IRC channel logs

2024-04-26.log

back to list of logs

<mekeor>hello. which package provides the info manual for "libstdc++"?
<futurile>podiki, ieure: it seems like you're both working on nss CVE's - I think Christina O'Donnel who's testing core-updates is also going to land-up with a patch for it - see #4036
<peanuts>"23.1; I can not understand a sentence." https://issues.guix.gnu.org/4036
<futurile>sorry #40316
<peanuts>"nss not reproducible" https://issues.guix.gnu.org/40316
<mekeor>mekeor: libstdc++-doc
<mekeor>ACTION is a guix-system-user who types "sudo rfkill block bluetooth" in the shell multiple times every day
<mekeor>how do i disable bluetooth for good, in operating-system definition?
<wakyct>mekeor I think there's an auto-enable option in the bluetooth service?
<wakyct>though the signature for that seems to imply it's false by default so I don't know
<mekeor>wakyct: hm, also i don't have the bluetooth-service at all
<wakyct>now that you bring it up I guess I never tested my BT before I configured the service. Maybe xfce sets it up as part of the desktop service?
<wakyct>perhaps you have it by default
<wakyct>for whatever desktop you use
<mekeor>i use plain xorg (started as user) (with xmonad window manager)
<mekeor>btw, libstdc++-doc is for gcc version 9... and also it does not include info manuals
<podiki>futurile: I think you should direct that to apteryx (I'm not currently doing anything with the nss updates/grafts)
<the_tubular> https://lists.gnu.org/archive/html/guix-patches/2024-04/msg01595.html
<the_tubular>Why is this a problem ?
<peanuts>"[bug#70569] [PATCH 3/3] system: Warn when multiple nss-certs packages ar" https://lists.gnu.org/archive/html/guix-patches/2024-04/msg01595.html
<coyote>the_tubular: because if you have `nss-certs' specified in your operating-system's packages it'll not build
<coyote>spits out an error about having two conflicting packages in the same profile, since now %base-packages includes it
<the_tubular>But can't you have many times the same program installed with guix ?
<coyote>well yes, but these are two versions of the same package, in the same profile
<coyote>in different profiles it'd work, i assume.
<the_tubular>I see
<jakef>dumb question: why is the version in %base-packages different?
<Kolev>Efraim: Can you help me with getting audio to work?
<Kolev>Is there a system service for Plymouth?
<sham1>This would probably be more appropriate for #guile, but is there a way for me to spawn a background process in Guile?
<KE0VVT>I did `guix install git` but had to run `git` directly from `/gnu/store`. O_O
<sham1>Is your profile correct
<KE0VVT>sham1, this si a fresh install of Guix System.
<KE0VVT>sham1: Weird! `git` is in `.guix-profile` but not in `PATH`.
<sham1>Huh
<sham1>What does your PATH shiw
<KE0VVT>sham1 /gnu/store/3dzz178civ4hjx041qd61iiy558x0rra-glib-2.72.3-bin/bin:/run/setuid-programs:/run/current-system/profile/bin:/run/current-system/profile/sbin
<sham1>Yeah, you don't have your .guix-profile set up properly in your path
<KE0VVT>That's weird. This is a freshly-installed system.
<sham1>Try logging out and logging back in
<KE0VVT>I'll do that, once `guix pull` finishes. :)
<adanska>reposting this problem in hopes that anyone might be able to help: https://paste.debian.net/1314906/
<peanuts>"debian Pastezone" https://paste.debian.net/1314906
<adanska>I have no idea what this `union-build: collision between file and directories` error means. if anyone would be able to give me some input id be really grateful :)
<KE0VVT>adanska: Did you posst to the mailing list?
<adanska>no, if i dont get any response i might do that later.
<KE0VVT>sham1, how do you suggest I fix my PATH? Just log in again?
<adanska>KE0VVT, you just need to have your shell restart. relogging works because it relaunches the shell and sources all the env variables it needs
<KE0VVT>adanska, no, I've tried opening another terminal window.
<adanska>if you run `bash` does the shell work?
<KE0VVT>adanska: If I run bash, I still get the weird PATH.
<adanska>hmm..
<KE0VVT>The path is /gnu/store/3dzz178civ4hjx041qd61iiy558x0rra-glib-2.72.3-bin/bin:/run/setuid-programs:/run/current-system/profile/bin:/run/current-system/profile/sbin
<adanska>i mean that does seem right, its got the current-system profile
<adanska>are you in the process of installing guix?
<KE0VVT>I just installed Guix System.
