<quiliro>xavierm02[m]: how did you find out where texlive-texmf was? *nckx guesses ‘grep define.*texlive-texmf gnu/packages/*.scm’ or equivalent. <quiliro>This has been great information to learn how guix is constructed. Thank you xavierm02[m] and nckx <quiliro>and of course thanks to bandali with your sugestion to use guix edit <gnu_srs>quiliro: I managed to start the guix-daemon. Next: guix build hello; guix build: error: could not find bootstrap binary 'tar' for system 'i586-gnu' <gnu_srs>Seems like the bootstrap binaries are needed? <quiliro>gnu_srs: i have no idea...but think you are progressing...keep on..i would love to see guix work with hurd <gnu_srs>I can do the cross-builds from a Linux box but how? The native tools are already there. Cross-compiling from a guix in e.g. amd64? <ruffni>i'm having trouble setting up my emacs config to evaluate whole buffers in guile (i'm using geiser). it says "ERROR: In procedure scm-error: no such language objcode" <rekado>Hey there, I’ll be unavailable for the next few weeks / months. Looks like our baby is coming earlier than expected. <PotentialUser-33>how can i add a channel as i am installing guix? i want to use some packages from a certain channel in my operating system configuration. i am manually editing the /etc/config.scm file during the installation process <Minall>I'm unable to share internet on Guix, maybe I need to install something? <apteryx>Minall: what do you mean by "share internet"? <Minall>I'm trying to share internet from my laptop with wifi, to another pc with ethernet. <Minall>apteryx: In other distros it works changing ipv4 mode to sharing, but in guix nope <Minall>Does guix system reconfigure need admin privileges ? <Minall>apteryx: Then this is strange... I can do it without admin privileges... <Minall>What's the installation of the gnome shell? <Minall>I'm trying to install gnome extensions, but the page of the gnome extensions says I don't have it installed... <apteryx>Minall: it shouldn't succeed unless you are root <Minall>apteryx: Maybe it has to do with: My latest reconfigure bringed an error, when I had to readd the passwords of my users <apteryx>As for your other questions (share internet + gnome shell), I'll defer to more knowledgable users; I haven't experimented with those yet. <apteryx>But I think there must be an operating system definition in the manual making use of Gnome (desktop.scm perhaps). <apteryx>you could have a look at it or try it in a VM (producing a QEMU VM launching script with 'guix system vm desktop.scm') <Minall>The gnome page says that the package for installation of what I need in other distros is: chrome-gnome-shell <Minall>But I don't see a package like that on guix... <apteryx>I cehcked teh manual; you need to add the gnome-desktop-service-type to your operating-system services list <Minall>You were right, reconfigure didn't complete <roptat>good news! I was finally able to split mariadb into 150 MB of "out", 40 MB of "lib" and 5 MB of "dev"! <roptat>guix size reported total: 456.1 MiB for perl-dbd-mysql and now only 201.6 MiB <roptat>still a lot, but we also got rid of a few libraries in the dependency graph :) <roptat>(ncurses, jemalloc and one version of zlib out of two that were used) <roptat>python-mysqlclient went from 537.0 MiB to 230.2 MiB <roptat>it lost a reference to gcc itself <sirmacik>is there a way to tell guix to download only precompilled packages and skip those it needs to build from source? <roptat>you can use -n (--dry-run) to see what would be built and exclude those <sirmacik>I don't have the fastest laptop so I want to skip on rust builds and so on, just download the latest prebuilt version :/ <roptat>if you're building, it means it was not built yet <roptat>rekado, did you know the https certificate for berlin.guix.gnu.org is incorrect? :) <roptat>I wonder when you see a succeeded build on berlin, how do you know which commit of guix it was (same for failed builds) <sirmacik>roptat: I know, it means that in those cases I just want to download the latest prebuilt package from repositories if I don't have it already not build latest available version ***benny is now known as Guest81054
