<{^_^}>
[mobile-nixos] @samueldr opened pull request #97 → Add release.nix for hydra consumption → https://git.io/JviLu
cosimone has quit [Ping timeout: 256 seconds]
<samueldr>
,callPackage Cadey
<{^_^}>
Cadey: If a Nix file foo.nix starts with something like `{ stdenv, cmake }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<samueldr>
this is how dependency injection happens in nixpkgs derivations
<Cadey>
:+1:
<Cadey>
samueldr++
<{^_^}>
samueldr's karma got increased to 182
<samueldr>
thogh, for actually more useful info, at the root of the nixpkgs checkout, nix-build -A the-attr-name, which the-attr-name is what would be in pkgs/top-level/all-packages.nix (or a language-specific set)
<samueldr>
though*
<samueldr>
not all paths end up mapping 1:1, so tab completion may help at guesttimating without looking at the file
justanotheruser has quit [Read error: Connection reset by peer]
nuncanada3 has joined #nixos
<blaggacao>
We've been trying to get sway running on raspberry pi 3 B, but enduringly runs at 65% cpu. It seems that vc4 driver does not kick in although nix seems to have vc4 support. Any clues?
bvdw has quit [Read error: Connection reset by peer]
thc202 has quit [Ping timeout: 256 seconds]
bvdw has joined #nixos
mrCyborg has quit [Quit: WeeChat 2.7.1]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
bhipple has joined #nixos
<evils>
blaggacao: grats, i haven't been able to get it to start xD, maybe ask in #nixos-aarch64?
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<Ashy>
so i just tried moving my desktop nvme nixos drive to a laptop and for some reason it won't boot, fails to mount zfs root in stage one and doesn't prompt for zfs encryption password and can't load stage 2
<Ashy>
desktop is ryzen 1600, laptop is an intel xeon
<lovesegfault>
that's what my config looked like before worldofpeace made me wage war on NIX_PATH
<lovesegfault>
I hate NIX_PATH
<lovesegfault>
so much
<adisbladis>
Each machine doesn't have much specifics
<lovesegfault>
adisbladis: I've been wanting to create modules
<lovesegfault>
but I've been busy waging war on NIX_PATH
bhipple has joined #nixos
<Ashy>
i guess i can just do a fresh install and add it as a new machine
medvid__ has joined #nixos
* lovesegfault
nods
<lovesegfault>
that's what I'd recommend
ash__ has joined #nixos
<{^_^}>
[nixpkgs] @alyssais pushed commit from @joachifm to master « nixos/firejail: use local runCommand »: https://git.io/JvimZ
<{^_^}>
[nixpkgs] @alyssais merged pull request #70477 → nixos/firejail: build wrapper dir with runCommand → https://git.io/JeCOa
<adisbladis>
My typical nixos installation process goes something like: a bootstrap installation, import hardware-configuration.nix into my repo and then `nixops deploy`
<lovesegfault>
adisbladis: stop tempting me with nixops
<lovesegfault>
I can't be tempted
<adisbladis>
:D
<adisbladis>
I'd really like for stateless deploys for my personal machines
<adisbladis>
Though that's on the way :>
<lovesegfault>
is NixOps not stateless?
<adisbladis>
It stores deployment state in a sqlite db
<{^_^}>
[nixpkgs] @alyssais pushed 2 commits to release-20.03: https://git.io/JvimX
day|flip has quit [Remote host closed the connection]
v88m has joined #nixos
eacameron has quit [Quit: Connection closed for inactivity]
<kalbasit>
in NixOS modules, is there a way to always include the default value in the final list when the type is a `listOf str`?
<kalbasit>
Say I have `default = [ "a" "b" ]`. If I set the value to `[ "c" "d" ]` in some config file then the final value should be `[ "a" "b" "c" "d" ]`
<kalbasit>
Trying to avoid having to create a custom type for it
<samueldr>
something else that is often dismissed, but has been the solution, try another usb drive, unless you know for a fact it's still good and boots other distros
<clever>
samueldr: hydra-queue-runner will poll every file in $NIX_REMOTE_SYSTEMS once per minute
<clever>
samueldr: and dynamically update the internal machine list when the last-mod timestamp changes
<samueldr>
yeah, looking at the source of hydra for NIX_REMOTE_SYSTEMS helps
ash__ has joined #nixos
<clever>
samueldr: HYDRA_DBI and PGPASSFILE are likely also used
ddellacosta has quit [Ping timeout: 258 seconds]
<samueldr>
hydra-in-a-bag
<clever>
ah
hoek has joined #nixos
<samueldr>
it works, I just couldn't find info about machines
<samueldr>
and tried lazy irc before source diving :)
<clever>
ive memorized the codebase :P
flodeskum has joined #nixos
<{^_^}>
[nixpkgs] @martinbaillie opened pull request #82537 → tailscale: init at 0.96-33 → https://git.io/JviOD
flodeskum has quit [Ping timeout: 258 seconds]
<Ashy>
hmm, now the nixos installer kernel isnt detecting the nvme drive
<clever>
Ashy: what does lspci say?
<Ashy>
yeah not listed there
<lovesegfault>
Ashy: is this a laptop?
<Ashy>
samsung 960 evo 250GB
<Ashy>
yeap, dell precision 3520
<lovesegfault>
Ah
<lovesegfault>
disable RAID in the BIOS
<lovesegfault>
pick AHCI mode
<clever>
Ashy: if its missing from lspci, then the bios has either disabled it, or its not seated properly
<lovesegfault>
clever: it's some bullcrap setting on new laptops
<lovesegfault>
they come with disks configured in some wacky way Linux hates
<lovesegfault>
you have to go in and set them back to AHCI
<Ashy>
ah yeap, quick google suggests that too
<clever>
lovesegfault: apple does even more whacky stuff
<lovesegfault>
clever: apple is the worst
<clever>
lovesegfault: apple has a custom nvme controller, that i think does transparent hw encryption
<clever>
lovesegfault: i suspect you will loose all data if you try to move the ssd to another mac
<Ashy>
doesnt make any sense to have raid enabled on this machine anyway since it only has a sinlge m.2 slot and the battery is too large to allow space for a sata drive on the single sata port
<clever>
and linux needs a patch to the kernel
<samueldr>
clever: no need to suspect
<lovesegfault>
clever: amaze
<samueldr>
clever: it's a property of their secure syste
<samueldr>
system*
<lovesegfault>
"security"
<samueldr>
it is
<samueldr>
no need for scare quotes
<clever>
samueldr: yeah, ive heard that the nvme is part of the T2 security chip
<samueldr>
yeah
<cyberwolf[m]>
c 4 ram nixos feels normal?
<clever>
ive also heard that such mac's cant boot with the SSD missing
rardiol has quit [Ping timeout: 265 seconds]
<Ashy>
lovesegfault: yeah ahci has it showing up now, cheers
<danderson>
is there a way to ignore the cached build version for just one derivation? I think a derivation is broken and I want to check it, but nix-build just grabs the cached build :)
<petersjt014[m]>
There's something in the nix config file
<petersjt014[m]>
~/.nix/nixpkgs.nix?
wrl has quit [Ping timeout: 268 seconds]
<petersjt014[m]>
Wait no
<danderson>
I'm looking at nix.conf settings
<danderson>
but I can only turn off the cache globally, afaict
<petersjt014[m]>
That'd be for all of them
<danderson>
which then causes a ton of dependencies to build from scratch
<danderson>
right :)
<petersjt014[m]>
There's a command flag
blaggacao has quit [Ping timeout: 256 seconds]
<petersjt014[m]>
Lemme check (dis gonna be a pain on mobile)
blaggacao has joined #nixos
toasted_avocados has joined #nixos
<danderson>
found a hack: nix-build once to pull from cache, then nix-store --delete <path to the derivation I care about>, then nix-build with caching disabled
<petersjt014[m]>
Okay, it'll be in the form `--option name value`, that I know
<petersjt014[m]>
That could work
Intensity has joined #nixos
<petersjt014[m]>
Just one more sec, I'm on the right manpage
<danderson>
(my hack didn't work, it's still building the dependency closure)
<petersjt014[m]>
Okay, think I found it: add `--option builders-use-substitutes false`
<lovesegfault>
What's the input type of builtins.filterSource?
<petersjt014[m]>
One stupid idea: setting `connect-timeout` to 0
<petersjt014[m]>
Wait
<petersjt014[m]>
I think you need to have some the first one I suggested along with `--option fallback true`
<petersjt014[m]>
> If set to true, Nix will fall back to building from source if a binary substitute fails. This is equivalent to the --fallback flag. The default is false.
