could help me understand this. I'm trying to build gnome-boxes for guix, initially libvirt-gconfig needed to be at version 2.0 so I updated libvirt-glib from 1.0 to 2.0 but libvirt-gobject needs 1.0 <Dynamicmetaflow>So libvirt-glib contains gconfig and gobject and I need different versions for each of them in order to build gnome-boxes <Dynamicmetaflow>Error that I'm receiving src/meson.build:122:0: ERROR: Could not generate cargs for libvirt-gobject-1.0: ***sushi_ is now known as ilikeheaps
<krusnus>hey everyone. just installed guix on my computer but cant seem to find vi. i assumed it would already be installed bu default as part of the base system but apparently not and there does'nt seem to be a package for it. what do i do? <krusnus>yes i know but i need vi for among other things <kori>but you shouldn't need to visudo? <kori>if your USER is in the wheel group <krusnus>yes but i want to configure the sudo timeout and such <krusnus>thanks anyway tho this will work for now :) <davidl>krusnus: try SUDO_EDITOR=nano visudo <krusnus>nope still get the message "no editor found (editor path = /usr/bin/vi) <rekado>krusnus: use the operating-system configuration to edit the list of sudoers <krusnus>rekado i just want to configur.e my sud <krusnus>configur the sudo timeout not the list of sudoers <rekado>krusnus: yes. We do that in the operating system configuration. <emacsomancer>stumpwm-specific workaround for my path issue: (setf (getenv "PATH") (concat "/home/emacsomancer/bin:" (getenv "PATH"))) <nckx>Also, visudo doesn't allow setting random $EDITOR/$VISUALs. <nckx>(Not that you should use it on Guix System.) <nckx>Saluton Dynamicmetaflow. <nckx>I'm not well-versed in glibbery. <nckx>What's the problem, exactly? <Dynamicmetaflow>So libvirt-glib contains three libraries libvirt-gconfig and libvirt-gobject <Dynamicmetaflow>to build gnome-boxes its requesting different versions of those two libraries, libvirt-gconfig 2.0 and libvirt-gobject 1.0 <nckx>And those aren't just sonames (‘library ABI versions’, kinda) but actual package versions? <nckx>Imma need a shower before I can face this. <Dynamicmetaflow>don't know how else to explain other than libvirt-glib includes other libraries and gnome-boxes needs different versions for libraries that are packaged together <Dynamicmetaflow>only ideas i have are to decouple the library and make them available separately or modify the requirements of the depdendency <kori>watching #hurd not talk makes me kinda saddu <kori>can I use xorg without it being setuid <kori>i.e. can i just startx from my user the usual way <rekado>Dynamicmetaflow: this doesn’t say that it needs different versions, no? <nckx>kori: You can try adding (service special-files-service-type `(("/bin/startx" ,(xorg-start-command)))) to your system configuration. <nckx>Whether it works is a different matter, but then you'd have a nice pretty bug to report. <rekado>Dynamicmetaflow: you have not added any input providing libvirt-gobject <nckx>Oh, your emphasis was ‘as a regular user’. No, I don't think so. <nckx>Dynamicmetaflow: If you can't find a way to make this work (rekado's suggestion seems eminently sensible), please paste your current package so we know where you stand. <Dynamicmetaflow>rekado: From my understanding libvirt-glib contains libvirt-gconfig and libvirt-gobject. When trying to build gnome-boxes the error beforehand was asking for libvirt-gconfig 2.0, after i modified the libvirt-glib to use version 2.0.0 I then receieve the error related to libvirt-gobject <Dynamicmetaflow>nckx: I don't fully follow rekado last statement of 'any input providing libvirt-gobject' <rekado>Dynamicmetaflow: make sure that the conditions mentioned in the pkg-config files provided by libvirt-glib (files ending in .pc) are met. <nckx>Crikey, is that what wgetpaste produces when you pipe a Scheme file into it? Something tries to evaluate it as bash? <rekado>Dynamicmetaflow: these files might say that certain libraries are “Required” are “private”. In