<raingloom>just out of curiosity: is cross compilation with mingw still supported? i tried `guix pack --target=i686-w64-mingw32 lua` but got an error about a missing readline header <g_bor>I was looking at how we currently do nginx config on berlin. <g_bor>I pondered a bit on the TODO list item of static-web-site-service, to extend nginx from there. That would be a great idea. <g_bor>Looking a bit around, I found that currently we could only extend nginx with server blocks, but having a look at the nginx master config we need something more sophisticated here. <g_bor>I would propose to modify the nginx extensions a bit along the following lines: <g_bor>1. either do a server-block merge at the extend procedure, by collecting server blocks having the same name. <g_bor>2. or create a server-block-extension type, similar to a server block, but more limited options, and merge these with the server block of the same name. <g_bor>I was also wondering if we could refine location a bit.... <roptat>qtbase went from 1331.7 MiB to 1083.1 MiB :) <roptat>and I guess I tested every direct dependent, so I can probably say it's safe. I'll send the patch later today :) <Minall>How can I install gnome extensions on guix? <Minall>I'm trying to mount a device, and I get an error: unknown type exfat... What should I do[? <g_bor>you could try installing fuse-exfat <g_bor>I belive it should be a system package, but not sure. <tune>youtube-dl is working for me but now newsboat is failing with ssl errors so maybe that problem wasn't totally fixed... <tune>I've been having intermittent ssl problems for weeks now <Minall>g_bor: Yep, I installed that package, but no luck <Minall>I'm still unable to mount exfat} <g_bor>Minall: did you install it on you profile, or using the packages fielsd in you system configuration? <PotentialUser-30>i am getting an error when i try to install gnu guix. the error says `<hash>-grub-efi-2.04/sbin/grub-install --boot-directory /mnt/boot --bootloader-id=Guix --efi-directory /efi` <PotentialUser-30>`error: <hash>-grub-efi-2.04/lib/grub/i386-pc-modinfo.sh doesn't exist. Please specify --target or --directory.` <g_bor>this does not look like an error message... <PotentialUser-30>it says error in red and stops the installation process while i am running `guix system init` <g_bor>I believe that mount should be /boot/efi <roptat>can you share the grub config from your config.scm? for instance on paste.debian.net? <roptat>and maybe include the file-systems section too <tune>sneek, later tell Minall are you still using the normal mount command? if so, try: mount.exfat-fuse <PotentialUser-30>i have tried mounting it at /boot/efi and changing the config to match, but i got the same error <tune>seems like I have a file-system section for efi and you don't <roptat>PotentialUser-30, try to mount to /efi instead of /mnt/efi? <nckx>Sisyphe[m]: Just a guess, but that smells like you have a ‘name’ in your (quoted) build-side code that should be a ‘,name’, since ‘name’ is indeed only defined on the host side. Could that make sense? <nckx>tune: It will work without the file-system section, you're just responsible for mounting it yourself before running reconfigure. Can be used as a sort of fail-safe, I guess. There are people who don't like to keep their EFI partition mounted all the time because it can corrupt in some corner (crash) cases. <nckx>So GRUB is complaining about a missing ‘i386-pc’ (non-UEFI) file when invoked as ‘grub-efi’ :-/ Weird. <PotentialUser-30>even mounting the efi partition at /efi instead of /mnt/efi, i get the same error <PotentialUser-30>i successfully installed guix a couple days ago, using the graphical installer, using efi bootloader <nckx>There may be some assumptions (in Guix) that /boot/efi is a subdirectory of /boot (where the meat of GRUB is kept), but I guess you already tried that. Did you get the *same* error then? <nckx>PotentialUser-30: Yeah, your configuration looks identical to mine (which has always worked), very strange. <PotentialUser-30>i can try /boot/efi again. i will mount my efi partition at /mnt/boot/efi and change the configuration to point there. <nckx>That moment where you accidentally hit C-c C-r and send rot13-encoded mail to your mother 🤦 <tune>is it an old machine that might lack efi support? <PotentialUser-30>when i successfully installed via graphical installer the other day, it used efi <nckx>The ‘solutions’ I find to fixing this bug on trad distroes is to ‘sudo apt-get install --reinstall grub-efi-amd64’, which Guix is obviously already using. *nckx just tried Sisyphe[m]'s code in a REPL; of course it works fine. Hm. <efraim>Isn't name previously defined as the name of the package? <efraim>I know it isn't unquoted but it's not a variable name I would've chosen <nckx>Sisyphe[m]: Are you getting any other warnings or errors? That's not the conventional way to import modules into build code, for example (grep for #:modules for examples), but I don't know off the top of my head if it could work. <nckx>efraim: It is, but it really shouldn't matter. <nckx>(Famous last words, I know; still.) <nckx>Easy to test by search-replacing all names in that snippet. <nckx>Replace them with ‘efraim’. *efraim goes back to BBQing <Sisyphe[m]>I already have #:use-module (ice-9 match) at the top of my module but removing (use-modules (ice-9 match)) in the lambda causes Unbound variable: match-lambda <nckx>Is this just for debugging while writing