<alasi>
How is NixOS's maturity as a daily driver OS for, say, a thinkpad? I'm considering moving from arch. Are there any features or bugs on the roadmap I should wait to resolve before investing time myself?
<palo>
looks like GrahamcOfBorg did not pick up my pull-request : https://git.io/vhDuA
<JasonGrossman>
alasi: the main thing to bear in mind about NixOS, IMO, is that it's very different from other linux distributions. It's not necessarily more work than Arch Linux, but it might be (depending on what you're doing), and when you need to fix things it feels more like you're starting from scratch. On the other hand, you get all the benefits - e.g. trivially easy rollbacks - straight away. A lot of people here moved from Arch and are
<JasonGrossman>
happy, but YMMV.
<elvishjerricco>
alasi: Is that issue unique to nixos? Seems any non-microsoft OS would have that problem. (I only skimmed the details)
<JasonGrossman>
I'm happy with it as a daily driver on a laptop.
<elvishjerricco>
alasi: Well it's good you found some Arch docs on it. In my experience, if there are Arch docs, it's not too hard to add it to NixOS.
<JasonGrossman>
I can imagine people moving from Ubuntu might not be happy. They're used to less involvement in the OS.
<alasi>
JasonGrossman: thanks for the perspective. I'm a student and casual (high level languages/tooling) dev
init_6 has quit [Ping timeout: 260 seconds]
<alasi>
The federation of packages on a monorepo makes me a little uncomfortable. Do security updates and such for packages get put in the registry at a comparable pace to arch?
asymmetric has quit [Ping timeout: 260 seconds]
<JasonGrossman>
alasi: Yes, they do. If you hang out in this room you'll see them being added in real time. :-)
<JasonGrossman>
It being a monolithic repo makes me uncomfortable too, for the long term, but the community's active enough that I'm execting it to decentralise before there are any major problems. That's only a bet, though, not a certainty.
<alasi>
how does stuff get bumped so quickly from so many different sources?
<alasi>
Haha nice
<JasonGrossman>
People are awesome, that's how!
<alasi>
Network effects are definitely good
<JasonGrossman>
Yes.
<elvishjerricco>
alasi: NixOS is largely volunteer driven, but IIRC ryantm recently invented some tooling for auto-updating things en masse.
<JasonGrossman>
I'm not sure what the long term will be. I would be talking about decentralisation a lot if I had any right to, but I'm new here and haven't contributed enough to be worthy of an opinion yet. :-)
<JasonGrossman>
My feeling is that people are relying on the very firm foundation (side-effect-free orthogonal package management, for example) to make decentralising relatively easy when the time comes.
<alasi>
Maybe I can't judge because I'm using a flavor of arch which I won't name and one of the maintainers' accounts just got compromised this afternoon
<alasi>
His passwords were in his dotfiles for a month
mahalel_ has joined #nixos
<alasi>
(public on GitHub)
<alasi>
So it'd be hypocritical to criticize haha
<JasonGrossman>
Sorry to hear that!
<alasi>
I'm definitely not running any updates anytime soon
<JasonGrossman>
Right.
tenten8401 has joined #nixos
<alasi>
Hmm
<tenten8401>
Is there a way to speed up package builds on NixOS unstable?
<tenten8401>
Pretty sure my system is trying to compile LibreOffice with 1 thread
<tenten8401>
and I have 12
adrianvovk_ has joined #nixos
<infinisil>
Libreoffice is failing to build in some recent unstable for some reason
<infinisil>
Which is the reason it's not in the cache
<tenten8401>
ahh alright
<tenten8401>
may just remove it for the time being
<tenten8401>
never use it
<infinisil>
It doesn't seem to be fixed still..
<infinisil>
There's isn't even an issue
<infinisil>
Maybe you could open one
adrianvovk has quit [Ping timeout: 256 seconds]
tenten8401_ has joined #nixos
andreabedini has joined #nixos
<tenten8401_>
graphical session died when I did the update
tenten8401 has quit [Ping timeout: 260 seconds]
<JasonGrossman>
alasi: I guess this (LibreOffice) answers part of your question too. Of course you can still use a previous version of LibreOffice.
<tenten8401_>
does unstable not have Plasma 5.13 yet?
ericsagn1 has joined #nixos
<alasi>
JasonGrossman: well, if it works on the stable channel :)
<tenten8401_>
I noticably did get a newer version of KDE but it doesn't seem to be 5.13 yet
<gchristensen>
Plasma is so big and our Plasma-maintaining team is so small that it can take a bit for those updates to be applied
<tenten8401_>
fair enough
<tenten8401_>
was just wondering if I updated wrong or something
<tenten8401_>
since I removed libreoffice and it updated instantly
<tenten8401_>
also, anyone know if there's a way to generate XDG directories on user creation?
<JasonGrossman>
alasi: Yes, the stable channel is good.
<JasonGrossman>
alasi: And the stable channel gets security updates.
<tenten8401_>
The Desktop, Documents, Pictures, Videos, etc. folders
<JasonGrossman>
alasi: And the claim that downgrading is much easier than on Arch Linux is absolutely true.
<gchristensen>
JasonGrossman: (so does unstable, but stable gets them faster)
<JasonGrossman>
alasi: (I mean rolling back, both packages and the whole OS.)
<JasonGrossman>
gchristensen: :-)
<alasi>
I have no doubts that rollbacks are difficult, that's the schtick haha
<JasonGrossman>
?
<alasi>
thats nixos's thing
<JasonGrossman>
Rollbacks are not difficult. That's the whole point (almost).
<alasi>
yeah sorry northeast tongue
<alasi>
I think ill take a look at my current packages and see what i'd be missing on nix
<JasonGrossman>
"no doubts that x" means x is false? Interesting! (Really interesting. Not being sarcastic.)
<alasi>
oh oopsies lmao
<alasi>
aren't* difficult
shabius_ has quit [Ping timeout: 256 seconds]
<alasi>
Does nixos have "legacy" support for, say, shell script installed bootleg packages that put stuff in /usr/bin? does that break any centralization of nix's special pkg directories?
asuryawanshi has joined #nixos
<gchristensen>
no it doesn't
<gchristensen>
have support for it
<gchristensen>
well ... it can make a FHS env just for a shell script, and it goes away when its done
<alasi>
hmm
jackdk has quit [Ping timeout: 264 seconds]
goibhniu1 has joined #nixos
<alasi>
and I know nix gives good docker-like "virtualization" for project-specific environments. Does nixOS make that even easier?
jackdk has joined #nixos
goibhniu has quit [Ping timeout: 264 seconds]
<alasi>
like nixpkg on mac vs nixpkg on nixos
<gchristensen>
sorry, I can't type much more tonight -- I should go.
<alasi>
no problem
<infinisil>
alasi: For this specific usecase, Nix already has those benefits
<tenten8401_>
figured I'd ask one more time to be sure as I didn't get an answer, is there a way to generate XDG user directories on user creation?
tenten8401_ is now known as tenten8401
Fare has quit [Ping timeout: 276 seconds]
<JasonGrossman>
tenten8401_: I don't think there's an easy way. You can always run a script.
<tenten8401>
ah fun
<tenten8401>
here's to hoping we'll get an /etc/skel implementation soon
<JasonGrossman>
Another way to use NixOS to manage files in the home directory is to symlink them to files in /etc, which can be managed by NixOS nicely. I can see many potential problems with that for some people, but it could work well for others.
