IRC channel logs


back to list of logs

<vagrantc>ACTION wonders why gnu/system/examples/vm-image.tmpl is apparently a desktop style image
<vagrantc>i guess a vm can be anything ...
<vagrantc>ACTION fires up guix on an aarch64/arm64 system for the first time in ages
<xisop>guix shell is game changing. i'm trying out the ungoogled chromium example from the docs.
<xisop>is it possible to cache the assets used for a guix shell? or is it just advised to use guix pack if i need it again?
<vagrantc>they should be cached out of the box
<vagrantc>e.g. the second time you run it, it should be much faster
<vagrantc>you could install the relevent packages in a profile ... e.g. guix install --profile=some/dir/my-profile PACKAGE1 PACKAGE2 ...
<vagrantc>as long as you don't run "guix gc" they should still be there, though
<xisop>vagrantc: thank you! that is very helpful
<vagrantc>i really like guix shell to experiment with new things without having to remember to remove them :)
<dthompson>vagrantc: yeah it's a wonderful thing :)
<vagrantc>anyone have an example of how to use the C.UTF-8 locale that is present in newer versions of glibc (2.35?) in guix?
<vagrantc>yup, apparently 2.35
<peanuts>"Proposals/C.UTF-8 - glibc wiki"
<xisop>i like the container aspect. it really helps me separate concerns between subsystems
<Kabouik>Does anyone have a private channel with Go 1.18?
<vagrantc>ACTION ponders why "guix system build config.scm && guix system init config.scm /mnt" ends up downloading lots of packages during init
<Kabouik>Oh but wait we have go 1.20.2 in the Guix channel. No I wonder why my package definitions pulls 1.17.13.
<Kabouik>Oh, we have (define-public go go-1.17) in golang.scm. Is 1.17 the current stable or should this be updated? The other packages are already there, just not default.
<Jaeme>Kabouik: It should be updated, I'm guessing that updating it would cause a World rebuild.
<Kabouik>But the world needs a rebuild.
<Jaeme>Later :yawn:
<Jaeme>Go versions can be specified in the arguments field of the go-build-system, so it's not stopping anyone.
<xisop>okay, that is really cool: guix shell --container only shares the specific cwd and nothing else. so this container wont be able to see parent dirs
<Kabouik>Yes, I'll add the version requirement to my package Jaeme.
<Kabouik>How does one set a minimal version for go-build-system Jaeme? I'm failing at finding the correct syntax in Guix sources
<vagrantc>Kabouik: never done any go packaging ... but gnu/packages/networking.scm:(define-public yggdrasil ... appears to use go 1.20 ... i think.
<vagrantc>passed in the arguments
<Jaeme>Kabouik: You can use the `#:go` keyword along with the go package variable.
<Kabouik>Oh, I see. Thanks. I'll try that. Actually my package builds fine with clang, but disables one of the options if a valid Golang version is not found, so I probably need two build systems
<Jaeme>Kabouik: `guix/packages/golang.scm` has a lot of examples.
<Jaeme>`gnu/packages/golang.scm` I mean.
<wdkrnls>Dear Guix, how do you automate tracking your config.scm?
<Jaeme>wdkrnls: In what way/what do you mean?
<wdkrnls>I'm just confused. I have two config.scm and I figure the one that is actually used is in my ~/.config/guix directory.
<wdkrnls>Well, the ~/.config/guix/channels.scm file looks right while /etc/channels.scm looks kind of old.
<Kabouik>I may just need specifying go-1.8 instead of just go in my package inputs Jaeme
<wdkrnls>But I want to get rid of doing anything with any ~/.config/ or any other dotfile.
<Kabouik>"Looking for include file xcb/xcb.h - not found" even with xcb-proto as input. Any idea? It's doesn't make build fail, but still, it itches my OCD.
<wdkrnls>My mind has trouble with keeping track of things if they have a global configuration.
<wdkrnls>I want project-specific configuration.
