tertl3 has quit [Quit: Connection closed for inactivity]
<drakonis>
isn't konqueror in practice unmaintained?
<drakonis>
rather, nobody really works on it anymore
silver has quit [Read error: Connection reset by peer]
Dedalo has joined #nixos
jackdk has quit [Disconnected by services]
jackdk_ has joined #nixos
aleph- is now known as Church_
hyp3rbor3ax_ has joined #nixos
hyp3rbor3ax_ has quit [Remote host closed the connection]
hyp3rbor3ax_ has joined #nixos
jperras has joined #nixos
hyp3rbor3ax has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #49571 → home-assistant: use SIGINT instead of SIGTERM to shut down → https://git.io/fxNZe
<colemickens>
drakonis: I don't know. I had wondered the same, but it still seems to be part of KDE releases?
hyp3rbor3ax_ has quit [Remote host closed the connection]
<colemickens>
But then again, someone must've chosen not to package it when they were packaging the rest of the KDE apps.
o1lo01ol1o has joined #nixos
<colemickens>
I guess the NixOS iso probably just includes Firefox.
<colemickens>
It's a shame, Konqueror and Epiphany are basically the only decent choices if you want a native Wayland client...
<drakonis>
no point shipping konqueror though
<drakonis>
firefox's wayland support is getting there
<colemickens>
Yeah, my pursuit of Konqueror is arguably niche.
hyp3rbor3ax_ has joined #nixos
<drakonis>
its not going to be a iso default unless you want to decrease the image size
<colemickens>
Eh, I suspect Chromium beats them. They've quietly been ramping up merging the ozone/wayland stuff.
<drakonis>
which then i'd rather have falkon instaed
<drakonis>
falkon over firefox if that's the goal
<colemickens>
Eh, that's not what I meant. I was just thinking through the ISO using KDE and what browser it must include. I don't really care what ships on the installer ISO to be honest.
hyp3rbor3ax_ has quit [Remote host closed the connection]
<ekleog>
drakonis: if KDE still ships Konqueror it would sound really weird to me for NixOS to refuse to includeit
<drakonis>
not against it, but its not exactly widely shipped as a default for anything
<drakonis>
maybe debian still ships it if you choose to install the kde metapackage
<ekleog>
and?
<drakonis>
debian still does it, fedora ships it with the kde image
<ekleog>
we have lots of programs that are in no defaults
<drakonis>
fair enough
<ekleog>
and thankfully, otherwise no one would be using nixos :p
alex`` has quit [Ping timeout: 245 seconds]
<ekleog>
colemickens: hmm, something is weird to me though, you use kdelibs4support but kde is kde5 now… right? (not sure whether the “4” here refers to the kde version, though)
<colemickens>
ekleog: I *think* it's common for kde5 apps to still pull in the kde4libs. I think I had started from nothing and only added stuff as the build errors came up. Let me spot check against some other packages and see if it's listed as a konq dep
justanotheruser has joined #nixos
<ekleog>
colemickens: hmm yeah I can confirm dolphin still uses kdelips4 too
<catid>
Hello! I'm attempting to get Gstreamer working with VDPAU hardware acceleration. gst-inspect-1.0 -a does not list vdpau
<catid>
sudo DISPLAY=:0.0 vdpauinfo does show that it's available
drakonis_ has quit [Ping timeout: 260 seconds]
Incin has quit [Ping timeout: 256 seconds]
drakonis_ has joined #nixos
davidak has quit [Quit: Leaving]
romanofskiWork has quit [Quit: leaving]
garbas has quit [Quit: WeeChat 2.2]
zopsi has quit [Quit: Oops]
zopsi has joined #nixos
jackdk_ has quit [Remote host closed the connection]
drakonis has quit [Quit: WeeChat 2.2]
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
justanotheruser has quit [Ping timeout: 240 seconds]
<catid>
So VDPAU is old and broken and NVDEC is the new way to do it. Looks like Gstreamer 1.14.0 that is currently in Nix doesn't support NVDEC - Version 1.14.2 or newer might
jackdk has joined #nixos
romanofskiWork has joined #nixos
justanotheruser has joined #nixos
catid has quit [Quit: WeeChat 2.1]
sanscoeur has joined #nixos
romanofskiWork has quit [Ping timeout: 240 seconds]
adamantium has quit [Remote host closed the connection]
<Kim>
So I'm currently looking at cross-compiling images for armv6l for some open-source wearables I'm working on. I'm currently looking through the information at https://nixos.wiki/wiki/Cross_Compiling -- does anyone know if there's anything else floating around? Or am I already in the right spot for that?