<{^_^}>
error: syntax error, unexpected $undefined, expecting ')', at (string):288:10
kolbycrouch has quit [Quit: Leaving]
ash__ has quit [Quit: ash__]
zupo has joined #nixos
<petersjt014[m]>
So overall: `--option builders-use-substitutes false --option fallback true`
<{^_^}>
[nixpkgs] @danderson opened pull request #82552 → libvorbis: document that patch for CVE-2017-14160 also fixes CVE-2018-10393. → https://git.io/Jvisb
<MichaelRaskin>
Well, last time it has bitten me was pre-5.0
<MichaelRaskin>
I run the /nix/store on it
<MichaelRaskin>
As for files you actually care about, like /home — do not trust BtrFS fully, but also there are zero storage devices in the market worth your trust
<lovesegfault>
It bit me pre-5.0 too
<MichaelRaskin>
So, backupds, do them
<lovesegfault>
but that one time was too much
<MichaelRaskin>
For a store, XFS was also horrible. I want a filesystem that mainline kernel maintainers will not mess up and smile
<MichaelRaskin>
(The store is the FS that gets 100% full during my use)
<MichaelRaskin>
ext4 — static inodes. Fine for /home, no thanks for store, I have previously run out of inodes
<MichaelRaskin>
And in any case, backups. I think I had more HDDs warranty-replaced in the same laptop than hard BtrFS failures.
opthomasprime has joined #nixos
fuzen2 is now known as Fuzen
<Ashy>
all filesystems scare me, I've learned to stop worrying and love the bomb
<Ashy>
also local laptop drives are just a cache of git repos and data that's also on various cloud storage products
<NobbZ[m]>
I lost a VM to btrfs… All of a sudden it started failing randomly turning the FS read-only…
* lovesegfault
runs LUKS + XFS and loves it
<lovesegfault>
my main gripe with ZFS is no s4 sleep support
<Ashy>
does luks give you cow snapshots?
<lovesegfault>
no
<lovesegfault>
LUKS = encryption
<lovesegfault>
if you want CoW snapshots use ZFS or BTRFS
<MichaelRaskin>
Or, you know, LVM2
<lovesegfault>
or that
<Ashy>
yeah I've just done btrfs on luks for this machine
<lovesegfault>
I'm waiting for bcachefs
<lovesegfault>
it's the only promising thing right now
<Ashy>
yeah I'm watching it too
<lovesegfault>
Although ZFS might gain s4 support before bcachefs reaches mainline, lol
<{^_^}>
[nixpkgs] @FRidh opened pull request #82560 → Python: introduce tests for interpreters → https://git.io/JviZv
<sphalerite>
ar: I'm not sure, because a bunch of things have to happen in the right order when waking, and some things that happen during a cold boot shouldn't happen
<kolaente_>
Hey there, when trying to run nixos-rebuild switch, I get the error "error: opening file '/root/.cache/nix/gitv2/17blyky0ja542rww32nj04jys1r9vnkg6gcfbj83drca9a862hwp/refs/heads/HEAD': No such file or directory" and then it just exists
<kolaente_>
Has someone seen this before?
dansho has joined #nixos
orivej has joined #nixos
<clever>
kolaente_: you could just `rm -rf ~/.cache/nix` to clear the cache, and nix will re-make it
<kolaente_>
clever: that seems to do something, thanks for the tip
<steevveen>
I'm having difficulties at https://nixos.org/nixos/packages.html for performing any query, they won't finish. Only the waiting dots in circle.
<steevveen>
Correction, the dots aren't in circle :-)
<Ashy>
uh oh, this is the first time i've attempted to use an nvidia gpu on linux for a very long time
c0c0 has quit [Ping timeout: 260 seconds]
helozjisky has joined #nixos
helozjisky has quit [Remote host closed the connection]
vishnusuresh has joined #nixos
vishnusuresh has quit [Remote host closed the connection]
<__monty__>
Hmm, how do I find out why my configuration depends on openssl-1.0.2u? I don't think I can use why-depends since nix is refusing to evaluate the unsafe expression, so I can't pass a store path.
kai_w has joined #nixos
<tilpner>
__monty__: You can try with NIXPKGS_ALLOW_INSECURE=1
<__monty__>
tilpner: So there's no way to figure out how to avoid installing unsafe software without installing said unsafe software? Talk about a catch-22.
<tilpner>
...
<tilpner>
That's not what I said
helozjisky has joined #nixos
helozjisky has quit [Remote host closed the connection]
cosimone has joined #nixos
helozjisky has joined #nixos
<{^_^}>
[nixos-homepage] @zimbatm pushed 0 commits to installers: https://git.io/JvicN
caeomuehh^ has quit []
andymandias_ has quit [Quit: ZNC 1.7.4 - https://znc.in]
<kai_w>
jtojnar: sorry, I'd gone to bed. Adding things to `targetPkgs` in a FHSUserEnv doesn't add them to /bin of the output (only `runScript` is), so they're not available on $PATH. Unfortunately `runScript` is generated in a way that swallows $0, so I can't even make links and match on them.
<MichaelRaskin>
__monty__: you do not need to install it, it is enough to instantiate it
<kai_w>
`targetPkgs` is put into /bin on the chroot, but that doesn't help unless I make `runScript = "bash";` or something, which makes things more verbose. because then I have to do `nix run ... -c quartus quartus_asm` instead of just `-c quartus_asm`, which is how it would work for a normal package
nbg^ has joined #nixos
<__monty__>
MichaelRaskin: But if it's instantiated wouldn't a distracted/mindless rebuild pick it up as already being installed?
<MichaelRaskin>
No, instantiated only writes a derivation file, not the actual output paths
<__monty__>
Hmm, and how do I find out the attribute's name? Neither `openssl` nor `openssl-1.0.2u` work. Do I have to pass an expression wrapping the expression with callPackage?
oida has quit [Ping timeout: 240 seconds]
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
opthomasprime has quit [Remote host closed the connection]
opthomasprime has joined #nixos
thc202 has joined #nixos
xelxebar has joined #nixos
absoluutely_ has joined #nixos
absoluutely_ has quit [Max SendQ exceeded]
cjpbirkbeck has quit [Quit: cjpbirkbeck]
flodeskum has joined #nixos
flodeskum has quit [Remote host closed the connection]
xelxebar_ has joined #nixos
flodeskum has joined #nixos
<eyJhb>
ANyone up for helping me finding the best way, when a vim plugin is actually a node package?
xelxebar has quit [Remote host closed the connection]
opthomasprime has left #nixos [#nixos]
oida has joined #nixos
<__monty__>
eyJhb: No packaged examples you could look at?
organixpear has joined #nixos
flodeskum has quit [Ping timeout: 260 seconds]
mrCyborg has quit [Quit: WeeChat 2.7.1]
<eyJhb>
__monty__: not really, there is a open PR that hasn't been merged yet
<MichaelRaskin>
__monty__: try reading all-packages.nix, I guess. Should be openssl_1_0_2
<eyJhb>
So I have no clue if it is the correct way
<organixpear>
hey all. i am trying to get xorg working on my linux from scratch distro using nix. i googled for help but found only nixos related info. i just need a few pointers on getting it installed and working
<__monty__>
MichaelRaskin: Hmm, yes that was the right guess.
zupo has joined #nixos
orivej has quit [Ping timeout: 268 seconds]
Diogo4189 has joined #nixos
<__monty__>
Next obstacle, I can't call why-depends on my configuration.nix because it's not in the nix store. Do I nix-instantiate it with callPackage to find the store path?
<adisbladis>
__monty__: nix-instantiate '<nixpkgs/nixos>' -A system
<eyJhb>
The hell, why doesn't it show up in `nix search nodePackages` ?
DiogoXPTO is now known as Diogo
<adisbladis>
Probably nodePackages is `dontRecurseIntoAttrs`
<adisbladis>
Indeed it is
<adisbladis>
We omit large generated sets from the search for performance
<eyJhb>
adisbladis: but should e.g. https://www.npmjs.com/package/coc-go be added to pkgs/development/node-packages/node-packages-v10.nix, or how is it done in a good way?
<adisbladis>
eyJhb: nodejs 10 is the default nodejs interpreter
<eyJhb>
adisbladis: I would enjoy having it done for all the coc-* things on npm, since they all need "compiling" -> https://www.npmjs.com/search?q=coc-*
Diogo4189 has joined #nixos
<eyJhb>
And then make overrides for them all in vimPlugins
<adisbladis>
I've merged it locally on top of the other coc PR. Now regenerating.
VanCoding has joined #nixos
<eyJhb>
Good, the generate.sh script doesn't play on my machine...
<eyJhb>
`./generate.sh: line 5: cd: /home/eyjhb/projects/git/nixpkgs/pkgs/development/node-packages/pkgs/development/node-packages: No such file or directory` weird
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « gitAndTools.stgit: 0.21 -> 0.22 »: https://git.io/JvilQ
ottidmes has joined #nixos
<ottidmes>
I have some dynamic config I inject in my NixOS configuration through a function call, but I sometimes forget to call this function and would like to give myself a reminder that I forgot through some throw someone. I tried a few things, but it is either to early (can't check options, would lead to recursion) or too late (fails on missing _module.args before the assertions get evaluated). Any ideas?
o1lo01ol1o has quit [Remote host closed the connection]
<kaliumxy1>
tilpner nope AMD
<kaliumxy1>
Linux nix.kalium.home 5.4.24 #1-NixOS SMP Thu Mar 5 15:43:52 UTC 2020 x86_64 GNU/Linux
flodeskum has joined #nixos
<Thra11>
If I *think* a PR will cause lots of rebuilds, should I target staging? Or should I target master and switch to staging if once I *know* it will cause lots of rebuilds?
<kaliumxy1>
CPU AMD Ryzen 5 2400G with Radeon Vega Graphics
<kaliumxyz>
The system seems to jsut randomly get rekt for no reason, thats not good.
reactormonk has quit [Ping timeout: 255 seconds]
<kaliumxyz>
I've had random program crashes, random reboots, random hangs, USB ports not working, all happening really quick after using the system and only for the last 5 days or so
<mrCyborg>
So it works normaly, the FPS count on games suggest that intel is being used. And when I set the offload environment variables it breaks...