<vagrantc>guix system describe will tell you the config that it *actually* uses, which is probably a copy of one of your other config.scm
<Kabouik>libxcb it is!
<wdkrnls>@vagrantc: Interestingly guix system describe suggests that neither are being used :/
<wdkrnls>I wish it showed what it was linking to in the output, rather just which channels were used.
<Jaeme>wdkrnls: It doesn't really matter where your Guix config files are located.
<vagrantc>guix system uses a config that is in the store
<Jaeme>*the ones you use to instantiate a `guix pull` and `# guix system reconfigure`
<vagrantc>why does guix system init pick a 6.0.10 linux-libre kernel? ... my guix is from the most recent pull ... would think that woudl be 6.6.x or something
<pkill9>what command would create a blank db.sqlite file?
<pkill9>with all the correct tables etc
<nathan__>i dont see "6.0.10" in guix anywhere. maybe you ran an guix from the root account instead of your user's guix. or the other way around.
<pkill9>whoa I fixed the sqlite database, this needs to be documented somewhere
<pkill9>sqlite3 db.sqlite.bak2 ".recover" | sudo sqlite3 db.sqlite
<vagrantc>nathan__: most certainly not.
<vagrantc>my gues is it is using guix from the "guix" package or something.
<vagrantc>oh, yes. from guix 1.4.x from the installed package.
<vagrantc>guess sudo is not using my user's guix ... that would also explain why it rebuilt almost everything
<vagrantc>so much for certainty! :)
<Kabouik>I have no experience with packaging programs that use go, but how do we deal with something that tries to access the Internet to download libraries? Logs:
<vagrantc>Kabouik: you package the libraries
<Kabouik>Uh oh
<vagrantc>recursively :)
<Kabouik>I'm out, leave me alone!
<Kabouik>I won't go down that rabbit hole, ni Whatsapp support in my nchat package then
<vagrantc>nathan__: wow. you were absolutely right. sorry for being so knee-jerk about it. :/
<Kabouik>Frustrating though, it would have been my only wai to finally see what people chit chat about on their Whatsapp groups that I've always refused to join by lack of an Android device.
<wdkrnls>The info page suggests it should be using the ~/.config/guix/channels.scm by default unless otherwise specified.
<wdkrnls>Is there still no way to get guix shell completions inside of shell-mode in Emacs?
<wdkrnls>Mine is failing with a note hinting that Guix macrology is not being loaded: ~/.config/guix/current/share/guile/site/3.0/guix/scripts/deploy.scm:176:7: Unknown # object: "#~"
<vagrantc>this reminds me of all the crazy stuff i did to get the linux-libre kernel working for aarch64 ... only i've forgotten everything :)
<pkill9>why does this keep failing every build? guile: error while loading shared libraries: cannot open shared object file: No such file or directory
<pkill9>guile failing to find/open
<Kabouik>So let's say I TRY to import the Go dependencies recursively to package them vagrantc. If I want to try that in a quick and dirty way in my private channel, can I just append the imported package definitions in my somepackage.scm I needed them for in the first place, or do they need to be in separate files?
<Kabouik>Also, when using guix build -f somepackage.scm, I need to add the name of the package at the end of the package definition. Do I need to do that for all dependency go libraries imported too if they are in the same file?
<Jaeme>pkill9: That's because you don't have libgc as an input.
<pkill9>this is when i build anything
<pkill9>or run guix pulll
<Jaeme>pkill9 Oh.
<Jaeme>pkill9: Are you on Guix System?
<wdkrnls>Kabouik: I think I understand what you are asking. I think not.
<wdkrnls>Kabouik: My understanding is that the expectation is that in a sense you are returning the scheme object.
<wdkrnls>Kabouik: Just as with any other scheme code, only the last object will be returned.
<Kabouik>Ok wdkrnls, so that would be one file.scm per package, which means I need to figure out the use-modules for each. Hum. I was hoping this could be just like those big .scm files in gnu/packages/ where all packages are together, but I assume this works only if this is pushed to a git repo.