<katyucha>
my configuration.nix is good with the 2 32bit active ..
<katyucha>
Have someone any idea .. Thanks
<srk>
you can try wiping ~/.steam if you don't mind downloading all the games again
<srk>
or maybe just moving it temporarily
vk3wtf has joined #nixos
<katyucha>
not a prolem but moving my saves before
jasongrossman has joined #nixos
<katyucha>
saves are on .local ... so, it's ok
<cocreature>
katyucha: how have you installed steam? that looks like the gl version that you’re running might have gotten out of sync with the one that your steam package is using
<cocreature>
which could happen if you installed it with nix-env
<katyucha>
Steam is installed with nix-env since the 17.03 and it's my first problem :( I used nix-env
<katyucha>
I try to uninstall , delete .steam and reinstall
worldofpeace_ has quit [Ping timeout: 244 seconds]
Thra11 has quit [Ping timeout: 244 seconds]
<katyucha>
ok installation failed ... curl: (22) The requested URL returned error: 404 Not Found
<katyucha>
error: cannot download libvulkan1.deb from any mirror
<{^_^}>
[nixpkgs] @megheaiulian opened pull request #49579 → brightnessctl: fix udev and clean unnecessary make flags → https://git.io/fxNrM
thc202 has joined #nixos
kreisys has joined #nixos
phreedom has quit [Quit: No Ping reply in 180 seconds.]
leotaku has quit [Ping timeout: 246 seconds]
phreedom has joined #nixos
lostman has joined #nixos
<lostman>
is it possible to merge the output directories of two derivations? I need to create a derivation where clang and llvm binaries are in the same place
<lostman>
so basically need to build clang and llvm and copy the outputs of both to the same directory
<arianvp>
gchristensen: how am I supposed to use terraform.withPlugins ?
CcxWrk has joined #nixos
<symphorien>
lostman: symlinkJoin or buildEnv may help
rprije has joined #nixos
<lostman>
@symphorien cool, will have a look at the docs
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @eadwu to master « pythonPackages.zstd: init at 1.3.5.1 (#49339) »: https://git.io/fxN6G
alex`` has quit [Ping timeout: 244 seconds]
alex`` has joined #nixos
ThatDocsLady has joined #nixos
<lostman>
@symphorien `symlinkJoin` does the trick! will also play with `buidEnv` to see if I can get what I want without creating a new derivation in `/nix/store`
<adamt>
johanot: Probably. Thanks. Silly me searched for rust and cargo in the nixos- and nix- manuals, and forgot all about nixpkgs. Time to get some sleep
sigmundv has joined #nixos
<srhb>
adamt: Make sure to take a look at the mozilla rust repo as well
<srhb>
adamt: Since they use nix extensively.
alex`` has joined #nixos
alex```` has quit [Ping timeout: 240 seconds]
<adamt>
srhb: Are the information in the nixpkgs manual not sufficient to get something basic working? It's just a quick proof of concept I'm working on
katyucha has quit [Ping timeout: 240 seconds]
katyucha has joined #nixos
teser has joined #nixos
<teser>
hi, I would like to know how to create a directory inside /var/log when building a package. I'm getting the following issue when running nixos-rebuild: "mkdir: cannot create directory '/var': Permission denied" ?
<teser>
I've seen that the module "nixos/modules/services/misc/jackett.nix" is doing the same and it works, not sure why it doesn't work on my side, I'm also doing it in preStart.
<srhb>
adamt: Yeah, definitely.
<teser>
Another example is in the pkgs/tools/backup/bareos/default.nix package, which specifies the ""--with-working-dir=/var/lib/bareos"" - this is exactly what I need, however I'm not sure what ensures that the directory will be created (that package doesn't create it itself).
