IRC channel logs


back to list of logs

***alMalsamo is now known as lumberjack123
<jonsger> "An experimental and incomplete implementation of Meson in C++, for solving Meson's bootstrapping issue." <-- sounds not bad :)
<lilyp>meson... bootstrap issues?
<lilyp>please tell me that's not a real thing
<singpolyma>lilyp: why not? I assume most build systems use themselves
<lilyp>Well, at the very least meson++ uses meson, so...
<akonai>singpolyma: usually not
<akonai>i believe that for example gnu make uses a shell script as a build system
<devmsv>djeis: nothing, I have pasted the examples and when trying to use the new package definition I get the following error:
<devmsv>I have put the package definitions before operating system declaration
<lilyp>you are probably referring to your package by name instead of just passing on the variable
<devmsv>Yes, I'm adding system packages with 'specification->package' procedure. Any way I can add this package as variable to that list?
<lilyp>devmsv: it's a scheme list, you can use (cons your-package rest-of-the-list) to generate a new one
<devmsv>It seems something is not working on the package customization:
<devmsv>I will continuous tomorrow as its really late . thanks everyone for the help and tips
***alMalsamo is now known as lumberjack123
<xelxebar>Hey, Guix!
<FlaminWalrus>Gentoo user that's hopped to the Guix System here; I
<FlaminWalrus>I'm used to using startx from a tty, or configuring the shell to automatically run startx upon console login from a particular tty; how to set up the X server to work without a display manager is not obvious/documented. %base-packages with xorg-service-type added didn't seem to make startx or xinit work at all. Am I missing the right command, or is there a configuration change I need to make?
<wargreymon2022>FlaminWalrus: it has been discussed on help-guix mailing list:
<jackhill>wooo, I'm exited about the racket updated and related improvements! Thanks lilyp (and the other folks whose nicks I don't know)!
<xelxebar>FlaminWalrus: FWIW, I am successfully running Guix without a login manager and just using startx.
<xelxebar>FlaminWalrus: It's been a long time since I looked at how I got this working, but IIRC, creating an appropirate .xserverrc in your $HOME should Just Work.
<xelxebar>Here's mine:
<xelxebar>Probably can just crib that as-is.
<xelxebar>I'm an old Gentoo user, myself :D
<xelxebar>Your operating-system configuration will need to use %base-services instead of %desktop-services, too.
<xelxebar>But then you'll probably want to manually add back in several quality-of-life things. Ping me if you want and I can share my services definition.
<Aurora_v_kosmose>Thank you Pierre Neidhardt for your work in debugging some Lisp asdf inputs.
<Aurora_v_kosmose>Turns out your fix works on essentially all fukamachi packages that have the "*-test" not found issue.
<Aurora_v_kosmose>For those wondering what I'm going on about: can be fixed by replicating the setup for this: which git-blame says Neidhardt figured out.
<Aurora_v_kosmose>This is mostly in case I forget to do a sweep, cleanup & fix a bunch and submit relevant patches.
<Aurora_v_kosmose>Yes, it's just loaded input ordering.
<Lumine>Good morning
<FlaminWalrus>xelxebar: mind sharing that service definiton? I'd like to be sure any errors justify a rigourous debugging, and aren't just stupid misconfiguration of config.scm :)
<roptat>I'm surprised how well I can build the core of intellij-sdk now :)
<roptat>reading xml files to figure out which modules I need to build works great
*ams-hash[m] uploaded an image: (99KiB) < >
<ams-hash[m]>Can anyone help with this error. I am unable to boot the iso
<lilyp>I don't think any of these errors ought to be boot-inhibiting ones
<ams-hash[m]>Well I've tried 2 different isos
<ams-hash[m]>Idk what the issue is
<lilyp>have you tried the minimal setup? the gnome one typically doesn't start if your graphics card acts up
<roptat>is this the iso you're booting, or the system that was installed from the iso?
*roptat is building a new chain of kotlins
<roptat>hopefully, I can build up to kotlin 0.6.470, from june 2015 ^^'
<lilyp>nice, could that then be used to build gradle from june 2015?
