<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/JvsJy
<flameslicer>
Could someone help me get my DVD burner working? Brasero, bashburn, and dvd+rw-tools can't burn anything with an error of "unable to mount /dev/sr0"
alex` has joined #nixos
alex` is now known as Guest60335
alexherbo2 has joined #nixos
<lovesegfault>
jtojnar: Wait from this it sounds like I _do_ want zipAttrsWithNames?
aveltras has quit [Quit: Connection closed for inactivity]
<jtojnar>
lovesegfault depends if you want only zip all attrs or just a subset
<lovesegfault>
jtojnar: would it be like foobar = name: lib.zipAttrs foo.name bar.name?
<jtojnar>
lovesegfault more like foobar = lib.zipAttrs foo bar
<lovesegfault>
Oh oh oh
<lovesegfault>
I see
<infinisil>
worldofpeace: Not sure what you mean
<lovesegfault>
awesome
<infinisil>
worldofpeace: The comment I made there is in relation to the one above that
<jtojnar>
also not sure if Nix recurses into lists inside attrset by default
Guest60335 is now known as alex``
<lovesegfault>
jtojnar: I don't need it to
<lovesegfault>
also thank you
<lovesegfault>
you just made my life 15% more awesome
leothrix has joined #nixos
<worldofpeace>
infinisil: ohhhh, will do that after I'm done porting a test to python.
<lovesegfault>
Can I make `nix build` (**NOT** nix-build) spit out the path?
<lovesegfault>
I want to `nix build` and then push to cachix
marcusr has quit [Remote host closed the connection]
marcusr has joined #nixos
<evanjs>
okay, here's a fun one. trying to get list of fonts minus the big guys, so I'm doing something like this: `lib.lists.filter (f: f.name != "nerdfonts") config.fonts.fonts`
<evanjs>
the problem is, one of the fonts is a string with context... not sure why it is, but I'm wondering if I should just filter that case out for now or actually try to solve it
<evanjs>
I figure it's some system thing that does that
delli3 has joined #nixos
delli3 has quit [Max SendQ exceeded]
bhipple has joined #nixos
tomberek has quit [Ping timeout: 265 seconds]
<evanjs>
/nix/store/18j2wb08w56klf7shdzhxfczcv1n167x-ghostscript-9.50/share/ghostscript/fonts is the string in question
<evanjs>
some very lazy nix-store -q queries show /nix/store/34d8jnnsbbygq5crw5g5g0316j5wrrlj-texlive-combined-2019 ... yeah, I think I know where this is going...
knupfer1 has joined #nixos
knupfer has quit [Ping timeout: 265 seconds]
knupfer1 is now known as knupfer
<evanjs>
guess I could just do `typeOf a != "string"` and that should do for now~
leothrix has joined #nixos
sigmundv has joined #nixos
<infinisil>
evanjs: You could use lib.isDerivation to filter derivations
flameslicer has quit [Ping timeout: 260 seconds]
leothrix has quit [Ping timeout: 260 seconds]
kreisys has joined #nixos
kleisli has quit [Ping timeout: 268 seconds]
alex`` has quit [Ping timeout: 268 seconds]
alexherbo2 has quit [Ping timeout: 265 seconds]
Rusty1 has quit [Remote host closed the connection]
hoobershaggus1 has joined #nixos
hoobershaggus1 has quit [Remote host closed the connection]
felixfoertsch has quit [Ping timeout: 260 seconds]
flameslicer has joined #nixos
wildtrees has quit [Quit: Leaving]
kleisli has joined #nixos
h0m1 has quit [Ping timeout: 265 seconds]
h0m1 has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #73547 → quarte: use qt5’s mkDerivation and an other fix 19.09 backport → https://git.io/JeoSP
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to release-19.09: https://git.io/JvsTX
<flameslicer>
I'm trying to run a binary I grabbed off the internet but it won't launch. Running it on an ubuntu PC works fine, and the only error that shows is "No such file or directory". Any idea how I troubleshoot htis?
<flameslicer>
*this
<infinisil>
Flameslicer: You need to patchelf it
<infinisil>
Flameslicer: Write a simple default.nix file that fetches the binary, then set `nativeBuildInputs = [ autoPatchelfHook ];` and install the binary in $out/bin in installPhase
<{^_^}>
[nixpkgs] @strager opened pull request #79067 → chatterino2: fix install on macOS → https://git.io/Jvsk7
kreisys has quit [Ping timeout: 265 seconds]
kreisys has joined #nixos
thongpv871 has joined #nixos
<{^_^}>
[nixpkgs] @dkudriavtsev opened pull request #79068 → vvvvvv: init at 2.3-git-2020-11-22 → https://git.io/JvskA
<thongpv871>
I'm configuring home manager for fzf integration. I want fzf to use my command as source for fzf history: "fc -l 1 | sort -r --key 2.1 -k1,1nr | uniq -f 1 | sort -rn". I set this command as "programs.fzf.historyWidgetCommand" but it does not work
<{^_^}>
[nixpkgs] @dkudriavtsev opened pull request #79069 → hercules: init at 3.13 (attempt 2) → https://git.io/JvsIJ
gyroninja has joined #nixos
<gyroninja>
Is anyone familiar with using Ruby and Nix? I am having trouble figuring out how to use a relative path to a gemfile. It complains with `/nix/store/1agnq6dzfm68n6dkk05m1j0bjc30ycw0-ruby-2.6.4/lib/ruby/2.6.0/bundler/source/path.rb:198:in `load_spec_files': The path `/nix/store/my-lib` does not exist. (Bundler::PathError)`
orivej has joined #nixos
<gyroninja>
In my Gemfile I am trying use a path of `../my-lib`
<flameslicer>
So I'm trying to package a python program in Nix, and the issue I'm running into is that the package isn't a binary. What ways are there to either package or compile the program into a binary?
chagra_ has joined #nixos
chagra has quit [Ping timeout: 265 seconds]
jmeredith has quit [Quit: Connection closed for inactivity]
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected. See: tofu-vim
<evanjs>
infinisil: right??? I mean it's great they have one, but....