Fare has joined #nixos
xcmw has joined #nixos
iyzsong has joined #nixos
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
oltoAltn has joined #nixos
olto has quit [Ping timeout: 240 seconds]
jtojnar has quit [Read error: Connection reset by peer]
JasonGrossman has quit [Ping timeout: 260 seconds]
<waynr>
i see that it's possible "to fetch store paths" by pulling via http using nix-serve
<waynr>
i guess i should have read further to learn about nix-copy-closure
<Yaniel>
you can also run a hydra server
<elvishjerricco>
waynr: That section could use some updates. `nix copy` is what I'd recommend these days
<gchristensen>
the caveat being `nix copy` isn't a stable interfrace right now
camsbury has quit [Ping timeout: 265 seconds]
camsbury has joined #nixos
b has quit [Ping timeout: 248 seconds]
<waynr>
i'll look into that. i really like the ability to build toolchain, dependencies, and package from source but some of my machines would take all day
Fare has quit [Ping timeout: 248 seconds]
<waynr>
i also like how easy it is to use :)
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<teto1>
alasi: as usual nix(os) lets you specify basic settings (for networkmanager) but for anything advanced, you still have to fiddle with the native config system
<teto1>
alasi: for the fingerprint-gui you would have to package it which can be very easy or hard depending on the software :)
<alasi>
@teto1: thanks. So there's always fallbacks
<teto1>
alasi: yeah, there is no way nix can cover all the settings of all software
<alasi>
This is a general disk/Linux question but if I have an encrypted partition eating up almost all of my ssd (but mostly unused filesystem) can I easily shrink it, partition Nixos, and if I like it, make it the primary OS that eats the whole ssd
earldouglas has quit [Quit: leaving]
<ghuntley>
Yeah it's wierd that you need Windows for the stream that's life tho. It's a company broadcast that I opened up to the public. If you missed it I'll be uploading a video later.
<JasonGrossman>
ghuntley: Thanks.
<JasonGrossman>
alasi: I don't know, but the NixOS way (and I think this is beautiful) is to play with it and then, if you want it on a different partition, install it on that partition from scratch, which is REALLY easy. You just install a base system and then copy the configuration over. (Needs good network access, though.)
<alasi>
JasonGrossman: do you mean play with it on a bootable device?
earldouglas has joined #nixos
earldouglas has quit [Client Quit]
earldouglas has joined #nixos
lassulus_ has joined #nixos
woodson has quit [Ping timeout: 240 seconds]
nlytend has joined #nixos
lassulus has quit [Ping timeout: 268 seconds]
lassulus_ is now known as lassulus
<nlytend>
Hi
<nlytend>
Is there any ebook reader which supports annotations?
jackdk has quit [Ping timeout: 260 seconds]
stepcut has joined #nixos
schoppenhauer has quit [Ping timeout: 268 seconds]
schoppenhauer has joined #nixos
pie_ has quit [Ping timeout: 260 seconds]
adrianvovk_ has quit [Read error: Connection reset by peer]
adrianvovk has joined #nixos
smallville7123 has joined #nixos
jackdk has joined #nixos
adrianvovk_ has joined #nixos
adrianvovk_ has quit [Read error: Connection reset by peer]
adrianvovk_ has joined #nixos
JasonGrossman has quit [Remote host closed the connection]
JasonGrossman has joined #nixos
<alasi>
are there any utilities that show the names / versions of your installed packages rather than ls'ing store? I can't see store being very friendly with all of the hashes
<alasi>
nix packages have so much browsable metadata
adrianvovk has quit [Ping timeout: 268 seconds]
<JasonGrossman>
alasi: A workaround is to install things declaratively and then just look at your configuration file to see what you've got installed.
<JasonGrossman>
alasi: Of course that won't list dependancies ... but the full list of dependencies gets overwhelming very quickly anyway (bearing in mind that you'll have multiple versions of lots of things).
<JasonGrossman>
alasi: Or `nix -q` will list what you've installed interactively (or `sudo nix -q` for things you've installed interactively as root, I guess - I haven't tried that).
<alasi>
@JasonGrossman is there a flag to install things through the shell and have it automatically register in the config? E,g npm has npm install --save mypkg which saves mypkg in the config
<JasonGrossman>
alasi: For versions, though, there is no easy way.
<alasi>
@JasonGrossman `nix -q` is probably what im looking for
<JasonGrossman>
alasi: No. Probably deliberately - you should hold your config deep in your soul.
<JasonGrossman>
alasi: You're right to think that the Nix Way is to have as much as you can in your config file though ...
<JasonGrossman>
alasi: Personally, what I do is install things with nix-env -i when I'm feeling lazy, but then move them (by hand) to my config. Alternatively, it's not hard to have everything in your config to start with.
<alasi>
assuming you meant `nix-store -q`, that seems like it's just ls'ing the store and not showing any names
camsbury has quit [Remote host closed the connection]
<JasonGrossman>
Sorry - I meant `nix-env -q`
<JasonGrossman>
And apparently that does list versions! I hadn't noticed.
camsbury has joined #nixos
<alasi>
ah perfect
<JasonGrossman>
With configuration.nix, the versions aren't listed but instead depend on the current state of the channel you're using.
<JasonGrossman>
Which is more complicated but is supposed to have advantages, especially when updating.
rardiol1 has quit [Remote host closed the connection]
<alasi>
I think that's fine
<JasonGrossman>
:-)
<alasi>
unless you arent allowed to explicitly declare what version you want when necessary..?
smallville7123_ has joined #nixos
<JasonGrossman>
Oh you are. People often do. It's just more complicated than with nix-env.
<alasi>
hmm ok
smallville7123 has quit [Read error: Connection reset by peer]
<JasonGrossman>
Because nixos-rebuild (for example) will take the most recent thing from whatever channel you're on, by default ...
<JasonGrossman>
... which is deliberate, of course, to make sure you get security upgrades etc. But it's less explicit than what you do with nix-env.
<alasi>
And if I make a new programming project in some random folder, i'd make a separate configuration.nix file
<JasonGrossman>
You can have whatever you want, but if you want old versions declared declaratively then you need to think about channels. There's a mechanism called "pinning" that can get you things out of old versions of the system.
<alasi>
or something
<alasi>
Hmm
<alasi>
I shold probably read stuff
<JasonGrossman>
alasi: I don't understand what you mean by a new project. If you're writing your own derivations then you need separate configuration, of course. Otherwise probably not.
<JasonGrossman>
alasi: All I have is /etc/nixos/configuration.nix (and the hardware configuration file, but that never changes).
<alasi>
If I wanted an isolated docker-like or virtualenv-like environment. Say a specific project needed python 2.7, while another project needed python 3
camsbury has quit [Ping timeout: 256 seconds]
<alasi>
I think that's what people use nix (the package manager) for on other OS's
woodson has joined #nixos
<JasonGrossman>
If you want them really ISOLATED then you need to write your own derivations, and then each derivation will need a separate configuration, right. Or you can use separate nix shells ... I'm not sure whether they count as derivations. I never do this myself, but I seem to be the only person chatting at the moment!
<JasonGrossman>
If you merely want to USE different versions for different projects then you can e.g. call python-2.7 or python-whatever without doing any prep at all except installing those versions.
<JasonGrossman>
Right. I'm the wrong person to answer detailed questions about that example, but yes that gets you isolation at the expense of extra config.
<alasi>
thanks so much for the help
<JasonGrossman>
nwvm!
<alasi>
I'm closer to sold on nix :)
<JasonGrossman>
Yay, I guess!
<JasonGrossman>
I know I like it but I don't think it'll be for everybody until it's more mature.
<JasonGrossman>
My musing about this is that I think it's likely that some big company (maybe Canonical?) will adopt it one day.
smallville7123_ has quit [Quit: rosa]
<JasonGrossman>
Or fork it.
<alasi>
I have installed arch... but yeah
<alasi>
without corporate backing itll be a tough uphill
smallville7123_ has joined #nixos
<JasonGrossman>
Personally I love NixOS + Arch Wiki. I'm glad Arch exists, but I don't want to use it myself any more.