<mrCyborg>
Right now I'm just hoping to get the NVIDIA part to be working at all. If that means I only run NVIDIA and intel is disabled I'm fine with that
<m1cr0m4n>
Hey folks. Anyone running rsyslogd and nixos 20.09/master? It is failing to build by default because rsyslogd is apparently compiled with an outdated openssl package. Has anyone else seen this?
<mrCyborg>
Anyone has a pointer? I'm really kinda lost here
<kaliumxyz>
oh we on 20 now?
<NobbZ[m]>
NemesisD: Something like that, yes, though that will enable dropbox autostart for all of your users, `~/.config/autostart` is the location for only a single user.
<LinuxHackerman>
kaliumxyz: not released yet, it's still in beta
<NemesisD>
NobbZ[m]: is there a way to write that with home-manager? i'm actually adding that to my home.packages. i'm sure it won't work
<Derek_S>
hi folks, is there any way to set a GRUB theme through NixOS config? Can't find any config options that suggest this and no results online.
rogue_koder_ has quit [Ping timeout: 256 seconds]
<NobbZ[m]>
I'm pretty sure there is a way to do it with home-manager yes. Though I'm not yet deep enough into all the nix stuff to come up with something :D
<tilpner>
mrCyborg: Did you enter the pci ids yourself, or did you copy them? You may need videoDrivers = [ "nvidia" ];
rogue_koder_ has joined #nixos
<mrCyborg>
tilpner: I calculated it myself using the guide in the docs
kaliumxy1 has joined #nixos
<kaliumxy1>
honestly this is making this PC unusable.
<kaliumxy1>
recommendations for how to go about fixing this? systemlogs tell me nothing.
<tilpner>
mrCyborg: In case you're not aware, the import-from-fetchgit gives msteen a lot of power over your system
<mrCyborg>
tilpner: Good point, I should lock that to a spesific commit
<kaliumxy1>
Also I get a lot of random conneciton requests from ??? might be because I was running some p2p software for a week last week and not yet everyone has blacklisted me for not running it anymore.
<kaliumxy1>
this machine is not exposed to the internet though.
mrCyborg_mobile has joined #nixos
<mrCyborg_mobile>
Opened a mobile session because I will probably be rebooting my laptop
blaggacao has quit [Ping timeout: 265 seconds]
kaliumxyz has joined #nixos
blaggacao has joined #nixos
<kaliumxyz>
It hung again.
<NemesisD>
NobbZ[m]: i forgot i can just add `dropbox start` to my startup hook with xmonad!
<kaliumxyz>
CPU borked?
<kaliumxyz>
Its been running 24/7 for over a year.
<kaliumxyz>
I had zero stability issues.
<tilpner>
Run a memory test?
<kaliumxyz>
How do I do that?
<NobbZ[m]>
I think its probably the way to go for `xmonad`, I'm not sure if it follows XDG autostart conventions.
<tilpner>
kaliumxyz: Boot memtest86, you may be able to write the nixpkgs-provided-one to an external drive (haven't done this)
<mrCyborg>
(will trade toilet paper for nvidia help :P /sarcasm)
<T_S_>
the tp folks won't be back soon.
<kaliumxyz>
I have a nix live boot USB here for recovery.
<tilpner>
mrCyborg: GPU debugging is painful enough with local access. I got that module working on unstable, and can provide values of a working system though
<mrCyborg>
tilpner: what kind of values?
<tilpner>
mrCyborg: "the FPS count on games suggest that intel is being used", what was the value?
<tilpner>
mrCyborg: Configuration, command output, etc.
abathur has joined #nixos
rogue_koder_ has joined #nixos
<mrCyborg>
26 fps on portal 2. I got 60fps easily on my arch install
<mrCyborg>
My arch install only used NVIDIA, I think.
<mrCyborg>
Which makes sense because my battery live improved on nix
<mrCyborg>
I'll run portal 2 again to check that I didn't cap the FPS
<mrCyborg>
tilpner: Portal 2 doesn't launch then. I just get the "prepairing to launch portal 2..." window, and then the window closes. Nothing happens after that
<tilpner>
mrCyborg: What's the error message?
<mrCyborg>
I don't think the game hangs because then steam would say I'm still playing it. Instead it seams like the game just exists
o1lo01ol1o has quit [Remote host closed the connection]
kapil_ has quit []
<Yaniel>
mrCyborg: journalctl -f
rogue_koder_ has quit [Read error: Connection reset by peer]
<tilpner>
No, not journalctl
<Yaniel>
keep that open while trying to launch the game
rogue_koder_ has joined #nixos
kaliumxy1 has quit [Remote host closed the connection]
<Yaniel>
I get steam games' output there
<tilpner>
Start steam from a terminal, or start a terminal from within steam
<Yaniel>
if memtest hangs then you most likely have a hardware problem
<Yaniel>
have you played with memory (or other) clock frequencies?
<kaliumxyz>
Not since I got the PC, I tweaked the fan setting to be more constant about 4 months ago when I replaced the CPU fan. I run this PC 24/7
reactormonk has joined #nixos
<kaliumxyz>
I think someone might have accendentally tilted it over while I was gone, I'll try taking out the ram and putting it back in again to make sure its in their sockets correctly.
<organixpear>
is it possible to use dbus from nix on a distro other than nixos? i can't seem to get dbus-daemon or dbus-launch to work they look for /etc/dbus-1/session.conf and clearly that doesnt exist
zeta_0 has quit [Quit: rebuilding]
mexisme_ has joined #nixos
fusion809 has joined #nixos
<kaliumxyz>
heh it hangs on 32% with two sticks of ram and 61% with one stick of ram
<avn>
Folks, have a silly question, how I can add a patch to result of fetchFromGitHub? (I need to apply fix, but won't modify `version` due assertions breakup)
elux has joined #nixos
<elux>
hi there
<tilpner>
organixpear: Are you building your own thing on LFS and nixpkgs? If so, I recommend to look the nixos modules anyway, they should describe how it's implemented on NixOS and could probably be translated to a format you can use
<elux>
has anyone here used fwupdmgr to upgrade firmware on their hardware? its been working for me for > 1 year, but all of a sudden my system has an update, downloads, fwup will reboot, but it never triggers the firmware updates? any ideas?
<fusion809>
Anyone else experiencing the error "error: attribute 'appimage-exec' missing, at /nix/store/rbb7w0h5ldw30ckqzvf3qa3am76j22ln-nixos-20.09pre216896.0729b8c55e0/nixos/pkgs/tools/package-management/appimage-run/default.nix:8:24" when they run nixos-rebuild switch --upgrade with appimage-run enabled in /etc/nixos/configuration.nix?
<adisbladis>
avn: You don't apply it to fetchFromGitHub, you apply it to the derivation that you're building.
<tilpner>
genesis: ^
mexisme_ has quit [Ping timeout: 255 seconds]
organixpear has quit [Remote host closed the connection]
<genesis>
ho
<tilpner>
genesis: You touched it last, maybe you can help fusion809 c.c
<avn>
adisbladis: usually -- yes. But I need inhect patch to a source which consist of patches (yes, I am about wine staging, lol)
<genesis>
sure but i've to do some shopping quickly, back in an hour
<avn>
adisbladis: and we have assert unstable.version == staging.version, so I want to grab "tag+1" commit, without update version field
<genesis>
fusion809 : it seems the file is not included, so that could be another issue , seems not to be with appimage-exec himself, never had missing issue
<avn>
adisbladis: well, I silly, sleepy and tired, re-examine expression and probably have a solution
<genesis>
could it be with merge issue from nixos ?
blaggacao has quit [Read error: Connection reset by peer]
<fusion809>
Definitely recent enough for it to be plausible.
blaggacao has joined #nixos
seneca has joined #nixos
cosimone has quit [Quit: Terminated!]
Mateon1 has quit [Remote host closed the connection]
Mateon1 has joined #nixos
seneca is now known as lanyap
elux has quit [Quit: leaving]
lsyoyom has quit [Quit: WeeChat 2.7.1]
shibboleth has joined #nixos
<bhipple>
electrocat: FWIW I'm running fwupd successfully on 20.03
<bhipple>
elux: ^
<tilpner>
fusion809: Which commit is it?
<fusion809>
tilpner: how would I determine which commit it is?
<bhipple>
elux: I have also run fwupd successfully on 19.09 in the past, but not in the last two months or so. (Lenovo X1 Carbon)
<tilpner>
Oh, it's from yesterday
<tilpner>
That's why the commit wasn't found locally
<tilpner>
That's master, not unstable
<iMatejC>
Well this suddenly does not work anymore .... `wget -O- -q https://nixos.org/channels/nixos-unstable/git-revision` ... is there any alternative link for this? I just want to get git revision of some channel with curl/wget
<tilpner>
fusion809: Can't reproduce locally. See if you can get the same error without your entire configuration.nix
<kaliumxyz>
anyone here has some good reading while I wait for memtest to finish running? I got a newer version and it doesn't hang c:
<fusion809>
What try removing (obviously backing it up) my configuration.nix and trying a rebuild? Guessing not as I think that'd result in an error, so are you suggesting I remove appimage-run from my configuration.nix?
lambdaclock has joined #nixos
disasm has joined #nixos
heatmiser has joined #nixos
<fusion809>
Just tried removing the appimage-run lines, and wow this update is a big one.
