<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #41939 → haskell generic-builder: Always use separate pkg db for custom setup → https://git.io/vhKxk
<adisbladis[m]>
tenten8401: I want some of those full laptop-size nixos stickers I've seen some people having
<tenten8401>
as in like a skin for the entire back of the laptop?
<tenten8401>
I'm probably going to end up putting one of these on the back of my laptop:https://www.redbubble.com/people/tenten8401/works/32301023-nixos-logo-w-right-text-sticker
<adisbladis[m]>
tenten8401: No, the NixOS logo is cut out
<tenten8401>
so it covers the back of the entire laptop but the snowflake is cutout from it?
<adisbladis[m]>
Yeah :)
<tenten8401>
that sounds cool
<tenten8401>
I wonder if my dad still has his sign cutter laying around
blahdodo has quit [Quit: Bye bye]
<tenten8401>
probably uses some old windows 95 drivers though
<adisbladis[m]>
tenten8401: I know someone owning a sticker factory. So might be able to get good pricing.
<tenten8401>
interesting
<tenten8401>
I'd like to see the sticker on a laptop first if anything though
<rizary>
so re chmod, where should i put it? is it in `buildPhase` or `installPhase` then should I revert it back to original condition?
<clever>
rizary: in the install phase, after you copy the frontend
<rizary>
oh re LOCALE_ARCHIVE, i am getting `not a valid identifier`
<clever>
you called it nixpkgs in your file, so nixpkgs.glibcLocales
<rizary>
yep, i try it in install phase and it works. I mean if is it possible that i am doing the copy and chmod in `buildPhase` then revert the chmod back again in `installPhase`?
<clever>
nix-daemon will revert all the chmod bits for you when the build is done
doyougnu has quit [Quit: WeeChat 1.9.1]
<clever>
everything will be made read-only and owned by root
<rizary>
i see, i thought when i `inherit (nixpkgs) pkgs;` i already get the `glibLocales`
<rizary>
thanks again clever
<clever>
ah, you do
<tenten8401>
Anyone know why I could be getting an "error: failed to get canocial path of /boot/efi" error when installing grub?
<tenten8401>
I'm installing it to 2 drives that are in a raid array
<tenten8401>
using the grub devices array in my configuration
<tenten8401>
while it does work on one drive, the other fails
<clever>
tenten8401: can you pastebin your entire configuration.nix file?
<tenten8401>
yeah, 1s
matthewbauer has quit [Ping timeout: 256 seconds]
<tenten8401>
slowly booting the live image and mounting again
<tenten8401>
any day now
<vaibhavsagar>
Has anyone seen "Failed to start systemd-journal-catalog-update.service: Transaction order is cyclic. See system logs for details." before?
<vaibhavsagar>
it's looking like "sysinit.targete/start" and "systemd-journal-catalog-update.service/start" depend on each other
<tenten8401>
if you'd like access to my server to screw around I can give it
<tenten8401>
there's literally nothing on it
<tenten8401>
bought it like a couple hours ago and have been trying to install NixOS since
schoppenhauer has quit [Ping timeout: 260 seconds]
tzemanovic has quit [Remote host closed the connection]
<clever>
tenten8401: not sure how to fix your /boot raid issues, but i can help you mess with kexec if you want to compare it to qemu and see which is simpler
schoppenhauer has joined #nixos
<tenten8401>
I really don't mind what I use
<tenten8401>
qemu seems to work well enough
<tenten8401>
I'm clueless on what to do for this NixOS install
<clever>
about all i can think of is to do a non-mirrored /boot, you can always recreate /boot from the rescue system
<jcrben>
Baughn: fyi, linked out to the IRC log in your issue (https://github.com/NixOS/nixpkgs/issues/42178) - hope that's OK. and I gave it a thumbs up, not that my opinion counts for anything :) I think minimal number of recommended glue languages is best - right now Python is pretty embedded into the nix world since https://github.com/NixOS/nixops is wr
<jcrben>
itten in it (as well as hydra-provisioner). but ofborg is in Rust and Rust is a language that a lot of people (including myself) would like to work with
tzemanovic has joined #nixos
<jcrben>
I suppose the compilation into a static binary is a big point in favor of Rust over Python
<hyper_ch>
clever: still can't install nixos from your kexec :( gnutls still fails
<bluej774>
So, what are the minimum requirements to install Nix? I know it needs curl. I think it also needs a version of groff, although only to run --help commands and you can install that from Nix after the fact. Is there anything else?
<jackdk>
patience and a good network connection?
blankhart has joined #nixos
<clever>
bluej774: there is also a bash script that just downloads a pre-compiled copy of nix
JasonGrossman has quit [Ping timeout: 264 seconds]
wchresta has joined #nixos
woodson has joined #nixos
rprije has quit [Ping timeout: 256 seconds]
MercurialAlchemi has joined #nixos
Guanin has quit [Ping timeout: 248 seconds]
ThatDocsLady has joined #nixos
Guanin has joined #nixos
sary has joined #nixos
<vaibhavsagar>
Has anyone seen "Failed to start systemd-journal-catalog-update.service: Transaction order is cyclic. See system logs for details." before? I'm getting it when upgrading from 17.09 -> 18.03
<colemickens>
Do the channels advance automatically? It's hard for me to correlate the last advancement with a specific hydra run. It seems like even for nixos-unstable, there's always some packages that fail.
camsbury has quit [Ping timeout: 268 seconds]
et4te has quit [Ping timeout: 260 seconds]
<cocreature>
colemickens: afaik whenever the tests pass, the channels advance
<etu>
The later link is the correct link to look at :)
<etu>
When all the ticks are green in a column and certain builds are populated in the cache the channel will advance
<etu>
And at the moment it looks very bad, but it's not as bad as it looks because the red circles means "canceled" and not actual fails
stepcut has quit [Remote host closed the connection]
<bluej774>
clever, I don't know how reliable it'll be in the face of updates, but this works for the time being: `wget -q -O - https://nixos.org/nix/install | sed 's/require_util curl/require_util wget/' | sed 's/curl -L "$url" -o "$tarball"/wget -q -O "$tarball" "$url"/' | sh`
stepcut has joined #nixos
stepcut has quit [Read error: Connection reset by peer]
stepcut has joined #nixos
endformationage has quit [Ping timeout: 248 seconds]
<manveru>
i'm wondering what all those `{memory:\d+}` gc roots are...
