<pie_>
has anyone looked at separating various stdenv phases into separate derivations to get greater granularity when packaging something incrementally
<clever>
pie_: ive seen chromium cripple hydra before, because there was a seperate unpackPhase derivation, and the source unpacked to over 10gig
<clever>
it was busy copying source around for days
Glider_IRC_ has quit [Ping timeout: 245 seconds]
<pie_>
clever, wah
freeman42x has quit [Ping timeout: 252 seconds]
drakonis_ has quit [Ping timeout: 248 seconds]
<pie_>
clever, maybe we should use more hardlinks >_>
<clever>
pie_: things like nix-copy-closure arent aware of hardlinks, and will still transfer dups around
<clever>
symlinks would be cheaper, but harder to create/manage
<clever>
it would be best if you can build from a read-only $src, but a lot of things dont play nicely with that
<pie_>
the last one was my first thought but yeah
freeman42x has joined #nixos
<pie_>
well, or moar overlay mounts
<pie_>
though i still find them sketchy
duoi has quit [Changing host]
duoi has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
<clever>
pie_: nix-store --optimize, will also hardlink all dups between things
<pie_>
clever, right but that only helps after the fact
ris has quit [Ping timeout: 260 seconds]
_e has quit [Quit: leaving]
ilikeheaps has quit [Ping timeout: 246 seconds]
_e has joined #nixos
drakonis has quit [Ping timeout: 264 seconds]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
drakonis has joined #nixos
ayerhart has quit [Ping timeout: 258 seconds]
ayerhart has joined #nixos
erictapen has quit [Ping timeout: 252 seconds]
drakonis_ has joined #nixos
buffet_ has joined #nixos
buffet has quit [Ping timeout: 244 seconds]
mabel has joined #nixos
drakonis has quit [Ping timeout: 252 seconds]
drakonis has joined #nixos
drakonis_ has quit [Ping timeout: 252 seconds]
hio has quit [Quit: Connection closed for inactivity]
ardumont has quit [Ping timeout: 258 seconds]
ardumont has joined #nixos
thc202 has quit [Ping timeout: 250 seconds]
Jetien has joined #nixos
simendsjo has quit [Remote host closed the connection]
jasongrossman has quit [Ping timeout: 252 seconds]
<pie_>
i dont suppose stdenv has some option that results in adding gcc flags to cmake
adamantium has quit [Read error: Connection reset by peer]
adamantium has joined #nixos
toppler has quit [Read error: Connection reset by peer]
toppler has joined #nixos
jasongrossman has joined #nixos
<lopsided98>
pie_: Not cmake specific, but you could use NIX_CFLAGS_COMPILE
MmeQuignon has joined #nixos
jasongrossman has quit [Client Quit]
<pie_>
lopsided98, tried that seems to break
<pie_>
but i googled some and ill probably just use a cmake specific variable
<pie_>
reading the changelog, " A new subcommand nixos-rebuild edit was added. " oh, cute
dsiypl4_ has quit [Ping timeout: 248 seconds]
jasongrossman has joined #nixos
mabel has quit [Ping timeout: 248 seconds]
phs^ has quit []
<dsx>
How to create GRE tunnels in NixOS?
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
alexherbo26 has joined #nixos
alexherbo29509 has quit [Ping timeout: 250 seconds]
alexherbo26 has quit [Client Quit]
alexherbo26 has joined #nixos
alexherbo268 has joined #nixos
MmeQuignon has quit [Ping timeout: 250 seconds]
jb55 has quit [Ping timeout: 268 seconds]
mabel has joined #nixos
alexherbo268 has quit [Client Quit]
alexherbo268 has joined #nixos
alexherbo26 has quit [Ping timeout: 250 seconds]
alexherbo2688 has joined #nixos
alexherbo2688 has quit [Client Quit]
alexherbo2688 has joined #nixos
alexherbo268 has quit [Ping timeout: 252 seconds]
alexherbo26886 has joined #nixos
jb55 has joined #nixos
drakonis_ has joined #nixos
drakonis has quit [Ping timeout: 248 seconds]
alexherbo2688 has quit [Ping timeout: 252 seconds]
alexherbo268866 has joined #nixos
alexherbo26886 has quit [Ping timeout: 246 seconds]
alexherbo2688667 has joined #nixos
init_6 has joined #nixos
<pie_>
huh, neat, "fish may be started in private mode via fish --private. Private mode fish sessions do not have access to the history file and any commands evaluated in private mode are not persisted for future sessions. A session variable $fish_private_mode can be queried to detect private mode and adjust the behavior of scripts accordingly to respect the user's wish for privacy."
alexherbo268866 has quit [Ping timeout: 246 seconds]
alexherbo2688667 has quit [Ping timeout: 250 seconds]
ardumont has quit [Ping timeout: 248 seconds]
adamantium has quit [Quit: Konversation terminated!]
drakonis_ has quit [Ping timeout: 246 seconds]
freeman42x has quit [Ping timeout: 252 seconds]
jb55 has quit [Ping timeout: 246 seconds]
drakonis_ has joined #nixos
wfranzini has quit [Remote host closed the connection]
<trevthedev>
aye, hello everyone. I'm trying to get Dwarf Fortress to run in text mode, but that requires me to hand edit the init.txt which is read only by NixOS
drakonis_ has joined #nixos
<trevthedev>
is there any way i can manage to hand edit that or modify the nix package to allow support for enabling text mode?
<trevthedev>
i want to use text mode so i can ssh into DF via my phone so i can play it when im away from my laptop haha
_e has quit [Client Quit]
_e has joined #nixos
<roconnor>
what sort of edits?
<trevthedev>
just want to change a line that's `[PRINT_MODE:TWBT]` to `[PRINT_MODE:TEXT]`
<trevthedev>
but it's in a read only file by Nix
<trevthedev>
and the package doesnt allow for enabling text print or not
drakonis has quit [Quit: Leaving]
Yaniel has joined #nixos
<lopsided98>
trevthedev: Have you tried putting it in ~/.dwarffortress/data/init/init.txt
<trevthedev>
that dir doesn't exist
<trevthedev>
init is in .local/share/df_linux/data/init/init.txt ^^
<jackdk_>
have you tried writing your own df derivation that does some sedding of init.txt?
<trevthedev>
no clue what options to add or change or anything
orivej has joined #nixos
<roconnor>
(pkgs.dwarf-fortress-packages.dwarf-fortress-full.override { enableTWBT = false; }) could maybe give you PRINT_MODE:2D ?
<trevthedev>
I entirely skipped over that option sorry, am rebuilding right now ^^
ardumont has quit [Ping timeout: 248 seconds]
drakonis has joined #nixos
drakonis_ has quit [Ping timeout: 258 seconds]
<trevthedev>
doesn't seem to work
<trevthedev>
changes printmode to 2D instead of text
<roconnor>
huge pain I know. If you are even more keen you can try to patch the nixpkgs expressions to add support for the mode you want. However DF is more complicated than your normal package.
<trevthedev>
hmmmm
<trevthedev>
might clone the package and add in the option myself and perhaps even push it? idk if that could be liable but idk
<trevthedev>
just need to edit lazy-pack.nix i think
<roconnor>
in your local copy of nixpkgs and build your hacked version.
<roconnor>
That's really a one-off approach. But may be good enough for whatever you want to do.
<trevthedev>
yeah
<roconnor>
I never could get into the DF game, though I've tried a couple times. Maybe it just isn't for me.
<trevthedev>
Ah fair enough haha
<trevthedev>
I've only been playing it on and off since october
<trevthedev>
It's pretty fun and I'd love to be able to pick back up on my phone whenever but there's no remote app
<trevthedev>
Found out i could just use screen and termux to just pop myself into it whenever though haha
<roconnor>
ha nice.
<trevthedev>
alright so i just run nix-build on this? i don't really remember sadly sorry
<trevthedev>
it's been a while since i wrote a package, was a Daggerfall Unity package i gave up on and used a shell XD
<roconnor>
so you've check out and edited a local copy of the nixpkgs repo?
<trevthedev>
or sorry, an environment(via nix-shell)
<trevthedev>
yup
<trevthedev>
edited it to where instead of print_mode:2D it's TEXT ^^
<roconnor>
you can do nix-build -A dwarf-fortress-packages.dwarf-fortress-full in the nixpkgs directory.
<trevthedev>
alright thanks
<roconnor>
that will make a result symlink to the nix-store
<trevthedev>
nice
<roconnor>
you can use the "-o" to name the symlink or put the symlink somewhare particular.
<trevthedev>
so now when i edit my config.nix declaratively for Dwarf Fortress it uses this instead of nixos.org?
