<vagrantc>appropriating widely used terms is hard enough, and the metaphor doesn't exactly seem strong :P <vagrantc>it's free in the sense you can typically look at the clouds (from very far away, without much ability to actually see what's going on), and make approximate copies with watercolor paintings ... but not really enough to actually functionally copy them ***duds-_ is now known as duds-
<pkill9>vagrantc: the metaphor i mean is that the data exists in the 'cloud' i.e. as copies between other people's computers <iskarian>Does anyone know why 'search-path-as-string->list' in (guix build utils) uses string-tokenize instead of string-split? <Gooberpatrol66>Is it possible to install firefox extensions from guix? I see chromium-extension.scm that lets you install chromium extensions, but I don't see an equivalent for firefox. <wirez>is guix made so there's always a bunch of stuff running and ports open n shit? or can you run a barebones config that has basically nothing running? <wirez>and hopefully tiny footprint too so it could go right onto embedded <Gooberpatrol66>wirez: guix requires guix-daemon to be running (though i hear there's plans to remove it). you can make a very minimal config though. i have a machine with 20 processes running. i don't know about ports though. and embedded is unlikely as guix uses a crapton of disk space. <Gooberpatrol66>because it's a functional distro, if something is built with different build dependencies, it duplicates it as another copy. <kittyblam>it really isn't very bloated compared to similiar projects imo <wirez>ya but if you only build once it should be able to be super tiny <wirez>or atleast no more than what's configured <abrenon>bien, aunqué tengo mucho que hacer hoy por la mañana : ) ***lukedashjr is now known as luke-jr
<efraim>did you make any changes? I don't know how the 6 got there <NicholasvonKlitz>Hmm, just pulled the master branch and for some reason I can't do a `make authenticate` <cbaines>NicholasvonKlitz, you probably need to ./bootstrap and ./configure ... again <NicholasvonKlitz>Hmm I did `./bootstrap.sh`, `./configure --localstatedir=/var`, followed by `make` <bdju>efraim: didn't make any changes <cbaines>NicholasvonKlitz, running make clean-go might help <NicholasvonKlitz><cbaines> "Nicholas von Klitzing, running..." <- Yep that fixed everything. Thanks again <bdju>did another pull and update and it worked this time... <leoprikler>I'm pretty sure that tests just need to stay disabled *leoprikler will be back later <liltechdude>leoprikler: Can you add iskarian's or your defenition to guix repo? <pkill9>does anyone have a guix system config for creating a guix system image for the raspberry pi? <mothacehe>pkill9: the rpi requires binary blobs to boot so you won't find this around <bdju>guix's nheko is a few versions behind latest release <apteryx>it's been a while! It was a bit quiet in here :-) <roptat>I plan to push the newest translations from weblate on Sep. 1 (or 2 depending on how much work I need for the LFS release :p) <roptat>Since the French translation is down (apart from package description), I ended up updating the Esperanto translation :) it's still missing ~1000 strings <roptat>my English is getting worse and worse... <abrenon>oh the french translation is complete ? that's great news ! also, cool that you can translate to esperanto <roptat>g_bor, I think you started the Hungarian translation for the website? Note that you're only adding suggestions <roptat>I'm not so good at esperanto, so hopefully someone can proof-read me :) <civodul>roptat: bonege, mi povus provi legi tion :-) <civodul>apteryx: hey! i was wondering: is "ansible-core" different from "ansible"? <roptat>mi ŝanĝis kelkajn vortojn, ekz. modifilo -> elekto (option) ĉar mi trovis ĝin en aliaj tradukoj <roptat>so is mine, I haven't spoken for a while ^^ <civodul>the glossary feature on Weblate is neat <roptat>yeah, I like that we can use it to coordinate and make sure translations are consistent <roptat>another thing I've been using is the search on all projets (at fedora's weblate, hosted.weblate.org and gnome's weblate) <roptat>I'd like to find something more centralized though, to search in all of them at the same time <apteryx>civodul: ansible has been split in two parts in recent versions; the core (the old ansible package) and the collections (the new ansible package) <roptat>so, we use sekurigo on the website, but everyone uses sekureco instead, so I'm wondering if we should change that? sekurigo is technically correct I think <civodul>apteryx: oh i see; so i guess we can keep the old one around for some time? <roptat>mh, more like sekurigo "the act of securing something", sekureco "security" <roptat>Esperanto is hard because of the nuances it can have ^^ <civodul>roptat: at first sight "serkureco" sounds more accurate to me, but i dunno <roptat>maybe sekurigo is closer to hardening in English <roptat>how would you translate "barebones"? <civodul>hmm good question, not even sure in Fr <cbaines>I hope you had a good break civodul :) <cbaines>barebones == minimal, and maybe that translates better? <roptat>ĉu kompreneblas? "Minimuma praŝarĝebla kruda kopio de sistemo Guix en Linukso por Pinebook Pro" <bsturmfels>hi, if I load up this system image config on a VM, cfdisk shows a 40M partition of type "EFI (FAT-12/16/32)" at /dev/vda1 and my root ext4 partition on /dev/vda2. Is that EFI partition generated by Guix? <bsturmfels>(I'm uploading the image to a VM provider, so not sure if that partition is related to them or me) <mothacehe> bsturmfels: the EFI partition is created if you use the efi-raw (default) image type. <mothacehe>if you choose the qcow2 image type it won't be present <bsturmfels>ah I see, thanks mothacehe. Machines provisioned at my VM provider only seem to have a single partition of type ext4, so I'm trying to figure out how to do something similar <mothacehe>we could have a "raw" image that would do that, but it should be possible with some hacking, let me see <bsturmfels>mothacehe: sorry, got disconnected, thanks very much I'll try iso9660 <bsturmfels>mothacehe: excellent, thanks for the link! very helpful to see someone else's certbot/nginx and such. I haven't figured any of that out yet <attila_lendvai>what am i doing wrong? i checked out guix.git, then git reset --hard to the commit that my guix install is at (as per guix describe). why does it fail for me then: guix build -L. bitcoin-core (with a not too useful error) <attila_lendvai>(all packages fail with the same, i'm trying to add go-ethereum) <attila_lendvai>am i missing some magic, like e.g.: guix environment guix ? damn, am i in an RTFM situation??? :) <attila_lendvai>it's a scheme error: Wrong number of arguments to #<procedure display-error (_ _ _ _ _ _)> <apteryx>why do you specify -L. ? what is your current directory? <apteryx>I'm guessing dongcarl may have an idea :-) <slyfox>'./pre-inst-env guix build ...' should work <attila_lendvai>apteryx, i'm in my local checkout of guix, where i'm adding my new package. (attempting to build bitcoin-core was just a test, that the problem is not with my new package) *attila_lendvai tries to generatea pre-inst-env <apteryx>if that's your local checkout, you want to do a slyfox suggested, yes (use pre-inst-env) <attila_lendvai>i seem to be getting the same error (trying bitcoin-core now, just to be sure) <attila_lendvai>guix environment guix; ./bootstrap; ./configure --localstatedir=/var; ./pre-inst-env guix build --keep-failed -L. bitcoin-core => Wrong number of arguments to #<procedure display-error (_ _ _ _ _ _)> <slyfox>'./bootstrap && ./configure --localstatedir=/var && make && ./pre-inst-env guix build bitcoin-core' should be enough <attila_lendvai>slyfox, why? then i get a missing package error (remember, i'm trying to add a new package, go-ethereum) <slyfox>if you added to existing .scm files as part of guix checkout it should be visible as is *attila_lendvai has realized he forgot to build guix (i.e. the make part) <slyfox>-L. should be needed only for non standard location <attila_lendvai>slyfox, do you recommend guix environment guix --pure, i.e. is the --pure part essential/advised? *attila_lendvai is making notes and plans to update the docs <leoprikler>iskarian: What definition was liltechdude talking about? <slyfox>--pure removes your existing PATH values (and other variables), it should not hurt <attila_lendvai>slyfox, i noticed one anomaly with it, so i stopped using: it seems to execute my .bashrc which contains tput and direnv stuff that are missing in --pure *attila_lendvai restarts using --pure <attila_lendvai>slyfox, another thing: without using --pure, i had to pass this to configure: ./configure --localstatedir=/var. with --pure a simple ./configure works <slyfox>--localstatedir=/var is needed for inplace guix to use /var/guix/daemon and friends <slyfox>lack of warning in --pure mode sounds like a bug <attila_lendvai>slyfox, hrm... FTR, it wasn't just a warning. ./configure simply errored out, prior to generating the pre-inst-env file <attila_lendvai>slyfox, do you think