<wdkrnls>Kabouik: For the first bit, I agree that I don't think you actually need a channel.
<wdkrnls>Kabouik: I think you can just make one big scheme file.
<vagrantc>substitute availability for aarch64 does not seem too terrible, yay! :)
<Kabouik>I did try that first wdkrnls, but indeed only the last one gets built if I use guix build -f
<wdkrnls>I like to define packages in my manifest.scm files rather than bother with channels most of the time since channels are a pain.
<Kabouik>Maybe I would need a (define-moduile (my channel file.scm) and then guix install -L. packagename
<Jaeme>pkill9: What distribution are you on and how did you install guix?
<pkill9>I used the binary installation
<Jaeme>pkill9: Installation script?
<pkill9>there's probably something funky with my machine, probably not a guix problem, hopefully
<Jaeme>It's not an selinux issue right?
<pkill9>ive set it to permissive
<Jaeme>Can you send the full log of guix pull in a pastebin:
<peanuts>"Debian Pastezone"
<pkill9>I'll send a log of `guix build hello` as that has the same issue
<pkill9>actually this is the full log: guile: error while loading shared libraries: cannot open shared object file: No such file or directory
<pkill9>nothing else in it
<pkill9>guix build: error: build of `/gnu/store/1qxiibc83lxxmxd40ixpl652fn09yg40-hello-2.12.1.drv' failed
<pkill9>also guix build guile returns guile: error while loading shared libraries: cannot open shared object file: No such file or directory
<pkill9>also when running `guix environment guile`: gzip -dc /var/log/guix/drvs/86/r725ya2p5kn58f6mp4q4zs8g6ygha4-bash-minimal-5.1.16.drv.gz
<pkill9>guile: error while loading shared libraries: cannot open shared object file: No such file or directory
<pkill9>not sure why guile isn't finding these things in the build environment, unless my drive is failing and the binaries dissappeared
<pkill9>but not sure how to check that
<Jaeme>pkill9: Is this a recent installation of fedora?
<pkill9>it's not terribly old
<pkill9>but not particularly new
<pkill9>guix could be corrupted somehow, but I'd like to know how
<pkill9>one of my fears about relying on guix is corruption
<pkill9>it can be finicky to fix such things, or to reinstall, which can also be finicky
<Jaeme>pkill9: sorry I mean what edition of fedora is it?
<pkill9>it's the normal version
<Jaeme>pkill9: Fedora Workstation 39?
<pkill9>fedora workstation 38
<Jaeme>pkill9: I wish I could help more, maybe you can try reproducing the issue by reinstalling Guix? Were you able to `guix pull` before or is this the first one after installation?
<pkill9>i could do so before
<pkill9>how would reinstalling guix reproduce the issue?
<pkill9>reinstalling it would fix it, i just want to know why it's doing this
<Jaeme>You can try emailing the mailing list `` so a maintainer could look over your issue.
<Jaeme>pkill9: I'm sorry that I couldn't provide much help.
<pkill9>no worries
<pkill9>thanks for trying to help
<Jaeme>You can subscribe to the mailing list here:
<peanuts>"Help-Guix Info Page"
<Kabouik>I'm making no progress with one of the go libraries I need to package for nchat. Build fails because " no Go files in /tmp/guix-build-go-go-mau-fi-libsignal-0.1.0.drv-0/src/"
<Kabouik>Yet I have an `(#:/import-path "go-mau-fi/libsignal") that looks exactly the same as everything else I could find in other Go packages.
<Kabouik>Minus the / typo just here
<Kabouik>And seeing the build logs, that folder seems full of go files:
<Kabouik>Oh, it seeems worked. However, I have absolutely no idea what I'm doing.