smallville7123 has quit [Quit: rosa]
smallville7123 has joined #nixos
smallville7123 has quit [Excess Flood]
smallville7123 has joined #nixos
agander has joined #nixos
woodson has quit [Ping timeout: 265 seconds]
vaninwagen has joined #nixos
{^_^} has quit [Ping timeout: 260 seconds]
abcrawf has quit [Ping timeout: 260 seconds]
palo_ has joined #nixos
palo_ has quit [Changing host]
palo_ has joined #nixos
thc202 has joined #nixos
smallville7123 has quit [Quit: rosa]
palo has quit [Ping timeout: 245 seconds]
<tfc[m]>
hey there. i have a project that i try to build in a nix shell but it does not work becasue the build system (`scons`) surpresses all the environment variables. so the result i am fighting against is the c compiler not finding <stdint.h> - which env variables do i need to restore to fix this?
<Mic92>
jarlg: it is in the bluez package: nix-shell -p nix-index --command 'nix-locate libbluetooth'
<Mic92>
bluez.out
smallville7123 has quit [Quit: rosa]
<yurb>
Hi all. I couldn't find a description of the process of security updates in libraries - what happens if something like the openssl heartbleed appears? Everything gets rebuilt?
<jarlg>
Mic92: Ah, thanks!
johann__ has joined #nixos
smallville7123 has joined #nixos
<JasonGrossman>
yurb: Yes. You pick up the updates (including new builds) when you use nixos-rebuild.
<JasonGrossman>
yurb: Hydra makes the new binaries and caches them for you to download (not instantly, but pretty quickly).
<yurb>
JasonGrossman: and if I'm using just nixpkgs on another distro? Just regular nix-env --upgrade?
troydm has quit [Ping timeout: 276 seconds]
smallville7123 has quit [Client Quit]
agander_ has quit [Ping timeout: 256 seconds]
smallville7123 has joined #nixos
smallville7123 has quit [Excess Flood]
tzemanovic has quit [Remote host closed the connection]
<JasonGrossman>
yurb: You're looking at it! It's not as well categorised as the debian list, but updates are announced here.
<JasonGrossman>
That's all updates, not just security updates.
<yurb>
JasonGrossman: do you mean this IRC channel? :)
<JasonGrossman>
Yes. There's no email list as far as I know. There is a NixOS security team that's responsible for rapid response on security problems, but as far as I know their only job is to fix things not to do any particular announcement ... although they might for something as big as heartbleed.
<yurb>
I see. Thanks.
<JasonGrossman>
yurb: Also, NixOS has an option to automate upgrades, but as far as I know nixpkgs doesn't ... although of course it would be a very simple cron job.
<{^_^}>
[nixpkgs] @Mic92 merged pull request #42182 → nixos-18.03/pgbackup: Fix bug in postgresql-backup module that causes to fail → https://git.io/ffCMt
<{^_^}>
[nixpkgs] @Mic92 pushed 2 commits to release-18.03: https://git.io/fUkpi
andymandias has quit [Ping timeout: 276 seconds]
<yurb>
JasonGrossman: I once had this funny situation where I had automatic upgrades setup on Fedora. I had an Ardour audio project setup for a performance. Then I travelled to another city and right before the concert I see that Ardour has been updated from 4 to 5 :) Luckily it all worked.
<JasonGrossman>
WOW!
<JasonGrossman>
Also, if you say that then you have to post a link to your music.
<JasonGrossman>
Also, that's why I like the Debian announcement mechanism. Maybe we'll have that for nixpkgs one day.
agander has joined #nixos
<manveru>
Lisanna: you can prefix your `let` with `with`, no?
astsmtl has quit [Ping timeout: 260 seconds]
<yurb>
JasonGrossman: sent in pm to not spam the channel. Recently I started using nixpkgs to install the audio stuff I use on Fedora - SuperCollider, jack, etc.
<Lisanna>
manveru I'm not trying to add it to my lexical scope, I'm trying to put it in the actual attrset
<Lisanna>
There doesn't seem to be a language feature for it
<manveru>
Lisanna: does `//` work for you then?
<Lisanna>
you have to do it from outside
<Lisanna>
or specify everything manually
tzemanovic has joined #nixos
<Lisanna>
it's not a /huge/ deal, would just be nice to have a feature for doing it with an inherit statement
<goibhniu1>
yurb, I'd also like to hear your stuff :)
<vaninwagen>
hi, i do have some script here, which i cannot control, which uses /bin/bash in its shebang. I need to make this work on my system. Is there a prefered way to link my bash to /bin/bash ?
<vaninwagen>
i know, i would love to have /usr/bin/env bash, but this is unfortunately not possible in this case :/
<vaninwagen>
Baughn: manveru unfortunately i cannot control how it is called. It is just called as an executable via execve. So i need something at /bin/bash. i just found that /bin/sh is also just a link to bash in my nix-store
<Baughn>
vaninwagen: You can't have something at /bin/bash, but you *can* run the whole shebang in a chroot via steam-run.
<JasonGrossman>
Baughn: You can have something at /bin/bash if you don't mind recreating it every time you rebuild NixOS. And of course you can automate that. Inelegant, but it would work.
<Baughn>
...very inelegant indeed.
<JasonGrossman>
Yes.
<Baughn>
Point is, there's something that calls this script.
<Baughn>
That "something" can itself be run via steam-run.