i should cancel the build, and retry with --pure, even though it seems to be running fine? or shall i wait for any errors? <slyfox>I'd suggest restarting with --localstatedir=/var <civodul>cbaines: hey! yup, i had a good time off <leoprikler>what's it with all those abi breakages lately? :( *attila_lendvai is finally testing it again, after some more wasted minutes with shell, the language <civodul>fun fact: "git log --since=last-month |grep ^commit |wc -l" says 145, which is like a 10th of what we have on a normal month! <leoprikler>It could also mean that we had 3 Rust packages less. <iskarian>ah, it looks like everyone is coming back from holidays :) <leoprikler>I've submitted my version as #50273, perhaps you may want to review it <lispmacs[work]>Hi, I use a manifest with inferiors and have recently got this error: <lispmacs[work]>The message is linguistically clear enough, but I'm a little confused about what is needed and why. Here is my manifest: <iskarian>leoprikler, ah, have we not patched sdl-config to produce the correct paths? <iskarian>Okay, it looks like because of sdl-union, sdl2-config uses its original prefix, not the union prefix; a fix seems to be to call sdl2-config with `--prefix=<sdl-union-directory>' instead <iskarian>See the jumpnbump package for an example <iskarian>(I wonder if sdl-union should be modified to make this change itself?) <iskarian>Anyway, I've got to go, but hope that helps! <leoprikler>Hmm, good point, that way we won't have to hardcode the flags themselves <dongcarl>attila_lendvai: If you need more help: what is the overall goal? <liltechdude>I cannot find `guix build` flag to forcing guix download source evert time when `guix build` invoked. Firsly all seems to be okay, but after this, guix take predownloaded sources from somewhere. How to fix this? <vivien>Dear guix, h-node says that I can run with 100% free software, but I "might have to blacklist the amdgpu kernel driver or else the system won't boot." How do I do that? <apteryx>leoprikler: your computationally inexpensive fix to srfi-9 records ABI breakage: $ git grep -l bootloader-configuration | grep .scm | sed 's/.scm/.go/g' | xargs rm ;-) <vivien>(to be clear, I’d like to "blacklist the amdgpu kernel driver", I know the system won’t boot without it, I’m not interested in a non-booting OS) <maximed>liltechdude: Why do you want "guix build" to download the source code again? <leoprikler>vivien you can pass a modprobe blacklist on the command line <maximed>This is by design. Not downloading it again saves disk and network I/O. <attila_lendvai>dongcarl, i have a go-ethereum package (as per `guix import go`, added to finance.scm), and i want to test-build the package. i have run the bootstrap, and i'm getting the same error from: ./pre-inst-env guix build --keep-failed -L. go-ethereum; and without -L. a missing package error <maximed>So I'm wondering what your reasons are for trying to download it again. <dongcarl>attila_lendvai: You have successfully `configure`d and `make`d guix inside your tree? <apteryx>liltechdude: if you want to force downloading/regenerating the *source* again, you can do: guix build --source --check <leoprikler>vivien: the kernel command line, you can pass options in through (kernel-arguments ) in Guix system or by pressing F2 in grub <attila_lendvai>dongcarl, yep: $ ./pre-inst-env which guix => /home/alendvai/workspace/guix/guix/scripts/guix <maximed>leoprikler: About the ‘I made a note to move png-file? from the minetest build system to (guix build utils)’: <maximed>I meant that I made such a note in my mind <maximed>I intend to move it some time after 'master' is merged into 'core-updates' again <dongcarl>attila_lendvai: Okay, first stop using `-L.`. Then, can you send me the error you get? <leoprikler>should have been a source code comment, but as long as you remember :) <leoprikler>on a related note, you don't have commit rights tho, do you? <attila_lendvai>dongcarl, i'm sorry, i was not attentive enough: i was on the wrong branch (because i started to patch the docs, and i got distracted)... thank you for the help! *attila_lendvai continues by dealing with a missing dependency error <dongcarl>attila_lendvai: I recommend that when there's a lot of confusion, try using `git clean -xdff` to wipe everything and start from scratch :-) <liltechdude>Okay, my real problem is getting new "actual hash" for git repo that reffered with "file://" method <attila_lendvai>dongcarl, it was a lame mistake that i rarely do, but it's all new now and i of course suspected