<Blackbeard>sirgazil: anyway, what's the problem with the file opening on thunar <sirgazil>Blackbeard: That they don't open in the specified application, I just get a message saying "No se pudo abrir el archivo XYZ. Falló al ejecutar el proceso hijo «gio-launch-desktop» (No existe el fichero o el directorio)" <sirgazil>So, if I double click on a file or press Enter while it is selected, I get that message. <sirgazil>The same happens when using Nautilus with sway. <Blackbeard>How does the xdg-open command know which application to use to open a file? - Unix & Linux Stack Exchange <Blackbeard>What happens if you use the command line with the same application? <Blackbeard>mime type files might be giving you the trouble too <sirgazil>No. In the case of XFCE I added (service xfce-desktop-service-type) to my OS config, and I would expect that operation to work without me having to configure anything. <sirgazil>Blackbeard: $ evince some.pdf works fine. <Blackbeard>Did you check mimeapps.list in your home directory? <Blackbeard>Open a directory in the default file manager and select a file - Unix & Linux Stack Exchange <Blackbeard>If this is happening across different desktops then I am inclined to think it is a problem with that <Blackbeard>Specially because with the command line it works <sirgazil>I've never used xdg-open directly because I normally use GNOME and its applications to specify default applications to open files. <sirgazil>So I don't have xdg-open nor xdg-mime in my profile. <sirgazil>I installed xdg-utils when in sway to see if that fixed anything, but didn't. <sirgazil>But it's odd that this doesn't happen when using Nautilus in GNOME. <sirgazil>Just a sec, I'm installing xdg-utils again. <Blackbeard>I am intrigued because you mention this is across DEs <Blackbeard>It is just my workflow is so completely different. I basically use emacs as my os <Blackbeard>I open libreoffice and from there I open the filosofía <sirgazil>Emacs alone is great for people working with text only. <Blackbeard>Anyway yes, so I would never even realize there is such a bug. But I can try to help <bdju>emacs can display images, even over tramp with remote images <sirgazil>Do you do 2D and 3D design in Emacs too :) <sirgazil>$ xdg-mime query filetype example.pdf does not return anything <Blackbeard>I think there is a problem with your mimeapps.list <sirgazil>But right clicking a file in Thunar lists as default application the application I used in GNOME as default, so where is it finding that information? <sirgazil>An funny thing. In XFCE, system sounds are not played either (same happens in sway). <sirgazil>Blackbeard: Yes, I added that to my config. <nly> (uri (git-reference <nly> (url "mirror://local/emacs-shroud.git") <nly>in guix/download.scm: <nly>fatal: unable to find remote helper for 'mirror' <nly>well, (method url-fetch) works <nly>with tarballs, not git ofc <nly>i don't think i want to substitute urls for all pkgs with "mirror://..." forms <nly>and wanted to use (method git-fetch) really, and possibly "file:///..." urls in %mirrors ***jonsger1 is now known as jonsger
<sirgazil>Blackbeard: find /gnu -name "mimeapps.list" takes forever. I'll let it run and tell you later about the results. <sirgazil>By the way. I installed MATE and I get the same behavior. <Blackbeard>sirgazil: yeah I figured it is related to something else <sirgazil>Blackbeard: That these DEs are not providing a default mimeapps.list? <sirgazil>For example, no $XDG_DATA_DIRS/applications/$desktop-mimeapps.list? <Blackbeard>sirgazil: of course you can copy paste one from the internet <Blackbeard>Change one or two file types to the ones you have installed <apteryx>bandali: I eventually figured out my mistake with the pdmp error, so that's one thing solved. As for the patch... Let me check :-) <bandali>apteryx, ah :-) so it wasn't due to my original definition? (i haven't pulled in recent guix in a week or two btw) <Gooberpatrol66>I am trying to set up git with libsecret. Archwiki says to use git config --global credential.helper /usr/lib/git-core/git-credential-libsecret <Gooberpatrol66>the /usr directory doesn't exist though. where do i find that program? <apteryx>bandali: no, your definition was alright! <guix-vits>Gooberpatrol66: `find /run/current-system/profile -name *git* 2>/dev/null` <sneek>Welcome