<roptat>most likely
<roptat>I want to see if I can go further, I was blocked at some 0.11 release, but managed to build the missing dependency
<roptat>although, at some point I need xmlbeans which I haven't bootstrapped yet
<roptat>I'll try to go as far as possible without building gradle
<lilyp>makes sense, one can of worms at a time
<roptat>gradle might not be too hard, but it wouldn't be useful to build it too soon
<roptat>I prefer to build it when it's needed
<ams-hash[m]>I get an error (file-system (device "/dev/nvme0n1p1") (mount-point "/boot/efi") (type "vfat")): invalid field specifier
<ams-hash[m]>Nvm fixed
<rndd>hi guix
<rndd>cannot access
***stampirl_ is now known as stampirl
<Lumine>I can
<rndd>Lumine: hi, dont work for me (ERR_CONNECTION_TIMED_OUT)
<rndd>Well, long story short - i am currently in Moscow. And site is accessible with vpn. So i came to ask is your site blocked for russians? Cause this problem is not about rusian isp i use (at least right now).
<ams-hash[m]>How long does it usually take to build, it's been in the checks phase for a while
<ams-hash[m]>Right as i say that it finishes the check phase xd
<roptat>ams-hash[m], it depends on the package it's building :)
<jlicht>guix-installed programs that have a gsettings-desktop-schemas entry added to XDG_DATA_DIRS in their binary wrapper seem to have issues running graphical host-distro programs (as those now try to start using guix's glibc, instead of the host's)
<jlicht>no, actually, it's the GIO_EXTRA_MODULES that gives trouble
<gnoo>there was a recent blog post about using guix with tex and latex with profiles but i can't find it anymore
<gnoo>does anyone have a link to it?
<Haider>I have noticed the new glad package in guix which is quite useful in my case. The only problem I have is that my c++ file cant actually find glad.
<SeerLite[m]>Hi! Currently vim-full tries to add a flag to #:make-flags with substitute-keyword-arguments on ones inherited from vim. But since vim has no make-flags, it doesn't get "modified" and no flag is added.
<gnoo>found it in the manual
<SeerLite[m]>Should substitute-keyword-arguments maybe create the argument if it doesn't exist?
<nckx>rndd: Some IP ranges are blocked by the institution that hosts the servers. Some of these abusive IPs are Russian, but not Russia as a whole I believe.
<phf-1>Hello Guix! So, it seems that I cannot get cuirass to show worker status... Here are the configs. cuirass: and worker: what's the problem?
<nckx>I can imagine that abusive traffic from Russia's through the roof right now so the effect might be the same.
<xelxebar>FlaminWalrus: Sure. Here's my whole configuration:
<xelxebar>You'll probably want to reference gnu/services/desktop.scm to see what some of those actually are.
<phf-1>On the server, this is the error I have on the logs:
<phf-1>So, the logs on the worker server is: and the configuration is:
<phf-1>where "" is the cuirass server ip
<pkal>What is the easiest way to modify the native-inputs of a package when inheriting?
<phf-1>pkal, there is a modify-inputs helper used in gnu/packages/ssh.scm
<phf-1>pkal, chances are that a similar helper exists for native-inputs
<phf-1>pkal, Well, there is that `(modify-inputs (package-native-inputs guile-ssh)' in `gnu/packages/ssh.scm' so, that's how it should be done I guess.
<phf-1>sneek later tell lfaim looking at the tests in `guix-cuirass/tests/remote.scm' we learn that the `(server "...")' field should be filled witht the ip and the port, not just the ip. I guess an example would help in the documentation like `(server "")'
<sneek>Got it.
<phf-1>sneek botsnack
<SeerLite[m]>pkal, phf-1: `info guix modify-inputs`
<pkal>SeerLite[m]: phf-1: Ok, from reading the manual I had the impression that that macro couldn't be used there.
<alethkit>raghavgururajan: Just remembered to up[load it now:
<ams-hash[m]>What are .drv files
<ams-hash[m]>How can I use the normal Linux kernel. Not libre
<ams-hash[m]>I cant boot thr libre
<singpolyma>ams-hash[m]: you will need to find a channel that supports it, such as nonguix
<singpolyma>Discussion of it is off topic here but other channels have their own discussion venues
<ams-hash[m]>I didn't know
<Lumine>Hello, I'm trying to figure out why system wide declaration of keyboard layout defaults to the US layout under i3 and xmonad, they ignore the configuration
<Lumine>I tried using xkbcli to change the layout but it does not find mine, and the documentation for xkvcli was very sparse
<nckx>ams-hash[m]: No need to apologise either :-)
<the_tubular>How hard is it to make a services file ?
