<monty>
sondr3: Those just happen to be the characters dropped. Otherwise there'd be too many for base32
<clever>
and i think they omited letters needed for certain words, so you cant spell bad words
<monty>
I don't get omitting u and t but not i or l though.
<monty>
Even v/w and m/n are more confusing.
<monty>
Ah, assumed it'd be optimized for least confusing.
<xel>
Hi, I am attempting to use NixOS to compile a Rust binary in NixOS. When I try to run it within a different OS (Ubuntu in this case), I get `No such file or directory` when it definitely is within the directory. I did verify that the binary is 64-bit. Could this be a problem with linking because of the non-standard filesystem structure of NixOS?
<monty>
Thankfully sass is a perfectly valid part of a hash : >
<monty>
Almost definitely.
<monty>
You need to provide the entire closure.
<clever>
xel: if you run `file` on the binary, youll see the path to the dynamic linker, and that must exist for it to run
<monty>
A static binary *might* be a way to get around that issue.
<monty>
Oh, you can just change the linker?
<xel>
Can you?
<clever>
monty: thats exactly what patchelf does, to make non-nix stuff work under nixos
CptCaptain has quit [Quit: WeeChat 2.6]
<clever>
and you could use patchelf to do the reverse, but you may run into compatability problems
<clever>
your basically throwing away half the benefits of nix when you do it
<xel>
so what would be the best solution for this? worst-case I could just compile it in a VM or container
<clever>
just tell nix to build a static executable
<clever>
then it works anywhere, and your still using the libc defined in nixpkgs
<xel>
how would I do that for Rust?
<clever>
xel: use pkgs.pkgsStatic instead of plain pkgs
<nh2>
clever: hmm, it seems like no matter how I override `libXtst` (top-level in `pkgs`, in `X11` or `xorg`), this reference to it in my statically linked GTK app https://github.com/nh2/static-haskell-nix/pull/67/files#diff-5a873d92c00133aa8d3f0a425a93aac2R839 (line 839 on the right side) does not refer to it. I can put arbitrary garbage into the my overridden libXtsts, they are not evaluated
<clever>
that will make all binaries static
t58 has quit [Quit: Leaving]
<xel>
alright, I'll give that a try
<ddima>
monty: congrats for fixing ;)
<xel>
thank you :)
<monty>
Didn't actually come up with any of the steps but I typed them over flawlessly : )
<monty>
nn, all
<clever>
nh2: due to the // going on, you might need to override both?
monty has left #nixos [#nixos]
vandenoever has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @globin opened pull request #74089 → nixos/tests: fail on failing subtests → https://git.io/JePmv
bvdw has quit [Read error: Connection reset by peer]
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
bvdw has joined #nixos
<nh2>
xel: if you build all your rust dependencies statically into the binary, so that it only depends on glibc at the end, then the patchelf approach should work (but make sure that you use an older-or-equals glibc in NixOS than in your Ubuntu, because glibc guarantees only that you can run stuff compiled against older versions on newer oens, not the other way around). I'd say try that first and report if it works.
<nh2>
xel: The alternative is fully static executables, as clever said; I have a project `static-haskell-nix` that works on this for Haskell, and right now I'm working on making statically linked Rust executables that link in GTK (https://github.com/nh2/static-haskell-nix/pull/67)
<clever>
xel: there is also nix-bundle, which will ship a nix closure, and automatically chroot for you
<nh2>
clever: locally I have overridden all 3 (top-level, `X11` and `xorg`) -- or do you mean something else?
<clever>
nh2: yeah, thats what i would have done as well
ng0 has quit [Quit: Alexa, when is the end of world?]
<clever>
nh2: sometimes, its simpler to edit a fork of nixpkgs, and once its working, transition to overrides
<xel>
nh2: I'll try those as well, I did just try using only patchelf to change the interpreter for it, though that does not seem to have fixed it so I will statically link most likely
<clever>
xel: you would also have to restore the RPATH to /lib:/usr/lib, and instruct the user to install the right libs
leungbk has quit [Remote host closed the connection]
<nh2>
clever: when I use `final.callPackage` as I do there, should that give it the `libXtst` from the same scope?
<clever>
nh2: yeah, that should be a good work-around
vandenoever has quit [Quit: Konversation terminated!]
<nh2>
xel: for static linking, keep in mind that for that you need `.a` files for all your system dependencies (which is where 60% of the work in static-haskell-nix goes into), and you will need to use `musl` instead of glibc (which is what `pkgsStatic` does)
<nh2>
clever: but that's what I'm doing, and it does not work
aveltras has quit [Quit: Connection closed for inactivity]
sondr3[m]1 has joined #nixos
<clever>
nh2: can you pastebin the entire error when it fails?
<nh2>
xel: if you need examples of adding `.a` files to your deps, static-haskell-nix has a lot of them
fenedor has quit [Read error: Connection reset by peer]
<clever>
nh2: ive also used glibc static, but i dont think its officially supported
<clever>
nh2: and if you find the libXTst in the drv for tutorial, and check its lib dir, what files are present?
<nh2>
clever: but the problem here is not in the error message; the key thing is that it doesn't use any of my 3 overrides. For example, if I build `nix-build survey/default.nix -A pkgsWithArchiveFiles.libXtst`, it builds my changes, but if I build `pkgsWithArchiveFiles.meson-tutorial-gtk`, then it does not build them, so it's using a non-overridden one
<sondr3>
p-h[m]: could you maybe do it with `nix-instantiate --parse`?
<nh2>
clever: interestingly `pkgsWithArchiveFiles.xorg.libXtst` also uses the non-overridden one, so it looks like my `xorg.override` is not doing anything
<p-h[m]>
thx
<clever>
nh2: try using pkgsWithArchiveFiles.extend (overlay) instead of overlays=[overlay] ?
<nh2>
clever: OK next question: Is there a good way to add `dontDisableStatic = true` to *all* the derivations in `xorg.*`, by mapping over the attrs without mapping over those that aren't valid (like `overrideScope` itself)?
<clever>
nh2: using newScope and makeScope together, you can create a new set of packages, that automatically gets its own scoped callPackage and overrideScope'
<clever>
i use that pattern any time i'm making a file with more then 2 expr
<nh2>
clever: that is useful, I have bookmarked it
<clever>
nh2: normally, daedalus depends on wallet, to build installers
<clever>
nh2: but for testing purposes, this makes wallet depend backwards on daedalus, and overrideScope itself in
selfsymmetric-mu has joined #nixos
<clever>
nh2: so you can now build daedalus installers, from the wallet repo, without having to commit wallet, push wallet, then update the pin in daedalus
<nh2>
clever: that makes sense
<clever>
the kexec stuff is using it a bit differently, i'm just abusing it as a fixed-point thing, so i can change a number of variables
<clever>
it could have also just been a single let block with a bunch of cross-referencing functions
Khetzal has quit [Remote host closed the connection]
Khetzal has joined #nixos
acarrico has joined #nixos
mexisme has joined #nixos
MmeQuignon has quit [Ping timeout: 240 seconds]
Rusty1 has quit [Remote host closed the connection]
Khetzal has quit [Quit: No Ping reply in 180 seconds.]
Khetzal has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JePmx
Rusty1 has joined #nixos
rople has joined #nixos
mmlb has joined #nixos
Khetzal has quit [Quit: No Ping reply in 180 seconds.]
Khetzal has joined #nixos
xelxebar_ has quit [Ping timeout: 260 seconds]
xelxebar has joined #nixos
silver has quit [Read error: Connection reset by peer]
Khetzal has quit [Read error: Connection reset by peer]
Khetzal has joined #nixos
sb0 has joined #nixos
mexisme has quit [Read error: Connection reset by peer]
Khetzal has quit [Remote host closed the connection]
Khetzal has joined #nixos
jluttine has joined #nixos
lunik1 has quit [Read error: Connection reset by peer]
lunik1 has joined #nixos
christianpoveda has joined #nixos
<christianpoveda>
Hi everyone, I'm trying to migrate my laptop to nixos and I'd like your opinion on a matter, I come from a rolling release distro and I wanted to use nixos-unstable for every package. Is that a good idea?
