<simpson>
It's pretty funny. Well done, for sure. I like how, from the start, there are extra packages not specified in LSB.
<gchristensen>
interesting, that is exactl ythe comment I had typed out
<gchristensen>
almost.
werner291 has quit [Quit: werner291]
Heirlung has joined #nixos
<lovesegfault>
gchristensen: That gave me a yikes, not sure why
<qyliss>
hmm yeah not sure I like that
<qyliss>
I'm not sure encouraging people to just enable that and go back to impure binaries is a good idea
<qyliss>
Which is what its presence will do
<gchristensen>
I'm on board with UX improvements, or a way to make a steam-run-like-thing more obviously the right answer, this feels bad
<lovesegfault>
Could we have an `lsb-run` instead?
<gchristensen>
that seems like a nice idea
<lovesegfault>
Also, we're two successful builds away from having trunk-combined all green on Hydra :D
<simpson>
lovesegfault++ I like that idea, especially if we actually limit it to LSB.
<{^_^}>
lovesegfault's karma got increased to 10
<lovesegfault>
simpson: Yeah, I wonder whether a setup like python.withPackages would be possible
mexisme_ has joined #nixos
<lovesegfault>
So you could do nix-shell -p lsb-run.withPackages ( pkgs: with pkgs; [ hello ] )
<lovesegfault>
Something like that
TwoNicToupper has joined #nixos
<simpson>
Interesting.
gustavderdrache has joined #nixos
<gchristensen>
sounds a lot like the FHS runner we have too
mexisme has quit [Ping timeout: 260 seconds]
wildtrees_ has quit [Quit: Leaving]
<lovesegfault>
gchristensen: kind of, yeah, just in cli-tool form
<gchristensen>
maybe there could be an autofhs CLI tool which uses `ldd` and fills in library dependencies with nix-locate
qubasa_ has joined #nixos
wavirc22 has quit [Ping timeout: 268 seconds]
wavirc22 has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvOG8
qubasa has quit [Ping timeout: 240 seconds]
genesis has quit [Ping timeout: 272 seconds]
fiddlerwoaroof has joined #nixos
<lovesegfault>
gchristensen: I fear that being misleading for non-linked deps
<gchristensen>
maybe so
<gchristensen>
qyliss instructed me to go take a break, so I'm going to do that
<drakonis>
qyliss: i'm cant agree that it is a bad thing, it gets rid of the annoying hoops that a user has to jump to use one off binaries that don't play ball with nix's structure
<samueldr>
I've had a good success rate, though minimal actual uses, of steam-run to run non-steam games
<samueldr>
why not lsb-run ?
<drakonis>
there's binaries that i have to use patchelf to run non game binaries
<drakonis>
the hoop jumping isnt fun because it broke the binaries
* samueldr
now catches up with scrollback
<evanjs>
drakonis: cases where steam-run doesn't work, either?
<drakonis>
yes
<evanjs>
dang
<samueldr>
drakonis: was it because steam-run was missing some libs?
<drakonis>
i have to chuck it into a container to make it work
<drakonis>
some libraries? there was a lot of proprietary libraries there
<samueldr>
rephrasing the question, what was the failure case that made it fail under steam-run?
<drakonis>
this was circa 2018, i won't remember the exact details
<drakonis>
its a xilinx editor i think?
<{^_^}>
[nixpkgs] @chkno opened pull request #78799 → gnupatch: Don't fetch from cgit URLs with unstable hashes → https://git.io/JvOGM
<drakonis>
the issue was the proprietary bits
valdosti has quit [Remote host closed the connection]
<thongpv87>
Does anyone use Thinkpad X1 Extreme (2nd). How can I make external display work with the machine? I tried https://github.com/NixOS/nixos-hardware, it take effect on the machine but external monitor still not working.
h0m1 has quit [Ping timeout: 272 seconds]
h0m1 has joined #nixos
marcusr has quit [Remote host closed the connection]
<petersjt014[m]>
<thongpv87 "Does anyone use Thinkpad X1 Extr"> If you're following exact instructions and not getting the expected results I'd submit an issue to the github page. Might be that you missed something, but hardware is picky so maybe not.
<thongpv87>
I'm new to NixOS so I thing I missed st. I just install nixos, configure network and GUI in the configuration.nix. Then I booted into the machine, run add channel and update chanel, then add the <nixos-hardware/lenovo/thinkpad/x1-extreme/gen2> into the configuration.nix and rebuild + reboot the machine
<thongpv87>
I noticed that the font size get bigger, and battery indicator work correctly
<thongpv87>
So I thing the hardware configuration really take effect on the machine
<thongpv87>
When I connect the external monitor to the thunderbolt port and HDMI port, it show the external monitor display mode (extend, external monitor only, duplicate) on the machine
<thongpv87>
I choose duplicated, but nothing displayed on the external monitor
<thongpv87>
When I connected the keyboard and mouse to the external monitor, then connect to the laptop via thunderbolt port, I could use these devices normaly
<Enzime>
I have: mkOption { type = types.attrsOf types.str; }
<Enzime>
is it possible to make it allow strings as well as null?
<petersjt014[m]>
Personally I'd avoid that if possible--the 'billion dollar mistake' (Horae's words, not mine) is a sneaky little bugger. If there was an Enum/Union type i'd suggest that.
<petersjt014[m]>
I am pretty sure what you want is possible though
<petersjt014[m]>
the nix manual contains things like `openssl = if sslSupport then openssl else null;`
<peelz>
does anybody know why memtest86+ fails to boot (hangs) with boot.loader.grub.memtest86.enable = true;
Rusty1 has quit [Quit: Konversation terminated!]
chagra_ has joined #nixos
<petersjt014[m]>
Enzime: try something like `types.str or types.null`. I can't test that rn but it seems like the intuitive answer.
<petersjt014[m]>
peelz: Does it dump an error in any logs that you know of?
<peelz>
logs?