<adanska>whats not working?
<KE0VVT>Programs I install into my profile don't go into my PATH.
<adanska>ahhh yes sorry .guix-profile isnt present in your path
<adanska>KE0VVT, try running this `GUIX_PROFILE="$HOME/.guix-profile" ; \
<adanska>source "$GUIX_PROFILE/etc/profile"`
<KE0VVT>adanska: THat worked! :) Thank you!
<adanska>awesome!! glad it worked for you :)
<adanska>I found out the issue!!! i had the binaries built as part of the project directory, and that was messing up guix!!
<adanska>is there a way to use `guix pack` with a package.scm file?
<KE0VVT>adanska HOORAY!
<adanska>going offline for a bit, so if anyone knows, sneek me!
<adanska>sneek: botsnack
<sneek>:)
<sneek>adanska: wb!
<KE0VVT>guix: guix: command not found
<KE0VVT>Oh! guix time-machine ... -- system reconfigure ...
<Franciman>does guix home allow to use different profiles?
<sham1>Franciman: Guix home and profiles are orthogonal
<ulfvonbelow>anyone else notice that many python packages these days will fail tests because of lacking certificates?
<ulfvonbelow>it's not even that they're missing any particular certificate, just that the file they expect to exist doesn't exist
<Franciman>sham1: yes, why?
<fnat>Good morning! https://packages.guix.gnu.org seems to timeout (504), in case someone who has access to it wants to give it a nudge.
<futurile>fnat: morning fnat - yeah it's been like that a couple of days - I think various people are away or at conferences
<futurile>fnat: there's the one that's associated with the science channel - I forget the url
<fnat>futurile: Hey, thanks, that's alright, I just thought of reporting it but not urgent on my side.
<futurile>fnat: ack - I've been using the toys one - it's really good as it finds packages across many channels
<fnat>Ah good point, I should have thought of that one!
<futurile> https://toys.whereis.みんな/ - I do find that name and URL impossible to remember though heh
<fnat>Haha, or even to spell for me! But I suppose that's the nice thing about it. :)
<cbaines>packages.guix.gnu.org is back
<cbaines>I think data.guix.gnu.org was having it's regular database connection pool problem
<fnat>Super, thanks cbaines. 🙏
<KE0VVT>How do I run this? https://paste.debian.net/1315248/
<peanuts>"debian Pastezone" https://paste.debian.net/1315248
<Franciman>is there a way to tell guix search a specific version of the package i want to search?
<nutcase>when I add an mcron job to my operating-system services, it removes already defined mcron jobs (e.g. rottlog). I did it as the manual told, which results in: https://paste.debian.net/1315251/ What am I doing wrong?
<peanuts>"debian Pastezone" https://paste.debian.net/1315251
<Franciman>i remember in older versions of guix, there were examples of using a GNU tool to filter guix search output
<Franciman>but i can't find the examples anymore
<Franciman>okay, solved! I have 126 rust crates i have to add
<Franciman>do i have to create a separate patch for each of them? lol
<Franciman>before being able to add texlab@5.15.0
<cbaines>Franciman, yep, one patch per new package
<Franciman>nifty
<festerdam>Hi, all.
<festerdam>«guix install glibc-locales» fails with «guix substitute: error: rename-file: No space left on device», but I have 8.1 Gs available on the partition / is mounted on, according to df. What could be wrong?
<futurile>festerdam: guix gc has some options to run to see if the DB has issues
<futurile>festerdam: sure the disks not failing and it's been remounted ro?
<festerdam>futurile: You mean / remounted as read-only?
<futurile>festerdam: yes - it should show up in the logs
<festerdam>futurile: Are you saying / should be read-only or that it shouldn't be? Because I don't think / being read-only would make much sense.
<cbaines>festerdam, if you have space, you probably have run out of inodes
<cbaines>check df -i
<festerdam>cbaines: 0 inodes out of 0 used. This is a btrfs file system.
<futurile>0 can't be right - you can't have used 0 inodes surely?
<festerdam>futurile: /dev/sda5 0 0 0 - /
<festerdam>Is the line that df prints for the file system.
<futurile>festerdam: can you write a file to that mount point - touch blah or whatever
<festerdam>futurile: Yes. I was able to create a text document, write some text to it and read it again.