the environment... :) <liltechdude>In my opinion changing commit field in origin should check if that commit actually exist <maximed>liltechdude: That would be nice to have <maximed>though I don't quite see how this could be implemented without decreasing performance <maximed>a database of "git commit <-> guix hash" would need to be kept <maximed>and when guix is computing derivations, it would need to check if they match <maximed>Isn't there 'git-checkout' or something like that? <maximed>'git-checkout-compiler' from (guix git) it is. <maximed>You just give it an URL and a commit, it doesn't need a hash <maximed>A downside is that it doesn't know the hash in advance, so the derivation of the package with that source code cannot be computed until it has been downloaded <maximed>(and on multi-user systems, each user will have to download it separately, unlike git-reference) <maximed>so the ‘these derivations will be built’ and ‘this will be downloaded’ output of "guix build" will be less accurate <attila_lendvai>dongcarl, do you have experience with packaging monsters like go and go-ethereum? i guess i should pin all its dependencies to mimic the exact same versions as per the go.mod file, right? <maximed>So I don't think it can be used in ‘official’ (gnu packages ...) <attila_lendvai>reproducible builds are crucial for such tools, so i thin the answer is yet <dongcarl>attila_lendvai: Sorry, I mean I don't have experience packaging go packages in Guix <attila_lendvai>yep, got it. i think i'll go with the exact versions, and drop a mail to guix-devel once i have something that is... worthy of a mail <sneek>nckx, MysteriousSilver says: hello! you earlier mentioned that you generate guix file separately and grub.cfg sources it, how do i do it? <nckx>MysteriousSilver: The opposite, really, and nothing fancy or complex: Guix was in charge of updating the GRUB configuration file as usual, but did not actually install the GRUB bootloader: (installer #~(const #t)) The GRUB that booted machine was one I installed once manually & uses its own grub.cfg that simply ran the equivalent of 'configfile (path,to)/guix/grub.cfg'. Past tense because I finally flashed Coreboot this week and no longer use that set-up. *nckx waves sleepily as if awakened from a dream. <ngz>Hello. I encounter a strange error: ice-9/eval.scm:293:34: In procedure abi-check: #<record-type <bootloader-configuration>>: record ABI mismatch; recompilation needed <ngz>I tried to recompile, to no avail. <ngz>(starting again from the bootstrap dance, after a make clean && make clean-go) <ngz>Do you know what I'm doing wrong? <attila_lendvai>how should i test a function call? i want to invoke (go-module-recursive-import "github.com/ethereum/go-ethereum" #:version "v1.10.8"); i opened a geiser repl, then typed ,use (guix import go), and called that function, but then guile started to load stuff painfully slowly and i killed it after several minutes. i guess it was in interpreted mode? <attila_lendvai>i'm happy for pointers to manuals, but both the manual, and the cookbook falls short right at this point <attila_lendvai>ok, i realised that the ,use was superfluous, and ,m was enough. but my emacs is at 100% CPU, and probably the crulpit of the slowdown *raghavgururajan rebases wip-gnome with master <attila_lendvai>i ran the same in guix repl, and it reached the error in reasonable time. something happens differently when i invoke it through geiser <nckx>sss1: Not really. Does "guix build --log-file /gnu/store/cvv6df1qfyjkgpapxn5256p2adwjxrpa-compute-guix-derivation.drv" result in anything useful? <nckx>Yeah, the name looked familiar, in a "that looks like it's looked familiar *before*" kind of way. <nckx>Whoever submits the fourth libx86emu patch buys a round of drinks. <nckx>But picking the best of all 3 should result in a very good package ☺ *raghavgururajan makes a cocktail <attila_lendvai>dongcarl, why is it that in a guix repl ,bindings only shows 5 variables, but in the .scm file i see several more? i want to call a function interactively that appears to be not bound in the guix repl context <sss1>nckx: guix build --log-file /gnu/store/cvv6df1qfyjkgpapxn5256p2adwjxrpa-compute-guix-derivation.drv - empty <maximed>attila_lendvai: Are these ‘several more’ variables exported? <maximed>If not, maybe guile optimised them out *attila_lendvai mimics the set! trick he saw for another definition to avoid inlining <maximed>sneek: later tell iskarian: About the git-fetch patches: I've written those let& and let*& macros and let them replace