<fusion809>
Lots of new goodies to enjoy after this update no doubt.
flodeskum has joined #nixos
<zeta_0>
i am going to print from an hp printer, am i going to need to install proprietary software for it?
sigmundv__ has joined #nixos
mrCyborg_mobile has quit [Ping timeout: 246 seconds]
<kaliumxyz>
not sure here is the snippet I use zeta_0
<cole-h>
Does `nix why-depends` only look for direct dependencies, and not dependencies-of-dependencies?
<zeta_0>
will i be fine with pkgs.hplip or will i need pkgs.hplipWithPlugin(proprietary code)?
o1lo01ol1o has joined #nixos
<Thra11>
zeta_0: I can print stuff from my HP printer with just hplip. It may depend which printer model you have though.
flodeskum has quit [Ping timeout: 255 seconds]
juhe has joined #nixos
shibboleth has quit [Quit: shibboleth]
lambdaclock has quit [Quit: WeeChat 2.7.1]
lambdaclock has joined #nixos
<bdju>
Getting a LOT of Firefox crashes lately!
<kaliumxyz>
same.
<Yaniel>
I'm not :3
<hexa->
huh, same.
o1lo01ol1o has quit [Ping timeout: 265 seconds]
<hexa->
especially when I'm interacting with addons via the toolbar.
<hexa->
on wayland
<kaliumxyz>
Good to know, then that might be inrelated.
<kaliumxyz>
unrelated*
<zeta_0>
thanks guys, i am using hp envy 7640'
<kaliumxyz>
firefox extensions also spam up my journalctl.
<bdju>
I'm also on wayland, I had a few from opening a file picker (both to upload and to save stuff), but one just now I had just opened a link from irc
lambdaclock has quit [Client Quit]
lambdaclock has joined #nixos
<{^_^}>
[rfcs] @Ericson2314 opened pull request #66 → content-addressed derivations RFC -- merge in new master → https://git.io/Jvi0z
<{^_^}>
[rfcs] @Ericson2314 closed pull request #66 → content-addressed derivations RFC -- merge in new master → https://git.io/Jvi0z
lanyap has quit [Remote host closed the connection]
<tilpner>
fusion809: If you add appimage-run back in, does it still error?
<iMatejC>
bhipple: thanks, I was searching for less hacky solution, I managed to shorten it down by one command, but I do not feel much better with `curl -sL https://nixos.org/channels/nixos-unstable | grep -Po "(?<=nixpkgs-channels/commits/)[^']*"`
turion has joined #nixos
<zeta_0>
i just want to wirelessly print something from my laptop
fusion809 has quit [Ping timeout: 240 seconds]
zupo has joined #nixos
<turion>
Anyone successfully using audacity with extensions?
juhe has quit [Remote host closed the connection]
bhipple has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @alyssais pushed commit from @dywedir to master « miniserve: 0.5.0 -> 0.6.0 »: https://git.io/Jvi0p
<genesis>
you can see some example how they will be easier, i think i will offer full stdenv , so extraPkgs will be buildInputs instead.
<fusion809>
Nice. Opens the door up to more Nix packages utilizing AppImages.
<colemickens>
Is that good? :/
<genesis>
better than no apps ?
<colemickens>
Is it closed source software or hard to build from source or why do we have to do that most times?
<genesis>
both
<colemickens>
(didn't mean to sound unappreciative, btw, currently using some proprietary software that uses those tooling)
<genesis>
btw, no one try to prevent people to do better packaging
o1lo01ol1o has joined #nixos
<colemickens>
yes, of course, sorry, I shouldn't have said anything.
<genesis>
i did what, 40 natives packages on nixpkgs
<genesis>
and i don't use appimage, it's an excercice.
<{^_^}>
[nixpkgs] @veprbl pushed commit from @knedlsepp to release-20.03 « python37Packages.vtk: Fix build (#82336) »: https://git.io/JviEf
<genesis>
it's kind of closed loop, i was one of main contributer @ appimagetool, then i discovered nixpkgs searching to solve appimage issue, then i rewrite a bit of nixpkgs appimage support
o1lo01ol1o has quit [Ping timeout: 258 seconds]
kaliumxyz has quit [Ping timeout: 256 seconds]
<genesis>
with more advertising on nixpkgs,, so nixpkgs people doesn't like me to doing such stuff, when appimage authors doesn't like me to promote nixpkgs over appimage
<genesis>
that'll never stop me to doing such stuff.
flodeskum has joined #nixos
gustavderdrache has joined #nixos
<colemickens>
genesis: I'm really sorry, it was a tiny tongue-in-cheek comment. I feel badly that my comment might have been taken as discouragement.
<colemickens>
I think anything that helps get packages into nixpkgs is great, and especially tooling that helps other people to do so as well. I also wish everything were FOSS and that it were easy to build everything from source, but obviously that's not the case.
<colemickens>
genesis: thank you for your work. I think I just updated a package that uses some of this tooling, in fact.
<genesis>
i'm fine, was more concerned on some comment on my PR, but btw, there is always some people who try to show their purity.
<genesis>
building stuff from source, i was a 15 years gentooist before switch to nix :)
delli3 has joined #nixos
<tilpner>
I don't think it's just purist
<tilpner>
There are many disadvantages to using appimageTools (before your changes, at least): Bundled dependencies can
<tilpner>
...'t be replaced, often limited to one or just two architectures
<tilpner>
appimageTools only exists for temporary "I need this now, will package properly Later(tm)" situations
<genesis>
yes, would be fun to have a look how electron apps are made to be able to write a feature to allow other arch to use their own electron on appimage
<genesis>
btw, lot of currently use appimage are only provide as appimage, like soulseekqt i maintained.
<ottidmes>
I am getting the following error, is there an easy way to determine who is to blame? error: Package ‘openssl-1.0.2u’ in /nix/store/hvyv8k385snbdp480x8lzny3ldy8s6l4-source/pkgs/development/libraries/openssl/default.nix:130 is marked as insecure, refusing to evaluate.
kenran has joined #nixos
<ottidmes>
I get I could probably use why-depends, but I would first have to allow it to build to get those paths, no?
<genesis>
i used to override old openssl to get deprecated ssl
<adisbladis>
And it hooks into fish via the prompt function iirc
<cole-h>
Strange -- I don't have that problem (also using fish and direnv)
<adisbladis>
So that would make sense
<__monty__>
cole-h: I'm on an older OSX, cause apple.
<cole-h>
Ah, I missed the "I'm on a mac" part haha
<__monty__>
Hmm, how do I troubleshoot direnv?
<cole-h>
Run it from the command line until it breaks? :P
<adisbladis>
elux: Download it locally and `(callPackage /path/to/hyper.nix { })`
<tilpner>
No, wait
<tilpner>
What do you want to modify? Maybe you can use overrideAttrs?
<adisbladis>
Even better: use nix-shell
<elux>
sure that'd be fine. i just dont know how to do that.. with nix, everything feels very different depending on the setup
<tilpner>
You still haven't said what you want to change
<__monty__>
cole-h: That's immediately.
<cole-h>
Now see if you can get a meaningful backtrace from it
<elux>
i dont mind managing it locally, i have a home.nix for home-manager where i have most of my pkgs, then at the top, `let hyper = import ./hyper; in { home.packages = with pkgs; [ htop etc hyper ]; }`
<cole-h>
And see if can debug from there.
<elux>
i just want to get the latest version of it, the pkg is grossly outdated
<__monty__>
cole-h: I can't pass any flags, it immediately crashes with that SIGILL error.
mrCyborg has quit [Quit: WeeChat 2.7.1]
<cole-h>
I meant see if go has any env vars that gives you a more meaningfull backtrace, like `RUST_BACKTRACE` does
<manveru>
__monty__: kinda looks like direnv is built against the wrong version of macos?
<manveru>
well, or go
<tilpner>
elux: Perhaps you can update it in nixpkgs, so it's no longer outdated
<__monty__>
manveru: Yeah, I'm thinking something like that too.
<{^_^}>
golang/go#37459 (by carlca, 2 weeks ago, closed): SIGILL: illegal instruction on any go tool under macOS
<ottidmes>
tilpner++
<{^_^}>
tilpner's karma got increased to 66
<cole-h>
__monty__: Now try to build it yourself by using `nix-build -A direnv --check`
<manveru>
__monty__: seems some catalina thing again :P
<tilpner>
elux: But if you insist on keeping your update local, callPackage on your copy was correct. overrideAttrs might work if you don't need to adapt libPath
<ottidmes>
your suggestion works, it is just that I should not be trying nix why-depends on it XD, then it will build
<tilpner>
ottidmes: The whole point of why-depends is answering the question "why is x in the runtime closure of y?", but you don't care about the runtime closure
<tilpner>
You care about x being used at all, and answering that doesn't require building
<manveru>
so solution is to upgrade to go 1.14 they say
<elux>
would be nice for the nix language to be rewritten and made to be more intuitive
<tilpner>
why-depends must build the arguments, because the runtime closure can't be determined without it
<elux>
or perhaps its the nixpkg core functions, whatever it is.. very inconsistent
<ottidmes>
too bad I now have to write it to a file, so I can see it easily in the whole tree, would have liked the same output as why depends but on a different subject
nuncanada2 has joined #nixos
<tilpner>
ottidmes: It's displayed in less for me, and then you just do /openssl.*1.*0.*2
<elux>
nixos is the best linux, the only thing that holds it back is how unintuitive the nix language is
<tilpner>
Probably with fewer quantifiers
<manveru>
elux: you tried guix?
