<jackjennings>
I'm using node2nix to install npm packages as a component of a nixops deployment with its own derivation. I have a problem where all of my node modules are removed when I garbage collect — the next deployment after garbage collection re-downloads all of the packages which takes a long time. I think this means that I need to explicitly declare the packages as build inputs (propagatedBuildInputs?) but not sure how to do so
orivej has quit [Ping timeout: 245 seconds]
johnnyfive has joined #nixos
Peetz0r has joined #nixos
romildo has joined #nixos
emily has joined #nixos
jperras_ has joined #nixos
<colemickens>
Is there a popular naming convention for overlays?
<colemickens>
I've seen `nixpkgs-mozilla`, `azure-cli-nix` so far.
drakonis1 has joined #nixos
drakonis1 has quit [Client Quit]
drakonis1 has joined #nixos
drakonis_ has joined #nixos
<pbogdan>
will there be a live stream of the upcoming nixcon?
drakonis has quit [Ping timeout: 264 seconds]
<CMCDragonkai>
dmj`: have a look at hnix, I've been using this for our Architect language
<pie_>
hyper_ch, yeah but i mean for actually retrieving it
<pie_>
well..maybe i can put something together for generating a tar file
ghasshee has joined #nixos
vk3wtf has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
jperras_ has quit [Ping timeout: 268 seconds]
ghasshee has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #48931 → misc pkgs: Darwin -> Linux cross compilation for 18.09 → https://git.io/fxXKM
drakonis1 is now known as drakonis
<{^_^}>
[nixpkgs] @jraygauthier opened pull request #48932 → nixos/xrdp: allow extra keymap files to be installed → https://git.io/fxX6I
<wykurz>
hi, having a really hard time getting fetchgitPrivate to work. I'm on nixos-18.09. despite my best efforts it keeps complaining about not being able to open ssh config file. could someone please point me to steps how to use fetchgitPrivate that work?
<elvishjerricco>
typetetris: Basically fetchGit uses one mega-rep where it caches all commits from all repos. Git does not much like this. Having many unrelated histories damages the performance greatly.
<typetetris>
I have two machines, each with nix/nixpkgs installed in multi-user-mode, if I try to use `nix-copy-closure` via ssh and an arbitrary user, it complains, that it doesn't find nix-store on the other machine. The problem seems to be, that nix-copy-closure doesn't try to load `/etc/profile.d/nix.sh` to get a usable `PATH`. (ssh will use bash in non-interactive mode, which doesn't load anything in my case :( )
<typetetris>
What can I do to get nix-copy-closure to work
<simpson>
typetetris: I think that there's nothing to be done short of fixing your SSH configuration on the target host. Is it not running NixOS?
astronavt has quit [Ping timeout: 268 seconds]
<typetetris>
None of the machines is running NixOS. Why does the SSH configuration need fixing? I thought its pretty standard, that ssh just lets bash handle the reading of startup files?
<typetetris>
Or is nix-copy-closure trying to BASHENV or similar?
<typetetris>
sorry it is BASH_ENV
<simpson>
I think that it calls standard Nix tools which are expected to be in PATH. Is doing a reverse nix-copy-closure possible?
reinzelmann has quit [Quit: Leaving]
<cocreature>
typetetris: probably something to do with login-shell vs non-login-shell. are you sourcing /etc/profile.d/nix.sh in .bashrc or .bash_profile? try moving it to the one that it’s not currently in
hyper_ch2 has joined #nixos
<cocreature>
specifically, if you pass a command to ssh it will run it in a login shell and not read .bashrc
<typetetris>
cocreature: Neither, I rely on /etc/profile.d/nix.sh to be sourced by /etc/profile
<typetetris>
Passing a command to ssh invokes a non-interactive shell, which loads nothing ...
<typetetris>
it is in the responsibility of the command to source what is needed.
<typetetris>
Maybe there should be a well known path to nix-store in /nix/var/nix somewhere and nix-copy-closure uses that. Having to manipulate ~/.bashrc of all the users I want to use is a bit ... unfortunate.
<typetetris>
its already there, it seems sshd-config is really broken for this machine ... I can't get it to source any of /etc/bashrc,~/.bashrc,/~/.ssh/rc,/etc/ssh/sshrc if I use ssh with a command ... :(
Ariakenom has joined #nixos
o1lo01ol1o has joined #nixos
<cocreature>
typetetris: that’s not an sshd thing, that’s a bash thing
<typetetris>
so bash runs non-interactive non-login and can't determine, that it is run by sshd, so it doesn't load anything :(
<srhb>
typetetris: If ~/.foorc works, then presumably /etc/foorc works too?
<srhb>
Though I remember that the rules are tricky.
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<typetetris>
srhb: No file tried so far worked ..
<colemickens>
wait do overlays work for nixos modules as well?
jackdk has quit [Ping timeout: 245 seconds]
<srhb>
colemickens: package overlays do, yes.
<srhb>
colemickens: There's no real way to define modules in an overlay, however (but those can just be in imports)
brejoc has joined #nixos
<typetetris>
okay, I configured a dedicated nix-ssh user now with sshd_config ForceCommand /nix/var/nix/profiles/default/bin/nix-store ... now I get: https://pastebin.com/HGLqyS6Y
<{^_^}>
nix#2127 (by antifob, 25 weeks ago, open): Failure to copy closures: ... lacks a valid signature.
<typetetris>
that is a little bit frustrating
nD5Xjz has quit [Ping timeout: 272 seconds]
<srhb>
typetetris: Did you actually sign with a trusted key?
<simpson>
At least it's progress.
<typetetris>
I didn't do anything. I just installed nix on two machines, build on one with nix-build and try to move the closure of the result to the other one ...
elgoosy has quit [Remote host closed the connection]
<colemickens>
though I'm actually a little unsure myself if ALL nix tools will respect that, or if nix.conf fiddling is required too. maybe someone will chime in.
<srhb>
colemickens: What does the list actually consist of? strings referring to keys? If so, it can be done, but it sounds a bit roundabout.
elgoosy has joined #nixos
<colemickens>
srhb: I have a list of packages in my overlay. I want to make a list of them as well, but then inherit each item in the list in the final list of pkgs surfaced as the overlay
<colemickens>
I can put a WIP commit up to try and illustrate the point...
<colemickens>
there's likely a better way :)
<typetetris>
`binaryCachePublicKeys` that option isn't even documented in the nix manual ...
<srhb>
typetetris: Because it's a NixOS-ism
<typetetris>
no nixos here ..
<srhb>
typetetris: The corresponding nix option is binary-cache-public-keys (and it's deprecated)
<srhb>
man 5 nix.conf
<typetetris>
srhb: Thanks for the pointer.
<colemickens>
srhb: hm, were the nixos options updated corresponding to the nix flag renames?
<srhb>
colemickens: ... not yet :-)
<typetetris>
Is there some timeline known, when there will be man pages for the new `nix` command?
<typetetris>
hmm, I don't understand that. I thought that '/nix/store/kmqiiwrvb04hqnxhpjxl2jkn3md19pp2-iana-etc-20180711' was downloaded from cache.nixos.org, so shouldn't that signature from cache.nixos.org still be valid?