<clever>
christianpoveda: i also use nixos-unstable on all of my machines
<clever>
the rollback stuff makes it easy to undo things if the rolling-release does break things
<christianpoveda>
clever: How do you use that channel instead of stable?
<christianpoveda>
I mean is it there a way to tell nix to use unstable by default in my configuration.nix file?
Khetzal has quit [Remote host closed the connection]
<christianpoveda>
Oh so thats just pointing the nixos channel to nixos-unstable, right?
<clever>
yep
h0m1 has quit [Ping timeout: 245 seconds]
<christianpoveda>
great thank you :)
h0m1 has joined #nixos
Khetzal has joined #nixos
tsrt^ has quit []
xqcw has joined #nixos
<gchristensen>
is there something like squashfs but faster to make?
<clever>
gchristensen: i think you can change the block size with squashfs, which would make it compress faster (but not as much)
<srid>
How do I uninstall some scripts from my ~/.nix-profile/bin? Even `sudo rm -rf` doesn't work. Specifically trying to remove `~/.nix-profile/bin/hie-*` (because it is conflicting with home-manager install of the hie package)
<clever>
gchristensen: mksquashfs is also very good at hammering all cores, so you could forward it to a machine with lots of cores
<srid>
I *believe* I had run `nix-env -i` at some point.
<clever>
srid: you must uninstall the conflicting package with `nix-env -e name`
<clever>
srid: `nix-env -q` will list the package names
<srid>
Ah, thanks. Didn't know about -q
<xqcw>
How often should you update nixpkgs channel as a nixos user? Is it recommended to do it daily? It seems like it might be more intensive than apt update && apt upgrade, or am I wrong
<srid>
Not very often. When I feel like it. Usually twice a month.
<srid>
(If things break, you can always rollback using `nix-channel --rolback`)
<xqcw>
Okay, got it. Thanks
Khetzal has quit [Remote host closed the connection]
Khetzal has joined #nixos
nfip^ has joined #nixos
mexisme has joined #nixos
xqcw has quit [Quit: leaving]
evils has joined #nixos
<gchristensen>
clever: do you know if Ican tell `mount` to mount everything in an fstab, but at a different root? (like: mnt/
<gchristensen>
(like /mnt-root/)
justanotheruser has joined #nixos
rople has quit [Ping timeout: 276 seconds]
buckley31002 has quit [Quit: Ping timeout (120 seconds)]
felixfoertsch23 has joined #nixos
rople has joined #nixos
buckley31002 has joined #nixos
felixfoertsch has quit [Ping timeout: 250 seconds]
<nh2>
well great github is down
bastion-tester has joined #nixos
pie__ has quit [Ping timeout: 240 seconds]
Chiliparrot has joined #nixos
rople has quit [Ping timeout: 240 seconds]
bastion-tester has quit [Quit: ZNC 1.7.4 - https://znc.in]
bastion-tester has joined #nixos
kolbycrouch has joined #nixos
rople has joined #nixos
<kolbycrouch>
anyone know how to install nixos-rebuild/nixos-install on non-nix hosts?
<srid>
kolbycrouch: what do you mean by 'non-nix hosts'?
<kolbycrouch>
srid: I meant non-nixos host. I have an armv7 sbc with a ubuntu image on the microSD. I figured that nixos-rebuild would turn in into a proper nixos system without having to write an expression to create an image
<infinisil>
The third point in that section
<samueldr>
just so you know, cache.nixos.org does not host armv7 binaries, so you'll be building the whole world
alex_giusi_tiri has quit [Quit: Leaving.]
<kolbycrouch>
infinisil: Thanks for the info! it looks like that might be what I want.
<kolbycrouch>
samueldr: Yes I know, it's not too much of a problem for me to build locally.
<samueldr>
great :) just saying in case
<lopsided98>
I don't know if anyone has managed to lustrate an ARM device
<kolbycrouch>
samueldr: It was a fair warning. Building a system on my rpi1 took nearly 2 weeks :D . My odroid-xu4 ( what I'm using this for ) is powerful enough to build the minimal system in reasonable time though.
shibboleth has quit [Quit: shibboleth]
christianpoveda has quit [Remote host closed the connection]
<kandinski>
how can I make my currently booted system-xxx-link profile the default boot before deleting all subsequent ones?
<kandinski>
I did `/run/current-system/bin/switch-to-configuration boot` as the manual shows, but when I list /nix/var/nix/profiles/ it still shows the `system -> system-{latest-non-working}-link` symlink
<{^_^}>
[nixpkgs] @anmonteiro opened pull request #74094 → Kubernetes cfssl: Allow configuring cfssl API server SANs → https://git.io/JeP30
domogled has joined #nixos
iqubic has quit [Quit: ERC (IRC client for Emacs 27.0.50)]
<kandinski>
I"m really confused here. I'm running `# /run/current-system/bin/switch-to-configuration boot` as it says in https://nixos.org/nixos/manual/index.html#sec-rollback, and on reboot it defaults to the expected profile, but the /nix/var/nix/system symlink still points to the latest, not the new default. Is this the expected behaviour?
<Ariakenom>
specifying outputs in services.xserver.xrandrHeads gives the right order to them, but wrong resolution (1440->1080). Specifying primary in the attribute freezes at login. Any tips?
<kandinski>
LnL sorry, but I don't understand. I'm running generation 109 now, which is the one that grub boots by default, but here's the output of ls -la /nix/var/nix/profiles/*system*: https://termbin.com/etrh
<kandinski>
shouldn't the symlink /nix/var/nix/profiles/system also be pointing to system-113-link?
gagbo has quit [Ping timeout: 246 seconds]
octe has quit [Ping timeout: 240 seconds]
gagbo has joined #nixos
<LnL>
you only booted the generation, nixos-rebuild switch --switch-generation 109 would update the profile
<kandinski>
LnL, thanks
<LnL>
also note that any future switch/build will be based on your configuration.nix not what the configuration of the current generation was
aveltras has quit [Quit: Connection closed for inactivity]
<rsa>
um, i dunno why i cant delete debug output, i did a 'nix-store -r /nix/store/2ppxd1plf2lc6id2nbqk9218l6pzqh7q-openssl-1.1.1d-debug' and got a warning that it can be garbage collected because no roots, but i can't even gc (--delete) it now
<{^_^}>
[nixpkgs] @filalex77 opened pull request #74123 → nixosTests.rabbitmq: port to python → https://git.io/JePnF
<ij>
This command works. Why? % NIX_PATH= nix-env -iA nixos.vitetris
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Ariakenom has quit [Remote host closed the connection]
Ariakenom has joined #nixos
<ij>
I feel like it's not supposed to.
zupo has joined #nixos
<adisbladis>
ij: nix-env is channel aware
<adisbladis>
ij: Indeed it shouldn't. And the `nix` 2.0 (I guess technically 2.4 or something like that) replacement for `nix-env` will not be channel aware and behave in less surprising ways.
<ij>
I guess it's root channel aware as well, because only the root user has that channel, not the user user.
octe has joined #nixos
<adisbladis>
Indeed it is
<ij>
I tried finding the translation book for nix 2.0, but couldn't. The cheatsheet didn't have nix 2.0
<skrzyp>
I've put NixOS on X230 for… fourth time(?) thorough last 6 years. And I can clearly say it has grown up a lot.
<adisbladis>
ij: There is no nix 2.0 replacement for nix-env yet
<adisbladis>
It's coming
zupo has quit [Client Quit]
<ij>
alright, alright
<ij>
I guess I got to start following the developments
<adisbladis>
Iirc it's `nix profile` or something similar
<skrzyp>
But I still miss the "regular" CLI for adding packages, where you can search, install, update, remove and rebuild without a weird parameter invocations like `nix-env -f '<nixpkgs>' -qAP 'regexp'` for searching packages
<adisbladis>
skrzyp: That's what I just mentioned :)
<etu>
skrzyp: "nix search" is great and fairly new :)
<ij>
adisbladis, nice to know that's how it works finally. I was so puzzled by nix-env not obeying NIX_PATH
<skrzyp>
etu: but is there `nix install` or `nix remove` too? The tooling is sadly scattered over few spearate binaries
<etu>
skrzyp: no :(
<etu>
skrzyp: It's comming as adisbladis said above :)
zupo has joined #nixos
<{^_^}>
[nixpkgs] @filalex77 opened pull request #74124 → nixosTests.riak: port to python → https://git.io/JePck
<skrzyp>
Anyways… can I somehow regenerate desktop menu entries cache? I've just added a package (claws-mail to be clear) and it didn't show up its desktop entry in KDE Plasma menu.