<flameslicer>
So flame is being a dumb. I'm trying to run a nix build on a python package and I get the error "anonymous function at default.nix called without required argument 'buildPythonPackage' ...", and I'm not sure how to actually call the argument
<samueldr>
,callpackage
<{^_^}>
If a Nix file foo.nix starts with something like `{ stdenv, cmake }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<samueldr>
it's how nixpkgs handle dependency injection
<lovesegfault>
How do I use recurseIntoAttrs?
<samueldr>
in all-packages.nix, all packages are being handled using `callPackage`, which in turn inspects the arguments and injects the right things
<samueldr>
nix doesn't know about the convention in nixpkgs, so it simply sees an expression that is a function with arguments it doesn't know how to fill
<lovesegfault>
Can I use the `nixos-unstable` channel in a non-nixos box?
<samueldr>
lovesegfault: yes
<lovesegfault>
Nice
<samueldr>
the main difference is that nixos-* channels wait on nixos tests
<samueldr>
while nixpkgs don't
<lovesegfault>
samueldr: do I call it `nixos` or `nixpkgs`?
<samueldr>
though nixpkgs-* will wait on darwin stuff
<samueldr>
not sure
<adisbladis>
Flameslicer: samueldr: Noteworthy thing about python packages: You'll want to use the python-specific callPackage function.
<lovesegfault>
Alright, let's see
<samueldr>
adisbladis: oh, right
<flameslicer>
What is that one? Just pythonCallPackage?
KeiraT has quit [Remote host closed the connection]
<adisbladis>
Flameslicer: python3.pkgs.callPackage (or python3Packages.callPackage, it's an alias of the former)
KeiraT has joined #nixos
Grauwolf has quit [Read error: Connection reset by peer]
m0rphism has quit [Ping timeout: 265 seconds]
Grauwolf has joined #nixos
<{^_^}>
[nixpkgs] @equirosa opened pull request #79076 → vimPlugins.bclose-vim: initi at 2018-10-10 → https://git.io/Jvst6
sigmundv__ has quit [Ping timeout: 265 seconds]
<flameslicer>
What should i do in the event of a package not having a setup.py? In my case the one I'm working with is run off of a .py file.
<flameslicer>
Thanks for all the help, by the way. I do appreciate it.
<adisbladis>
Flameslicer: You could supplement your own setup.py or override phases and just chuck the .py file into the output
<infinisil>
adisbladis: /nix is messed up in my browser
<adisbladis>
Oh, messed up here too
<adisbladis>
That's new
<adisbladis>
Flameslicer: For this expression to work on ARM you'd have to cross-build the entire expression (or build it natively on ARM)
<adisbladis>
Though I don't think you'd need any ARM cross stuff at all since the arm binary is distributed in the repo
<flameslicer>
So the main differences I'm seeing between mine and yours are using stdenv with a pythonenv that has the packages needed, and simplifying the buildinputs, right?
mexisme has joined #nixos
mexisme_ has quit [Ping timeout: 260 seconds]
<adisbladis>
Flameslicer: Yeah, since the package is not really a proper python package it's easier to not use the python build infrastructure like buildPythonPackage
<adisbladis>
Instead we create a python environment with python3.withPackages. That gives you a wrapped python interpreter with all the packages that we can stick in the script shebang
v88m has quit [Ping timeout: 265 seconds]
<adisbladis>
Normally this would be handled by patchShebangs (a nixpkgs stdenv function), but since we don't even have that we create our own shebang
<slack1256>
I got a nixos iso I want to install on an old laptop from 2010. It has a BIOS but not uefi. I burned the iso to a pendrive via `dd` but it isn't working, it's dropping me onto a rescue shell . So I can guess UEFI support isn't the problem.
<slack1256>
Has anyone problems installing a current nixos iso?
minicom is now known as minicomocinim
minicomocinim is now known as minicom
<slack1256>
Alternatively, where can I find old releases of nixos?
alex`` has joined #nixos
<flameslicer>
The current nixos iso worked fine for me when I used it, though I can't find any archived verisons of older releases.
alexherbo2 has joined #nixos
<flameslicer>
Unhelpful as this sounds, maybe using a different flash drive, or maybe even a dvd might work instead.
<slack1256>
The error said it couldn't find something on such inode table etc
<slack1256>
I will try another pendrive
peelz has joined #nixos
bvdw has quit [Read error: Connection reset by peer]
<flameslicer>
slack1256: Glad to hear that fixed it!
nckx has quit [Ping timeout: 268 seconds]
Heirlung has quit [Ping timeout: 240 seconds]
<ongy[m]>
Is there a good example of a session systemd unit that uses the session dbus?
Heirlung has joined #nixos
rogue_koder has quit [Ping timeout: 265 seconds]
chagra has joined #nixos
nikola_i has quit [Ping timeout: 265 seconds]
<ongy[m]>
hm, found one weirdness. The way I started sway i thas a session dbus service running. The way I wrote my service, it started on in the session systemd's control
<{^_^}>
[nixpkgs] @vcunat pushed 94 commits to staging-19.09: https://git.io/JvsYt
<magnetophon>
When I do "nix-shell -I $NIXPKGS FaustShell.nix", I end up with the system versions of my dependencies, instead of the NIXPKGS ones. NIXPKGS is correctly set and works when used in other commands. Any ideas?
aaa_ has joined #nixos
o1lo01ol1o has joined #nixos
ar has joined #nixos
o1lo01ol1o has quit [Ping timeout: 265 seconds]
<LnL>
you probably want -I nixpkgs=$NIXPKGS
<magnetophon>
LnL: thanks, that was it!