<petersjt014[m]>
I think they'd go in either dmesg or /var, hold on
chagra has quit [Ping timeout: 268 seconds]
<petersjt014[m]>
Hmm, not sure which if either of those
<petersjt014[m]>
dmesg logs everything hardware-wise from boot
<peelz>
uhh
<peelz>
well seeing as that I have to hard reset my machine after attempting to boot into memtest86+, wouldn't dmesg get cleared?
domogled has joined #nixos
<peelz>
I'm not too sure how to debug this
<petersjt014[m]>
Yeah. Sorry I'm just kinda throwing things out there
<peelz>
lol
<petersjt014[m]>
I haven't boot nixos on a physical machine in awhile
<duckfullstop>
I could be on a red herring here, but I fixed this exact issue with the latest kernel on a Kaby Lake system by updating microcode, so I'm presuming I have the same problem
jboy has quit [Remote host closed the connection]
<duckfullstop>
(kernel 4.19.98)
jboy has joined #nixos
<petersjt014[m]>
peelz: Has it done anything with the systemd version enabled? I just noticed the warning about needing `allowUnfree` is only on one of those and I'm not sure if that's a mistake in the docs or not
<peelz>
I haven't tested with the systemd one
<peelz>
I'll go do that
TwoNicToupper has quit [Ping timeout: 268 seconds]
<peelz>
petersjt014[m]: not sure if I did something wrong but it's not even showing up in the systemd one
<peelz>
I executed `sudo nixos-rebuild switch && reboot`, went to my BIOS and manually selected "UEFI OS" from my boot override menu
<peelz>
uhh... it fails to boot if I pick any of the generations from the systemd bootloader. Could it be that the systemd bootloader might not be getting updated? Is my /boot stale? o.O
<clever>
peelz: first, you must run `efibootmgr -v` and see if systemd is listed there
<petersjt014[m]>
Could be--it's happend to me before. I dont remember the circumstances but they were rare. I know how to test for that tho
<peelz>
clever: uhh efibootmgr: command not found
<peelz>
why is it not installed by default?
<clever>
peelz: because you may not be using efi
<peelz>
I don't have CSM enabled
<peelz>
I'm definitely on UEFI
<clever>
peelz: but nixos cant magically know that, and you need to chooe to install it
<peelz>
so you're saying I should add it to my system packages?
<clever>
peelz: yes
<peelz>
aight, I thought my boot.loader settings would've made it install it
<clever>
nixos could be modified to do that
<peelz>
oh is there a setting for that?
<clever>
the nix code within nixos could be modified, so it would do that
<clever>
but its not an option currently
<peelz>
ah gotcha
<peelz>
clever: so what name should I expect to see for the systemd bootloader?
<peelz>
I have NixOS-boot, Linux Boot Manager and UEFI OS
<clever>
peelz: probably nixos, but you should also see the partition uuid and path to the .efi file
<peelz>
ah yeah I see it
<peelz>
it's the linux boot manager
<clever>
Boot0004* UEFI OS HD(1,GPT,27c99b08-455d-4dfe-a44f-6150cbc09ef8,0x800,0x100000)/File(\EFI\BOOT\BOOTX64.EFI)..BO
<clever>
peelz: it will do grub first, and then try systemd next
<peelz>
right, but what is \EFI\BOOT\BOOTX64.EFI ?
mexisme_ has quit [Ping timeout: 260 seconds]
<clever>
not sure
<peelz>
does that just boot directly into the latest nixos gen?
<clever>
both grub and systemd will give a menu
marcusr has quit [Remote host closed the connection]
<clever>
entry 7 i'm not sure, run `file` on that .efi file
marcusr has joined #nixos
<clever>
and/or strings
<peelz>
uhh what's the package for strings?
<clever>
peelz: `nix-shell -p`
<clever>
its in the default stdenv
<peelz>
I don't have it for whatever reason o.O
<clever>
the stdenv isnt installed globally
<peelz>
ah
<peelz>
clever: so if I wanted to install it globally, should I add pkgs.stdenv?
<clever>
,libraries peelz
<{^_^}>
peelz: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<clever>
peelz: strings is part of binutils, but i really dont want compile-time tools installed globally, so this will copy out just strings and readelf
<clever>
so i can then just install "stuff" and get just those 2 programs
<peelz>
so what packages should be installed globally?
<clever>
anything you use on a daily basis, that isnt directly involved in compiling
<peelz>
hmm well strings is something I use every now and then
<peelz>
and same for file
<peelz>
so no problem in installing it globally then?
<clever>
yeah
<peelz>
ok thanks lol, thought I was doing something really wrong
<{^_^}>
[nixpkgs] @jonringer pushed commit from Farkas, Arnold to master « pythonPackages.djangorestframework-simplejwt: init at 4.4.0 »: https://git.io/JvO8I
nikola_i has joined #nixos
<peelz>
clever: anyway doesn't really matter since I'm going to be using grub... still weird though
<peelz>
also memtest86+ still fails to boot (hangs my GRUB)
<clever>
peelz: last i checked, memtest doesnt support efi
<raboof>
I'd like to be able to debug firefox in gdb. I've used `overrideAttrs` to set `debugBuild = true; dontStrip = true;`, but still don't see any method names between `__libc_start_main` and libglib/libc
<colemickens>
wedens: isn't that done via allowedIPs?
<wedens[m]>
afaik, allowedIPs is what goes though the tunnel
<wedens[m]>
* through
<colemickens>
Well, I used to think that, and used to have it working, but now on Android when I enable WG to my server, even if I restrict allowedIPs, my other Internet doesn't work. :/ Hm.
<colemickens>
wedens: oh I'm sorry, are you asking for subtractive logic in the list of allowedIPs?
<wedens[m]>
yeah
* colemickens
doesn't know about that
<colemickens>
I could be making it up, but I thought maybe there was a bang syntax?
<wedens[m]>
I need to exclude some subnets from being routed through the tunnel
tilcreator has joined #nixos
<wedens[m]>
I'm sure it's doable on a lower level via iptables or manually specifying routes, but I lack the skill ;)
<colemickens>
And you can't do it via regular subnet masking?