<roconnor>
nope
<trevthedev>
probably worded that poorly but hopefully you getwhat i mean haha
<trevthedev>
ah
<roconnor>
this is really one-off.
<trevthedev>
fair enough
<roconnor>
nix-build uses the default.nix file of your local directory.
<trevthedev>
ahh
<roconnor>
since you are in your own nixpkgs it is using the nixpkgs that you've checkout for this command.
<trevthedev>
ye
<trevthedev>
makes sense then
<roconnor>
so nothing else about nixos, etc. is affected.
<trevthedev>
fair enough yeah
orivej has quit [Ping timeout: 268 seconds]
<trevthedev>
so once this is done building it shouldjustwork:tm: right?
<roconnor>
when you get your symlink, it will automatically be registered as a garbage collector root, so it won't be deleted until you delete the symlink.
<trevthedev>
alright
<trevthedev>
is still building haha
<roconnor>
however you should be able to nix-env -i "the-symlink-file" to add it to your user profile, if you manage your software that way.
<trevthedev>
i do it declaratively, but i can do that yea
<roconnor>
well I'd be hesitant to mix declariative profiles with imperative profiles.
<trevthedev>
hmm
<roconnor>
I might just nix-build -o ~/dwarf-fortess and run the program from there by hand.
<trevthedev>
oh dear
<trevthedev>
didn't compile
<roconnor>
oh ya. wanna post your git diff?
<roconnor>
(assuming you used git to checkout the nixpkgs)
<trevthedev>
only 2 differences is my daggerfallunity package and the one line changed from default.nix
<roconnor>
did you checkout the master branch of nixpkgs?
justanotheruser has quit [Ping timeout: 252 seconds]
m0rphism has joined #nixos
<isaac1>
ivan: I don't understand what you are suggesting
<ivan>
1) if the installer doesn't boot with UEFI, your own UEFI install is unlikely to work
<ivan>
2) try a newer BIOS for UEFI or give up and use CSM/MBR
<isaac1>
what is UEFI and why should I care?
<isaac1>
does it matter after boot?
<isaac1>
or is it just a nicer way to boot?
tsrt^ has joined #nixos
<yuken>
EFI is newer than BIOS, offers faster boots times and is pretty much just better than BIOS.
<ivegotasthma>
hello, how do I get the fix for the firefox plugins?
<yuken>
Except it has even more control over your system and can do even more, which is bad if IME turns out to be an Intel baackdoor ;^)
<ivegotasthma>
the studies can't be installed on the firefox build from the nix cache
<isaac1>
yuken: I think I am going to go with legacy then
<isaac1>
If it wont boot of usb in uefi
<isaac1>
then it's not worth the pain
<yuken>
Avoid legacy if you can.
<yuken>
How did you write the image to your flash drive?
<isaac1>
yuken: cp nixos.iso /dev/mydisk
<yuken>
Well that's probably why it didn't work.
<isaac1>
yuken: ??
<yuken>
Make a FAT32 partition anywhere on your system - either on your HDD/SSD, a flash drive, whatever
<yuken>
mount the ISO somewhere, copy its contents to the partition, sync it, reboot.
<isaac1>
I am on my laptop
<isaac1>
is there a command for this?
<isaac1>
I have always used cp or dd for this
<yuken>
There isn't. Do you know how to make filesystems & use fdisk?
<isaac1>
I use gdisk
<ivan>
nixos installer should work fine when copied to a block device, no?
<yuken>
Didn't work with me, ivan. And that's how I handle all my EFI media anyway.
<yuken>
isaac1, use whatever you want, just make a FAT32 partition capable of holding the image on it. Then, mount your ISO or image for Nix, and copy the contents to this partition.
isaac2 has joined #nixos
<isaac2>
yangm97: I am back
<isaac2>
sorry
<isaac2>
how can I make a usb of nixos that will boot uefi?
isaac1 has quit [Ping timeout: 255 seconds]
<yuken>
Just make a FAT32 partition, aand dump the contents of the ISO onto there.
<isaac2>
so reformat the drive
<isaac2>
to fat 32
<isaac2>
mount the iso
<isaac2>
and cp the contents?
<yuken>
yep. then umount/sync is and it should work.
<yuken>
If your board is modern enough, it will just show up as a bootable device, aassuming it isn't in legacy mode obviously.
<edcragg>
hi, if i have the free choice to use either openssl or gnutls when packaging a project for nixos, are there any opinions on which i should choose?
<yuken>
and obviously, make sure you properly sync & unmount that
<yuken>
so you don't lose everything :P
<isaac2>
what do you mean by sync?
<isaac2>
I usually just unmount
<yuken>
unmounting will sync by default iirc
<yuken>
Because when you copy files to a device, it will finish faster than it actually copies files, caching them in memory in the mean time while it keeps on copying. Or something, that might not be the way Linux does it.
<isaac2>
yuken: it's not booting
<isaac2>
it's cut off
<isaac2>
but it says ...solinux.bin missing or corrupt
<yuken>
Did you unmount & make sure all data was synced? If so, then something is messed up somewhere, not 100% sure.
<isaac2>
yuken: I unmounted in gnome-disk
<isaac2>
maybe I didn't do it right
<isaac2>
recopying
<yuken>
Just sync & umount it.
<isaac2>
yuken: what is the command?
<yuken>
those are the commands.
<yuken>
umount to unmount it, sync to sync it. umount should sync it, but just in case.
<isaac2>
this is very interesting
<isaac2>
ok so it finished
<isaac2>
and I ran "sync"
<isaac2>
rebooting
<isaac2>
same
<isaac2>
...solinux.bin missing or corrupt
<tilpner>
I don't know why you're copying the ISO file-wise
<isaac2>
me neither
<tilpner>
Did a block-wise copy not work?
<yuken>
That's all any EFI board should need to boot, tilpner.
<tilpner>
Just cp nixos.iso /dev/sdx
<isaac2>
tilpner: that's what I did before
<tilpner>
And how did that go?
<isaac2>
it would not boot uefi off usb
<tilpner>
And did what instead?
_e has quit [Quit: Lost terminal]
<isaac2>
tilpner: it would drop me into bios
<isaac2>
I could boot legacy off the usb though
<tilpner>
Where you chose the USB stick as a boot device, right?
<isaac2>
yes
lambda-11235 has quit [Quit: Bye]
<tilpner>
Oh, I don't remember if I have secure boot enabled
<isaac2>
tilpner: I have secure boot disabled
_e has joined #nixos
<tilpner>
And you did a sync after the block-wise copy before you unplugged it?
<tilpner>
My devices never needed too much hand-holding for this, I can't help with device-specific workarounds. Can you boot other distros the same way?
<isaac2>
I am going to bed
<isaac2>
I have spent 5 hours now with the same error
<immae>
Hello there! What is the currently recommended way to write a derivation "with plugins"? An example derivation or a wiki-page is fine, it’s just to get started (I want to re-package nextcloud but add plugins ("apps") support to it)
Dema has joined #nixos
<Dema>
Hi! Is there any beautifier/formatter for .nix files? I couldn't google any.
<tilpner>
adamantium: autoOptimiseStore just passes auto-optimise-store to Nix, while optimise.automatic calls nix-store --optimise once a day by default
<tilpner>
adamantium: Either will work, and you shouldn't need to set both. I have only autoOptimiseStore set
ixxie has joined #nixos
gagbo has joined #nixos
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<tilpner>
klntsky: That depends on what you want to build with it. Instantiating systems is going to need more memory than instantiating hello
thc202 has joined #nixos
<tilpner>
klntsky: And the builds are going to need memory too. You would limit the number of concurrent builds to one, and then you need to consider if you want to build llvm, chromium, etc.
<edcragg>
i'm having trouble building a package for which the configure script has a `usr/bin/env perl` shebang. i've included perl in buildInputs (also tried nativeBuildInputs) but i'm getting `./configure: /usr/bin/env: bad interpreter: No such file or directory` when building with nix-build. is there anything i can do about this?
<tilpner>
klntsky: 2GB might be possible, but it's not going to be fun
<tilpner>
edcragg: You could try to use patchShebangs
<tilpner>
Search nixpkgs for examples
<edcragg>
ah, this is something i was blissfully unaware of, thanks
<fex42>
hi there, I'm missing libgl.so in nixos 19.03 - is there an easy way to get it back into /run/opengl-driver/lib ?
<etu>
fex42: What are you trying to do? Compiling something or?
adamantium has quit [Quit: Konversation terminated!]