<srhb>
teser: You cannot do that.
<teser>
srhb, why does the pkgs/tools/backup/bareos/default.nix do it then?
<srhb>
teser: I'm assuming those are used at runtime.
<teser>
srhb, ok, so it's ok to use such flags at runtime, but not at compiletime?
Spooktober has quit [Quit: Spooktober is over.]
<srhb>
teser: At runtime your program can do whatever you like (given normal permission restrictions) -- at build time, it's extremely sandboxed and cannot even see most of the filesystem. :)
grumble has joined #nixos
<srhb>
teser: Essentially all a build can do is put contents in its $out (or other outputs), that is, one or more /nix/store/xxx...-name-ver/ paths
<teser>
srhb, is there a way to enter the sandbox, so I can see what part of the filesystem it sees, or if there is documentation about that?
<wirew0rm>
teser: if your service package complains if the dir does not exist, you need a wrapper or if it is a service use systemd preStart to create the directory
<srhb>
teser: There is, though to my knowledge it's not well documented yet. If you want to experiment the easier thing is probably to use ls, find, etc. in your build to test what paths are there
<symphorien>
we want lightdm to look for config files in /etc but to install its sample config files in $out/etc
rprije has quit [Ping timeout: 250 seconds]
Akii has joined #nixos
<teser>
symphorien, srhb what if I use the following:
turion has joined #nixos
<teser>
configureFlags="--with-sqlite3=${pkgs.sqlite.dev} --with-postgresql=${pkgs.postgresql} --with-logdir=$out/log --with-working-dir=$out/working" -> this will compile the package, but not sure what happens at runtime, since it won't be able to actually write the files into those folders.
<srhb>
teser: It probably fails.
<srhb>
teser: since it'll try to do that and error out hard because it's read only.
<srhb>
Definitely not the way to go. symphorien's suggestion above is excellent :)
<teser>
Ok, I'll take a look at that, thanks
<lostman>
how to find sha256 for builtins.fetchTarball?
<symphorien>
,tofu
<{^_^}>
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.
<srhb>
lostman: With a recent enough nix, it will output the hash in the error message, instead of the unhelpful earlier message.
<srhb>
eg. Nix 2.1.3
<lostman>
oh, I have the older version. how do I upgrade?
<srhb>
lostman: Depends how you got Nix in the first place
<lostman>
I'm on Ubuntu
<lostman>
installed using the script from the website
<teser>
symphorien, so I should pass the configure script a path in /etc/bacula, while the installFlags should specify $out/etc/bacula?
<srhb>
lostman: I'm not sure if nix upgrade-nix is safe yet, but either that or rerunning a newer installer.
<teser>
symphorien, can you explain this in little more details, since configure/make install is still at build time, what about at runtime when the bacula is actually run?
<symphorien>
lostman: can you run `nix-shell -p nix-info --run nix-info`
<srhb>
teser: Presumably, make bakes the path you specify into the binaries/scripts
<symphorien>
teser: if you pass --with-working-dir=/var/lib/bacula bacula will be compiled with this path hardcoded but even if the path does not exist in the sandbox
<symphorien>
only make install will break
<infinisil>
lostman: Currently the tofu thing doesn't work with builtins.fetchTarball, you need to use `nix-prefetch-url --unpack <url>` (in the terminal) to spit out the sha
<srhb>
infinisil: It does, on 2.1.x
<infinisil>
Ahhh, must've changed then
<srhb>
Yes, thankfully :)
<symphorien>
so you pass somevariableuesdbyautotools=$out/whatever to tell the makefile to install what should go to /var/lib/bacula elswhere instead
<symphorien>
in the case of /etc with autotools the variable is called sysconfdir
<srhb>
prefetch-url is too error prone..
<infinisil>
srhb: It is?
<lostman>
update-nix worked fine!
<srhb>
infinisil: Well, only due to user error
<symphorien>
for your working dir, well, read the Makefile :/
<lostman>
@infinisil what does it compute sha of? I tried downloading the tarball and computing sha of that but it's different
<srhb>
infinisil: It's easy to generate a sha for the subtly wrong thing and not notice that your expression is not getting the url you thought it was. Happens all the time :-P
<srhb>
lostman: It's the unpacked source.