<elux>
neat. i will check it out
<ottidmes>
tilpner: ah, did not realize I could do that in less, only used that in manpages, I just opened it in my code editor on my desktop and its mongodb that is the cause
<MichaelRaskin>
Note that a part of conceptual load is not really dependent on the language (although there are language trade-offs made differently between Nix and Guix)
<zeta_0>
hello everyone, i am going to contribute to xmonad and darcs, but i am not sure how to build them in a nix-shell, is there some instructions anywhere that show some default.nix configurations on how to do it correctly?
<elux>
hyper = pkgs.callPackage ./hyper/default.nix { }; might have just worked..
<adisbladis>
elux: I was just gonna say that :)
<adisbladis>
Happy to be your rubber duck
<elux>
yep it works.. yay for trial+error brute-force technique with nix :/
<elux>
thanks for helping
<zeta_0>
the #darcs and #xmonad channels cannot help me with the nix part, that is why i am asking here
<adisbladis>
zeta_0: You should generally be able to use the nixpkgs expressions to hack on pretty much anything packaged in nixpkgs (barring bugs and new dependencies).
rsoeldner has quit [Remote host closed the connection]
<Thra11>
zeta_0: You'll want to use cabal2nix I think. There are some tutorials (for haskell dev with nix) online somewhere, though I can't remember which ones are good/up-to-date
rsoeldner has joined #nixos
<zeta_0>
adisbladis: ok, with ghc it was pretty easy to install it from nixpkgs as a user, but to contribute to ghc i had to clone and build it in a nix-shell using ghc.nix, which was very difficult to setup, so i wasn't sure if i have to something similar with darcs and xmonad
<MichaelRaskin>
With xmonad you can probably just do a complete build every time?
<judson_>
Oh, that looks like it could do exactly what I want :)
johnw has quit [Ping timeout: 268 seconds]
<judson_>
Now I just have to find the hashes in e.g. nix-build errors
* judson_
task add project:Nixpkgs ...
<cole-h>
evanjs: btw if you wanted to avoid borg evaling + building, you can add WIP to the title I think
<cole-h>
idk if it changes after the fact, though
<evanjs>
eh still should help
<zeta_0>
ok, i have a generic default.nix file to develop haskell in a nix-shell, i'll just tweak it to build xmonad and darcs, to develop them in an isolated environment
<evanjs>
cole-h: it does build now, anyhow, so that should be fine
<zeta_0>
hopefully that works
<colemickens>
if anyone has experience with laser cutting the nix logo into something, I want to do the same when I get my new laptop (maybe soon? rumor is they're announced on monday)
<colemickens>
Although, I probably won't be able to go to any maker space any time soon... I keep forgetting.
<samueldr>
what new laptop is "announced on monday" per the rumor?
<evanjs>
adisbladis: feel free to add any errors I missed in my initial write-up if you do end up having time or etc
<cole-h>
colemickens: I'm afraid to ask about the rumored price range
<adisbladis>
evanjs: Once I added usbutils to the closure it's actually working :)
<evanjs>
wtf wait without creating the directory??
<adisbladis>
evanjs: No, still have to create the directory manually
<adisbladis>
But can't enter the effects tab
<evanjs>
adisbladis: oh okay and derp that makes sense :D
<evanjs>
Yeah if you switch to console view you might see errors like "cant get current status"
<evanjs>
forgot to mention that prints it out again when switching tabs
o1lo01ol1o has joined #nixos
<evanjs>
interesting, though. I didn't need usbutils, but maybe that's because it's installed systemwide :D
elux has quit [Quit: leaving]
<adisbladis>
Man.. These applications are buggy :P
reactormonk has quit [Ping timeout: 255 seconds]
<evanjs>
Ah wait it'
<evanjs>
it isn't printing anything out when switching tabs. And yeah, I chose a dev version because it was way easier to package, might be somewhat related :P
<{^_^}>
[nixpkgs] @jonringer pushed commit from @zimbatm to master « folding@home: 6.02 -> 7.5.1 »: https://git.io/JviuA
<adisbladis>
Same same
<adisbladis>
And tbh I don't really care enough about the colour of my periphirals to debug :P
<adisbladis>
Hm, I just got a really cool idea
jonten has joined #nixos
aveltras has joined #nixos
blaggacao has quit [Ping timeout: 258 seconds]
elux has joined #nixos
alexherbo27 has joined #nixos
Jonathan87 has joined #nixos
alexherbo2 has quit [Ping timeout: 258 seconds]
alexherbo27 is now known as alexherbo2
<elux>
im trying to use https://gist.github.com/pkieltyka/67d5eff53ea38ecc75dd73bf86bfe526 (hyper.nix) in my home and import it with `hyper = pkgs.callPackage ./hyper/default.nix { };` and use it as a pkg to install. This will build with nix and install, but when I run `hyper` it returns the error: Error: `node-pty` failed to load. Typically this means that it was built incorrectly.
<elux>
seems i need the `node-pty` pkg, but doesnt seem like this is part of the nodePackages .. any suggestions on how to add this to nodePackages as part of this nix?
<Jonathan87>
Gah! NixOS is the most amazing operating system in existence. But right now I want to strangle someone because of how impossible it sometimes seems to do a basic task.
<zeta_0>
in order to add non-haskell pkgs to this default.nix file, do i just need to place pkg.my-package next to the haskell packages? https://hastebin.com/razimeyafi.bash
<Jonathan87>
zeta_0: I believe you need it next to `cabal-install` and `cabal2nix`
<Jonathan87>
zeta_0: does it work when you put it with them?
<Jonathan87>
I am trying to build a C library to use with a D application but the build instructions on the git repo end up with build errors. I am not very familiar with D nor Nix enough to know how to coorerse the two to work together.
<Jonathan87>
I have a nix-shell with what are supposed to be the dependencies, am I just going about it wrong, or am I missing something?
<Yaniel>
do you also have pkg-config?
<Yaniel>
the paths of dependencies are not the same as usual so the configuration scripts need some help from pkg-config usually
od1n has joined #nixos
<od1n>
hello, when trying to start a machine with an iso attached, libvirtd chowns the iso with the user it runs under (root), which should not be a problem - yet I get permission denied on accessing the iso. there's a ticket on this from 2018 (https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1784001). did anyone else hit this?
b has joined #nixos
<od1n>
tried qemuRunAsRoot = false already
<evanjs>
adisbladis: yup haha. on another note... moved to python3Packages.callPackage, but for some reason it can't find build.ninja now lol. hrm. wonder if meson isn't doing its thing
<adisbladis>
I dunno
<adisbladis>
I'm battling dbus :(
<adisbladis>
Trying to get my emacs to talk to openrazer
<adisbladis>
I had this idea to set colours based on flycheck and erc buffers
rsoeldner has quit [Remote host closed the connection]
rsoeldner has joined #nixos
<evanjs>
oooo that sounds interesting
<evanjs>
I've been meaning to look into how I can tie my WM and RGB stuff together more
<adisbladis>
I don't know why that snippet isn't working..
sudoforge has quit [Ping timeout: 256 seconds]
<adisbladis>
I can call other services on the session bus just fine
linarcx has joined #nixos
delli3 has quit [Read error: Connection reset by peer]
jtojnar_ has quit [Read error: Connection reset by peer]
<evanjs>
adisbladis: like exwm?
<adisbladis>
Yep
<evanjs>
I remember... I think it was ashkitten? experiencing similar issues. I want to say enabling services.dbus.socketActivated helped but I can't quite remember
jtojnar_ has joined #nixos
<zeta_0>
i am trying to setup a default.nix file to develop darcs with but it is throwing errors, could you guys help me fix it, thanks: https://hastebin.com/umiduxujeb.bash
<cole-h>
Anybody know the best way to overlay a package inside a custom NixOS ISO? e.g. apply the overlay to the ISO boot environment (and also the subequently installed system)
<evanjs>
does python3Packages (vs callPackage) interfere with meson/ninja or something like that?
<evanjs>
*python3Packages.callPackage vs pkgs.callPackage
<{^_^}>
[nixpkgs] @alyssais pushed commit from @jpotier to master « bitlbee-mastodon: 1.4.2 -> 1.4.3 »: https://git.io/Jvizx
alexherbo2 has quit [Ping timeout: 265 seconds]
<tomberek>
Anyone familiar with rust packaging? I've got a rust build that errors out with "fatal error: 'errno.h' file not found." I've added glibc.dev (and glibc, just in case) to buildInputs in buildRustPackage. Not sure why errno.h is not available to the build. Specifically this fails while building the dav1d-sys crate.
<tomberek>
evanjs: The build gets farther than when i started, been adding deps one by one, so I'm making progress, but this errno.h doesn't seem to respond to being provided glibc.dev.... okay, looking...
<kaliumxyz>
I'm not happy, I need to find some error else this PC is just cursed.
<kaliumxyz>
which is bad.