<fex42>
wanted to build Haskell Gloss with cabal and it fails with "Missing (or bad) C library: GL"
<etu>
,libraries fex42
<{^_^}>
fex42: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<ambro718>
fex42: NixOS now used libglvnd, a single OpenGL library that dispatches to the appropriate vendor library. You need to link to libGL which is part of the libglvnd package.
<{^_^}>
#40001 (by tiramiseb, 1 year ago, closed): libGL.so missing in the mesa-drivers package
<ambro718>
or better, use the "libGL" attribute which I think on Linux just points to libglvnd
<ambro718>
(it's not exactly the same things but a separate package with symlinks and some other stuff, anyway this is what stuff should depend on and link to)
jD91mZM2 has quit [Remote host closed the connection]
<fex42>
(I guess it's not cabal related, but on arch linux I could compile the gloss stuff with cabal, so I try to do the same with nixos now)
<ambro718>
fex42: just compiling things on NixOS doesn't work well, you almost always need to integrate with nix in some way so that it solves dependencies for you
rprije has quit [Ping timeout: 250 seconds]
orivej has joined #nixos
kshaa has joined #nixos
<kshaa>
Is there a way to add some extra output for nixops deploy? I.e. if I have some service which will auto-generate a key and I want to see it right in the nixops deploy log.
<eliasp>
struggling with overlay-based overrides as well... all I'm trying to do, is to use a different version for a pkg, e.g.: http://ix.io/1I62 - the expression is definitely evaluated during `nixos-rebuild build`, but it seems it doesn't result in anything being built. Am I doing something obvious wrong here?
<tilpner>
> xorg.xorgserver.src.override
<{^_^}>
attribute 'override' missing, at (string):253:1
<LnL>
eliasp: override is for dependencies / flags, you want overrideAttrs for this
<tilpner>
There is no such attribute. You are trying to re-call xorgserver with different arguments, and it doesn't take a src argument, so this doesn't errrr for you
<eliasp>
LnL: thanks a lot - I'll give this a try - makes ense
orivej has quit [Ping timeout: 252 seconds]
linarcx has joined #nixos
<LnL>
since it's a nested attribute you also want to make sure to extend the xorg set instead of replacing it
adamantium has quit [Quit: Konversation terminated!]
<eliasp>
ok, this is most likely what previously gave me weird issues of completely different pkgs (e.g. gtk+) not finding xorg components, such as libXinerama etc
<eliasp>
still, using your gist results in no build results... :/
<LnL>
if the references to xorgserver don't go trough the fixpoint nothing except the attribute itself will change
<eliasp>
hmm, so I'd have to check how xorgserver is referenced by other pkgs and then have my override follow this "path"?
<talqu>
hi, i a.nix and i install it to my system `nix-env -i $(nix-build a.nix)`. How should I include the same via my configuration.nix environment.systemPackages?
<pie_>
aswanson, i ran into something weird, if I use pyStuff = (python3.override { self = enableDebugging pkgs.python;} ).withPackages (p: with p; [ Keras tensorflow ]); instead of pyStuff = python3.withPackages (p: with p; [ Keras tensorflow ]); i get a different error https://bpaste.net/show/ed6d269724f6
<pie_>
im not sure if thats an improvement , but i dont know why enabling debugging should have changed that...
linarcx has joined #nixos
<pie_>
oh maybe i messed up the python version i passed...rebuilding
<eliasp>
LnL: could it be, that the problem I'm seeing is caused by the fact that I have `services.xserver.enable = true;` in my cfg and the corresponding module (nixos/modules/services/x11/xserver.nix) pulls in `xorg.xorgserver.out`, not `xorg.xorgserver`?
__monty__ has quit [Quit: leaving]
sb0 has quit [Quit: Leaving]
the-kenny has quit [Ping timeout: 258 seconds]
tboston_ has joined #nixos
ris has joined #nixos
hyper_ch has quit [Read error: Connection reset by peer]
<flokli>
yes, some tests seem to succeed, or at least get very far
<jakkn>
Hi, any advice on managing nixos config files? I've looked around at Google/reddit/HN and only found opinionated alternatives that required file copying and chowning and whatnoot. Any semi-official recommendations around?
o1lo01ol1o has quit [Ping timeout: 252 seconds]
tboston_ has quit [Ping timeout: 248 seconds]
<gchristensen>
jakkn: NixOS doesn't care where you put it. for al ong long time I kept it in /etc/nixos, now I keep it in ~/projects/nixos
<flokli>
but I don't really feel confident enough to start reading that code…
MmeQuignon has joined #nixos
buffet has joined #nixos
<flokli>
maybe try to find an easy repro and update an create a bug report upstream about it?
<ambro718>
ah it says "on host systems that run AMD CPUs", my CPU is Intel and the checkbox in the GUI is disabled even
buffet has quit [Client Quit]
mkoenig has joined #nixos
buffet has joined #nixos
<jakkn>
gchristensen: then you symlink those files to /etc/nixos/, correct?
cfricke has joined #nixos
<pie_>
is it sufficient to put a library in buildinputs for dlopen to work?
<pie_>
i think its not sufficient but i dont remember how this is supposed to work
MmeQuignon has quit [Ping timeout: 246 seconds]
<manveru>
ryantm: you around to talk about adding gem support to nixpkgs-update?
MmeQuignon has joined #nixos
<pie_>
ok i temporarily fixed something with ld preload.... nix packages need a dev mode that contains checks for dlopens that ultimately dont find a result :P
<ambro718>
flokli: so I am assuming that in these tests virtualbox runs without HW virt support, can you say that's true?
<ambro718>
and this is the reason that a 32-bit VM is started
<ambro718>
flokli: if you look closely at the error (https://paste.ubuntu.com/p/2tqFrXVrXt/) you see it happens just as the kernel is initializing the serial port
<ambro718>
I'm not sure over what channel those kernel logs from the guest are shown anyway
cosmo_ has joined #nixos
buffet has joined #nixos
<flokli>
ambro718: virtualbox vms are initialized with com port and proper console= params
englishm has joined #nixos
<flokli>
grep for vboxtestlog in nixos/tests/virtualbox.nix (and above)
buffet has quit [Remote host closed the connection]
npmccallum has quit [Ping timeout: 258 seconds]
<ambro718>
flokli: but /run/virtualbox-log-${name}.sock lives on the kvm machine right, not inside the vbox guest?
buffet has joined #nixos
<flokli>
ambro718: that's the file socket living in the nixos vm. on the other side, it's attached to the serial port of the virtualbox vm (which is running inside the nixos vm)
<ambro718>
vmFlags in createVM are plassed to qemu?
<flokli>
yes. sub createVM_… calls modifyvm with these flags
<flokli>
L219
<ambro718>
ah I see
<flokli>
commented out the uart options from vmFlags, and rerunning tests… let's see
<ambro718>
well it might not fix it if the uart hardware exists by default (not sure)
<flokli>
well, most of the things are failing again with that Details: code VBOX_E_IPRT_ERROR (0x80bb0005) error I posted on your PR
<ambro718>
I never seen this error
nD5Xjz has quit [Ping timeout: 248 seconds]
Ariakenom has quit [Quit: Leaving]
nD5Xjz has joined #nixos
shabius has quit [Quit: Leaving]
mvnetbiz_ has joined #nixos
<ambro718>
time to find a 32bit VM *somewhere* to try just running with SW virt, outside of tests
nD5Xjz_ has joined #nixos
nD5Xjz has quit [Ping timeout: 246 seconds]
fex42 has joined #nixos
jnetod has joined #nixos
dsiypl4_ has joined #nixos
cfricke has quit [Quit: WeeChat 2.4]
gnzh has joined #nixos
nD5Xjz_ has quit [Ping timeout: 252 seconds]
<jnetod>
how do I chroot into an arbitrary distro from nixos?
nD5Xjz has joined #nixos
<jnetod>
I have an existing fedora installation and I want to chroot into it.
<gnzh>
hello everyone, i have a small question. was trying to make swapfile work for hibernation with nixos, is it even possible? I did try to manually add resume=/var/swapfile and set remuse_offset in kernel options without any luck. I saw in init script that there is a section that cleans swap partition on boot, might this be cause of issues for me?
lightbulbjim has quit [Ping timeout: 264 seconds]
lightbulbjim has joined #nixos
__monty__ has joined #nixos
kshaa has quit [Ping timeout: 250 seconds]
jchia_ has joined #nixos
<jchia_>
which nixos setting can be used to set the console font size in text mode?