back Gooberpatrol66, you have 1 message. <sneek>Gooberpatrol66, guix-vits says: `find /run/current-system/profile -name *git* 2>/dev/null` <Gooberpatrol66>that command doesn't return anything. i'm currently running a find on / <apteryx>is it possible to have Geiser work well over Tramp? <guix-vits>bandali: is there a way to scan packages with `guix` to search for a file? <bandali>guix-vits, hmm, i'm not quite sure; i don't think so. there's ongoing work improving `guix search'; i wonder if this may be in scope for that (if it isn't already there) <guix-vits>Gooberpatrol66: "Use git-credential-gnome-keyring instead of this." <Gooberpatrol66>i'm trying to exclude it but i don't understand the find command syntax <guix-vits>Gooberpatrol66: try to search in /run/current-system/profile or /var/guix/profiles/per-user <guix-vits>Gooberpatrol66: in meantime you can just download the script git-credential-libsecret. <Gooberpatrol66>it seems like you might not need to put the full path to the command <sneek>Gooberpatrol66, you have 1 message. <Gooberpatrol66>git: 'credential-gnome-keyring' is not a git command. See 'git --help'. <guix-vits>"GNOME keyring support is now included in git, so this project is deprecated." <guix-vits>So, git-credential-libsecret is deprecated in favor of git-credential-gnomekeyring, which is deprecated due to above. <guix-vits>Gooberpatrol66: i'll invite you to join me in lookup of git's man pages. <guix-vits>"1. Find a helper , 2. Read its description, 3. Tell Git to use it" <atw`>I've got a rudimentary synapse package for guix! <atw`>I will make a WIP patch series tomorrow <noeu>Can I ask a question about using Guix here? <noeu>I'm sorry if you make the wrong place. <noeu>I wanted to create a channel for Common Lisp, including cl-sdl2 etc., and I understood how to create a simple channel. <noeu>However, since the channel is a GIT repository, I can't test if the channel is written correctly without pushing it once. <noeu>Could you tell me if there is a way to test the behavior of the channel locally? <noeu>(English is not my native language; please excuse typing errors.) <bdju>Yes, it can be local. You put a file path in place of the url, but you still "push" and such. <noeu>Thank you for your response. <noeu>I write a test channel repository in channels.scm as follows, <noeu>If you change the URL to "~/path/ to/ cl-guix/packages/cl-packages.scm" and do guix pull, an error will occur. <noeu>Updating channel 'cl-guix' from Git repository at '~/Projects/Hobby/LispProjects/guix-package/my-private-channel/gl-guix/packages/cl-packages.scm'... <noeu>guix pull: error: Git error: unsupported URL protocol <noeu>I get an error like above <noeu>I'm sorry I misunderstood. When testing locally, is it not possible to read the package definition file in scm format directly, but is it necessary to create and read a GIT bare repository? <noeu>I will try different things, thank you. <noeu>bdju: Thank you. I'm using "file: //", but I can't read it because it's doing something weird, so check a little more. <noeu>I'm sorry, I mistakenly realized that the path I specified should be where the GIT repository is (where the .git directory is). <noeu>I was able to load with guix pull by specifying the directory up one level as the path. ***apteryx is now known as Guest34434
***apteryx_ is now known as apteryx
<bdju>Does anyone know how I can find out if the Atheros QCNFA435 will work with linux-libre? <bdju>No, I was thinking of buying it. <bdju>I'm running NixOS on this machine right now because I didn't want broken wireless on Guix System. I had Guix System on my previous laptop and want to use it again soon. <bdju>I had replaced the WLAN card in the old laptop to work with Guix System, however the new one uses an M.2 slot instead of mPCI-e, so I can't use the same card again. <bdju>Currently in this newer machine I have an Intel 7260, which I am assuming does not work. <bdju>I will probably only buy it if it will work out of the box... I don't want to pull in non-free stuff <bdju>I see ath10k-firmware mentioned. I have never really looked into drivers like this before. <lfam>bdju: h-node says it doesn't work with free software <bdju>Ah, okay... I checked there but couldn't find it. Thanks. <lfam>The mainline kernel includes nonfree drivers that