Hunterkll has joined #nixos
<Hunterkll>
howdy! I was just told by someone that NixOS has something similar to Solaris's boot environment management, IE I can run a BE clone on solaris and have two grub entries that are identical OS boots, but i can modify both independantly without it affecting the other, is that the case?
<magnetophon>
Hunterkll: yes, kind of: you can make separate boot entries, but you can't edit them once built. That is one of the key features of nixos: immutability
<magnetophon>
Hunterkll: every time you modify your config and build it, you get a new grub entry.
<Hunterkll>
magnetophon, ah okay, i was thinking of Solaris's ability to run a be clone, and i have two identical BE's that I can modify independantly, only taking up the storage space of the changes
<Hunterkll>
but that makes sense
M-Gregoire49 has quit [Ping timeout: 240 seconds]
<LinuxHackerman>
Hunterkll: you can use zfs for that sort of stuff, but it may need some legwork in terms of getting nixos's config to mount the right datasets.
<Hunterkll>
LinuxHackerman, yea, that's about how solaris does it, via ZFS shenanigans, okay.
<Hunterkll>
Friend mentioned that it had that functionality when i showed him the BE docu from solaris, and wanted to ask/confirm
<Hunterkll>
Thanks! He made it seem like there was a native command/config to do it
M-Gregoire49 has joined #nixos
<magnetophon>
Hunterkll: The new configs will also just take space for the changes, but on a package granularity, not via the filesystem
<Hunterkll>
right, of course
<Hunterkll>
he was just convinced there was an 'easy mode' BE clone command like on solaris, and if such a thing existed in a linux system i'd be very happy
<{^_^}>
[nixpkgs] @flokli merged pull request #78556 → nixosTests: Port more tests → https://git.io/Jvq5d
<magnetophon>
Hunterkll: afaict they are functionally equivolent, just a different way to get there.
<magnetophon>
note: i have 0 knowledge about solaris, just what you just told us.
<Hunterkll>
magnetophon, basically, in solaris, I just run a BE clone, and now have two entries in grub which are the same install but any modification in either doesnt' affect or is seen by the other
gour has joined #nixos
<Hunterkll>
magnetophon, so i can test even major OS upgrades/patches, or just minor config changes, and then I can blow away with another one line command the environment and reclaim any used disk space
<gour>
is it possible to use nixos with some non-systemd init system?
<{^_^}>
[nixpkgs] @sikmir opened pull request #79084 → tt2020: init at 2020-01-05 → https://git.io/JvsY6
o1lo01ol1o has joined #nixos
<sphalerite>
gour: not currently.
aaa_ has quit [Quit: Leaving]
<gour>
sphalerite: it's considered for the future?
<sphalerite>
gour: you could build an openrc (or whatever your preference of service manager is) based system with nix pretty easily, but it wouldn't be nixos and you wouldn't have all the pre-written modules you have with nixos.
<magnetophon>
Hunterkll: yeah, you can do that. But instead of cloning and then making changes, you make changes in your config and then rebuild. I think it's better, cause you can track your changes in git.
o1lo01ol1o has quit [Remote host closed the connection]
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
UndefinedIsNotAF has joined #nixos
BjrnForsman[m] has left #nixos ["User left"]
<sphalerite>
gour: I don't see it happening anytime soon. Quite afew people have talked about it and maybe started implementing it, but AFAIK none of those efforts ever went anywhere.
<gour>
sphalerite: ok. is it possible to repackage binary *.deb into nix for pkgs not avaialble in src form?
<sphalerite>
gour: there's this "church of suckless nixos" thing, they don't really do much in public though so it's unclear what the status of SLNOS is.
<gour>
sphalerite: ok, so if i want non-systemd distro, better to look elsewhere
<sphalerite>
gour: yes, it's how skype, some printer drivers, and a few other things are packaged. It does generally take quite a bit of patching.
<gour>
ok
<sphalerite>
gour: yes. I would recommend giving nixos a try though, a number of systemd haters still use it because of its other advantages :)
<sphalerite>
gour: or you could have a look at guix, the "GNU Nixos" which uses lisp and the GNU shepherd
<gour>
well, systemd broke (again) my system (manjaro) and have enough, so looking for alternatives
<gour>
sphalerite: guix does not any non-free pkg, afaik
<sphalerite>
nixos is a bit more breakage-resistant, regardless of systemd.
dingenskirchen has joined #nixos
<gour>
sphalerite: well, latest systemd update on manjaro (stable) require that i pass 'iommu=soft' due to my older gigabyte mobo
<gour>
and systemd-udevd hogs my cpu (100%) :_(
<Hunterkll>
magnetophon, ah, see, i split BE's then make a single change, see effects, than commit back or delete the previous BE - though a lot of the changes are potential 'render non bootable' changes
<Hunterkll>
but they're also changes that require kernel parameter changes in hex
<Hunterkll>
so it gets ... interesting
chagra has quit [Ping timeout: 260 seconds]
isHavvy has joined #nixos
<magnetophon>
Hunterkll: so, no pb: if you want a quick test, tel nixos-rebuild to not write a new grub, and if you wanna reboot the test, tell it to write one. when you are done you can delete the generations that you don't want and reclaim the space. Wouldn't that have the same advantages as what you are describing?
<Hunterkll>
Hm, let me process that for a *second* but
<magnetophon>
you can manually delete generations by number, or automatically everyting but the last X days, or just everything but the last.
<Hunterkll>
magnetophon, i can see one issue that might not be addressed - split off the new BE's config and save it, then deelte the originating BE
isHavvyGhosting has quit [Ping timeout: 260 seconds]
<magnetophon>
Hunterkll: why would that be an issue?
<Hunterkll>
but i don't want to do a rebuild
<magnetophon>
why not?