<jchia_>
gnzh: I tried Lat2-Terminus16 and Lat2-Terminus32 and there seemed to be no difference
<gnzh>
@jchia_, interesting, i never actually checked if this works :)
kreisys has joined #nixos
<gnzh>
@jnetod, to chroot afaik you need to mount dev proc sys and then it should work
ardumont has joined #nixos
<__monty__>
gnzh: Those @'s are super confusing. Makes everyone look like ops. IRC custom is to just not.
dsiypl4__ has joined #nixos
kreisys has quit [Ping timeout: 246 seconds]
<DigitalKiwi>
also @ is likely to trigger bots
dsiypl4__ has quit [Client Quit]
dsiypl4_ has quit [Ping timeout: 252 seconds]
fenedor has joined #nixos
kipari has joined #nixos
fendor has quit [Ping timeout: 245 seconds]
<eliasp>
jnetod: `systemd-nspawn -D /root/of/your/distro` - you can also "boot" it by adding the `-b` param
<jchia_>
gnzh: I suspect the problem is because I don't have Lat2-Terminus32 on my system or there is no such thing. I wonder how I can list all the available console fonts.
ixxie has quit [Ping timeout: 250 seconds]
<jchia_>
It's also unexpected that nixos-rebuild didn't give an error or warning about a missing font.
kreisys has joined #nixos
ma27 has quit [Quit: WeeChat 2.4]
isaac2 has joined #nixos
<AlexRice[m]>
Does anyone use spacemacs and use the nixos layer on it?
ma27 has joined #nixos
<infinisil>
,ask AlexRice[m]
<{^_^}>
AlexRice[m]: Just ask your question. It's the best way to know if anybody can help. Or look at other ways of getting in touch with the Nix community here: https://nixos.wiki/wiki/Get_In_Touch
<pie_>
AlexRice[m], i dont but ive been meaning to set it up, ping me if you get anything cool :)
<AlexRice[m]>
I guess my question was whether there was any magic you needed to add to configuration to get indentation working
<AlexRice[m]>
it currently wants to force no indentation anywhere in a nix script
<manveru>
AlexRice[m]: i use nix-mode, but indentation is all kinds of broken here as well
<manveru>
definitely no auto-indent :(
nD5Xjz has quit [Ping timeout: 245 seconds]
<AlexRice[m]>
manveru: that's a shame, I'm sure the mode claims to indent stuff nicely so I assumed I had just failed to turn it on
<jnetod>
gnzh: I've tried this already, no success.
<jnetod>
eliasp: systemd-nspawn worked out. thanks!
<manveru>
AlexRice[m]: yeah, it definitely has indentation support, it's just not very smart i think
<manveru>
no clue why
<manveru>
matthewbauer: ping :)
<AlexRice[m]>
manveru: just looked in the customize options, and there is a nix-indent-function option, I might give the option that's marked smart but buggy a go
nD5Xjz has joined #nixos
npmccallum has quit [Quit: npmccallum]
<manveru>
godspeed!
<isaac2>
I am back up
<isaac2>
err awake
<isaac2>
I still have an Maximus Code motherboard that I can't get to boot a nixos USB in UEFI
<jchia_>
"sudo nix-collect-garbage -d" cleans up stuff from the filesystem, apparently, but at boot time, I still have many generations and the early ones no longer work. (I get kernel panics if I selecet them, presumably because the necessary files have been deleted.) How can I finish the work of the nix-collect-garbage command and remove the options that no longer work?
<AlexRice[m]>
isaac2 is that an asus one?
<isaac2>
AlexRice[m]: yes
freeman42x has quit [Ping timeout: 258 seconds]
<AlexRice[m]>
isaac2: I had a similar problem with an asus motherboard but I did get it to work
<gnzh>
__monty__, sorry, my bad. havent used irc for couple of years now. slack took over in most of the communities
<manveru>
jchia_: the boot entries are removed on the next nixos-rebuild
Melkor333 has joined #nixos
<AlexRice[m]>
isaac2: unfortunately cannot remember what I did, but could look through irc logs if you want
<AlexRice[m]>
isaac2: is it giving an error to mount something then fails stage one problem
<isaac2>
AlexRice[m]: I wish
<jchia_>
manveru: I see, and I notice that if the config didn't change, nix-rebuild doesn't add a new generation.
<isaac2>
AlexRice[m]: isolinux.bin missing or corrupt
npmccallum has joined #nixos
<AlexRice[m]>
isaac2: Have you got a different computer you can try the usb on?
<qyliss>
Where does nix-shell get its environment variables from?
<AlexRice[m]>
My only guess is that it had some sort of extra kernel module that my motherboard needed
kisik21 has quit [Remote host closed the connection]
<AlexRice[m]>
Your error sounds different but it is probably worth a try
<AlexRice[m]>
You can always change back to stable channel once installed
<isaac2>
AlexRice[m]: how so if its a kernel module?
gnzh has quit [Quit: Leaving]
domogled has joined #nixos
<AlexRice[m]>
isaac2: Good point, I haven't really investigated as I use unstable anyway
mabel has quit [Ping timeout: 248 seconds]
<isaac2>
AlexRice[m]: well lets see if it boots
<AlexRice[m]>
manveru: from my 3 seconds of testing nix-indent-line seems a lot better
justanotheruser has quit [Excess Flood]
domogled has quit [Ping timeout: 250 seconds]
justanotheruser has joined #nixos
o1lo01ol1o has joined #nixos
endformationage has joined #nixos
alexherbo2 has quit [Remote host closed the connection]
<isaac2>
AlexRice[m]: ok it just booted the USB in UEFI
<AlexRice[m]>
isaac2: great
alexherbo2 has joined #nixos
<isaac2>
AlexRice[m]: so now I have the choice of running unstable on this build
<isaac2>
or accepting that the problem was nixos all along
<isaac2>
and using legacy and stable
<AlexRice[m]>
isaac2: it might be that stable with latest kernel works or something
<AlexRice[m]>
isaac2: or it could be only a problem with the install iso, I really have no idea
<isaac2>
AlexRice[m]: how would I obtain an install usb like that?
<isaac2>
AlexRice[m]: what is the risk of using unstable?
tboston_ has quit [Quit: leaving]
<AlexRice[m]>
isaac2: I don't think you would obtain an install image with stable and latest kernel, but you could install your system like that using the unstable usb iso
<isaac2>
AlexRice[m]: how? Do tell!
<isaac2>
AlexRice[m]: you want to help out? I can send you a one time ssh!
<AlexRice[m]>
isaac2: Just install it as you normally would
jnetod has quit [Ping timeout: 256 seconds]
<isaac2>
AlexRice[m]: wont that install unstable?
<AlexRice[m]>
isaac2: It might do but you could boot into the new system, change the channel and rebuild
<ajs124>
isaac2, building an iso like that is pretty easy, do you want a .nix file that builds one?
<AlexRice[m]>
or do that ^
drakonis has joined #nixos
<AlexRice[m]>
as for using a more recent kernel I believe you want `boot.kernelPackages = pkgs.linuxPackages_latest;
silver has quit [Read error: Connection reset by peer]
<AlexRice[m]>
` in your config
silver has joined #nixos
<AlexRice[m]>
also fwiw I couldn't get the archlinux iso to work either so I believe this is an asus problem not a nixos problem
<AlexRice[m]>
<isaac2 "Alex Rice: what is the risk of u"> The risk is that packages have a higher chance of breaking
countingsort has joined #nixos
countingsort has quit [Client Quit]
shabius has joined #nixos
jomik has joined #nixos
kuznero[m] has joined #nixos
camsbury- has quit [Remote host closed the connection]
npmccallum has quit [Remote host closed the connection]
<ambro718>
what kernel parameter to add to see full kernel output in the console?
<ambro718>
never mind
jomik has quit [Quit: WeeChat 2.4]
mabel has joined #nixos
o1lo01ol1o has quit [Ping timeout: 248 seconds]
o1lo01ol1o has joined #nixos
JosW has quit [Ping timeout: 258 seconds]
<talqu>
I have a.nix and I can install it into the system `nix-env -i $(nix-build a.nix)`. How can i reference this local file in configuration.nix, to get the same effect as putting the package name in environment.systemPackages?
<ambro718>
Can I change a 64bit system to 32-bit just with nixops, or will horrible things happen?
<immae>
or environment.systemPackages = [ (callPackage /path/to/a.nix {}) ]; depending on how your a.nix is written
<talqu>
thank immae
<talqu>
*thank you
<ambro718>
what do I set in nixops machine to make it 32-bit?