<adisbladis>
skrzyp: Invoke `kbuildsycoca5`
<adisbladis>
That nukes the desktop item cache for KDE :)
<adisbladis>
And rebuilds from scratch
<skrzyp>
Hm. Apparently it didn't helped.
<skrzyp>
But I can still call the claws-mail binary itself from term. Maybe the expression is broken in some way.
<adisbladis>
Maybe it doesn't provide a .desktop file?
<ij>
now everyone will complain about pasting stuff directly into irc
mananamenos_ has quit [Remote host closed the connection]
<adisbladis>
skrzyp: $XDG_DATA_DIRS will not point directtly at claws. It will point at your user profile and your system profile.
MmeQuignon has joined #nixos
<{^_^}>
[nixpkgs] @adisbladis pushed commit from @elohmeier to release-19.09 « nixos/containers: fix handling of cfg.additionalCapabilities »: https://git.io/JePcG
<skrzyp>
tj: depends on the "acceptable number of lines", I thought 7 are still okay
lukash_|away is now known as lukash_
<adisbladis>
skrzyp: I expect something like /home/adisbladis/.nix-profile/share:/etc/profiles/per-user/adisbladis/share:/nix/var/nix/profiles/default/share:/run/current-system/sw/share
<adisbladis>
If you installed something with nix-env you'll find it's .desktop file symlinked in /home/$USER/.nix-profile/share/applications/
<skrzyp>
adisbladis: really?
<skrzyp>
$ echo $XDG_DATA_DIRS | sed 's/\:/\n/g' | wc -l
<skrzyp>
141
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<adisbladis>
:O
<adisbladis>
That's huge
<skrzyp>
Yet I still have only 7 pkgs in user env, zero in global.
<adisbladis>
Anyway, the important bit is that ~/.nix-profile/share is included
<ij>
what's that feature called where you scroll and when you release fingers, it keeps going?
<skrzyp>
adisbladis: Yes it is, and Claws desktop entry is symlinked into it
<skrzyp>
tj: "natural scrolling"?
zupo has joined #nixos
<adisbladis>
skrzyp: Hm, ok. Then I really think kbuildsycoca5 should do the trick...
<skrzyp>
adisbladis: But I ran it like… 4 times while we talked. Guessed it might be delayed due to some syncing or anything, but no - it's still not in menu
<raboof>
I have a password-protected ~/.ssh/id_rsa, and indeed 'git pull' uses that key and prompts me for a password.
o1lo01ol1o has quit [Ping timeout: 252 seconds]
DigitalKiwi has joined #nixos
<raboof>
I'd like it to cache the key for a while though. I've set `default-cache-ttl` and `default-cache-ttl-ssh` in ~/.gnupg/gpg-agent.conf` and restarted, but this didn't have an effect yet
<eyJhb>
raboof: isn't that the job of our ssh agent?
<raboof>
any idea? (I also have `use-agent` in `~/.gnupg/gpg.conf` but not sure if it matters)
<adisbladis>
If you use gpg-agent you don't want to use the "regular" ssh agent at all
<adisbladis>
SSH knows which agent to speak to through the environment variable SSH_AUTH_SOCK
Khetzal has quit [Remote host closed the connection]
<adisbladis>
raboof: If you're not using any of the nixos modules for ssh agents you can add `enable-ssh-support` to ~/.gnupg/gpg-agent.conf
<adisbladis>
Though I'm not sure what the default path for the socket is
Khetzal has joined #nixos
m0rphism1 has quit [Ping timeout: 265 seconds]
<raboof>
I'm using the nix modules, and `ssh-add` works with either the gpg-agent with ssh support or the ssh agent. Caching the key still didn't work, but I guess `ssh-add` isn't too bad
<eyJhb>
raboof: if you have the ssh agent, you shouldn't need to ssh-add it each time
<eyJhb>
As far as I know
m0rphism1 has joined #nixos
<adisbladis>
eyJhb: With the regular old ssh-agent you need to add it every time as it doesn't persist state (every time being once per agent session)
<adisbladis>
Iirc gpg-agent imports the key into your regular gpg keychain, so that will persist (though I might be wrong on this one)
<eyJhb>
yeah, once per agent session, which usually for me lasts untill reboot
<eyJhb>
Which was what I thought was somewhat the wanted case
<skrzyp>
adisbladis: How weird. Runing kbuildsycoca5 --menutest shows claws-mail.desktop in menu tree dump, yet it's still not visible in the actual KDE menu
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<skrzyp>
I've installed qpdfview too and it has the same issue.
<skrzyp>
However, when I installed firefox and telegram yesterday, it somehow worked.
<Ashy>
hmm, `nix-channel --update` is stuck waiting for a lock and seems to stay this way whenever i've tried in the last couple weeks
inkbottle has quit [Remote host closed the connection]
philr has quit [Ping timeout: 240 seconds]
Ralith_ has quit [Ping timeout: 240 seconds]
eraserhd has quit [Ping timeout: 240 seconds]
amfl has quit [Ping timeout: 240 seconds]
eraserhd has joined #nixos
philr has joined #nixos
noudle_ has joined #nixos
<Ashy>
oh, log has this error unit-script-nixos-upgrade-start[23139]: warning: unable to download 'https://nixos.org/channels/nixos-19.09': Couldn't resolve host name (6); retrying in 314 ms
<inf>
hey. I'm starting to get more and more private nixos boxes... I'd love to get some tips on how to properly lay out my nixos-config repository :) I know this is all down to personal preferences, but I'd really like to get some inspirations...
drakonis_ has quit [Ping timeout: 246 seconds]
civodul has joined #nixos
<gchristensen>
how many do you have, and are they isolated or edployed to with nixops or something?
<{^_^}>
To search public NixOS/Nixpkgs/NixOps configs, use https://search.tx0.co (Ping ${"til" + "pner"} if it acts up again)
<inf>
gchristensen: ~5.
drakonis has quit [Read error: Connection reset by peer]
<infinisil>
{^_^}: That was really slow!
<inf>
nice, that is exactly what i was looking for.
christianpoveda has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
Bryophyllum has quit [Quit: Bryophyllum]
drakonis_ has joined #nixos
<christianpoveda>
Hi everyone, how can I remove a package i installed with `nix-env -iA`? i tried `nix-env -e` but it didn't work
<christianpoveda>
nvm I was writing the name of the package wrong
puck has joined #nixos
drakonis1 has quit [Ping timeout: 250 seconds]
cfricke has joined #nixos
drakonis has joined #nixos
<tilpner>
Does anyone have nginx serving files straight out of a DynamicUsers StateDirectory? I don't know how to set up the permissions properly
<gchristensen>
sounds tricky
drakonis_ has quit [Ping timeout: 245 seconds]
ffl^ has joined #nixos
horner has quit [Quit: zzZzZzZzZZZzzz…]
<mananamenos>
clever, i did `nix-shell default.nix -A dart` but dart binary is not visible from the shell
<clever>
mananamenos: if you want the dart binary, then you need `nix-shell -I nixpkgs=. -p dart`
christianpoveda has quit [Remote host closed the connection]
drakonis1 has joined #nixos
<mananamenos>
clever, when i change a version of package, but do not change its sha256 attr in fetchFromGitHub, nor the the modSha256, it still builds. Shouldn't it throw some error?
fendor has quit [Remote host closed the connection]
<clever>
mananamenos: if it can find something with the given sha256 in /nix/store, it will blindly use that, even if its an older version
fendor has joined #nixos
<clever>
mananamenos: because you lied to nix and claimed that the output has the same hash
<tilpner>
Or if it's something else entirely
<tilpner>
"You wanted the source for your project, that has this hash? Look, I have a cat picture with that hash already, you can have that"
<clever>
mananamenos: if you replace a few characters with 0's, it will try to download and find the real hash
<clever>
tilpner: :D
* tilpner
actually does have cat pictures in the store
Khetzal has quit [Quit: No Ping reply in 180 seconds.]