<Hunterkll>
I want to BE clone, then reboot, select new one, and play around, then reboot back into original
<Hunterkll>
might take me all of a minute for the test i want to do
<Hunterkll>
i might be missing something that'll make it easy here, of course
<symphorien>
A rebuild does not mean recompiling the world
<Hunterkll>
I can also save a BE and keep it frozen
<magnetophon>
I don't see the problem, could you explain
<symphorien>
It can be a question of seconds
<Hunterkll>
yea, see, i'm trying to work it out in my head right now
<sphalerite>
edit config, nixos-rebuild boot, reboot, reboot again, this time selecting the previous generation, undo changes to config, nixos-rebuild boot
<Hunterkll>
under solaris, i don't have to edit anything
<magnetophon>
Hunterkll: and frozen is the only state nios knows
<Hunterkll>
and i could potentially have 20 BEs from one starter
<Hunterkll>
and i can even delete the starter, and still boot just fine
<Hunterkll>
maybe I am not grasping the Nix process properly , but it seems similar to powershell DSC ... but with a lot more flexibility
Diogo has joined #nixos
<magnetophon>
that's no problem in nios either: you can delete any generation you want and the rest will still be there.
<symphorien>
Hunterkll: on nixos, changing the system configuration means creating a new BE
<symphorien>
So you can havd hundreds
<magnetophon>
they are independent, but still share disk space
<symphorien>
That's the default mode of operations
<Hunterkll>
OK, that does make sense
<Hunterkll>
i'm asuming it's using either BTRFS or ZFS tricks to make that apparent?
<sphalerite>
no, it's using the nix store
<sphalerite>
independent of which file system you use
sbetrium has joined #nixos
<Hunterkll>
ah, see, in Sol I can bang up everything, even critical system directories
<symphorien>
The only thing you have to take care of is things like databases
<symphorien>
And the content of /home
<Hunterkll>
i guess thinking more like custom /dev stuff, kernel modifications, etc
<Hunterkll>
in a BE i can maintain custom kernel modules and injections
<sphalerite>
Hunterkll: you can `rm -r /etc /dev` on nixos too and it should still boot the next time
<Hunterkll>
to the point i can be running a 10 year old kernel
<Hunterkll>
inside the BE
<symphorien>
Kernel configuration belongs to nixos textual configuration so can be rolled back whenever
<Hunterkll>
OK
<sphalerite>
(discarding stateful /etc stuff like ssh host keys, but still)
<symphorien>
You can even track it in git for extra convenience
<Hunterkll>
sphalerite, that's stuff i'd be interested in having preserved
<sphalerite>
Hunterkll: then use zfs and snapshots :)
<Hunterkll>
sphalerite, but i already am! :)
<sphalerite>
Hunterkll: (or btrfs) but nixos's system generations are orthogonal to that
<Hunterkll>
Well, i am seeing that a lot of evolution has happened in the past few years - it may not be equivalent, but there's stuff here that means linux *may* be viable for our use case
<Hunterkll>
We'll still need Solaris for storage (SMB implementation) but there seems to be a lot of differences from 3 years ago
<magnetophon>
Hunterkll: everything we've discussed has been working on Nixos for many years
<sphalerite>
I guess another way of looking at it is this: nixos lets you build your system and its config from a declarative expression, meaning that you don't need BEs in order to recover breaking your system software — you can roll back _above_ the filesystem level, or just rebuild it completely
<Hunterkll>
sphalerite, ah, we don't use BE's as recovery, but as experimentation land - it provides a 100% true lab environment to test in
<Hunterkll>
and we can export them and import to another system on a different newtork
<magnetophon>
Hunterkll: no problem for nix either
<Hunterkll>
magnetophon, even files in home directories, or custom /opt stuff?
<symphorien>
Not home, but you can package custom /opt stuff properly with nix
<magnetophon>
what he said! ^^ :)
<Hunterkll>
magnetophon, well, i'm not exactly a Nix veteran, but everything i've seen makes me want to investigate it :)
<Hunterkll>
is there a commercial support organization?
<sphalerite>
huh, where did the support page go…
<Hunterkll>
symphorien, see, what I want from a BE split is *everything* except mounts. so anything in the BE's filesystem .... regardess of where/why it is, including commercial applications
<sphalerite>
there's no one canonical support company, but several who offer support for building what you need with nixos
<Hunterkll>
if i need to split a test environment, it should be 100% minus database mounts
chagra has joined #nixos
<Hunterkll>
sphalerite, yea, currently we use RHEL and OEL
<Hunterkll>
i figure if there's something close to what I do with solaris (which we also use) and powershell DSC it's worth looking into
sbetrium has quit [Quit: sbetrium]
<magnetophon>
Hunterkll: not sure if I understand: you want fstab untouched, right? that's easy: don't change it's config. We already told you nix will ignore home (but there are tools for that)
<magnetophon>
everything else is under nixos's control.
<Hunterkll>
magnetophon, the end goal is that I run 'be clone' essentially, reboot, and grub now has two independant entries
<sphalerite>
Hunterkll: I'll phrase it this way: nixos doesn't provide exactly what you want
<sphalerite>
Hunterkll: but maybe you'll end up wanting what nixos does provide :p
<Hunterkll>
sphalerite, OK. but from what you've told me, it's close. so that warrants further investigation
<Hunterkll>
sphalerite, because a powershell DSC-like linux system would be .... useful, at least
<magnetophon>
Hunterkll: it sure is close. you just gotta wrap your head around the immutable and declarative aspects
<Hunterkll>
but i have a yoke of regulatory compliance, lol
<magnetophon>
Hunterkll: where are you working, if I may ask?
<Hunterkll>
magnetophon, yea, like i said, it's like DSC but a bit more ... down to earth?