<tilpner>
vaninwagen - If you want to replicate /bin/sh behaviour, start in nixos/modules/config/shells-environment.nix+175
tzemanovic has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 merged pull request #42211 → Grub: default is signed int. Fixes #42152 → https://git.io/ff8vR
<infinisil>
vaninwagen: The User setting there isn't needed (default is root), and the enable isn't needed too actually (default is true)
<vaninwagen>
infinisil: but why a timer for it? running it and updating /bin/bash on rebuild to ensure i have the most recent bash linked it just fine, no?
<tilpner>
infinisil - I think you got the wrong person
<infinisil>
Oh
<tilpner>
An activation script is the best (yet still ugly) way to add that symlink
<vaninwagen>
:)
<infinisil>
Sorry, I meant sir_guy_carleton :P
<vaninwagen>
np
<infinisil>
(I just woke up, I just saw "using activationScripts" and panicked)
<infinisil>
vaninwagen: Such a symlink is a smell already though
<infinisil>
vaninwagen: On nixos you want to link to a specific bash version by using ${pkgs.bash}
<infinisil>
But a single string should work too: nix.optimise.dates = [ "10,22:00:00" ]; (if i remember the combined syntax correctly)
<joko>
Hello, could anyone explain to me how using a private SSH key works on fetchgitPrivate and Hydra? I have noticed this: https://github.com/NixOS/nixpkgs/issues/4004 , but it unclear how the key should now be used
<sir_guy_carleton>
sir_guy_carleton: oh, the comma there seems to work
<sir_guy_carleton>
i was trying to put the whole time after the comma
johann__ has joined #nixos
<LnL>
joko: I would discourage using fetchGitPrivate, it's a pain to setup
init_6 has joined #nixos
<LnL>
joko: hydra inputs and builtins.fetchGit both support using rsa keys
<joko>
LnL: I know, I remember doing something similar in the past, but I am at work and I don't have my notes. If I recall correctly, I had to add an expression to Nix path, so that the builders get the key. But even then, there were issues with the file ownership and permissions and I don't remember anymore how to deal with them. Set a specific number of builders? I don't remember...
<joko>
LnL: Really? How so? I want to build it in Hydra
<LnL>
this part is handled by the hydra-evaluator IIRC so it's the hydra user
<joko>
LnL, ah, I see, the hydra user already has the key
<tilpner>
Filesystem Hierarchy Standard. It's the layout most applications expect and not adhereing to it is what causes so many problems for NixOS users trying to run foreign applications
<tilpner>
In this case, steam-chrootenv uses buildFHSUserEnv, and you can add packages to steams usage of that
andymandias has quit [Ping timeout: 240 seconds]
<mupf>
Alright, thanks. I will have a look. Is there some further documentation about the topic somewhere?
<kreetx>
another question: is there a difference when adding remote builders when being in single- or multi-user mode? I'm wondering if on macos I should set the NIX_REMOTE_SYSTEMS env variable for the user, or I should do it as root somewhere
semihonest has joined #nixos
johann__ has quit [Quit: Leaving.]
fendor has joined #nixos
raynold has quit [Quit: Connection closed for inactivity]
xy2_ has quit [Ping timeout: 264 seconds]
winem_ has joined #nixos
johann__ has joined #nixos
ericsagn1 has joined #nixos
steell has quit [Ping timeout: 264 seconds]
<Baughn>
gchristensen: Sounds like the consensus is "Wait for the upstream solution", if one can call this consensus. I guess I can hack up a wait-for-all-devices script with zdb.
<Baughn>
...probably a good idea anyway, we need something fast.
<Turion>
vim noob here. How do I activate the vim haskell plugin I installed via nix-env -iA nixos.vimPlugins.vim-hdevtools?
jwynn6 has quit [Ping timeout: 245 seconds]
xy2_ has joined #nixos
<LnL>
you need to do that declarative
igo95862 has joined #nixos
<LnL>
mutable operations like nix-env -i don't influence other packages
<{^_^}>
[nixpkgs] @mnacamura opened pull request #42232 → lf: available on the other UNIX-like platforms → https://git.io/fImwd
<{^_^}>
[nixpkgs] @mnacamura opened pull request #42236 → esh: init at 0.1.1 → https://git.io/fIuZn
<Turion>
LnL: Ah I see :) makes sort of sense
sorixelle has quit [Ping timeout: 255 seconds]
<Turion>
LnL: It doesn't say in the manual where to add these lines to... am I supposed to do that system-wide in /etc/nixos/configuration.nix? And how to deal with vim packages that should only load when opening specific file endings?
<Baughn>
Turion: It's not like it *couldn't* auto-discover things in $HOME/.nix, but that wouldn't be very Nixy. The behaviour of any one package isn't supposed to depend on what other packages may or may not be installed -- it would be spooky action at a distance, and also stop you from having multiple vim installs with different settings.
devx is now known as DevX
alex``` has joined #nixos
<Turion>
Yes that makes sense
<Turion>
I'm just wondering how and where to install vim now
<Turion>
I've got it installed system-wide declaratively
<Baughn>
System-wide is probably fine.
<Baughn>
If it's a single-user system. I tend to put everything in configuration.nix too.
<Turion>
Harrharr, there is a section in the manual "Running Emacs as a service" :D I wonder whether it's possible to run atom as a service as well ;)
alex```` has joined #nixos
alex```` has quit [Client Quit]
<etu>
Turion: Nah, Atom is a graphical program. Emacs can run as a non-grahpical deamon and then you can connect clients to it :)
smallville7123 has quit [Quit: rosa]
<LnL>
Turion: you can do this in multiple places
<Turion>
etu: Right, people did spend some thought when designing emacs ;)
jperras has joined #nixos
alex``` has quit [Ping timeout: 256 seconds]
<Turion>
LnL: Good :) is there a place where I can add a vim with my plugins without having to understand home-manager now (and without having to shoot myself in the knee when I want to learn home-manager later)?