<thefloweringash>
sorry if this is a FAQ, but if I opened a PR against the wrong base branch, can I change it to another branch (like staging instead of master) without spamming everyone that ever touched nixpkgs?
rauno has quit [Ping timeout: 276 seconds]
fusion809 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #74052 → glava: fix glava wants to load shaders from /etc/xdg/glava → https://git.io/Jeij7
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JePBl
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #74060 → Regroup all themes under data/themes → https://git.io/JePJR
<{^_^}>
[nixpkgs] @worldofpeace pushed 6 commits to master: https://git.io/JePB8
waleee-cl has joined #nixos
MmeQuignon has quit [Ping timeout: 250 seconds]
sigmundv has quit [Ping timeout: 276 seconds]
horner has joined #nixos
<worldofpeace>
hmm, I think we need some part of our review process or a label like timokau mentioned for "needs new reviewer assigned". I have a lot of PRs that I don't think I can continue to review.
<worldofpeace>
In the meantime, is there anyone up to the task of helping me out? A lot of them I can't act on.
<betaboon>
worldofpeace do you have a list somewhere ?
<worldofpeace>
I could make a list betaboon. The hard part is for some PRs, I can see a lot of issues others can't see. It just takes to much time to actually follow up on them.
<ddima>
worldofpeace: I can try to have a look as well if there is some list. Ofc no idea if I can handle any of them yet before seeing them.
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JePBb
<betaboon>
worldofpeace: yeah thats been my biggest concern on getting involved in reviewing PRs, that i am unsure if i could maintain the expected level of quality
<ddima>
betaboon: well, this has to be learned somehow ;)
<worldofpeace>
betaboon: in nixpkgs current state, it asks a lot on the reviewer. I'm also a committer so it can turn into: review, review, review, test, test test, approve/reject, merge, determine if you can backport.
<betaboon>
ddima: yeah. as i suggested/asked a while ago: i think it would be best to have some kind of mentoring-program. as in: new reviewers get reviewed by old/proven reviewers for a while as to "train" the new ones in order to maintain a level of quality and scale the usergroup that does reviews
<worldofpeace>
if even one of those parts in the chain could be lessened it would help tremendously
<betaboon>
worldofpeace: i am very willing in helping with reviews, but considering that https://github.com/NixOS/nixpkgs/pulls is my expected entry point to do so it is just realy hard as where to start, and what to do to get started
<worldofpeace>
betaboon: totally understand.
<worldofpeace>
I'm not sure I could give good advice on that currently.
<worldofpeace>
In general, I think we accept to many packages. So much so I'm not sure I want to review them. Hopefully that's not taken the wrong way.
ardumont has quit [Ping timeout: 245 seconds]
<ddima>
I kinda agree from my limited exprience, I think many things could nicely live outside of the core repo.
<worldofpeace>
Totally. I starting to think there should be a restriction on binary packages especially. Kinda understand why fedora has this restriction.
<worldofpeace>
i.e 100+ electron blob packages :(
CptCaptain has joined #nixos
<gchristensen>
+1
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<CptCaptain>
I've added services.openssh.enable = true; to my configuration.nix. Yet the service is not available and there is also no error
<Shoubit>
Most of the Electron packages can probably delete their provided binary in their install phase, use a Nixpkgs provided one, and save a bunch of space.
<CptCaptain>
What could be the reason for that?
<worldofpeace>
we even have to do some pretty poor backports policy stuff just for them. they depend on webservices so it's like I have to backport all their updates to stable.
<gchristensen>
CptCaptain: (just checking) have you run nixos-rebuild switch?
<CptCaptain>
ghuntley: yes
<worldofpeace>
Shoubit: I've tried this, and the thing is what something was compiled with might not match with what's in nixpkgs. so it'll error out.
NobbZ has joined #nixos
<betaboon>
CptCaptain: is the sshd service running ?
alexherbo2 has quit [Ping timeout: 252 seconds]
<CptCaptain>
betaboon: No. The Systemd service doesn't even exist. ss -lt doesn't output anything
NobbZ has quit [Client Quit]
<clever>
CptCaptain: does sshd.service exist in /etc/systemd/system/ ?
m1sosoup has quit [Remote host closed the connection]
<CptCaptain>
clever: yes it does
m1sosoup has joined #nixos
<clever>
CptCaptain: try `systemctl status sshd.service` ?
<clever>
CptCaptain: and is this on the installer iso or a normal install of nixos?
<ddima>
Shoubit: but blobs aside, this is still quite a massive maintenance burden and other implications like security, right? we're basically #2 behind AUR but we have all of that all together, with additional security implications, where AUR is basically "yeah no, you review every package yourself at your own risk. no trust".
<CptCaptain>
clever: It was entirely empty
<CptCaptain>
But I was able to start it using systemctl start
<CptCaptain>
The image is an aarch64 image
<CptCaptain>
Maybe it's an issue with that
<jschievink>
is anybody using Nix on Arch Linux?
<lordcirth>
jschievink, what is your actual question?
<clever>
CptCaptain: it may have already had ssh enabled, but auto-start disabled
ardumont has joined #nixos
<jschievink>
I keep running into problems where problem segfault because they get linked against mixed dynamic libraries from Nix and Arch
<CptCaptain>
clever: What does that mean?
<jschievink>
*programs, not problem
<clever>
CptCaptain: you have to manually `systemctl start sshd.service` until you stop including the installer profile
<ddima>
Shoubit: and some ecosystem just drag in a lot of work with them. Python being kinda one of them, but some we can't avoid in the core set, others we probably could.
<clever>
jschievink: nix is supposed to ignore the dynamic libraries the host provides, but LD_LIBRARY_PRELOAD can leak in
<jschievink>
I've had that happen with system tools like `ar`, but it mostly happens with things I build myself
<jschievink>
clever: can `LD_LIBRARY_PATH` also be a problem?
<clever>
jschievink: yes
<jschievink>
I had that enabled a while ago, but now it's blank and the problem still happens
<ar>
people mistake me for a tool again
<jschievink>
I think it caused a few problems a while ago
zupo has quit [Ping timeout: 250 seconds]
<CptCaptain>
clever: Do you mean I have to restart?
<clever>
CptCaptain: restarting wont make much of a difference
<jschievink>
Now my binary seems to use Nix' dynamic linker, but dynamic libs from Arch
<clever>
jschievink: sounds like you used the wrong -L flags when compiling
<clever>
jschievink: is the nix sandbox enabled?
silver_hook has joined #nixos
<clever>
jschievink: was it built manually or with nix-build?
<jschievink>
clever: it might be because I've built outside of Nix, but `ld` is still the one from Nix
<jschievink>
`gcc` is from Arch though
<CptCaptain>
clever: What will make a difference?
<jschievink>
I might have messed up my environment
<clever>
jschievink: you must use cc-wrapper (a shell script around gcc) for things to correctly use nix
captn3m0 has quit [Read error: Connection reset by peer]
<clever>
jschievink: and installing any compiler tools will not work with nix, you must always build inside nix-shell or nix-build
<jschievink>
clever: ah, I see
captn3m0 has joined #nixos
<jschievink>
would be great to improve the UX here, things segfaulting for no reason is no fun
<clever>
jschievink: the moment you start using the arch gcc, nix has lost control and cant stop you from doing things the wrong way
<jschievink>
clever: I'd have expected that `gcc` to link against Arch's dynamic linker
<jschievink>
but if it uses Nix' `ld` I can see why it might not
<ddima>
worldofpeace: thanks for the link, I've missed that thread. maybe there is some way to make progress there, as it seems dearly needed.
noudle has joined #nixos
andreas303 has quit [Remote host closed the connection]
<gchristensen>
clever: did you have a nix expression like make-disk-image but to create a ZFS root?
<ddima>
worldofpeace: btw, wrt usage metrics (this ofc only applies to existing packages), thoughtpolice was iirc considering to work a bit on that, so maybe there is already something for at least possibly cleaning out existing packages?