<the_tubular>I'd be interrested in one for rtorrent, for example
<lilyp>the_tubular: it's mostly boring declaration work
<the_tubular>I might try my hand at it, but I'll probably run into a few errors ^^'
<ams-hash[m]>After I've made a change to my cofnig file how do i rebuild the system
<Haider>sudo guix system reconfigure /path/to/config.scm
<roptat>mh, kotlin is giving me some troubles :/
<roptat>I get a methodnotfoundexception, but the method both in the bootstrap and in the source code I'm compiling...
<helaoban>Hey guix, I'm looking for clarificatino on the use of `guix home`, the docs say that the packages listed in the 'packages' field of 'home-environment' are installed into the user's profile.
<roptat>the first version that has this method is 0.11.153, and it builds fine, but then it can't rebuild itself...
<Lumine>Okay, didn't realise setxkbmap was in the GNU packages, just not on the system by default, so my problem is not a problem anymore
<helaoban>But from what I gather the "user's profile" that is referred to here is the profile saved in ~/.guix-home/profile, and is distinct from ~/.guix-profile
<helaoban>So I need to make sure that ~/.guix-home/profile/bin is in my system to pick up the binaries, and this is not managed for me by `guix home`, is that correct?
<helaoban>*system path
<maximed>sneek: later tell apteryx: I sent a v5 of the patch series for reducing the number of false positives of 'check-tests-true' (+ some #:tests? fixes): (only difference with the v4 is a rebase)
<maximed>sneek: botsnack
<helaoban>nevermind! I am an idiot, literally the next page in the documentation addresses this!
<Aurora_v_kosmose>So I have a series of commits adding a bunch of dependencies for a package & a package. How would you recommend I submit those as patches?
<Aurora_v_kosmose>Individual patch series "Adding XYZ" or one "Adding XYZ & dependencies [n/m]"?
<lilyp>Aurora_v_kosmose: individual patches please
<unmatched-paren>what do i do if a git clone for a package returns an error about http/0.9?
<unmatched-paren>error: Received HTTP/0.9 when not allowed (curl_result = 1, http_code = 0, sha1 = 0d1a5ca50d2ecd81e145a83f2661516d840c59f9)
<unmatched-paren>i guess there's some flag to enable HTTP/0.9...
<Aurora_v_kosmose>Does curl_result map onto curl's exit codes?
<unmatched-paren>no idea
<Aurora_v_kosmose>> 1 Unsupported protocol. This build of curl has no support for this protocol.
<unmatched-paren>hm, okay, that makes sense
<Aurora_v_kosmose>If it does map, then that'd make sense. If it doesn't map, then back to step 1.
<unmatched-paren>this is the repo i'm trying to clone:
<unmatched-paren>git:// works, but bizarrely the resulting repo is completely empty
<unmatched-paren>weirder, when i `git clone` the above git:// link, it works fine, but not when guix does it
<Aurora_v_kosmose>I don't think it maps. With curl I get an HTTP/1.1 connection with TLSv1.3
<unmatched-paren>yeah, that's really strange: `curl --include | less` sends back `HTTP/1.1 200 OK
<Aurora_v_kosmose>Using dumb-http. Apparently it doesn't support shallow cloning.
<unmatched-paren>yeah, but it looks like the git clone is retried full if it fails to do a shallow clone
<unmatched-paren>in the guix log: Failed to do a shallow fetch; retrying a full fetch...
<Aurora_v_kosmose>Oh interesting. It fails to fetch entirely just with git clone -vvv
<Aurora_v_kosmose>Same error you received.
<Aurora_v_kosmose>> error: Received HTTP/0.9 when not allowed\n\t(curl_result = 1, http_code = 0, sha1 = f95661aa437f3bcdaf88a1c64dc408ee79fbe436)
<unmatched-paren>well, well:
<unmatched-paren>i am not alone, apparently
<Aurora_v_kosmose>There's something wrong with that git endpoint.
<unmatched-paren>"I copy pasted so no typo, but I had a weird http 0.9 error."
<unmatched-paren>the two github mirrors i can find are unmaintained and don't have the latest commits
<Aurora_v_kosmose>git:// This one works. However you can say good-bye to transport security.