your package? TBH (and sorry for implying you should ignore bugs instead of confronting them head-on), I just always use ‘(invoke "ls -…")’ in such cases. Less typing than doing it properly in Guile. <Sisyphe[m]>nckx: thanks, I would have liked to improve my guile skills but it's actually a lot simpler <nckx>Yeah, I feel appropriately dirty for suggesting that… <Sisyphe[m]>nckx: too late... I'm gonna use that everywhere from now on... <nckx>I've always found FTW a bit baroque for my few casual use cases. <nckx>Typo: "ls" "arg"…, not one single string. <mbakke>I will be without internet and computers until Sunday, have a nice weekend Guixers :) <sneek>Welcome back Minall, you have 3 messages. <sneek>Minall, tune says: are you still using the normal mount command? if so, try: mount.exfat-fuse <sneek>Minall, nckx says: For now, you can invoke ‘mount.exfat /dev/…’. <sneek>Minall, nckx says: (or mount.exfat-fuse, I forget the exact name.) <Minall>When I installed some emacs package from guix, do I have to make extra configuration? <nckx>You need to add it to your .emacs, how that is to be done is probably a religious argument as any other. I use ‘use-package’ (from Guix). <janneke>crap, why does webkitgtk depend on python2 *and* python3 (and 135 other requisites) <Minall>nckx: use-package?, what do you mean? <nckx>Minall: For example, to actually use emacs-yasnippet in emacs after installing it with Guix, I write: (use-package yasnippet :ensure t :init (yas-global-mode 1)) <janneke>nckx: ;-) still, i wonder why guix pack adds (seems to add?) python2 from native-inputs; <efraim>Or it was ffmpeg, I don't remember <nckx>janneke: I admit to just using grep -r on the outputs of ‘guix size’ in such cases. <janneke>iwbn if this is a guix pack bug, we don't need any native-inputs in a pack <Minall>How can I add extensions to the gnome shell? I installed the page extension on icecat, but nope <nckx>janneke: Oh, I thought you meant that webkitgtk actually depended on both Pythons. You're right; it doesn't. IWBN. <janneke>nckx: they are both used in the build, and they both end up in my pack; when i remove them after untarring the pack, it still runs fine <janneke>hmm, adding python to gnu hello's native-inputs does not have it end up in a pack -- that's good <Minall>How can I install gnome shell extensions?, I added the extension to icecat, but nope <janneke>so, some dependency somewhere has python in inputs or propagated-inputs... <nckx>Minall: We're not ignoring you, we simply don't know 🙂 <janneke>it would be so nice to have one gnome developer use guix <Minall>Welp, I found a package named: gnome-shell-extensions, let's see if that's the one <tune>finally my updates are finished after a few hours. I had to build cataclysm-dda and icecat <ilyaigpetrov>Where can I read about maintaining my own substitutes server (packages repository)? <sneek>ilyaigpetrov, you have 1 message. <sneek>ilyaigpetrov, vagrantc says: RPi boards require non-free firmware to work, so probably not a good candidate for guix <ilyaigpetrov>I plan to try building my own substitute server that works over IPFS protocol... <ilyaigpetrov>And the first step, I think, is to read the guide on maintaining a substitutes server over https <nckx>ilyaigpetrov: I guess the place to start would be ‘8.8.21 Continuous Integration’ in the Guix manual. <nckx>(which I was just editing 🙂 ) <nckx>Hm. Probably a bit over-engineered for a personal 'toot server, but can still be helpful. <nckx>Oh yes, that will be very helpful. <nckx>I've heard people say that you don't actually need to learn Scheme when installing Guix System, and fervently disagree. Frustrated users & dreams are thattaway. <bandali>it’s kinda like “do i need to learn haskell to use xmonad” <bandali>many would say no, and it’s partly true, but you’d pretty much be trying to copy paste and salvage things from others’ configs <bandali>whereas knowing haskell (or in this case guile) would actually be helpful in understanding to some degree what is actually going on, and what the expressions you write in your config file(s) actually do <nckx>bandali: Yeah, I wasn't sure how to word that. If you're willing (and, hopefully, eager) to learn Scheme ‘hand in hand’ with Guix, you'll be fine. But saying that it's optional (as I've seen done) will send the wrong message. <nckx>(Ex-xmonad user here 🙂 ) <yu44>Hello to everyone! I have a question. I cannot start .run, .elf, .appimage (compiled for Debian). Is this a problem with dynamic libraries? I cheated on LD_LIBRARY_PATH, but still won 't work. <ilyaigpetrov>yu44: did you check that architectures match and did you add +x flag? P.S. I never had guix running myself :-) <nckx>yu44: Running such binary blobs isn't really supported on Guix System, although I've heard of an FHS service floating around that