<ikwildrpepper>
gchristensen: would love to have that too :)
<gchristensen>
ikwildrpepper: :D
<gchristensen>
ddima: I talked to somebody on Friday who desperately wanted a "popcon" like feature for Nix/Nixpkgs/NixOS, based on derivation names instead of store paths
andreas303 has joined #nixos
knupfer has joined #nixos
<betaboon>
worldofpeace: it seems like almost half of all PRs are package updates, i wonder why those have to be review by hand anyway as most packages have tests. lets say: if its a package-update + it has doCheck + it causes < N rebuilds + all rebuilds have doCheck aswell -> automerge. (havent read RFC0050 completly yet)
<gchristensen>
a key point here is that they patch nixpkgs at a very low level, and so their usage would be completely unrepresented if the count is by store path
<betaboon>
gchristensen: nixos/tests/installer.nix has something on zfs-root
<gchristensen>
yeah, that does the installation to ZFS. I guess that could be made in to a library sort of
<ddima>
gchristensen: yeah, its not gonna be without problems. caches/mirrors etc will get in the way too. more like as a noisy metric among many others.
<gchristensen>
I think we have the technology to do something a bit more smart
<gchristensen>
until recently I didn't realize that usersactually wanted to do it
<clever>
gchristensen: yes
<thoughtpolice>
gchristensen: I've thought about something like that, too
<gchristensen>
I saw the value as mostly one-sided (on the maintainer side) but this was a case where the user was hurt badly by nobody knowing they were using something in nixpkgs
<thoughtpolice>
ddima: Yes, part of the motivation is to get ideas about what closures/paths are popular, though we'd need to do a bit more work to tie it to deriv names, etc
<gchristensen>
and importantly, they want to send contact information too -- because if they had known something was going to be deleted, they would have taken over maintainership
<clever>
gchristensen: its loosely based on the AMI generation, but amazon-image.nix is problematic, it has both aws things and ext3 things mixed together
<gchristensen>
awesome!
<gchristensen>
ikwildrpepper: ^
<thoughtpolice>
gchristensen: That's a bit more work, I think. It's almost something you want CODEOWNERS for, but without the "ownership" part
<adisbladis>
CODEUSERS ^_^
<thoughtpolice>
More like a "subscriber" to one file. Which in theory would notify them on changes like renames/deletes, etc via GitHub
<gchristensen>
they seemed to want to push to us derivations they use
<ikwildrpepper>
clever, gchristensen: <3
<gchristensen>
automatically
romildo has quit [Quit: Leaving]
<ddima>
gchristensen: is there some more in-depth discussion on that somewhere or concrete proposals for implementations? I have some time on my hands but maybe not the sufficient context for what an ideal solution could look like for things. But I also see this situation as one of the biggest challenges - sometimes feels like ~10 (ofc jk) people are spread over 10_000 things.
<gchristensen>
no, and I agree with you
<thoughtpolice>
gchristensen: Ok, that's even a bit weirder, but I can kinda see it
<gchristensen>
thoughtpolice: well, that is how popcon works
<adisbladis>
gchristensen: That's probably going to be a component of (un)trustix for other reasons :)
<thoughtpolice>
Oh, I see
<adisbladis>
Re users pushing derivations they use
<ddima>
iirc also somebody was pondering the thought of having some automation/comms channels to detect inactive maintainers and call them to action or so. so its not really the lack of ideas.
<gchristensen>
adisbladis: derivation names, specifically, not paths
<thoughtpolice>
ddima: I think my work on logging requests to the cache is more or less the only current work going on in that direction right now
<clever>
ikwildrpepper: what where you thinking of with zfs disk images?
<adisbladis>
gchristensen: Yeah you'd need to map paths back to attribute names
Khetzal has quit [Ping timeout: 245 seconds]
<gchristensen>
clever: personally, I want to make rpi images with ZFS roots
<adisbladis>
But there will be a source of that data at some point
<clever>
gchristensen: ah, ive wanted that too, but havent looked into it yet
<gchristensen>
adisbladis: as long as it is calculated locally, true
<adisbladis>
For my use case it's to prevent split-view attacks, but it can surely be used for other things
<clever>
gchristensen: ive been curious if 32bit zfs would work better there, since it has less ram and the heap size stuff wouldnt be an issue
<gchristensen>
interesting
Khetzal has joined #nixos
<ikwildrpepper>
clever: I mostly want to replace my current ec2 workers image (which use unionfs) with zfs version
<ikwildrpepper>
(using all available ephemeral disks )
<clever>
gchristensen: you might be able to get away with using an x86 runInLinuxVM to generate an arm image, the only tricky part is line 72 of my gist, where you fix the bootloader
FRidh has quit [Quit: Konversation terminated!]
<worldofpeace>
Ooh ddima If thoughtpolice has some work on that or ideas on how we could make it a more systemic change that could help.
<clever>
ikwildrpepper: youll likely have less to change, since my gist is already able to generate and upload working AMI's, i forget what was left to finish
<thoughtpolice>
adisbladis: The deriver is included in the .narinfo, so you can get it that mapping done "on demand" by the logging infrastructure. Unfortunately such a setup would possibly require a JOIN, which is (somewhat) problematic for my infra.
<{^_^}>
[nixpkgs] @bcdarwin opened pull request #74164 → ocamlPackages.biocaml: init at 0.10.0 → https://git.io/JeP0F
<raboof>
is there a way to pass the derivation hash to the builder script? I guess you could derive it from the `out` environment variable, but perhaps there's a nicer way?
<thoughtpolice>
Well, it's not really ultra problematic. But it introduces a resource requirement where one didn't exist before in the infrastructure right now for ClickHouse (JOINs are in memory only)
<clever>
raboof: builtins.placeholder "out"
silver_hook has quit [Ping timeout: 265 seconds]
boxscape95 has joined #nixos
<ddima>
thoughtpolice: well, I think this can't hurt as a somewhat low-hanging fruit and if the basic infra/tools are in place, it likely will be possible to switch it to derivation names too once other parts are adjusted. looking forward to it and hope some basic stats become available soon (even if just trivial csv with frequencies or so).
<thoughtpolice>
No, it will be far better than some CSVs. :)
<ddima>
well, incremental delivery and all ;)
<thoughtpolice>
It works already! You can use it today!
<ddima>
thanks, I will look at the updated docs. the last thing I saw was your initial proposal from a while ago.
<clever>
gchristensen: lines 39-71 of my gist dont care about the target arch, you just need to swap out the config.system.build.toplevel reference (or force lines 37&33 to use an x86 pkgs on an arm build)
<worldofpeace>
thoughtpolice: the expressivity of the emoji there is truely on point
<worldofpeace>
like both graceful and refined.
<clever>
gchristensen: and then use one of the sd-image scripts as an example to fix the bootloader, so it can boot
<gchristensen>
worldofpeace: Notion.so encourages that exact level of emojis, it is nice
<thoughtpolice>
:D
<adisbladis>
thoughtpolice: In my case I don't have narinfo files, only hashes.
<worldofpeace>
it's really fab
<thoughtpolice>
I really like it a lot. If anyone wants guest permissions on that page to comment etc feel free to ask, too
boxscape95 has quit [Remote host closed the connection]
<clever>
gchristensen: also on the subject of packet.net, what do you think about `curl | zfs recv` vs `nixos-install` ?
<thoughtpolice>
Need to put the fancy ASCII diagram I drew of the whole pipeline up there, too...
<clever>
gchristensen: i can see some issues with recv, since the IP's still have to be customized
knupfer has quit [Remote host closed the connection]
<clever>
gchristensen: that reminds me, there was a similar solution for digital ocean, let me find it...
<gchristensen>
clever: hmm maybe..... that would be really nice, but only an optimisation for customers who select zfs
<clever>
gchristensen: the network auto-config could help others too, less to bake into your configuration.nix files
<gchristensen>
it has to be baked somewhere
<clever>
gchristensen: i could see some less experienced users just breaking everything if they copy a configuration.nix over, and dont import your auto-generated stuff, and having it auto-config on bootup
<clever>
gchristensen: the digital ocean thing i linked above, is a systemd service that will configure the network on bootup, without baking it in
<gchristensen>
good thing the ycan rollback :)
<clever>
gchristensen: and packet.net does have similar api's
<clever>
gchristensen: it disables dhcp, and runs systemd-digitalocean-generator on bootup
<edef>
oh hi
<ddima>
thoughtpolice: if you see some work to help with progress that can be split out sufficiently from your deeper context, maybe poke me.
