IRC channel logs


back to list of logs

***apteryx_ is now known as apteryx
<nckx>sneek: later tell ebrasca: htop works just fine on Guix System, but your terminal emulator or fonts might be the problem.
<sneek>Will do.
<nckx>sneek: botsnack and don't ever leave me again.
<apteryx>sneek: where had you been?
<nckx>Look at that face. That's a face that's seen things, man.
<apteryx>eh, I hope we can meet one day so that they can tell me about it, over a beer.
<wednesday>yoo whats going on with this error I assume it's something to do with the color stuffs that happened a few days ago
<nckx>wednesday: Is this running from git?
<nckx>Have you made clean-go and remade?
<wednesday>I'll do that
<apteryx>eh, is down, that's why the user count of this channel lost ~50 users.
<brendyyn>anyone here use spacemacs with holy-mode for hacking guile or lisp?
<brendyyn> was cracked, maybe twice
<apteryx>I wonder if that means that IRC passwords of users of its IRC bridge got their password stolen, I hope not.
<nckx>wednesday: Wurked it?
<wednesday>nckx: still building after a make clean heh
*nckx builds their 13th kernel of the night.
<wednesday>Why do you build so many kernels ha
<nckx>I don't know what else to do with this fancy new CPU.
<nckx>while :; do build kernel; done
<wednesday>why not do it recursively in guile =^)
*nckx was just going to mumble something about tail-call optimising that sucker.
<brendyyn>Has anyone thought about how guix's principles might be extended towards managing configuration files in ones home directory? dotfiles remain something of a wild west
<wednesday>Well guix sort of does manage config files, just only really system ones heh
<wednesday>nckx: yea my stuff works fine now heh
<nckx>Cool in da pool.
<wednesday>well I gotta close my stuff down for a while, so catch ya later
*nckx → 😴
<wednesday>off to sleep for me now, cya
<brendyyn>weird, when using inherit, the source field which refers to `name` says causes the error Ubound variable: name. I have to add (name "tinc") to the package to make it work, even though it should be inherited?
<salotz>hello, anyone awake?
<vagrantc>given the span of timezones, and variable sleeping habits, i suspect so
<salotz>well unless your a bot...
<salotz>I got guix installed and followed the locale instructions but I am still getting that annoying message everytime I do something
<salotz>was wondering if there were any more tricks
<salotz>I have the environment line in the systemd unit file too
<brendyyn>It's a curse that one
<brendyyn>You have to join the Church of Emacs before it will go away
<salotz>I already drank that koolaid
<salotz>how come I'm not saved???
<brendyyn>did you reboot?
<Blackbeard[m]>hello guix :)
<Blackbeard[m]>I am trying to update this package
<Blackbeard[m]>so I got the code and put it on a file with all the #:use-module
<salotz>did you see the matrix hack?
<Blackbeard[m]>salotz I did not
<salotz>ya somebody got into their build servers or something because of ssh tunneling misuse or something
<Blackbeard[m]>the first problem I find is that it says
<salotz>just like yesterday
<Blackbeard[m]>error: emacs-a: unbound variable
<brendyyn>so there really was two separate attacks?
<Blackbeard[m]>salotz: ah I know what you mean, I am not worried about it .)
<Blackbeard[m]>so I have this
<salotz>just spreading the word, I am a matrix fan myself
<Blackbeard[m]>(propagated-inputs `(("a" ,emacs-a)
<Blackbeard[m]>but I get the error
<Blackbeard[m]>error: emacs-a: unbound variable
<Blackbeard[m]>what might be happening??
<brendyyn>why should emacs-a exist?
<atw>I am also curious about what package you mean by emacs-a
<Blackbeard[m]>well it is a package
<Blackbeard[m]>if you do guix package -s emacs-a it is there
<Blackbeard[m]>and it is a dependency :/
<Blackbeard[m]>but if I remove it then it happens with the next dependency
<Blackbeard[m]>what can I be doing wrong?