can make them work. <nckx>LD_LIBRARY_PATH won't help binaries find their ELF interpreter (usually hard-coded to /lib/ld-linux…, for example). <yu44>I am not running firefox-bin, balenaetcher.appimage. Error: File not found <pinoaffe>I made some changes to guix (my tree is at https://gitlab.com/pinoaffe/guix_fork/tree/fpga ), but when I try to run `guix pull` it throws seemingly unrelated build-time errors for guix-extra, stating that it's unable to find patches for icecat - any idea what the cause might be? <nckx>yu44: ‘I am not running firefox-bin’ — why mention that explicitly? Are you referring to a previous discussion? <nckx>OK. I don't see how Firefox would be a special case, though. <str1ngs>nckx: the dynamic linker is hard coded INTERP header of the elf. correct <nckx>Not about that, but I am unsure whether these nublob formats act as a single executable or whether you need to ‘unpack’ them and patch every single embedded binary. I don't expect ‘patchelf foo.appimage’ to Just Work. With some luck, I'm wrong. <nckx>Oh, wait, I forgot, I'm never wrong. Damn it. <nckx>str1ngs: Do you have experience with these formats? <str1ngs>how much is patchelf used in guix. I would think guix would rely more on using gcc and ld to set the dynamic linker and run paths <nckx>Haskell & julia might be exceptions, but it's nothing like NixOS. <str1ngs>I'm familiar with elf format some. but I have never heard of nublob. I'm assuming it's thats what appimage uses? <nckx>I mean that AppImages & Flatpaks are the new blobs. <str1ngs>aww. what does literal program file describe there format as. e.g. file ~/the.app.image <nckx>(And whatever ‘.run’ and ‘.elf’ refer to above.) <str1ngs>elf though I guess you could call a blob. Is more technically a executable format, like windows PE and MacOS DWARF. though I'm not familiar with appimage format <str1ngs>if you can ./app then my guess it's an elf. but it probably use some .data magic <str1ngs>elf is extensible so probably they just extended it to embed dependencies the real executable into the elf file <str1ngs>though if it's an elf then in theory you could use a file server to provide the dynamic linker. and the appimage should work. since it's self hosting *nckx can't bring themselves to care about these fads. It's design by PR: lobby upstreams to use your ‘universal’ binary blob format, knowing full well that it's the *distributions* that will have to do all of the work. But that's because I'm a cranky zealot who cares about practical software freedom. <nckx>str1ngs: I think that's what the proposed FHS service did. <str1ngs>it probably did that as well, I thought though it did some more stuff. particularly with containers <str1ngs> http://dpaste.com/3QV2588 should provide a standard dynamic linker locations for FHS binaries. it has extra's in there that is probably not required just for this <nckx>str1ngs: Heh, I do the same thing (but with /lib instead of /lib64, I thought…). <str1ngs>also /lib64 is not smart is assumes x86_64 lol <str1ngs>each arch has it's on dynamic linker path of course <nckx>I can't even remember why I added that line, time to get rid of it, I guess. What do you use it for? <str1ngs>I mainly use this service when I'm using something third party that I don't want to hack to work with guix <str1ngs>like a trivial bash script for example. or I have a go binary that I don't want to rebuild <gnu_srs1>Hi, sorry to trouble you again: guix build bootstrap-tarballs --system=i586-gnu on GNU/Linux amd64 <gnu_srs1>guix build: error: could not find bootstrap binary 'tar' for system 'i586-gnu' <str1ngs>pkill9: right I meant to check that out. this was a quick and dirty on my part. <nckx>I knew it was some unix command named person but all I could think of was rm -rf. <str1ngs>this might be good for inclusion into guix? :) <str1ngs>pkill9: also this will only work on x86_64 just an fwi <pinoaffe>nckx: probably part of the guix system, it gets built(or downloaded, ofc) when guix is updated <nckx>pinoaffe: I guix pulled your repository earlier, it worked fine. <pinoaffe>and I think I figured out the issue, seems like the diff creates a circular dependency <nckx>Thought guix-extra was some other channel that I needed to add. <nckx>Yeah, that's what the backtrace smells like. <nckx>(But then why didn't I hit it…?) <pinoaffe>nckx: this only recently started happening, and in the python build system code I found what looks like a workaround for this issue <nckx>Well, by ‘earlier’ I meant ‘when you asked this earlier this evening’, and I pulled from master, so… 🤷 <nckx>Glad that you've found the culprit. <pinoaffe>nckx: my master is clean, no patches of my own on there :) <nckx>Eh, sorry, I meant ‘faber’, but yeah, looks likely that something was still wrong (or, well, right). Took bloody ages to pull but did.