<futurile>festerdam: there's something off with that - it should be telling you the number of inodes and used blocks
<cbaines>maybe it's significant that rename-file is failing here
<cbaines>given that shouldn't be taking up more space, maybe there's a metadata issue
<cbaines>or the filesystems are such that rename-file can't work (e.g. you can't rename a file from one filesystem to another)
<festerdam>futurile: This stackoverflow answer seems to indicate that it's normal for a BTRFS filesystem to report 0 inodes.
<festerdam> https://unix.stackexchange.com/questions/404933/why-is-mounted-nas-showing-0-inodes-on-users-partition
<festerdam>cbaines: I will simply try rebooting to see if that fixes it.
<attila_lendvai>when locally building guix, i get "Error: Could not set character size". it's some error coming from graphwiz... any hints?
<attila_lendvai>this is all i could find online (not very useful): https://gitlab.com/graphviz/graphviz/-/issues/863
<peanuts>"[Dot] The simplest example doesn't work (Error: Could not set character size) (#863) ? Issues ? graphviz / graphviz ? GitLab" https://gitlab.com/graphviz/graphviz/-/issues/863
<festerdam>Nope. Still fails after reboot. Also made a quick disk check from the bios screen, which the disk passed.
<futurile>festerdam: are you on Guix or using Guix on top of another Linux distro?
<futurile>festerdam: if you're on top of another Linux distro I'm wondering if systemd (or whatever) can tell you anything useful
<festerdam>futurile: I'm running on Fedora (I heard from someone that they got it working even after nscd was removed, so I'm trying again)
<futurile>festerdam: ok, is the systemd guix service running correctly?
<futurile>(I basically run a similar set-up - guix on top of Ubuntu in my case)
<festerdam>futurile: I'm running the guix daemon in a sepate shell. (I ran «sudo guix-daemon --build-users-group=guixbuild»)
<festerdam>And it is receiving connections.
<festerdam>I also just tried to do a guix pull but it also prints that error.
<KE0VVT>Is there a way to restrict access to the Guix daemon?
<futurile>festerdam: It has to run as root - so I think the instructions tell you to install the systemd unit
<festerdam>futurile: Doesn't sudo do that?
<festerdam>Run as root I mean
<futurile>festerdam: no I don't think so - the binary install script should have done this for you - it has special requirements about mounting the /gnu/store FS correctly
<festerdam>futurile: Ah, I forgot to say: I didn't install via the install script, but via this copr: https://copr.fedorainfracloud.org/coprs/lantw44/guix/
<peanuts>"lantw44/guix
<futurile>festerdam: OK - I'm at the limit of my understanding - I've only used the binary script - and then used the systemd unit to mount /gnu/store correctly on my btrfs partition
<futurile>festerdam: maybe there will be someone along later who's used the Fedora package - or you could try asking on the Guix mailing list / presumably in the Fedora community :-)
<apteryx>didn't we have a sysdig package?
<lechner->Hi, what's the easiest way to apply patches from Mumi to a Git tree, please?
<rekado>lechner: something like "wget -qO- https://issues.guix.gnu.org/issue/$1/patch-set $2 | git am --patch-format mbox -3"?
<lechner->rekado / thanks! what happened to your b4 work flow?
<lechner->rekado / also what's $2 ?
<rekado>nothing happened. The above is just the easiest way to apply patches from mumi to git.
<rekado>the optional patch set number.
<rekado>when there's a reroll count (e.g. v2 or v3) the /patch-set/n endpoint should get you the nth set of patches.
<lechner->it's a separate argument to wget?
<lechner->okay, there may be a slash missing
<rekado>oops, missing slash
<lechner->rekado / thanks! also thanks for Mumi, which is great
<futurile>just put out a post on package inputs: https://www.futurile.net/2024/03/29/guix-package-structure-inputs/ - very interested in feedback (and if there are any mistakes in my understanding!)
<futurile>I also discovered that I thought DAG stood for "Directed ACRYLIC Graph" - that's acrylic the material/paint - not "Directed Acyclic Graph" - and I never questioned it in my head heh
<futurile>Google was all "don't you mean acyclic" and I was "Waat?"
<lechner->yeah, i also do a double-take on that one
<KE0VVT>efraim: How do I run this again? https://bpa.st/PELA
<peanuts>"View paste PELA" https://bpa.st/PELA
<Aristide>Hello everyone, I am currently trying to create a package for an astronomical code named idg that is build using cmake. Everything was handled perfectly until a problem occured. It appears that, in the cmake instructions, there is a git clone command to load and build (directly inside idg source folder) some dependencies. However I get an error
<Aristide>message stating that git clone "Could not resolve host: github.com". Hence my question : What should I add to my code so that I have access to the internet during the cmake build ?