<etu>
Turion: Yes, emacs is not built on google chrome :p
<LnL>
Turion: it's a common pattern, there's something similar to make a python interpreter that's aware of a set of pypi packages
npmccallum has joined #nixos
agander has joined #nixos
<LnL>
we do this to avoid global dependencies, otherwise installing a tool wouldn't work if you didn't also install one of it's dependencies
<LnL>
or a user might upgrade one of the dependencies to an incompatible version at some point
<Turion>
Ok, maybe my beginner's question is: How do I install stuff declaratively on a per-user basis?
<LnL>
etc.
<LnL>
you'll need some place for the expression so an overlay is probably the easiest
<Turion>
LnL: Thanks! It seems I don't really know how to use that... I get "error: undefined variable 'vim_configurable' at /home/turion/.config/nixpkgs/overlays/vim-overlay.nix:11:11"
<Turion>
Err error: 'customize' at /nix/store/l9p58f99rbll6grpy7ay8l7jyg4w8r09-nixos-18.09pre142796.4b649a99d84/nixos/pkgs/misc/vim-plugins/vim-utils.nix:328:17 called without required argument 'name', at /home/turion/.config/nixpkgs/overlays/vim-overlay.nix:11:11
<LnL>
hrm, guess the example in the manual isn't complete
<LnL>
should probably just inherit the name by default, updated
fendor has joined #nixos
Neo-- has quit [Ping timeout: 240 seconds]
<Turion>
infinisil: Yeah, I guess I'm just at the hard stage of going from beginner to advanced nix user. When I read the installation section of home-manager, I'm intimidated by the many things I don't completely understand.
<infinisil>
The installation is a bit weird admittedly heh
<Baughn>
Turion: Personally, I much prefer keeping everything in a single configuration (i.e. the system-wide one). :)
<Baughn>
Rebuilding the system only takes twenty-thirty seconds anyway, and that's with the extremely over the top complexity of my own configuration.
<Baughn>
Though I already use a build script instead of calling nixos-rebuild directly, so it's not like throwing in a `home-manager switch` in that would've hurt me.
<Turion>
Baughn: With my low-memory machine, I have to switch off browser and any nontrivial applications before upgrading, and it takes 15 minutes or so :/
<Baughn>
Not the best case.
<Baughn>
Do you have a more powerful machine you might run the build on?
<Baughn>
nixos-rebuild has a --target-host option.
wotan_gs has quit [Remote host closed the connection]
<Baughn>
(If you use it, you need to also set --build-host localhost.)
<gchristensen>
with low memory you may need to be instantiating on a different host, though
<gchristensen>
which nixos-rebuild doesn't have a flag for :)
<Baughn>
Well, the idea I had was to run nixos-rebuild on a different machine.
<infinisil>
Baughn: Turion: home-manager supports building from your configuration.nix
<Turion>
Baughn, gchristensen, that's a cool idea, but I wouldn't know what machine I could legally use for that :/
<Turion>
infinisil: What does that mean?
<gchristensen>
legally? interesting
<infinisil>
So a nixos-rebuild will make the home-manager thing as well
<infinisil>
Then you can assign the `home-manager.users.infinisil = { home.packages = with pkgs; [ hello ]; };` in your configuration.nix
<Baughn>
Neato. I might use that.
<manveru>
might be worth switching... that'd make my administration even easier :)
ryantrinkle has joined #nixos
<Turion>
(gchristensen: I think I can't use university servers for that kind of thing. Haven't thought of any other machines I could use)
dbmikus has joined #nixos
johanot has quit [Quit: Leaving.]
<Turion>
Ok, zeroth home-manager question: Why not nix-env -iA nixos.home-manager?
<Turion>
Or what would I achieve by that?
<manveru>
because home-manager is more or less independent of nixpkgs
<manveru>
it has its own release cycle
<infinisil>
The nixpkgs version needs a PR for every module change, which would be rather annoying..
fendor has quit [Ping timeout: 255 seconds]
<infinisil>
would need*
<manveru>
i think it could be added for 18.09... not many changes recently
stepcut has joined #nixos
chrisaw has joined #nixos
<chrisaw>
ls
<chrisaw>
oops
<infinisil>
manveru: True
jensens has joined #nixos
<manveru>
i just wish it was possible to simply reference the home-manager repos default.nix from nixpkgs :P
<chrisaw>
Quick question - how do folks usually setup their public and private keys on NixOS? I know it can be done manually but if I want to maintain both between re-installs for example?
<gchristensen>
do it manually
<gchristensen>
imo :)
endformationage has joined #nixos
<infinisil>
Hmm.. I'm doing it manually too
<{^_^}>
[nixpkgs] @fpletz pushed 2 commits to release-18.03: https://git.io/fLrTT
<infinisil>
Putting the private keys into the nix store doesn't sound right, so I think optimally it would be a dynamic thing with some key server thing, is there software for this?
asuryawanshi has quit [Ping timeout: 264 seconds]
andymandias has quit [Ping timeout: 265 seconds]
lassulus_ has joined #nixos
lassulus has quit [Ping timeout: 260 seconds]
lassulus_ is now known as lassulus
<manveru>
might be fun adding per-user encrypted volumes to nix...
<Turion>
manveru, infinisil: But couldn't I install the initial home-manager like that and then let it manage itself?
<infinisil>
not sure
<hyper_ch2>
hmmmmm, does the initrd also have already firewall activated?
<Turion>
Ok, I'll do it by the book then ;) but another day. Thanks to everyone for the help!
<Baughn>
Turion: You can, indeed, install home-manager with nix-env and have it manage itself.
<Baughn>
But then you have one step which isn't declarative.
johann__ has joined #nixos
<Turion>
Baughn, yes, but as soon as I've installed it and let it manage itself, I can uninstall it again, right?
<yeme>
hello everyone, how to I get my nixos installation to use a local checkout of nixpkgs?
<yeme>
i imagine somewhere in the configuration.nix the nixpkgs folder is on the same machine
<yeme>
configuration.nix, *
<clever>
yeme: you have to use -I nixpkgs=/path/to/nixpkgs
<Baughn>
yeme: configuration.nix can't[1] do it, because it's the nixpkgs code that reads configuration.nix.