<srhb>
typetetris: Yes, it should.
lonokhov has joined #nixos
<typetetris>
so `nix-copy-closure` doesn't copy the signature along?
<colemickens>
(i think you have to respecify the nixos key if you supply your own)
<typetetris>
`nix copy` yields the same error
<typetetris>
Can I check wether I downloaded `...-iana-etc-` or wether it was build on my machine?
<srhb>
typetetris: You can check whether it has a signature with nix path-info --json /nix/store/... if that's what you intend to try
<colemickens>
If I just put the pkgs into an attribute set and then return that, the `fmt` package isn't used as an override for the `waybar` packaeg as needed.
<typetetris>
That yields: `/nix/store/kmqiiwrvb04hqnxhpjxl2jkn3md19pp2-iana-etc-20180711 cache.nixos.org-1:QTE2q2N6aUPq7YctBJsbw7UGMoDSDOszukU3GzaM9TZIaXx/+zFJCyg1Bx9aOYB0U3NClvZ0AXsu6iGPyyDBDA==` so at least `nix copy` should work?
<colemickens>
If I make the change I just linked to, it builds properly, but now is not as nicely organized and easily accessible via the 'swaypackages' handle I've made.
<typetetris>
I just noticed, that the installer for nix/nixpkg really only provides a very limited /etc/nix/nix.conf without any trusted-public-keys listed, not even the one for cache.nixos.org so maybe that is the problem. Sorry for the noise if so.
<srhb>
colemickens: So eg. waybar gets fed pkgs.fmt, but expectes pkgs.swaypkgs.fmt
<srhb>
If instead you depend on swaypkgs in waybar, it'll work
<srhb>
(Or alternatively if you create a new callPackage with the scope of swaypkgs added in)
elgoosy has joined #nixos
<colemickens>
srhb: okay, good excuse to go read the source/docs for callPackage. I'm also curious if there's a nicer way to achieve what I'm doing
<srhb>
colemickens: Your current callPackage works like this: Look at the function called, get all arguments. Feed it all the attributes from pkgs.* scope that have the same name
<srhb>
but you're expecting arguments to be autofilled from pkgs.swaypkgs
<colemickens>
(the whole point is to get "swaypackages" exposed so taht `build.nix` can access and test building against a nixpkgs snapshot)
<colemickens>
srhb: I see.
<srhb>
I don't think I really understand the purpose of swaypkgs. Seems like you're making things harder by introducing a new scope?
<srhb>
Consider that you're not actually making things more generic for build.nix because you're already defining the attrset of packages you want to include in default.nix explicitly
<srhb>
You might just as well override the pkgs.* things and include the explicit attrset in build.nix instead.
<srhb>
It's no more and no less generic.
<srhb>
(But it'll work and won't duplicate existing attributes)
reinzelmann has joined #nixos
<srhb>
(Meaning, build.nix will just be `blah blah ... in { inherit (pkgs) fmt grim slurp waybar ...; }`)
<srhb>
I like paranthetical remarks. Sorry. :-P
vk3wtf has joined #nixos
mmercier has joined #nixos
<colemickens>
srhb: don't judge me, I just was trying to avoid duplicating the pkg list in build.nix
<ekleog>
we can handle a system with mixed openssl / libressl without issues, right? (just checkin')
<colemickens>
That's really the only point it serves.
<gchristensen>
ekleog: yea
<ekleog>
great :)
<srhb>
colemickens: Right. I'm not judging, just pointing out that you're not really deduplicating anything. :)
<ekleog>
so I can tell opensmtpd developers it's ok for us if next version depends on libressl instead of leaving the choice between libressl and openssl :)
<ekleog>
:D
nD5Xjz has quit [Ping timeout: 246 seconds]
<colemickens>
Hm. I might be misunderstanding. The attrset or list of packages in default.nix is to conform as on overlay. I can't avoid it there. But the `build.nix` needs to reference all of them somehow, preferably without relisting? (I'm fine with not introducing another scope. it certainly seems preferable)
<Arahael>
Hmm, seems I have a bit of a catch-22. I have a wifi network here, but the nixos installer does not have the firmware required for my network adapter.
<Arahael>
I do, however, have a raspberry pi I could use as a router. How do I configure it to route?
<srhb>
colemickens: Okay, I guess I am advocating relisting at least the attrnames _once_. The alternative is to actually provide the new callPackage, which I don't recommend.
<typetetris>
I have a bunch of machines and I want them to use one big nix store together, so every machine should be able to place store paths there and use them from there. How could I best achieve that?
<sphalerite>
Arahael: if you have network-manager on the pi, the quickest thing would probably be to create an ethernet connection with the sharing option enabled
Guanin has joined #nixos
<simpson>
typetetris: Hm, like over NFS? How did you want to make each machine's local /nix/store manifest?
<srhb>
typetetris: You should look at what CERN and co. do. It is possible to essentially NFS the whole thing, as long as you're not on NixOS, but that costs you provisioning headaches of course.
<Arahael>
sphalerite: Well, I've set the sysctl ip_forward option, but it doesn't seem to be sufficient.
<srhb>
Ideally it could be done with NixOS and a proot somehow, but that seems less trivial.
<Arahael>
sphalerite: The Pi doesn't have a screen. :)
<Arahael>
srhb: You're here, too!?
<srhb>
Arahael: I AM EVERYWHERE.
<srhb>
:-)
<typetetris>
srhb: It would be sufficient if there was one dedicated machine and the others could push theirs build products there ... I tried to do that with nix-copy-closure ..
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.rx: 1.6.0 -> 1.6.1 »: https://git.io/fxXdF
<simpson>
typetetris: Can you be more specific about why you want this? Compare and contrast, say, running your own Hydra, which is unpleasant but more akin to traditional build-farm design.
<sphalerite>
Arahael: you need to set up IP addresses on the pi and the laptop as well
<simpson>
typetetris: Also, at the end of the day, how much build time are you actually trying to save? https://xkcd.com/1205/
<Arahael>
sphalerite: That, too, and I can ping the pi's wifi connection, but not the other computers on the wifi network.
<typetetris>
simpson: So I would generate a key pair for the hydra server and set that as trusted public key in each of the machines?
<srhb>
Musings on the topic: It shouldn't be impossible to share only /nix/store and still boot from it, but it requires distributing the nix db and gcroots somehow.
<Arahael>
sphalerite: So, how do you set up the routes between *networks*?
<Arahael>
sphalerite: I mean, one 10.1.0.x is routed to that network, and the other is 10.0.0.x for the other network, but.
<sphalerite>
Arahael: oh actually you could also bridge them. But I'm not very familiar with how to do that
<Arahael>
The pi can ping 10.0.0.138, but the nixos system can't, not even via the pi (which is, from nixos's perspective, the 10.1.0.2 gateway.
<sphalerite>
Arahael: I think it would make more sense to have them all in 10.0.0.x
<Arahael>
sphalerite: Bridge? How is that different to routing?