<Hunterkll>
magnetophon, F100 defense contractor
<Hunterkll>
er, not down to earth, but bare metal
<Hunterkll>
magnetophon, this should be enough to give you an idea - we put captain james kirk in a ship ;)
<Hunterkll>
but beyond that, this has given me a LOT to think about. I'm still going to push for SLES instead of RHEL and OEL, but since there's organizations that can provide contract support, that makes it an option for me
<Hunterkll>
one of the things I have to have is vendor support and patch timelines
<sphalerite>
Hunterkll: that's still a bit lacking in nixos. We have our releases, with ~7month support periods, and no LTSes currently!
<Hunterkll>
sphalerite, hence my question about commercial support :)
<sphalerite>
Hunterkll: you won't currently get LTS from any of the companies listed there either, there just isn't anyone with enough customers paying enough to finance long-term maintenance.
<sphalerite>
Hunterkll: we've discussed offering LTS at Mayflower in the past, but haven't come up with a model that would finance our maintenance of an LTS release yet.
<Hunterkll>
sphalerite, ah, see, we pay red hat and suse for long term stuff because of our out of cycle instlls
<Hunterkll>
I suppose ya'll aren't quite *mainstream* enough yet :)
<sphalerite>
yep, I understand why LTS is a thing :) we just don't have a red hat or a suse with the resources to maintain an LTS
<Hunterkll>
oh i totally understand
knupfer has quit [Ping timeout: 265 seconds]
<Hunterkll>
I definitely will be trying this at home
<Hunterkll>
if i can get to 75% of how I use BE's at work, that'll be something that'll be worth pitching to ditch more oracle products
<sphalerite>
yep, I can definitely recommend trying it at home :D
<Hunterkll>
I really do thank you all for the information though! I hadn't heard of NixOS before it was mentioned in a conversation i was having about solaris :)
h0m1 has quit [Quit: WeeChat 2.7]
<Hunterkll>
You still can't beat Sol 11.4's SMB, but that's because Samba is behind the times ;)
infty has joined #nixos
h0m1 has joined #nixos
<sphalerite>
how so? (I think this is a good point in the conversation to move to #nixos-chat though since it's not about nixos anymore)
zaeph has quit [Ping timeout: 260 seconds]
gour has left #nixos ["Leaving"]
<{^_^}>
[nixpkgs] @Ma27 pushed 2 commits to oraclejdk8u241: https://git.io/JvsOb
<{^_^}>
[nixpkgs] @Ma27 pushed 0 commits to oraclejdk8u241: https://git.io/JvsOA
M-Gregoire49 has quit [Remote host closed the connection]
<tilpner>
gchristensen has/had an elaborate STT setup involving a Windows VM, he can probably tell you what not to try too
o1lo01ol1o has joined #nixos
Yannik_Sc has quit [Ping timeout: 265 seconds]
<ixxie>
tilpner: that already sounds like more work to setup than my hand can muster :(
<ixxie>
but ill ping him if hes in the hood, so thanks
Chiliparrot has joined #nixos
<wedens[m]>
I use nix-serve on my desktop as a binary cache for a laptop, but it still takes ages to do channel updates on the laptop even though it fetches everything from cache. it takes less time to fetch from the default caches. what might be the problem?
Jackneill has joined #nixos
chloekek has joined #nixos
coco has joined #nixos
<raboof>
should the homepage in the 'meta' section be quoted or not? a quick grep suggests it's unquoted in 9933 cases but quoted in 2626 cases...
M-Gregoire49 has joined #nixos
fendor has joined #nixos
<tilpner>
wedens[m]: What's the content of /nix-cache-info?
<wedens[m]>
I'm currently trying to check if local network is ok. both computers are on wifi :(
<tilpner>
wedens[m]: That might do it. There is no compression with nix-serve, so you need fast local links
<tilpner>
wedens[m]: The compression ratio for hello is 5, so the cache needs to read only a fifth of the data from disk that your desktop needs to read from disk
<tilpner>
wedens[m]: If you desktop has spinning metal drives, that doesn't help either, especially with concurrent requests
<tilpner>
If your internet connection is fast, and your desktop is connected wirelessly and serving from slow drives, it seems reasonable that it would be slower
coco has quit [Quit: WeeChat 2.6]
coco has joined #nixos
kenjis1 has joined #nixos
<wedens[m]>
there are only ssds on the desktop
<tilpner>
Well, that doesn't help my argument c.c
coco has quit [Client Quit]
<tilpner>
But do a throughput test for both anyway, together with (lack of) compression it could explain the speed difference
<wedens[m]>
I'd gladly just change cache's priority so that it serves only something that is not in the default cache
<sphalerite>
there's a fairly recent nix change that allows setting it client-side as well
coco has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<wedens[m]>
nice. thanks!
CMCDragonkai1 has joined #nixos
<tilpner>
I don't know how I feel about it being encoded as a query parameter, but I agree client-side priorities are better here
<wedens[m]>
is it released?
<tilpner>
No
<tilpner>
Hmm, there was a release 28 days ago
<tilpner>
But it doesn't seem to include this commit
<pie_[bnc]>
28 days later
<wedens[m]>
does `nixos-rebuild` respect NIX_CONF_DIR ?
<aveltras>
how do people usually handle nixops state and secret credentials ? I'd like to keep all the projects self-contained to get maximum reproducibility. I've read some use git-crypt to encrypt a directory in the project with all credentials and nixops state. Is there any obvious drawback with this method ?