the 'let' and 'let*' macros and it seems to work (with some changes). I should be able to submit them sometime this week <nckx>sss1: Just to simplify, can you retry with --no-offload? <sss1>it works without offloading <nckx>Damn, I was going to blame your unsupported 3rd-party chanchans but they're probably not the problem then. <sss1>i think i had similar problem in past, but unfortunately do not remember how to debug/fix it <sss1>i have done guix pull ; guix package -u on builders recently, probably related ... <nckx>Even if the problem is caused by an unsupported channel (which seems likely), the lack of an error message is squarely on Guix. <sss1>no, problem not caused by third-party channel <sss1>for test sake i can disable third-party channels <nckx>If you run 'guix build ...drv' on the .2 node itself, does it produce more useful output? <attila_lendvai>maximed, are you the one working on the go importer? if so, i have identified an issue i don't know how to fix: (fetch-module-meta-data "github.com/prometheus/tsdb") returns #f because the repo was moved and http-fetch follows a redirect. because of that the string-prefix? call doesn't match it anymore at the end of fetch-module-meta-data <maximed>attila_lendvai: No, I'm not working on the go importer <maximed>I have some things in mind for the minetest importer though (turning it into an updater) <maximed>* things that require (guix upstream) to support git-reference <maximed>atittla_lendvai: Maybe Maxim Cournoyer? <maximed>Possibly http-fetch could be extended to optionally return the URL it redirected to <attila_lendvai>maximed, yep, it's him. i wish people used the same name on irc as in their email, and preferrably IRL, too if it's not dangerous... <attila_lendvai>i think it would be enough to fix this specific issue. i'm a seasoned CL'er, but i'm rather new to scheme. <attila_lendvai>shall i really import every random go library as a guix package? with specific versions demanded by the go.mod of my target package? doesn't sound like something that scales <attila_lendvai>nixos does some vendoring. i think it means that it captures all the sources of every dependency into an archive <sss1>nckx: calling gc was mistake, downloading ..... <civodul>maximed: supporting git-reference in (guix upstream) would be great! <maximed>civodul: The code for supporting that in (guix upstream) and (guix import minetest) already works, I tested it a little <maximed>(Incrementing the revision isn't done yet though) <maximed>it might need some ‘finishing touches’ though <maximed>A problem was packages defined like (let ((commit "...")) (package (name "stuff") (source (origin ... use 'commit'')) ...) <maximed>It's not getting better sitting in my worktree, I'll just send it <maximed>The approach might be a bit dubious though, so maybe I'll use [RFC PATCH] instead of [PATCH] <dstolfa>civodul: did you find out what tool they will use to run the conference? <civodul>i just noticed that attendance is not gratis, bummer <civodul>dstolfa: nope, i told them about my expectations but didn't hear back :-/ <civodul>i hope i'm right in being confident... <dstolfa>my worry is that it's just going to end up being zoom :/ <maximed>civodul: I just sent it to guix-patches@gnu.org and added you to X-Debbugs-CC <vagrantc>wow, i read so many incorrect things just in the name ... packaging con, as in "pros and cons" ... and also "packaging is a con" <pkill9>mpv -> ffmpeg -> sdl2 -> fcitx -> extra-cmake-modules -> qtbase -> mariadb <pkill9>but it tries to build that on my pi :/ not sure if it fails <civodul>dstolfa: i want to believe some of the people involved care (and if it's Zoom, i won't attend) <civodul>but it's better if it's clearly stated upfront, no doubt <dstolfa>i certainly hope it won't be zoom... <danialbehzadi[m]>I sent bugs 50088 and 50089 in a couple of weeks ago. But no reply since then. How can I ping the authors? <maximed>danialbehzadi: Those two patches have been applied two weeks ago by Efraim Flashner <sss1>nckx: looks like it working after gc ..... <maximed>You should have received a mail (one by Efraim Flasher, one by ‘the debbugs bug tracking system’) <leoprikler>sometimes they are on my mailserver when the sender comes from a .xyz domain for example <sss1>does guix already support separate boot partition ? <sss1>and/or boot from luks2 encrypted partitions <attila_lendvai>sss1, yep. in my current setup /boot is on the root, and it's luks2 (or at least i didn't specify --luks1). the password is asked twice, by grub, and then by initrd