<ambro718>
I know I did this a while ago, it must be a simple option one sets to "i686-linux"
<immae>
talqu you’re welcome
o1lo01ol1o has quit [Remote host closed the connection]
mthst has quit [Quit: the bouncer died]
o1lo01ol1o has joined #nixos
mthst has joined #nixos
<fyuuri>
Hi. After upgrading to NixOS 19.03, my fish shell (which was upgraded to 3.1), does not load my fish_prompt anymore. It load the one user /nix/store/.... How can I change the bahaviour back?
o1lo01ol1o has quit [Ping timeout: 255 seconds]
<aswanson>
fyuuri: you can define the fish_prompt function in config.nix under `programs.fish.promptInit`
<aswanson>
configuration.nix, sorry for the abbreviation
Jetien has joined #nixos
MmeQuignon has quit [Ping timeout: 252 seconds]
<fyuuri>
@ aswanson thx! Did that change in the last upgrade?
<fyuuri>
Also, it does not load functions that are defined in .config/fish/functions/ before loading .config/fish/config.fish
<ambro718>
How do get the ssh public key generated by nixops so I can add it as an authorized key on the target machine
<jchia_>
After I "nixos-rebuild switch", "nix-collect-garbage -d" and then "nix-store --gc" as root, I see a directory "/nix/store/fwvfi3shxx71fzk5igwbqcynjbxf5ndy-terminus-font-4.47/share/consolefonts/" but terminus-fonts is not in my config's environment.systemPackages. How can I track down why this directory exists?
<jchia_>
Am I doing the garbage collection wrongly?
<ajs124>
jchia_, you can try using "nix why-depends"
<aswanson>
fyuuri: no idea how recent that change was, but I found it the last time I tweaked my prompt/greeting/aliases about 3 weeks ago
mthst has quit [Quit: the bouncer died]
fusion809 has quit [Remote host closed the connection]
<aswanson>
I share pieces of my configuration.nix between my desktop and laptop so having fish stuff defined there is really convenient for me
<fyuuri>
aswanson: That was after the 19.03 upgrade most likely
<fyuuri>
I have a config git repo... so for me its really annoying right now... I just wanted to do a quick upgrade..
<fyuuri>
Do you also have the rest of your fish config there? I have a quite long fish config file
<fyuuri>
Is there a list with all the things you can define for fish.* `
<simpson>
m1cr0man: Do you perhaps have a more specific question? Many folks want to know stuff like "where are all the packages?" or "can I get a list of what's in my profile?"
<m1cr0man>
simpson, "Where are all the packages?" is kind of what I'm wondering, but at a higher level. I've had this notion recently that the fact that most deployment strategies nowadays involve packaging an entire operation system to run a single binary is a bit barbaric. I'm wondering does NixOS (even partially) solve this problem in the way it structures packages and runs containers. My logic is that if 2 pieces of containerised software need the
<m1cr0man>
same version of a dependency, it should only be present on the system in one place to make the updating process less daunting
srid has quit [Ping timeout: 248 seconds]
<gchristensen>
yeah Nix does that
<joepie91>
m1cr0man: to be clear, NixOS does not use containers at all, unless you explicitly configure it to
<joepie91>
m1cr0man: a lot of deployment strategies conflate dependency isolation and runtime isolation, and they apply runtime isolation (Docker etc.) to get dependency isolation as a side effect
<joepie91>
Nix only provides dependency isolation
<joepie91>
(which is /usually/ what you want)
<joepie91>
if you want to also have runtime isolation for some reason, then you can either use declarative containers (in NixOS), or your favourite other containerization technology
<simpson>
m1cr0man: Ignoring NixOS for a moment (or, at least, treating it as nixpkgs + systemd), the packages are in a heap, much like spare dynamic memory allocations in a process. This heap, the "Nix store", is garbage-collectable, has roots, etc.; the analogy is quite complete.
isaac2 has quit [Quit: WeeChat 2.2]
<simpson>
You have whichever packages are currently at your fingertips.
<m1cr0man>
Oh right, yeah that makes a lot of sense.
<simpson>
Nix treats your disk like Firefox treats your RAM~
<edcragg>
is there any advantage to declaring a list as singleton in Nix? i see a number of packages using `users.users = singleton { }`, but presume that it's not compulsory
<tilpner>
edcragg: singletion is defined with: singleton = x: [x]
<tilpner>
edcragg: It's really just a function that returns the element in a list, nothing special
<tilpner>
edcragg: The comment above the definition says this can be more convenient for indentation
jakkn has joined #nixos
mthst has quit [Quit: the bouncer died]
Melkor333 has quit [Quit: WeeChat 2.4]
knupfer has quit [Ping timeout: 248 seconds]
<edcragg>
ok, cool. thanks
marusich has joined #nixos
<edcragg>
so it's not something to do with adding an extra user to users.users in a module?
<gchristensen>
nope
<gchristensen>
> :p lib.singleton 1 == [ 1 ]
<{^_^}>
true
<edcragg>
hmm
gagbo has joined #nixos
lambda-11235 has joined #nixos
<lambda-11235>
Is there a way to get appimages to run?
<bifunc2>
Did anyone manage to successfully complete "12.5. Completing graphviz with gd" ?
cyris212 has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
<bifunc2>
I keep getting the "Format: "png" not recognized" error no matter what. I tried on both Linux and macOS.
<bifunc2>
Any ideas?
cyris212 has joined #nixos
freeman42x has quit [Ping timeout: 264 seconds]
rabbit-ng has joined #nixos
rabbit-ng has left #nixos [#nixos]
lambda-11235 has joined #nixos
<JaakkoLuttinen[m>
If an executable is wrapped with setuid using `security.wrappers`, does this wrapped setuid executable appear somewhere in `/nix/store` or only in `/run/wrappers`? I'm asking because I'd like to create a script which uses this setuid'd executable with a fixed path without relying it being in PATH.
<JaakkoLuttinen[m>
It would feel "wrong" to use `/run/wrappers/bin/foobar` in a script instead of `${pkgs.foobar}/bin/foobar` but so that this latter would be just the setuid'd executable..
<maurer>
JaakkoLuttinen[m: My understanding is that /nix/store is intended to be mountable from a nosuid source
<maurer>
(and that we don't want suid bits showing up in .nar etc)
Ariakenom has joined #nixos
<gchristensen>
a NAR doesn't have the ability to store the suid bit
<maurer>
JaakkoLuttinen[m: Can you give more context on where you want to be able to call "the suid version of foobar"?
<gchristensen>
and SUID bits are prevented from being set on macOS and Linux during builds, and are stripped after the build completes
<maurer>
If it's just a personal script or something, just call the one from wrappers/bin. If it's packaged, you may want to contrive for yourself to be root, then exec the script
<JaakkoLuttinen[m>
maurer: I'm creating my i3 config file with nix and there I want to define key binding for locking the screen with slock. For this, I need to use the wrapped slock executable.
acarrico has quit [Ping timeout: 258 seconds]
<maurer>
JaakkoLuttinen[m: Ah, I do the same thing for xmonad. I just call the slock in path
<maurer>
Doing /run/wrappers/bin/slock is totally clean there, fear not
<JaakkoLuttinen[m>
ok, thanks!
<maurer>
gchristensen: And given the shenanigans e.g. calibre has engaged in in the past, I think that is a good decision
<maurer>
gchristensen: Too easy for a suid to slip past the package maintainer otherwise
<gchristensen>
:o what'd calibre do? :o
<maurer>
They made a "mount helper" binary to make it possible to autoconnect to kindles and such
<maurer>
but they are not exactly low level systems people
<maurer>
and it had lots of vulns
<maurer>
and they made their mount helper binary suid so they wouldn't need to ask for permission
<maurer>
etc.
<maurer>
2011-4124,4125,4126 at a minimum, I think there was another batch reported at some point
<gchristensen>
my word
<maurer>
I mean, I see how they got there, but this is why it's good to have processes to help guard against this kind of thing
<gchristensen>
yeah.