<etu>
,locate libQt5Core.so.5
<{^_^}>
Found in packages: robo3t, qt5.qtbase, qt59.qtbase, qt511.qtbase
<clever>
gchristensen: that will then query http://something/metadata/v1.json, find out what IP it should have, and generate systemd units automatically
<edef>
i started writing a packet version of systemd-digitalocean
<edef>
but i think i might've lost that code
<gchristensen>
ah, that will only work on the first boot
<edef>
not quite
<gchristensen>
oh?
<edef>
custom PXE will leave it enabled for the lifetime of the machine iirc
<gchristensen>
yeah, but these are not custom iPXE
<clever>
edef: oh, that was you in the nixops issue over 2 years ago! heh
<edef>
actually, no, the metadata service remains available
<CptCaptain>
services.openssh.enable = true; Only links the openssh service and doesn't enable it. How do I fix that?
<betaboon>
CptCaptain: it should start the service, it does for me since forever. can you provide some more context as to what you are doing / trying to achieve ?
<thoughtpolice>
For now I actually redirected the beta cache to that page for updates. Not sure what info should be permanent there vs Notion.
<clever>
CptCaptain: which enables ssh, but disables auto-start!
mexisme has joined #nixos
<clever>
CptCaptain: so yes, sd-image-aarch64.nix is your problem
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #73134 → nixosTests.plasma5: port to python → https://git.io/JeVMW
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/JePEM
<CptCaptain>
clever: Should I switch to base.nix?
<CptCaptain>
or something else?
<thoughtpolice>
Some things like design notes, TODO are probably best left on Notion. Things that users and maintainers want to know are probably best left on the nixos wiki
<clever>
CptCaptain: but if you remove that, it may cease to boot, so a new option must exist, that sets the right options for an rpi
<thoughtpolice>
Like who to contact, diagnostic info, etc
<CptCaptain>
clever: What would that line be for an amd64 machine?
<clever>
CptCaptain: you may need to try removing it, see what breaks, and then find a working minimal config to replace it with, and get the docs updated
<clever>
CptCaptain: try just imports = []; to start with, and see what happens from there
<{^_^}>
[nixpkgs] @jonringer pushed commit from @ajs124 to master « nextcloud: 17.0.0 -> 17.0.1 »: https://git.io/JePEy
fendor has quit [Ping timeout: 265 seconds]
<thoughtpolice>
gchristensen: Also, IDK if Notion emails you about this, but I did add you to some tickets on the BUGS.txt page for the long run. No pressure to comment but something you might want to look over
<thoughtpolice>
(I've added details to most of the bugs, by now)
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « calamares: 3.2.15 -> 3.2.16 »: https://git.io/JePEH
<lemsip>
hrm, within an FHS shell, pkg-config can't seem to find libraries that I definitely am providing, such as dbus. Is there something special I need to do to make this work?
<clever>
lemsip: ive found that pkgconfig only works right under normal nix-shell
justanotheruser has joined #nixos
<clever>
lemsip: and you generally dont need to use FHS
<CptCaptain>
clever: Oh boy. How does nixos handle changes in filesystem settings? Does it just reformat?
<clever>
CptCaptain: it wont format things on its own, it will just stop mounting, or try to mount with the wrong flags, causing it to not boot
<etu>
,locate libQt5WebSockets.so.5
<lemsip>
clever: well shit. That means that pretty much most rust things can't be built from an fhs env if they rely on system libraries (dbus in my case)
<{^_^}>
Found in packages: qt5.qtwebsockets, qt59.qtwebsockets, qt511.qtwebsockets
<clever>
lemsip: why are you trying to build things in an FHS?
<lemsip>
I can build my rust things from a regular environment, but other things rely on an FHS thing for node_modules, electron and the like. Since I have a single bash script that produces installers for my project, I'd very much like to be able to run pkg-config
<gchristensen>
thoughtpolice: yeah I know, I really like Notion too
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « catch2: 2.10.2 -> 2.11.0 »: https://git.io/JePuZ
<clever>
lemsip: for example, line 31-43 generates a fake electron cache, so electron-builder doesnt have to fetch anything
<raboof>
clever well, my *real* problem is that I want to generate a deterministic GUID - while the path might work, the hash seemed like a neater starting point.
drakonis has quit [Remote host closed the connection]
lukash_ is now known as lukash_|away
<rydnr>
I'd need some help. I lost my original grub menu after I chose to install Ubuntu in a partition, trying to recover my NixOS. The original problem was that my wifi card triggers a kernel panic in kernel 4.19. My current problem is that, in a chroot inside Ubuntu, `nixos-rebuild switch` fails with a error: opening pseudoterminal master: No such
<rydnr>
device`.
<rydnr>
I'd need some help. I lost my original grub menu after I chose to install Ubuntu in a partition, trying to recover my NixOS. The original problem was that my wifi card triggers a kernel panic in kernel 4.19. My current problem is that, in a chroot inside Ubuntu, `nixos-rebuild switch` fails with a error: `opening pseudoterminal master: No such
<rydnr>
device`.
<rydnr>
(sorry for the duplicate. I thought I was editing the first message)
orivej has quit [Ping timeout: 246 seconds]
<rydnr>
When booting from latest NixOS livecd I get the same kernel panic. I was able to boot from a NixOS 18.03 (I think), and with `nixos-enter` I got a different error (cannot write to file, without indicating which file), but stopped at the same point.
<rydnr>
My idea was to downgrade the kernel version, but I'm not able to make nixos-rebuild finish successfully.
<tilpner>
What's the newest kernel you've tried?
<rydnr>
Another option would be to start over with a clean installation and use my current configuration.
<skrzyp>
If my package has a dependency only usable for this particular package ever (some author decided to split his program into "actual program" + "libprogram"), can I nest mkDerivation inside build like: libprogram = mkDerivation {... \n buildInputs = [ libprogram... ?
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @r-ryantm to master « bibutils: 6.7 -> 6.8 »: https://git.io/JeP2i
o1lo01ol1o has quit [Remote host closed the connection]
halfbit has joined #nixos
o1lo01ol1o has joined #nixos
iqubic` has quit [Ping timeout: 246 seconds]
iqubic` has joined #nixos
MinceR_ is now known as MinceR
m1sosoup has quit [Quit: Leaving]
<steveeJ>
I'm trying to use buildRustPackage to build a member crate of a repo which is configured as a workspace. if I set `sourceRoot` to the path of it I get an error that `Cargo.lock` doesn't exist. is this use-case really an exception or am I doing something wrong?
<vaibhavsagar>
does anyone have experience with jupyter nbextensions in Nix?
<vaibhavsagar>
I have something that was working a few months ago but doesn't currently seem to work
<steveeJ>
to answer my own question. using `cargoBuildFlags = [ "-p workspace-member" ];` did the trick. now I'm stuck because we don't seem to have unstable or nightly rust in the tree :-D
<tilpner>
steveeJ: You don't have to use the NUR, you can directly reference the mozilla overlay if you want
mexisme has joined #nixos
<steveeJ>
tilpner: thanks, I was going to look for that again
<rydnr>
Finally I booted with the 20.03 livecd. I got the same kernel panics until I blacklisted both iwlwifi and iwlmvm modules. But I get "error: writing to file: Operation not permitted" when I try to do "nixos-rebuild switch" after "nixos-enter"
<tilpner>
rydnr: What about nixos-rebuild boot?
<rydnr>
If I use "chroot /mnt" instead of "nixos-enter", I get "error: opening pseudoterminal master: No such device"
fendor has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @d-goldin opened pull request #74190 → pythonPackages.m2crypto: conditional `typing` dep → https://git.io/JePaD
<rydnr>
I get the same error messages in both cases (nixos-enter and chroot /mnt)
<tilpner>
rydnr: What about nixos-rebuild boot?
<rydnr>
tilpner: yes, I tried and got the same error messages
<tilpner>
Can you strace it?