<JasonGrossman>
Actually I think it only needs the amount of work it's currently getting. It's getting more mature all the time. But corporate backing might help to make sure the work doesn't tail off ... or it could last forever with volunteer work, like Debian. Maybe.
lukelau has joined #nixos
aarvar has quit [Ping timeout: 240 seconds]
igo95862 has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
<rotaerk>
I just don't like arch's model for system updates
adrianvovk_ has quit [Remote host closed the connection]
robstr has joined #nixos
adrianvovk has joined #nixos
irisarce has joined #nixos
irisarce has left #nixos [#nixos]
Guanin has quit [Ping timeout: 240 seconds]
rprije has quit [Ping timeout: 265 seconds]
adrianvovk has quit [Remote host closed the connection]
adrianvovk has joined #nixos
Guanin has joined #nixos
xy2_ has quit [Ping timeout: 245 seconds]
Fare has quit [Ping timeout: 245 seconds]
<sorbet>
Any tips on searching for function definitions in `nixpkgs`? I find myself wanting to know more about stuff like `fetchPypi`, but searching for it brings too many results in packages.
<sorbet>
Or even what directories to include/exclude when searching for stuff like that would be helpful.
<elvishjerricco>
sorbet: Generally, I just follow the attr path. e.g. for `(import <nixpkgs> {}).foo.bar`, I look in `all-packages.nix` for `foo`, then look in its definition to find `bar`.
<elvishjerricco>
Sometimes you can just do `nix edit nixpkgs.foo`, for a lot of derivations (though not functions)
endformationage has quit [Quit: WeeChat 1.9.1]
smallville7123_ has quit [Quit: rosa]
JasonGrossman has quit [Ping timeout: 240 seconds]
<sorbet>
elvishjerricco: Thanks, that helped. Just took me a bit to find `python-packages.nix` :P
smallville7123_ has joined #nixos
andymandias has quit [Ping timeout: 260 seconds]
Lisanna has quit [Quit: Lisanna]
andymandias has joined #nixos
sary has joined #nixos
Ariakenom has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hotfuzz has joined #nixos
jasongro` has joined #nixos
hotfuzz_ has quit [Ping timeout: 264 seconds]
justan0theruser has quit [Quit: WeeChat 2.0]
justanotheruser has joined #nixos
joehh has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 268 seconds]
Mateon3 is now known as Mateon1
sorixelle has joined #nixos
<ekleog>
grep obadz
<ekleog>
oops
* ekleog
missing a /
smallville7123_ has quit [Quit: rosa]
tzemanovic has quit []
Ariakenom has quit [Quit: Leaving]
michiell2 has joined #nixos
asuryawanshi has joined #nixos
hyper_ch2 has joined #nixos
hamishmack has joined #nixos
jackdk has quit [Ping timeout: 264 seconds]
kreetx has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
asuryawanshi has quit [Client Quit]
<{^_^}>
[nix] @edolstra merged pull request #2234 → libstore/gc.cc: ignore ESRCH when reading symlinks in /proc → https://git.io/vh194
<fuzzy_id>
is there some easy way to define an additional(!) pre-start step for a service
asymmetric has joined #nixos
<fuzzy_id>
i want to wipe out all configuration-related files before jenkins.service starts up
smallville7123 has quit [Quit: rosa]
smallville7123 has joined #nixos
<fuzzy_id>
but jenkins.service already has a pre-start script defined…
asuryawanshi has quit [Remote host closed the connection]
<fuzzy_id>
i'm currently trying to write a service which does the job and let jenkins.service depend on that job
<fuzzy_id>
wondered if there was an easier way
asuryawanshi has joined #nixos
<sphalerite>
fuzzy_id: if you make sure it has the same priority level as the existing definition I think they should be concatenated
thc202 has joined #nixos
<sphalerite>
if that's correct, you can also use lib.mkBefore/lib.mkAfter to control the order
<fuzzy_id>
is that documented in nix?
vaninwagen has joined #nixos
smallville7123 has quit [Client Quit]
<woffs>
fuzzy_id, or just set systemd.services.yourservice.postStart = "yourscript"
<woffs>
preStart
Zoom has joined #nixos
hyper_ch2 has quit [Quit: Page closed]
<Plato[m]>
Basic question: is nix 2.0 supposed to completely replace all other nix-* commands?
hyper_ch2 has joined #nixos
atu has joined #nixos
<adisbladis[m]>
Plato[m]: Not replace, supplement
<adisbladis[m]>
And replace a few
<fuzzy_id>
cool, simply setting systemd.services.jenkins.preStart worked :) i didn't expect that behaviour, thx!
Zoom has quit [Client Quit]
<{^_^}>
[nixpkgs] @globin pushed to master « libvirt: make xen support optional »: https://git.io/vjuUD
Zoom has joined #nixos
aarvar has quit [Ping timeout: 240 seconds]
<Zoom>
A tool screwed up my partition table so I need to rebuild a config of a non-booting system. How can I do it without reinstalling nixos (it downloads the whole thing again)? nixos-enter requires network connection to rebuild the config, even though nothing has changed besides the partitioning.
<Zoom>
The only option I see is to reinstall with the most basic config just to boot, but functioning chroot would be much better. What am I doing wrong?
<hyper_ch2>
Zoom: not sure what you're trying to do
<hyper_ch2>
what's wrong with nixos live iso?
<Zoom>
hyper_ch2, how can it help besides reinstalling? I'd like to just rebuild a config of a system I can't get to boot due to partitioning scheme changing.
<Zoom>
Backup help restoring files or whole partitions. I didn't lose anything, just the partition scheme changed. Should I rewrite the hole partition just to change a single UUID in a config, you think?
<{^_^}>
[nixpkgs] @jokogr opened pull request #42154 → jetbrains.editors: Fix update script and update versions → https://git.io/vjKbA
<{^_^}>
[nixpkgs] @adisbladis pushed commit from @Ekleog to release-18.03 « opensmtpd module: allow changing the package »: https://git.io/vj6eO
<Zoom>
In my mind the easiest way to solve my problem would be booting anything that has nix, chrooting into my existing installation and rebuilding the hardware-configuration. But this doesn't work due to nixos-enter requiring network connection. Am I missing something at this step? Do I use nixos-enter the wrong way?
atu has quit [Ping timeout: 276 seconds]
<{^_^}>
[nixpkgs] @jokogr opened pull request #42156 → libreoffice-fresh: 6.0.3.2 -> 6.0.4.2 and don't use system xmlsec → https://git.io/vjPez
<hyper_ch2>
or (1) get nixos live iso (2) boot live iso up (3) mount original root in /mnt and mount original boot in /mnt/boot and mount additional partitions if necessary (4) make backup copy of /mnt/etc/nixos/configuration.nix and /mnt/etc/nixos/hardware-configuration.nix (5) nixos-generate-config --root "/mnt/" (6) nixos-install
<Zoom>
hyper_ch2, that's what I started my question with. I've said it's the only way that would definitely work that I know of. I was just curious if I miss some way one can chroot to an existing installation.
<Zoom>
hyper_ch2, Thanks anyway.
atu has joined #nixos
andreabe_ has joined #nixos
andreabedini has quit [Read error: Connection reset by peer]
andreabe_ has quit [Read error: Connection reset by peer]
asuryawanshi has quit [Remote host closed the connection]
smallville7123 has joined #nixos
asuryawanshi has joined #nixos
<booglewoogle>
hey! I'm on a quest to get my haskell program (the simplest possible "print hello world" example) that I compiled on this nixos machine run on another linux mint one. i've used "patchelf --set-interpreter /lib/ld-linux.so.2 program" which makes the error when running it on the mint machine go from "no such file" to "Accessing a corrupted shared library". what can I do here?
<LnL>
so you copied only the binary?