<srhb>
lostman: A recursive hash.
<infinisil>
There's the nix-hash binary you can use to get the same hash I'm pretty sure
<lukego>
Do all nix command line options also have configuration option equivalents? I'm not immediately seeing 'nix.extraSandboxPaths' and 'nix.allowNewPrivileges'. should I?
<lostman>
@srhb ah, got it. prefetch-url with --unpack does indeed work! thanks
<srhb>
lukego: Usually they are foo-bar-baz in nix.conf instead of fooBarBaz
<infinisil>
lukego: Check out `man nix.conf`
<srhb>
lukego: man nix,conf
<srhb>
eh..
<symphorien>
lukego: there is --option name value
<srhb>
lukego: In general, the NixOS options are quite out of sync with the actual nix names.
<symphorien>
ah sorry misread
<infinisil>
Nix 2.0 changed these --option's without backwards compatibility too :(
<srhb>
There's only one non-backward compatible one I think? And it used to be, but there were problems..
<lukego>
nix.extra-sandbox.paths = [ ... ] seems to work, but not 'nix.allow-new-privileges = true' ?
<srhb>
Which was it again..
<symphorien>
iirc there is nix.extraConfig, lukego
<srhb>
lukego: Where's that nix coming from?
<lukego>
this is a nixos config
<lukego>
symphorien: thanks, I'll try it that way
<srhb>
nix.extraOptions is indeed the way to go then.
<betaboon>
infinisil: yeah i was looking for `submodule ({ name, ... }: { options.foo.default = name; })` and i just remembered that we talked about that a couple of days ago. and i already wrote it to a notes file xD
init_6 has quit [Ping timeout: 240 seconds]
<infinisil>
betaboon: options.foo.default isn't valid though, you want config.foo = mkDefault .., and you don't want `options.foo = mkOption { default = ...; ... };` because that could rebuild the manual, I think
<turion>
I installed nixos on this machine without disk encryption. I'd like to change that and use LUKS. Is there anything (script, blog post) that guides me through the steps in NixOS?
GDPR is now known as emerson
<joko>
turion: I think you would need to re-install again
<etu>
turion: There's several ways to do this
<turion>
My main aim is to prevent data leakage in case the laptop is stolen, so I'm not really concerned with an encrypted boot partition
<etu>
turion: One is to reinstall and build a system with the same config. Another is to make a copy of the partiton from a live usb, create a luks partition and copy things back.
<turion>
Couldn't I shrink my partitions, create an encrypted partition, move everything there, then delete the old partitions and resize again
<etu>
Once I've used luksipc to convert an unencrypted partition into an encrypted one :D
<etu>
But that doesn't seem to be in nixpkgs :(
sb0 has joined #nixos
<etu>
luksipc is actually really impressive
katyucha has quit [Ping timeout: 268 seconds]
<turion>
etu: luksipc sounds cool! I wonder how one would put that in nixpkgs. I mean, just including it as an executable script should be trivial. But it would be nice if nixos-rebuild switch would automatically realise that you've replaced the unencrypted / with an encrypted one, and call luksipc. Or is that too magical?
periklis has quit [Remote host closed the connection]
<etu>
turion: Good find! Then I know for the next time I need something like that :)
iyzsong-x has joined #nixos
krgn has joined #nixos
<turion>
What I don't really understand yet is the "--reduce-device-size" option.
<turion>
It says: First, be sure you have space added to disk. Or alternatively shrink filesystem in advance. Here we need 4096 512-bytes sectors (enough for 2x128 bit key). fdisk -u /dev/sdb # move sdb1 partition end + 4096 sectors (or use resize2fs or tool for your filesystem and shrink it)
<turion>
What now. Am I supposed to enlarge or shrink now before using that option?
<srhb>
Make the (virtual) device larger or shrink the fs to make room for the luks headers, I suppose
<krgn>
heyhey! I'm using nixos, and I'm looking to get to libclang.so, but apparently its wrapped (clang-wrapped-6.*) and no `lib` output is defined. how can I refer to the real clang, if you will?