<Blackbeard[m]>here is my file:
<Blackbeard[m]>I just changed the commit, version and hash
<Blackbeard[m]>i didn't add emacs-a, it was already there
<Blackbeard[m]>I just want to update to a new version :)
<atw>when you load this file, where are you putting it? Are you putting it, like, on your GUIX_PACKAGE_PATH? Or is your paste an abbreviated version of the entirety of (gnu packages emacs-xyz) and you're making this change inside your checkout of guix from the git repo?
<vagrantc_>emacs-a has been there since before january, at least
<atw>my bad, should have looked for it
<vagrantc_>was added in d4f7b7e127e2392db93afc7252d0758c22361f06 ... ~November 2018
<atw>anyway, the reason I'm asking the questions above is because I suspect that the definition that Blackbeard[m] wrote is unable to find (gnu packages emacs-xyz)
***jonsger1 is now known as jonsger
***ng0_ is now known as ng0
<Blackbeard[m]>atw: hi, sorry I was out
<Blackbeard[m]>atw: sounds like a good question
<Blackbeard[m]>I wrote the file
<Blackbeard[m]>and I run the terminal
<Blackbeard[m]>guix package -f emacs-matrix.scm
<Blackbeard[m]>and it gives me a backtrace
<brendyyn>Blackbeard[m]: in inputs you put the symbol that refers to a package, not its name
<brendyyn>its possible for them to be different
<Blackbeard[m]>brendyyn: ok, so is there something wrong? :/
<Blackbeard[m]>I am using the definition that already existed
<brendyyn>Blackbeard[m]: it looks like your module has the same name as a module that already exists in guix
<atw>guessing here, but what if instead of (define-module (gnu packages emacs-xyz ) ...) you do (use-modules ((guix licenses) #:prefix license:) (guix packages) ...), ie don't define a module? I am thinking that maybe your use of define-module prevents the original (gnu packages emacs-xyz) from being found, and that's where emacs-a and emacs-anaphora are defined
<brendyyn>and emacs-a was in taht
<Blackbeard[m]>this is the backtrace
<brendyyn>but your file does not contain that. did you delete it?
<atw>brendyyn: I like the way you think ☺
<Blackbeard[m]>brendyyn: ahh sorry
<Blackbeard[m]>I am just learning, didn't know that
<brendyyn>thats fine
<Blackbeard[m]>this would be my first package if I do it :)
<Blackbeard[m]>ok I changed it and it still fails :/
<brendyyn>how did you change ti
<Blackbeard[m]>i replaced the names with blackbeard-emacs
<Blackbeard[m]>let me show you
<Blackbeard[m]>(define-module (blackbeard-emacs)
<atw>same error after changing to use (use-modules ...)? I think I know what it might be: make sure (gnu packages emacs-xyz) is in the (use-modules ...)
<Blackbeard[m]>(define-public blackbeard-emacs
<Blackbeard[m]>atw: ah... sorry :/
<brendyyn>and then you need #:use-modules (gnu packages emacs-xyz)
<atw>^ yup!
<Blackbeard[m]>it will work now :D :D :D
<Blackbeard[m]>it is building
<Blackbeard[m]>seems like it was installed
<atw>does it work?
<Blackbeard[m]>seems like it was installed and it kinda works
<Blackbeard[m]>but I think I am missing some dependency
<Blackbeard[m]>a new one i think
<Blackbeard[m]>ohh no it is parsing
<Blackbeard[m]>I will give it a couple minutes to see if it syncs with the matrix server
<Blackbeard[m]>matrix can be slow after all
<Blackbeard[m]>it is working and it is installed \ø/
<Blackbeard[m]>emacs-matrix-client 0.0.0-4.0f561ef out /gnu/store/jcmjy45kbdwd8lvd66iigjk2z6zcs9j5-emacs-matrix-client-0.0.0-4.0f561ef
<atw>excellent! and if you want to contribute upstream, there's a section of the manual that I found helpful
<Blackbeard[m]>atw: yes I want to contribute
<Blackbeard[m]>although it hasn't work yet
<Blackbeard[m]>let me try with regular emacs instead of emacsclient
***jonsger1 is now known as jonsger
<Blackbeard[m]>it keeps having the same problem
<Blackbeard[m]>I wonder if I am missing some dependency
<puoxond>Hello Guix!