these cases you should probably have libvirt-glib propagate those inputs. <kori>all I want in life is to be able to run custom WMs easily <nckx>Pretty much the definition of custom. <kori>that could work but like <nckx>(It's what I use so I am hella biased.) <kori>let's suppose i don't want a display manager <rekado>kori: I’d be happy if Guix could make that setup simpler. Could you outline what the problems are? Maybe we can figure out on guix-devel what needs to change to support this setup. <kori>rekado: guix could definitely use an actual barebones "%desktop-services" <kori>no gdm, no nothing, maybe xdm, at most <nckx>But then what's left to be started? <nckx>There's also a world of difference between ‘no DM’ and ‘OK maybe xdm’. <kori>nckx: no clue. maybe a %xorg-services isn't really needed even <kori>it's weird? that %desktop-services is so big <kori>it could be called %gnome-services and it would be a better name <rvgn>kori For that setup you can use `%base-services` and `xorg-service-type`; instead of `%desktop-services`. <kori>tl;dr I want to do "echo "exec your-wm" >> ~/.xinitrc; startx" and be happy with that <nckx>I personally use a slim (boo obsolete for ages spooky spooky) that logs me in automatically and execs .xsession so I never see a DM, I ‘boot’ straight into i3. Would that do for you, for now? <kori>I don't exactly mind a display manager... i guess xdm is "okay" <kori>but i usually don't use one at all <nckx>I'd second the suggestion for a mail in which you clearly state your goal(s), since they're all quite legitimate but it wasn't always clear to me what the priority was. Rootless X? Any working ‘startx’? Just hate GDM? Etc. <kori>i like a very very slim system <kori>i don't mind maintaining these packages if it's what it takes <kori>i signed up for savannah a bit ago <nckx>That's great, although it's not a requirement to interact with Guix in any way (until you want commit access, but that requires a bit of history and trust). *ng0 has xdm lying around.. sort of. <ng0>never went beyond the crucial thing, which was getting a service done <ng0>or did I sent it in and someone fixed something up? <nckx>‘guix system search xdm’ says no. <nckx>(If that is of course what it's called.) <ng0>hrm.. when do I find the time to upstream all of this. <kori>ng0: I plan to gain that trust :) <kori>nckx: what's your slim setup look like? <nckx>kori: Super basic, and probably not up to current Guix Best Practices, go caveat yourself, etc.: (service slim-service-type (slim-configuration (auto-login? #t) (default-user "nckx") (xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout) (modules (list xf86-video-intel xf86-input-libinput)) (extra-config (list %xorg.conf)))))) <nckx>%xorg.conf configures my input devices but isn't essential to the concept. <kori>a base-devel package that includes all the niceties (gcc, make, etc.)? <kori>that i can include in my $PATH <kori>i do know gcc is installed <kori>i get command not found when i try to run it <nckx>gcc-toolchain is the bloatiest one (it gives you a working gcc zomg bloat). <nckx>kori: …then maybe that's *not* how gcc is installed? <nckx>(Sorry, couldn't resist.) <nckx>Are you using gcc-toolchain? <kori>i know gcc is in my store somewhere <kori>quite a few places actually <nckx>There's no ‘build-essential’ like (I think Debian has) since, to be honest, that's an obsolete concept nowadays. Build *what*? Go? Rust? npm? Good luck. <nckx>kori: Sure, but ‘in store’ is not ‘installed’. <kori>is there still interest in ressucitating scsh-on-guile? <nckx>(And other ‘essentials’ like make & autotools are haemorrhaging ground to cmake, meson, and friends.) <kori>i should mention im like <kori>i was/am interested in making a scheme-based distro <kori>wrote a blog post about it on medium.com/@kori <kori>which is why im asking all these sorts of questions and why i want to get involved with the development of guix <kori>i figured a few days ago it'd be much better to share code with guix