<kaliumxyz>
downgrade to an older linux kernel and pray/
<evanjs>
tomberek: yeah typically errno/glibc shouldn't need to be added IIRC
<evanjs>
tomberek: what does your current expression look like, anyhow?
<{^_^}>
[nixpkgs] @alyssais pushed commit from @r-ryantm to master « tautulli: 2.1.44 -> 2.2.0 »: https://git.io/JvigG
<chrisaw>
I'm trying to create an overlay of a few interdependent python packages but I'm struggling to use one of my custom python packages as a dependency for another. Can anyone help? I've posted a quick Gist here: https://gist.github.com/cawilliamson/b4f22189fa5431a7cfc533f90b91087e
<{^_^}>
[nixpkgs] @Ekleog opened pull request #82614 → xfce4-12: remove alias → https://git.io/JvigZ
knupfer has joined #nixos
<yorick>
what are us mortals who constantly hotplug ethernet dongles supposed to do about networking.useDHCP deprecation?
<adisbladis>
evanjs: I'm going insane :S
cosimone has quit [Quit: Quit.]
reactormonk has quit [Ping timeout: 268 seconds]
<adisbladis>
I don't understand this..
<{^_^}>
[nix] @LnL7 opened pull request #3415 → build: disallow root without build users → https://git.io/Jvigc
<genesis>
thanks, people would be happy on that one
opticnerve has quit [Client Quit]
<genesis>
but dbus speaking, not easy to satisfy completly.
<adisbladis>
Holy hell, this is a rabbit hole
<adisbladis>
5 rabbit holes a day keeps the doctor at bay
<adisbladis>
I hop
<adisbladis>
e
<genesis>
more you learn about dbus, the more you see it's completly broken stuff.
linarcx has quit [Ping timeout: 256 seconds]
<genesis>
then kbus dead too, no more on the rabbit hole.
fling has joined #nixos
<genesis>
there is a rewrite, that pretend to have some solve
Neo-- has joined #nixos
reactormonk has quit [Ping timeout: 268 seconds]
<adisbladis>
I think I'm gonna admit defeat and do something more meaningful
<danderson>
adisbladis: that's how I feel working through a small piece of the nixpkgs bug backlog. It's a huge field full of wonderful little rabbit holes :)
<danderson>
but it's nice. Kinda relaxing, in a meditation kind of way. Chipping away at a tiny piece of a huge mountain.
<genesis>
because spending time on white trailing space (today i've 2 / trailing, yesterday a commit that should splice in 3 etc)
<genesis>
see how we spend humane time.
<genesis>
improve dbus service providing would be much more interesting btw
<adisbladis>
danderson: What's really cool about Nix is that it lets you enter the FOSS rabbit hole at almost any point in the maze ;)
<genesis>
(since noone look at contributer review stat, but only their friend doing-something-stat, i wonder how many good contributer nixpkgs missed to upgrade)
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
genesis: What does that mean?
<{^_^}>
[nixpkgs] @adisbladis merged pull request #82453 → Python: introduce NIX_PYTHONPREFIX in order to set site.PREFIXES → https://git.io/Jv6Pr
<MichaelRaskin>
Refactoring often requires getting a consensus (or at least indifference) on trade-offs, have I mentioned it is hard to make a decision quickly in Nixpkgs?
<gchristensen>
as a person who does sometimes merge PRs, I don't look at who wrote a review - just what it says and if I agree with their points. if I agree, I note who said it for later
<MichaelRaskin>
Well, I would say that Nixpkgs is an impressive project from the point of view of collaboration despite voiced disagreements
<MichaelRaskin>
gchristensen: I guess that last part mostly happens if you do not recognise them, though?
<adisbladis>
gchristensen: Exactly :)
<genesis>
anyway, i think review is a lost of time for people with no commit bit.
<gchristensen>
MichaelRaskin: right :)
<LnL>
yeah, also a review from somebody that knows the specific tool/project is often more useful than their nix knowledge
<gchristensen>
genesis: well, you can think that, but I don't think anybody else thinks that
<genesis>
i speak from experience, you have commit bit for years i guess, so are you sure it's false?
<genesis>
LnL : people with no knowledge but with commit bit can close your PR ... that happens too
<gchristensen>
as someone with commit bit, reviews form people who don't have it are useful to me -- so not sure how it is possible for it to be lost time
<gchristensen>
not all PRs should merge
<genesis>
for example , i've some experience of appimage since i'm the 3rd contributor
<gchristensen>
yeah, appimage is going to be hard since they generally go against the Nix model of things
<genesis>
i review ripcord one, get some good result
<genesis>
was just a recent example since less than 2 days
<gchristensen>
yeah, it happens
fenedor has quit [Read error: Connection reset by peer]
<MichaelRaskin>
Hm, self-merge of a leaf package addition? Yeah, that happens, the PR seems to be just a way of running basic ofborg checks
<danderson>
to try and provide a counter-example: I've been going through bug backlogs in the last couple of days. Response time is not instant, 5-minute turnaround, but on a timescale of a few days I'm getting what I need to make progress.
<danderson>
it's slower than I would like, but nixpkgs is also huge with a lot of activity. Given that, the turnaround is quite good. I just have to keep notes to remind me to ping if something falls through the cracks.
<cole-h>
Gonna ask once more, then wait for it to become more active here if nobody has any ideas :P
<cole-h>
Anybody know the best way to overlay a package inside a custom NixOS ISO? e.g. apply the overlay to the ISO boot environment (and also the subequently installed system)
<danderson>
sphalerite: usually I have more context than that in my notes. My notes also track bugs where I asked for information or feedback
<sphalerite>
danderson: ah, fair enough
<MichaelRaskin>
cole-h: just add an overlay and build the ISO? or what are you trying to do
<danderson>
I guess my point is, it works fine. I just need to remember to check back with old stuff occasionally, because I tend to forget about stuff easily :)
<sphalerite>
cole-h: not a perfect answer to your question, but if you need that level of persistence I'd probably use a nixpkgs fork instead
<cole-h>
MichaelRaskin: Build an ISO with said overlay applied (so that the install using that ISO will have that overlayed package instead of the nixpkgs one)
<cole-h>
sphalerite: Probably a good idea if the resident wizards have no idea
<MichaelRaskin>
Hm, true, might be easier to just build using a fork
<cole-h>
MichaelRaskin: My use case is: I want to build the ZFS PR adding zstd compression support and install my system with that
<cole-h>
But since that PR is still in-progress, I can't just PR it to nixpkgs and be done :P
<MichaelRaskin>
Oh, it is FS, so you do indeed need overlayed system all the way and cannot just rebuild later
<colemickens>
Why not just build the ISO from your fork instead of overlaying?
<colemickens>
It should copy your nixpkgs to the iso as teh channel and thus install with that nixpkgs too, right?
<cole-h>
Is that how it works? That might be what I want to do, then.
blaggacao has quit [Ping timeout: 240 seconds]
<colemickens>
I think it basically has to be how it works.
<colemickens>
Someone can correct me, but otherwise you'd have to reference some other channel for it to use to do/seed the install
<cole-h>
Clarification: so it will use the `nixpkgs` from my NIX_PATH as the base for packages installed into the system?
<cole-h>
What would happen if I `nix-channel --update`? Would my fork's changes be clobbered?
<colemickens>
Idk don't use channels.
<sphalerite>
colemickens: yes, it would
<sphalerite>
err cole-h
<sphalerite>
your shared prefix is too long!
<colemickens>
sphalerite what does the installer do for the initial nix-channel registration?
<cole-h>
Alright. Well, since it is just for playing around in a VM at the moment, shouldn't matter too much
<sphalerite>
colemickens: it copies the channels profile from the installer system, which in turn is copied from pkgs.path when the installer system is built
lovesegfault has quit [Quit: WeeChat 2.7.1]
<colemickens>
yeah, just pulled up channel.nix and see that
<sphalerite>
(or if you're installing from a non-installer system, it's just whatever's in channels at nixos-install time)
vykook- has quit [Ping timeout: 272 seconds]
Thra11 has quit [Ping timeout: 268 seconds]
<colemickens>
trying to understand the last bit, is that where the channel is actually registered? Trying to figure out where its' going to pull from when someone runs nix-channel --update with a custom nixpkgs.
<colemickens>
anyway, cole-h, by time you've done the install, I'd recommend using a custom nixpkgs and not using channels anyway.
<colemickens>
it's pretty easy to just always use your own nixpkgs, and if you're going to use nixpkgs and contribute, you really probably just want to do that anyway
<cole-h>
(MichaelRaskin: that's what spectrum is :P)
<gchristensen>
for context, the removal of genesis is temporary, and in response to their comments about a dearly appreciated contributor.
<MichaelRaskin>
cole-h: very much no
symphorien has joined #nixos
<kaliumxyz>
any way to do pinning of the nix channel in the nix config?
<MichaelRaskin>
Spectrum is KVM
<jumper149>
Hi, is it possible to hide plain-text passwords from the nixos-option query?
<MichaelRaskin>
And apparently a lot of Wayland more or less as is
o1lo01ol1o has joined #nixos
<cole-h>
Maybe I've misunderstood spectrum then.