<pkill9>does anyone have an example of modifying the kernel to add realtime support?
<pkill9>looking at the wiki it only requires a config change, but I thought it also needs patches
<ArneBab>sneek later tell civodul: I have a hard time interpreting the strace output. Anything specific I should look for?
<sneek>Got it.
<ArneBab>sneek later tell civodul: (regarding java failing to start)
<ArneBab>sneek later tell civodul: a guess I would have is that java might not find fontconfig, but that’s installed.
<ArneBab>sneek later tell civodul: ok, yes, that’s it. adjusting LD_LIBRARY_PATH works, but that’s more a bad hack than a solution: LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/run/current-system/profile/lib/:~/.guix-profile/lib/:~/.guix-profile/lib/nss/:~/.guix-profile/lib/lib/:/gnu/store/69x60a1pn0mf5jv68al8awjfkyp1miwi-gcc-8.3.0-lib/lib/:. java -jar new_installer_offline.jar
<sneek>Will do.
<ArneBab>sneek: later tell civodul: to reproduce: wget '' -O new_installer_offline.jar
<sneek>Will do.
<ArneBab>sneek: later tell civodul: to reproduce: wget '' -O new_installer_offline.jar; java -jar new_installer_offline.jar
<sneek>Will do.
<roptat>what is the best way to test the installer?
<roptat>you mean guix system vm ? what's the right config file to use?
<pkill9>i don't know
<pkill9>maybe generate the image using `guix system vm-image` then mount it in a VM and install that way?
<pkill9>as for the config file, the config file for the installer image is somewhere
<pkill9>im not much help, sorry
<puoxond>I've noticed that there was a recent change to the sddm service, so I gave it a try. Unfortunately I still get a black screen.
<puoxond>This is the config I passed to `guix system vm-image':
<wednesday>Damn puoxond is gone, wanted his X log
***buffet_ is now known as buffet
<vixus>hmm, it looks like the USB I made with guix system disk-image won't boot
<vixus>it has 2 partitions, one ext4 flagged boot and one fat16 flagged esp
<vixus>ah, can't boot it in qemu either so something worse must be happening
<joshuaBPMan>hello, I'm trying to figure out why my custom channel isn't working...I've got the file jmacs.scm working. ie: guix package -f jmacs.scm works. But when I try to add that git repo to a channel and do a guix pull I get this error in the build log:
<joshuaBPMan>(exception misc-error (value #f) (value "~A ~S") (value ("no code for module" (jmacs))) (value #f))
<geostarling>hi, I'm trying to package some rust crates and I've stumbled on some issues with the `cargo-build-system`.
<geostarling>Are there any example packages that show proper use of cargo-build-system? I haven't found any in main guix repo.
<joshuaBPMan>geostarling: Have you tried importing ? guix import
<geostarling>joshuaBPMan: I've completely forgot about importing, thank you!
***benny is now known as Guest30717
<rendaw>Hello! I'm getting "wrong number of arguments for action 'init'" when I do "guix system init -e '(my-func arg1 arg2)' /mnt" - is there something obvious I'm doing wrong? I googled around but the only result I found was logs from 2 years ago that are gone now
<rendaw>Also, the docs aren't specific in this regard - will the filesystem created by "guix system init" be a bootable system? Like disk-image but on a mounted empty filesystem? The other way I could interpret the docs for that command are that "init" copies all the files required to build the system (source files, guix binaries, etc) into a directory for backup purposes
<kmicu>Hi rendaw. why do you pass -e flag to system init?
<rendaw>To pass arguments into the build
<rendaw>(drive uuids, secrets I don't want to commit to the system config, etc)
<rendaw>I suppose I image it's the same reason one would want to pass -e to system disk-image or system-vm or whatever
<joshuaBPMan>rendaw: guix system init my-os-config.scm /mnt
<rendaw>Thanks @joshuaBPMan - can you elaborate on that? It looks like you aren't using -e
<joshuaBPMan>rendaw: It's from the manual.