<Kabouik>Wrong paste URL:
<Kabouik>I believe I have all go dependencies packaged and listed as inputs of the main package, but the compilation of the main packae still shows go trying to download libaries from the Internet
<ulfvonbelow>I keep getting errors in test suites when trying to package python stuff, which looks like this:
<ulfvonbelow>E ModuleNotFoundError: No module named 'tests.utils'; 'tests' is not a package
<Kabouik>If anyone knows how to make Go packages build offline when inputs are already packaged, please ping me. I'm going to be mostly away from PC from the next 24h but I'll check later, and I think I'll just give up and revert my commits about this whole Go build if this is too complicated, as currently my private channel is broken with them.
<ardraidi>Hello Guix!
<lalebarde>I have installed icecat on a fresh guix. I am surprised because I cannot run it directly and which does not find it. Guix tells me at the end of the `guix install icecat` : "Advice: thiink about setting up environment variables by running: `GUIX_PROFILE="/home/lalebarde/.guix-profile" . "$GUIX_PROFILE/etc/profile"` otherwise, consider `guix package
<lalebarde>--search-paths -p "/home/lalebarde/.guix-profile"` . I am lost: I have expected just install & use. Is Guix only for experts? Shall I have to spend days and weeks to just figure out how to use it? I am disapointed.
<ardraidi>I'm writing a patch to update labwc to 0.7.0, which requires wlroots >0.17. Unfortunately, wlroots >0.17 breaks a bunch of WMs, including Sway.. What's the best approach here? Create a wlroots-0.17 package and once the other WMs catch up, remove it? Or just wait until wlroots >0.17 gains more support then update it and labwc?
<ardraidi>lalebarde: The last two statements are not constructive, I think. Anyway, did you try restarting the terminal or rebooting your PC? Not an expert, but IIRC these environment variables get update automatically.
<lalebarde>ardraidi Yes, I was provocative purposely to draw attention because I am always sad to see genious solutions die because not targetted to everyone. I am not completely a newbie with 20 years Linux practice and at this stage I find Guix difficult to use. I miss it does not work out of the box and there should not be the need to a
<lalebarde>Windows-disease-reboot after a package installation to setup some environment variables.
<ardraidi>lalebarde: I'll skip the provocation part. IIRC, the reboot is rarely needed. I think you can just source the variables. There are a bunch of factors why you might need or need not to reboot, like if you're using Guix the distro or Guix on a foreign distro. There's also the first pull, which would replace the guix command with another that is specific to the user, for example.
<ardraidi>If you think it's a bug, you can file one. Or even better, submit a patch. Sorry I can't help much.
<lilyp>lalebarde: Guix warns you, because it doesn't do those envvar modifications for you when you freshly install a package. If you logout and back in again, you can use your Icecat, same with a reboot. The warning is only needed if you want to want to use Icecat right after installing it.
<lilyp>Guix is quite clever about this and warns you about any envvar modification, not just PATH. For example, if you install emacs from scratch, it will warn you as well due to EMACSLOADPATH.
<lalebarde>lilyp ok, so I can set up the environment variables as suggested by guix to use the package now and it will be managed by guix at the next session. It would be nice that guix includes that in its post install advice.
<lilyp>You cannot appease everyone. If Guix were to print out a verbose explanation for everything it did, people would complain that it's too verbose.
<lalebarde>lilyp One feature of verbosity is to have different levels, there could be one for newbies ;D . I stress the point and I am provocative because I think conceptually, Guix is the best distro today. I have used in the past Mandrake (my first one), Ubuntu, Gentoo, Debian and AlmaLinux (at work). I was traumatized by distro upgrades falls. So I want
<lalebarde>Guix to succeed and have massive adoption.
<lilyp>I mean, it's not like the manual doesn't already tell you about this.
<lilyp>You can try to provoke me into reading it for you, but I'm not sure how much it helps you learn.
<rekado>lalebarde: you have a unique opportunity here that we all no longer have: you are new to this, so your point of view when it comes to the manual is valuable. If you have recommendations or patches to the manual to make it address things you would have found helpful we’d be happy to discuss and possibly incorporate them.
<ardraidi>You guys are awesome. :)
<ardraidi>Maybe I'll submit the wlroots patches as RFC.