<yeme>
I would like to avoid that everytime I issue a nix-command, especially nixos-rebuild. is there a better wai?
<Baughn>
[1]: It can, but that would imply running nixos-rebuild twice.
<Baughn>
Yes, write yourself a shell script.
<clever>
yeme: you can also use nix.nixPath in configuration.nix to change the default NIX_PATH, but you still need to -I nixpkgs=/path/to/nixpkgs the first time
<Baughn>
Mm. And keep in mind that if you ever change that path, you need to rebuild before it'll use the new one.
<yeme>
so, I edit configuration.nix, add nix.nixPath = /local/nixpkgs/checkout, then run nixos-rebuild switch -I nixpkgs=/path/to/nixpkgs. From then on the local nixpkgs will be used for all nix-commands?
<Baughn>
Hmm. Three of my last ten commits to that repository are--
<Baughn>
"Misc, stuff."
<Baughn>
"More stuff."
<Baughn>
"even more stuff."
<Baughn>
yeme: Yes, unless you use -I nixpkgs. Note that the -I nixpkgs in your command, above, has no bearing on how the configuration setting is interpreted, and may be pointless.
<Baughn>
(Unless of course you edit nixpkgs to change the meaning of nix.nixPath, heh.)
<yeme>
got it. thank you Baughn and clever
<Baughn>
yeme: Oh, and hopefully you know this, but you *don't* want to base your system off a raw checkout of the nixpkgs repository.
<Baughn>
Especially not master.
<Baughn>
That would mean you get untested code for which there is no binary cache.
<Baughn>
Given we seem to be doing much the same thing, you might be able to copy that wholesale.
<Baughn>
(Look in particular at line 11.)
<yeme>
Baughn: I'll check out the script. I forked nixpkgs/release-18.03 and did small changes to accommodate my needs. I use that fork to build some software, and I thought it was a good idea to also tell nixos/nix to use that fork for everything nix-related. Is there soemthing wrong with that?
chrisaw has quit [Quit: WeeChat 2.1]
<tilpner>
Baughn - It's possible to have configuration.nix use a custom nixpkgs with a single call to nixos-rebuild, by replacing <nixpkgs/nixos> with a two-stage evaluation
<tilpner>
(Which doesn't require forking/modifying <nixpkgs>, as your wrapping <nixpkgs/nixos> can call the real one)
jperras has quit [Ping timeout: 240 seconds]
orivej has quit [Read error: Connection reset by peer]
orivej has joined #nixos
jperras has joined #nixos
<Baughn>
yeme: Nah, that's sensible. It's the exact same thing I do.
<Baughn>
Though if you're only changing packages, then an overlay would be an option.
<Baughn>
tilpner: That seems like it'd work, but I never figured out how to make nix-channel point at a local directory.
<tilpner>
Oh, I don't use channels
<tilpner>
(Well, nix-channel)
<Baughn>
Then you're overriding the default behaviour anyway.
<Baughn>
tilpner: What sort of changes are these? Anything you could PR?
<init_6>
it's possible with nixos-generate-config make a some more output with all default configs/variables?
<Baughn>
Er, yeme
<Baughn>
init_6: No, but check 'man configuration.nix' for defaults.
<Baughn>
Basically there are too many.
halfbit has joined #nixos
logiq has joined #nixos
<Baughn>
init_6: I suppose we could write the manpage into it, but once you take a look, you'll realize why we don't. :)
<init_6>
Baughn: now i switsh from gentoo to nixos and more verbose verbose output from nixos-generate-config would be simply divine
camsbury has quit [Remote host closed the connection]
camsbury has joined #nixos
fendor has joined #nixos
<rycwo>
Thanks gchristensen, any idea why my posts were flagged?
<gchristensen>
nope :) looking in to that
camsbury has quit [Ping timeout: 240 seconds]
erasmas has joined #nixos
yyadavalli has joined #nixos
<yyadavalli>
Hi, I am simple question. I am trying to install an updated version of a package (scalafmt) by overriding the existing derivation. Is this possible, how should I go about doing it?
<yyadavalli>
Here is the derivation in question, I started `nix repl`, loaded nixpkgs and tried `scalafmt = scalafmt.override { version = "1.6.0-RC3"; }`
<{^_^}>
[nixpkgs] @Moredread opened pull request #42254 → Platformio: 3.5.2 -> 3.5.3 and fix for click dependency → https://git.io/fqFYt
Lev50 has quit [Quit: This computer has gone to sleep]
tzemanovic has quit [Remote host closed the connection]
tenten8401 has quit [Ping timeout: 260 seconds]
fresheyeball has quit [Quit: WeeChat 2.0]
woodson has joined #nixos
orivej has joined #nixos
emmanuelrosa has joined #nixos
stepcut has joined #nixos
mmercier has quit [Quit: mmercier]
<emmanuelrosa>
I'm trying to work around this (https://github.com/NixOS/nixpkgs/issues/38808) issue to continue deploying a machine via Nixops. To work around the out-of-memory issue I tried using nix-store export/import using a list of paths I know are missing at the remote machine, however this results in "cannot add path 'foo' because it lacks a valid signature: nix-store --export $(cat new.paths.txt) | bzip2 | nixops ssh -d deployment machine -- bzc
<emmanuelrosa>
at | nix-store --import --option require-sigs false
elasticdog has quit [Ping timeout: 276 seconds]
<gchristensen>
the --bzcat | nix-store ... the pipe there is causing a new nix-store process on your _host_ not on the target
<gchristensen>
you probably need -- /bin/sh -c '"bzcat | nix-store --import --option require-sigs false"'
btbytes has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
btbytes has joined #nixos
btbytes has quit [Client Quit]
btbytes has joined #nixos
jperras has joined #nixos
jperras has quit [Ping timeout: 276 seconds]
camsbury has joined #nixos
jperras has joined #nixos
<emmanuelrosa>
Aaaah, thank you gchristensen. It produced an out-of-memory error; So much for that work-around. I'm now trying another way: bind-mounting the remote nix store and extracting a tar archive of the paths.
grp has joined #nixos
<{^_^}>
[nixpkgs] @dezgeg opened pull request #42255 → WIP: Log stage-{1,2} output to secondary consoles → https://git.io/fYIQe
<inquisitiv3>
Why does the symbolic link `~/.nix-profile` lead to the path `/nix/var/nix/profiles/per-user/user/profile` that doesn't exist?