than not do that, and id like to contribute <nckx>Dynamicmetaflow: Yes actually, as spice-gtk. <nckx>(spice-client-gtk-3.0 is the name of the pkg-config file that spice-gtk version 0.36 provides. I know, it's confusing.) <Dynamicmetaflow>most recent error with gnome-boxes src/meson.build:122:0: ERROR: Vala library 'spice-client-gtk-3.0' not found <pkill9>kori: the gcc package was hidden as people kept using that instead of gcc-toolchain <nckx>Be sure to add comments to as many propagated inputs as possible. <kori>pkill9: i see there's a "base" package module <kori>where's make and binutils and pkg-config and all that stuff ; <rekado>kori: the gnu-build-system provides them. <nckx>And if you want to use them yourself (keep in mind that most non-trivial projects are easier to package Guix than to attempt build by hand), ‘guix install make pkg-config’. <nckx>A *working* binutils is already part of the (gcc-)toolchain. ‘guix install binutils’ will probably disappoint, so don't. <nckx>But TL;DR write packages not command lines. <nckx>(I get this vague impression you like Scheme a bit so that shouldn't be hard.) <kori>nckx: yes, i want to write packages, but i'd also like my regular tooling (st, firefox, swm + Xlambda [and eventually just Xlambda]) <kori>this is a weird situation i find myself in <nckx>I can't believe st isn't in Guix. <kori>it's weird that it would be... <kori>it being configured at compile time and all <kori>i have my own st with a bunch of patches that i use on another machine <nckx>Well, you can inherit it and add a phase to configure it. <kori>so ideally id like to locally package that <nckx>And Guix takes care of a working base package that's always up to date. So it makes sense. <nckx>I run a (very mildly) customised IceCat but I'm very glad it can just inherit from Mark's hard work. <kori>i appreciate his hard work but i very much so want just vanilla firefox, i don't care for any of icecat's differences <nckx>Then you'll have to do your own hard work. <kori>i know it's not gonna make upstream but a local package is fine <nckx>I'm sure there are people running Firefox though, I'm not sure how much it's a taboo on here (being non-FSDG but also, c'mon, not evil) but it's moot since I don't know them anyway. <kori>at least this time around of me putting on the packager hat i feel like my work is probably going to be more widely used <kori>packaging for crux was fun but... eh, it was stuck on crux <kori>packaging for guix means my packages might run anywhere guix runs <nckx>kori: I think it's best to start with ‘upstreamable’ packages first, then writing your own, because upstreamable ones receive free unlimited (well…) review rounds 🙂 <kori>font packages are easy to make <kori>and theres lots of fonts i like <nckx>Indeed. We can use more fonts. I'm lucky that my favourite (Hack) is already in Guix, but we need more. <kori>i packaged like 15~20 fonts for crux :) <kori>i need a working X server first lol <kori>i was using xfce but i got tired of it, i want my st :< <nckx>It's a lot easier to customise the boat if you first get on the boat, even though you don't like things about it, than to swim around with a hammer. <nckx>Wow, that was a bizarre metaphor. <nckx>What I meant was: just install Guix, customise a working system 😛 <nckx>Dynamicmetaflow: Looking at /gnu/store/…-spice-gtk-0.36/lib/pkgconfig/spice-client-gtk-3.0.pc, I see: <nckx>Requires: gtk+-3.0 >= 3.22 spice-client-glib-2.0 <nckx>Are both those packages propagated by spice-gtk? <nckx>I know, I could check myself, I'm busy. *nckx checks themself before they wrecks themself. <rvgn>Any update on core-updates-->master merging? <genericspider>Hey everyone, I need a little help with building GNUnet. I'm fairly new to Guix. When I'm compiling GNUnet libextractor fails to build. I posted where the error seems to occur in the logs here: https://pastebin.com/T2LeSHpD <genericspider>What exactly is going on here, and how would I go about resolving the issue with the build?