<turion>
srhb so basically partition != filesystem, so there is some 4096 sectors leftover in the partition that aren't used by the filesystem?
<etu>
turion: yeah, shrink the fs a few megabytes (more than enough) to make space for luks headers
<srhb>
krgn: llvmPackages.clang-unwrapped.lib I think
<krgn>
oh, that is where its hiding! thanks!
c0bw3b_ has joined #nixos
<etu>
luksipc operated by copying the first block from the disk. Writing luks headers, copying second block. Encrypting first block and put it where second block was. Copy third block. Encrypt second block and write where third was. And so on.
<etu>
turion: ^
<srhb>
krgn: Might want to look at nix-index/nix-locate for tasks like that :)
<etu>
I assume that cryptsetup-reencrypt does more or less the same kind of operation
<krgn>
srhb: nice one, I didn't know those :)
<symphorien>
krgn: llvmPackages.libclang.lib
LnL has quit [Ping timeout: 245 seconds]
<turion>
etu: Makes sense. I wonder... I can't leave my disk mounted while doing that, right?
<turion>
That's what's meant by "offline"?
<srhb>
turion: Right.
<etu>
You need to have it unmounted
<etu>
Since you're "changing block device", the filesystem may be confused :p
<srhb>
Clearly we just need a generalized splitfs kernel module
<srhb>
"this part is some fs, this other part is this other fs, please merge" -- what could go wrong
LnL has joined #nixos
<{^_^}>
[nix] @janpath opened pull request #2506 → Fix aws-sdk-cpp patch can not be applied → https://git.io/fxNxi
<nh2>
arianvp: from my recollection, the failed stops also happened also on a non-server node. What was your suspicion?
alex```` has quit [Ping timeout: 240 seconds]
<arianvp>
nh2: that suspision I have thrown away again
<arianvp>
unfortunately
<arianvp>
but what I do see, is that the "systemctl stop" command itself failed, from your logs
<arianvp>
which can happen when the stop job gets cancelled by systemd
<nh2>
you can still say what the suspicion was so that I don't have to come up with that myself and then find out that it doesn't make sense
<arianvp>
why it got cancelled is the mystery now
<arianvp>
I confused the two logs, and thought that in the failure scenario there was an "exit code 1" for consul
<arianvp>
in which a restart would trigger. If a restart and a "systemctl stop" is executed at the same time
<arianvp>
then systemd cancels the systemctl stop with the message "Stop job for consul.service canceled"
<arianvp>
but the log of the failed node had no failing process in it, so it's probably not that
<arianvp>
but, something caused the "systemctl stop consul.service" to cancel still... but I can't figure out why
<arianvp>
I tried asking in #systemd but that wasn't very helpful
<nh2>
hmm I see
<arianvp>
and googling the error only ends up in one bug tracker :P
<arianvp>
but yeh I have a feeling it has to do with consul, not nixos
<arianvp>
but thats a hunch
<nh2>
also, when I read your consul-server-question I didn't really follow why you thought this would be dependent on a consul-specific setting, as the `Stopping ...` part of switch-configuration already doesn't appear
<arianvp>
Yes but look further down the logs!
<nh2>
OK let's look together, where exactly shall I look?
<arianvp>
node-2..............> Job for consul.service canceled.
<arianvp>
this means the actual "systemctl stop consul.service" command failed
<arianvp>
which then means the job (the internal name for queued tasks in systemd) was never actually scheduled by systemd
<arianvp>
hence the lack of "Stopping consul.serivce"
erictapen has joined #nixos
<hodapp>
are many folks using Wayland on NixOS? just curious. I have no particular need
<arianvp>
hodapp: not many, but I have heard some people succesfully running Sway
<turion>
Einheiten: Sektoren von 1 * 512 = 512 Bytes
<turion>
I got "loop" from parted
<clever>
can you switch it back to english?, just need to tweak one of the LC_ vars i think
<nh2>
arianvp: do you know what exactly switch-to-configuration does after it prints `stopping the following units: cdn-rsync.service, server.service, consul-alerts.service, consul.service`?