<MichaelRaskin>
I mean, there are similarities
vykook has quit [Ping timeout: 255 seconds]
<danderson>
spectrum seems to be using crosvm, which supports virtio-wayland to let isolated apps still share a GUI system
<danderson>
(from a very quick read recently)
blibberblob has joined #nixos
<MichaelRaskin>
I also hope SpectrumOS ends up sharing chunks of store instead of building hermetic VMs
<cole-h>
MichaelRaskin++ Thanks for correcting me.
<{^_^}>
MichaelRaskin's karma got increased to 21
<cole-h>
And colemickens++, sphalerite++, and MichaelRaskin++ for the ISO assistance earlier (still in-progress though)
<{^_^}>
MichaelRaskin's karma got increased to 22, colemickens's karma got increased to 17, sphalerite's karma got increased to 79
o1lo01ol1o has quit [Ping timeout: 260 seconds]
blaggacao has joined #nixos
<misuzu>
kaliumxyz: not really, i had to write a script that outputs something like this `-I nixpkgs=/nix/store/8x90sd5m33zhz96z72zgydgy11nc6wb2-source -I home-manager=/nix/store/pk81mh4djhpzbf5jqbgnfspn1517zckk-source`
vykook has joined #nixos
<misuzu>
and then running it like this: `nixos-rebuild $(deploy/echo-nix-path.sh) switch`
<misuzu>
but this does not cover channels used by nix-env
<adisbladis>
MichaelRaskin: There were discussions around how to achieve that at the CCC camp
<{^_^}>
[nixpkgs] @jonringer pushed commit from @jluttine to master « pythonPackages.sphinxcontrib-katex: init at 0.5.1 »: https://git.io/Jvi2b
<notgne2>
I think my brother in law had something similar to that, and I had to package some random driver off github to make it work, lemme figure out if it's the same deal
<blibberblob>
Also I had this guide but I'm not sure how easy it is to adapt to nixos
<blibberblob>
I don't know much about nixos under the hood (i.e. how it handles drivers etc)
<notgne2>
nixpkgs has rtl8812au, but it didn't work for me, so what I did is override it to use a different source, I think I'm a bit out of the loop when it comes to these drivers but eventually after trying 4 or 5 different repos one worked
vykook has joined #nixos
<makefu>
blibberblob: rtl8812 is already packaged, no? you should be able to enable the kernel module via "boot.extraModulePackages = [ config.boot.kernelPackages.rtl8812au ];"
o1lo01ol1o has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
Thra11 has quit [Quit: WeeChat 2.7]
<makefu>
it worked for me at least
<blibberblob>
makefu: thanks a ton, I'll give that a shot
<NobbZ[m]>
As far as I remember, you need to blacklist all the other rtl kernel modules, when you have a EU variant… Autoloading will always pick the wrong one…
<notgne2>
there's also a chance that NixOS already fixed the issue, because last time I checked was a long time ago
<cole-h>
Mic92: ping
symphorien has joined #nixos
<NobbZ[m]>
Hmmm… Though if "Au" isn't a typo, things can be totally different to what I just said about "Eu"…
<blibberblob>
notgne2: thanks!!
<notgne2>
fwiw the device I was testing it with was purchased in the US, but I'm not sure what the regions really mean for these things
wrl has joined #nixos
<NobbZ[m]>
Regions for wireless technology is 90% about frequency ranges.
<blibberblob>
I'm in the US as well
<NobbZ[m]>
The remaining 10% are probably allowed transmission power
* adisbladis
always sets US
<adisbladis>
Higher transmission power and channel 13
__monty__ has quit [Quit: leaving]
<NemesisD>
hi all. i'm trying to build a tool that uses haskell stack. stack seems to understand its on nixos and is trying to get nixpkgs.ghc821. however i think that version isn't in the packageset anymore. any idea how i could backport (or forwardport) it into my nixpkgs?
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<blibberblob>
hmm, adding that extraModulePackages didn't work... wpa_supplicant failed to start
<notgne2>
NemesisD: you could add the older nixpkgs release as a channel and do something like `(import <oldpkgs> {}).ghc821`
vykook has quit [Ping timeout: 258 seconds]
<colemickens>
Isn't virtio-fs in 5.4?
<blibberblob>
/sys/class/net doesn't show the wireless card...
<notgne2>
or even `(import (builtins.fetchGit { url = ".."; rev = ".."; }) {}).ghc821`
h0m1 has quit [Quit: WeeChat 2.7.1]
<NemesisD>
notgne2: in that version would i still need to add the channel?
h0m1 has joined #nixos
<sphalerite>
misuzu: yes, on the condition that you have nix on the system and are running systemd
<{^_^}>
[nixpkgs] @jonringer pushed commit from @B4dM4n to master « cargo-udeps: 0.1.9 -> 0.1.10 »: https://git.io/JviaT
<sphalerite>
misuzu: and the kernel supports everything you need
<notgne2>
NemesisD: with the bottom one you can just fill in the details for fetching your desired nixpkgs commit from git, channels are just a way of managing this externally if you want to (so you can upgrade them "impurely")
o1lo01ol1o has quit [Remote host closed the connection]
<notgne2>
not using a channel just means you will need to update the rev/hash every time you want it to update
aveltras has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « fio: 3.18 -> 3.19 »: https://git.io/Jviak
<NemesisD>
notgne2: ok that might be usable. i guess i can go through old versions until i find it. here's the trick though, stack is somehow demanding the ghc821, i'm not writing it anywhere. so my best guess is i'd have to merge this into my nixpkgs somehow so t can see it
<NemesisD>
or rather i probably don't understand at all how stack is automagically trying to install ghc
<blibberblob>
I feel like I must be missing some step that's preventing nixos from detecting this wireless card...
vykook has joined #nixos
<NemesisD>
all i get is error: attribute 'ghc821' missing, at (string):1:43
symphorien has quit [Quit: WeeChat 2.7.1]
<notgne2>
if you have something like `pkgs = import <nixpkgs> {};` you can do `pkgs = import <nixpkgs> { overlays = [(self: super: rec { ghc821 = <code from earlier to fetch it>; })]; };`
<notgne2>
I actually don't really remember what either boot.extraModulePackages or hardware.firmware are for, or the differences between them, but I guess it was added to both of them
<notgne2>
probably only 1 is needed
<blibberblob>
haha ok... lemme play around with that :) thanks a bunch
<MichaelRaskin>
colemickens: One has been usable since forever, as 9p support in kernel is old and stable. The new fancy FUSE-like one — I might have missed that it has landed recently, but I would be a bit surprised
<notgne2>
to be clear about that snippet, you'll need to put the `let .. in` code at the top of your configuration.nix, and the stuff inside `{ .. }` where you would normally expect
<danderson>
virtio-fs is quite exciting, assuming it ends up being small enough to audit
<tomberek>
but the lack of a Cargo.lock seems to be stiffling my progress
blaggacao has quit [Ping timeout: 240 seconds]
<danderson>
I look forward to giving my VMs ZFS directories to use instead of zvols :)
<MichaelRaskin>
colemickens: oh they did push it through, interesting, thanks
<colemickens>
tomberek I have found a number of rust projects need a "Please commit a Cargo.lock" issue, especially depending on when they were initially "cargo init"d.
<tomberek>
colemickens: i've been trying to patch that fix into the src
<colemickens>
yeah, that's what I do short term too.
symphorien has quit [Quit: WeeChat 2.7.1]
<blibberblob>
still no luck, it's just not detecting the card
<blibberblob>
I don
<tomberek>
colemickens: but i still must be doing something wrong, after generating a local Cargo.lock and using "src = ./.;" i get an updated cargoSha256, after that the build keeps saying "cargoSha256 is out of date". Even though it's right there.
<blibberblob>
I don't have really any expertise with what's going on under the hood
<blibberblob>
it's detecting the USB, but it can't seem to figure out that it's a wireless card
<notgne2>
blibberblob: what kernel version are you on
<colemickens>
tomberek: hm, not sure, I usually clone the repo, generate the Cargo.lock, and then git add/commit/format-patch and include that as a patch in the derivation.
<colemickens>
Or just generate the lock, copy it to the package dir, and then copy it in prePatch to the src dir.
abathur has quit [Ping timeout: 255 seconds]
<blibberblob>
notgne: 5.4.24
<colemickens>
(and not sure I totally understand how src = ./.; would work)
<tomberek>
this repo has lots of subdirectories with their own Cargo.toml's, do those need their own Cargo.lock (i'm new to Rust packaging)
<notgne2>
huh, that's the version it was working for me on
<colemickens>
I can't say, I'm actually literally working on splitting my project into a multi-project repo and nix-ing it, so maybe I'll find out soon.
symphorien has joined #nixos
<notgne2>
blibberblob: it might be worth experimenting with trying some different git repos, if there was a one-size-fits-all nixpkgs would probably already have it (and have solved this problem)
vykook has quit [Ping timeout: 240 seconds]
<blibberblob>
I'm just pretty lost because I don't know how network interface detection or Linux device drivers work
<notgne2>
lol me neither
<{^_^}>
[nixpkgs] @m-scr opened pull request #82619 → unison: fix build with ocamlPackages_4_09 → https://git.io/JviaE
<danderson>
If it's USB, the happy path is that the device reports itself as a generic, well-known device type through USB
<danderson>
and then it all just works.
<danderson>
Sounds like this dongle isn't doing that, and needs explicit support of some kind
<blibberblob>
How does that reporting process work?