linux-libre removes <lfam>I would look for something that works with ath9k drivers <bdju>Yeah, just wasn't clear to me if this was a non-free driver or not. It's very annoying to find info on this stuff. <bdju>If the 9 vs 10 is an age thing I might not have a lot of options since I need something for an M.2 slot, which is not all that old. <lfam>There are some options but not a big variety <lfam>And the performance is starting to fall behind <bdju>Searching this stuff is not working well for me... Anyone able to tell if Atheros QCNFA222 would be any better? I get linked to that wiki when searching also, but then can't find it anywhere so not sure why it matched a search result <bdju>I'll be back in a bit, gonna go shower and eat breakfast <guix-vits>2 of them is, actually, "mobile broadband"... <bdju>guix-vits: Thanks! Looks a bit pricy, but I see cheaper things with the same chipset on eBay. Just gotta make sure they're still m.2. <guix-vits>bdju: read the store notes (about key-form and... <bdju>Yeah, saw that also... Not sure which slot the T440p uses. I will have to look into that. <guix-vits>that some "defective-by-design" manufacturers make their machines rejects some free devices. It recommended to use an USB-thingy for some. <bdju>and I am aware I likely have to deal with a whitelist. I can flash coreboot to get around that. <guix-vits>bad that pine wiki sometimes not accesible via tor... (cloudflare). <guix-vits>manoj-dark, better for my case, in meantime. *guix-vits notices a noldor o/ <DamienCassou>I would like to have 2 couchdb develpoment servers running in my Guix OS, on 2 different ports. Can I configure that within my config.scm? <DamienCassou>there is no CouchDB package in Guix. So I would have to start with that. But my question remains: when I'm done creating a package and a service for CouchDB, how do I configure VMs in my config.scm file? <apapsch>Damien Cassou: you create a different operating-system file. you can even define it in a guile module and use (inherit) like in packages <DamienCassou>guix-vits: I don't understand the link between my question and mcron <DamienCassou>apapsch: yes, I will create an additional config.scm for both of my CouchDB servers. How do I modify my main config.scm so the 2 other VMs are handled as services? <apapsch>Damien Cassou: I would try to create some shepherd-service with one-shot? #t <DamienCassou>ok, so there is no support for running Guix VM builtin. Thank you <DamienCassou>apparently, I won't be able to pursue this route: the sources for the CouchDB I need doesn't seem to be available anymore, it's a very old version. So I guess I will have to go the docker route <DamienCassou>I'm trying to package couchdb and there will be a lot of trial and failure. How should I try my package definition? <DamienCassou>I don't want to git commit, guix pull and guix install a hundred times :-) <nckx>DamienCassou: pre-inst-env guix build. <nckx>(Read (guix)Building from Git if you haven't yet done so.) *nckx already replied to ya bug. <nckx>‘don't listen to them, they don't fave a fedora workstation :p’ — true fact. I won't be of much help. <apapsch>guix-vits: that's what software heritage is mirroring :-) <DamienCassou>`configure.ac:89: error: possibly undefined macro: GUILE_MODULE_AVAILABLE` <raghavgururajan>DamienCassou: IIRC, there come packages that you need pass as '--ad-hoc'. They are mentioned in manual. <nckx>DamienCassou: That should just work. I just ran ‘guix environment --pure guix -- ./bootstrap’ on Guix System with great success. Is your ‘guix’ command up to date? <nckx>I don't understand. Those arguments should not not be positional (and a quick ‘env’ test seems to agree). <nckx>DamienCassou: <When running `./boostrap` in `guix environment --pure`> What was the exact command you meant by that? <nckx>‘guix environment --pure guix’ from the manual or just ‘guix environment --pure’? <nckx>The latter will create an empty environment so nothing working is to be expected. <nckx>I wonder if Guix should warn about creating an empty environment. Not refuse, just warn. <DamienCassou>when can I start running `./pre-inst-env` to build my own package definition in the guix repo? `guix environment --pure guix -- make` is now running tests <nckx>DamienCassou: When that's done. <nckx>You can probably run