<srk>
aveltras: sqlite database appears as blob to git/git-crypt
<srk>
there were some efforts to have a text-based backend for nixops that would make it better
gnidorah has quit [Quit: Connection closed for inactivity]
coco has quit [Quit: WeeChat 2.6]
<aveltras>
srk: what do you mean ? i don't get the implications of the state being appearing as a blob
<srk>
well you either add it to commit or don't but you won't see any meaningful diff out of it. it's more of a comsetic issue as it screws your git history a bit
<srk>
whether to add it to commit depends on what changed and you need to be aware of that, e.g. if it's only toplevel path which changed you can safely skip commiting these
<aveltras>
what i mean is i'd like the nixops state to be shared between people working on a project but without exposing credentials if the project is in a public repo
<srk>
yes you can do that with git/git-crypt, all of the above is still true tho
<{^_^}>
nixops#865 (by mogorman, 1 year ago, open): Key/Value backend
iMatejC has quit [Quit: WeeChat 2.7]
iMatejC has joined #nixos
<aveltras>
srk: what you mean is that the nixops state can be modified with irrelevant changes (like maybe timestamps or sth like this in the sqlite db) and commiting those would be noise in the git history ? if so, that's fine for me, im only concerned about security for now
<srk>
aveltras: yes, pretty much
<aveltras>
srk: ok i get it then :) thanks for the explanation
ixxie has quit [Ping timeout: 265 seconds]
m0rphism has joined #nixos
<wedens[m]>
anyone experienced problems with dnsmasq (I run it separately, bot via NM) after resume? It doesn't respond even when I restart the service
iMatejC has quit [Quit: WeeChat 2.7]
o1lo01ol1o has joined #nixos
iMatejC has joined #nixos
dm9 has quit [Quit: WeeChat 2.7]
CMCDragonkai1 has quit [Ping timeout: 268 seconds]
<tilpner>
wedens[m]: Was all this triggered by the "my cache substitutes slowly" thing?
<wedens[m]>
tilpner: no, unrelated
<tilpner>
Did you get any numbers for that?
bhipple has joined #nixos
<wedens[m]>
my network is likely to blame for slow substitute. somwthing fishy is going on. speed jumps, packet loss, ping jumps, etc
<evanjs>
Okay, trying to make a tex config module atm. if I have `inherit (pkgs.texlive) scheme-full collection-langjapanese algorithms cm-super;` and I want to move the packages out, how do I pull them into the statement with inherit?
<evanjs>
the texlive packages end up trying to reference files that don't exist and etc. I'm pretty sure it's due to the way I translate inherit (pkgs.texlive) scheme-full collection-langjapanese algorithms cm-super; to `inherit (pkgs.texlive) <packages>`
<evanjs>
the end goal: I want to make it so that, if I do want to override the tex packages, or even just the scheme package from another part of my config, I don't have to do texlive.combine or etc all over
nikola_i has joined #nixos
nikola_i has quit [Max SendQ exceeded]
nikola_i has joined #nixos
nikola_i has quit [Max SendQ exceeded]
elux has joined #nixos
<elux>
hi there
nikola_i has joined #nixos
nikola_i has quit [Max SendQ exceeded]
<elux>
does anyone here use bluetooth on their systems? hehe, i imagine so. how do i actually get to check the manufacterer name of a MAC address from bluetooth? how does anyone ever pair anything without any hints from the tools
<elux>
i just see a list of devices by MAC address
nikola_i has joined #nixos
nikola_i has quit [Max SendQ exceeded]
<globin>
wedens[m]: we are aware of that, the current 99-main will be removed at some point
<evanjs>
Okay, drawing a blank again, how do I remove packages like node-_at_angular_slash_cli via nix-env? path is `nodePackages."@angular/cli"`
<evanjs>
node-_at_angular_slash_cli is also what shows up via tab completion :D
erictapen has joined #nixos
chloekek has joined #nixos
Addison[m]1 has left #nixos ["User left"]
noudle has joined #nixos
gkmngrgn has quit [Ping timeout: 265 seconds]
<evanjs>
well... I guess `nix-env -e <store path>` works :D though I really do wish there was a better way to handle packages with unique names like "@angular/cli", etc
<mkg20001>
I have a python app that requires wrapGAppsHook
<mkg20001>
Building it works fine, but what do I need to do so I can launch it in nix-shell aswell?
<WhatisRT>
I'm trying to setup a headless rpi 3 with nixos. Almost everything works fine, but for some reason uboot waits until I have connected a monitor via HDMI
<WhatisRT>
I've tried lots of different configurations, but that issue persists and seems to be documented nowhere. Does someone here have an idea what might be the issue?
Chiliparrot has quit [Ping timeout: 265 seconds]
Chiliparrot has joined #nixos
WhatisRT has quit [Remote host closed the connection]
WhatisRT has joined #nixos
yvan-sraka has joined #nixos
<evanjs>
mkg20001: hold on, I added some stuff to the openrazer docs showing how it can be used in nix-shell, might be useful
<aveltras>
anyone using nixops with virtualbox on a nixos unstable host ? I have an error "Error: Failed to become guest control master: VERR_VERSION_MISMATCH" when trying to deploy to virtualbox
<{^_^}>
[nixpkgs] @FRidh pushed commit from @bhipple to staging « waf: 2.0.18 -> 2.0.19 »: https://git.io/Jvs0z
<monadplus>
Then, here is my question: discord client doesn't open links on click and I am trying to set the default browser but I haven't found how to do this. I took a look at discord/base.nix but I couldn't find anything related to the default browser. Does anyone has fixed this ?
lord| has quit [Read error: Connection reset by peer]
<clever>
astk17: you only need nix, so you could just use the curl script to install it
<astk17>
thanks
<clever>
if its a single-user install, you can ssh into that user, instead of root
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
<astk17>
yeah, i think ssh needs a .pem keys file aswell
erictapen has quit [Ping timeout: 268 seconds]
<astk17>
is there a similar option (its -i for ssh) to link these certificate keys
<astk17>
to nix-capy-closuer
<astk17>
copy*
<astk17>
closure*
<astk17>
and does the AWS need eg. nix-env --install cabal2nix
<astk17>
nix-prefetch-git and cabal-install
<astk17>
which are needed on the host to use create the exe
<astk17>
well, maybe not git
cosimone has joined #nixos
<astk17>
nix-env --install cabal2nix
<astk17>
GC Warning: Failed to expand heap by 8388608 bytes
<astk17>
maybe it will work without
<astk17>
yeah, and it defiantly cant connect without the keys file: Permission denied (publickey,gssapi-keyex,gssapi-with-mic).