<Aristide>I am using the command : "guix build -K -f idg.scm"
<Aristide>And my file idg package is for now :
<Aristide>"
<Aristide>(define-public idg
<Aristide>  (package
<Altadil>Aristide: my understanding of packaging is limited, but I believe the network access is removed on purpose, for safety and reproducibility purposes. The best way is probably to package the dependencies separately, and then add them as inputs.
<Aristide>Altadil It makes perfect sense but unfortunately I am not part of the team that develop idg and it will be very difficult for me to create (and maintain) a branch of the project that do not build this dependencies internally and just assumes that they are already there so I was looking for a solution in which I can leave the source code untouched.
<Altadil>Aristide: I know it can be done directly from the guix package code, but that’s too advanced from me, I’m afraid. Hopefully, someone more experienced can help. :)
<Altadil>s/from/for
<futurile>Altadil: do you know https://mastodon.social/@sharlatan - they package a lot of astronomical code - if there's no-one who can help here they might be able to
<peanuts>"Sharlatan (@sharlatan@mastodon.social) - Mastodon" https://mastodon.social/@sharlatan
<futurile>I don't know if they ever come to IRV
<futurile>uh IRC even
<Altadil>futurile: I think you meant this for Aristide
<futurile>oh wow - I really need to go have a lie down
<ieure>Aristide, Guix does indeed block all network access during package builds, to enforce reproducability. You'd have to patch its build system to not use the network in order to package in Guix. Most software build systems are a complete shambles, so this is a pretty normal thing one has to do when packaging for Guix.
<jakef>and to clarify, you patch the build system in the guix package definition using guile, not by forking the project
<Aristide>Thank you all for your answers, I will try to patch this using guile then :)
<PotentialUser-79>Why *nobody* talks about emacs-guix? Its like first working gui program for installing programs on linux
<jakef>only time i hear about it is people saying it doesn't work for them
<rekado>:)
<rdrg109>[Question] I have been using guiix in Ubuntu without problems for the last 3 months. Today, I decided to create a new user on Ubuntu to troubleshoot a, what I believe is, a bug in a Guix package, so to avoid using my default profile which contains multiple configurations that could add complexity to the debugging process, I decided to create a new user. I logged in as the new user, when I execute
<rdrg109>"guix pull", this error is shown: "guix pull: error: Git error: unexpected http status code: 502". Does anyone know how to solve that problem?
<ieure>rdrg109, Wait.
<rdrg109>Ok, I'll wait. Here's more info: When I execute "guix pull" on my original user, the command doesn't report that error (i.e. it proceeds to "Computing Guix derivation for 'x86_64-linux'...)
<hiecaq>rdrg109: Well, why not "guix shell -C" in the first place? For the problem I think maybe it is related to per-user networking rather than Guix.
<ieure>rdrg109, Do you have a different channel configuration between the two users? Error indicates that the Git repo for one or more channels is broken -- 502 is an HTTP error which usually comes from the load balancer in front of a cluster of servers.
<rdrg109>ieure, hiecaq: Thanks for the help. I think it works now. I did not change anything, I executed the command again now the command proceeds to "receiving objects XX% [#### ]. I did have to wait approximately 30 seconds for that message to appear.
<ieure>rdrg109, Guix relies heavily on network availability, but networks are unreliable.
<acrow>Hmm.. problem with substitutes? Routine guix pull is doing some major rebuilds for mes-boot, tcc, binutils and many more. Makes me go hmmmmm.
<cbaines>acrow, can you share the specific derivations/outputs you're building?
<acrow>Though -- building from source is always the best; but substitutes do save time on the lesser systems.
<acrow>sure. just a sec.
<cbaines>substitute availability is generally good at the moment https://qa.guix.gnu.org/branch/master
<peanuts>"Branch master Guix Quality Assurance" https://qa.guix.gnu.org/branch/master
<acrow> https://paste.debian.net/1315301/
<peanuts>"debian Pastezone" https://paste.debian.net/1315301
<cbaines>acrow, this line in the paste is relevant: guix substitute: warning: ACL for archive imports seems to be uninitialized, substitutes may be unavailable
<cbaines>which suggests you're not getting any substitutes because no substitutes are authorised
<acrow>that's new.
<rdrg109>hiecaq: Thanks for the recommendation, I wasn't aware of it. I did a bit of research and came up with this command; => $ guix shell --container --preserve='^TERM$' --expose="$HOME/.config/ibus/rime" emacs emacs-rime -- emacs <=. However, this triggered some errors that are related to the database that is modified by the emacs-rime package: only one user can be connected. So, I will stick to creating
<rdrg109>a new system user and debug there.