it now but it will run much slower. <nckx>Which is why you should run ‘guix environment --pure guix -- make -j`nproc`’ whenever you've modified a lot of .scm files and the ‘note: source file … newer than compiled …’ warnings start becoming numerous. <nckx>That will recompile the .scm files to .go files and put the fast back in. <DamienCassou>`guix environment --pure guix -- make -j4` finishes happily. But `guix environment --pure guix -- make check` doesn't. It complains about failing tests <DamienCassou>I've tried again and tests/lint.scm and tests/pack.scm` keep failing <NieDzejkob>DamienCassou: sounds like a bug, could you sent an excerpt of the output and a short note to bug-guix@gnu.org? <nckx>DamienCassou: There's no need to ‘make check’, just ‘make’. <NieDzejkob>you need to be in the guix environment to use pre-inst-env <nckx>My fedora box has ‘dnf install libgcrypt-devel’ in its bash history. <nckx>NieDzejkob: That would be a bug <NieDzejkob>nckx: oh, I didn't realize. I had to do that since the guile3 migration IIRC <NieDzejkob>might be related to the fact that I replaced guile2.2 with guile3 in my operating-system's packaged <nckx>Sure, there was probably no shortage of bugs/unforseen gotchas during that transition. <nckx>Yeah, something like that. <nckx>But the ‘env’ in ‘pre-inst-env’ is just that. <nckx>You can double-env all you want but understand that it's a hint that something ain't right. <DamienCassou>I've just install libgcrypt-devel but this doesn't change anything to the result of `pre-inst-env`. Do I have to try again the `./configure` and make in the pure environment? <nckx>Don't bother running ‘make check’ this time. It would be very nice of you to report the bugs later, but it's not required to use ./pre-inst-env. <DamienCassou>nckx: I did the bootstrap, configure and make again in the pure env. Still `./pre-inst-env guix build couchdb` complains `no code for module (gcrypt hash)` <nckx>I can't really help you, I got Guix ‘running’ on Fedora without having a clue what I was doing. I never got that error. <DamienCassou>nckx: ok! I've sent a new bug report. I hope the community won't get upset about me. I'm just trying very hard to use Guix as my main OS <nckx>Why'd we get upset? Bugs are bugs & if it's frustrating not being able to help, that's not your fault. 🙂 <DamienCassou>ok :-). I really hope I will be able to install Guix System on my upcoming computer <DamienCassou>I'm a bit worried for now because there are many things that require quite some work (such as packaging couchdb) and I face many issues preventing me from making much preparation <DamienCassou>are there people using Guix System as their main OS for their day-to-day job? <nckx>Plenty, including myself. <nckx>Source: myself. I don't think we have any ‘data’ on that. <nckx>I have not made it sufficiently clear how clueless I am about Fedora. It just came with a VM. *janneke made some progress on hurd (by using an old workaround by phant0mas...) <nckx>My main OS before 2016 was NixOS, before that Exherbo, go back until you hit Slackware. <nckx>janneke: Yay. Seems like Hurd support is recently hip again. <DamienCassou>thank you nckx. Someone answered on my bug regarding ./pre-inst-env. It seems this one has to be run under a pure environment too. It seems to work fine now <janneke>nckx: Yes! Well, we hit 1.0 quite a while ago, so it's about time :-) <nckx>DamienCassou: Damn it. That's not the case on Guix System. Sorry for steering you wrong. (NieDzejkob ☝) <nckx>Nor on my Fedora system, but as we've established that's just a black box I paste dnf command in so 🤷 <kirisime>Guix now supports authentificated channels, but I'm not sure how to use one since you can't use user@host:repo as a URL. <DamienCassou>building mozjs (in gnuzilla.scm) failed. Isn't there a build server checking that all packages can be built? <NieDzejkob>DamienCassou: do you have substitutes enabled, or are you building everything from source? also, what architecture? <DamienCassou>I haven't enabled any substitute but I hope the official one is enabled. I'm on Fedora 31, amd64 <nckx>DamienCassou: ‘guix describe’ is a command. <DamienCassou>`guix describe` will tell you that I'm using file://...guix-git-repo. The commit is just 1 above current `master` branch as I'm trying to add a package for couchdb <DamienCassou>I indeed used the guix install script to install on Fedora. But, since then, I have cloned the git repo and added a ~/.config/guix/channels to point to that <nckx>DamienCassou: Does your /etc/guix/acl contain 8D156F295D24B0D9A86FA5741A840FF2D24F60F7B6C4134814AD55625971B394? <DamienCassou>nckx: indeed, the file contains some scheme and this string appears in the middle <nckx>DamienCassou: <Isn't there a build server checking that all packages can be built?