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<judson>
I'm noticing that passExtensions all seem to have a file conflict on lib/password-store/extensions - additionally I get an error from builder.pl about different permissions. Is this a known issue?
o1lo01ol1o has quit [Remote host closed the connection]
astk17 has quit [Remote host closed the connection]
<sondr3>
how usable are flakes as of now? I've been meaning to take a look at them once they had some time to brew
<gchristensen>
they're not merged in to Nix
<qyliss>
they work, but the RFC is still open and anything about them is therefore subject to change at any time
<sondr3>
so in an alternative universe where they are in Nix, how usable are they? :P
o1lo01ol1o has joined #nixos
<sondr3>
but fair, I've just seen a few mentions of flakes pop up in nix-community so figured it had become at least experimental
<qyliss>
you can use it if you want
<qyliss>
pkgs.nixFlakes
<qyliss>
but i don't know what you mean by "become experimental"
<sondr3>
merged into nix but not become mature
<qyliss>
that's a long way off
<qyliss>
at least at the current pace
<gchristensen>
it is?
<qyliss>
I would say so
doyougnu has joined #nixos
<qyliss>
the current RFC only specifies the flake format, which is a pretty small part of the overall implementation, and it hasn't moved in a while
<gchristensen>
what about the implementation present in the flakes branch? what if that merged as-is?
o1lo01ol1o has quit [Ping timeout: 260 seconds]
MmeQuignon has quit [Ping timeout: 268 seconds]
fragamus has joined #nixos
<qyliss>
I don't know much about the implementation in its current state
<gchristensen>
the nixos infra is using it for deployments
<qyliss>
But the reason the RFC was scaled back was that there was a lot of discussion still needed about the interface, etc.
<gchristensen>
yeah
<{^_^}>
[nixpkgs] @pulsation opened pull request #79111 → fldigi: update fltk dependency to 1.4 → https://git.io/Jvsud
<qyliss>
And the last shepherding meeting we had was in october
justan0theruser has joined #nixos
justanotheruser has quit [Ping timeout: 265 seconds]
<qyliss>
there's apparently been "work ongoing" on the RFC since then, from what I can tell from RFCSC minutes.
<qyliss>
so all things considered i think we're a long way off
fragamus has quit [Ping timeout: 240 seconds]
<gchristensen>
qyliss: I'd like it if you tried out the CLI as is
<qyliss>
I have, a bit
<qyliss>
Or at least I did a while ago
<qyliss>
but no matter how well it works, I still think it's a long way from being merged.
<gchristensen>
okay
<{^_^}>
[nixpkgs] @dtzWill opened pull request #79112 → kdeApplications: 19.08.3 -> 19.12.1, new fetch base URL? → https://git.io/Jvszk
<lovesegfault>
I have a meson build that also needs cmake, but when I add cmake to nativeBuiltInputs it start building the project as if it were a CMake, and not a Meson, one
<lovesegfault>
How can I add CMake to the nativeBuildInputs without having it coopt the entire build strategy
<evanjs>
hrm. any ideas on where the 78 user packages neofetch is telling me exist might be? Nothing via nix-env -q :P
<Yaniel>
lovesegfault: override configurePhase
<lovesegfault>
Yaniel: for my package or for CMake?
<Yaniel>
for your package
<Yaniel>
hmm, do cmake and meson have variables for their default phases?
<Yaniel>
that would be nice
<Yaniel>
so you could do stuff like configurePhase = cmake.configurePhase;
<tilpner>
evanjs: Does this match? nix-store -q --requisites "/run/current-system/sw" | wc -l
<tilpner>
evanjs: nix-store -q --requisites "$HOME/.nix-profile" for users packages
<evanjs>
tilpner: yes on system
<lovesegfault>
How do I find out what meson was doing in it's configurePhase?
<tilpner>
evanjs: But really, vim $(nix-build --no-out-link '<nixpkgs>' -A neofetch)/bin/neofetch +/nix-store
<astk>
but then it complains on writing ssh-add key.pem, that It is required that your private key files are NOT accessible by others. This private key will be ignored.
<flameslicer>
Is there a good way to tie my revision to my version number in a .nix derivation? Or will I just be stuck grabbing the revision # from github?
<evanjs>
tilpner: oh XD that's what you meant. gah instantiating stuff like that is nice
<astk>
Permissions 0777 for 'kp1.pem' are too open.
<astk>
is that a chmod thing?
<evanjs>
tilpner: I just found it odd as one computer has 2 and the other has 78 users packages, and a large portion of the configs are shared. hrm
<astk>
seems to be, if i write chmod 755 then it shows up in the error
<evanjs>
okay whoah I see something now. stuff outside of hm. what is this doing here
<astk>
anyone the chmod value for keys for ssh-agent
<evanjs>
WOW. that was _all_ nixpkgs-review??
<Yaniel>
lovesegfault: you probably have to dig into the meson package, possibly even the build tools nix files
<Yaniel>
or you can try your luck with the manual/wiki
<astk>
700 seems to do it...
<evanjs>
anyway! user packages match across systems now, and I did not know I could do that instead of `nix edit`ing (esp with e.g. +/nix-store)
<evanjs>
tilpner++
<{^_^}>
tilpner's karma got increased to 62
<AmandaC>
astk: 600 would probably be best, you don't really need to execute them
<astk>
thanls
<astk>
thanks*
<astk>
now when i do nix-copy-closure it says; bash: nix-store: command not found
<astk>
which at least is not a keys error, i think it might have worked
<astk>
i guess the nix-store error is on the remote machine side
<tilpner>
evanjs: Also useful, cd /tmp; tar xvf $(nip sudo.src), where nip is an alias for nix-build --no-out-link '<nixpkgs>' -A. Makes looking at the actually used sources quick, instead of finding random git repos at different revisions
<astk>
since it is available on my local machine...