<unmatched-paren>ah, but it doesn't work with guix
<unmatched-paren>it clones, but the checkout dir is empty
<Aurora_v_kosmose>unmatched-paren: Hm, would Guix have chosen not to implement git:// support? Seems possible. In any case, I strongly recommend emailing the author. They're still active
<Aurora_v_kosmose>Or at least they were until last month.
<unmatched-paren>they have a list on, but no source mirror, apparently
<Aurora_v_kosmose>unmatched-paren: Check the commit log for email.
<unmatched-paren>while i wait for a response i'll try to maintain a mirror
<unmatched-paren>i've mirrored it; thanks Aurora_v_kosmose:
<unmatched-paren>(well, it's not exactly a mirror, since it's manually uploaded...)
<Aurora_v_kosmose>A mirror's a mirror)
<mekeor[m]>hi guix! :)
<nckx>Hi mekeor[m]!
<f1refly>when i try to execute ./bootstrap on my freshly cloned guix repo autoconf complains about ' error: possibly undefined macro: PKG_CHECK_MODULES'
<f1refly>are there additional steps to do before executing ./bootstrap?
<f1refly>I'm following the instructions from running-guix-before-it-is-installed to update a package from the reop
<jpoiret>f1refly: did you make sure to use `guix shell -D guix`?
<f1refly>no, I didn't know that was required
<f1refly>thank you
<jpoiret>you need guix's dependencies to build it
<f1refly>would I need to do that again befor invoking guix refresh --update?
<f1refly>for the package in question i mean
<f1refly>because my invokation fails with "guix refresh: error: mkstemp: Read-only file system"
<maximed>f1refly: "guix refresh -u" modifies the package definitions in place
<f1refly>yes, and i'm in my own checkout of the guix repo in my user dir, so that should work, right?
<maximed>as such, you can only run in from a git checkout (./pre-inst-env guix refresh --update) (with some exceptions, e.g. GUIX_PACKAGE_PATH & -L)
<maximed>f1refly: yes, just make sure to do the ./pre-inst-env
<maximed>Such that it tries to modify the package definitions in the git checkout and not the package definitions of the guix in the user profile
<maximed>f1refly: also, I don't think that `guix shell -D guix` is strictly required, but it can be very useful (and probably better-tested)
<f1refly>guix shell -D guix was required for ./bootstrap
<f1refly>to get the guix utils independent from the system
<maximed>f1refly, eer, interesting. Maybe try "make && ./pre-inst guix refresh --update alacritty"?
<f1refly>okay, making first
<nckx>It's not strictly required, but without it it's your own responsibility to manually install the compatible versions of each Guix dependency.
<florhizome[m]>hi guixers... (full message at
<f1refly>when submitting the package I'd send the patch to the patches mailing list, right?
<nckx>If it's a single patch, yes. Otherwise, also yes, but different.
<nckx>To wit:
<ryanprior[m]>I want to include various dotfiles and git repositories as part of my home configuration. I read the Guix home manual and don't see a way to include arbitrary dotfiles, directories, git repos - am I missing something? Do I need to write my own services for those things?
<civodul>ryanprior[m]: hi! currently you can do that by extending home-files-service-type
<civodul>which is really about dot files
<civodul>that might change in the future because it's a source of confusion
<ryanprior[m]>It would be handy to document it under home-services too, I read all the nodes there and home-files-service isn't mentioned
<ryanprior[m]>I looked in the repo for mentions of `home-files-service` and don't see much in the way of definition or examples either XX I do see in the tests for home a "simple-service" that extends home-files-service-type.
<ryanprior[m]>It uses "plain-file," I wonder if there's "plain-directory" or similar, or maybe some way to specifically declare a checkout of a git repo like you can in a package origin.
<civodul>yeah some Home things are still missing from the manual, in some cases because things are still in flux
<civodul>regarding Git checkouts, you can use an origin record
<civodul>that's a "file-like object"
<civodul>hmm doesn't work as expected
<civodul>jonsger: i was going to suggest that to you ↑
<Noisytoot>How can the gcc-objc package actually be used? gcc fails with "Objective-C compiler not installed on this system" if I try to compile any Objective-C code.
<f1refly>jesus christ rust crates sprawl like a plague
<f1refly>i feel like i'm currently defining packages for all of
<drakonis>ain't that a hoot