Fare has quit [Ping timeout: 240 seconds]
Plato[m] has quit [Remote host closed the connection]
PLPD-Bot has quit [Remote host closed the connection]
<booglewoogle>
yea
<LnL>
making a tarball including it's dependencies would be much easier
<booglewoogle>
oh, no, this is more a learning journey than anything else, so both would be fine!
<LnL>
in that case
<LnL>
nix-store -qR $(nix-build '<nixpkgs>' -A hello) | tar -T - -czf hello.tar.gz
<LnL>
that created a tarball with the hello binary and all it's runtime dependencies
<booglewoogle>
okay
<booglewoogle>
i'll try that
smallville7123 has joined #nixos
crmlt has quit [Ping timeout: 260 seconds]
norfumpit has quit [Ping timeout: 265 seconds]
smallville7123 has quit [Client Quit]
<booglewoogle>
now, what sort of dependencies would that be in such a simple binary? or rather: why isn't just pointing it to the ld-linux.so.2 enough? sorry, trying to wrap my head around this from pretty much zero
lsix has quit [Ping timeout: 265 seconds]
<manveru>
booglewoogle: it also depends on shared libraries
<manveru>
try using `ldd` on your binary to see them
<guibou>
I'm trying to use docker on nixos (nixos-unstable channel) and adding `virtualization.docker.enable` in my configration (and `nixos-rebuild switch`) doesn't work (no docker service, no docker binary, no docker group): I had to install docker through `nix-env -i`, run `dockerd` as root and run my `docker` command as root. Is there an undocumented step I missed other than `virtualization.docker.enable = true`?
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
<gchristensen>
it sounds like nixos-rebuild switch didn't work, you sure it finished successfully? (maybe try againN)
stepcut has quit [Read error: Connection reset by peer]
stepcut has joined #nixos
smallville7123 has quit [Quit: rosa]
<CapsLock>
guibou: it worked for me, 2 days ago, and I've just set virtualisation.docker.enable to true
<guibou>
gchristensen: CapsLock: damned, "virtualisation", not "virtualization"... My bad...
<joepie91>
fetchTarball documentation: "Download the specified URL, unpack it and return the path of the unpacked tree. The file must be a tape archive (.tar) compressed with gzip, bzip2 or xz. The top-level path component of the files in the tarball is removed, so it is best if the tarball contains a single directory at top level."
<joepie91>
is there a way to make it *not* remove the top-level path component?
<gchristensen>
guibou: as an american who uses NixOS *a lot*, I've started to spell some words with an s over a z :P
<joepie91>
(assume that I cannot modify the tarball structure)
<johanot>
what is the easiest way of getting a list of all derivations that references derivation X ? Easier than: grep -r 55ccl83zw3srqqggmmfandmr44irc1v1-docker-image /nix/store
tzemanovic has quit [Remote host closed the connection]
crmlt has joined #nixos
tzemanovic has joined #nixos
<nico202>
anybody using fprintd? It stopped working a couple of months ago but never had the time to investigate. fprintd-enroll works, but then fprintd-list username does not list saved fingers
tzemanovic has quit [Remote host closed the connection]
<joepie91>
how do I set up nar signing for Hydra? it'
<joepie91>
how do I set up nar signing for Hydra? it's not in the manual *
<sphalerite>
joepie91: I don't actually know about hydra, but maybe by setting secret-key-files on the builders
acarrico has joined #nixos
<sphalerite>
That might be wrong though
atu has quit [Ping timeout: 240 seconds]
dbmikus has joined #nixos
bollu has quit [Quit: Ping timeout (120 seconds)]
<{^_^}>
[nixpkgs] @adisbladis pushed 3 commits to release-18.03: https://git.io/fenkc
camsbury has joined #nixos
camsbury has quit [Remote host closed the connection]
camsbury has joined #nixos
<aminechikhaoui>
Hi, any idea why would a cryptsetup luksOpen of an encrypted device won't fire up the required uevent so that systemd can continue handling it
<aminechikhaoui>
(my question is probably not clear though)
<aminechikhaoui>
but basically what I have is a hanging dev-mapper-bla-bla.device that waits for a /dev/mapper/foo which already exists
<aminechikhaoui>
also the odd thing (this is a GCE volume), if I choose for example a 10Gb volume everything works fine and udevadm monitor -e shows the uevent of the change and the device is mounted properly
bollu has joined #nixos
<aminechikhaoui>
with a 500Gb volume it just hangs though (both encrypted with luks)
reinzelmann has quit [Quit: Leaving]
<joepie91>
sphalerite: I can't find any documentation on this...
<joepie91>
ah hold on, it's referenced in the nix manual
jmeredith has joined #nixos
<{^_^}>
[nixpkgs] @adisbladis pushed commit from @yegortimoshenko to master « nixos/networkmanager: add extraConfig »: https://git.io/fecyV
ericsagn1 has quit [Ping timeout: 240 seconds]
<joepie91>
sphalerite: do you happen to know what user the builds will be signed under?
<{^_^}>
[nixos-org-configurations] @edolstra pushed 2 commits to master: https://git.io/feCfE
<LnL>
joepie91: I use store-uri = foo?secret-key=/run/keys/example.org-1/sk
bollu has quit [Quit: Ping timeout (120 seconds)]
<LnL>
and no not root
bollu has joined #nixos
<LnL>
hydra-queue-runner IIRC
mmercier has quit [Ping timeout: 256 seconds]
crmlt has quit [Ping timeout: 265 seconds]
johanot has quit [Quit: Leaving.]
<joepie91>
eh... my nixops spontaneously started compiling git, perl, and python?
<joepie91>
o_O
<crystalgamma[m]>
hi, I have a box running nix-on-Arch-Linux, and when I try using sandboxing for the builders I get "error: while setting up the build environment: getting attributes of path '': No such file or directory". Any ideas how to fix this?
vaninwagen has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 0 commits to yegortimoshenko-patch-4: https://git.io/fe022
sk8forether has joined #nixos
Fare has quit [Ping timeout: 256 seconds]
asuryawanshi has quit [Ping timeout: 248 seconds]
<crystalgamma[m]>
seems like it's trying to bind mount '' to the sandbox's /bin/sh, but why?
PLPD-Bot has quit [Remote host closed the connection]
Plato[m] has quit [Remote host closed the connection]
<rizary>
hi, when i add/execute program under `buildPhase`, then i run it in another derivation result, i got permission denied. where should i put such command?i.e:
<rizary>
`${node}/bin/npm run build --prefix ${frontend.package}/lib/node_modules/abc`
PLPD-Bot has joined #nixos
<rizary>
which by the looks, the execution will create `app` folder under `${frontend.package}/lib/node_modules/abc` like this ${frontend.package}/lib/node_modules/abc/app/bundle.js
Plato[m] has joined #nixos
jperras has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c merged pull request #42156 → libreoffice-fresh: 6.0.3.2 -> 6.0.4.2 and don't use system xmlsec → https://git.io/vjPez
<rizary>
i'm using `node2nix` for generating .nix file
<infinisil>
You mean the frontend one?
<rizary>
yep
<infinisil>
So what's exactly the error?
<hodapp>
weird. I can install 'wings', but I can't run it through Applications / Run nor by "wings &" from a commandline... but running it just as "wings" via commandline is fine
<rizary>
infinisil: i have updated the gist for the error.. sorry forgot about it in the first place
<infinisil>
rizary: Hmm, maybe try copying the frontend stuff instead, and/or chmod/chown it
<infinisil>
hodapp: Any error?