<lalebarde>lilyp I don't want to hurt anybody. There is a time to learn and one to use. There are people who want to learn and others who don't. You should be able to use without learning and learn when you want if you want. Remember that M$ succeeded with a bad product because they targetted the masses
<lalebarde>rekado Alright, I will do my contribution (y)
<lilyp>Targetting the masses is quite distinct from targetting the masses (M$).
<lilyp>M$ succeeded, because it made itself the default and most people could not if their life depended on it figure out how to change that.
<rekado>still no movement on
<rekado>no idea what’s wrong :-/
<lilyp>No new builds succeeding?
<rekado>all the machines already went through a reboot due to the firmware upgrades, so it’s unlikely to be weird state.
<rekado>they do succeed
<rekado>but only a handful is pushed to nodes
<rekado>all the nodes keep asking for work and are not given anything to work on.
<rekado>just a handful of i686 jobs and the odd rust job for x86_64
<lilyp>That's annoying, given that I'd really like them to build webkit like right now.
<rekado>I have this bulk R update that I really want to push to master
<lilyp>Better yet yesterday, as it stalls gnome-team.
<lilyp>Maybe there's something wrong in the db?
<rekado>I don’t know exactly how cuirass decides if there’s work to be done. Someone would have to start a REPL and query the db to see where it’s stuck
<rekado>yes, that’s my guess
<pkill9>wow I think i fixed it, I recovered the database, then ran guix gc --verify=contents,repair and then guix gc
<tschilp>just received 502 "Bad Gateway" for „“!
<tschilp>from guix substitute
<lilyp>pkill9: "it" being?
<pkill9>basically whenever i tried to build anything, giule kept failign to load shared libraries
<pkill9>including running guix pull
<tschilp>How would I let a unreachable substitute (like linux-libre-module-builder-6.6.8 right now for me) let be built on my computer with guix system reconfigure, without having all substitutes being skipped (I just know that there exists a --no-subtitutes or similar flag)?
<pkill9>tschilp: --fallback
<tschilp>pkill9: thanks, will try it right away
<tschilp>mhm, I run 'sudo guix system -L my-guix/packages/ --fallback reconfigure my-guix/system-configuration.scm', but still it fails after substitution fails. I actually recall having this module-builder being built here on my pc most of the time anyway, so I wonder if anything has changed in this regards last month?
<tschilp>The actual error is (sorry, my shell speaks german :)): 'guix system: Fehler: Beschädigte Eingabe, während das Archiv aus „#<closed: file 7ff5927c39a0>“ wiederhergestellt wurde'
<tschilp>this is guix f24b14767d362a84e6469682b4fe303b50f4b589, sorry forgot to mention.
<peanuts>"guix.git - GNU Guix and GNU Guix System"
<tschilp>it is interesting -- 'guix substitute' seems to contact for the substitute download, receives 502, then it tells to try with another substitute URL and actually calls the same one with the same error and then fails with the information 'there couldn't be an alternative substitute'.
<tschilp>'s|substitute|substitute be found|'
<lalebarde>Where shall I raise an issue or a proposal?
<lalebarde>Is it normal that `guix system reconfigure /etc/config.scm` reloads all the packets?
<lrustand>What do you mean with "reloads all the packets"?
<lalebarde>all the packages (not packets, sorry)
<lrustand>I understood that, but not the question
<lrustand>What do you mean by reload?
<lalebarde>I have a fresh install (less than one day), so I assume most packages are up-to-date. I have modified `/etc/config.scm` to add users and groups. Then `reconfigure` makes Guix download again all the packages like btrfs-progs-static-4.2
<lrustand>That shouldn't happen unless you have pulled since the last time you reconfigured
<lrustand>But if you didn't pull after you installed you probably should, since I think it is often outdated on fresh installs
<lalebarde>Understood, thanks lrustand. But that would be nice that even after a `pull` it downloads only what is obsolete.