<joshuaBPMan>It's the easiest way to do ti.
<rendaw>Yes, thank you -- I'm specifically interested in using -e though
<joshuaBPMan>May I ask why you want to use -e ?
<joshuaBPMan>Isn't it easier to just write your os in a config file?
<kmicu>rendaw: could you move -e before init and try again?
<rendaw>kmicu: then it says there aren't enough arguments (since I'm not specifying the config file)
<rendaw>joshuaBPMan: kmicu just asked the same question, but specifically to pass arguments into the build such as drive uuids, secrets I don't want to commit to the system config, etc -- I imagine it's the same reason one would want to pass -e to system disk-image
<kmicu>rendaw: then add a dummy config file. That part is not optional.
<kmicu>(Also keep in mind that in Nix/Guix all secrets end up in a readable store so -e can only protect against secrets in config file only, nothing more.)
<rendaw>The concern is access to the config file and not the end system
<rendaw>Hmm, well I won't say I'm excited about putting secrets in temp files to build but I suppose I'll try that path
<rendaw>Thanks kmicu for the help!
<samplet>rendew: It looks like “init” requires exactly two arguments: the OS configuration file and the target path. I guess it doesn’t support “-e” for some reason.
<samplet>(That’s what the source code looks like, anyway.)
<rendaw>Awesome, thanks stamplet! I actually spent 10m looking at the source code before coming here but my lisp isn't quite at that level yet
<kmicu>Ah, now I see the issue
<samplet>Near the bottom of “guix/scripts/system.scm” it has a “case” form that checks if the action is “init” and then makes sure the argument count is exactly two.
<samplet>It would be pretty easy to fix, but I recommend contacting the mailing list before doing anything, in case there is a good reason for it.
<samplet>(If you’re interested in contributing. If not, that’s perfectly fine!)
<rendaw>I'm interested in contributing! I'm not confident in my abilities to fix the lisp code yet though :)
<rendaw>I have lots of questions like this though and I'm not sure at what point they'll just be annoying
<rendaw>Er, fix the lisp code -> submit my changes to the lisp code
<samplet>To be honest, people who work on Guix are generally friendly but busy, so feel free to ask questions. No one will be annoyed, but you might not get an answer right away.
<kmicu>rendaw: ‘-e’ contributor didn’t handle init case, intention was ‘(guix-system): Allow commands taking a file as an argument to use an expression instead.’ but there is a check for init that doesn’t allow that currently.
<joshuaBPMan>rendaw: why are drive uuids considered a secret? What could a malicious attacker do with that information?
<kmicu>rendaw: more at 5a72ddf176d53a7f4df922985d9d7fd4cfa160f5
<rendaw>kmicu: So I guess it sounds unintentional?
<rendaw>joshuaBPMan: the drive uuids were separate from the secrets - secrets are like api keys for external services, etc
<samplet>rendaw: The main issue seems to be that the “process-action” procedure assumes the target to be the second argument. If there were no configuration file, it would have to be the first.
<samplet>I’m not certain, but I bet with a little tweak to “process-action”, you could remove the check and everything would work fine.
<rendaw>I suppose none of the other actions have a target parameter
<kmicu>rendaw: yep, you are basically the first user using init with -e. Congrats xD
<rendaw>Well I'm not completely sure I want init yet... I'm setting up images myself with loopbacks and then dding to disk so disk-image might work, but I'm not sure how that works with the filesystem section (how to target the drives, how it knows to set up a fat32 /boot partition, etc)
<rendaw>Er, disk-image is closer to what I want but I wasn't sure how to target the filesystems it generates
<rendaw>and the system build script I'm migrating from was already setting up loopback images to build the system
*kmicu wonders whether Mathieu Othacehe (--expression contributor) is on IRC.
<rvgn>Hello Guix!
<rvgn>What is the command to view packages installed as "system", that is packages in system profile?
<nckx>rvgn: guix package --list-installed -p /var/guix/profiles/system/profile
<nckx>(/run/current-system/profile works too, if that's what you want.)
<nckx> /var/guix/... might be newer.
<rvgn>nckx Thank you!