<colemickens>
ex: allowedIPs: 192.168.1.0/17, 192.168.129.0/24 would exclude 192.168.128.0/24
<colemickens>
but I guess that only works for certain subnets
<wedens[m]>
I want to route everything except some subnets through the tunnel, so whitelisting is not an option
* colemickens
understands more, but doesn't have any ideas
<hyper_ch2>
how about running two WG devs? one routes 0.0.0.0 and one routes only the disallow subnets and in the second one, with the disallowed subnets, make sure the endpoint is not reachable.... haven't tried this though
<xa0[m]>
the biggest problem I have with nix is the expression language tbh. Its extraordinarily weird. It makes me want to try guix, but everything else about guix seems to be done worse
<{^_^}>
[nixpkgs] @rnhmjoj opened pull request #78843 → nixos/networking-interfaces: change preferTempAddress to allow disabling temp addresses → https://git.io/JvOzj
<Shoubit>
xa0[m], if you know any ML-like language (Haskell, OCaml, Elm) it's more familiar, but I can see it being different and harder to understand coming from a procedural/OOP language. FWIW Nix is a pretty small functional language compared to normal ML-likes
phyfey[m] has quit [Quit: User has been idle for 30+ days.]
copy` has left #nixos ["Kicked by @appservice-irc:matrix.org : User has been idle for 30+ days."]
virus_dave has joined #nixos
zurdo has quit [Quit: User has been idle for 30+ days.]
dywedir[m] has quit [Quit: User has been idle for 30+ days.]
octowan[m] has quit [Quit: User has been idle for 30+ days.]
freerangechicken has left #nixos ["Kicked by @appservice-irc:matrix.org : User has been idle for 30+ days."]
satdev[m] has quit [Quit: User has been idle for 30+ days.]
l33[m] has quit [Quit: User has been idle for 30+ days.]
macerbi[m]1 has quit [Quit: User has been idle for 30+ days.]
<mudri>
xa0[m]: Are you talking about the namespace management or something? I can't think of so many differences between Nixlang and Scheme, other than the macro system.
virus_dave has quit [Ping timeout: 268 seconds]
Makaveli7 has joined #nixos
knupfer has quit [Ping timeout: 260 seconds]
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @andir merged pull request #78706 → buildRustCrate: do not overlay OUT_PATH on src/ → https://git.io/JvYqm
nbardiuk has quit [Quit: User has been idle for 30+ days.]
leothrix has joined #nixos
nanashi0x74[m] has quit [Quit: User has been idle for 30+ days.]
<mudri>
My question: has something been done to fonts recently on nixos-unstable? I'm having difficulty debugging why Firefox has started resolving bold Japanese to shit.
gxt has joined #nixos
o1lo01ol1o has quit [Ping timeout: 265 seconds]
leothrix has quit [Ping timeout: 245 seconds]
o1lo01ol_ has quit [Ping timeout: 246 seconds]
<mudri>
(By “shit”, I mean Unifont.)
<hyper_ch2>
you can check the nixpkgs repo and see if there was a change
clefru has quit [Quit: User has been idle for 30+ days.]
o1lo01ol1o has joined #nixos
kecia[m] has quit [Quit: User has been idle for 30+ days.]
whateverthis has quit [Quit: User has been idle for 30+ days.]
<mudri>
hyper_ch2: where *in it*? I can't even find the changelog these days.
<hyper_ch2>
git clone
<hyper_ch2>
find . -iname "*.....*"
oida has quit [Quit: byez]
<hyper_ch2>
or grep -iRl xxxx
lisprez[m] has left #nixos ["Kicked by @appservice-irc:matrix.org : User has been idle for 30+ days."]
balsoft has quit [Quit: User has been idle for 30+ days.]
aurielmp has quit [Quit: User has been idle for 30+ days.]
m4ts has joined #nixos
<mudri>
It's fine to not help if you aren't going to.
<hyper_ch2>
what makes you think I know the location of something that you are looking for?
Sisyphe[m] has quit [Quit: User has been idle for 30+ days.]
isgy[m] has quit [Quit: User has been idle for 30+ days.]
leothrix has joined #nixos
manveru[m] has quit [Quit: User has been idle for 30+ days.]
<mudri>
Surely the changelog is in a standard place.
m4ts_ has joined #nixos
m4ts has quit [Ping timeout: 246 seconds]
<Shoubit>
AFAIK there is no changelog for unstable. If you're on unstable you're on your own
<etu>
If there's things worth noting that we change in unstable, for example breaking changes. We tend to put that into the changelog for the next release.
icetan has quit [Quit: User has been idle for 30+ days.]
icetan[m] has quit [Quit: User has been idle for 30+ days.]
<Raito_Bezarius>
I'm using node2nix to generate nix package files
<Raito_Bezarius>
and I got this error: error: The path name '.drv' is invalid: it is illegal to start the name with a period. Path names are alphanumeric and can include the symbols +-._?= and must not begin with a period. Note: If '.drv' is a source file and you cannot rename it on disk, builtins.path { name = ... } can be used to give it an alternative name.
<Raito_Bezarius>
I understand that means some package has a ".xxxx" in its name
<Raito_Bezarius>
But I cannot find any of them in my node-packages.nix
<Raito_Bezarius>
I'm using nixops and --show-trace tells me only it happens when recursively composing packages seems like
<Raito_Bezarius>
ah the path name is '.drv'
mexisme_ has joined #nixos
<Raito_Bezarius>
sorry for all of this :D…
<Raito_Bezarius>
so I guess it means empty name
o1lo01ol_ has joined #nixos
<Raito_Bezarius>
So my question becomes, can we trace evaluation of nix ?
<Raito_Bezarius>
I'm in nix-repl and I'd like to trace all evaluations when I try to evaluate some attrset
<clever>
Raito_Bezarius: there is the builtins.trace function, which you can insert anywhere in your code
<Raito_Bezarius>
clever: I know but this is generated code and manual tracing didn't give me useful results
<Raito_Bezarius>
I have like thousands of dependencies
<Raito_Bezarius>
(yes, Node.js…)
<clever>
Raito_Bezarius: what does `--show-trace` output?
<Raito_Bezarius>
some stacktrace leading to "composePackage" where I've put a tracing function which gives me in turn some values but the names given are valid
<Raito_Bezarius>
I'll give it a try again with the tracing to give you a 100% certain answer
<kiloreux>
I am getting `Consider using the `--user` option or check the permissions.` when installing wheel through nix (the expressions are generated through pypi2nix). Any idea what could be the problem?