<hodapp>
infinisil: can't see one if I run in the two ways I mentioned
<infinisil>
hodapp: Try using strace
<hodapp>
not sure how to use strace to trace a process that's immediately backgrounded
<foldingcookie>
sudo strace -p $pid
<foldingcookie>
or strace -f the process that forks to spawn it (e.g. the shell)
<hodapp>
the process dies immediately
ryantrinkle has quit [Ping timeout: 268 seconds]
<infinisil>
Not a single strace output? That seems unlikely
<hodapp>
if you want to see what I mean, just install "wings" and then try to run with "wings &"
<{^_^}>
[nixpkgs] @xeji pushed commit from @mkaito to master « WoeUSB: 3.1.5 -> 3.2.1 (#42158) »: https://git.io/fvpuR
<earldouglas>
nix-store --optimize is giving me a bunch of "skipping suspicious writable file" errors. Is there a way to mount the store so that I can fix the write bits?
<joepie91>
infinisil: ... apparently this does not work for options.nix.binaryCachePublicKeys? error: attribute 'default' missing, at /home/sven/projects/nixrc/systems/desktop.nix:214:33
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 2 commits to master: https://git.io/ffCZM
<LnL>
don't understand the problem, assigning works fine
aveltras has joined #nixos
benley has quit [Remote host closed the connection]
benley has joined #nixos
benley has quit [Remote host closed the connection]
benley has joined #nixos
<aveltras>
hi everyone ! has anyone experience about applying a firmware patch to the kernel in nixos ? i have all the data (got the files from ubuntu patch) but don't know how to properly use it for nixos
<gchristensen>
sorry I meant copying around drvs from store to store
<gchristensen>
I figured it out I think, though ... now I just have a funny output from `nix-diff` http://ix.io/1dOg
<kolb>
Hi, so I have builds in hydra that write runtime data to /var/run/foo (and delete it afterwards) but somehow when run under hydra they fail to clean up after themselves. I assume this is due to some kind of nix-build sandboxing? Can I turn that off somehow?
<LnL>
might also suffer from nix#2227, but you could work around that if needed
<LnL>
euh, /var/run is owned by root so builds can't touch that
<kolb>
LnL: my builds run as root
<LnL>
no they don't
<kolb>
well they have sudo
<kolb>
I know its not intended that way, I am abusing hydra as a CI server
<LnL>
??
fendor_ has joined #nixos
* LnL
stops poking the bear
<infinisil>
Um, isn't hydra a CI server?
<Baughn>
It can be used as one.
<infinisil>
kolb: But really, why run it as root?
<kolb>
infinisil: yeah I guess that came out wrong... I use it as a CI server for a piece of software that directly messes with hardware via DMA and stuff
<infinisil>
Why write to /var/run?
Drakonis[m] has quit [Changing host]
Drakonis[m] has joined #nixos
Drakonis[m] has joined #nixos
<kolb>
infinisil: runtime data...
<Dezgeg>
you can set build-extra-sandbox-paths in nix.conf
<kolb>
Dezgeg: is that equivalent to using nix.SandboxPaths?
<Dezgeg>
oh hmm, you have them already running as root, no need for that then I guess
<LnL>
kolb: make your stuff put it's data in $NIX_BUILD_TOP
<kolb>
LnL: is that a tmpfs?
<LnL>
probably
hoshineko has joined #nixos
<kolb>
LnL: what does that point to?
<LnL>
a unique tmpdir where builds can write to
<LnL>
that way you don't have conflicts
<kolb>
LnL: Will try, thanks.
tenten8401 has joined #nixos
<tenten8401>
Anyone know how I could install grub to a soft raid 1 array using the configuration.nix?
<tenten8401>
also, somewhat related, but what key do you press to exit the NixOS manual, the one you get to with ALT F8
MichaelRaskin has joined #nixos
<LnL>
it's something like /tmp/nix-build-hello-2.10-0, created by nix before the build starts
<tenten8401>
If I need to download a new VNC client I can
dbmikus has quit [Ping timeout: 260 seconds]
<aveltras>
is there anything special to do to be able to use a local derivation in my configuration.nix file ? the rebuild says it's not of type package but nix-build my-package.nix does not seem to encounter any issue
<infinisil>
aveltras: gist the config?
<infinisil>
It should work by doing `the.option = import ./my-package.nix`
<infinisil>
Or add a {} if the file takes arguments
stranger___ has quit [Quit: Connection closed for inactivity]
<aveltras>
trying right now
slyfox has quit [Quit: no need to cry]
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #42174 → darwin stdenv: Put back "man" attribute on clang and llvm → https://git.io/ffC8r
Fare has joined #nixos
<tenten8401>
well after mashing the keyboard for a while I somehow exited the documentation, but now I've got a completely blank tty
<Guest12>
Is there a way to install the Nix package manager on macOS without it changing anything in my `/etc` directory? I tried `curl https://nixos.org/nix/install | sh` but it warns me about updating `/etc/bashrc`, `/etc/zshrc` etc. and I am afraid that I will lose control over all changes that will be made.
ryantrinkle has joined #nixos
<Guest12>
Also I get the impression it wasn't like that in the past. I recall installing Nix some time ago without it messing with my system at all.
<Guest12>
And now it wants to do it...
<MichaelRaskin>
Well, it's the question of defaults. I think you can just compile it without much trouble, for example.
<tokudan[m]>
I've got a strange penomena since about 2 to 3 weeks: I can't get a dynamic ipv6 address anymore in my network through network manager on 18.03. does it work for someone else?
<kreetx>
Guest12: you need the nix paths in your env, no?
<kreetx>
isn't that what it does under /etc
<{^_^}>
[nixpkgs] @Ericson2314 pushed 16 commits to staging: https://git.io/ffCET
<Guest12>
@kreetx yeah, but I'd like to add them manually every time I want to run something installed via nix
<__monty__>
Guest12: Fwiw it should leave your current copies of those around as .backup's or something. All it's trying to is add paths for your environment.
<__monty__>
Guest12: Sounds like a chore but you can just change back your rc files after the install.
civodul has joined #nixos
<Guest12>
is there really no way to have a 'local' nix install??
<infinisil>
Define "local"
<kreetx>
but you still want ~/.nix-profile/bin in your PATH to use it?
<tenten8401>
Anyone know how I'd go about installing NixOS on a soft raid 1 setup without using the /dev/mdX names, since those change?
<kreetx>
Guest12: in a sense I get what you mean: you just want to install nix things without it affecting any other users, and perhaps add the env stuff into user bashrc
* Baughn
tears his hair out.
<Baughn>
gchristensen: Do you think anyone would object if I make nixos-generate-config call out to a Rust program for some of its work?
<kreetx>
Guest12: not an expert, but since macos installs nix in multi-user mode, so that might have something to do with it
<kreetx>
Guest12: it used to be single-user some time ago
<Guest12>
kreetx: Ah right, so something changed some time ago?
<tenten8401>
I guess what I'm really asking is if there's a more "permanent" way to reference a raid array
<tenten8401>
I'm not real great with raid
<gchristensen>
Baughn: dunno, I wouldn't necessarily object, as long as its closure is tiny
<Baughn>
Its runtime closure is tiny. Build-time is something else again.
<Baughn>
I mean, the runtime closure is ['glibc'].
<kreetx>
Guest12: yes, they now install a multi user nix onto macos
jperras has joined #nixos
<gchristensen>
pretty sure building nixos requires rust somewhere in there anyway :/
<kreetx>
Guest12: so there's a daemon that services user requests to install nix things etc
<tokudan[m]>
tenten8401: that works if the filesystem stays the same. if you create another filesystem on it, the uid will change
<tenten8401>
filesystem should stay the same
grumble has quit [Ping timeout: 600 seconds]
<Guanin>
Sorry to bump this question, but I did not find anything related to this in the options list: Does NixOS provide any facility to configure an LTE modem?
<Baughn>
gchristensen: I realized that one of my earlier patches has opened NixOS users up to data loss, due to an erronous assumption on my part. So I'm feeling a bit frazzled atm.
<Baughn>
`zpool import` has the "import pool despite missing devices" option... little did I realize, that doesn't apply to missing *redundant* devices.