> Not really. The build server (ci.guix.gnu.org) tries to build all packages. That's it. <nckx>There are no negative answers. If your guix doesn't find a successfull build there it will try building locally. <DamienCassou>nckx: I don't see the difference between what you say ci.guix.gnu.org does and what I asked <nckx>Your question ‘Isn't there a…?’ implies that it would somehow affect your build failure. It won't. <Blackbeard>Damien Cassou: how did you handle selinux and guix <DamienCassou>I thought that building packages couldn't fail because a build server was regularly checking that all packages build <nckx>DamienCassou: There is such a server, but there's an unstated assumption (A because B) that it prevents packages from breaking that I don't get. <nckx>mozjs (/gnu/store/rsgg6kif5ii7xf2hlhmpsa8pd7yxh2ih-mozjs-60.2.3-2) builds fine on currentish master. It's also cached on ci.guix.gnu.org, so if your substitutes were working Guix should just download it for you. <nckx>Currentish being f76c16d220e6c349441c08bf25a5197037490fa5. <DamienCassou>you are talking about mozjs60, I'm talking about mozjs, which seems to be version 17. I also tried mozjs-24, without more success <nckx>I'm talking about ‘guix build mozjs’, yes. <DamienCassou>Blackbeard: what do you mean by start guix? I'm on Fedora and using Fedora packages <nckx>It starts to build so it's prolly gonna fail. <nckx>It fails with ‘error: ISO C++ forbids comparison between pointer and integer [-fpermissive]’ That's something a human needs to fix, either by applying a patch, or (probably) just by removing what is apparently an unused and broken old package. <DamienCassou>what about an option to gcc? Or a different version of gcc? <nckx>Then I fear you've been appointed volunteer to fix it. <Blackbeard>Damien Cassou: oh I thought you had Guix on top of fedora :) <DamienCassou>I indeed installed guix on my Fedora to prepare myself for Guix System <nckx>DamienCassou: Adding -fpermissive as CFLAGS would be the first thing to try. <nckx>mozjs uses the gnu-build-system so #:configure-flags (list "CFLAGS=-fpermissive") might just work if you're lucky. <Blackbeard>I ask because I haven't figured how to start the Guix daemon on fedora <Blackbeard>And I found the only way that works is to disable SELinux entirely <Blackbeard>Damien Cassou: if you didn't disable SELinux then Guix won't work on fedora <raghavgururajan>DamienCassou: Could you try your bootstrap steps under `guix environment guix --pure --ad-hoc help2man git strace` <DamienCassou>raghavgururajan: you mean `guix environment guix --pure --ad-hoc help2man git strace -- ./bootstrap` ? <roptat>For selinux, I think we have a cil file in etc in the repo <roptat>Not sure how to use it, but it's supposed to help <NieDzejkob>DamienCassou: you do realize it's possible to start a shell in the environment by just leaving off the -- command part, right? <Blackbeard>I use fedora in my laptop because it has icecat in the repos and I need proprietary drivers :c <DamienCassou><NieDzejkob "Damien Cassou: you do realize it"> I do but I find it more convenient to always do everything in one command. This gives me confidence that I'm ding the right thing. Also, I stay in eshell this way :-) <nly>I made guix use local git repository for 1 package, now for some more packages <nly>I guess it's not the best idea to make changes in "guix/build/*" where it causes a full rebuild i think <atw`>sent my patches but forgot to guix lint, whoops <efraim>lfam: by not skipping the builds of rust packages we can check that the inputs are correct and see that they actually build correctly when placed together. I actually ran into an issue while doing the last 3 patches or so of alacritty <lfam>I figured it was like that. In that case, why do so many of the crates packages skip the build? <efraim>Historical