<simpson>
typetetris: I guess? It's a pain though. It would be easier, and more parallel, to let each machine grab stuff from hydra.nixos.org directly. Does your network topology allow that?
<sphalerite>
Arahael: so for example the pi is 10.0.0.100 and the laptop is 10.0.0.101. so on the pi you have 10.0.0.100/24 on the wifi and 10.0.0.100/31 on the ethernet
<Arahael>
sphalerite: And why would it makes sense to have them all in 10.0.0.x? I mean, how would the *two* networks communicate? How would it know which subnet is which?
<sphalerite>
Arahael: then on the laptop you have 10.0.0.101/31 and default route as 10.0.0.100
camsbury has joined #nixos
<Arahael>
sphalerite: Remember, I'm trying to setup the Pi as a wifi adapter.
<sphalerite>
Arahael: yes
<Arahael>
sphalerite: If say, 10.0.0.138 is on the same subnet, how would it know it actually needs ot use a gateway?
<Arahael>
sphalerite: Because that's one hop away.
<typetetris>
simpson: hydra.nixos.org does build glibc.static ?
<srhb>
typetetris: Yes.
<simpson>
typetetris: Sorry, I need to get to sleep soon. Could you reveal what you're actually working on?
<sphalerite>
Arahael: by having 10.0.0.100/24 on the wifi adapter, the pi knows to reach 10.0.0.138 via wifi. By having 10.0.0.100/31 on ethernet, it knows to reach the laptop via ethernet
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.kazoo: 2.2.1 -> 2.5.0 »: https://git.io/fxXbB
<typetetris>
simpson: I just want to use nix for some internal tool, which needs to stay in sync with some other product, and don't want to waste our 50 developers time with all the identical rebuilds.
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.schedule: 0.3.2 -> 0.5.0 »: https://git.io/fxXb1
<srhb>
Also Hydra...
<Arahael>
sphalerite: Yes, the pi knows where everything is. But isn't the rule for using a gateway simply that the subnet has to spcify that it's needed?
<Arahael>
sphalerite: I mean, "I have a packet I need to send to 10.0.0.138". I could either blurt it out of my default route, or observe that I have a route for 10.0.0.x for eth0. Since it's matching, I send it to eth0.
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.typing-extensions: 3.6.5 -> 3.6.6 »: https://git.io/fxXNc
<Arahael>
sphalerite: I think there's something I'm clearly missing, and I don't think I'm going to understand it from IRC without more real examples, so let me show you the output of the commands you asked for earlier:
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.validictory: 1.0.0a2 -> 1.1.2 »: https://git.io/fxXNa
<Arahael>
symphorien: The only computer running linux is the RPi. :)
pie_ has quit [Remote host closed the connection]
<sphalerite>
Arahael: right. By making the ethernet LAN a subnet of the wifi LAN, you avoid the need for any NAT or for the wifi router being aware of the 10.1.0.0/24 net
<Arahael>
symphorien: (Well, I do have the nixOS live usb, but there's still the matter of network access)
pie_ has joined #nixos
<Arahael>
sphalerite: Just to clarify, which wifi router are we talking about? The Pi, or the home wifi router?
<sphalerite>
Arahael: the home wifi router
<Arahael>
sphalerite: So I'm wondering how the routing even works if I put both devices in the same subnet?
<Arahael>
sphalerite: Lets suppose the home wifi router has a response for 10.0.0.2, how does it know it has to send it through the Pi?
<Arahael>
sphalerite: And hwo does hte Pi know it has to pass that on?
<sphalerite>
Arahael: they're not the same subnet. One is 10.0.0.0/24 and one is 10.0.0.100/31
<Arahael>
sphalerite: Ah!
<sphalerite>
Arahael: but the latter is within the former, so the home router knows it has to send stuff to 101 through the wifi where the pi will get it
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.simpleeval: 0.9.6 -> 0.9.8 »: https://git.io/fxXNx
<sphalerite>
might have been easier to bridge it at this rate :')
<Arahael>
sphalerite: Well, the Pi's wifi has to respond to it, recieving the packets that aren't actually directed to it. I guess this means it has to be promiscious.
<ocharles>
Can anyone help me do some Hydra debugging? I have a jobset that should contain an attribute `networks.internal.coordinator` but it's no longer showing up in evaluations since I updated Hydra
<ocharles>
The weird thing is the jobset has no evaluation errors
<sphalerite>
Arahael: network-manager is probably more of a hindrance than help on the nixos machine just now
<sphalerite>
Arahael: I'd suggest systemctl stop network-manager, then set up the interfaces with ip
<ocharles>
ah wait, that one I renamed actually, that's fine
<ocharles>
but another one that's missing is `networks."circuithub.com".api`
<Arahael>
sphalerite: That's disappointing, but will do.
<ocharles>
I *think* it might be because of the "circuithub.com" attribute name
<ocharles>
That is, it really is `networks."circuithub.com".api` and not `networks.circuithub.com.api`
<Arahael>
sphalerite: How does the UI there get it so badly wrong?
<sphalerite>
Arahael: so… ip link set enpfoo up; ip a add 10.0.0.101/31 dev enpfoo ; ip route add default via 10.0.0.101
<sphalerite>
Arahael: I don't think it's necessarily getting stuff wrong, just adding a bit of confusion by hiding more of what's going on
<sphalerite>
oops, default via 10.0.0.100 not 101
<Arahael>
I still seem to have too many routes.
<sphalerite>
ip route flush
<sphalerite>
:)
zarel has joined #nixos
mayhewluke has quit [Ping timeout: 240 seconds]
mayhewluke has joined #nixos
<Arahael>
sphalerite: Why 10.0.0.100?
<Arahael>
sphalerite: I mean, shouldn't that one be on the Pi?
<sphalerite>
yes, 10.0.0.100 shoudl be the default route on the big computer, since it's supposed to send stuff through the pi
<sphalerite>
and 101 should be the big computer's address
<Arahael>
Hmm, back to square one.
zarel has quit [Remote host closed the connection]
<Arahael>
Big machine: IP address 10.0.0.101/31; route -> default via 10.0.0.100 dev eno1; 10.0.0.100/31 dev eno1 proto kernel scope link src 10.0.0.101
<Arahael>
pi: IP address Well, too many, by now. I should clear it, but amongst them on this link is 10.0.0.100/31, and the routing is clearly working (I mean, I can ping, browse network, do everything, just fine)
<Arahael>
I can also ping the big machine from the pi, that's fine.
<sphalerite>
can you ping the pi from the big machine?
<Arahael>
Yep, I can ping both the pi's interfaces frmo the big machine - curiously, including the wifi.
<Arahael>
I just can't ping anything else on that wifi.
ghasshee has quit [Remote host closed the connection]
<sphalerite>
what does cat /proc/sys/net/ipv4/conf/all/forwarding on the pi say?
<{^_^}>
[nixpkgs] @xeji pushed commit from @veprbl to master « arrow-cpp: fix for zstd 1.3.6+ (#48928) »: https://git.io/fxXp5
<sphalerite>
hmmmeh
Mateon1 has quit [Ping timeout: 264 seconds]
Mateon2 is now known as Mateon1
<Arahael>
sphalerite: Feelsl ike I'm guessing, is there an actual *test* that shows: "The problem is on the pi"?