<tilpner>
Which file does it fail to write to?
<tilpner>
strace -o /tmp/log -vfefile ...
<rydnr>
I'm not sure which file it is by looking at /tmp/log...
Guest55348 has joined #nixos
<rydnr>
674 stat("/nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/x86_64", 0x7ffc2d6c9fb0) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/librt.so.1", O_RDONLY|O_CLOEXEC) = 3674 openat(AT_FDCWD, "/nix/store/mam3z7sv5csby0qccvn4qak9ri8r9vwr-acl-2.2.53/lib/libacl.so.1",
<rydnr>
O_RDONLY|O_CLOEXEC) = 3674 openat(AT_FDCWD, "/nix/store/mam3z7sv5csby0qccvn4qak9ri8r9vwr-acl-2.2.53/lib/libattr.so.1", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/w6gxny55j7pmxm0kkc9fdvjw86zyk3n4-attr-2.4.48/lib/libattr.so.1", O_RDONLY|O_CLOEXEC) = 3674 openat(AT_FDCWD,
<rydnr>
"/nix/store/mam3z7sv5csby0qccvn4qak9ri8r9vwr-acl-2.2.53/lib/libcrypto.so.1.1", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/w6gxny55j7pmxm0kkc9fdvjw86zyk3n4-attr-2.4.48/lib/libcrypto.so.1.1", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD,
<rydnr>
"/nix/store/2q1izv872frqih56akkf4n22z67q8qj2-openssl-1.1.1d/lib/libcrypto.so.1.1", O_RDONLY|O_CLOEXEC) = 3674 openat(AT_FDCWD, "/nix/store/mam3z7sv5csby0qccvn4qak9ri8r9vwr-acl-2.2.53/lib/libpthread.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD,
<rydnr>
"/nix/store/w6gxny55j7pmxm0kkc9fdvjw86zyk3n4-attr-2.4.48/lib/libpthread.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/2q1izv872frqih56akkf4n22z67q8qj2-openssl-1.1.1d/lib/libpthread.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD,
<rydnr>
"/nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/libpthread.so.0", O_RDONLY|O_CLOEXEC) = 3674 openat(AT_FDCWD, "/nix/store/mam3z7sv5csby0qccvn4qak9ri8r9vwr-acl-2.2.53/lib/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/w6gxny55j7pmxm0kkc9fdvjw86zyk3n4-attr-2.4.48/lib/libc.so.6",
<rydnr>
O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/2q1izv872frqih56akkf4n22z67q8qj2-openssl-1.1.1d/lib/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)674 openat(AT_FDCWD, "/nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/libc.so.6", O_RDONLY|O_CLOEXEC) = 3674
<samueldr>
am I missing something or is udev.extraHwdb not taking effect on rebuild?
<samueldr>
tried to udevadm hwdb --update too, just in case
<bdju>
I just realized pcmanfm doesn't show the icons I selected in lxappearance. what should I check into? also pcmanfm in general seems very buggy on nixos. no right click menu, and when typing to search it starts the search over after each letter
<samueldr>
though I would feel more satisfied if udevadm would print where it looked at files
growpotkin has joined #nixos
<steveeJ>
tilpner: any chance you've packaged a rust crate before which has a "link" dependency?
<tilpner>
steveeJ: I don't understand the question
<samueldr>
and rebooting *does* refresh the hwdb appropriately, seemingly
<kraem>
long shot: is there a way to check the progress of hydra's progress of building a binary cache of a specific package?
selfsymmetric-pa has joined #nixos
justanotheruser has joined #nixos
<LnL>
hydra not really but nix-build --dry-run will show you how many paths will be substituted vs built locally
<ddima>
Is there some documented syntax to run ofborg build just for a particular platform/builder type (for resource conservation)?
run500 has joined #nixos
<kraem>
yeah i guessed so. that's actually what i am doing with nixos-rebuild but sometimes i want to know if it's worth waiting for hydra for large builds like chromium and libreoffice to update the kernel. thanks LnL
<clever>
kraem: nixos hydra or your own hydra? which channel are you on?
<run500>
im running a "nix copy --to file://..." and "nix copy --from file://..." command to move a closure between machines. i get an error on the second step saying "path is not valid". any ideas what could cause that?
<kraem>
clever: nixos hydra. i'm on unstable-small (iirc i switched from unstable to receive kernel updates faster while unstable was stuck for ten days)
<clever>
run500: you can also use `local?root=/foo` instead, and it will create a complete store that could be chroot'd into, or copied from
civodul has quit [Read error: Connection reset by peer]
<clever>
kraem: if you switch back to normal nixos-unstable, then hydra will always have things built
<run500>
clever: that seems to work better. i think the problem was something to do with signatures - one of the paths was missing a sig, and the "copy to" step succeeds with no error message, the local variant fails
<clever>
kraem: tricky question, you would have to review the release notes and see what has been fixed and what type of problems your open to
<clever>
run500: there is also `nix copy-sigs`
<run500>
clever: in my case i dont care that much, and using --no-checks sigs seemed to do it
<clever>
run500: and nix copy can be told to not check sigs
<jared-w>
Hmm, speaking of flakes, is anyone using flakes on nixos yet in their configuration.nix file as a full replacement for channels (or manual pinning/updating/etc)?
<jared-w>
nice. How are you liking it so far? I'm thinking of switching to that if I can. I have a vaguely janky setup that involves just importing tarbells that I'm finding slightly annoying to use
<pbb>
it has it's advantages and disadvantages :))
<pbb>
so I do "nix run --recreate-lock-file .#nixosConfigurations.amalthea.config.system.build.toplevel -c sudo switch-to-configuration switch" to update my system
knupfer has quit [Quit: knupfer]
<jared-w>
heh, I guess that command could look slightly prettier lol
<jared-w>
I can imagine. I'm hoping nix becomes a bit more ergonomic eventually. Right now I'm still trying to find a balance between taking advantage of nix for projects and not rendering the entire git repo unusable for non-nix users
<jared-w>
(likely unrelated to flakes but it seems to be part of a general trend)
<pbb>
hehe
<pbb>
currently I found no way of rebuilding my system with a remote build host like with "nixos-rebuild --build-host", which makes it tempting to switch back right now
<pbb>
have to compile some stuff on my slow laptop
<{^_^}>
[nixpkgs] @jonringer opened pull request #74202 → python3Packages.papis: remove restrictive version bounds → https://git.io/JePoI
<jared-w>
just spitballing here, but would it be possible to setup something like a cachix cache on your fast PC and then reference it from the laptop?
knupfer has quit [Client Quit]
<clever>
jared-w: its called nix-serve, and the module is already in nixos
<pbb>
ah yes but the next problem would be how to ergonomically make that fast computer build my system config
fendor has joined #nixos
<pbb>
and I would have to set up the signing keys and stuff, which is not necessary with nixos-rebuild --build-host
<clever>
pbb: configure it as a build machine
<clever>
pbb: just a normal build machine, not --build-host
<clever>
pbb: and if the build user is marked as a trusted user, you dont have to setup the signing
<pbb>
hmmm yeah I guess, but it has so many limitations like I need root/trusted access on the remote machine
<steell>
clever: i can just add that to my imports list?
<clever>
red[evilred]: what was the problem again?
<clever>
steell: a variation of that, yes
<run500>
im running "nix copy --from ... --no-check-sigs" and yet i still get an error saying "cannot add path because it lacks a valid signature"
<red[evilred]>
What's the liklihood of being able to port nixos to this device (odroid HC-2)
<clever>
run500: did you run it as root (or a trusted user) ?
<{^_^}>
[nixpkgs] @magnetophon opened pull request #74204 → noise-repellent: init at unstable-29-12-2018 → https://git.io/JePow
<clever>
red[evilred]: should be fairly simple, but there wont be a binary cache building for it, youll first need to get nix installed on the device, and use nix-build to build things like the closure of nixos
<m3lt99034>
I got a udev files, that i need to add to services.udev somehow. but service.udev.packages does not directly accept a text file :c
<run500>
clever: i did not, and it works when i do! thanks.