<arianvp>
it executes the command " systemctl stop -- cdn-rsync.service server.service consul-alerts.service consul.service"
<arianvp>
that command then failed with "Job for consul.service canceled"
<arianvp>
whilst completing the stop for the other units
iyzsong-x has quit [Ping timeout: 268 seconds]
<nh2>
arianvp: what if due to some systemd configuration, like `Restart=always` or something like that, consul is immediately restarted? So systemd queues the stop command, but then it immediately cancels it itself because something else restarts it?
<turion>
clever:
<turion>
Disk /dev/sdb: 3.8 GiB, 4026531840 bytes, 7864320 sectors
<nh2>
arianvp: I can actually reproduce this! If I just type `systemctl stop consul` on my deployed system, I immediately get `Job for consul.service canceled.`
<arianvp>
wow
<clever>
infinisil: ,pastebin doesnt work
<arianvp>
that's! good to hear
<arianvp>
:D
<nh2>
arianvp: maybe that will help us figure out what happens
<arianvp>
now we just need to figure out why it's happening
<clever>
turion: yeah, thats not even partitioned, "partition 1" starts at offset 0
<clever>
infinisil: ah
drakonis has joined #nixos
<arianvp>
can you 'systemctl cat consul' for me ?
<arianvp>
and 'systemctl show consul'
erictapen has quit [Ping timeout: 246 seconds]
<arianvp>
nh2: also can you type "systemctl list-jobs" ?
<nh2>
arianvp: also just comented on the issue
<turion>
clever, weird, how did it even have a volume label?
<{^_^}>
[nixpkgs] @vincrusher opened pull request #49588 → More Types of CPUs Added. → https://git.io/fxAe7
<arianvp>
Could you slide into #systemd on irc?
<clever>
turion: the label is part of the filesystem, not the parititon tables
<nh2>
arianvp: also what we don't know is: When systemd prints `Job for consul.service canceled.`, it doesn't actually say what exact jobs was cancelled. Is it really talking about our stop-job, or perhaps it's cancelling some other consul-related job?
<turion>
clever: Ahhh :) thanks! Until today, I couldn't really appreciate the difference
<clever>
turion: gpt makes things more fun, by allowing labels on the partitions as well as the filesystems
alex``` has joined #nixos
__Sander__ has joined #nixos
alex`` has quit [Ping timeout: 252 seconds]
<yurb>
Hi everyone. I've submitted my first little pull request - https://github.com/NixOS/nixpkgs/pull/49556 But to be honest, I'm not sure how to interpret two of the checkmarks - "Tested compilation of all pkgs..." and "Determined the impact..."
<{^_^}>
#49556 (by yurkobb, 16 hours ago, open): supercollider: build with alsa (midi) support on non-darwin
<turion>
clever: Way too much confusion potential for me ;) Now I have created an msdos (~MBR-capable?) partition table and an ext4 partition, and formatted it. Beginners question: Do I have to set the grub device to /dev/sdb or /dev/sdb1?
<clever>
turion: sdb still
<turion>
But then there is no deterministic way to address that like I can address volumes with their UUID?
<turion>
This has confused me in the path as well, I think. The grub.device I need to specify, is it the path in the installing system or later in the installed system? Or do I possibly need to change it after reboot?
<schopp0r>
hi. I am (still) trying to build the nix package manager as a user on a non-nixos-computer. for some reason it builds "libaws" … I never used amazon web services. why is this a hard dependency?
<clever>
schopp0r: nix has an s3:// url it supports
<betaboon>
is there an easy way to generate the manual of a custom module that is not part of nixpkgs ?
<joko>
betaboon: sth like nixos-option <custom>?
<nh2>
arianvp: btw for later, we should vet more of nixpkgs systemd units, I think more might have `TimeoutStartSec=0` which is probably very wrong
<arianvp>
yes, noted
<betaboon>
joko: like the html stuff that is hosted at nixos.org/nixos/manual etc but for my custom module that i import
<schopp0r>
clever, can I disable this? because it takes ages.