<gchristensen>
I am definitely +1 on revoking their privilege to do such things by default
<maurer>
(I'd also say this is an argument in cap-based perms if any of you are making a new kernel :P
<clever>
and chroot can be done without root, by just pulling off a mitm attack on the fs channel
<clever>
setuid binaries become an issue though
<maurer>
gchristensen: It's a combination of a BSD port and a hardware design, but still in academic world
<MichaelRaskin>
Wait, if we are talking of mounting Kindle, the correct way with Linux would be just to allow the FUSE helper to get the USB device FD as #4
<maurer>
MichaelRaskin: That doesn't let them scan for the kindle
<maurer>
This wasn't a fuse helper
<maurer>
it was a thing that scanned USB/bluetooth etc and then executed an appropriate "mount" command
domogled has joined #nixos
zeta has joined #nixos
<maurer>
gchristensen: If you are further curious about it, I know one of the postdocs on the project and can forward questions or make an introduction
andreas303 has quit [Quit: andreas303]
<MichaelRaskin>
I think unprivileged process can learn enough of USB to ask the helper for a specific device
<maurer>
MichaelRaskin: this may be true
<maurer>
I'd have to go reread exactly what thye were doing
<gchristensen>
maurer: oh dear I am not educated enough :)
<gchristensen>
thank you though
andreas303 has joined #nixos
<MichaelRaskin>
But yeah, all these crap devices are not a well-specified enough problem, so the solutions tend to be hacky and with holes in other directions
<maurer>
As much as I dislike a lot of Android system design, I think their vendor/system split was actually pretty genious
<maurer>
*genius
<zeta>
is this error talking about home.nix or do i need to create this default.nix directory ? error: getting status of '/home/wolf/default.nix': No such file or directory
shibboleth has joined #nixos
<maurer>
zeta: Context? It sounds like you ran nix-build or nix-shell in your homedir
<JaakkoLuttinen[m>
I enable xss-lock: `programs.xss-lock.enable = true`, but still there's no xss-lock systemd service found with `systemctl status xss-lock`.. What could be wrong?
<MichaelRaskin>
maurer: I have a feeling that the proper vendor/system split was called «BIOS Interrupts»
<maurer>
MichaelRaskin: So, the vendor/system split in this case allows the vendors to have userspace daemons as well, which provide generically specified APIs
<maurer>
MichaelRaskin: And then enforces it with SELinux
<MichaelRaskin>
I will believe it when phones get routinely upgraded to new Android versions without any vendor action
alexherbo29 has joined #nixos
<MichaelRaskin>
Android is a graveyard of good design decisions that are never actually allowed to serve any useful purpose
<maurer>
So, the split is already there in the architecture. I can't comment on whether you will be able to do independent upgrades in the future
<sphalerite>
mumble mumble chrom(ium|e)os
sigmundv_ has quit [Ping timeout: 248 seconds]
<zeta>
i am setting up cachix, the last command(nix-build) is throwing the error, if you guys could take a quick look at the instructions, thanks
<maurer>
My point was mostly that I think the notion of having a split between hardware-related daemons and general daemons both in privilege and design is smart
<clever>
MichaelRaskin: i liked that you need a special gid to use network, but modern android doesnt even bother asking for net perms when installing
<MichaelRaskin>
Would be great if they were actually used
<maurer>
zeta: It is an example command, e.g. if you go into a directory with a default.nix and run nix-build, it will use cachix
alexherbo2 has quit [Ping timeout: 252 seconds]
<maurer>
(assuming the previous commands completed successfully)
<maurer>
MichaelRaskin: I can't really fully endorse intents - in an intensely privacy relevant context, it makes it very hard to know where information will go
<clever>
i also like how you can serialize the state of an app and terminate it
<clever>
then restore it later
<clever>
but few apps actually use it properly
<maurer>
zeta: tl;dr whatever you were configuring cachix for, go try to do it, and cachix *should* be invoked
MmeQuignon has joined #nixos
<maurer>
clever: Yeah, I'm hoping that Optane + some clever libs may bring that to general purpose computing
knupfer has quit [Remote host closed the connection]
<maurer>
clever: Basically, if your data structures are designed in a semi-position-independent way, you could imagine being able to essentially "warm-boot" an application that was closed if you kept some of its data in optane rather than main memory
<zeta>
oh, that makes sense, i am setting up cachix for the haskell-ide-engine, so does that mean i need to create a `default.nix` for this or will it be done automatically ?
<maurer>
MichaelRaskin: Only kind of - it would be a usability nightmare if I always had topick a browser to open links, for example
<maurer>
MichaelRaskin: but at the same time, that very ability (opening links in a browser) is one of the way an app can leak data
<clever>
maurer: many years ago, i made a program that used shm to share some cache structures between many procs, and the pie structures where fun to work with
o1lo01ol1o has joined #nixos
<MichaelRaskin>
(I actually don;t have a default PDF handler set on my smartphone)
<clever>
eeeh, without the e on pie, lol
<maurer>
clever: Yeah, you'd probably need some fancy support libraries to make everyone not need to personally write everything that way)
alexherbo291 has joined #nixos
<maurer>
e.g. you could imagine a rust proc macro that converted a datastructure that used boxes/arcs etc into one that used offset based pointers
alexherbo29 has quit [Ping timeout: 248 seconds]
<maurer>
it would require some ninja voodoo from the library maintainer, but then you could allocate one of those in the pseudo-nvram space, and be able to get it back later
<maurer>
which would be pretty cool
o1lo01ol1o has quit [Remote host closed the connection]
<zeta>
maurer: thanks, i'll give it a try
<zeta>
maurer++
<{^_^}>
maurer's karma got increased to 3
o1lo01ol1o has joined #nixos
pingiun has quit [Quit: bye]
<das_j>
,locate gnustep-config
<{^_^}>
Found in packages: gnustep-make
fenedor has quit [Quit: Leaving]
shibboleth has quit [Quit: shibboleth]
parsnip has joined #nixos
bart` has joined #nixos
zeta has quit [Quit: WeeChat 2.4]
ixxie has quit [Ping timeout: 246 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<parsnip>
hmm, i have ~/.nix-profile/Applications/Emacs.app, but not sure the idiomatic way to make this available to Spotlight, etc
<parsnip>
i guess i could add that to the very few locations i let spotlight index
zeta has joined #nixos
<parsnip>
(i only let spotlight index /Applications/, to avoid cpu usage on dumb things like Maildir, git repos, etc
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<zeta>
i am still not sure `where` and `when` i need to do the last command(nix-build) ? https://hie-nix.cachix.org/
<maurer>
So, cachicx is a generic service for nix binary build caches
<maurer>
*cachix
o1lo01ol1o has joined #nixos
<maurer>
The instructions on cachix.org are generic for cachix - the hie-nix you see in the subdomain is just filling in which cache to enable
<maurer>
Many people adding a cache to nix are doing so to facilitate builds of some project, which is usually done via nix-build
<maurer>
Since you're just installing HIE, you are unlikely to run nix-build in a way that will need cachix
deadloko has quit [Quit: WeeChat 2.4]
<mog>
can i see easily whats taking up space in my /nix ? like which derivations are fat?
kshaa has quit [Ping timeout: 258 seconds]
<zeta>
well it says if i do not use cachix, hie will take a long time to compile ?
<clever>
mog: du -h --max=1 /nix/store/ | sort -h
<maurer>
zeta: Yes. That nix-env -i command I pointed you at is the one that would have taken a while
<maurer>
That describes a way to *build* hie - but since you enabled the cache, it sees that the build is cached, and just downloads it instead
shibboleth has joined #nixos
<maurer>
zeta: If you really want to see what the nix-build it's talking about is, you can clone https://github.com/domenkozar/hie-nix , cd into that directory, and run nix-build
<maurer>
but it will just create a results/ directory symlinked to the thing you just installed
o1lo01ol1o has quit [Remote host closed the connection]
<zeta>
so would recommend me to just remove cache and just compile hie, or to keep using it, is it worth the hassle ?
o1lo01ol1o has joined #nixos
<maurer>
You've already enabled the cache, why disable it?
o1lo01ol1o has quit [Remote host closed the connection]
<maurer>
Unless you're worried about trust issues, in which case you shouldn't just be blindly typing commands anyways
<asdf__>
hi can some one help me i'm trying to uncrypt disk that is a root device
<asdf__>
that isnt i mean
Guest75181 has quit [Quit: WeeChat 2.4]
orivej has joined #nixos
hedning_ has quit [Quit: hedning_]
<zeta>
infinisil: thanks, i did not know hie-nix was outdated
trevthedev has quit [Quit: WeeChat 2.4]
<infinisil>
all-hies is rather new :)
<lambda-11235>
For packaging, what should you do if an open source game is bundled with non-free libraries, secifically the discord and steam APIs?
<pie_>
cut out the cancer? :P
<lambda-11235>
Can't. Not without rewriting the game.
<maurer>
lambda-11235: Are they just depending on those APIs, or actually shipping libraries?
<mog>
clever, doesnt work very well because of hard links id think?
hedning_ has joined #nixos
andreas303 has quit [Remote host closed the connection]
nD5Xjz has quit [Ping timeout: 244 seconds]
<lambda-11235>
maurer: It has the .so and .h\ files included.
<lambda-11235>
Would the solution be to just add a nonfree license to the meta?