<gchristensen>
Baughn: take a second and calm down, take a deep breath, everything will be fine,
<Baughn>
Certainly it will be.
<Baughn>
As soon as this is fixed.
<gchristensen>
I would advocate for you learning a bit of perl first
<kolb>
yeah please
<Baughn>
Not going to happen.
<kolb>
(not that I know any perl)
<Baughn>
I mean, I can read it, but learning it would be an enormous waste of time.
smallville7123 has joined #nixos
<gchristensen>
what does the rust thing need to do?
tzemanovic has quit [Ping timeout: 248 seconds]
<Baughn>
*Ideally*, everything nixos-generate-config does today.
<__monty__>
If nix uses perl6 then don't you have opt-in types?
<Baughn>
Short-term? Parse the json output from lsblk.
<Baughn>
__monty__: Perl 5.
<gchristensen>
and then?
<Baughn>
Output, for each ZFS pool, the list of devices backing it.
aveltras has quit [Quit: Page closed]
<Baughn>
Which modules/filesystems/zfs.nix can then use to generate appropriate device dependencies.
<Baughn>
Oh, it needs to do some parsing of /dev/disk/by-id as well to get stable names. Nothing comple.
<Baughn>
x.
<gchristensen>
this seems complicated, what happens if a device fails out?
<gchristensen>
or is replaced?
<Baughn>
Systemd waits 90 seconds for it, then fails the device unit and moves on.
<Baughn>
The import will still be attempted, and may or may not succeed.
<gchristensen>
but won't it fail the tree of deps?
<gchristensen>
what about that automatic systemd unit file generator in the new (or coming in the new) zfs?
<Baughn>
Not if we don't use Required.
<Baughn>
Well, I mean...
<Baughn>
That generator does this same exact thing, but it's been "up and coming" for the last two years.
<Baughn>
And anyway, as I said, I'm against using Perl for core software that needs to keep working.
<tenten8401>
how long does grub typically take to install?
<Baughn>
Fraction of a second.
<tenten8401>
I've been waiting for like 3 minutes
<Baughn>
That means it's broken.
<tenten8401>
hm
<gchristensen>
(a) maybe we could contribute it to ZFS? (b) I think it is a bad idea for you to reimplement the config generator in Rust if you're not interested in learning Perl (and perl isn't going away any time soon) (c) I don't have any specific problems with you doing this thing in Rust, but I'm certainly not the only person to have an opinion on it, and I would prefer you learn the perl to do it in perl. perl isn't so
<gchristensen>
hard
<tenten8401>
seems like I/O is maxed on the system
<tenten8401>
might be a raid array syncing
<tenten8401>
idk
<Baughn>
That means it's *really* broken.
<MichaelRaskin>
tenten8401: I guess you have a sync
<MichaelRaskin>
Baughn: if you have a sync backlog, it might take some time…
<Baughn>
I suppose... 3 minutes is a bit much, but sure.
<MichaelRaskin>
Even without a RAID, just a backlog of sync=ing the store
<MichaelRaskin>
It depends
<MichaelRaskin>
Ig you have lots of RAM and install on a poor-throughput storage…
<tenten8401>
yeah the raid array is being resynced
<tenten8401>
42% complete
<tenten8401>
on 2 TB drives
<tenten8401>
so that'd probably be why if I had to guess
<Baughn>
gchristensen: I can bash my way through implementing it in Perl. I was halfway through doing so, in fact; that isn't my concern.
logzet has quit [Remote host closed the connection]
<MichaelRaskin>
Yes, that explains a lot
catern has joined #nixos
<tenten8401>
is it safe to reboot while an array is resyncing? or should I wait it out
<Baughn>
This is: Implementing a single program in Perl isn't the same thing as *learning* Perl, and I'm sure you know that.
<MichaelRaskin>
tenten8401: I wouldn't bet that GRUB is already installed
<Baughn>
If your proposal is that I spend half a year working on Perl software in my spare time in order that I can reliably work on this, then... okay, I can see where you're coming from, but my answer is still going to be a flat 'no'.
<tenten8401>
seems I have to write grub to each individual drive instead of going through the raid array
<MichaelRaskin>
I am not sure it won't wait until the resync
<tenten8401>
which is really dumb
<tenten8401>
I'll wait it out
<gchristensen>
that surely isn't what I'm saying ...
<Baughn>
tenten8401: I'd suggest looking into ZFS, if you haven't already. Using it on NixOS is pretty painless.
<MichaelRaskin>
It is GRUB not supporting RAID markup, I think
<tenten8401>
what are the advantages of using ZFS over ext4?
<tenten8401>
for a server install
<Baughn>
Checksumming and snapshots, mainly.
<tenten8401>
would it just be as easy as overwriting my ext4 partition with a zfs partition and installing to that?
<Baughn>
Sssort of. You do need to understand the underlying concepts of zfs.
crmlt has joined #nixos
<Baughn>
There is no "mkfs.zfs". It combines LVM/RAID with a filesystem, effectively; you wouldn't use it on top of mdraid or lvm.
<Baughn>
gchristensen: Then I'm not sure what you're saying, but the fact is that I won't feel comfortable submitting code in a language I don't feel solid on.
<tenten8401>
sounds confusing
<tenten8401>
I'll look into it
<Baughn>
Especially when it has no typing.
<gchristensen>
I mean, I didn't tell you _not_ to do it in Rust... I just said I'd _rather_ you do it in perl :P
<Baughn>
Yes, a lot of people told me to do it in Perl.
<Baughn>
I get the idea. It's fine. I'll just leave the bug report and ask someone else to fix it.
<MichaelRaskin>
gchristensen: why increase the amount of code in a patented unreadable language?
<gchristensen>
I don't know why you're taking my feedback and turning it in to a no :/
<Baughn>
...um, I'm reading it as "You'd prefer this in Perl".
<__monty__>
MichaelRaskin: Which of the two languages are you talking about? Both sound plausible ; )
<Baughn>
There are people in here who understand Perl. I'm not one of them.
<gchristensen>
yeah, but preferences are not requirements
<Baughn>
So the reasonable action is to leave it to one of them.
<MichaelRaskin>
Perl
<gchristensen>
okay
<MichaelRaskin>
I sincerely think that Perl is less readable than J.
<Baughn>
If you'd said something like "Well, we'd prefer using (any language other than Perl), but it's all in Perl for historical reasons", then I'd have been happy to help with that.
<Baughn>
It's what I was expecting, in fact. :P
<gchristensen>
I'm honestly confused why my preference is blocking you from doing it in rust
<Baughn>
Um... because I don't want to create a maintenance nightmare?
<Baughn>
Which is what having it in multiple languages long-term would be.
<gchristensen>
ah, I see
<gchristensen>
yes, I agree
<Baughn>
Also because I tend to take maintainers' preferences are iron-clad law, as the alternative gets aggravating to the maintainers, and I don't want to be That Guy.
smallville7123 has quit [Quit: rosa]
<gchristensen>
yeah, that is hard, and I can appreciate that
<infinisil>
I'm all for using more modern and safe languages in nixpkgs
<gchristensen>
ultimately we'd need to get Eelco's opinion on this
<Baughn>
nixos-generate-config is only 650 lines. I thought it might be a usable test-case.
<gchristensen>
then let's get his opinion :)
<Baughn>
That'd be why I wrote that bug, yes.
<Baughn>
...I'm sorry if it's a bit meandering. Best I can say in my defence is I'm doped up on anti-allergy medication.
<MichaelRaskin>
Rest assured that getting any opinion out of Eelco is hard (including Eelco who is expected to give more opinions than he has), and there is no consensus among maintainers on any issue.