<sphalerite>
ok really hacky idea which might just work… on the pi, ip a add 10.0.0.101/24 dev wlan0 && ip route add 1.1.1.1 via 10.0.0.1 src 10.0.0.101 && ping -c 5 1.1.1.1 && ip route del 1.1.1.1 && ip a del 10.0.0.101/24 dev wlan0
cmacrae has joined #nixos
<Arahael>
Let me try cut&pasting that. Heck, if it doesn't work, I'll just reboot it.
<sphalerite>
if it doesn't, tcpdump will probably be helpful
<Arahael>
Just did some more googling. It seems that wifi devices often don't reliably do promiscious or monitor mode.
<typetetris>
how do i specify hydra.nixos.org as binary cache? Just adding https://hydra.nixos.org to substituters I get the error, that it isn't a binary cache ?
<Arahael>
sphalerite: So maybe I simply can't easily do it with the pi's wifi stack?
<{^_^}>
[nixpkgs] @FRidh pushed 2 commits to python-unstable: https://git.io/fxXhh
<sphalerite>
you could do NAT
<sphalerite>
but I can't help you with that because I need to get ready for my flight to london for nixcon, sorry!
<Arahael>
sphalerite: No worries, I learnt a lot from teh discussion anyway. :) Thanks for your time!
nD5Xjz has quit [Ping timeout: 252 seconds]
Myrl-saki has joined #nixos
ostkind has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @xeji merged pull request #48641 → mytetra: support opening urls under wayland → https://git.io/fx2A6
<{^_^}>
[nixpkgs] @xeji pushed commit from @gnidorah to master « mytetra: support opening urls under wayland (#48641) »: https://git.io/fxXjS
<typetetris>
srhb: I use that: https://pastebin.com/j5VnZB81 and checking 'nix path-info -I nixpkgs=$(pwd) --json nixpkgs.glibc.static gives `/nix/store/s58hg86980680l8rr9yp9i8m7czzacvj-glibc-2.27-static`, but building that shakeTestbett-thingy it insists on needing `/nix/store/sk2y03klwqxhnf9xjqyc1lw39qd6r7vy-glibc-2.27` I am really confused at the moment.
pie_ has quit [Read error: Connection reset by peer]
<srhb>
colemickens: it.. Sounds like you want buildEnv and nix-build?
<__monty__>
How do I get w3m with images on mac os? I don't mind if I have to run xterm in xquartz to get them. nix-shell -p w3m doesn't seem to be enough.
<srhb>
colemickens: The question doesn't really make sense to me, so I'm extrapolating :P
Guanin__ has quit [Ping timeout: 252 seconds]
<colemickens>
srhb: yeah, it's a lazy script I'm writing
<colemickens>
it lists the contents of an azure container
<colemickens>
lists my store/cache and then uploads the differing files after symlinking them into a directory that the azure cli can bulk upload w/o conflicts.
<srhb>
Like.. Non-deterministically, whatever is in the store currently?
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « python36Packages.outcome: 0.1.0 -> 1.0.0 (#48978) »: https://git.io/fx1IG
<srhb>
Taneb: Bigger than the GHC closure by 2 MiB. :P
<Arahael>
Hmm, the base install is very minimal. :)
<sphalerit>
Libreoffice is pretty horrific.
<srhb>
Arahael: Yup.
<Arahael>
I suppose I want firefox now.
<Arahael>
Does nix-env -i w3m store this into the config file?
<srhb>
Arahael: No
<srhb>
,imperative
<{^_^}>
nix-env has multiple drawbacks as an imperative package manager. nix-env -u will sometimes upgrade to the wrong thing; the outputs to install are very finicky to override; and packages that have been removed or are otherwise unavailable will remain in your profile without any warnings. Consider using a ,declarative setup instead.
<srhb>
Arahael: Consider nix-env equivalent to apt-get or the likes for many purposes. It's completely imperative.
<Arahael>
Odd that this is the first introduction to nixos. :(
<srhb>
Arahael: NixOS means /etc/nixos/configuration.nix :) nix-env is just nix
<Arahael>
How do I use a ,declarative setup' instead?
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « python36Packages.pudb: 2016.2 -> 2018.1 (#48975) »: https://git.io/fx1IH
<srhb>
You _can_ make reasonably declarative setups with nix-env (that's what NixOS does under the hood) -- by eg. creating a single metapackage for your user that contains everything you need. Or by using home-manager
<srhb>
Arahael: But for NixOS, especially on a single user system, you probably just want to get started by adding things to environment.systemPackages in configuration.nix and then nixos-rebuild switch :)
strobelight has joined #nixos
<Arahael>
srhb: Scenario: I've just installed nixos for the first time. How do I declaratively install firefox globally? :)
<srhb>
^
<srhb>
environment.systemPackages
<Arahael>
Reading my mind :)
<srhb>
Basically if I'm crunching for time I might nix-env -iA nixpkgs.firefox -- but later I'll remove it from my imperative env and add it to one of my declarative configs instead.
<srhb>
(Or I might just use nix-shell -p firefox for one-offs)
<srhb>
fsvo firefox :P
<Arahael>
fsvo? :)
<srhb>
Sorry, "for some value of"
<srhb>
Meaning, I'll do that for whatever package that I need ad-hoc right now.
<Arahael>
Ah, yeah, fair enough.
silver has joined #nixos
<Arahael>
What's the best way for me to get firmware into nixos? I suppose I should find that package and put it in as a system package.
<srhb>
depends on the firmware I guess?
strobelight has quit [Ping timeout: 240 seconds]
<srhb>
But yes, that approach sounds more or less right.
<srhb>
Arahael: You probably had a wpa_supplicant.conf from the installers /etc
<Arahael>
Ah, yes, I did...
<Arahael>
sphalerit: No, I rebooted.
<Arahael>
sphalerit: But the originall install did have it....
<Arahael>
(Erm, the live image had it)
<typetetris>
Do I need to setup a cron job for the nix garbage collection or will nix-daemon run it occasionally ?
<sphalerit>
,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 <link to do, sphalerite!>; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
<sphalerit>
Argh I still haven't done that example for 2
<srhb>
Arahael: I don't know that one specifically, but be advised that non-free things are not going to show up unless you configure your system to allow them.
<Arahael>
srhb: Ah, so it's in non-free.
<srhb>
I don't know, just a guess. The UX around discovering non-free is an issue and permitting it is a little horrendous, but it's documented :)
<srhb>
It'll throw an error at you if it's required.
<ekleog>
Lisanna: you mean, originalAttrset // { things not inherited } ?
<Arahael>
srhb: So, literally just: 'firmwareLinuxNonfree' to the hardware.firmware line?