<maurer>
lambda-11235: I don't even know if you're allowed to redistribute those
<maurer>
Even "nonfree" in nixpkgs has to at least be redistributable
<manveru>
maurer: but we have discord and steam?
andreas303 has joined #nixos
nD5Xjz has joined #nixos
<maurer>
manveru: It is frequently the case that those kinds of things are licensed in such a way that you can redistribute them for running their app, but not separately, etc.
<maurer>
manveru: I do not know the licenses for the libraries in question
<maurer>
So, I would advise lambda-11235 to go read the licenses before uploading them
<MichaelRaskin>
I would say that if it cannot work without bundled Steam, the package as a whole is unfree
jb55 has joined #nixos
<MichaelRaskin>
Which is still acceptable for submission to Nixpkgs as an expression — won't be built on Hydra, of course
<asdf__>
what is the best way to decypt a drive after boot during systemd
<JaakkoLuttinen[m>
nixos doesn't seem to restart a systemd user service after `nixos-rebuild switch` although `ExecStart` of the service has changed. I need to manually run `systemctl --user restart foobar`. Is this right or can I make it restart automatically?
<JaakkoLuttinen[m>
`xss-lock` service in particular
<zeta>
infinisil: so just to make sure that i am doing this correctly, i need to remove hie-nix file: `rm -rf hie-nix.nix`, then: `cachix use all-hies` ? after that do i still need to clone: https://github.com/infinisil/all-hies and do a `nix-build` in that directory ? or do i continue with your instructions instead ? if i understand correctly: your instructions show how to clone `all-hies` within a nix file ? i'm
<zeta>
confused !!!
<zeta>
infinisil: so just to make sure that i am doing this correctly, i need to remove hie-nix file: `rm -rf hie-nix.nix`, then: `cachix use all-hies` ? after that do i still need to clone: https://github.com/infinisil/all-hies and do a `nix-build` in that directory ? or do i continue with your instructions instead ? if i understand correctly: your instructions show how to clone `all-hies` within a nix file ? i'm
<zeta>
confused !!!
<zeta>
sorry i accidently pasted more than once
<infinisil>
zeta: I'm not sure why you have a hie-nix.nix file to begin with, it shouldn't have been needed. For all-hies you just follow the instructions in the readme
<infinisil>
zeta: There's a section on installing with NixOS and one for installing with nix-env, choose whichever you prefer
<ambro718>
Does someone have a clue why this library fails to be copied (apparently)?
shibboleth has quit [Remote host closed the connection]
<ambro718>
My understanding is that the initrd generation code will automatically include linked libraries.
<infinisil>
zeta: Let me know if the instructions are unclear
<infinisil>
(specifically what is unclear in that case)
orivej has quit [Ping timeout: 248 seconds]
<clever>
ambro718: yeah, it will try to copy the libs into the initrd, and if it cant copy just the libs, it will just copy the entire closure
bifunc2 has quit [Quit: -]
<Ralith>
is there an easy way to get a 19.03 system onto kernel 4.20?
o1lo01ol1o has quit [Ping timeout: 248 seconds]
<zeta>
infinisil: ok, like i said, i was reading outdated instructions, so i think i need to just remove the old `hie-nix` and just continue with your instructions on `all-hies`
<zeta>
infinisil: i was talking to rycee a while back and he said currently there is not a way to install and configure haskell-ide-engine with home-manager(`home.nix`) so i have to use `nix-env` instead. so your instructions are on how to install and configure haskell-ide-engine globally in `configuration.nix` ? is this correct ?
<infinisil>
zeta: Ah, actually the instructions for NixOS can be used for home-manager the same way, by just using `home.packages` instead of `environment.systemPackages`
<colemickens>
firefox is out, is someone racing to send the PR already?
<ambro718>
What is buildCommand supposed to do when specified for mkDerivation? I do not find documentation in the manual.
m0rphism has quit [Ping timeout: 245 seconds]
sondr3 has joined #nixos
oida has quit [Remote host closed the connection]
<symphorien>
buildcCommand overrides all the phases
<sondr3>
I just updated my nix-channels and now I cannot install Emacs because it fails when installing emacs-rust-mode, anyone else experiencing this?
<symphorien>
For derivations so simple you only need one phase
oida has joined #nixos
<sondr3>
nevermind, there's a few others, now helpful doesn't want to compile
<sondr3>
I get a 403 from the download apparently
<sondr3>
err, nevermind again, now it worked
<sondr3>
when in doubt run nixos-rebuild thrice
lambda-11235 has quit [Quit: Bye]
sheeldotme has joined #nixos
eraserhd has joined #nixos
zeta has quit [Quit: WeeChat 2.4]
<sondr3>
in any case, I think you might get rate limited from GitHub now
eraserhd has quit [Quit: WeeChat 2.4]
<sondr3>
I get the occasional `curl: (7) Failed to connect to codeload.github.com port 443: Connection refused`
nD5Xjz has quit [Ping timeout: 246 seconds]
eraserhd has joined #nixos
Ariakenom has quit [Quit: Leaving]
<sheeldotme>
I'd like to use a derivation available in the nixpkgs/unstable repository, is there a standard way to do that?
nD5Xjz has joined #nixos
<sondr3>
sheeldotme: through configuration.nix or just via nix-env?
<sheeldotme>
I'd like to add it as a nativeBuildInput in my derivation in default.nix
<Ralith>
that does not sound like the right way to override it, then
<qyliss>
oh lol I didn't even notice
<qyliss>
I found it by reading the module source
<clever>
Ralith: ive set that option before to use a variant of libGL with debug symbols
<qyliss>
I've used it to use a custom set of drivers
<Ralith>
overriding mesa_drivers is getting the dep in somehow but it's unclear exactly how
<qyliss>
I have an in-progress branch for making basically every mesa dependency optional
<clever>
Ralith: if you only put your override into hardware.opengl.package, you can avoid the mass-rebuild it would otherwise cause
<qyliss>
I believe mesa_drivers is the default value of hardware.openGL.package?
<qyliss>
Possibly through an alias
<clever>
i think its a buildEnv of drivers and something else
<Ralith>
I don't have a lot of stuff in my systemPackages, so hopefully not a huge deal
ilikeheaps has quit [Read error: Connection reset by peer]
ilikeheaps has joined #nixos
nD5Xjz has quit [Ping timeout: 255 seconds]
<Ralith>
turning off all the drivers I need in the options annoyingly seems to have had no effect, leading to a very slow build
nD5Xjz has joined #nixos
<Ralith>
(i.e. it's still building them)
<Ralith>
probably the switch to meson that did it
<sheenobu>
building nix v2.2.2 on fedora. should build fine but i'm not sure what to do after that.
<Ralith>
and of course it fails at the last step
<Ralith>
argh
<sheenobu>
setup nixpkgs and switch my desktop to something nix-based while keeping my fedora base system... or start hacking together some things on top of the fedora base sysstem
nD5Xjz_ has joined #nixos
Glider_IRC_ has joined #nixos
nD5Xjz has quit [Ping timeout: 258 seconds]
shibboleth has joined #nixos
Rusty1 has joined #nixos
<lopsided98>
Is there any way to make a setup hook refer to its own package?
ilikeheaps has quit [Read error: Connection reset by peer]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
ilikeheaps has joined #nixos
devalot has joined #nixos
Glider_IRC has quit [Ping timeout: 268 seconds]
<sheenobu>
lopsided98: i'm a little rusty. what are you trying to accomplish?
dustinm has quit [Quit: Leaving]
ivan has quit [Quit: lp0 on fire]
<m1cr0man>
Hey again quick question. Trying to build a netboot image. I'm running nix-build from an install I made in a VM. Where do I find the "nixos/release.nix" file?
<gchristensen>
you have nixpkgs?
<gchristensen>
in nixpkgs/nixos/release.nix
dustinm has joined #nixos
e_linq has joined #nixos
<sheenobu>
alright since i'm waiting on nix to build i'll go ahead and checkout nixpkgs. but i won't be happy about it =P
<e_linq>
hi
<sheenobu>
hullo
<m1cr0man>
Hm I don't think I do, will read some more :P
<e_linq>
what is it?
<e_linq>
it may first time in irc
<e_linq>
its fucking amazing
<lopsided98>
sheenobu: I have a package that needs to set an environment variable that points at itself
<lopsided98>
Mostly for nix-shell purposes
<colemickens>
do you just need $out ?
ivan has joined #nixos
<lopsided98>
colemickens: Normally setup hooks are defined in another derivation, so that doesn't work
<lopsided98>
I guess I could manually create $out/nix-support/setup-hook
rprije has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
<qyliss>
that's not unheard of
<qyliss>
Ruby does it, for example
<clever>
i think mkDerivation also accepts a setupHook = ./hook.sh;
<clever>
and it will be copied to the right spot for you
<sheenobu>
hrm. i've got a thumb drive sitting here. wondering if i should just break down and blast away fedora for nix on here.