<gchristensen>
your profile has nothing in it, so it doesn't exist (this is normal)
<inquisitiv3>
gchristensen: Is this because I (probably) haven't installed packages locally to my own user?
<gchristensen>
yeah
<inquisitiv3>
Okey, thanks!
Cale has quit [Ping timeout: 276 seconds]
<gchristensen>
you're welcome!
<inquisitiv3>
gchristensen: Reading through Nix pills now. Seems that I'll have to learn which things is most relevant for those only using Nix.
<hodapp>
hmm, I should re-read Nix pills
<gchristensen>
it is loaded with good info, nix-env is the only thing I don't regularly use
<gchristensen>
though my work has caused me to use it a lot lately (dealing with bootstraping Nix in new environments)
marc__ has quit [Ping timeout: 248 seconds]
<inquisitiv3>
If I'm not mistaken the link `/nix/var/nix/profiles/system` links to the current generation?
<inquisitiv3>
And if I do a rollback the link only changes from pointing from generation n to generation n - 1 if I don't specify a specific generation?
camsbury has quit [Remote host closed the connection]
<philandstuff>
Hello! I'm trying to fix pkgs.haskellPackages.dhall-json_1_2_0 (which seems to have a broken dependency on dhall). I've got a local fix. I just wanted to confirm that it's broken for everyone else before issuing a PR. I can't find the job in hydra - can someone help me?
<philandstuff>
i can find dhall-json, just not dhall-json_1_2_0
<philandstuff>
thanks LnL... reading the comment at the top of the file, I'm not sure if it should be in there or in configuration-common.nix
agander_ has joined #nixos
Neo-- has quit [Ping timeout: 248 seconds]
vcunat has joined #nixos
<LnL>
yeah one of those, not sure which is more appropriate both will work
agander has quit [Ping timeout: 240 seconds]
<tilpner>
fresheyeball - As I'm sure you've figured out by now: Yes, either that or with callPackage. Note though that expressions like that are not allowed in nixpkgs
<fresheyeball>
tilpner: actually I am still trying to figure out what fields are allowed on the set for fetchFromGitLab
<tilpner>
fresheyeball - You can look for example in nixpkgs, or check the source in all-packages.nix
<philippD>
Ok, maybe I am trying to do the wrong thing. I want to use a boolean variable in a buildPhase shell script as the condition of an if, without turning the variable into a string variable. Using `toString` doesn't seem to work there
<fresheyeball>
found it
<fresheyeball>
thanks tippenein
<fresheyeball>
thanks tilpner
troydm has joined #nixos
<philippD>
> toString false
<{^_^}>
""
emmanuelrosa has left #nixos [#nixos]
<philippD>
Is there a built in function that turns a boolean into either the string "true" or "false"?
<LnL>
builtins.toJSON
logzet has quit [Read error: Connection reset by peer]
<philippD>
thanks LnL that's excatly what I was looking for.
<infinisil>
Ah, so it only tests whether the channel could be created successfully
btbytes has joined #nixos
pip3000 has quit [Ping timeout: 260 seconds]
agander_ has quit [Ping timeout: 256 seconds]
<infinisil>
Even though an actual channel update only occurs when all those jobs succeed
rihards has quit [Quit: rihards]
<gchristensen>
yeah `tested` is also a specific derivation, which is an "aggregate derivation" (special to Hydra)which dependss upon all the constituent jobs
<samueldr>
hmm, I'm thinking you could maybe work with ./., infinisil, and convert that
<samueldr>
if you keep it next to your configuration
oida has joined #nixos
Thra11 has quit [Quit: IRC for Sailfish 0.9]
<infinisil>
Hmm true..
Thra11 has joined #nixos
<samueldr>
though, building from another machine may not work
<samueldr>
(as in nixos-rebuild --target-host)
<samueldr>
path will not exist there, so nix-shell/nix-env may have issue, would they?
<infinisil>
nixpkgs gets evaluated on the local machine
<samueldr>
for --target-host, local is localhost or target-host?
simukis has quit [Ping timeout: 256 seconds]
<samueldr>
I would think localhost
<samueldr>
(I know for --build-host, it's localhost)
<infinisil>
nix always evaluates on the local machine, only builds and deployments can be on others, is that what you mean?
<infinisil>
localhost local
<samueldr>
`nixos-rebuild switch --target-host $other-machine`, "system activation will happen on the remote host instead of the local machine."
<samueldr>
"If --build-host is not explicitly specified, --build-host will implicitly be set to the same value as --target-host."
<samueldr>
I see
chessai has quit [Remote host closed the connection]
<infinisil>
Ah and yes, the nixpkgs=/local/nixpkgs won't work then
<infinisil>
For the deployed machine
<samueldr>
ah right
<infinisil>
I do use nixops for all machines though
Fare has joined #nixos
<lejonet>
Anyone that have gotten the wifi module working in nixos on a pine a64 lts? I can't get the interface to appear despite the module (r8723bs) loading fine and all...
<infinisil>
I should make it so that it clones my system config if it's not existent on the machines I want to deploy from
<infinisil>
And the ones I don't want to deploy from, it should set `nixpkgs=${lib.cleanSource <nixpkgs>}`
sary has quit [Ping timeout: 260 seconds]
tmaekawa has joined #nixos
<samueldr>
lejonet: hadn't realized there was wifi on the A64-lts
<Dezgeg>
maybe r8723bs needs firmware?