<rdrg109>^ This leads to another inquiry:
<afm-victoria>Anybody running Emacs Cider here? I am having issues running a minimal clojure program with Cider.
<ieure>afm-victoria, Yes, I use CIDER often.
<afm-victoria>It runs just fine with `clj -X core/main` but Cider chokes when I try to jack-in
<ieure>Can you pastebin the error?
<afm-victoria>Here is the pastebin: https://pastebin.com/tkjmGQVQ
<peanuts>"Cider error message - Pastebin.com" https://pastebin.com/tkjmGQVQ
<ieure>afm-victoria, Oh, I vaguely recall seeing this, but I don't remember what I did to fix it.
<afm-victoria>It happens on guix linux but also on Fedora with Guix as a package manager
<ieure>afm-victoria, Ah, issue, is that Guix needs the "jdk" output of openjdk.
<afm-victoria>Great! I'll give that a whirl!
<ieure> https://issueantenna.com/repo/clojure-emacs/cider-nrepl/issues/857
<ieure>That's what I found which jogged my memory.
<ieure>...because it had the answer.
<afm-victoria>*blushing* as I tried to look the issue up with Google but, as we all know, Google is not great anymore...
<afm-victoria>ieure: thanks for the tip! I think it's the second time this week you've thrown me a life line!
<ieure>Glad I could help!
<ieure>I use SearxNG these days, run my own instance. It's far from ideal, but I don't hate it.
<KE0VVT>I installed font-google-noto-emiji but I can't see color emoji.
<KE0VVT>I logged out and in.
<nutcase>Hi, when I add an mcron job to my operating-system services, it removes already defined mcron jobs (e.g. rottlog). I did it as the manual told, which results in: https://paste.debian.net/1315251/ What am I doing wrong?
<peanuts>"debian Pastezone" https://paste.debian.net/1315251
<jonsger>ACTION tries to revive his icedove@115 patches...
<afm-victoria>dthompson: congrats on your new role!
<afm-victoria>dthompson: you might know me as monkey1 on fosstodon
<dthompson>afm-victoria: oh hey! thank you :)
<ulfvonbe`>it would be nice to have a sort of "global" package transformation ability. right now I'm building 22 rust versions for the second time because there's no way to transform the packages used in profile hooks.
<ulfvonbe`>likewise with e.g. packages used in building origins
<jackhill>KE0VVT: don't know if this will help, but have you treid an `fc-cache -r`?
<jakiki6>the commit 95937506983 broke my system configuration because I can no longer specify specific package outputs in my os configuration without it failing with "In procedure struct-vtable: Wrong type argument in position 1 (expecting struct)"
<acrow>For some time now, on guix system, /etc/guix/acl was linked to the store and substitutes seemed to work. Recently, guix weather complains about the lack of /usr/local/etc/guix/acl... I suspect this is part of a work in progress.
<wingo>it seems that we need "libspectre" to have evince read ps files
<wingo>ah it is there
<wingo>i wonder why evince isn't loading https://web.archive.org/web/20030313122854if_/http://www.ccs.neu.edu:80/home/matthias/Scheme2000/hilsdale.ps
<wingo>ah. we actually need to add -Dps=enabled to the configure flags, as it currently defaults to off
<wingo>tested via guix build evince --with-configure-flag=evince=-Dps=enabled
<wingo>if someone has commit access, plz push that :)
<wingo>i.e. add that flag to gnome.scm:2628
<dariqq>acrow: did you run sudo make install on a guix checkout by any chance?
<afm-victoria>ieure: I have another head-scratcher. I run into a 'class not found' error when using clojure-tools in Fedora with guix as the package manager. If I use the 'clj' tool that I installed on Fedor, the program runs with no issues. Here is the pastebin: https://pastebin.com/t8stJsr1
<peanuts>"Clojure java class error - Pastebin.com" https://pastebin.com/t8stJsr1
<afm-victoria>I tried with a couple of versions of clojure-tools: clojure-tools 1.11.2.1446 out bin-guix/packages/clojure.scm:22:2
<afm-victoria>and clojure-tools 1.11.1.1413 out gnu/packages/clojure.scm:204:2
<afm-victoria>the first one is from a channel that you run; the second one is from the regular Guix channels
<afm-victoria>both show the issue
<afm-victoria>I have a work-around by using the tool under Fedora, so this is not a pressing issue.