<evanjs>
tilpner: was literally brainstorming easier ways to do this today :O
<evanjs>
tilpner++
<{^_^}>
tilpner's karma got increased to 63
<clever>
astk: when you run ssh-agent, you must set the env vars it prints, to the values it prints
<astk>
so for example how would i set the nix-store ?
<clever>
astk: the .bashrc on the remote end, must put ~/.nix-profile/bin/ into $PATH
<astk>
oh, so just like; export PATH=~/.nix-profile/bin:$PATH
<clever>
yeah
<{^_^}>
[nixpkgs] @dtzWill opened pull request #79117 → ktimetracker: init at 5.0.1 → https://git.io/Jvsgt
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « socat: 1.7.3.3 -> 1.7.3.4 »: https://git.io/Jvsgs
<jtojnar>
lovesegfault also Meson supports cmake as a method of finding dependencies but unless you need a dependency that does not provide pkg-config definitions, cmake is not necessary
<flameslicer>
Is there a good way to tie my revision to my version number in a .nix derivation? Or will I just be stuck grabbing the revision # from github?
judson_ has joined #nixos
nek0 has joined #nixos
<astk>
so im still getting a "bash: nix-store: command not found" error...
<infinisil>
Flameslicer: lib.commitIdFromGitRepo
<infinisil>
Gives you the revision for whatever git-tracked directory you give it
<flameslicer>
Thanks
judson has quit [Read error: Connection reset by peer]
<astk>
oh, there is more error message, it also says; error: cannot connect to 'ec2-user@ec2-1-2-3-4.us-east-2.compute.amazonaws.com'
<clever>
astk: you can test that .bashrc was done right with `ssh user@host nix-store --version`
<astk>
but the adding of the key using ssh-add defiantly works because now i dont need the -i option to ssh
<clever>
astk: bash will run different files for login and non-login shells, so you must have ssh directly run nix-store
<flameslicer>
Though shouldn't I be able to just run ${version} from within my fetchfromgithub?
<astk>
yeah, that doesnt work
<astk>
bash: nix-store: command not found
<astk>
so its using a different path if i just echo to add it to the path
<astk>
were you meaning that .bashrc should be modified directly?
shibboleth has quit [Quit: shibboleth]
<aterius>
I'm a little confused about UEFI/grub, I'm moving my nixOS install from my VM to workstation which currently has fedora and windows on it, using UEFI, but managed via grub. I wanted to keep the other OSes installed, however, when I mount /boot to the EFI partition (which is 100 mb) there isn't enough space, and when I mount it to the /boot partition fedora creates (which mounts the EFI partition to /boot/efi) it fails to
<aterius>
detect it as a valid boot partition...
<astk>
hmm there is a warning in /etc/bashrc not to change it directly!
joshmeredith has quit []
<clever>
astk: try both .bashrc and .bash_profile, and maybe add an echo to them temporarily, to confirm they even run, and when
joshmeredith has joined #nixos
lally has quit []
lally has joined #nixos
<astk>
no, cant edit the readonly file...
Naptra has quit [Quit: Leaving]
<clever>
astk: are you editing the one in the home folder?
<{^_^}>
[nixpkgs] @bhipple opened pull request #79127 → stdenv: default unpack can unpack bzip2 as well → https://git.io/Jvsan
Naptra has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @volth opened pull request #79129 → oraclejdk8: remove dead code → https://git.io/Jvsa4
doyougnu has joined #nixos
[Leary] has joined #nixos
Lears has quit [Read error: No route to host]
<sondr3>
huh, I'm attempting to use niv to manage some overlays but the taffybar one just throws `fatal: '/nix/store/6nzd8hxc170384l68qrf7yqp7x4wkgpg-source' does not appear to be a git repository` when I attempt to import it
chloekek has quit [Ping timeout: 240 seconds]
<sondr3>
all I'm doing is `(import "${sources.taffybar}/overlay.nix")`
<clever>
sondr3: does overlay.nix do anything with builtins.fetchGit?
<evanjs>
sondr3: any reason you’re using the overlay for taffybar? Seems to work fine without on my end
<sondr3>
evanjs: funny you ask, I've been looking at your configuration trying to wrap my head around it
o1lo01ol1o has quit [Ping timeout: 265 seconds]
kraem has joined #nixos
<sondr3>
and yeah, it's marked as broken
<evanjs>
sondr3: yeah I know it’s marked as broken, but im able to build in spite of that. Sorta lazy and busy right now but hold on I’ll grab my config
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<bhipple>
astk: I've seen that error before, but I can't for the life of me remember what the solution was :/
<bhipple>
but if I recall, it may have had something to do with `nix-copy-closure` not using a login shell and sourcing the bashrc to setup PATH to find the remote nix (?)
<bhipple>
though since your ssh does it, that's probably not what's happening in this case
<gchristensen>
when you `ssh usr@host nix-store --version` are *any* bytes sent back over stdout beyond what nix-store --version prints?
<bhipple>
Ah yes, that was it! Something was echo'ing in the user's zshenv
<astk>
ok, it was echoing "hello" as a line in the ~/bashrc
<astk>
with it removed it says its copying paths
<astk>
i had to edit the ~/bashrc to add; export PATH=~/.nix-profile/bin:$PATH
<astk>
the echo "hello" was just to see that it was calling this when ssh logging in
<clever>
2020-02-02 17:14:57 < clever> astk: try both .bashrc and .bash_profile, and maybe add an echo to them temporarily, to confirm they even run, and when
<clever>
astk: i too have broken scp by leaving echo's like that in place
astk has quit [Ping timeout: 260 seconds]
sigmundv__ has joined #nixos
Rusty1 has quit [Remote host closed the connection]