<lrustand>I don't think it downloads anything that is not updated
<lalebarde>ok, then I will test. After it has finished, I will do again `pull` and `reconfigure`.
<tschilp>OK, some kind of forcing with '--fallback --substitute-urls=""' downloaded linux-libre-6.6.8.tar.xz, which started the derivation chain. It feels a litte weird though, that '--fallback' alone did not do the trick! Is it possible that a failed substitute download on bordeaux currently redirects to bordeaux itself?
<lrustand>I also have problems with substitutes from time to time and have to disable one of the substitute urls
<lrustand>I think it is way too brittle, and a failed download from a substitute shouldn't be a fatal error
<unfroq>Hey Guix!
<unfroq>I get some error I can't solve, when trying to reconfigure my system. I tried to build the package which was mentioned in the error, but it fails and I don't know how to solve this.
<unfroq>guix build -q ghc-fgl
<unfroq>guix build: error: reading file `/gnu/store/137l8cmib0b254di8p79g2ch0pr0s1q5-ghc-fgl-': No such file or directory
<unfroq>I've got a similar error on another package a few days ago.
<snape>lalebarde: just passing by and reading the log... it seems the whole discussion is because you didn't source ~/.guix-profile/etc/profile in your ~/.bash_profile?
<snape>you should be able to run programs right after installing them
<snape>fun nobody told you
<snape>they were more focused on countering your provocation
<peanuts>"Getting Started (GNU Guix Reference Manual)"
<snape>also, "guix home" (if you choose to use "guix home") does that automatically
<lrustand>Oh, if there was more to it than what I answered I didnt see the previous messages, since I just logged in right before answering
<lrustand>I have removed a channel from my channels.scm, but this channel is still being pulled every time I run guix pull. Why is it doing that?
<lrustand>ah, figured it out. My own channel had specified it as a dependency
<jsbiff>Hello. I'm wondering if anyone has figured out a way to get the VirtualBox Guest Additions to install in guix?
<jsbiff>I suspect I'm either missing some packages, or the guix filesystem layout is just a bit different than most other Gnu/Linux distros?
<jsbiff>I get this error:
<jsbiff>"Unable to determine correct library path"
<unfroq>jsbiff: how do you try to install the guest additions?
<jsbiff>I used sudo -i to become root, I mounted the Guest Additions iso to /mnt/cdrom, then:
<jsbiff>cd /mnt/cdrom
<jsbiff>I'm reviewing that file now, and I discovered that the reason for the failure is that this bash script tries to invoke ldconfig and that's not installed? I thought ldconfig was an essential part of any Gnu/Linux distro?
<jsbiff>I've been using various flavors of gnu/linux for many years, but just starting to dabble with Guix
<jsbiff>I tried running: find / -name ldconfig, and there are a couple hits under /gnu/store
<jsbiff>so, is there a guix package I need to install that provides ldconfig? How is it that Guix can even run without that?
<unfroq>jsbiff: I am not sure how much effort it takes to get the guest additions running with Guix. You could check out for some hints.
<peanuts>"GitHub - palfrey/guix-vm: Scripts and support necessary to make a GuixSD Virtualbox image"
<jsbiff>What is GuixSD?
<jsbiff>Interestingly, I quickly found on that page you two linked, a note that VBox GA doesn't work because of exactly the issue I ran into
<jsbiff>I'm confused how Guix can work at all without a ldconfig lol. That would probably be an interesting bit of engineering to read up about
<lilyp>GuixSD is an old old name for Guix System that just won't die :)
<jsbiff>I understand that largely, the stuff a user wants to run is installed in the store and linked in their home dir, but, there must be at least a little bit of stuff to bootstrap the whole thing, that must be system wide. Like, for example Wayland or (not sure which guix is using) must have to be installed at a system level, right, because it
<jsbiff>starts up before a user logs in?
<jsbiff>Thanks lilyp
<lilyp>Guix is still on X for Gnome by default atm
<jsbiff>Also, Shepherd has to be installed at a system level, because it's the init system - literally the first thing to run - although, it runs as root .. . maybe all the system level stuff is just installed in the root user profile?