<blibberblob>
It does show up in lsusb
<notgne2>
blibberblob: you could try `hardware.enableRedistributableFirmware = true;` too
<danderson>
in `lsusb -v`, I think it'd show up as some kind of CDC (Communications Device Class) descriptor
<danderson>
which is USB-speak for "I'm presenting you a generic interface for a well known kind of device (e.g. mouse, keyboard, cellular modem...)"
<danderson>
The alternative is that lsusb -v doesn't show any interesting CDC descriptors, and instead just lists a Vendor ID and Product ID
fendor_ has joined #nixos
<tomberek>
colemickens: does "cargo generate-lockfile" only create lockfiles at the top of a git repo? (is it git-aware?) I've tried to generate .lock's in the subdirs, to no avail.
o1lo01ol1o has joined #nixos
<danderson>
then what would normally happen is that udev+kernel (not sure which one precisely) would load the correct driver for that vendor/product, and it handles the custom communication
<danderson>
Does `dmesg` show anything other than generic USB "device attached" messages?
<danderson>
That would be how I'd see if a specific driver noticed the device and grabbed it
<danderson>
(and also where I'd look for errors like "waaah I don't have a firmware for this", which is where enableRedistributableFirmware=true comes in)
<blibberblob>
danderson: dmesg has a lot of things in it
<danderson>
well dmesg is the kernel log for everything, so yeah. If you unplug/replug the USB device, you should see a burst of messages at the bottom of dmesg about it
<blibberblob>
ah ok lemme try
<danderson>
I forget the exact messages, but it'll be something about USB detecting a new device and configuring it
<colemickens>
tomberek: sorry, we're out of my depth of rust experience here
fendor has quit [Ping timeout: 255 seconds]
<danderson>
if you can paste those into gist.github.com or some other pastebin, I can take a look and see if anything jumps out at me
<danderson>
(I'm by no means an expert on any of this, but I've done my share of debugging "augh why device no work", so I'm happy to try :) )
<danderson>
those messages are all generic USB stuff
<blibberblob>
ah
o1lo01ol1o has quit [Ping timeout: 258 seconds]
<danderson>
the way USB works is that there's 2 levels: the lower layer is just "oh look a new device, I'll ask it who it is and what it does"
symphorien has quit [Quit: WeeChat 2.7.1]
<danderson>
and then there's a higher layer of "oh, I understand what this device is, I'll load a driver for it"
<blibberblob>
so the first part worked but not the second?
<danderson>
the messages you're seeing there are all the lower layer USB handshakes
<blibberblob>
I see
<blibberblob>
is there another log I can look at for that second part?
<danderson>
i.e. your kernel is going "well, it's definitely a USB device, and here's the vendor and product ID, and various strings I pulled out of its device descriptor"
<danderson>
but if there's no other messages, that means the process stopped there - nothing else went "oh, I know about that device, I'll take it!"
<danderson>
so the process just stopped
<danderson>
so, there's two main possibilities here: you're missing the driver for this device, or there is no linux driver for this device yet
vykook has joined #nixos
<danderson>
let's try to find out which one it is...
<danderson>
The RTL8812au driver from github can be compiled with support for RTL8814au"
<danderson>
okay, so, now, we have to figure out if NixOS already has that somewhere, or if you have to do some work
<blibberblob>
I think that driver exists
<blibberblob>
Or at least something with that name does, which I put in my `boot.extraModulePackages`
<blibberblob>
yeah that's the same card :)
rsoeldner has quit [Remote host closed the connection]
<danderson>
what's the module name you put in extraModulePackages ?
rsoeldner has joined #nixos
<danderson>
I'm reading more, and it looks like the "default" rtl8812au doesn't support rtl8814au, but the aircrack-ng driver (https://github.com/aircrack-ng/rtl8812au) does
<{^_^}>
[nixpkgs] @alyssais pushed commit from @r-ryantm to master « eksctl: 0.13.0 -> 0.15.0 »: https://git.io/JviaH
<danderson>
So at this point, it looks like the driver for this chipset is super new and bleeding edge, so I'm guessing the nixpkgs module isn't right...
<{^_^}>
[nixpkgs] @alyssais pushed commit from @r-ryantm to master « clash: 0.17.1 -> 0.18.0 »: https://git.io/Jviad
<blibberblob>
building now
<danderson>
grepping through that repo, I _do_ see ifdefs for RTL8814A_SUPPORT
<danderson>
so it looks like there is some support... Maybe the derivation just needs updating? Let's see if that logic is in the version nixpkgs builds...
rardiol has joined #nixos
<sphalerite>
Ericson2314: I just got a build to work by setting NIX_aarch64_unknown_linux_gnu_CFLAGS_COMPILE=-Wno-error=address-of-packed-member — should I ever need to do that?
blaggacao has joined #nixos
<danderson>
blibberblob: okay, in the source code, that product ID is not one of the listed supported ones :(
<blibberblob>
it seems to have found it tho :))))
<danderson>
so the nixpkgs driver definitely won't work right now.
<danderson>
oh, the aircrack-ng driver?
<blibberblob>
with that aircrack-ng driver
<blibberblob>
yeah
<danderson>
awesome!
<NemesisD>
if i'm using the `imports` attribute to break up my nixos config, how would i append things to say `environment.systempackages` in some of those files?
<blibberblob>
now I just need to enable wireless and hopefully it works....
<Yaniel>
NemesisD: just append whatever in what file you like
<danderson>
it looks like they already have 8814 chipset support, so it might be as simple as "add the product ID to the list"
<blibberblob>
danderson: is that preferable to updating nixpkgs to use the aircrack-ng repo?
<danderson>
I think it is, because the current nixpkgs driver supports a lot of different chipsets (5-6 rtl chips)
<danderson>
so if it's a simple fix, it would be nice to just add your dongle to that
<blibberblob>
for sure, that's a good call
zeta_0 has joined #nixos
<danderson>
if you file a bug with `lsusb -v` output, and mention that the aircrack-ng driver seems to work, the maintainers can probably figure it out
<danderson>
and then we can update the nixpkgs derivation and nobody else has to deal with this ever again :D
vykook has joined #nixos
<blibberblob>
hahaha ok will do
<blibberblob>
now it's failing to find /etc/wpa_supplicant.conf but hopefully that's an easier fix
kaliumxyz has quit [Quit: leaving]
<blibberblob>
I've gotta run but this was SOOO helpful cc: danderson, notgne2, makefu
<danderson>
hah yeah
<NemesisD>
Yaniel: oh! that's pretty cool. thanks
<danderson>
if you're joining a regular wpa2 wifi network, I can give you a trivial wpa_supplicant.conf, one sec...
<blibberblob>
ohh that'd be helpful
<sphalerite>
blibberblob: as can wpa_passphrase $ssid $psk
<sphalerite>
:)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<danderson>
oh yeah, for a quick test that's probably easier
<{^_^}>
[nixpkgs] @aanderse opened pull request #82620 → nixos/ssh: silence ssh-keygen during configuration validation → https://git.io/Jviaj
Rusty1 has quit [Remote host closed the connection]
<danderson>
replace MYSSID with your network name and passphrase with the wpa password
<blibberblob>
ok, wpa_supplicant started correcly
<blibberblob>
if I just yank my ethernet cable presumably it'll just keep working? or is there a better way to test that I'm connected successfully
<danderson>
and if you're lucky, `dhcpcd <interface name>` will give you an IP
<danderson>
and then everything will work
<danderson>
careful, wpa_supplicant only connects you to the wifi network, it doesn't do DHCP :) You need to run `dhcpcd` separately
<blibberblob>
oh jeez
<danderson>
once you've verified that this all works, you can do all this in nixos configs as well
<blibberblob>
kk
<danderson>
this is just crappy manual testing to check that it works :)
<blibberblob>
no such file /etc/dhcpcd.conf
vykook has quit [Ping timeout: 260 seconds]
<danderson>
hm, was that a warning or an error? Does `ip addr show <wlan interface name>` show an IP addr?
<cole-h>
tomberek: AFAIK, Cargo only generates lockfiles in the root of the project (aka next to the Cargo.toml)
<danderson>
(I have to go soon here, but hopefully you at least have some clues for how to continue)
<blibberblob>
hmm, the interface appears in ifconfig but the `ip addr` command says the device does not exist
<tomberek>
cole-h: Yeah, found that out, but things still didn't work. Copying the Cargo.lock to be next to every sub-directory Cargo.toml seems allow the build to go a lot further.
reactormonk has quit [Ping timeout: 260 seconds]
<tomberek>
( this repo has a bunch for Cargo.toml's in a bunch of subdirectories )
<{^_^}>
[nix] @Ericson2314 opened pull request #3416 → WIP more rust → https://git.io/JviVv
<cole-h>
So if you have multiple workspaces, only the toplevel Cargo.toml will have an accompanying Cargo.lock with all the deps of the workspaces inside (IIRC)
<blibberblob>
nm found it, just was a formatting issue
<blibberblob>
ip says it's down, even if I run `ip link set dev up` it still says it's down
<{^_^}>
[nixpkgs] @aanderse opened pull request #82621 → nixos/sshd: only include AuthorizedKeysCommand and AuthorizedKeysComm… → https://git.io/JviVU
Neo-- has quit [Ping timeout: 258 seconds]
<blibberblob>
alright well I gotta go but again thanks, seems like I'm getting close :)