<{^_^}>
[nixos-org-configurations] @edolstra pushed to master « Stop updating the 18.09 channels »: https://git.io/JvOVC
delli3 has joined #nixos
virus_dave has joined #nixos
tertl3 has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @petabyteboy merged pull request #78558 → mindustry,mindustry-server: make it possible to choose what to build (server and/or client) → https://git.io/JvqdM
<{^_^}>
[nixpkgs] @petabyteboy pushed commit from @fgaz to master « mindustry,mindustry-server: make it possible to choose what to build »: https://git.io/JvOVo
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<worldofpeace>
,
<{^_^}>
Special commands: find tell locate expand inclusive-language random-pr - Commands sorted by use count, page 0 (use ,<n> to view page <n>): help libraries unstable escape'' declarative library howoldis paste unfree escape" ping callpackage pinning -a profiling ask overlay stateversion pr which-channel tofu xml nixlang++ ifd wololo botsnack channels cache fancy-uninstall invite xy exec imperative pills haskell loot home-manager pointers random-pr
<worldofpeace>
uhhn, what's the command to reveal the magic way to change a pr base for staging without the ping attack
<{^_^}>
[nixpkgs] @Mic92 pushed to master « bpftool: fix compilation with 5.5 »: https://git.io/JvOr0
<infinisil>
,avoid-the-review-spam worldofpeace
<{^_^}>
worldofpeace: If you're retargetting a pull request and want to avoid notifying the code owners for all the unrelated changes, first rebase it against the merge base of master and staging, e.g. `git rebase --onto $(git merge-base origin/master origin/staging) master` if changing from master to staging.
ronny has left #nixos ["User left"]
<infinisil>
(I cheated by ripgrepping through the directory of these commands :))
<worldofpeace>
lol, I just need to practice doing this correctly now :D
<infinisil>
The command listing isn't very useful admittedly.. there's just too many
<rihardsk[m]>
hi, just noticed that `fetchgitPrivate` has been removed from nixpkgs. what should i use if i need to retreive stuff from a private git repo and i need to fetch submodules? `fetchGit` doesn't support `fetchSubmodules`
<Shoubit>
If I import, say a config containing environment.systemPackages, and I specify that myself too, do the two lists get concatenated? Or does the latter override the former?
<xa0[m]>
they get merged
smatting has joined #nixos
<Shoubit>
That's ideal for me now -- I am curious though, is there a way to override it? Would mkForce help?
<clever>
Shoubit: mkForce would undo all other things setting it, likely breaking many things
<clever>
Shoubit: your better off using mkIf and a custom option, to allow not setting it
<{^_^}>
[nixpkgs] @drewrisinger opened pull request #78867 → pythonPackages.marshmallow-polyfield: init at 5.7 → https://git.io/JvOim
<Shoubit>
What things would break? Is that specific to an option that's a list that accumulates a bunch of stuff, rather than just a boolean value?
<FRidh>
aanderse: are you using gitea with ldap? I am looking for a sample extraConfig (no idea where the authentication fields go)
mac10688 has quit [Ping timeout: 240 seconds]
dingenskirchen has quit [Quit: dingenskirchen]
<aanderse>
FRidh: i am pretty sure at one point i tested gitea with ldap
<aanderse>
not using it in production or anything
Tucky has quit [Quit: WeeChat 2.7]
<aanderse>
it might still be kicking around on a server
eyenx has joined #nixos
<aanderse>
oh... boo, looks like i would have configured that in the web ui instead of in my nixos config :(
<FRidh>
aanderse: ah that's unfortunate. So far all I've seen were examples using the UI.
<FRidh>
well, no LDAP for now :)
<aanderse>
sorry :(
<aanderse>
also... looks like i nuked it at some point, so yeah i guess that is gone :\
<FRidh>
heh
<FRidh>
it's strange by the way, that their sample ini includes loads of options, but nothing regarding ldap
<aanderse>
hmm yeah that is weird
camsbury has joined #nixos
<aanderse>
gosh i wish i could have migrated the gitlab i admin to gitea, but i think the group situation wasn't quite matching how our gitlab instance was built out
<Shoubit>
WinchellsM, nix-index is a package that has nix-locate and can be used to find specific files in nixpkgs
<infinisil>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<infinisil>
,locate GLES3/gl3.h
<{^_^}>
Found in packages: emscripten, mesa_noglu.dev, driversi686Linux.mesa.dev
cosimone has quit [Quit: Terminated!]
<{^_^}>
[nixpkgs] @drewrisinger opened pull request #78878 → pythonPackages.pylatexenc: init at 2.1 → https://git.io/JvOPP
iqubic` has quit [Quit: ERC (IRC client for Emacs 28.0.50)]
<{^_^}>
[nixpkgs] @drewrisinger opened pull request #78879 → pythonPackages.stestr: init at 2.6.0 → https://git.io/JvOP9
<fyuuri>
Hi. When using fish under NixOS, fish cannot access functions in .config/fish/functions even if this path is in $fish_function_path. Does anybody know what to do? Thanks in advance :).
<Ankhers>
Does anyone here use sway with nixos? Is https://nixos.wiki/wiki/Sway the most up to date information.
vidbina has joined #nixos
candicelove[m] has joined #nixos
<candicelove[m]>
Hi, are you already on a free sex dating site? This is something incredible! Registration is just a couple of minutes and most importantly everything is free, try http://bit.do/sosaty
candicelove[m] was banned on #nixos by gchristensen [*!*@gateway/shell/matrix.org/x-ujfiouccimufxfzv]
candicelove[m] was kicked from #nixos by gchristensen [candicelove[m]]
<xa0[m]>
oh dear
<Yaniel>
wow, when did spammers discover matrix
<xa0[m]>
@wait, that was over matrix?
<xa0[m]>
oh wow, it was
<xa0[m]>
interesting
<{^_^}>
[nixpkgs] @Mic92 opened pull request #78886 → Restic fixes: pruning, process substitution (take 2) → https://git.io/JvO1k
<xa0[m]>
it's russian. i guess they're everywhere
bvdw has quit [Read error: Connection reset by peer]
<gchristensen>
I feel impressed every time I take the very tip of master, build a package with a ton of dependencies, and can substitute every single one
lsix has quit [Quit: WeeChat 2.6]
Jackneill has quit [Read error: Connection reset by peer]
<zeta_0>
I never had this issue when installing emacs in home.nix, how strange
<rycee>
zeta_0: Unfortunately the package option is not in the 19.09 branch of HM. The manual on the web site is built for the master branch so it can be a bit misleading for 19.09 users.