<lilyp>(we should maybe switch to wayland on gnome-team, but …)
<lilyp>You shouldn't populate the root user profile unless you are camping out there with a root shell open all the time.
<lilyp>The system profile is its own profile distinct from root's and everyone else's profile.
<lilyp>You get to it from init, which gets mapped into the initramfs
<lilyp>Anyhow, the canonical way to modify it is per the packages and services fields in your operating-system configuration.
<lilyp>I don't know much about VirtualBox to give you more hints on what to do, but you might want to use qemu instead :)
<apteryx>oh, mumi seems disconnected from
<apteryx>rekado: ^ is rsync running?
<apteryx>see e.g. vs
<peanuts>"#57068 - Resizing mcron job in vm-image.tmpl interferes with settings - GNU bug report logs"
<peanuts>"Resizing mcron job in vm-image.tmpl interferes with settings"
<unfroq>Anyone has any ideas, why I get a 'No such file or directory' error regarding a derivation (.drv) when I try to build a package? Or how to get rid of the error? Keeps me from upgrading my system...
<tschilp>unfroq: how do you build the package? How does it prevent you from upgrading?
<unfroq>tschilp: guix build -q ghc-fgl guix build: error: reading file `/gnu/store/137l8cmib0b254di8p79g2ch0pr0s1q5-ghc-fgl-':
<unfroq> No such file or directory
<unfroq>tschilp: when I reconfigure, the reconfiguration results in this error as well.
<tschilp>and the file does exist on your system? I guess you are on a local guix checkout?
<unfroq>tschilp: actually, no, it doesn't. I am not sure, what you mean with local guix checkout? I have a system installation fetching from the main online guix repos and a few other channels.
<tschilp>Sorry -- I'm just trying (didn't notice that ghc-fgl actually exists in the guix repos, that's why I thought it was something custom!)
<unfroq>tschilp: I am not sure if this is specifically related to ghc-fgl.. I had another package yielding in the same error a few days ago. Can't remember which one it was. I did a guix gc and tried again after pulling latest guix. Now it fails at ghc-fgl which seems to be a dependency for something in my config.
<tschilp>mhm, here it actually builds (guix f24b14767d362a84e6469682b4fe303b50f4b589), do you get any more information ommitting '-q'?
<jsbiff>The main thing I wanted Guest Additions for is dynamic resolution and clipboard copy/paste to host. For resolution, I can just manually set a suitable resolution in the settings for xfce/gnome/whatever DE I'm using. I'm sad not to have clipboard integration though.
<jsbiff>I might be able to mostly get away with using ssh windows for copy/paste to host.
<jsbiff>Thanks all
<unfroq>jsbiff: good luck with your explorations. I can recommend using qemu as well with Guix.
<unfroq>tschilp: Sure, here is the output:
<peanuts>"debian Pastezone"
<jsbiff>Well, I'm using WOE 10 64-bit as my host, so Qemu is not really an option, unless I run VMs in VMs and that seems overly complicated and like just asking for performance and other issues lol
<jsbiff>I'm running guix pull for the first time and it's taking a long time (I guess it's pulling the source for basically all packages and building them, is that right?). I presume this is one of those things that takes a long time the first time, but subsequent runs should be a lot faster as it only needs to pull and build things that are different?
<lalebarde>Thanks snape . Why isn't it included by guix in *bash_profile* directly?
<unfroq>jsbiff: Yes, the first runs take more time than subsequent ones, usually. It also depends a little bit which substitutes are available for the guix revision you are using. But that gets relevant after pulling.
<tschilp>unfroq: interesting, I don't really see, why it wants to get this derivation, when it actually could get it from somewhere. But today I did have to add '--fallback --substitute-urls=""' to prevent guix getting something from the bordeaux substitute server, which was not there. Maybe this could help (or at least it would not any harm I can think of)
<tschilp>(to my guix system reconfigure command)
<tschilp>thanks again to pkill9 again for the '--fallback' flag at this point :)
<tschilp>like guix system $FLAGS reconfigure ...