<manveru>
we can have a competition, the most readable and safe language wins :D
<Baughn>
Rust ain't all that readable, but it's definitely *safe*. And small-at-runtime, and fast, all of which I think are important.
<infinisil>
manveru: I nominate Nix
<Baughn>
Well, not so important in *this* case but...
<{^_^}>
[nixpkgs] @xeji merged pull request #42172 → connect: works on darwin → https://git.io/ffCGH
<{^_^}>
[nixpkgs] @xeji pushed commit from @clacke to master « connect: works on darwin (#42172) »: https://git.io/ffCaR
<MichaelRaskin>
Somehow some people are still not good at reading reasonable Shell, even though we have to debug upstream build scripts all the time
<Baughn>
Total # of languages in nixpkgs is a number to reduce if possible.
<Baughn>
Think about Rust is, it's fairly hard for a beginner to accidentally break something.
<infinisil>
Baughn: For maintenance? Or are there also other points?
<manveru>
infinisil: that's why i rewrote my nixos-option in nix instead of bash :P
<Baughn>
infinisil: Plenty of points. It has a powerful type system, which apart from making it memory-safe also makes it possible to write inherently thread-safe code.
<Baughn>
I.e, assuming you don't go out of your way to break the system, it's *easy* to safely write multithreaded code.
<Baughn>
(That was one of the design criteria, way back.)
nico202 has joined #nixos
<Baughn>
And also there's, yes, that if your patch *compiles* it'll probably work. Or at least not break anything outside of itself.
<kolb>
How do you clean up old Hydra builds to make space for new ones?
<infinisil>
Baughn: Ah yeah, I meant: Are there other points for having a low # of total languages in nixpkgs other than maintenance?
<Baughn>
Oh. No, it's basically maintenance.
lysk has joined #nixos
<Baughn>
Some people enjoy the smaller closure, I suppose, but I'm not one of them.
<Baughn>
The size of the minimum system closure doesn't matter so much to me when the size of my *actual* system closure is >10 GB. :P
<{^_^}>
[nixpkgs] @xeji pushed commit from @mrVanDalo to master « pencil : fix udev dependency (#42147) »: https://git.io/ffCaF
<infinisil>
I'd honestly rather learn Rust than Perl if I'd contribute to nixpkgs programs. And I'm sure the number of such people will go up with time
<gchristensen>
its really important to keep the minimum closure small for initial user experience and initial deployments
<manveru>
but nixos-generate-config would just be a binary, right?
<gchristensen>
I know nixos systems very rarely have a small system closure beyond the firsts 5min, but those first 5min are importan
<infinisil>
manveru: Yeah, I don't know where this increased size would be coming from either
<gchristensen>
yeah, I'm addressing just "The size of the minimum system closure doesn't matter so much to me when the size of my *actual* system closure is >10 GB. :P"
<lysk>
Hey folks! While installing NixOS I accidentally confused my swap and root partitions. Since 8GB are not really enough for root, I deleted swap, resized root and created a new swap partition. Unfortunately the new swap partition no has a UUID different from the one in hadware-configuration.nix. Does anyone of you guy maybe have an idea how to fix this?
Slabity has quit [Remote host closed the connection]
<manveru>
i just don't like writing CLI tools in rust, but that's only my opinion :)
<infinisil>
manveru: Why not?
<Baughn>
gchristensen: It doesn't matter much to *me*. I can appreciate that it matters to others.
<infinisil>
lysk: Regenerate it with nixos-generate-config, hardware-configuration.nix is meant to be updated like this on hardware changes
<manveru>
infinisil: my native language is ruby :P
<gchristensen>
Baughn: understood :) thank you
<infinisil>
manveru: (I don't see the point just yet)
<infinisil>
Is it super easy to write CLI's in ruby you mean?
<Baughn>
gchristensen: Oh, incidentally. If you look around on the internet, you'll find people saying this ZFS import problem is fixable by setting `scsi_mod.scan=sync`.
<lysk>
infinisil: Will nixos-generate-config also overwrite my configuration.nix or just hardware-configuration.nix?
<Baughn>
*This is false*.
<infinisil>
lysk: Just hardware.
<manveru>
infinisil: pretty much, yes, around the same level as perl
<gchristensen>
yikes
kreetx has quit [Quit: leaving]
<Baughn>
It seems to work on BIOS systems, but not on any of my UEFI systems.
<Baughn>
Presumably because UEFI doesn't hand-off the devices to the kernel, it just expects the kernel to probe them again.
<Baughn>
Or something like that. I haven't looked at it in detail.
<manveru>
infinisil: and with crystal you get nice binaries and a type system, so that's my go-to these days
<{^_^}>
[nixpkgs] @xeji pushed commit from @bkchr to master « vlc: Fixes build with fribidi version 1.0 (#42130) »: https://git.io/ffCVB
<{^_^}>
[nixpkgs] @xeji merged pull request #42130 → vlc: Fixes build with fribidi version 1.0 → https://git.io/vhMBC
<infinisil>
Baughn: I actually fixed this import issue by updating my hardware-configuration.nix once, a kernel module got replaced, hold on I'll check
<Baughn>
My native language is Kotlin, but I doubt anyone would be fine with putting *that* on the installation disk. :P
<__monty__>
Yes, so obviously we should rewrite everything in haskell : >
<manveru>
:D
<Baughn>
Heh.
<Baughn>
You joke, but Rust and Haskell are really very similar.
<Baughn>
(Alas, no monads. ...yes, I'm upset about that dangit!)
<infinisil>
Baughn: Ah found it, boot.initrd.availableKernelModules switched "usb_storage" out for "uas"
<Baughn>
..okay. I have no idea why that would help.
<infinisil>
This solved the problem of a boot not importing all devices
<manveru>
Baughn: i'm just upset that they kept semicolons
<MichaelRaskin>
Rust is waay more pragmatic than Haskell about side-effects.
<fresheyeball>
__monty__: that is there. But I dont see a client
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<__monty__>
fresheyeball: Are you sure that isn't the client?
<fresheyeball>
__monty__: yes
<fresheyeball>
I downloaded the client from the website and its mattermost-desktop
<__monty__>
Hmm, weird cause that's from the webapp tarball the derivation downloads.
<__monty__>
I'm not referring to mattermost-plaftform btw, that seems to be the binary for mattermost-server.
<fresheyeball>
hu
<fresheyeball>
I dont get it
<__monty__>
The derivation adds two binaries "platform" and "mattermost-platform".
work_ has joined #nixos
nschoe has joined #nixos
<nschoe>
Hi everyone,
<nschoe>
I have an issue with the Arduino IDE with nixOS, quite often, apparently randomly, I've got this error message when I try to compile a sketch: "fork/exec /home/nschoe/.arduino15/packages/arduino/tools/avr-gcc/4.9.2-atmel3.5.4-arduino2/bin/avr-g++: no such file or directory"
jtojnar has quit [Remote host closed the connection]
<nschoe>
Obviously this files _does_ exist, so I don't understand why it says this.
<nschoe>
I've noticed that by removing the ~/.arduino15 directory and starting the IDE again it worked, but really this is weird. Anybody else has this?
jtojnar has joined #nixos
iMatejC has quit [Remote host closed the connection]
iMatejC has joined #nixos
iMatejC has quit [Remote host closed the connection]
<Ralith>
nschoe: that "no such file or directory" error usually means the binary references the wrong dynamic linker
<Ralith>
this typically happens when you obtained the binary from a non-NixOS source
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « ace: 6.4.8 -> 6.5.0 (#42038) »: https://git.io/ffCid
slyfox has joined #nixos
den_ has joined #nixos
xy2_ has quit [Ping timeout: 248 seconds]
Lev50 has joined #nixos
greg_ has joined #nixos
den_ has quit [Client Quit]
Fare has quit [Ping timeout: 248 seconds]
Fare has joined #nixos
<greg_>
Hello. What is the convenient way to program using nixos? For example i have some simple haskell projects, and i want to have all the smart features in my editor like intero, formatting tool, etc. Should i run my editor from nix-shell every time (im scared of this a little, because previously i have had some troubles with tmux running from it) or install package management tool globally and then run nix-shell just for compiled project?