Reasons™ Really because we needed to add in so many crates to be able to actually build and test them all that we didn't actually build most of them for a while <lfam>The go-build-system has a different mechanism for packages that we don't really need to build. I wonder if it's worth trying to make it a common-build-option <lfam>That's not a priority but it is annoying to see similar use cases addressed differently in different build systems <[df]>have I done something wrong or is the key used to sign binary distro expired? <roptat>Don't know, but have you checked your computer's date? <[df]>pub rsa4096 2014-08-11 [SC] [expired: 2020-01-30] <[df]> 3CE464558A84FDC69DB40CFB090B11993D9AEBB5 <nckx>[df]: Where did you obtain the key? <roptat>Mh, maybe the expiration date was modified since you got it? <[df]>um... can that happen? I'll try deleting the key from my keyring and refetching <nckx>[df]: Try running gpg --refresh-keys, the URL you used serves an expired key. <nckx>Somehow I had an up-to-date copy so it's out there™. <lfam>Might need to go out of your way to pick the right keyserver <lfam>It's surprising to me that it's expired on Savannah. I wonder if something happened on Savannah <nckx>Like restored from an old back-up, you mean? <lfam>That's what I'm thinking... <lfam>Anyways you might want to email civodul <lfam>I know he is supposed to be on vacay <[df]>doesn't that mean it was signed with an old key too though? <lfam>It's the same key. With PGP you are supposed to keep changing the expiration dates... <nckx>[df]: No, that's not how GPG works. You don't change the key, you just sign a new ‘valid thru’ statement. <lfam>I think I got it from <hkps://keyserver.ubuntu.com> <nckx>Is what an attacker would say. <nckx>lfam: Interesting choice. Any particular reason? <lfam>nckx: Before <keys.openpgp.org> was a thing, it was always the most reliable one for me <lfam>I think the default was MIT's server but it almost never worked :/ <nckx>Then there was a gnupg.org(?) one that just went away one day. <lfam>And then while keys.openpgp.org was being created and promoted, many keys were not on there so I still didn't start using it <lfam>The Ubuntu server just worked for me <lfam>Well there is always room for improvement <efraim>the keys on Savannah are manually uploaded, not everyone remembers to upload it there also after uploading it to another keyserver <lfam>Yeah I know, but the policy of making Savannah the "single source of truth" for our keys came from civodul, so I'm surprised he would have forgotten <lfam>We don't require people to add their keys to the keyservers, but do instruct them to add the key to their Savannah page <lfam>We do ask them to use the key servers too, I was mistaken (Commit Access in the manual. Rip HACKING) <DamienCassou>why doesn't `#:phases` aligns itself under `#:parallel-build?` <nckx>lfam, efraim: Either of you know the right way to human-dump a key like that? I did some hacky file renaming, must be a better way. <DamienCassou>each time I touch a package in a file, the indentation gets messed up, making it really hard to prepare patches <lfam>I love the TITLE in HACKING though, never noticed that before :) *lfam makes raspberry sound <nckx>What I did was <mv keyring away, kill the agent to be sure> && curl | gpg --import && gpg --list-keys ludo. <nckx>That worked but can't be right. <lfam>It won't import the key in that case <lfam>Or maybe with --list-packets <nckx>lfam: No, you were right, ‘gpg’ is ‘show’. <lfam>Does it show the expiration dates though? <lfam>The expiration dates that we don't need for our use case <nckx>sa4096/0x090B11993D9AEBB5 2014-08-11 [SC] [expired: 2020-01-30] <nckx>If that doesn't work for you, it's probably due to one of the lines in my 20-line gpfoo.conf <nckx>‘gpg’ means ‘go ahead and type your message’ <nckx>then it's suddenly show me the key. <nckx>When I say it's a mess I say so as a loving user but god is it a mess. <lfam>It's late now but I'd like to see a GSoC for Git regarding code-signing <atw`>"...but god is it a mess." there needs to be a word for that. I feel the same way about org <lfam>Well we all love our own messes. That's why my process for "installing" a new laptop is copying the hard drive from the old one with "my Debian" <lfam>But it's not good when you have to share the