<unfroq>tschilp: thanks, tried it. But it does not result in something different, unfortunately. Getting the same error when building the package in the guix repl.
<tschilp>unfroq: :(
<unfroq>I am not sure what my actual problem is.. but I try to build ghc-fgl (not modified) and it complains, that a .drv is missing from the filesystem (which is correct, it is missing). Is there a way to force a rebuild or something like that, in order to create the .drv file which is missing?
<tschilp>maybe there is a mismatch with the guix commits going on. I recall a cleaning session with '--delete-generations' for home, system and guix itself in combination with gc, which brought things out of sync pretty much for me. If you check with 'guix describe' and 'guix system describe', it's probably not the same at the moment -- does the problem of not-being-able-to-reconfigure persist, if you bring guix back to the same commit as your
<tschilp>system is at right at the moment (guix pull --list-generations, guix pull --switch-generations=$NUMBER_WHICH_CONTAINS_SYSTEM_COMMIT, sudo guix system reconfigure ...), just for checking...
<unfroq>tschilp: There actually is a mismatch. The system revision is two weeks old.. I will look into this.
<unfroq>I am using 'sudo -E guix system reconfigure ...' to apply the system config. What is the recommended way to use sudo for this? Without -E, just plain sudo?
<graywolf>unfroq: I am using just: sudo guix system reconfigure foo.scm
<unfroq>graywolf: I'll try this. The docs also recommend just using sudo without any flags.
<unfroq>tschilp: I was able to reconfigure after switching to the same user level generation as the system one. Thanks for the commands.
<gnucode>hey guix! do ya'll support becachefs as a filessystem yet? Just curious
<unfroq>gnucode: there are some packages available.
<peanuts>"Search ? Packages ? GNU Guix"
<meorly>is the guix documentation website not up to date? i'm seeing different context in guix info page and the website
<gnucode>unfroq: well the packages are there...but that doesn't really mean that you can do a (file-system (type bcachefs) ...)
<rekado>apteryx: it’s running now
<rekado>but cuirass still isn’t building much at all
<Jaeme>meorly: The guix manual and cookbook both have a "latest" and "stable" version (1.4.0) in the website.
<Jaeme>meorly: Needless to say you most likely want to be reading the latest development manuals.
<meorly>ah, i landed on it from google that must be it
<Guest14>If I send in a patch, is the qa ci still building the package for various platforms?  My PC takes ages
<jsbiff>So, I notice that it seems like the download repo that guix is using is a bit on the slow side (topping out occasionally at around 2MBps). Is it possible to point guix at a faster mirror?
<jsbiff>It appears to be downloading packages from:
<rekado>jsbiff: it a tiering issue as far as we know. The server at gives me 50MB/s downloads from AWS in Frankfurt.
<rekado>I don’t know of a mirror.
<rekado>Guest14: currently, there are weird problems with Only a handful of builds are scheduled at a time, so the queue is continuously growing.
<Guest14>rekado: Ah thanks, good to know.   I thought is hosted in berlin.  Since when is AWS a thing?
<rekado>no, I’m saying that from an AWS EC2 instance in Frankfurt I get 50MB/s when downloading from
<Guest14>Ah, my bad.
<ennoausberlin>Hello. I am starting another attempt to build the pharo vm on guix. Does anyone know which package contains ndir.h?
<Guest14>did not find it in glibc but apparently that is the place it lives according to the internet.
<ennoausberlin>Found it in glibc-bootstrap but ran into another error :(
<loopback___>how to set the mtu for an interface via /etc/config.scm?
<tschilp>unfroq: happy to help, I hope you can upgrade to the latest version again from there on (at this stage you are at the version from the two weeks ago as you said) after a pull (the 'mismatch' comes naturally after the pull and 'equalizes' after the reconfigure)!