<{^_^}>
[nixpkgs] @veprbl pushed commit from @lo1tuma to master « runLaTeX: add additional rerun criterion (#78318) »: https://git.io/JvOyD
<xa0[m]>
because on the scale of things i'd rather not have to work out how to pull this branch into my config when it's essentially a 1 line change in nixpkgs
<zeta_0>
rycee: ok, thanks for letting me know, i use emacs but i wanted to give vscodium a try, oh well, maybe some other time
<infinisil>
Ohh neat, didn't know there was a pkgs.texFunctions.runLaTeX
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<jperras>
Is there a way to ensure that a nix-shell is always spawned with a particular timezone identifier? In the shell I can do `export TZ=utc` and the timezone is changed, but doing so in the shellHook doesn't take effect.
<jperras>
additionally, the TZ is forgotten on shell exit
<simpson>
jperras: I use direnv for that.
* jperras
nods
<jperras>
I use direnv with lorri already, so I guess that's my real option
<jperras>
thanks, simpson
<jperras>
was wondering if I was missing some directive similar to time.timeZone that worked in shell
<simpson>
Oh, maybe. I don't know.
<jperras>
all good, the direnv approach works :)
<gchristensen>
hmm Lorri explicitly declines to export TZ
<jperras>
I actually export TZ in .envrc
<jperras>
before the lorri direnv eval
<gchristensen>
yeah that is a solution, but what you're saying may suggest lorri should let you decide to export it even though it would normally not
<jperras>
ah
knupfer has joined #nixos
<gchristensen>
(and the same mechanism for nix-shell)
<jperras>
in my particular situation I have a shell that's setup for a project, and that particular project is effectively assumed to run in UTC
<gchristensen>
like all good software
<jperras>
agreed :)
<jperras>
but my host is set to my local TZ, because I'm not insane
<jperras>
hence the slight disconnect
<gchristensen>
aye
<jperras>
I can file a bug or feature request in lorri if you like
<gchristensen>
sure
ayerhart has quit [Read error: Connection reset by peer]
<zeta_0>
i just installed the solargraph nix package for ruby ide support in emacs, there is not an emacs-solargraph nix package so am i going to have to clone it in order to get solargraph working in emacs?
<chloekek>
qyliss: I am passing a command if [[ $A = Y ]]; ... to Hivemind, and it works on NixOS, on Ubuntu it complains that /bin/sh: 1: [[: not found
<qyliss>
That's your bug, then
<ivan>
Ubuntu uses dash instead of bash
<qyliss>
[[ isn't POSIX
<cransom>
[[ isn't posix
<chloekek>
AFAIK /bin/sh shouldn’t accept [[ since it’s Bash-specific.
<cransom>
jinx
<chloekek>
Why does /bin/sh on NixOS accept [[ ]]?
<qyliss>
Because it's Bash
mexisme_ has joined #nixos
<chloekek>
Does Bash not emulate sh when invoked with argv0 = /bin/sh?
<qyliss>
It does to some extent, but not very well
kleisli has joined #nixos
<chloekek>
Thanks for the help, I’ll add some sort of check. :)
zeta_0 has quit [Quit: rcirc on GNU Emacs 26.3]
<cransom>
unless you need some kind of wildcard/globbing thing, you may be able replace the [[ with [ without any other changes.
<chloekek>
I switched to the inferior [ ] and now it works on Ubuntu as well.
<chloekek>
I think I’ll spend some time later into patching hivemind so it takes sh from Nixpkgs, and not upstream that.
<chloekek>
(For my own sanity.)
<qyliss>
It would be nice if /bin/sh on NixOS was Dash, I think
<gchristensen>
chloekek: do you know shellcheck?
<chloekek>
Nix is amazing, only had to jump through two hoops (LOCALE_ARCHIVE and /bin/sh) to get integration tests to work on GitHub Actions.
<chloekek>
gchristensen: I use it for a bunch of things, but not yet for Procfile, where it’s a little tricky.
<gchristensen>
ah
<chloekek>
Ironically, the script that generates my Procfile is covered by shellcheck. :^)
<adisbladis>
chloekek: I'll patch up hivemind
<adisbladis>
qyliss: Indeed it would :)
<chloekek>
adisbladis: Oh cool, thanks.
leothrix has joined #nixos
Jarva[m] has quit [*.net *.split]
yoho[m]1 has quit [*.net *.split]
cornu has quit [*.net *.split]
phittacus has quit [*.net *.split]
jonge[m] has quit [*.net *.split]
Ox4A6F has quit [*.net *.split]
MarkOtaris has quit [*.net *.split]
domenkozar[m] has quit [*.net *.split]
nckx has quit [*.net *.split]
hexagoxel has quit [*.net *.split]
packer has quit [*.net *.split]
talyz has quit [*.net *.split]
nikola has quit [*.net *.split]
ericnoan has quit [*.net *.split]
JohnAZoidberg has quit [*.net *.split]
djahandarie has quit [*.net *.split]
braingain has quit [*.net *.split]
Luker has quit [*.net *.split]
_cyril_ has quit [*.net *.split]
ericnoan has joined #nixos
Luker has joined #nixos
nckx- has joined #nixos
packer has joined #nixos
nikola has joined #nixos
talyz has joined #nixos
JohnAZoidberg has joined #nixos
braingain has joined #nixos
hexagoxel has joined #nixos
<adisbladis>
Probably the same issue with overmind.
<Ankhers>
I just switched to using sway. I have networking.networkmanager.enable = true in my configuration.nix. However, my wifi networks are not automatically being connected to. Should I be installing a separate package or something in order to connect automatically?
<adisbladis>
Overmind does not have the same issue :)
<xa0[m]>
that sounds like a networkmanager problem, not a nix problem
<chloekek>
adisbladis: probably because it goes through tmux
<gchristensen>
xa0[m]: indeed, it may be a nix or not nix problem, but I think it is reasonable and okay to ask questions like that here
<gchristensen>
Ankhers: that "just worked" for me. does nmtui show anything interesting, or dmesg, or ... not sure what else? :)
<qyliss>
Ankhers: does it work when you manually connect to them? What program are you using to do so?
<xa0[m]>
oh, i wasn't saying it wasn't
<xa0[m]>
was just pointing out the context
<gchristensen>
sure
f0i has left #nixos [#nixos]
UndefinedIsNotAF has joined #nixos
<Ankhers>
I have used nm-applet and nmtui and those both connect just fine. They are set to connect automatically. I'm just not really sure where to look to debug this.
<{^_^}>
[nixpkgs] @adisbladis opened pull request #78912 → hivemind: Patch /bin/sh to runtimeShell → https://git.io/JvOQf
<adisbladis>
chloekek: ^
<qyliss>
adisbladis: is that a good idea? we don't patch glibc (to my knowledge) to tell it not to use /bin/sh?
<gchristensen>
qyliss: I'm not sure I understand your question
kenjis1 has joined #nixos
<qyliss>
Is avoiding /bin/sh something we generally try to do?
<qyliss>
I know lots of programs that don't and assumed that we were okay with such an impurity since it's POSIX
orivej has quit [Ping timeout: 260 seconds]
<gchristensen>
I guess it depends
<gchristensen>
if hivemind needs bash, it should probably refer to ${bash}
<adisbladis>
In this instance it makes sense
<qyliss>
It doesn't need bash though
<qyliss>
A user provided it with a Bash script
<gchristensen>
hm
* gchristensen
will butt out
<qyliss>
I'm not sure it makes any more sense in this instance than any other?
<adisbladis>
Imho it's a reasonable assumption that nix-provided tooling uses bash since we're already using bash all over the place?
<qyliss>
If we consider use of /bin/sh an impurity, we should consider it so universally
<adisbladis>
qyliss: While I agree I think it's good that hivemind pulled in from nixpkgs behaves the same on NixOS and non-NixOS.
<qyliss>
But you could say that about any other program
<gchristensen>
I do say that about any other program
<rycee>
Anybody experienced with Nipkgs's `nodePackages`?
<rycee>
I wanted to add a package yaml-language-server and followed the instructions in the Nixpkgs manual but it seemed to update the whole world instead of just adding my package :-)
o1lo01ol1o has joined #nixos
<qyliss>
In that case, I think we should be trying to fix this across the board rather than just for Hivemind
<qyliss>
rycee: yeah that's unavoidable
<qyliss>
the best thing you can do is update the whole world, make a commit, then add your package and make another
<qyliss>
and hope nothing else in the world updates in between
<adisbladis>
I actually _do_ think that keeping /bin/sh around is a design mistake.
<qyliss>
generally touching nodePackages always involves updating the whole world
<qyliss>
I can do either way on whether /bin/sh is a good idea
<qyliss>
But Hivemind is definitely not the first place I think we should be focusing on if we want to remove it.
<rycee>
qyliss: Right, I was afraid I had made some mistake. I will do that and make a PR with two commits. Thanks :-)
<adisbladis>
Gotta start somewhere, why not where it's obviously causing issues?
tux has quit [Quit: tux]
<gchristensen>
adisbladis: especially since the kernel will assume an otherwise un-executable file which is +x should be executed with /bin/sh :)
<samueldr>
I assume the fix is by prepping a patch for upstream hivemind?
<qyliss>
samueldr++
<{^_^}>
samueldr's karma got increased to 158
<samueldr>
(note: I don't know what it is, only saw that it relies on /bin/sh and bad assumptions)
<qyliss>
Although again, there's nothing to patch in upstream Hivemind, because this issue comes down to user error
leothrix has joined #nixos
<adisbladis>
I'm not sure what the patch would be though.
<adisbladis>
A user installing hivemind with `go install` on lets say openbsd may have completely different expectations than someone pulling it in via nixpkgs.
Chiliparrot has joined #nixos
<adisbladis>
If pulling it in via nixpkgs I expect 0 impurities and consistent behaviour.
<colemickens>
Are there (enough) other Nix(OS)ers in Seattle to justify a meetup?
Fare has joined #nixos
Guest68037 has joined #nixos
<qyliss>
Why is /bin/sh an impurity when running on a non-Nixpkgs kernel isn't
<qyliss>
*?
<qyliss>
Or, say, using different OpenGL drivers?
<qyliss>
Some things are outside the scope of Nixpkgs, and I think /bin/sh is one of them, at least as things stand.
<adisbladis>
And I obviously disagree :)
<qyliss>
It's certainly not something that's been considered in-scope up to this point, AFAICT
leothrix has quit [Ping timeout: 260 seconds]
<qyliss>
If we want to make it in-scope (which I would probably be okay with), we should do that explicitly rather than just for Hivemind.
<adisbladis>
Come to think of it; Is there a shell which is strictly POSIX-compliant only? No extensions.
<adisbladis>
I'm not sure if dash fits the bill
<qyliss>
Supposedly yash in strict mode is strictly POSIX-only
<qyliss>
Hard agree on that
<samueldr>
yes
<adisbladis>
It would be pretty nice if the NixOS /bin/sh was POSIX-only.
<samueldr>
not positive, but I think busybox's sh when ash is not enabled might be
<gchristensen>
qyliss: well the opengl stuff is an impurity, and it is ugly and hurts our users
o1lo01ol1o has quit [Remote host closed the connection]
<samueldr>
it's the pragmatic side of faking it 'til we can make it (libGL)
<gchristensen>
yeah
<gchristensen>
the kernel is uniquely successful with its commitment to not break userspace
<gchristensen>
(^just musing)
<qyliss>
Basically I think there needs to be wider community discussion about making /bin/sh in-scope for Nixpkgs, rather than modifying leaf packages every time a user is surprised that it generally isn't.
<adisbladis>
I can buy that argument :)
<gchristensen>
okay
<{^_^}>
[nixpkgs] @robertodr opened pull request #78915 → highfive: init at 2.1.1 → https://git.io/JvOQP
jakobrs has quit [Quit: Lost terminal]
<qyliss>
I suggest (and would probably support) an RFC
<qyliss>
But if we were to do that I think we should generally rewrite /bin/sh to point to a non-Bash shell.
<qyliss>
And we should start with glibc.
o1lo01ol1o has joined #nixos
<gchristensen>
well ... not sure ... an experimental cautious approach with incremental wins might also be a perfectly valid approach
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @sikmir opened pull request #78916 → gpxlab: enable on darwin → https://git.io/JvOQQ
<qyliss>
(In reference to starting with glibc. Rewriting /bin/sh to ${bash}/bin/sh would be okay to start with.)
<qyliss>
I don't think so, if it ends up being impossible to go all the way I think it would be better we don't get halfway before we realise.
<Palpares>
hi guys. any idea why qt is still 5.12 while new version exist. should I upgrade myself and provide a patch ?