<lopsided98>
clever++
<{^_^}>
clever's karma got increased to 127
<zeta>
infinisil: could i a see a configuration of `all-hies`, every time i try to use: `let in` in configuration.nix i get a syntax error, is it better to create another nix file for `all-hies` and use it as a submodule ?
<sheenobu>
if so, i mostly need to make sure my drivers work and my LUKS/LVM doesn't get destroyed
<infinisil>
zeta: Can you show your nix file?
<zeta>
infinisil: sure, let me pastebin it, one second
<sheenobu>
how is wayland support?
Anton-Latukha has quit [Ping timeout: 246 seconds]
<qyliss>
I can only speak for sway, but it works perfectly, and is super easy to set up
<sheenobu>
cool. i'm not using sway now because 1. no out of the box touchscreen. 2. no wayland redshift for fedora. but nixos is a little easier for me to fight with
<maurer>
So, out of curiosity, what are y'all trying to use wayland for? Has it grown improvements over X? I know X is kinda legacy, and wayland will probably *eventually* succeed it, but I've not actually heard of benefits to using wayland.
<infinisil>
zeta: Yeah, just add the version you want to home.packages like in the NixOS section of the readme
<Qubasa>
Does someone know how to get the original installPhase if you are trying to override a derivation? I am trying this but it doesn't work sadly: pkgs.lib.overrideDerivation pkgs.firefox (oldAttrs : { installPhase="${oldAttrs.installPhase}"; })
<simpson>
maurer: Wayland explicitly *cannot* succeed X11, as it doesn't replace every use case. Most notably, Wayland isn't network-transparent and everybody involved has accepted that.
<maurer>
simpson: by succeed here I mean become the default display server in linux distros. I use X forwarding sometimes, but I'm aware that most people don't
<maurer>
It would need to offer some kind of other benefit though, and I'm not sure what that is beyond "the code doesn't smell as bad"
talqu_ has quit [Remote host closed the connection]
<simpson>
maurer: Okay. However, since Wayland can *host* an Xorg server, I imagine that Wayland supremacy would mean that we all still run Xorg for a few applications which haven't/won't migrate, as well as any network-transparent needs.
<Ashy>
maurer: probably the most noticable improvement is no screen tearing
talqu has quit [Ping timeout: 268 seconds]
<maurer>
Ashy: What's an example of an application where I'd expect to see screen tearing?
bepvte has joined #nixos
<maurer>
(I may just be missing this because I'm a luddite who spends pretty much all his time in either lilyterm or chrome)
<zeta>
infinisil: ok sounds good, also the terminal is telling me some cabal commands are outdated and do not work, why is that ?
<infinisil>
zeta: Read the full message
nD5Xjz_ has quit [Ping timeout: 248 seconds]
<bepvte>
is the correct way to set musl as libc '--arg crossSystem '{ config = "x86_64-unknown-linux-musl";}''
<simpson>
Ashy, maurer: Note that for many applications that can tear (Firefox and mpv come to mind for me), a bare-bones compositing manager like compton can do the same thing for Xorg.
<maurer>
Oh, I'm running xcompmgr
hio has quit [Quit: Connection closed for inactivity]
<maurer>
so maybe that's why I do not see tearing
nD5Xjz has joined #nixos
<simpson>
(Same codebase, different fork name. compton is a fork of xcompmgr.)
<maurer>
simpson: So uh, are there benefits beyond the code being nice to wayland? I'm not trying to be an ass, I'm genuinely curious, since I frequently run fun cutting edge software on my machine, but never understood the pull of wayland
<colemickens>
it actually works *well*? performance is consistent. hidpi and mixed dpi monitors works *well*
<bepvte>
x11 lets any window read any windows contents, so theres a security boost, as well as VSYNC stuff never being a problem
<simpson>
maurer: IMO there aren't really any benefits other than the memes: clean-slate, homesteading opportunities, every frame a painting, etc.
<simpson>
The rendering model's certainly nicer than X11's! But it's not like that wasn't already known.
<Ashy>
simpson: a compositor can help, but with 4k monitors screentearing is still fairly bad in say youtube in firefox
<maurer>
colemickens: I thought mixed DPI problems were usually due to the apps not X?
<maurer>
e.g. if I ask X about DPI of my two monitors, it does correctly register them as different numbers
<maurer>
but applications can only seem to understand a single DPI, for whatever reason
<simpson>
bepvte: Don't worry too much about that; without ARB_robustness, any (E)GL user can read GPU memory! (Even with ARB_robustness, security is not exactly a GPU driver priority.)
<colemickens>
No, X fundamentally isn't setup to support it. Even in a model where it could, Xorg still needs to change to adapt. KDE folks are working on it because it's needed for XWayland and mixed DPI to work properly as well.
mabel has quit [Ping timeout: 245 seconds]
<maurer>
Not to mention without selinux or something similar, anything executing on your X already has full permission to act as you
<maurer>
and go poke other things
<colemickens>
It might know the *monitors* are different DPI, but the screen definitely isn't.
<simpson>
Ashy: Sure. It's all down to the rendering model. Wayland's every-frame-a-painting model is very nice when the expected FPS is consistently above, say, 20 frames/s.
<simpson>
maurer: We have a traditional analogy for this. DPI settings are like babies.
<bepvte>
is there a way to set localsystem permanently without having to --arg every time i use nix-env
<simpson>
You can't just pick one baby. You can't mix-and-match babies. You can't try to reinterpret one baby as another.
<Ralith>
apparently overriding src on mesa doesn't work
<Ralith>
just one thing after another
<maurer>
Yeah, if I can put two different DPI monitors next to each other, and then move chrome from monitor A to monitor B and have it not look fucky (either way to big or way to small), that'd move me to wayland immediately
<colemickens>
maurer so, that's the case today, except Google kinda forgot to invest in Wayland at all, oops
<colemickens>
so Chromium is basically the only thing that doesn't work well in Wayland.
<zeta>
infinisil: will the import work locally in home.nix or does it have to be placed globally in configuration.nix, i am just curious ? https://hastebin.com/lomehusiso.bash
<colemickens>
and thus all Electron apps too. Otherwise, what you described is exactly what happens with Sway.
<maurer>
colemickens: Will Firefox work?
<bepvte>
there is a lot of work going on on wayland but it is not built
<qyliss>
firefox has good wayland support
<bepvte>
firefox has a wayland port and its pretty good yeah
<colemickens>
Yes.
<maurer>
I specifically mean the DPI thing, not general wayland support
<colemickens>
I actually switched back to stable today too.
<Ralith>
or it did work and then the old version just got built anyway???
<colemickens>
maurer yes
<colemickens>
I specifically am saying mixed DPI works well with firefox and sway, as a daily user of all three
<qyliss>
(make sure you use the firefox-wayland package though)
<infinisil>
zeta: The `let in` syntax defines a variable for use in that file only, so you can reference `all-hies` within that file but not outside
<colemickens>
qyliss wait what
<qyliss>
It's the same, except it forces GDK_WAYLAND
<qyliss>
GDK_BACKEND
<qyliss>
whatever
<colemickens>
Oh, right on.
<infinisil>
zeta: Did you manage to install it?
<qyliss>
So if you have an X11 program that tries to spawn firefox, you won't get X11 firefox
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 248 seconds]
mbrgm_ is now known as mbrgm
<ambro718>
How do I get NixOS test logs (including machine screenshots) if the test fails?
<ambro718>
I see this in the output: machine: sending monitor command: screendump /nix/store/dhsn4sryl7w8dlxa9xxqy01aajl7q8ci-vm-test-run-virtualbox-simple-gui/gui_manager_started.png.ppm
<ambro718>
no idea how to get to that picture
<zeta>
infinisil: i'm almost done, i just finished configuring cabal, now back to all-hies, i'll let you if i got it to work or not
shibboleth has quit [Quit: shibboleth]
<zeta>
infinisil++
<{^_^}>
infinisil's karma got increased to 97
<infinisil>
Alright :)
sheenobu has quit [Remote host closed the connection]
zeta has quit [Quit: WeeChat 2.4]
sl0wz has joined #nixos
detran has joined #nixos
slowz has quit [Ping timeout: 252 seconds]
silver has quit [Quit: rakede]
zeta has joined #nixos
<zeta>
infinisil: it did not work, maybe i am supposed to put it in the `hpkgs` set of packages instead ? here are my pastebins: