<aleph->
Wondering if I'm laying it out wrong layout wise somehow.
codygman has quit [Ping timeout: 264 seconds]
<aleph->
Basically copying it verbatim from the example under the option page and yet it's not checking out.
codygman has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JfWNG
user_0x58 has quit [Remote host closed the connection]
user_0x58 has joined #nixos
slack1256 has joined #nixos
m0rphism has joined #nixos
<emily>
cole-h: thanks >_> i wish that was changed
<betawaffle>
is there a way to configure a kernel module to be loaded *before* /init runs?
hlisp has quit [Remote host closed the connection]
morgrimm has quit [Ping timeout: 272 seconds]
<eadwu[m]>
stage-1? I'm not what /init refers to
<betawaffle>
yes, stage-1
<betawaffle>
sorry, i was looking at the boot log and that's that the kernel calls it
shreyansh_k has quit [Ping timeout: 272 seconds]
<aleph->
There we go, got that solved. Guess it's just easier to keep it in a zone file and read it in via builtins.readFile
hlisp has joined #nixos
proofofkeags has quit [Remote host closed the connection]
KindTwo has joined #nixos
<eadwu[m]>
Not sure about before stage-1, though if you really need it you can just compile it into the kernel not as a module
<betawaffle>
how do i tell nixos to do that for me?
KindOne has quit [Ping timeout: 272 seconds]
bhipple has joined #nixos
<eadwu[m]>
You need to patch the kernel
<cole-h>
emily: On my TODO list to investigate -- unless Graham has a good reason for it.
<eadwu[m]>
i.e. if you need zswap, add something like `{name = ""; patch = null; extraStructuredConfig.ZSWAP = lib.kernel.yes; }`` in `boot.kernelPatches`
<betawaffle>
ah, perfect
KindTwo is now known as KindOne
vidbina has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @zowoq opened pull request #87564 → tmsu: use $out instead of $bin → https://git.io/JfWNP
<betawaffle>
eadwu[m]: is is `structuredExtraConfig`?
<eadwu[m]>
Uh it might be
<eadwu[m]>
Yes
<eadwu[m]>
You also do an `extraConfig = ''
<eadwu[m]>
''`
<eadwu[m]>
ZSWAP y
<eadwu[m]>
Well multiline got butchered
rogue_koder_ has quit [Ping timeout: 256 seconds]
bhipple has quit [Ping timeout: 260 seconds]
fivehole has quit [Quit: Leaving]
<betawaffle>
i love how fast i can build the kernel on this machine
<clever>
cole-h: NIX_SHOW_STATS only prints the stats upon successful exit
<bqy>
Will try once it next unlocks
<bqy>
Ty
<bqy>
Ah
<clever>
bqy: try using ulimit to limit how much ram it can consume
<finnwww[m]>
pistache: Updated using that fancy termbin thing u taught me ab :)
<bqy>
<clever "bqy: try using ulimit to limit h"> Thats what I was looking at trying, yeah
hlisp has joined #nixos
bqv has joined #nixos
felixfoertsch23 has joined #nixos
<cole-h>
clever: Oh. Heh.
felixfoertsch has quit [Ping timeout: 256 seconds]
felixfoertsch23 is now known as felixfoertsch
<keithy[m]>
Using syncthing I am attempting to switch over to declarative folders/devices. Not sure what to expect to happen.
Peter_Storm has quit [Quit: Connection closed for inactivity]
<bqy>
What would be a normal heap size for a nix eval?
<pistache>
finnwww[m]: does your ".xsession-errors" contain anything ?
<finnwww[m]>
pistache: root directory or home?
<finnwww[m]>
nm neither exist
user_0x58 has quit [Quit: Leaving]
hmpffff_ has joined #nixos
hmpffff has quit [Ping timeout: 244 seconds]
hlisp has quit [Remote host closed the connection]
<pistache>
finnwww[m]: you can try to enable the "xterm" window-manager at the NixOS side, and run manually your .xsession from there to know what the problem is
<pistache>
finnwww[m]: ah, or even easier, enable services.xserver.displayManager.job.logToFile so that the errors are logged in .xsession-errors
jumper149 has quit [Quit: WeeChat 2.8]
hlisp has joined #nixos
<bqy>
With `ulimit -v 10Gb` nix still errors out
pjt_014 has joined #nixos
<clever>
bqy: which process is consuming most of the ram? what if you run it with `-vvvv` ?
<finnwww[m]>
pistache: still no xsession errors
<keithy[m]>
I removed a folder from the gui, and mmanually form the xml config. Added that folder to the declarative folders in nix, and , it remains deleted.
<keithy[m]>
* (using syncthing) I removed a folder from the gui, and mmanually form the xml config. Added that folder to the declarative folders in nix, and , it remains deleted.
user_0x58 has joined #nixos
hlisp has quit [Ping timeout: 246 seconds]
hlisp has joined #nixos
<bqy>
clever: nothing at the point of crash. The crash is in GC though: `GC Warning: Out of Memory! Heap size: 9210 MiB. Returning NULL!`
<bqy>
so its during eval, I guess
<clever>
bqy: did you add something recursive to your config?
<clever>
bqy: what did you change most recently?
<pistache>
finnwww[m]: try using "${pkgs.bspwm}/bin/bspwm -f /nix/store/69a67hykj3jgjmr97wgzwc77alzxfx10-bspwmrc 2>&1 > ~/.xsession-errors" as "windowManager.command", so that you can see what is causing the errors
<bqy>
Nothing should be recursive I think. My most recent change was switching from nixpkgs#nixFlakes to using the nix flake directly
<bqy>
Is that an absurdly high vmem size for nix? If so I can bisect my config
<bqy>
But I have no idea what usage is normal, nobody answered
<danderson>
so, how would I pull in a cross-compiler gcc to a nix shell, but without making the entire shell set itself up for cross-compilation? I literally just want the compiler toolchain in $PATH
<clever>
bqy: yeah, id give bisect a try, it shouldnt be using that much for 1 machine
hlisp has quit [Remote host closed the connection]
<danderson>
(in particular, I don't want environment variables set up for cross-compiling, I just need the programs)
<danderson>
clever: yup, just found it too. Thanks!
h0m1 has joined #nixos
<lafrenierejm>
clever: I'm not setting `boot.loader.grub.device` that at all in my configuration.nix. I have `boot.loader.grub.devices` (not the "s") set to `["/dev/disk/by-id/..." "/dev/disk/by-id/..."]`.
<lafrenierejm>
s/not/note
<clever>
lafrenierejm: what is the full value you set it to?
<pistache>
finnwww[m]: also, check for any errors in "sudo journalctl -f -u display-manager"
<danderson>
oh good, there's no cached builds for any of this. Wheee compiling gcc :(
<lafrenierejm>
clever: The ellipses are just the IDs of my two disks (/dev/sda and /dev/sdb) as retrieved from /dev/disk/by-id.
<clever>
lafrenierejm: can you pastebin the full output when it fails?
<lafrenierejm>
The IDs worked during a bunch of automated `sgdisk` stuff, so they should be good.
<danderson>
hrm, finding a bunch of bugs about how mingw is broken in nix. That might end up being the thing that forces me back to arch, damnit :/
<danderson>
oh well, let's see how it goes first, I guess.
<clever>
danderson: iohk is using nixpkgs to crosscompile a lot of things to windows
<cole-h>
danderson: What are you trying to cross-compile? (Just curious, I can't offer any help :D)
mumuluxi has joined #nixos
<finnwww[m]>
pistache: It's in a pastebin on the forum
<lafrenierejm>
clever: Yes, /boot is a fat32 partition on top of GPT. I'm also mirroring it across the two disks with `mdadm --create --metadata=0.90 --level=1 --raid-devices=2 /dev/md/esp ...`.
<danderson>
cole-h: some glue goop for Tailscale on windows
<clever>
lafrenierejm: the mdadm in the middle might be confusing systemd-boot
<cole-h>
danderson: Tailscale on Windows :o Does Wireguard (I think you guys use Wireguard, right?) even support Windows?
<danderson>
trying to get this laptop to be a viable full dev environment, which means I have to be able to build the windows installer packages for distribution. I just fixed nsis, next step is "oh, I don't have a windows cross-compiler" :)
wucke13 has joined #nixos
<danderson>
cole-h: wireguard-go works on Windows, via the "wintun" driver
<clever>
danderson: iohk already has nsis fully cross-compiling
<cole-h>
Wow, that's crazy (and cool)!
<lafrenierejm>
clever: Sure. I'll try without the RAID1 and see if it helps. Many thanks for taking a look.
<danderson>
nsis works natively, it's fine. The problem is some random .dll in our build that needs a cross-compiler.
waleee-cl has quit [Quit: Connection closed for inactivity]
<cole-h>
emily: Sorry to ping again. It doesn't look like borg saw your "pls test" message. If you edit it and add single space at the end (or anything that makes GH notice the message has been edited), it should pick it up...
<emily>
done, but nothing happened. are we sure ofborg isn't just broken again? :p
<cole-h>
I asked so I could see if anything pops up in the logs ;^)
<cole-h>
And it popped up this time, so should be g2g
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<cole-h>
emily: Wait jk, apparently the earlier one WAS picked up...
<emily>
are you sure it doesn't work in PR bodies? I swear I've had it pick those up before but maybe I'm just senile
<cole-h>
lol
<emily>
distributed systems are the worst
<cole-h>
Well, I was looking at the logs before I pinged you and I didn't see anything show up
<finnwww[m]>
pistache: something is indeed very wrong. issue with home manager?
<pistache>
finnwww[m]: you ran "home-manager switch" and the xsession script did not update ?
<cole-h>
emily: Something does seem wrong, but there's nothing in the logs...
<cole-h>
emily: Yeah, so the service that posts check progress is dead I think. Nothing to be done until Graham comes back/wakes up and restarts it :P Sorry
<bqv>
my understanding is that only recent sets of builds are kept
<emily>
bqv: and?
<energizer>
bqv: it wouldnt make sense to keep builds from years ago? space is cheap
nixy37179 has joined #nixos
<emily>
I mean, maybe I'm wrong, I just remember hearing that once
<emily>
I rarely use old enough branches to test this hypothesis
<emily>
I wouldn't be surprised if, even if there was no expiry policy, truly ancient stuff is lost due to cache migrations
<bqv>
i mean building the entirety of nixpkgs is huge enough as it is, let alone maintaining hundreds if not thousands of different versions of each build
<bqv>
it's obvious enough to me because when i let my nixpkgs get stale enough i stop getting cache hits
<emily>
I mean the nixos foundation is a valuable S3 customer yes :p
<bqv>
(and also when it gets new enough)
<bqv>
energizer: i really hate that phrase, fwiw
<bqv>
but that's a whole other story
<energizer>
bqv: why's taht?
<emily>
I guess tarballs.nixos.org never expires at least?
<emily>
it'd be a shame to not have the ability to reproduce old commits imo
<bqv>
well, the point of nix is reproducibility, so it shouldn't *require* everything to be binary cached
<emily>
(also: remember that nixpkgs was smaller in the past so old builds would take up correspondingly less space. that's not to say it would be economical, just it's not quite as simple as "current nixpkgs times a billion")
<energizer>
bqv: i mean why do you hate the phrase
<emily>
(+ there's significant amounts of sharing between commits)
<emily>
bqv: external dependencies exist though
<bqv>
energizer: because it breeds laziness and lazy thinking
<bqv>
really? i feel like i recall seeing some code to remove old builds based on freshness
<cole-h>
danderson: I haven't looked, but maybe all those _TARGET vars are set by the wrapper -- did you try `gcc-unwrapped` and friends?
<energizer>
would be nice if the infra situation were more transparent
<danderson>
cole-h: TIL gcc-unwrapped :D
orivej has joined #nixos
<cole-h>
emily: Taking a peek now. Do you know of more elegant way to generate that sudoers file without having to build the entire iso, cancelling, and building the drv I want manually?
<bqv>
hydra.nixos.org has been hanging on my search for 10 minutes straight now
<bqv>
180TB, jesus christ
<cole-h>
Yes, h.no.o is slow lol
<emily>
bqv: ...as I said, the nixos foundation is a very valuable S3 customer :p
Acou_Bass has joined #nixos
<bqv>
well, i guess i dreamt up whatever i saw then, huh. that begs the question, why is webkitgtk on 20.03 a cache miss
<energizer>
well, 180T is about $3300 on amazon right now
<emily>
so that's, like, $4k/month in S3 storage costs alone
<emily>
I guess $48k/year is a lot lower than many startups' AWS bills. but the transit must hurt
<emily>
I guess the CDN in front helps
<energizer>
some of that is probably donated?
<bqv>
sounds like a huge incentive to get IPFS sorted to me
<energizer>
ipfs would be cheaper you think?
<emily>
ipfs is futile unless it can keep up with the write volume
<emily>
since S3 costs will just inflate regardless
<energizer>
i find it hard to believe that any storage medium will be cheaper than aws + cdn, but nonprofits can sometimes play with funny money
<emily>
(but also yeah ipfs doesn't make storage free necessarily still)
user_0x58 has quit [Quit: Leaving]
<pjt_014>
energizer: siacoin has some dirt-cheap storage currently
<energizer>
eg PyPI's bandwidth cost is quoted at ~$800k/mo but it's donated
<bqv>
no, it doesn't, but it seems smarter to me to distribute load like that over the network of users than centralize it. I wouldn't be willing to donate money, but i'd quite happily donate quite a lot of disk space
<energizer>
by the provider
<bqv>
pjt_014: that's used by dgit, it's quite neat
<bqv>
and yes, it's built on a network of servers that you can join, rather than some expensive server farm
Orbstheorem has quit [Ping timeout: 265 seconds]
<energizer>
bqv: disks at your house?
<energizer>
or in the cloud
<bqv>
energizer: both :p
<pjt_014>
bqv: what's that? searches just give me dairy queen
<emily>
cloud providers absolutely nickel-and-dime in some regards, esp with regards to egress
<energizer>
emily: exactly
<bqv>
pjt_014: basically github, but completely decentralized
<emily>
I think "cloud pricing is perfectly optimised efficient market hypothesis-style" is a pervasive and naive assumption... it's more like you pay a premium for scalability/fancy deployment models/lack of ops requirements, which of course is valuable in many cases, but it doesn't mean it's necessarily the most frugal option
<energizer>
how much bandwidth would people be willing to donate? that's usually a pretty limited quantity for me at home
<bqv>
in the vein of sia, something like that could be incentivised, but i'd be quite happy to donate a fair bit personally, since i have an unlimited plan
<energizer>
i believe unlimited will become "unlimited" fairly quickly after you start hosting cache.nixos.org
<emily>
bqv: well, your disk space is kind of useless because your home connection sucks too much to be of any use for the nix cache
<emily>
maybe not yours specifically, but people's in general. once you get toa people donating random servers it starts to become a kind of ad-hoc S3. I haven't heard that S3 costs are a huge worry for the foundation or anything, so it doesn't seem worth the complexity and effort
<bqv>
on it's own, yes it does, but as part of a well-balanced decentralized network, not really
<bqv>
and yes that's my point, but yeah i know it won't be done
<cole-h>
infinisil: Say I want to make a list item in a module come before all other normal list items, but still allow someone to `mkBefore` it... What should I do? `mkOrder 501`? Something else?
<emily>
ngl, a world where we depend on the like 10 people who care enough to set mirrors up on their random hetzners really doesn't seem better than the status quo of relying on S3 and basically not having to worry
<emily>
but I admit I have a substantial dose of cynicism about these kinds of projects
<bqv>
lots of people had cynicism about bitcoin too
<bqv>
they're poor now
<infinisil>
cole-h: Yeah something like that. Maybe 600 to give some wiggle room
<bqv>
aw. after all that time, hydra.nixos.org gave me a 404
cmk_zzz has joined #nixos
<emily>
I mean, I use bitcoin on a regular basis and still have a healthy dose of cynicism about it, so it's up to you to decide whether that's cognitive dissonance or not basing all my reasoning on mere capital :p
<emily>
but this is super off-topic and I need to sleep
<bqv>
:p
<emily>
(and fwiw I was, actually, hyped about bitcoin age ~14 when it came out, though not quite hyped enough to actually buy a bunch (I mean, I was 14))
<cmk_zzz>
How can I list dependencies of a package? I see that I gout pinentry-1.1.0-gtk2 installed (which I assume comes either from gnupg or gpg-agent service) but I don't want pinentry-gtk2 installed. I only want pinentry-curses
<infinisil>
pjt_014: That really is rather cheap hmm..
<pjt_014>
cmk_zzz: finding the nixfile that builds it would work
<danderson>
okay well gcc-unwrapped, unsurprisingly, can't compile anything :)
<pjt_014>
also nix-store -qR /nix/store/foo I think?
<emily>
it's not really an equivalent product to AWS because the reliability model is all different
<danderson>
okay, so, I'm really down to: the go toolchain mistakenly picks the windows cross-compiler when building cgo, instead of the host compiler. I just need to figure out why it does that, and how to make it stop.
<emily>
in terms of "cheaper object storage" something like Wasabi already gets you a bunch of savings in return for an (arguably less than Sia/whatever) reduction in "trust that your data won't randomly go poof"
<bqv>
lopsided98: i've mentioned that to gchristensen before, explicitly, and he's told me "no, it will build on hydra, give it a day or two", despite me checking on hydra.nixos.org and not seeing any build in progress
<bqv>
please tell me i'm right to be frustrated now
<lopsided98>
yeah, I would expect it to build because a lot of things depend on it. I was trying to figure out why hydraPlatforms was set
<cmk_zzz>
pjt_014: ok, I think I found in the the nixpkgs. Thanks
<bqv>
emily: nix-store -j0 -r (nix-instantiate channel:nixos-unstable -A epiphany)
<bqv>
lopsided98: right, so graham was wrong, that's satisfying to know
<cole-h>
infinisil++ Thanks.
<{^_^}>
infinisil's karma got increased to 287
<danderson>
oh that sneaky bastard! The go toolchain understands CC_FOR_TARGET and CXX_FOR_TARGET
<danderson>
naughty toolchain, no cookie.
<danderson>
(admittedly a weird case where I want to break nix's automagic cross-compilation setup for most of my build)
<cole-h>
bqv: That seems a little rude to Graham...
<bqv>
well i tried to get an explanation at the time but he wasn't very cooperative
<lopsided98>
bqv: I'm not sure. Based on the comments on that commit, and my understanding of the channel system, I would still expect it to be available in the cache before the channel updates
<emily>
that line should just be removed tbh
<emily>
if gnome is making it be built anyway
<bqv>
yeah that commit is pointless if it still gets triggered
<emily>
ok i should really sleep now
<cole-h>
Graham is really busy. He basically runs ofborg, nixops, has to do stuff for the NixOS org, has to do stuff for his job at Tweag...
Acou_Bass has quit [Ping timeout: 260 seconds]
<lopsided98>
Also, I'm getting a cache hit for webkitgtk in the 20.03 channel right now
<cole-h>
He was wrong once, and it's satisfying to know? Sorry, that rubs me the wrong way, especially knowing how much he does for NixOS in general...
pistache has quit [Quit: leaving]
knupfer has joined #nixos
<bqv>
i understand that. i was posing a general question to the room as to why it didn't look like like hydra was building it, he decided to respond to me telling me i was wrong, with no actual explanation why, even when i asked. i took his word for it because i didn't want to cause a fuss, but i think its fair enough that i'm a bit miffed for doing that if he wasn't actually correct
<lopsided98>
bqv: I ran: nix-build channel:nixos-20.03 -A webkitgtk . It gave me /nix/store/3kj434dj18187f9ks0qdna19kysy99d4-webkitgtk-2.26.4, using commit 14dd961b8d5
cmk_zzz has joined #nixos
<bqv>
ah, 20.03 is trying to build /nix/store/p9vrpb0q3pchx6shzgdlqi49qscr7xgi-webkitgtk-2.26.4.drv which is /nix/store/66365iz7xi5vic42ish649v5zw14cp66-webkitgtk-2.26.4 for me
<bqv>
am i more up to date?
<bqv>
no, you are...
<bqv>
hm
pistache has joined #nixos
<{^_^}>
[nixpkgs] @cole-h opened pull request #87579 → nixos/sudo: default rule should be first → https://git.io/JflvW
plutes has joined #nixos
proofofkeags has quit [Ping timeout: 272 seconds]
gerschtli has quit [Quit: WeeChat 2.7.1]
gerschtli has joined #nixos
<bqv>
lopsided98: would you be able to paste /nix/store/k0bz0x7zj5kl4i3b081qj2i0pnvsf6by-webkitgtk-2.26.4.drv somewhere for me?
<cole-h>
infinisil: Mind taking a look at #87579? Maybe we don't actually want to do that and instead want the user to use `lib.mkAfter`... idk
justanotheruser has quit [Ping timeout: 244 seconds]
cmk_zzz has quit [Remote host closed the connection]
<bqv>
that's interesting.
<bqv>
i'm going to assume nix{-,}build -A uses the system nixpkgs for everything besides the attribute pointed at, because i think one of my configs overrides is leaking through
<bqv>
so that command wasn't a very good test
<srhb>
bqv: You're using some default.nix, right?
<srhb>
bqv: Eg. the nixpkgs root one
<srhb>
bqv: The logical that import config and overlays is contained within the impure bootstrap parts of nixpkgs, so it's not really about a "system nixpkgs"
<srhb>
The logic*
<srhb>
bqv: To try and be pure, something like `nix-build -E 'with import path-to-nixpkgs { config = {}; overlays = []; }; fooPackage` is needed
<bqv>
no, i wasn't involving any nixpkgs tree, i thought the point of `nix-instantiate channel:nixos-20.03` was that it would fetch it from that channel
<{^_^}>
[nixpkgs] @cole-h opened pull request #87581 → nixos/doas: default rule should be first → https://git.io/JflvD
<srhb>
bqv: That's essentially the same thing, it'll still fetch the default nix of the head of whatever channel and do the normal default.nix instantiation, which includes all of the impurities in nixpkgs bootstrap, config and overlays and system and all.
<bqv>
would it be sufficient to unset NIX_PATH?
<srhb>
bqv: No.
<bqv>
ok, ty
<srhb>
bqv: take a look at pkgs/top-level/impure.nix to see why. :)
mac10688 has quit [Ping timeout: 272 seconds]
<srhb>
bqv: There's all sorts of "if this variable is unset, but the user has a file which is called [...] then... else... "
<bqv>
hmm, yeah
<srhb>
bqv: But in general, explicitly calling nixpkgs with config and overlays set to the empty set/list almost always suffices.
<bqv>
makes sense :)
<bqv>
i guess i'll just make a shortcut for that
pistache has quit [Quit: leaving]
remirol has joined #nixos
lorimer has quit [Ping timeout: 256 seconds]
hlisp has quit [Remote host closed the connection]
palo1 has joined #nixos
<colemickens>
srhb: does lib.mkForce override recursively? For example, if I wanted a non-empty config but wanted to prevent any existing config from being merged?
cjpbirkbeck has quit [Quit: cjpbirkbeck]
<srhb>
colemickens: nixpkgs config is not a NixOS module at all, just a dumb attrset.
<srhb>
colemickens: So there is no mkForce.
<colemickens>
oh, hm, ok
<srhb>
iow you'd have to explicitly create some expression that has the exact config you want.
hlisp has joined #nixos
lafrenierejm has quit [Ping timeout: 258 seconds]
palo has quit [Ping timeout: 256 seconds]
palo1 is now known as palo
cole-h has quit [Quit: Goodbye]
thonkpod has quit [Quit: WeeChat 2.6]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #87564 → tmsu: use $out instead of $bin → https://git.io/JfWNP
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @zowoq to master « tmsu: use $out instead of $bin (#87564) »: https://git.io/JflfG
hlisp has quit [Ping timeout: 272 seconds]
jlv has quit [Ping timeout: 245 seconds]
pistache has joined #nixos
Orbstheorem has joined #nixos
hlisp has joined #nixos
pistache has joined #nixos
pistache has quit [Client Quit]
pistache has joined #nixos
pistache is now known as Guest8011
Guest8011 has quit [Client Quit]
pistache_ has joined #nixos
pistache_ has quit [Client Quit]
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
knupfer1 has quit [Client Quit]
knupfer has joined #nixos
knupfer has quit [Client Quit]
knupfer has joined #nixos
mallox has joined #nixos
hlisp has quit [Ping timeout: 256 seconds]
knupfer has quit [Client Quit]
knupfer has joined #nixos
knupfer has quit [Client Quit]
knupfer has joined #nixos
agsdheidjd has joined #nixos
Peter_Storm has joined #nixos
<la-s>
has anyone experienced nix search duplicating results, once for nixos and once for nixpkgs?
<srhb>
la-s: Once for each channel -- that's normal.
<la-s>
but nix-channel --list only has nixos
<la-s>
am I stupid?
<srhb>
la-s: What about sudo nix-channel --list ?
<la-s>
that is for root
<la-s>
.nix-defexpr/channels doesn't contain any nixpkgs either
<srhb>
la-s: Not afaik. At least not on a daemonized system.
<la-s>
it's a normal system
<srhb>
la-s: As in, NixOS default setup? That should have a nix-daemon then.
<quinn>
i know that if you set a list to something inside of a module it is actually concatenated with the value outside of the module, is there a similar feature for attributes?
<la-s>
thanks srhb
<la-s>
that makes sense
<srhb>
quinn: Yes. It all depends on the rules of the specific option type.
<srhb>
quinn: (Hint: attributes doesn't give enough information to really answer this.)
<quinn>
srhb: so you mean if i'm setting environment.variables, i should look at the source code for that specific option?
<srhb>
quinn: Though frankly the "friendly name" is probably getting in the way.
alp has joined #nixos
<srhb>
quinn: (It's attrsOf (either str (listOf str) -- which the source link will also tell you)
proofofkeags has joined #nixos
shreyansh_k has joined #nixos
<quinn>
sorry, was still looking through the documentation. this is quite cool, thank you. i think i'll be able to figure out the next problem like this myself which is awesome
codygman has quit [Ping timeout: 246 seconds]
<quinn>
srhb++
<{^_^}>
srhb's karma got increased to 105
<srhb>
quinn: Great :-)
codygman has joined #nixos
<quinn>
i have another question, where can i find out more about mkOption.apply? it's not documented in the manual
rauno has joined #nixos
proofofkeags has quit [Ping timeout: 256 seconds]
<srhb>
quinn: I think you should consider that an implementation detail, but you can look in lib/modules.nix if you're curious.
<quinn>
okay, cool. third weird question, why is it nixpkgs.git/lib/modules.nix and not nixpkgs.git/nixos/lib/modules.nix ?
<srhb>
quinn: Because naming in computering is hard. :-P That said, the module system _is_ usable outside of NixOS (see eg. home-manager) so you can definitely justify this layout.
<quinn>
yeah, that makes sense. i actually remember being really confused module development was under the nixos manual not the nixpkgs module because of this.
<quinn>
thank god for ripgrep and fd right
<srhb>
quinn: Yeah. The split is rational most of the time, but not always.
<quinn>
yeah, a lot of programmers really want to make everything rational and simple, but you still don't want to see how the sausage is made in my experience
<srhb>
Haha.
<srhb>
I think there's just a lot of mindsets you can enter the nix ecosystem with
<srhb>
For a developer of packages, and only packages, on some, non-NixOS distribution, you generally don't care about NixOS or modules at all. Until you do.
hlisp has quit [Remote host closed the connection]
hlisp has joined #nixos
pistache has joined #nixos
pistache has quit [Client Quit]
waleee-cl has joined #nixos
codygman has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @etu pushed commit from @ccellado to master « phpPackages: phpmd (#87387) »: https://git.io/JflL7
Acou_Bass has quit [Ping timeout: 246 seconds]
Acou_Bass has joined #nixos
opthomasprime has left #nixos [#nixos]
justanotheruser has joined #nixos
asheshambasta has joined #nixos
codygman has quit [Ping timeout: 256 seconds]
<quinn>
srhb: that's true. my moment was packaging something in haskell. one of the dependencies was out of date, and i spent 5+ hours trying to get nixpkgs up to date because the script kept crashing my computer
<quinn>
that was the day i learned to fear nixpkgs
<srhb>
quinn: Ouch. :)
codygman has joined #nixos
lsix has joined #nixos
dnlkrgr has joined #nixos
thonkpod has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed 114 commits to staging-next: https://git.io/Jflq6
negaduck has joined #nixos
Darkmatter66 has joined #nixos
Darkmatter66_ has quit [Ping timeout: 258 seconds]
jkachmar has quit [Quit: Connection closed for inactivity]
eoli3n__ has quit [Remote host closed the connection]
<Mic92>
Izorkin: could be an optional attributes in nginxModules for modules that don't support MemoryDenyWriteExecute = false; i.e.: MemoryDenyWriteExecute = builtins.any (mod: mod.needsWriteableExecutuableMemory) pkgs.nginx.modules
<Mic92>
maybe you find a better name instead of needsWriteableExecutuableMemory
<flokli>
Is there a better/more automatic way of tracing dlopen() calls inside .so files than `strings` and grokking at the binary stream? :-D
<Mic92>
Actually it should be: MemoryDenyWriteExecute = !(builtins.any (mod: mod.needsWriteableExecutuableMemory) pkgs.nginx.modules)
<Mic92>
flokli: LD_DEBUG?
<flokli>
Mic92: I don't want to execute it
iyzsong has joined #nixos
<flokli>
I basically just want to analyze the .so to see what things it might want to dlopen() at runtime
<Mic92>
flokli: this is not so trival. Maybe this could implemented using radare2 scripting to inspect all call sites of dlopen that use a constant string.
hlisp has quit [Ping timeout: 258 seconds]
Neo-- has joined #nixos
<{^_^}>
[nixpkgs] @rnhmjoj pushed commit from @vojta001 to release-20.03 « monero: fix rcp.restricted option »: https://git.io/JflCc
<{^_^}>
[nixos-search] @garbas opened pull request #16 → Hide details when clicked on result row again → https://git.io/JflWM
cfricke has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<taika>
adisbladis: Thanks! What does 'format = "other"' mean?
<turlando>
I'm not sure how boot.initrd.secrets is supposed to work
<taika>
adisbladis: Your version builds, but it wont run: ModuleNotFoundError: No module named 'gi'
<taika>
mine did exaclty that too
<turlando>
I'm getting "boot.loader.initrd.secrets values must be unquoted paths when using a bootloader that doesn't natively support initrd secrets" then it prints an example, which uses boot.initrd.secrets and not boot.loader.initrd.secrets. I then copy and paste the very same example from the error message into my configuration.nix and when I rebuild I get the same error message back again
eoli3n____ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @flokli merged pull request #87414 → nixos/test-driver: Specify /bin/sh shell when running a bourne shell script as the user → https://git.io/JfCAs
<Mic92>
taika: custom buildsystem, instead of setuputils
<Bryan[m]>
I'm doing a nixos-rebuild right now after updating to the nixos-20.03 channel, and it looks like i'm building coreutils from scratch right now. (1) What's the best way to know what's actually getting built? My current method is to `ps afx` and squint a lot. (2) How can I troubleshoot a cache miss like this? (3) How could I prevent my computer from rebuilding the world in situations like this? I'd rather have nixos-rebuild
<Bryan[m]>
just error out right now
<betaboon>
I'm trying to boot a qemu-vm via pxe using pixiecore, it fetches kernel and initrd but then seems to get stuck on "Booting the kernel" any suggestions ?
<{^_^}>
[nixpkgs] @romildo opened pull request #87597 → jgmenu: optionally install a xfce4 panel plugin → https://git.io/Jfll3
eoli3n__ has quit [Remote host closed the connection]
<wnklmnn>
hi
<wnklmnn>
I'm trying to setup a nex-shell and try to add optional buildFlag to pythonPackages.pyqt5. But it seems that I got the Syntax wrong. buildInputs = [ pkgs.python3Packages.pyqt5.override{ withWebSockets = true; }];
<wnklmnn>
But I get the Message "cannot coerce a set to a string", can someone give me a hint how to solve this?
<betaboon>
wnklmnn: can you share the hole derivation ?
<sephii>
I'm trying to get gdal to work on a Django project. I've installed django and other dependencies in a virtualenv, and I have a shell.nix with some packages such as `gdal`, but I'm still getting "OSError: libgdal.so.26: cannot open shared object file: No such file or directory" when starting Django. Any hint on how to fix this?
hlisp has joined #nixos
knupfer1 has joined #nixos
<finnwww[m]>
pistache: .xsession remains the same as on the website
knupfer1 is now known as knupfer
fendor_ has joined #nixos
<{^_^}>
[nixpkgs] @sjau closed pull request #85939 → easysnap: changing format for samba shadow_copy2 to work → https://git.io/JfLBU
U-armin has quit [Quit: killed]
fadenb has quit [Quit: killed]
sasyctu has joined #nixos
dnlkrgr has quit [Ping timeout: 272 seconds]
hlisp has quit [Remote host closed the connection]
<raboof>
sephii: I'm not too familiar with python, but it'd be helpful to know if it cannot find libgdal.so.26, or it can find it, but it encounters a 'no such file/dir' while trying to load it. In any case using 'buildFHSUserEnv' might help
<sasyctu>
So I'm trying to get some GTK applications to display icons on i3-gaps, and I'm having no luck. I've got lxappearance installed like suggested, but I don't appear to have an icon theme, or at least not one which is being used by default by any application. I'm still pretty new to nixos, so is there something super simple I'm missing, like installing a GTK theme and creating a gtk config file with that theme named, or what?
cfricke has quit [Read error: Connection reset by peer]
cfricke has joined #nixos
morgrimm has joined #nixos
dnlkrgr has joined #nixos
dugur has joined #nixos
<FRidh>
sephii: patch the reference to gdal or use the nixpkgs package for django
U-armin has joined #nixos
fadenb has joined #nixos
fadenb has quit [Client Quit]
U-armin has quit [Client Quit]
quinn has quit [Read error: Connection reset by peer]
<sephii>
FRidh: the reference to gdal is in Django :/ and I need to have the same Django version installed locally than the one in the requirements file (the site is not deployed on a nixos server). How can I see which django versions are available in my channels? I tried `nix-env -qa django` but it gives no results
<betaboon>
Mic92: yeah seems like i might have to go that route :/
morgrimm has joined #nixos
<{^_^}>
[nixos-search] @garbas pushed to package-and-options-details « package and option detailed view »: https://git.io/JflRD
<sephii>
raboof: I'm not sure how I could find out which one it is?
<amanjeev>
May I ask if it is ok to change "system.stateVersion = "19.09"; # Did you read the comment?" this line after running --upgrade or before?
<{^_^}>
[nixos-search] @garbas pushed to package-and-options-details « package and option detailed view »: https://git.io/JflRd
<{^_^}>
[nixos-search] @garbas opened pull request #17 → package and option detailed view → https://git.io/JflRb
U-armin has joined #nixos
fadenb has joined #nixos
<immae>
amanjeev: a good way to check that is to be up to date, change only that value, and then re-build your system. This will show you everything that changes during the update (another way is to read the release notes)
<immae>
check that your system*
fadenb has quit [Client Quit]
U-armin has quit [Client Quit]
<immae>
(the big change recently is postgresql default version)
<{^_^}>
amanjeev: Changing stateVersion doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you need to change stateVersion for some odd reason regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value
<immae>
note that the ctrl-F won’t spot the postgresql change here ^
<amanjeev>
thank you! tbh I have already upgraded to 20.03. I just saw this line and got a bit worried if I had missed something.
<amanjeev>
but given that i have spent more time fixing my macbook than two laptops that run nixos in past 3 months, i should not have been worried at all!
cfricke has quit [Read error: Connection reset by peer]
cfricke has joined #nixos
proofofkeags has quit [Ping timeout: 258 seconds]
<amanjeev>
a friend who uses arch is asking me why should he move to nixos. anyone here who has used to be arch fan moved to nixos and why?
<niso>
amanjeev: well your friend is probably used to having arch/software break - having rollbacks should be a convincing argument :P
<immae>
I did, mostly because I was fed up of being afraid to break my services on my server at each upgrade. with nixos I may break things but I can downgrade them locally if necessary which is all I want
<Yaniel>
amanjeev: I did after the 2nd time of having to spend half a day reinstalling things after an update bricked my bootloader
<amanjeev>
:D
U-armin has joined #nixos
U-armin has quit [Client Quit]
bqv has joined #nixos
shabius_ has quit [Client Quit]
fadenb has joined #nixos
pingiun has joined #nixos
U-armin has joined #nixos
<amanjeev>
glad to be a part of community that is not ready to take "break" for an answer! :)
<pingiun>
can I make 2 instances of a module easily?
invra has joined #nixos
gustavderdrache has joined #nixos
<pingiun>
should I just use containers?
<niso>
pingiun: of what do you want to run 2 instances?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
agsdheidjd has left #nixos ["WeeChat 2.8"]
<amanjeev>
others can correct me if i am mistaken but if one of those two instances is only going to live for a short duration, one could use nix-env or something local right? talking about pingiun usecase.
numkem has joined #nixos
<pingiun>
amanjeev: I think so, but I was talking about setting up something permanent in configuration.nix
numkem has quit [Client Quit]
<niso>
amanjeev: i was kinda wondering if it might even work via the module system, however in pingiun 's usecase he probably needs matrix-synapse, coturn, postgress (and maybe something like nginx+riot)
<pingiun>
ya the dependencies screw it up here
<niso>
*posgres
o1lo01ol1o has quit [Remote host closed the connection]
<niso>
nvm
<pingiun>
otherwise i could do something like systemd.services.matrix-synapse-new = systemd.sercies.matrix-synapse // {things to change}
o1lo01ol1o has joined #nixos
cfricke has quit [Read error: Connection reset by peer]
<niso>
well, the dependencies kinda make sense (however it would be imo cool if the matrix module could e.g. include the coturn stuff). i'd go with containers so you can use the same ports among those services, and NAT to them on the host
cfricke has joined #nixos
wfranzini has quit [Remote host closed the connection]
<pingiun>
niso: yep, doing containers right now with bind mounts for postgres
wfranzini has joined #nixos
<pingiun>
just so I don't have to configure passwords for postgres
<pingiun>
hmm this assertion is kinda messing things up: hasLocalPostgresDB -> config.services.postgresql.enable
philr has quit [Ping timeout: 272 seconds]
<pingiun>
I could add the enable = true to my container config and then force the service to stay off but that seems hacky
<niso>
pingiun: oh, is it save to use postgres with bindmounts?
<pingiun>
niso: why not?
<pingiun>
I'm not doing containers for security
<pingiun>
just so I can run two instances easily
<niso>
pingiun: i'm not sure how bindmounts are working (and which type of containers you are using) .e.g in docker it's unsafe to use a database within the container (i think that was due to the filesystem)
<pingiun>
I'm using nixos containers, so that works via systemd-nspawn
<niso>
then running a database in the container should be fine, however i'm not too sure about bind-mounts (never looked into it)
<pingiun>
well I want the database to run on the host
<pingiun>
so both instances can use it
is_null has joined #nixos
<niso>
pingiun: guess i'd share the database via port
<pingiun>
niso: probably a better idea yes, but then I still have to circumvent the assertion (github link shows which)
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
Darkmatter66 has quit [Ping timeout: 240 seconds]
cfricke has quit [Quit: WeeChat 2.8]
<niso>
pingiun: do you even need to set hasLocalPostgresDB if you share the database via tcp?
<pingiun>
niso: good point, didn't see that. thanks
ddellacosta has joined #nixos
agsdheidjd has joined #nixos
qbit[m] has joined #nixos
jumper149 has quit [Quit: WeeChat 2.8]
jumper149 has joined #nixos
philr has joined #nixos
<betawaffle>
when using `pkgs.writeTextFile` is it reasonable to use `$n` in my `checkPhase` to refer to the generated file?
<{^_^}>
[nix-mode] @matthewbauer pushed 6 commits to master: https://git.io/JflgU
o1lo01ol1o has joined #nixos
qbit has joined #nixos
<qbit>
hi!
<qbit>
I have a kernel patch I am attempting to apply with boot.kernelPatches, but it isn't clear to me where the actual patch file should go
<hexa->
if it is in the same directory you can reference it with ./foo.patch
<qbit>
same directory as configuration.nix?
o1lo01ol1o has quit [Ping timeout: 246 seconds]
Khetzal has quit [Ping timeout: 272 seconds]
<qbit>
(sorry - pretty new to nix in general)
<qyliss>
You can put it wherever you want
<qyliss>
You just need to give Nix the path to it
<qyliss>
Relative paths, like ./foo.patch, are resolved relative to the file you put them in
nopsled has joined #nixos
<qbit>
hm. I have it set to the full path but i get "no such file or directory"
<rasmusm>
qbit: a good place would be with your configuration.nix in nix ralative path are from the source file so ./file.patch are file.patch in the same dir
<nopsled>
hello; recently whenever I upgrade my system I get the message "malformed JSON string, neither array, object, number, string or atom, at character offset 0 (before "cups lightdm message...") at /nix/store/z9a0mg0qg4xhlih0wix950xgq285fbzh-update-users-groups.pl line 11."
<nopsled>
line 11 of that file is "my $uidMap = -e $uidMapFile ? decode_json(read_file($uidMapFile)) : {};"
<nopsled>
this fails because /var/lib/nixos/uid-map is of the format "cups lightdm messagebus mpd mtx nixbld1 nixbld10 ......"
Desetude has joined #nixos
<rasmusm>
qbit: and its where the source file are after symlinks are expanted
<nopsled>
What is the purpose of /var/lib/nixos/uid-map, and how do I fix this?
<Desetude>
Is pkgsCross.i686-embdded.buildPackages.gcc the right package for cross-compiling i686-elf binaries?
Khetzal has joined #nixos
<{^_^}>
[nixos-search] @garbas pushed to show-details-should-be-preserved-on-refresh « preserving item displaying details on page refresh »: https://git.io/Jflgr
o1lo01ol1o has joined #nixos
<{^_^}>
[nixos-search] @garbas opened pull request #19 → preserving item displaying details on page refresh → https://git.io/Jflg6
proofofkeags has joined #nixos
<nopsled>
Does anyone know the purpose of /var/lib/nixos/uid-map?
shafox has quit [Remote host closed the connection]
<qbit>
maybe i am doing something dumb - I get "no such file or directory" with relative or absolute paths
<nopsled>
qbit: there is a line " boot.loader.grub.device = "/dev/vda";". Is there really a /dev/vda, or did you mean /dev/sda?
<qbit>
nopsled there really is
<nopsled>
ok
<b42>
nopsled: comment in source says "Keep track of deleted uids and gids" so i'm guessing the purpose is for deleted and re-added users to get the same uid
<melling>
can anyone point me in the right direction to install dictionaries for libreoffice? i tried adding 'hunspellEnv = pkgs.hunspellWithDicts [ pkgs.hunspellDicts.en_GB-large ];' to environment.systemPackages, but can't see the dictionary in libreoffice
<nopsled>
b42: for some reason nix complains whenever I upgrade because "/nix/store/z9a0mg0qg4xhlih0wix950xgq285fbzh-update-users-groups.pl" expects that file to be in JSON format when the file is really just a space seperated list of names
<b42>
melling: just put pkgs.hunspellDicts.en_GB-large in your systemPackages
hlisp has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @wamserma opened pull request #87604 → libfm: fix duplicate inclusion of libfm-extra → https://git.io/Jflgp
<nopsled>
qbit: is nix giving you a line number that it's failing at? I'm not seeing any syntax errors at first glane
<nopsled>
glance*
<qbit>
nopsled : /nix/store/vilh5ays3ymz3xkwk0fri2a70lha7pfc-stdenv-linux/setup: line 952: ./pd.diff: No such file or directory
<qbit>
other patches apply
<qbit>
and even if I have the full path, i get "No such file or directory"
<b42>
nopsled: that sounds like a bug maybe ... perhaps you can delete delete/move the file?
knupfer has quit [Quit: knupfer]
<melling>
@b42 thanks, that does not seem to make a difference though. i would expect to see that language under libreoffice tools -> language settings -> writing aids, but it does not appear
<nopsled>
arent most things like that usually configured with their own special nix options? I know that e.g. fonts are
<b42>
melling: are you using NixOS? what version?
<melling>
b42: NixOS, 20.03
<nopsled>
b42: for some reason in /var/lib/nixos uid-map is not JSON but gid-map is
<nopsled>
I would guess that if it does turn out to be a bug, the script is not the problem
<chatterbox>
I am trying to install PyTorch with OpenBLAS (instead of MKL). I thought something like https://pastebin.com/WCWFkAwP should work. But I get an error complaining about unexpected argument 'blas'.
<bqv>
(replace an existing unused one? or just append?)
shafox has joined #nixos
<asbachb>
Hi. Is there a way to prevent https certifacation validation?
vidbina has joined #nixos
kkd has joined #nixos
<nopsled>
What's the proper procedure to upgrade nixos's state config? I am running 19.09 right now
<asbachb>
For nixos-rebuild.
<emily>
hm, wonder if I can just go to the latest revision hydra built and hope it'll be in the cache
<sphalerite>
asbachb: that sounds like something you shouldn't really need to do! WHy do you want to do that?
hlisp has joined #nixos
epl692 has joined #nixos
<asbachb>
sphalerite: There's a https bypass by my proxy.
corpix has joined #nixos
<sphalerite>
nopsled: read the release notes, and generally leave the system.stateVersion option as is unless there's something in the release notes speficially suggesting to change it.
<emily>
asbachb: you're probably best-off getting its CA into the root certificate store if you really want
<emily>
but might I recommend a VPN of some kind? :/
<nopsled>
sphalertie: ok
<asbachb>
emily: I guess I need to get nixos-rebuild work before adding something to certifcate store?
<emily>
hm, true, that's awkward
<sphalerite>
asbachb: you can prevent nixos-rebuild from trying to get anything from the cache by using --option substitute false
<sphalerite>
then it should build everything locally, and if you're not doing an upgrade or have major changes from your current config, it should work fine without internet access
<asbachb>
Is it possilbe to pass proxy configuration per command line for rebuild?
proofofkeags has quit [Ping timeout: 256 seconds]
<armin>
hi, can i easily mount a btfrs volume from the live ISO?
<sphalerite>
armin: yep should work without any further ado
<finnwww[m]>
do you recommend configuring a window manager in home manager or in /etc/configuration.nix?
proofofkeags has joined #nixos
justanotheruser has joined #nixos
mcwitt has joined #nixos
<nopsled>
Is there a way to explicitly configure what channel nixos-rebuild should use by default? I changed mine to nixos-20.03 but it won't find it unless I name it "nixos"
sarcasticadmin has joined #nixos
<melling>
finnwww[m]: i do it with home-manager for my i3 desktop as it lets me declare all my config files, not sure how much you could really do from configuration.nix other than ensure it is installed
<finnwww[m]>
nice
<mcwitt>
Has anyone had luck building agda-stdlib on darwin recently? I'm getting a failure with "B.b != q₁ of type Level" (same message referenced in this issue https://github.com/agda/agda/issues/4351) on latest nixpkgs-unstable (cfe68f2b68b7)
<{^_^}>
agda/agda#4351 (by andreasabel, 18 weeks ago, closed): Regression affecting standard library: Data/List/Relation/Unary/Any/Properties: B.b != q₁ of type Level
<{^_^}>
[nix] @domenkozar merged pull request #3582 → doc: consistently refer to 'fixed-output' with a dash → https://git.io/JfCQT
<melling>
finnwww[m]: xsession.windowManager.i3 = { ... } for all my i3 stuff, and a bunch of stuff in xresources.properties for urxvt
<dramaturg>
How long does it usually take for reviewed nixpkgs PRs to get merged? Is there some process I need to follow? (asking for this one: https://github.com/NixOS/nixpkgs/pull/86927)
<{^_^}>
#86927 (by dramaturg, 6 days ago, open): exoscale-cli: init at 1.12.0
sigmundv__ has joined #nixos
<melling>
finnwww[m]: but unless you also want i3 your config might be quite different from mine
<melling>
finnwww[m]: and i haven't set up any wms other than i3
<invra>
hello, im on my first nixos install, i am still booted into the iso, did i setup my config properly? for xfce and lightdm? http://dpaste.com/25Y50BF
simba1 has joined #nixos
<melling>
invra: it /looks/ ok, does it build? does it boot? you can always reboot into the installer and edit your config if you hit any issues that prevent it from booting
<sphalerite>
invra: looks fine, the indentation's a bit weird but that won't stop it from working :) and if it doesn't work — don't worry! Your work writing it won't have been in vain, you can just boot the installer again, mount the partitions, edit the config and rerun nixos-install if you missed something.
<invra>
idk the indentation was liek this when i first opened that file
<cole-h>
invra: You forgot a semicolon after `networking.hostName = "nix"`
simba1 has quit [Client Quit]
<makefu>
invra: it seems you are not setting the password for your user or root, be sure to do this via the password options
<gchristensen>
grub-install: error: cannot find a GRUB drive for /dev/disk/by-id/ata-MICRON_M510DC_MTFDDAK480MBP_160511B060AD. Check your device.map. grrr
simba1 has joined #nixos
<asbachb>
Interesting that `nixos-rebuild test` on a freshly imported virtualbox ova compiles a lot of stuff.
<sphalerite>
makefu: well, since nixos-install runs passwd that's fine
<sphalerite>
makefu: in fact it's safer not to specify it in the config since that means the hashes won't end up world-readable :)
<invra>
makefu: yes i am no tsure how to setup password from the config file
<cole-h>
invra: Also, you forgot to close the curly braces for `desktopManager`
<invra>
i guess ill have to chroot to change them
MarcWeber has joined #nixos
<sphalerite>
invra: nixos-install will ask you to set a root password when building the config succeeds
MarcWeber has quit [Client Quit]
hlisp has joined #nixos
MarcWeber has joined #nixos
<makefu>
sphalerite: fair point, initialHashedPassword is also an option here. but yeah i forgot that nixos-install asks at the end
<cole-h>
Means it will be removed in some future release.
<invra>
can i ctrl C the install process?
<cole-h>
Using deprecated things won't cause any issue *now*
<cole-h>
However, it might in a future release.
<invra>
i rather fix it
<invra>
dont want to learn doing things like they used to be done in 1918
<invra>
:P
philr has quit [Ping timeout: 256 seconds]
<sphalerite>
invra: yes, you can ctrl-c it safely. Or let it finish and then rerun
<sphalerite>
invra: the latter will probably save a little bit of time, since it's downloading xfce and stuff while you edit :)
<invra>
sphalerite: i know, the idnentation sucked from the beginning.. it was messed up like that since i started
<invra>
i removed all the commented stuff tho cause with all that aswell it was a mess
dhess has joined #nixos
<cole-h>
emily: tbh I'm a little surprised that was never caught in the 2 years it's been there...
<T0pH4t>
has anyone created a base image of NixOs for docker? I didn't see any public containers and was wondering if there was something preventing this.
pingiun has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
T0pH4t: nixos requires systemd, and systemd doesn't like running in docker
<invra>
http://dpaste.com/3N5650E anyone be so kind to fix the indentation on this? its a mess sphalerite
hlisp has quit [Ping timeout: 258 seconds]
<T0pH4t>
@sphalerite, aren't there other distros (eg ubuntu) using systemd in docker?
<armin>
i just used that iso because i had it lying around, i have to fix a centos 7 vm with btrfs on some vmware hypervisor at work actually (yes, shame on me)
orivej_ has quit [Ping timeout: 260 seconds]
MichaelRaskin has quit [Read error: Connection reset by peer]
ntqz has joined #nixos
asbachb has quit [Remote host closed the connection]
<i077>
i know i can use the config and options sets to add my own modules to nixos, but how do i do the same in home-manager in the same context
<invra>
i cant figure out how to fix that, i tried like 15 different things alredy
<cole-h>
nopsled: It's complaining about the key itself, not the value (yet)
<invra>
tricky namings
<betaboon>
anyone knows what might be the reason for: "rootfs image is not initramfs (write error); looks like an initrd" ?
<clever>
betaboon: how much ram? how big is the initrd?
<nopsled>
betaboon: What does your configuration.nix look like?
<invra>
wah this is all it is to it.. xD installed an OS in 5 minutes... wah
<invra>
and it even works :)
<cole-h>
:)
<nopsled>
I spend 99% of the time installing nixos re-learning how to setup an internet connection. Once that's done, nearly the whole process is automated
hlisp has quit [Ping timeout: 258 seconds]
o1lo01ol1o has joined #nixos
<invra>
nopsled: thankfully everything works out of the box here
<betaboon>
clever: initrd-size: 397M rm: 1024M
<invra>
with jsut that simple config i have X working and network too
melling has quit [Quit: WeeChat 2.7.1]
<clever>
betaboon: might not fit after you uncompress, can you give it more ram?
<betaboon>
clever: increasing ram to 4G helped. but why would it need that much ?
hlisp has joined #nixos
smatting has quit [Ping timeout: 258 seconds]
<nopsled>
Why does some virtualization software create its own group (e.g. libvirtd)?
<Yaniel>
a lot of software does
<invra>
so users without root privilege can use libvirt
<danderson>
usually that's for controlling who has permission to talk to the root daemon
<danderson>
since it's very privileged
<nopsled>
ok
<turlando>
Sorry for the noob question, I'm trying to compile ffmpeg with libfdk-aac and I'm not sure how to do it. The nix package has an appropriate flag but I'm not sure how to enable it
<clever>
betaboon: that 397mb initrd is compressed, if you `cat initrd | gunzip > temp ; ls -lh temp`, how big does it inflate to?
<danderson>
a common way to do that is to talk to the daemon over unix domain sockets, which allows the server to ask the kernel what user ID is talking to it
<danderson>
then it can use the group for auth
codygman has joined #nixos
<clever>
danderson: though for group stuff, you can just chgrp the socket, and use normal unix permissions to allow/deny connection, seperate from querying the uid
codygman has quit [Read error: Connection reset by peer]
<betaboon>
clever: unpacked to 442M oO
zeenk has joined #nixos
<clever>
betaboon: and it may need both the 397 and 442 at the same time, so thats 839mb needed, plus some overhead and your getting close
codygman has joined #nixos
<betaboon>
clever: yeah you might by right
<betaboon>
clever: still thanks. brought me a step closer :D
<{^_^}>
[nixpkgs] @FRidh pushed commit from @Atemu to master « jc: 1.10.7 -> 1.10.10 »: https://git.io/JfloY
<nopsled>
Is there a way to install software for a user where I need to sudo to manipulate it (so normal software can't overwrite itself), but do not need root privileges?
<invra>
what?!
hlisp has quit [Ping timeout: 264 seconds]
<gchristensen>
nopsled: sorry, I don't understand your question
bogdb has joined #nixos
FRidh has joined #nixos
<nopsled>
Basically, I would like to install some of the user-level software in such a way that I don
<invra>
i dont understand your question but you can modify /etc/sudoers
<nopsled>
't need root privs to upgrade
<nopsled>
But at the same time if I run the software with normal user privs, it cannot be modified
<invra>
works the same way on any system with sudo installed
<cole-h>
,declarative
<{^_^}>
There are multiple ways of managing declarative profiles. 1) Attrset, compatible with imperative use of nix-env https://git.io/fAQHW ; 2) buildEnv, providing more control over the paths that are linked into the profile https://git.io/fp0aU ; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
<gchristensen>
nopsled: paths in /nix/store should not ever be modified
<danderson>
clever: yeah, as I was writing it I realized that, then gave up and said the wrong thing anyway :)
<nopsled>
gchristensen: Yes, but it requires root for me to install/remove/upgrade software, right?
<cole-h>
chloekek: Yes, it comes from `stdenv.mkDerivation`. If you don't define a `checkPhase`, it'll try to autodetect one by running various check/test targets (which may or may not be in the Makefile, if it even has one)
<cole-h>
betaboon: Ah.
ravndal has joined #nixos
<clever>
betaboon: probably
<bogdb>
yes that's the one, mingw should be added since it works perfectly fine
<chloekek>
cole-h: Thanks. Weird, my Makefile does have a test target.
<cole-h>
chloekek: As in, `make test` works?
jumper149 has quit [Client Quit]
proofofk_ has quit [Ping timeout: 256 seconds]
<cole-h>
chloekek: Ah, I think you need to set `checkTarget` to be that target
<chloekek>
cole-h: Ah, a dependency of this target is missing.
<cole-h>
It doesn't actually check "known test targets"
<{^_^}>
[nixpkgs] @veprbl pushed commit from @MetaDark to master « texlab: 2.0.0 -> 2.1.0 (#87565) »: https://git.io/Jfl6k
<cole-h>
bogdb: Then modify it and send a PR! :D
<chloekek>
cole-h: Thanks for the help! I’ll set checkPhase = "make test"; explicitly so it will fail if something is wrong with the Makefile.
devonzh has joined #nixos
hlisp has joined #nixos
<bogdb>
I will try, I am looking at half a year old issue adding support for new AMD arch and nothing happened
<cole-h>
chloekek: It will run `make $target` automatically, so you can just set `checkTarget`
<bogdb>
@cole-h, but first to get it working locally
<chloekek>
cole-h: Amazing!!
<cole-h>
chloekek: Last element in `flagsArray` is `${checkTarget}`, and stdenv calls `make $makefile $flagsArray`
ris has joined #nixos
<betaboon>
bogdb: as i wrote above: i would suggest overriding that package in an overlay to do it locally, before opening a pr. or with forking nixpkgs
<cole-h>
bodgix_: `NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1 nix-build ....` should work
<cole-h>
bogdb: ^
jumper149 has joined #nixos
<bogdb>
but just academically, can it be achieved with overrides?
<morgrimm>
Anyone know what the easiest way to pin a dependency version in a derivation is? For example, I need a specific version of libusb that's not in nixpkgs (1.0.0, for somereason)
<morgrimm>
Do I have to write my own libusb derivation to pass in?
<dramaturg>
How long does it usually take for reviewed nixpkgs PRs to get merged? Is there some process I need to follow? (asking for this one: https://github.com/NixOS/nixpkgs/pull/86927)
<{^_^}>
#86927 (by dramaturg, 6 days ago, open): exoscale-cli: init at 1.12.0
<{^_^}>
[nixpkgs] @jonringer pushed commit from @acairncross to master « onnx: 1.6.0 -> 1.7.0 »: https://git.io/JflPM
<{^_^}>
[nixpkgs] @vbgl opened pull request #87617 → ocamlPackages.labltk: init at 8.06.8 for OCaml 4.10 → https://git.io/JflPD
<{^_^}>
[nixpkgs] @bjornfor pushed 2 commits to release-20.03: https://git.io/JflP9
<kalbasit>
what's the proper way to gdb nix?
shibboleth has quit [Quit: shibboleth]
hlisp has quit [Ping timeout: 256 seconds]
D_ has quit [Ping timeout: 240 seconds]
w3emperor has joined #nixos
D_ has joined #nixos
shreyansh_k has quit [Quit: Konversation terminated!]
hlisp has joined #nixos
wunderbrick has quit [Quit: Leaving]
user_0x58 has joined #nixos
<emily>
dramaturg: "whenever a committer reviews and merges it", pretty much; it can vary dramatically. pinning committers to relevant packages might be a good idea
<emily>
dramaturg: (doesn't look like doronbehar has commit access)
<dramaturg>
right - i'll just wait and it will happen at some point in time? There is no hurry - I just don't want to miss a step I'm supposed to take
hlisp has quit [Ping timeout: 256 seconds]
<emily>
no particular step, but it can take indefinitely long unless you let people know about it, nixpkgs gets a lot of PRs. posting to the discourse might be a reasonable option too
<{^_^}>
[nixos-search] @garbas pushed to master « preserving item displaying details on page refresh (#19) »: https://git.io/JflXW
<{^_^}>
[nixos-search] @garbas pushed 0 commits to show-details-should-be-preserved-on-refresh: https://git.io/JflXl
mallox has quit [Quit: WeeChat 2.8]
hlisp has quit [Ping timeout: 246 seconds]
<OmnipotentEntity>
Can someone help me troubleshoot a package? I'm getting a build failure when attempting to build katagoWithCuda. Specifically, cudatoolkit-10.1 is failing to build citing a failed sanity check in a test compile during the configuration stage. It's a linker error: "undefined reference to symbol 'dlclose@@GLIBC_2.2.5'"
<{^_^}>
[nixpkgs] @chkno opened pull request #87618 → [20.03] nixos/test-driver: Specify /bin/sh shell when running a bourne shell script as the user → https://git.io/JflX4
<OmnipotentEntity>
I'm the package maintainer, so it'd be nice to understand what's failing now. It wasn't failing before.
<OmnipotentEntity>
I tried looking on hydra, but hydra doesn't do katagoWithCuda, or anything cuda, look like? Probably because it's so weird and impure.
<{^_^}>
[hydra] @knl opened pull request #755 → Convert all tests to the Python DSL → https://git.io/JflXR
<{^_^}>
[nixos-search] @garbas pushed to pagination « Pagination of search results »: https://git.io/JflXa
<{^_^}>
[nixos-search] @garbas opened pull request #20 → Pagination of search results → https://git.io/JflXV
<{^_^}>
[nix] @edolstra pushed 4 commits to bash-completion: https://git.io/Jfl1m
axx[m] has joined #nixos
aveltras has joined #nixos
<raboof>
hrm. /nix/store/v7ddsmcfvsqbxigb8p8ih1byhi9wvni0-terminology-1.6.0/bin/terminology on my system segfaults when displaying a 'bell', but the same executable in (any) nix-shell doesn't :/
<{^_^}>
[nixos-search] @garbas pushed to master « Pagination of search results (#20) »: https://git.io/Jfl1B
<{^_^}>
[nixos-search] @garbas pushed 0 commits to pagination: https://git.io/Jfl1R
<quinn>
raboof: is your system fully up to date?
pingiun has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<raboof>
quinn: just updated to f83c5d3432d
<devalot>
Is there any universal way to make systemd services wait for the NixOps keys service? There are a lot of modules these days that expect secret files to exist somewhere (e.g. security.acme.credentialsFile) but don't provide any mechanism to wait for that file.
<{^_^}>
[nix] @edolstra opened pull request #3587 → Generic shell completion support for the 'nix' command → https://git.io/Jfl1z
<clever>
devalot: nixops creates a -keys service automatically, you should see it in /etc/systemd/system and the nixops docs
<clever>
devalot: then you just flag your service as depending on that one
<devalot>
clever: Right, but I have to hack a bunch of systemd services to wait for that service.
<raboof>
quinn: I guess I can update to nixos-unstable-small
<tokudan>
devalot, that's just one line per service: services.postfix.after = [ "acme-selfsigned-certificates.target" ];
<devalot>
clever: That's what I do now, but figuring out the service names is sometimes a pain. For example, the names of the acme services which are generated by a private function.
<quinn>
raboof: my first guess is that there's some impurity in the derivation and it's getting a different version of something, don't update to unstable
<raboof>
(I was already on unstable)
<clever>
devalot: yeah, that could be an issue...
<quinn>
raboof: oh, right. beyond that, i don't have any ideas so gl
<raboof>
quinn: possibly, but weird that it only reproduces outside of nix-shell. I was so happily bisecting until I found that out :D
<devalot>
I just wish all services added a dependency on the key service if they are given a file in /run/keys.
splintah has joined #nixos
<quinn>
raboof: do you use a shell besides nix-shell? if you don't have any-nix-shell iirc you get switched from your normal shell to bash if you run it
splintah has quit [Client Quit]
<quinn>
sorry, do you use bash or another shell normally?
<raboof>
quinn: I use zsh. when starting terminology from nix-shell, it'll start zsh in that terminology window though
<tokudan>
devalot, how would these services know that its their file and not for a different service
<quinn>
raboof: i'm saying that you should launch terminology *from* bash and see what happens
<raboof>
quinn: ah, good one. still segfaults ;)
<quinn>
also there may be other shenanigans going on with $PATH, you may want to 'which' the terminology binary in both environments to make sure you're actually running the same one
o1lo01ol1o has joined #nixos
hlisp has quit [Ping timeout: 264 seconds]
<raboof>
it's the same one - so yeah, it's *something* with the environment...
<devalot>
tokudan: For the services that I write I have it check the prefix of the file. So, for ACME/lego, if credentialsFile starts with "/run/keys/" then it adds a dependency on ${name}-keys.
<{^_^}>
[nixpkgs] @FRidh merged pull request #86493 → gzip: add setup-hook to stop gzip from adding timestamps to headers → https://git.io/Jf3tE
<{^_^}>
[nixpkgs] @FRidh pushed commit from @prusnak to staging « gzip: add setup-hook to stop gzip from adding timestamps to headers »: https://git.io/JflME
<{^_^}>
[nixpkgs] @matthewbauer pushed 4 commits to master: https://git.io/JflDc
hlisp has quit [Ping timeout: 258 seconds]
chloekek has quit [Ping timeout: 258 seconds]
hlisp has joined #nixos
<hexa->
i'm trying to package matrix-appservice-irc using node2nix and building the resulting derivation fails with "sh: tsc: command not found", can anybody help me?
klntsky has quit [Remote host closed the connection]
klntsky has joined #nixos
leungbk has joined #nixos
hlisp has quit [Ping timeout: 272 seconds]
<leungbk>
My only nix-channel appears to be nixos-unstable, yet whenever I try upgrading my pkgs via `nixos-rebuild switch --upgrade`, I get nothing beyond the stuff in the stable release. How can I guarantee the upgrade uses the unstable channel? Do I need to set `system.stateVersion` to unstable?
<cole-h>
What is the output of `nix-channel --list` and `sudo nix-channel --list`?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
fusion809 has joined #nixos
shabius has quit [Remote host closed the connection]
FRidh has quit [Quit: Konversation terminated!]
<energizer>
why would a channel not update?
<cole-h>
That's the million dollar question. Could be for any number of reasons.
<energizer>
in this case, nobody knows why it hasn't updated?
<cole-h>
Or at least, nobody with the power to change it.
<cole-h>
s/power/power or ability
est31 has quit [Remote host closed the connection]
shabius has joined #nixos
est31 has joined #nixos
<gchristensen>
nixos-unstable: release is: nixos-20.09pre225009.085282aa4ea (build 118714565) -> error: in package ‘php’: this derivation has bad 'meta.outputsToInstall'
<cole-h>
Oh, I figured it was another issue since the fix went in a while ago.
hlisp has joined #nixos
<energizer>
is nixos-unstable supposed to advance automatically when nixpkgs/master tests pass, or is it a manual intervention?
<gchristensen>
it would be cool to have a page like: latest eval to pass `tested` and how many are queued and must be finished before it can be the new channel
<cole-h>
That would be really nice...
<cole-h>
gchristensen: And to see the number of queued jobs, you go to the eval link of /latest, right?
<gchristensen>
yeah
hlisp has quit [Ping timeout: 260 seconds]
<cole-h>
And having completed the queue is required to be a channel candidate, because then it poinsto /latest-finished, which is what the channel updater looks for
<cole-h>
s/whatever that was/points to/
<gchristensen>
you can curl -H "Accept: application/json" to see the API replies too, or ... oh gosh, thanks to gilligan https://editor.swagger.io/?url=https://raw.githubusercontent.com/NixOS/hydra/master/hydra-api.yaml
<gchristensen>
exactly
<{^_^}>
[nixos-search] @garbas pushed to index-as-keywords « index as keywords »: https://git.io/JflHT
<{^_^}>
[nixos-search] @garbas opened pull request #24 → index as keywords → https://git.io/JflHL
shabius has quit [Remote host closed the connection]
w3emperor has quit [Quit: Connection closed for inactivity]
<cole-h>
Wow, that's actually pretty simple. But thanks for explaining it to me -- I've been stumped by what hydra does for the longest time. gchristensen++
<{^_^}>
gchristensen's karma got increased to 291
wnklmnn has quit [Read error: Connection reset by peer]
<gchristensen>
:)
i077 has joined #nixos
shabius has joined #nixos
hlisp has joined #nixos
wnklmnn has joined #nixos
<{^_^}>
[nixos-search] @garbas merged pull request #24 → index as keywords → https://git.io/JflHL
<{^_^}>
[nixos-search] @garbas pushed to master « index as keywords (#24) »: https://git.io/JflHg
<{^_^}>
[nixos-search] @garbas pushed 0 commits to index-as-keywords: https://git.io/JflH2
<i077>
getting a problem with evaluating my nixos config as a flake
<i077>
`nix flake check` succeeds, but when i do `nixos-rebuild build`, it complains about the flake not providing config.system.build.toplevel
<mcint_>
where do nixpkgs come from? I can see from the convenient url+hash of nix-prefetch-url, that nix might only need to keep around a name, description, url, and hash.
hlisp has quit [Ping timeout: 240 seconds]
<mcint_>
does nix have no source distribution of its own? is it preferred to fetch things from project-canonical source repositories?
cosimone has quit [Ping timeout: 244 seconds]
<drakonis>
uh
<MichaelRaskin>
There is a tarball mirror. And binary cache also serves source packages.
akymaky[m] has joined #nixos
<drakonis>
ah there it is
zeenk has quit [Quit: Konversation terminated!]
<MichaelRaskin>
But the definitionally correct origin of the source tarball is upstream; binary cache is just a cache that lets you get the results without performing the originally defined steps.
<dsal>
I'm trying to migrate some builds from stack to cabal. In stack, I have to add a 'nix:\n packages: [zlib]' to pretty much anything I build. Is there an equivalent to this for cabal?
<{^_^}>
[nix-pills] @jonringer pushed commit from @dmedinag to master « Add missing word »: https://git.io/JflFz
<{^_^}>
[nixpkgs] @jonringer merged pull request #87628 → python3.pkgs.ipykernel: skip test_unc_paths on darwin → https://git.io/JflQi
<{^_^}>
[nixpkgs] @jonringer pushed commit from @badmutex to master « python3.pkgs.ipykernel: skip test_unc_paths on darwin (#87628) »: https://git.io/JflFw
hlisp has joined #nixos
wnklmnn has quit [Quit: Leaving]
Serus has quit [Ping timeout: 256 seconds]
hlisp has quit [Ping timeout: 246 seconds]
Serus has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
hlisp has joined #nixos
mcint_ has quit [Remote host closed the connection]