<Lisanna>
ekleog sure
<ekleog>
Lisanna: well, this syntax works then :p
<Lisanna>
ekleog it's ugly
<Lisanna>
sometimes you want to position your inherit statements in specific spots in the attrset, to keep like with like. can't do that with that syntax
mmercier has quit [Quit: mmercier]
<colemickens>
I wish I were going to nixos so badly
<Lisanna>
unless you start building a chain of //
<Lisanna>
but that is pretty verbose
<ekleog>
chains of // work well, but I see your point
<srhb>
Arahael: By default your configuration.nix is a function { pkgs, ... }: -- so pkgs is the name of the nixpkgs attribute set.
<Lisanna>
ekleog if the author of some nix code wants to use that feature, that's just a caveat they have to keep in mind. no actual retro-compat would be broken for code that doesn't use it.
<Lisanna>
there are many ways today to write unstable / unsafe nix code.
kim0 has joined #nixos
<kim0>
Hello folks .. I see xmr-stak is updated on github to 2.5.1 .. but 5 days later, it's not yet showing in unstable channel .. When should it land ? Thanks
<ekleog>
Lisanna: I mean, should it trigger an evaluation error? should it pick the version of the upper set? what if two inherit (set) *; have keys that conflict?
<ekleog>
Lisanna: OTOH, { foo = 1; } // set // { bar = 1; } is not much longer than { foo = 1; inherit (set) *; bar = 1; }
<Lisanna>
ekleog it should do whatever doing { foo = bar; inherit (baz) foo; } does
<ekleog>
this eval-time fails
<Lisanna>
then that's what it should do
<Lisanna>
ekleog also, your example is wrong
<Lisanna>
it needs to be ({ foo = 1; } // set // { bar = 1; })
<Lisanna>
(note the parens)
<ekleog>
well, it depends on how you use it
<Lisanna>
half of the time it's going to be arguments to a function of some sort
<ekleog>
> let set = { baz = 1; } in { foo = 1; } // set // { bar = 1; }
<{^_^}>
error: syntax error, unexpected IN, expecting ';', at (string):205:24
<Lisanna>
which will screw you over if you don't have the parens
<ekleog>
> let set = { baz = 1; }; in { foo = 1; } // set // { bar = 1; }
<{^_^}>
{ bar = 1; baz = 1; foo = 1; }
<kim0>
srhb: Thanks .. very useful! But I'm still confused .. nixpkgs-unstable (I guess this is what I want, as I'm on Ubuntu) is showing 19hours old .. but xmr-stak which was updated 5 days ago is still not showing ?!
<srhb>
kim0: Are you really on nixpkgs-unstable, not nixos-unstable?
<ekleog>
Lisanna: then, you can always open an issue on nixos/nix :)
<srhb>
kim0: OK, yes, it's not in nixpkgs-unstable yet.
<kim0>
Ok how / why
<ekleog>
it's trading one footgun for another so I'm personally not in favor of adding new syntax for its sake, but other people may disagree with me :)
<kim0>
What does this 19hours old thing mean then
<srhb>
kim0: The update time refers to when the channel updated, not which "commit time" it has reached
Guanin__ has joined #nixos
<srhb>
If you look at the commit it has updated to, xmr-stak is not updated.
<kim0>
srhb: so it was updated 19hours ago, but took an old commit (not tip of master) ?
<srhb>
It was the tip of master at the point where the hydra build/test commenced.
<srhb>
Building all of nixpkgs might take a long time, however, so there's always latency.
<kim0>
Aha Ok
<kim0>
so I guess a few more days of waiting
<Izorkin>
Mic92: How to need to merge PR 48875 ?
<Mic92>
Izorkin: can you remind me later about that, maybe in the weekend? I need to prepare something.
<schopp0r>
Hi. I am trying to compile nix myself (I need another store directory, that is why). There is gcc 4.9.4 installed which can *definitely* compile c++14 source code (I tried). anyway, the configure-script insists on g++ not being able to do so. can I somehow enforce this test to be passed?
<srhb>
ghasshee: nix-shell -p coq coqPackages.ssreflect -- seems to be the way to go.
<ghasshee>
srnb: maybe I could install with the packaged version of nix/store but I do not know the way of including ssreflect then.
<ghasshee>
I tried on archlinux with opam package manager which easily worked..
<srhb>
We're very, very different from Arch. :)
<ghasshee>
Yep :-) !!
<ghasshee>
I am graduating it now !
<srhb>
A lot of work has to go into bringing some obscure builds to work in nixpkgs. Someone already did this work for you, so I recommend you try to make good use of that work, instead of trying to essentially replicate what they had to do but in an impure way :)
<ghasshee>
Thanks, I will do another try on the packaged nix binaries.
<srhb>
ghasshee: Doesn't the above command work for you?
<srhb>
nix-shell -p coq coqPackages.ssreflect <-- this one
jasongrossman has joined #nixos
<ghasshee>
maybe I tried with my shell.nix ... but I try it again now
<srhb>
It looks like it works to me, but I've no experience with the coq ecosystem.
<ghasshee>
and then I opened coqide
<ghasshee>
but coqide does not seem to include ssreflect ..
<ghasshee>
(coqide:24022): GtkSourceView-WARNING **: Failed to load '/nix/store/zajfp56idzicrsz2fcdp1wz8xjqd9m97-coq-8.7.2/share/coq/coq.lang': could not find the RelaxNG schema file
<hyper_ch2>
I'm creating my own snapshot tool and I still wonder what format I should use for the snapshot naming.... I'm currently pondering something like: @1540380794_2018-10-24-13h33-CEST_easysnap_hourly --> the unix time in the beginning will allow easy sorting; the time representation will give an impression when the snapshot was created; the last part "easysnap" "hourly indicate my tool was used and which kind of snapshot it is... but it seems s
o1lo01ol1o has quit [Ping timeout: 264 seconds]
<hyper_ch2>
infinisil: any thoughts on that?
<infinisil>
hyper_ch2: Why not just easysnap-2018-10-24-13:33 ?
<hyper_ch2>
infinisil: because I also want to indicat hourly, daily, weekly, monthyly and be able to easily delete them
<srhb>
Someone already invented this wheel for you. :)
<infinisil>
Ah yeah, then just add hourly, but the unix timestamp doesn't do much
<hyper_ch2>
srhb: I fail to see the connection
<srhb>
It's a timestamp format that is human readable, is standardized, and where lexicographical sort = chronological sort
<infinisil>
But really, this is offtopic for #nixos, how about #nixos-chat?
<hyper_ch2>
the unix timestamp is the best way to indicate when it was created - if you change timezones you might have conflicting information of which one was taken earlier
<Arahael>
srhb: Sorry, I had to head out. So, I eventually tried `hardware.firmware = [firmwareLinuxNonfree]`, but that doesn't appear to be correct syntax.
<symphorien>
try pkgs.firwareLinuxNonfree
<srhb>
Arahael: I wrote a message. You probably have your pkg set called pkgs
<srhb>
Arahael: Unless you really did get a syntax error, and not just an undefined reference. :)
<Arahael>
srhb: Ah, and every line has to end in ; :)
<srhb>
No. But every attribute being set does. :-)
<Arahael>
srhb: Makes sense. :)
<Arahael>
Rebooting to test - ha, five generations already. :)
<srhb>
:-)
<Arahael>
Woo, and wifi magically works!