<pbb>
I am getting "error: unable to start any build; either increase '--max-jobs' or enable remote builds" with --max-jobs=0, even though I have specified remote builders. what could cause this?
<clever>
pbb: i think it still wants to do some builds locally, and then ran out due to a limit of 0
<pbb>
meh
<pbb>
I don't want to build the new kernel on a slow dual core
<clever>
pbb: then use the supported-features flag
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JeP6W
hmpffff has quit [Quit: nchrrrr…]
hmpffff has joined #nixos
hmpffff has quit [Client Quit]
noudle has quit []
drakonis_ has quit [Ping timeout: 246 seconds]
sigmundv has joined #nixos
kaba has joined #nixos
<kaba>
Hi, how can I start plasma in wayland on current unstable? Apparently it should work, just not yet available as a session file (though even that seems to be available, just not yet merged https://github.com/NixOS/nixpkgs/pull/56988)
magneticduck has quit [Remote host closed the connection]
hlolli has quit [Remote host closed the connection]
steell has quit [Ping timeout: 240 seconds]
fenedor has joined #nixos
kaba has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @kalbasit opened pull request #74216 → bazel.tests: update the protobuf test to the latest protobuf version → https://git.io/JeP6a
mizmor has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<mizmor>
Hi! My configuration.nix is such a mess. Are there any ressources/best practices that could help me refactor it? Thanks!
fendor has quit [Ping timeout: 265 seconds]
fenedor is now known as fendor
lux4 has quit [Ping timeout: 245 seconds]
tilpner has quit [Ping timeout: 252 seconds]
lux4 has joined #nixos
<jared-w>
That gets complicated. There are a lot of different ways to go about it and everyone uses a different way :p
Ariakenom has quit [Quit: Leaving]
<ddima>
mizmor: I dont know a good resource - but maybe just start pulling things apart in some "topical" sense into separate files, like "desktop.nix" for X and WM stuff etc, "network.nix" etc. Ideally you should find 2-3 people on github who's config looks nice to you and just get inspired ;)
<jared-w>
you can use flakes, you can use channels, you can manually pin everything, you can use fetchTarbell, etc., and that's just for getting "pkgs". Some people abstract their entire configuration.nix into a series of modules. Some people have it separated by configuration per user + configuration per machine and their configuration.nix imports a file that picks which machine/user combo you want to use for your particular
<jared-w>
machine
<jared-w>
Some people have it all mostly inlined, some have 900 different files... Some use home-manager, some don't, some use their own home brewed version of a "user env" manager
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<mizmor>
Thanks to both of you :) I think I'll go the separate file way for now! I only use Nixos on 1 machine at the moment and that would suit me well i think!
<jared-w>
and then there's configuring your applications and terminal stuff. Some people use normal files and "link them in" with nix. Some people inline their entire rc files into their nix files. Some do both in the same file. Still others upload their user configs to the NUR and reference it from there
<jared-w>
then there's the difference between having your setup be agnostic to whether or not you're running NixOS vs nix on, say, macOS or some other linux distro. That can change your config setup significantly
tilpner has joined #nixos
<mizmor>
jared-w: I see why you said it could get complicated haha!
<mizmor>
Thanks a lot, very enlightening!
sigmundv has quit [Ping timeout: 276 seconds]
<ddima>
mizmor: once you have managed to split things like that you can start thinking about making actual nixos modules out of those (parametrizing and stuff) and what you need as new packages, I'd suggest to place in some folder called "overlays" or so and use the overlay mechanism with those. After that you've probably seen enough to proceed a bit more on your own.
<jared-w>
np. Just wait until you have to figure out how different code projects use hercules-ci, cachix, lorri, direnv, niv, and different philosophies of setting up nix integration with build systems for code repos
levdub has quit [Remote host closed the connection]
jlv has joined #nixos
<ddima>
mizmor: many people have their nix config in a git repo and symlink the files to the locations in /etc and ~/.config/nixpkgs (for home-manager) - some just have /etc/nixos as a git repo.
<jared-w>
Don't worry, all of this is extensively documented by 3 people in four articles, a few magical git issues, and a few vaguely worded discord threads spanning 5+ years of constantly-changing best practices :p
<jlv>
Is there a recommended / easy way to overlay a package with a version from a different channel?
<mizmor>
jared-w: hahahaha good to know!
<ddima>
jared-w: jeez, let him make his own discoveries ;)
<ddima>
err, s/his/their
<mizmor>
ddima: i like the /etc/nixos as a git repo idea, are there any downside to it?
<jared-w>
Nope
<{^_^}>
[nixpkgs] @tfc opened pull request #74218 → Nixos test containers python port → https://git.io/JeP6S
<jlv>
mizmor: you need to do things as root, which is a little annoying.
<mizmor>
jlv: what if i chown /etc/nixos to my user? is this bad?
<jared-w>
You can chown /etc/nixos as your user then everything works great. That's what I do
<jlv>
I think that answers your question, lol
<jared-w>
/insert "vague security bla bla"
<ddima>
mizmor: well, I use home manager, so I have various locations, so for me it makes sense to link to the repo instead of having all of it in /etc/nixos. but this is all rather easy to change, so just go with it.
jb55 has joined #nixos
<jared-w>
Alternatively, you can stick home manager inside your configuration.nix file so that you don't have to run `home-manager switch` and just `nixos-rebuild switch` is sufficient
<jlv>
Personally, I have a separate directory, and a simple bash script to copy and activate the configuration.
fendor has quit [Quit: Leaving]
<jared-w>
no such thing as too many options amirite /s
<mizmor>
oof
monsieurp has quit [Ping timeout: 246 seconds]
<jlv>
I don't think there's really a "right way". Whatever works and is easy for you.
<ddima>
mizmor: dont overthink it. esp since you only have one box, just go with the simplest thing that works and extend
<mizmor>
i was interested in using home manager too but i wanted to configure my machine first
<jared-w>
I want to kinda write an article on this and sorta why I discovered and settled on my particular setup
<ddima>
there is really no point in looking at all options upfront, probably just too many issues to deal with at once :)
<mizmor>
and yeah that's what i understand now!
<mizmor>
jared-w: would be very cool to read
<jared-w>
https://github.com/jared-w/nixos-configs/ you can look at my setup here. It differs from a lot of others online in that it's intentionally low abstraction to the point of almost-being-annoying
<ddima>
mizmor: I can only second what jared-w said about running home-manager from your nixos config - much nicer and less confusing than dealing with separate home-manager generations. thats more for mental note for later.
<jared-w>
(the only file you need to read is nixos/configuration.nix)
<mizmor>
ddima: duely noted :)
<mizmor>
jared-w: cool! i'll check it out!
kraem has quit [Quit: outta here]
run500 has quit [Remote host closed the connection]
<jared-w>
Just pushed what my configuration.nix looks like as of 2 minutes ago so you might wanna refresh the page if you already opened it
werner291 has quit [Quit: werner291]
kaba has joined #nixos
CptCaptain has quit [Ping timeout: 265 seconds]
<mizmor>
jared-w: i see what you meant! It's really straightforward, everything in configuration.nix! that's cool!
werner291 has joined #nixos
<jared-w>
Some of the weakness of my setup are: 1) a lot of unnecessary conflation of levels of abstraction. Things like wanting ghcide for neovim are pushed into the global configuration when they shouldn't necessarily be. Lots of globally installed packages so that "local concerns" can take advantage of it, etc. 2) My setup is purposefully non idempotent. If you run it on one day vs another day you're not guaranteed the _exact_
<jared-w>
version of some packages; that comes from using fetchTarbell for most everything
<{^_^}>
[nixpkgs] @flokli pushed commit from @kylesferrazza to master « ipscan: init at 3.6.2 (#74209) »: https://git.io/JeP6N
<mizmor>
jared-w: good to know!
<jared-w>
I don't need nearly all of those fonts in there and I should simplify that. My global packages list is horribly organized as well. There's basically no rhyme or reason to it
<jared-w>
My neovim and zsh configuration are both still completely outside of nix and updated manually and some of my configuration.nix reflects that (having 'lsd' installed so that zsh works vs having a zsh "module" responsible for installing dependencies that my aliases/functions depend on)
<mizmor>
jared-w: whats bad about having config outside of nix?