* colemickens
is totally not tracking this conversation. Nixpkgs uses a /bin/sh from the env by default?
mDuff has quit [Ping timeout: 265 seconds]
<qyliss>
colemickens: some packages use /bin/sh at runtime
<{^_^}>
nix#1759 (by muji786, 2 years ago, closed): Nix on Mac OS 10.13.2
<colemickens>
qyliss: Oh I gotcha.
<qyliss>
simpson: I always just modified the loop from the installer :/
<gchristensen>
qyliss: well... I disagree, I think getting 90% there is better. if there is a 100% solution, even better, of course
<gchristensen>
I don't think things need to be done 100% to be worth doing
<gchristensen>
I don't think all* things need to be done 100% to be worth doing
<samueldr>
the approaches can also meet halfway, supporting eachother
<gchristensen>
yeah
<qyliss>
right, but I don't think it's worth doing if we can only get 40% of the way there, for example
<gchristensen>
but I definitely agree that whackamole isn't a fun game to play, and a systemic fix is definitely nicer
o1lo01ol1o has joined #nixos
<qyliss>
I don't think some applications using /bin/sh, and some not is going to be any good for anybody
<gchristensen>
ah
<qyliss>
Especially not if it comes down to "does the application manually start /bin/sh, or does it use system(3)", which is an extremely low-level implementation detail users shouldn't have to worry about.
knupfer has quit [Remote host closed the connection]
<qyliss>
Right now the rule is that if an application spawns a shell, it'll use /bin/sh. That might not be ideal, but at least there's a general rule that applies.
<samueldr>
gchristensen: idea: run wrappers with SUID
<gchristensen>
now you're talking! with an explicit remount-as-rw!
<gchristensen>
but really, I'm going to set /bin/sh to dash and see what happens
<simpson>
qyliss: Interestingly, I found that /etc/nix/nix.conf had 32 build users, but the machine only had 10 build users. Changed nix.conf to say 10 instead, and now things appear to be working.
<colemickens>
I've wondered if dash would make things faster :) or using ripgrep in places
<gchristensen>
jtojnar: oh neat ... :)
<colemickens>
I wonder how much low hanging fruit there is in hot paths in nixos
kleisli has quit [Ping timeout: 272 seconds]
<qyliss>
colemickens: I had in intresting conversation with the Distri developer at CCCamp about this
<qyliss>
Do you know Distri?
<qyliss>
He said that the big slowdown he'd found in NixOS is that we iterate through lots of library directories every exec
<qyliss>
And theorised that if we created a directory for each package, containing symlinks to all the libraries it needed, it would be a lot faster to start programs on NixOS
<gchristensen>
I've heard that before. I'd be interested in seeing rigorous research in to that
<colemickens>
Distri is the one that keeps popping up recently - the recent initrd posts I think.
nurelin_ has quit [Quit: WeeChat 2.6]
<gchristensen>
qyliss: those symlink farms could also only be 1 hop, whereas currently we might have to explore multiple hops for each rpath entry
nurelin has joined #nixos
* colemickens
had never even considered much runtime/startup costs for Nix-packaged software
<jtojnar>
gchristensen what do you mean?
<adisbladis>
jtojnar: As opposed to many rpaths you'd only have one pointing to a symlink farm
<qyliss>
gchristensen: exactly
<jtojnar>
I meant what is neat
<gchristensen>
jtojnar: speechd, never heard of it
tilcreator has quit [Ping timeout: 240 seconds]
<jtojnar>
oh, everything uses that (well, orca, firefox and chromium at least)
<adisbladis>
qyliss: Do you know if the distri guy has done any research into how much runtime overhead their fuse thingy incurs?
<qyliss>
Yeah of course
<qyliss>
that's the whole point of the project :P
<qyliss>
IIRC it was pretty small, but not unnoticable
<lovesegfault>
I mean, it does seem like something that is easy to remedy, even though from that ticket it's unclear to me whether you need to send WITH -L or WITHOUT -L for it to work
<adisbladis>
elvishjerricco: Which isn't really comparable to zfs crypto
<elvishjerricco>
I know. They have different use cases and drawbacks
<elvishjerricco>
But if you're just looking to encrypt your whole disk, LUKS is probably more appropriate.
<gchristensen>
I'm still on luks myself, though I wish I had the foresight to encrypt where my backups live
wildtrees has quit [Remote host closed the connection]
wildtrees has joined #nixos
<lovesegfault>
I'm on LUKS + XFS right now
<lovesegfault>
I'd love a benchmark of LUKS+XFS, LUKS+ZFS, XFS, ZFS, and ZFS encrypted
* lovesegfault
emails the phoronix guy
<elvishjerricco>
I need to read up on XFS. I hear lots of people like it
<adisbladis>
My laptop is on XFS... I regret it sooo much
<elvishjerricco>
Why?
<adisbladis>
Because it corrupts when the battery runs out
<elvishjerricco>
Oof that's pretty bad
<adisbladis>
Tbh anything not CoW is pretty bad.
<elvishjerricco>
I know stratis or whatever it's called is based on xfs.
<elvishjerricco>
Iirc, stratis is the thing that red hat dropped btrfs for
<lovesegfault>
adisbladis: Does it? I'm on XFS+LUKS and it works great
Jackneill has quit [Remote host closed the connection]
<lovesegfault>
including powerloss failure
waleee-cl has quit [Quit: Connection closed for inactivity]
<lovesegfault>
My main issue with XFS is that you can't shrink it
<elvishjerricco>
Well you can't shrink ZFS either
<colemickens>
@lovesegfault ZFS on NixOS was a very painless migration for me, other than the part where I did it in place without even thinking about some data I had exclusively archived on my laptop. I'm just using native ZFS encryption now (though I'd probably stick with LUKS if I were prescribing to other people)
<lovesegfault>
Yes, that annoys me to elvishjerricco
<elvishjerricco>
Also I thought the reason stratis chose xfs was that they could shrink it...
<lovesegfault>
Unfortunately I've been bitten by btrfs too many times to be willing to try again
<lovesegfault>
colemickens: I'm backing up my `/home/` right this moment :P
<adisbladis>
lovesegfault: Not every time
<lovesegfault>
colemickens: Do you have compression enabled too?