* Arahael
wonders if video is correct.
<srhb>
Arahael: This laptop is one week old, generation 40 :-P
<Arahael>
So I've debugged my wife's program (she's playing with 'micro:bit'), and installed nixos on my new computer, and fixed an email issue for a small business... I think I should call it a night. :)
<srhb>
Arahael: Sleep well and keep having fun. :-)
<Arahael>
Thanks. :)
TweyIV has quit [Ping timeout: 252 seconds]
astronavt has joined #nixos
<rawtaz>
Arahael to his wife in bed when it's time; "- Sorry hun, i've had enough sex for today"
astronavt_ has joined #nixos
pie_ has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
pie_ has quit [Remote host closed the connection]
<srhb>
I don't think that's appropriate for this channel, in the future. :)
<__monty__>
How do I get w3m with images on mac os? I don't mind if I have to run xterm in xquartz to get them. nix-shell -p w3m doesn't seem to be enough.
<schmittlauch[m]>
What is the best way to see how input parameters of an expression have been/ are set? I want to figure out the current values of the vim arguments (e.g. what is given as the pythonSupport argument)
<infinisil>
> fun = { foo ? throw "will never occur", ... }?args: if args ? foo then "Foo was set to ${foo}" else "Foo unset"
<{^_^}>
error: syntax error, unexpected '?', expecting ':' or '@', at (string):80:47
<infinisil>
> fun = { foo ? throw "will never occur", ... }@args: if args ? foo then "Foo was set to ${foo}" else "Foo unset"
<{^_^}>
fun defined
<infinisil>
> fun { foo = "foo"; }
<{^_^}>
"Foo was set to foo"
<infinisil>
> fun { }
<{^_^}>
"Foo unset"
<infinisil>
schmittlauch[m]: Like this? ^
<schmittlauch[m]>
infinisil (IRC): preferrably without changing the expression itself, but like inspecting it from the nix repl
ghasshee has joined #nixos
Itkovian has joined #nixos
<infinisil>
What expression is it?
jperras_ has joined #nixos
<schmittlauch[m]>
infinisil (IRC): e.g the vim_configurable expression has an argument `multibyteSupport ? config.vim.multibyte or fals;`
<schmittlauch[m]>
*false
<schmittlauch[m]>
I want to obtain how this argument set looks either from my current system configuration or for a built package (depending on what's easier), independently from examining config.vim.
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python36Packages.Nuitka: 0.5.25 -> 0.6.0.4 »: https://git.io/fx18h
<Guanin>
Hi, before I break anything by rebooting: `nixos-rebuild switch` prints the line "Failed to start local-fs.target: Unit boot.mount is masked." - can that cause any problems?
<infinisil>
layus: Oh, that might work after all, sorry
Theuni2 has joined #nixos
Theuni2 has quit [Client Quit]
<infinisil>
Because cfg is an attribute of the mkDerivation
<infinisil>
Didn't realize
Theuni2 has joined #nixos
crmlt has joined #nixos
<layus>
Yes, it is an argument of vim_configurable, not vim
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « sslscan: 1.11.11 -> 1.11.12 »: https://git.io/fx1BW
<layus>
Well, I cannot seem to make it work.
<Guanin>
ottidmes, I did not mask anything, that's why I'm asking. besides temporary manipulation of units (systemctl start/stop), I only use nixos-rebuild
<colemickens>
emily: I actually have it working, with lots of automation too, and a halfway decent README. I'd really appreciate any feedback. https://github.com/colemickens/nix-overlay-sway
iyzsong has joined #nixos
amosbird has quit [Max SendQ exceeded]
amosbird has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<schopp0r>
hi. what do I need to pass the nix ./configure script to prevent it from building the documentation? because it fails to build and I do not need it.
amosbird has joined #nixos
<schmittlauch[m]>
Ah, I guess I found the problem…
<schmittlauch[m]>
does environment.SystemPackages have precedence over user environment packages?
sir_guy_carleton has quit [Ping timeout: 272 seconds]
amosbird has quit [Max SendQ exceeded]
endformationage has joined #nixos
amosbird has joined #nixos
amosbird has quit [Max SendQ exceeded]
amosbird has joined #nixos
amosbird has quit [Max SendQ exceeded]
amosbird has joined #nixos
astronavt_ is now known as astronavt
<disasm>
how do I just `nix-build` a single nixos module without actually doing anything with it?
amosbird has quit [Max SendQ exceeded]
amosbird has joined #nixos
<ottidmes>
disasm: not sure if this is what you mean, but I have this in one of my scripts to make sure everything is build for that piece of config: nix-build --no-out-link --expr '(import <nixpkgs/nixos> {}).config.hosts'
<ottidmes>
where hosts is something I have defined in one of my modules
<disasm>
schmittlauch[m]: that'll depend on the order of your $PATH. In mine it does not (just checked)
<disasm>
ottidmes: more of I have a module foo.nix and I want to build it with services.foo.enable = true; but I don't actually want to create a configuration for it, but it takes a { config, lib, ... } as arguments.
<disasm>
if it's not easy, I can create a stub configuration.nix to test it with, just was wondering if there's some way to do it.
jperras_ is now known as jperras
amosbird has quit [Max SendQ exceeded]
Guanin has quit [Remote host closed the connection]
schopp0r has quit [Quit: Page closed]
amosbird has joined #nixos
<layus>
Any informal gathering tonight before NixCon ?
<ottidmes>
disasm: still not 100% sure what you want to achieve (or do you mean using a module outside of the module system?), so I would go with a stub if that works for you
<greymalkin>
I need to figure out how to make sure the other dependent derrivations don't disappear first... it's a haskell build, and I've already gone through the near-day-long pain of building the profiling version of "the world" for these projects.
<nyanloutre[m]>
if you keep the result seemlink of your app, dependancies should stay in the store
pie_ has quit [Ping timeout: 264 seconds]
amosbird has quit [Max SendQ exceeded]
amosbird has joined #nixos
slack1256 has joined #nixos
Tucky has quit [Remote host closed the connection]
<exarkun1>
are the derivations given in hackage-packages.nix built on CI or anything like that? is there any expectation that they will Just Work, as-is? or should I be unsurprised if they fail because there's no guarantee a human has ever actually looked at any of them?
<IslandUsurper>
I have a home server running Debian Jessie that I'm looking at installing NixOS on. If I put /home, /var, and /etc in NIXOS_LUSTRATE, how many problems will I have?
<tilpner>
Do you have backups?
amosbird has quit [Max SendQ exceeded]
<IslandUsurper>
I'm not too worried about keeping it intact. It would just be nice if it worked.
sb0 has joined #nixos
amosbird has joined #nixos
<symphorien>
IslandUsurper: I am not sure you can keep /etc
<IslandUsurper>
tilpner: I'm not sure what that gets me.
amosbird has joined #nixos
<tilpner>
IslandUsurper - It allows you to repartition and format freely, and then install a clean NixOS
<IslandUsurper>
But I can already do that with the live CD, right?
amosbird has quit [Max SendQ exceeded]
<tilpner>
If you attach the live CD, sure, go for that
<IslandUsurper>
Sorry if I'm being obtuse. I probably need more sleep.
amosbird has joined #nixos
amosbird has quit [Max SendQ exceeded]
<tilpner>
But if you can do that, I don't see why you asked abotu LUSTRATE in the first place
amosbird has joined #nixos
<tilpner>
kexec works great with remote servers, where you can't attach a live medium
<IslandUsurper>
I wanted to keep the existing package configurations and settings while changing the fact that they were installed by Nix instead of APT.