<samueldr>
Dezgeg: I have it on a cherrytrail laptop working without specific handling
<samueldr>
(though I had to do things for bluetooth to work)
tmaekawa has quit [Client Quit]
<samueldr>
it does load a firmware (looking at dmesg)
<samueldr>
firmwareLinuxNonfree has that firmware (and rtl8723bs-firmware)
<meowmeow>
Im still trying to run tmux or nvim terminal from nix-shell, but i alwsays get problem of these programs running /nix/store/...-bash-4.4/bin/bash even when i have bashInteractive in scope of nixshell
<meowmeow>
so prompt still broken in bash instances
<meowmeow>
but running simple bash works just fine
<samueldr>
ah, I wouldn't be able to see the wifi interface, it's an external module
<samueldr>
I would assume lejonet does have the module
<samueldr>
Dezgeg: how likely would it be that the device tree needs to know about the wifi module?
wchresta has quit [Ping timeout: 260 seconds]
LFF^ has quit [Ping timeout: 245 seconds]
<Dezgeg>
quite high
<Dezgeg>
not 100% though
amir has quit [Ping timeout: 265 seconds]
tzemanovic has joined #nixos
tzemanovic has quit [Client Quit]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
andymandias has quit [Ping timeout: 240 seconds]
taktoa has quit [Ping timeout: 265 seconds]
grp has quit [Quit: box shutting down...]
<tnks>
I'm just trying to reverse engineer how some things work in Nix, and I encountered something unexpected.
<tnks>
If I do 'sudo LS_COLORS=sdf env' then it overwrites the LS_COLORS environment variable.
<tnks>
but that doesn't work for 'sudo NIX_PATH=sdf env'
amir has joined #nixos
<clever>
tnks: does /etc/sudoers mention either var?
<tnks>
clever: no, but it does have this: Defaults env_keep+=SSH_AUTH_SOCK
<tnks>
clever: yeah, so that's the same with me. It's what makes it give me a warning with a 'sudo nixos-rebuild' because it can't find the unexpanded "$HOME" directory.
<tnks>
so I just started using "-i", but that seems to not be what other people are doing, which made me feel like I didn't have someting configured right.
<clever>
i just `sudo -i` and then `nixos-rebuild switch`
<tnks>
clever: yeah, me too. Mostly I just like understanding how things work, because the minutea sometimes comes in handy in other contexts.
<clever>
i'm guessing its something to do with /etc/profile or /etc/bashrc
<tnks>
clever: I ran down that route. there's something special about NIX_PATH, though that doesn't allow overriding.
<clever>
tenten8401: this option allows you to do an efi install without access to the efivars
<tenten8401>
oooh that might work
<clever>
it uses the same filename as removable drives, so the bios is tricked into thinking its a bootable usb stick
<clever>
and those cant be in the efivars, because they are removable
<tenten8401>
and from there I can boot to the system and fix it manually
<tenten8401>
hopefully
<clever>
yeah
<tenten8401>
I don't know if the BIOS even supports UEFI but it's a fairly new processor so it should
<clever>
once it boots with efi once, you can unset that, and then nixos will poke around in the efivars
<tenten8401>
I say fairly new, it's like 2015 processor
<tenten8401>
thanks for the pointer, gonna try it out
__monty__ has quit [Quit: leaving]
<tenten8401>
do I still need an efiSysMountPoint if I'm installing as removable?
<tenten8401>
I'd assume so but
<clever>
yeah
<clever>
it still needs to know where to put the .efi binaries
<tenten8401>
installing... here goes
<tenten8401>
then I'll reboot the server and see if it wants to boot
tzemanovic has joined #nixos
deepfire` has joined #nixos
astrofog has joined #nixos
deepfire` has quit [Ping timeout: 260 seconds]
<rizary>
In `buildPhase`, how you specify custom variables? Like I want to run `$CABAL_DIR=$Src/config/ cabal new-build .` or similar to that. I try with export but fail.
<worldofpeace>
infinisil: "commit actually triggering the rebuild directly", what do you mean by that?
<clever>
rizary: same way you do it in bash, export VAR=value
<clever>
without the $ on VAR
<infinisil>
worldofpeace: nixos-rebuild in .git/hooks/post-commit
<infinisil>
Not gonna do it after all though, would be annoying with rebases
sbdchd has quit [Remote host closed the connection]
<tenten8401>
welp, install finished
<tenten8401>
here's to hoping it boots
deepfire` has joined #nixos
<tenten8401>
starting to wonder if the BIOS even supports UEFI
<tenten8401>
so either these hdds are *really* slow or the bios doesn't support UEFI
<tenten8401>
or I configured SSH wrong
<tenten8401>
pretty sure my machine is ded, gonna fire up rescue system and see if I can do legacy boot
lukelau has left #nixos [#nixos]
<tenten8401>
providing I have an EFI partition mounted on /boot, what would I have to do for legacy booting?
<tenten8401>
do I just delete the partition and expand everything to 1 partition, then install grub?
<tenten8401>
or do I even have to do anything?
ryantrinkle has quit [Ping timeout: 245 seconds]
<Dezgeg>
I guess the closer you do it how the existing distros on hetzner do it the more likely it will work
<Dezgeg>
some annoying biosen don't support legacy boot from GPT partitioned disks, for instance
<tenten8401>
yeah, I wasn't sure if the existing distros used EFI or not
<tenten8401>
I have 2 TB drives, it better support GPT
astrofog has quit [Quit: Quite]
Lev has quit [Quit: Leaving]
<tenten8401>
can you have a hybrid grub configuration?
<tenten8401>
as in install EFI and legacy
camsbury has quit [Remote host closed the connection]
camsbury has joined #nixos
<tenten8401>
going to try with EFI support and with /dev/sda1
<tenten8401>
and see if it'll work
asymmetric has joined #nixos
<tenten8401>
may try booting under qemu from the disk first
<tenten8401>
anyone know why I could be getting a 'This GPT partition label contains no BIOS boot partition" error?