<infinisil>
stack works without needing to be in a nix-shell
lord| has quit [Quit: WeeChat 2.1]
lord| has joined #nixos
<infinisil>
So I'm only using that with emacs + haskell-ide-engine
<greg_>
Oh thats nice. But thats just the special case of stack. What about other languages? Like f.e. rust
fendor_ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @xeji merged pull request #41470 → singularity-tools: create mount points for image building → https://git.io/vh4nL
<{^_^}>
[nixpkgs] @xeji pushed commit from @jbedo to master « singularity-tools: create mount points for image building (#41470) »: https://git.io/ffCXl
asuryawanshi has quit [Ping timeout: 245 seconds]
<__monty__>
greg_: Maybe look at johnw's config, I know he does at least some haskell and I'm sure he uses emacs when he does ; )
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andymandias has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @xeji merged pull request #41227 → tesseract: make tessdata a fixed output derivation → https://git.io/vh37G
<{^_^}>
[nixpkgs] @xeji pushed commit from @symphorien to master « tesseract: make tessdata a fix output derivation (#41227) »: https://git.io/ffC1B
andymandias has joined #nixos
__monty__ has quit [Quit: leaving]
hph^ has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @markuskowa opened pull request #42182 → nixos-18.03/pgbackup: Fix bug in postgresql-backup module that causes to fail → https://git.io/ffCMt
tzemanovic has joined #nixos
noobly has joined #nixos
<tenten8401>
is fetchFromGithub still a thing in nix?
<tenten8401>
says undefined variable when trying to use it
<fresheyeball>
tenten8401: pkgs.fetchFromGitHub
<fresheyeball>
capital H
JasonGrossman has quit [Ping timeout: 264 seconds]
phdoerfler1 has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @IvanMalison opened pull request #42184 → pasystray: add libappindicator-gtk3 for appindicator support → https://git.io/ffCSx
<infinisil>
I can help you with the module system :D
phdoerfler has quit [Client Quit]
<tenten8401>
I'll take any help I can get
<tenten8401>
currently all I've got done is the package (I think)
<infinisil>
tenten8401: Looks like it, a couple suggestions to the nix file you linked
<infinisil>
pkgs shouldn't be an argument, replace that with `fetchFromGitHub` to not needing to use pkgs for it
<infinisil>
Nitpick: The / after $out is unnecessary
<tenten8401>
yeah, a lot of it was copied from the nextcloud package
<infinisil>
It is somewhat of a convention to use `meta = with stdenv.lib; {`, so you can use `with maintainers; [ ..`, `licenses.mit` it the declarations
<infinisil>
Ah yeah
<infinisil>
And the fetchurl and fetchpatch argument isn't used, so you can remove those
<infinisil>
Oh also, does this package really need no build phase? So it's only a source unpacking really?
<tenten8401>
it's just a set of PHP scripts
<tenten8401>
but it does need to be configured after installation though
<{^_^}>
[nixpkgs] @ariutta opened pull request #42185 → docs: minor rewording for easier reading. → https://git.io/ffC9s
<tenten8401>
which is what I thought the module was for
<infinisil>
Oh nvm, my idea was to use pkgs.srcOnly, but it doesn't propagate the meta field, so that's no good for this
<infinisil>
The module must be thought of to make lives easier on NixOS what would have to be done manually on other distros
<infinisil>
The package should be usable in of itself for all Nix users
<tenten8401>
the installation process is pretty much just unzip, install dependencies, and configure the .env file
<tenten8401>
the package should take care of the unzipping as long as the user has the dependencies I would think
<{^_^}>
[nixpkgs] @nek0 opened pull request #42186 → update vym to 2.6.11 → https://git.io/ffC9R
<infinisil>
There is somewhat of a convention to embed a configuration file into packages
<{^_^}>
[nixpkgs] @jameysharp opened pull request #42187 → openjdk: fix build for !enableGnome2 → https://git.io/ffC9E
<tenten8401>
hmm..
<tenten8401>
I have no clue how I'd do that
<infinisil>
It's not as nice as the module system, but it does the job: Adding an argument `config` at the top, defaulting to `null`, and making a wrapper of sorts to make the package use the given config
<tenten8401>
it does already have an example configuration that comes with it
<infinisil>
I really suggest to not worry about this, and instead just focus on the nixos module
<infinisil>
People will have to expect to configure things on their own when they don't use nixos :)
<infinisil>
It's the same for all packages really (unless you use some trickery)
<{^_^}>
[nix] @dtzWill opened pull request #2241 → progress-bar: refresh occasionally even if no updates are received → https://git.io/ffC96
<infinisil>
Eventually I'll implement that tag system, just wait!
earldouglas has joined #nixos
<samueldr>
infinisil: in that case, tags shouldn't be in the FS, use a specific system, like per first letter
<offline>
infinisil: also if you don't consider the tagging part of the filesystem, there's ways.
<infinisil>
samueldr: Yeah that would be the idea
<samueldr>
b/bash h/hello
<tenten8401>
would a webapp be considered a service?
<offline>
samueldr: sed: unterminated s/ command.
<tenten8401>
would I have to inject into nginx somehow when it's enabled?
<offline>
well that was a b command.
<infinisil>
tenten8401: Not sure what you mean, but nixos is very flexible with such things
<samueldr>
eh, that was a directory listing of a small subset of nixpkgs :)
<samueldr>
tenten8401: if it runs its own webservice, it is fine to expect the end-user to configure the reverse proxy themselves
<tenten8401>
like, could I put "services.pterodactyl.enable" and have it autoconfigure nginx?
<infinisil>
samueldr: The basic idea of a tagging system would be: put all packages in a flat directory, add a `meta.tags` field to each one of them, create some functions to query tags
<tenten8401>
it doesn't run it's own webservice, it relies on a webserver
<samueldr>
it's not right to expect nginx will be used
<samueldr>
oh
ericsagn1 has quit [Ping timeout: 260 seconds]
<tenten8401>
so it shouldn't be classified as a service?
<infinisil>
tenten8401: Yeah nginx can be used for that
<infinisil>
(Not entirely sure what it exactly uses it for though)
erasmas has quit [Quit: leaving]
<samueldr>
ah, php-based
<infinisil>
Ah yes, that seems fine
<samueldr>
I'm kind of bummed how for apache there is/was a system to manage webapps, but not for nginx
<samueldr>
and it's not designed in a way to be inter-compatible
<fresheyeball>
elvishjerricco: I wish it was working!
<elvishjerricco>
fresheyeball: FWIW, you may need to use `127.0.0.1` instead of `localhost`. IIRC, I had some problems with the `localhost` name before.
<fresheyeball>
hmm
<infinisil>
fresheyeball: It would be helpful if you posted the error message..
<elvishjerricco>
Similarly, <3 `journalctl` :P
<fresheyeball>
infinisil: unfortunately I dont have one
<fresheyeball>
I checked the journal and its clean
<infinisil>
Well that makes it really hard to help then
<fresheyeball>
browser just says "timed out"
<fresheyeball>
I agree
<elvishjerricco>
fresheyeball: Deson't sound like a symptom of the issues I had with `localhost` resolution, but trying `127.0.0.1` may be worth a shot anyway
dbmikus_ has quit [Client Quit]
<infinisil>
fresheyeball: Try it with a non-proxying nginx config: set the root of the virtualhost to some directory and put some index.html in it
logiq has quit [Ping timeout: 260 seconds]
<fresheyeball>
if I comment out all the ssl stuff, it works