<ieure>afm-victoria, Not sure on that one. Looks like a missing dependency, maybe this library assumes you'll explicitly depend on the right thing and doesn't explicitly do so?
<ieure>afm-victoria, I haven't used deps.edn that much, because I don't think highly of it, but if there's an equivalent of `lein deps :tree', I'd start there.
<afm-victoria>I think it's a deps.edn issue for sure. The odd thing is that the code works fine when I download the Clojure tools from their site and use it to run my program.
<ieure>The clojure-tools in Guix are way out of date.
<afm-victoria>I decided to focus on Guile because I am tired of all these java issues that are ubuquitous in Clojure.
<afm-victoria>I am using clojure only to do some reports for work because they are done now.
<afm-victoria>Moving forward, no Clojure for me, especially under Guix
<afm-victoria>Thanks for your suggestion, anyway
<nutcase>Hi, I am running a cron job as a user that is not root. Running the command in question as user from shell works fine. If mcron issues the command, I have TLS/SSL issues: "certificate verify failed: unable to get local issuer certificate". What is the reason? Is nss-certs missing in the mcron job's profile?
<afm-victoria>nutcase: have you looked to make sure the environment has the right environment variables?
<nutcase>afm-victoria: I only checked $PATH and it only contains /run/current-system/profile/sbin
<nutcase>sorry /run/current-system/profile/bin
<afm-victoria>maybe your regular shell environment has variables that are set for the job but you don't have them available in the other environment. Look at the code blobs in https://guix.gnu.org/manual/en/html_node/X_002e509-Certificates.html. They export variables so that the program that needs them can find them.
<peanuts>"X.509 Certificates (GNU Guix Reference Manual)" https://guix.gnu.org/manual/en/html_node/X_002e509-Certificates.html
<afm-victoria>The R example is telling: one needs to have that special CLUR_CA_BUNDLE variable set
<acrow>dariqq: No, I'm digging around to find out why guix pull is unexpectedly no longer successfully pulling substitutes. It is a long standing guix system that is now behaving differently. I did git pull the latest sources but just throwing out observations bc it might help someone else.
<nutcase>afm-victoria: I think https://guix.gnu.org/manual/en/html_node/X_002e509-Certificates.html is outdated, since nss-certs is now part of %base-packages. Therefore I thought that it is included in whatever profile the mcron job is running. But maybe I'm mixing profile and operating-system up a bit. SSL_CERT_DIR and SSL_CERT_FILE are not set in the mcron job
<peanuts>"X.509 Certificates (GNU Guix Reference Manual)" https://guix.gnu.org/manual/en/html_node/X_002e509-Certificates.html
<afm-victoria>I think you still have to worry about the environmental variables because there is no way for nss-certs to know which ones to set
<afm-victoria>and depending on what you do with your job, you might have to set special variables as is the case with R
<afm-victoria>I had to do it for Racket--raco couldn't pull packages until I told it where to look
<afm-victoria>for the certs
<ieure>nutcase, afm-victoria, Note that you're linking to the manual for Guix release 1.4.0. The current manual is here: https://guix.gnu.org/en/manual/devel/en/html_node/Application-Setup.html#X_002e509-Certificates-1
<peanuts>"Application Setup (GNU Guix Reference Manual)" https://guix.gnu.org/en/manual/devel/en/html_node/Application-Setup.html#X_002e509-Certificates-1
<ieure>IMO the release manual shouldn't be linked on the website, it regularly causes this kind of confusion.
<afm-victoria>ieure: thanks for the reminder.
<afm-victoria>nutcase: I think my comments still stand as the development manual has much the same content under 11.12 X.509 Certificates
<coyote>hm, i don't know if i understand how 'guix gc' works, because i did `guix gc' and now it's redownloading nearly everything from my profiles, even though they should be already there (i assume! since i am using said things)
<coyote>well, the gc isn't redownloading, but a subsequent reconfiguration
<ieure>coyote, I think this is a known thing, something something grafts?
<ieure>Not fun, whatever it is.
<nutcase>ieure: thanks for the clarification.
<nutcase>afm-victoria: yes, you're right the part with environment variables did not change
<coyote>ieure: that makes sense
<coyote>well that'll teach me to not run guix gc without arguments ;-)
<nutcase>afm-victoria: prefixing the command with "SSL_CERT_DIR=/run/current-system/profile/etc/ssl/certs" let my mcron job suceed. Thank you!
<afm-victoria>nutcase: glad you figured it ou!
<afm-victoria>out