<adisbladis>
But 3 times now I lost my shell history and got some corrupt files in the nix store
<lovesegfault>
adisbladis: Yikes
<colemickens>
Yeah, me too. I forget that I lazily use /data sometimes when working with docker so, woohoo
<elvishjerricco>
Migrating my desktop to ZFS was pretty painless, yea. NixOS has good support for it. I rsync'd my ext4 data to my laptop, reformatted with ZFS, then used zfs send to get it off my laptop and back to the desktop. Make minor config changes, nixos-install, and bam I'm back where I started except with ZFS
<colemickens>
lovesegfault: I don't remember and I can't tell which option it is in `zpool get all`. I mostly just did what the nixos zfs wiki page said.
<colemickens>
I didn't encrypt my top-level pool either, there was a reason for that.
<elvishjerricco>
colemickens: It'd be zfs get compression on one of the datasets
<colemickens>
lz4
<elvishjerricco>
colemickens: You definitely don't want to encrypt the root dataset because that means EVERY dataset HAS to be encrypted; you can never opt out after that
<colemickens>
FWIW, but that wasn't a decision I made deliberately, so ymmv
<elvishjerricco>
Actually no, it used to be that way. They later changed it to allow unencrypted children
<colemickens>
oh. I was actually going to say that some people may WANT that.
bbarker has quit [Ping timeout: 268 seconds]
<colemickens>
remove the footgun of potentially winding up with one unencrypted child
<colemickens>
but yeah, you're right that's what I'd read too.
<lovesegfault>
I can't think of something other than SWAP that I want unencypted on my box
<elvishjerricco>
lovesegfault: Pro tip, don't put swap on zfs. It basically just doesn't work
<lovesegfault>
elvishjerricco: Yep, I just use `linux-swap`
<elvishjerricco>
Also I feel like swap is one of the *more* important things to encrypt.
<elvishjerricco>
Since it can contain runtime secrets, including your disks' master keys
<lovesegfault>
Hm
* lovesegfault
goes figure out how to encrypt swap
<qyliss>
Yeah even on servers where I don't use disk encryption I encrypt swap
<elvishjerricco>
Like imagine leaking your gpg-agent state through swap.
* elvishjerricco
shudders
<lovesegfault>
That is terrifying
<lovesegfault>
I want encrypted swap with suspend-to-disk
<lovesegfault>
Oh, can I just use a swap file?
Palpares has quit [Remote host closed the connection]
chromiumBishop has joined #nixos
Palpares has joined #nixos
* colemickens
just doesn't run swap muahaha
<colemickens>
zfs, no swap, intel gpu with linux 5.4 (or 5.5). I like to live dangerously.
<lovesegfault>
colemickens: I have a Intel(R) Xeon(R) E-2176M and I'm on 5.4 too, lol
<lovesegfault>
randomly freezes
<lovesegfault>
I love it
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<colemickens>
I was just really disappointed to finish building 5.5, reboot and almost immediately get one. As long as I avoid certain versions of Chromium though, I can go a very long time between hangs.
<colemickens>
Sway has trained me to avoid Electron (and know painfully when I'm using it) and so I've noticed that Chromium seems to really trigger the bug.
<lovesegfault>
colemickens: Yeah, I've noticed the slack electron app triggers it
michalrus has joined #nixos
<michalrus>
Hey, can I `nix-build` all attrs in `default.nix` symlinking their out links in current directory?
stites has quit [Ping timeout: 272 seconds]
noudle has quit []
<michalrus>
Okay, just `nix-build` does this, but how to control symlink names separately?
Fare has quit [Ping timeout: 265 seconds]
drakonis has joined #nixos
stites has joined #nixos
<lovesegfault>
michalrus: Thet get created as `result`, `result-2`, `result-n` IIRC
<elvishjerricco>
lovesegfault: Definitely don't put a swap file on a ZFS dataset. The two just don't play nice.
rajivr___ has quit [Quit: Connection closed for inactivity]
<lovesegfault>
elvishjerricco: Hmmm
<michalrus>
Okayyy
<lovesegfault>
Can I have swap on a zvol
<gustavderdrache>
michalrus: you can pass -o foo to nix-build, which will create foo, foo-2, foo-3, etc
<gustavderdrache>
if you want to rename each result separately, you'll have to build a specific attr with -A and pass the name you want via -o
<lovesegfault>
Apparentl zvol swaps cause deadlocks
<lovesegfault>
ffs
stites has quit [Remote host closed the connection]
bbarker has joined #nixos
<lovesegfault>
@zfs users: how do y'all do encrypted swap? (cc. gchristensen qyliss)
gustavderdrache has quit [Quit: Leaving.]
<gchristensen>
I use LUKS for swap
zebrag has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
zfsonlinux/zfs#260 (by behlendorf, 8 years ago, open): Support freeze/thaw
<lovesegfault>
ffs
<lovesegfault>
I can't win
<infinisil>
You can still use a separate partition for swap
<infinisil>
It's what I do
<elvishjerricco>
But swap is of suspect utility if not being used for suspend to disk
<infinisil>
I occasionally need it when running something that takes a bit too much memory
<elvishjerricco>
I got rid of my swap partition when I switched to ZFS because I found any time I needed the space to swap memory, my machine would lock up and become close to unusable.
kiloreux_ has joined #nixos
<elvishjerricco>
Swap really only helps if there's a non-negligible amount of data in RAM that's almost never accessed. Such data can be swapped, increasing available active memory. But swapping active memory is a complete and totally disaster.
* colemickens
is seeing BT failures post-suspend in 5.5 too
m0rphism has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @rycee opened pull request #78926 → WIP: Updating node packages and adding yaml-language-server → https://git.io/JvOd2
kiloreux has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #76157 → nixosTests.gnome3: port to python → https://git.io/JvOdr
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/JvOdo
<{^_^}>
[nixpkgs] @alyssais merged pull request #77444 → nixos/mailman: lots of big improvements → https://git.io/Jvv7f
<dashkal>
I'd like to append some kernal modules to boot.initrd.availableKernelModules in configuration.nix, without needing to copy over the existing ones from hardware-configuration.nix. Is there any syntax like ++= for this? Or is it safe to refer to the existing attribute without worrying about recursion?
sigmundv__ has joined #nixos
<samueldr>
dashkal: most types in the nixos configuration implictly merge by default
<ivan>
dashkal: lists are merged, you can just use =
<dashkal>
Danke
jb55 has quit [Remote host closed the connection]
jb55 has joined #nixos
mexisme_ has quit [Ping timeout: 272 seconds]
o1lo01ol1o has quit [Remote host closed the connection]