mess <nckx>I've sent Ludo' a mail by the way, I just forgot to CC a list because today is that kind of day. <rekado_>we ran out of space on ci.guix.gnu.org again. <rekado_>it’s probably just a whole bunch of disk images at 2G each. <kirisime>Is there a way to set directories with certain permissions in the config.scm file? <noobly>"/usr/bin/env: command not found" after running an npx command (`npx create-react-app my-app`).. any tips? <janneke>kirisime: you can probably add a computed-file with mkdir as special-file or something <lfam>janneke, noobly: We have the special-files-service-type for adding things like /usr/bin/env <lfam>Actually, the extra-special-file is more appropriate. See the manual section Base Services <janneke>lfam: yeah, kirisime asked about adding a directory with specific permissions <lfam>Sorry janneke, I thought you were replying to noobly! <janneke>lfam: np, related issues -- the extra-special-file is probably helpful there too :) <lfam>NieDzejkob: I've noticed that its log files are often truncated :/ <lfam>NieDzejkob: Can you check if there is already a bug report and make a report if not? I'm building audacity locally now <NieDzejkob>lfam: I feel like you'll have more info to provide after you finish your local build of audacity <lfam>NieDzejkob: I mean, a bug report about the truncated logs from Cuirass <noobly>lfam: thanks, I'll check that out <lfam>NieDzejkob: Thanks. I think we should send a followup message to say "It's still happening" <noobly>lfam: first time editing my config.scm, so within the 'services' field I previously had only '%desktop', I've since added `(extra-special files "/usr/bin/env" (file-.... "../env"))`. Now it is time to run `guix system build config.scm`? <lfam>NieDzejkob: Audacity built for me on x86_64. I think the failure was probably caused by ci.guix.gnu.org running out of space, as mentioned by rekado above <lfam>Hey rekado_, do you know if builds that failed on ci.guix.gnu.org due to lack of space will be restarted? <lfam>I need to make a hidden-package or maybe just an origin that downloads 5 files so that they can be used in another package's test suite. Do we have any examples of this? Multiple files being downloaded in one origin? <lfam>Unfortunately the test files are not available as a single file archive <lfam>Or, do I have to make 5 different origins? <lfam>I guess I can do like the "drops" in the icedtea packages <noobly>I'm still trying to get npx to behave normally, currently it returns '/usr/bin/env: command not found'. From what I understand /usr/bin/env does not exist, so I need add the 'special file' /usr/bin/env, but how do I determine what it's 'target' should be? The info manual gives this code: (("/bin/sh" ,(file-append BASH "/bin/sh")) ("/usr/bin/env" ,(file-append COREUTILS "/bin/env"))) <noobly>should the above code just go in the 'services' field in config.scm? my first attempt failed <lfam>noobly: Yes, you have the right idea <lfam>You should add something like (extra-special-file "/usr/bin/env" (file-append coreutils "/bin/env")) to you services <lfam>I copied that from the manual section Base Services <noobly>well, my first attempt was adding `(extra-special-file "/usr/bin/env" (file-append coreutils "/bin/env")) after (services %desktop-services .. but no luck. <lfam>It should be like (services (cons* (extra-special-file ...) %desktop-services)) <lfam>I can tell that your first attempt did not use the correct syntax <lfam>You can to explicity add the extra-special-file service to desktop-services. This is what cons* does <lfam>cons* takes two arguments, both of which are lists. It adds them together to make a new list <lfam>This "cons*" stuff is sort of obscure old Lisp jargon. <lfam>You can also use (services (append (list (extra-special-file ...) %desktop-services))) which is less jargon-y <lfam>If you load the manual as one web page and search for "services (append" you'll find some examples <noobly>lfam: ah, thanks. i was wrongly assuming services did this on it's own. setting up geiser right now so I can read the docstring next time :^) <lfam>Sorry if my advice is a little unfocused. I'm quite far from an expert on Scheme <noobly>the advice was great, thanks for the help <lfam>Oof, some of the test data for this package I am working on is hundreds of megabytes