<IslandUsurper>
So, I'll just backup those directories and put them back afterwards
<IslandUsurper>
more or less
<tilpner>
Oh, maybe you're talking about data in e.g. /var/lib, and want to keep that
<tilpner>
I read that as if you wanted to automatically translate installed debian packages to their nixpkgs counterparts
<simpson>
IslandUsurper: /etc is managed by NixOS though. What, precisely, is your plan for migrating your configuration?
<drakonis>
tilpner: lustrate works fine btw
<IslandUsurper>
simpson: starting a convo here and seeing what more experienced people would say. :P
<drakonis>
i have used it myself
<tilpner>
drakonis - It might, and I didn't say it can't work, it just didn't the one time I tried
Ariakenom has joined #nixos
<thefloweringash>
What's the meaning of NIX_PATH under Nix 2.0? If I have NIX_PATH=/foo, where /foo/nixpkgs is the contents of the main package set, should I expect `nix build nixpkgs.ncdu` to succeed?
<IslandUsurper>
So I guess the things I've changed in my current /etc files will need to be added to the nix-expressions I use on NixOS. And /home and some subdirectories of /var can be restored from backup as needed.
<kreisys>
halp! I keep hitting that same thing from different angles... so my release.nix uses `builtins.fetchGit ./.` for the src as seems to be the current way to do it; however, I'd like to filter out the *.nix files from my source before I start building. That is, to avoid nix unnecessarily rebuilding the entire source just because a non-source file changed. However, I get that annoying 'string cannot refer to other paths'
<kreisys>
error. Is there no "correct" way to filter a store path?
<mdash>
kreisys: this is what the filterSource function is for
<{^_^}>
[nixpkgs] @etu opened pull request #49052 → dmrconfig: init at 2018-10-20 → https://git.io/fx1F8
<Taneb>
I'm back at it again, trying to install NixOS on my new laptop
<bpye>
Taneb: good luck :)
<Taneb>
Now I get a few messages while booting that seem scary
<Taneb>
And then it stops before giving me a prompt
<{^_^}>
[nixpkgs] @lokke-me opened pull request #49054 → homely: init at 0.15.3 → https://git.io/fx1Fj
<Taneb>
"sp5100-tco sp5100-tco: Watchdog hardware is disabled", "mmc0: Unknown controller version (3). You may experience problems." and "kvm: disabled by bios"
<Taneb>
Any of those look familar
slack1256 has joined #nixos
Mr_Keyser_Soze99 has joined #nixos
slack1256 has quit [Remote host closed the connection]
slack1256 has joined #nixos
<ottidmes>
Taneb: If you dont use virtualization, the kvm error can be safely ignored
acarrico has quit [Read error: Connection reset by peer]
acarrico has joined #nixos
zazzi has left #nixos ["Leaving"]
hotfuzz_ has quit [Read error: Connection reset by peer]
_kwstas has joined #nixos
hotfuzz_ has joined #nixos
<_kwstas>
I'm working on a haskell project, i have already configured project.nix and default.nix. I want to configure shell.nix as well, but I want shell.nix to have some extra dependencies (e.g. hlint), but I do not know how :(. I tried to do something like: devEnv = (import ./project.nix).project.overrideAttrs (oldAttrs: rec { libraryHaskellDepends = oldAttrs.libraryHaskellDepends ++ [ pkgs.haskellPackages.hlint ] ; }); but it seems that thi
<_kwstas>
s does not work. any help?
xeji has joined #nixos
<{^_^}>
[nixpkgs] @veprbl opened pull request #49055 → pyarrow darwin debug → https://git.io/fx1AQ
<{^_^}>
[nixpkgs] @Tomahna opened pull request #49056 → scalafix: init at 0.9.0 → https://git.io/fx1xC
boxscapeR has joined #nixos
sigmundv has quit [Ping timeout: 264 seconds]
<boxscapeR>
For some reason, I sometimes cannot access the nixos.org domain, but only from machines that have nixos installed. Has anyone else experienced this?
<{^_^}>
[nixpkgs] @bendlas pushed 6 commits to release-18.09: https://git.io/fx1pK
<boxscapeR>
ottidmes: weirdly, it differs. Right now, I can access nixos.org with firefox, but ping says "Name or service not known". I can't access nixos.wiki, and ping says "connect: Network is unreachable". Sometimes when I can access nixos.org, ping finds the IP, but then just times out.
<Taneb>
OK, I seem to be stuck on "Starting Load/Save RF Kill Switch Status..."
JosW has joined #nixos
<ottidmes>
boxscapeR: what does your /etc/resolv.conf file look like, maybe try the DNS of Google 8.8.8.8 and 8.8.4.4 or of Cloudflare 1.1.1.1 and 1.0.0.1
<ottidmes>
Taneb: what kind of laptop do you have?
<Taneb>
ottidmes: Thinkpad E486
<Taneb>
*E485
<boxscapeR>
ottidmes, huh, yeah, that should work, actually, thanks for the tip. I just tried pinging nixos.wiki on my phone, and if I enter the IP directly, I can access it. I didn't change the resolv.conf file from the default
statusfailed has joined #nixos
<ottidmes>
boxscapeR: then try this setting: networking.nameservers = [ "8.8.8.8" "8.8.4.4" ];
<boxscapeR>
that's in /etc/nixos/configuration.nix?
<ottidmes>
Taneb: you just tried a basic install (no real config done yet, except bootloader and basic hardware)? I see the E485 uses Ryzen thus you might need the latest Linux kernel for everything to work properly
<kingemp>
Hello, I’m a new nixos user and just managed to break my boot process. Fortunately due to the glory that is nixos, I can still boot an old generation. However I haven’t figured out how to fix my configuration.nix / hardware-configuration.nix
<kingemp>
ottidmes: I deleted the new /data mount in hardware-configuration, otherwise I *think* it looks good
xeji has quit [Client Quit]
orivej has joined #nixos
kingemp_ has joined #nixos
<ottidmes>
kingemp: as long as you mounted your root, and boot (if you have this separately, as most have), your swap (not necessary, but if you plan to use it, dont forget about it), then the rest should work
<kingemp>
ottidmes: fixed! Turns out the nixos-generate-confit
<emily>
nixos-generate-confit :D
<kingemp>
g command switched to UUID instead of label, and using label works
<kingemp>
No confit was had :(
<kingemp>
Hint was in feta
<kingemp>
Wow autocorrect killing me
<kingemp>
fstab
<kingemp>
iPad knows it’s lunchtime
cmacrae has joined #nixos
<{^_^}>
[nixpkgs] @Izorkin opened pull request #49069 → systemd: enable extra configuration to the [Service] section → https://git.io/fxMvG
fragamus has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<hyper_ch>
people still here or everyone on their way to london?
slack1256 has quit [Remote host closed the connection]
slack1256 has joined #nixos
o1lo01ol1o has joined #nixos
<palo>
tilpner: seams like you can only download the newest version from the jetbrains servers. the current version works fine
<ottidmes>
kingemp: gratz! then I assume you remade the file system on one the partitions causing the UUID to change
<kingemp>
ottidmes: I added a label to the /data drive (sdb), but didn’t touch sda,
<kingemp>
So it seems like for some inexplicable reason the UUID changed between generations, although I never could boot to verify. In any case, happy to use label if things boot!