<tenten8401>
./boot is mounted and exists
<MichaelRaskin>
You are trying to install GRUB2 for Legacy/BIOS boot on GPT
<MichaelRaskin>
Unlike MBR GPT doesn't have good empty spaces to stash the loader
<MichaelRaskin>
So a special partition — BIOS boot partition — not suitable for anything but directly dumping real bootloader there for the minimal stub to load — is required
init_6 has quit []
<tenten8401>
I have a /boot partition MichaelRaskin
<tenten8401>
am I going to need something else?
<tenten8401>
Do I need to set my boot device to /dev/sda1 instead of /dev/sda?
<MichaelRaskin>
Do you understand that «/boot partition» might mean two quite different things?
<tenten8401>
I have a partition that's mounted on /boot
<tenten8401>
formatted as FAT32
rprije has joined #nixos
<MichaelRaskin>
I have a feeling that you are doing a mix of UEFI and Legacy boot configuration
<MichaelRaskin>
Just add a new partition of type BIOS boot partition, that should be enough
woodson has joined #nixos
sbdchd has joined #nixos
<MichaelRaskin>
I think 16MiB is plenty of space for it
<tenten8401>
yeah, I'm aiming for legacy boot since apparently this bios doesn't support uefi
<MichaelRaskin>
You should not format it, mount it, or do anything with it
<MichaelRaskin>
In case of BIOS boot, why format /boot as FAT32, but oh well…
<tenten8401>
ahh alright
<tenten8401>
I've always had a UEFI capable system, I don't think I've ever had to boot legacy before
<tenten8401>
which is why I'm so clueless
<tenten8401>
alright, switched my EFI partition to BIOS boot type and wrote to disk
<MichaelRaskin>
Argh
<MichaelRaskin>
Wasn't your EFI partition /boot?
<MichaelRaskin>
Because /boot and BIOS boot partition are _different_ things
<MichaelRaskin>
And you need both
<tenten8401>
I just straight up unmounted it from /boot, going to see if it'll store /boot in the raid array
<tenten8401>
or do I need a seperate /boot?
<MichaelRaskin>
Well, what is your /
<tenten8401>
ext4 raid array
<tnks>
clever: I guess I found more variables that can't be overridden with sudo. These include: LD_LIBRARY_PATH, LOCALE_ARCHIVE, NIX_PATH, SANE_CONFIG_DIR, AND TZDIR
<tenten8401>
raid1
<MichaelRaskin>
old or new metadata?
<tenten8401>
not a clue
erasmas has quit [Quit: leaving]
<tenten8401>
probably new
<MichaelRaskin>
Then I would make a separate ext4 /boot with 0.90 metadata
<tenten8401>
think 240M is enough for /boot?
<MichaelRaskin>
Not sure
<tenten8401>
guess I'll find out the hard way
<MichaelRaskin>
NixOS might have large initramfs images
<MichaelRaskin>
And it has more of them than some other distributions
philippD has joined #nixos
<MichaelRaskin>
On NixOS /boot overfull is actually a thing that happens
<tenten8401>
that worries me a bit
<tenten8401>
does a raid array resize take that long?
taktoa has quit [Ping timeout: 264 seconds]
<MichaelRaskin>
Any change in stage1 settings means an additional initramfs…
meowmeow has quit [Quit: Konversation terminated!]
<philippD>
Is there an easy way of getting texlive packages more selectively? As soon as I want to do some fancier stuff I have to use texlive-full which is HUGE.
Fare has joined #nixos
<tokudan[m]>
tenten8401: if you have trouble installing on hetzner, have a look at nixops. that works pretty well to deploy nixos on a hetzner server for me
<MichaelRaskin>
So you could have multiple for the same kernel version
<MichaelRaskin>
(until a GC + rebuild, both are necessary)
<tenten8401>
problem isn't hetzner, it's the raid configuration
<tenten8401>
and the fact that I was trying UEFI instead of BIOS
deepfire` has quit [Ping timeout: 256 seconds]
<tenten8401>
I'll try 512MB for /boot
chasevasic has joined #nixos
<tenten8401>
alrght, sp O jad ,y
<hodapp>
I made LuxCoreRender build! \o/
<tenten8401>
oh man I butchered that
* hodapp
spikes pull requests onto the ground
<tokudan[m]>
Ttenten8401: ah, ok. anyway, http://paste.debian.net/hidden/07e1bb3e/ that creates new raid and two big filesystems i used as placeholders for a zfs zpool
<tenten8401>
So I have my 16MB BIOS boot partition, 512MB ext4 partition for /boot, and a 1.8 TB RAID partition. Does that sound decent?
<hodapp>
on one disk?
<tenten8401>
replicated on both disks
<tenten8401>
how would I do the "old metadata" on /boot?
<tokudan[m]>
if you've got 2 TB disks in there... does it really make a difference to you if /boot is 512 MB or 1024 MB?
<tenten8401>
nope
<tenten8401>
but it was 256 before
<hodapp>
hm. sometimes I just stick that stuff on USB or flash and use the whole disk(s) for RAID
<tenten8401>
yeah, but I don't have physical server access
<hodapp>
ohhhhh
<tokudan[m]>
tenten8401: you can specify the version when you create the raid, something like --version=0.9 or something similar
<tenten8401>
ahh yeah I see
<tenten8401>
--metadata=0.90
<tenten8401>
can I leave the new metadata on for the root partition?
<tokudan[m]>
yes, it's only important for /boot, as there the mdadm superblook needs to be at the end
<tokudan[m]>
(and possible for a bios boot partition)
<tenten8401>
alright
<tenten8401>
I take it the BIOS partition shouldn't be made a raid array?
<tokudan[m]>
root is mounted by the linux kernel that should be able to understand it
<tenten8401>
and is /boot alright being ext4?
<tokudan[m]>
you can make it a raid 1 in linux, though the bios will never see an array there
<tokudan[m]>
should be good
<tenten8401>
I'll give it a shot so I can have redundant boot drives
<tenten8401>
actually, would raid metadata be an issue on the bios partition?