<emily>
colemickens: I like your binary cache's hostname
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « cabal2nix: fix build with ghc 8.4.x »: https://git.io/fxMm3
drakonis_ has joined #nixos
ottidmes has joined #nixos
astronavt has joined #nixos
drakonis has joined #nixos
drakonis1 has joined #nixos
drakonis_ has quit [Ping timeout: 260 seconds]
markuskowa has joined #nixos
Ariakenom has quit [Quit: Leaving]
<Taneb>
AHA! I've finally got it!
<Taneb>
Now to see if I can get NixOS on here before nixcon :)
vaibhavsagar has joined #nixos
revtintin has quit [Quit: WeeChat 1.9.1]
vaibhavsagar__ has quit [Ping timeout: 246 seconds]
acarrico has joined #nixos
dmc has quit [Quit: WeeChat 2.3]
<kalbasit[m]>
rycee: I'm having issue booting nixos vm `nixos-rebuild build-vm` when the home manager is installed as a nixos module. The home-manager-<username>.service fail because it can't write to the nix store. Have you ever had success with that?
<drakonis1>
oh yeah, nixcon later this week
<drakonis1>
tomorrow, even.
<rycee>
kalbasit[m] (IRC): Unfortunately you need to use the nixos-module-user-pkgs branch for build-vm support.
dmc has joined #nixos
<kalbasit[m]>
oh
michalrus has joined #nixos
<kalbasit[m]>
let me give that a try
<michalrus>
Hello, is there some `stripHash` *Nix* function, not a Bash one?
lor|d is now known as lord|
<drakonis1>
is there any reason why the nixcon domain isn't just *just* nixcon instead of nixcon+year?
<kalbasit[m]>
rycee: are you planning to merge that branch to master?
<rycee>
kalbasit[m] (IRC): I'd like to merge that branch into master at some point but there are some issues with it, at least there were a long time ago. Need to revisit and dig into it in the future.
<kalbasit[m]>
cool
<rycee>
It won't happen in the near future, though.
<rycee>
I try to rebase it over master occasionally.
<kalbasit[m]>
rycee: could you possibly open a PR and list out the TODOs and the problems? Maybe I can help get it there
<rycee>
kalbasit[m] (IRC): The problem was mainly concerning Qt programs.
Thra11 has joined #nixos
<rycee>
That couldn't find Qt plugins when they were installed in the user-pkgs directory, or something along those lines.
<kalbasit[m]>
I see
<boxscapeR>
is there a canonical directory to mount drives to in nixos? /mnt maybe?
<rycee>
The fontconfig issue might be resolved.
<kalbasit[m]>
so probably the QT issue is left
<ottidmes>
boxscapeR: I believe that convention is mostly dictated by the desktop environment and whatever mounting software is being used
<kalbasit[m]>
I'll see if I can replicate
<boxscapeR>
ok, thanks ottidmes
boxscapeR has quit [Remote host closed the connection]
<ottidmes>
boxscapeR: for manual mounts I tend to use /mnt, but my file manager for example uses /run/media to mount devices
boxscapeR has joined #nixos
<greymalkin>
If I'm doing haskell overrides (super.callPackage ../custom_package.nix {}) -- where should I put a 'dontCheck' (the build environment won't have credentials to run the tests, so it will have to be done locally only)
<greymalkin>
I've been trying to read through haskell-modules/lib.nix and have not gotten anywhere towards understanding for the last hour or so.
<kalbasit[m]>
any way to change the default RAM size of 384 when `nixos-rebuild build-vm`?
<rycee>
michalrus (IRC): I don't know if such a function. I guess it would be possible to implement one using builtins.match.
<michalrus>
rycee: okay, thank you!
sigmundv has quit [Remote host closed the connection]
<rycee>
kalbasit[m] (IRC): That would be much appreciated!
boxscapeR has quit [Remote host closed the connection]
<o1lo01ol1o>
Is there a command to get the list of all users on a system?
<symphorien>
getent passwd
<ottidmes>
michalrus: let stripHash = path: let m = builtins.split "^[a-z0-9]{32}-" (baseNameOf path); in if builtins.length m == 3 then builtins.elemAt m 2 else path; in [ (stripHash "/nix/store/88h69bc0cs956z57xipiy1af8risxyih-etc.drv") (stripHash "etc.drv") ]
<michalrus>
ottidmes: do you agree tho that this should go to lib/filesystem.nix?
<michalrus>
And if not, why?
<ottidmes>
michalrus: I wanted to make the argument that NixOS tends to be conservative with the lib to only keep general stuff in it, and stuff that is actually being used, but looking at lib/filesystem.nix I find haskellPathsInDir...
acarrico has quit [Ping timeout: 260 seconds]
<michalrus>
Yea… :p
<ottidmes>
michalrus: but personally I agree with you, it should be there, but you could make the case, if it really was needed often, it would already have been added in the past
<michalrus>
I see. =)
<michalrus>
But you’re not going to make a PR?
<ottidmes>
michalrus: no, I did not plan to
<michalrus>
Okie. *taking notes to do so maybe*
luigy has joined #nixos
Thra11 has quit [Ping timeout: 272 seconds]
camsbury has quit [Remote host closed the connection]
Khetzal has quit [Quit: No Ping reply in 180 seconds.]
<colemickens>
I am typing `nix-env -f ... -iA wlroots.examples`, then `rootston` appears on PATH, but it's clearly in the wlroots-bin package, not -examples?
<clever>
pie_: nix-env -q does exactly what you want
<pie_>
hm ok
<ottidmes>
This is offtopic, but on NixOS I use pass and rofi-pass to manage my passwords, but I sometimes need to get access to them outside my NixOS installs. I was wondering what others use to manage this problem
astronavt has quit [Changing host]
astronavt has joined #nixos
endformationage has quit [Ping timeout: 268 seconds]
sigmundv has quit [Read error: Connection reset by peer]
<Arahael>
Hmm, I use keepass, but I use dropbox to synchronise it.
<clever>
i just use lastpass
<arianvp>
pass has git support
<arianvp>
for syncing
sigmundv has joined #nixos
<Arahael>
clever: I stopped using lastpass when LogMeIn acquired them. :)
<fiatjaf>
what about bitwarden?
<pie_>
is it possible to disable imperative configuration?
<ottidmes>
pie_: You mean installing things through nix-env, right? Why the need to disable it? I simply don't use it