<schopp0r>
and I don't need it
<clever>
schopp0r: it should have a flag for it in ./configure
<Unode>
hi all, I'm trying to use pygobject + gtk in nixos and I can't get past importing Gtk. http://dpaste.com/1Y1NZ0J am I missing something in my shell environment?
<{^_^}>
[nixpkgs] @jtojnar pushed 161 commits to gnome-3.30: https://git.io/fxAqk
<romildo>
clever, with this I am getting the stable version. I want to test a development version on my cloned nixpkgs repository. What else should I change on that command line
<clever>
romildo: with import ./. {}; instead
<romildo>
clever, thanks.
silver has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @c0bw3b opened pull request #49593 → dropwatch: init at 1.5 → https://git.io/fxAmG
romildo has quit [Quit: Leaving]
<dhess>
atlas, I hate you
<dhess>
anyone know why atlas would ignore a `doCheck = false` override?
<dhess>
something strange about its build system?
<symphorien>
doInstallCheck = false ?
<dhess>
interesting, maybe that's it
<dhess>
it's broken upstream btw and has been for about 2 weeks
<{^_^}>
[nixpkgs] @dotlambda merged pull request #49571 → home-assistant: use SIGINT instead of SIGTERM to shut down → https://git.io/fxNZe
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @peterhoeg to master « home-assistant: use SIGINT instead of SIGTERM to shut down (#49571) »: https://git.io/fxAOM
sanscoeur has quit [Remote host closed the connection]
sanscoeur has joined #nixos
dbmikus_ has joined #nixos
cacatoes has joined #nixos
sanscoeur has quit [Ping timeout: 252 seconds]
cacatoes has left #nixos [#nixos]
dbmikus__ has joined #nixos
dbmikus_ has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @Synthetica9 opened pull request #49595 → fetchFromGitHub: add goPackagePath similar to meta.homeapage → https://git.io/fxA3j
peacememories has joined #nixos
<{^_^}>
[nixpkgs] @c0bw3b opened pull request #49596 → pmtools: remove and alias → https://git.io/fxAsR
<{^_^}>
[nixpkgs] @jtojnar pushed 20 commits to gnome-3.30: https://git.io/fxAs1
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 245 seconds]
<schopp0r>
ok, "./aws-cpp-sdk-core-tests" runs since 186 minutes now. is this … ok?
lonokhov has quit [Quit: WeeChat 2.2]
<c0bw3b_>
yes aws-sdk-cpp is a long build
<schopp0r>
c0bw3b_, I build a nix-shell. "nix" doesn't seem to have an argument *not* to build it.
<schopp0r>
c0bw3b_, how can I determine whether it actually does something and is not just busily looping on something?
<dtz>
aws-sdk-cpp has a bug where it goes into infinite loop in tests on newer-ish versions
<dtz>
there's an upstream issue about it
johanot has quit [Ping timeout: 244 seconds]
hyper_ch2 has quit [Quit: Page closed]
<dtz>
no idea how that's remotely acceptable, the response was "yeah we know we'll fix eventually" lol
<dtz>
FWIW I encounter it regularly on one machine (hefty server but older kernel) and never on another (laptop, new kernel, NixOS) -- not sure how relevant any of that is but might help
<dtz>
schopp0r ^
<schopp0r>
dtz, meh. so what can I do?
<schopp0r>
dtz, "nothing", I guess :(
<dtz>
suppose you can disable the tests :/
<schopp0r>
I tried ;_;
<schopp0r>
I set the variables, I set the flags. but it still tests everything.
<schopp0r>
compiling nix is really a mess.
<c0bw3b_>
some can confirm it is ok to have underscore in a package version
<Mic92>
Has somebody by chance a working nix expression to run or build Spec benchmark?
peacememories has joined #nixos
<{^_^}>
[nixpkgs] @dtzWill opened pull request #49602 → pyjwt: fix tests to work with pytest >= 3.9.0 → https://git.io/fxAnM
<{^_^}>
[nixpkgs] @dtzWill opened pull request #49603 → slack-term: init at 0.4.1 → https://git.io/fxAnF
LnL has joined #nixos
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<dtz>
okay that's all I got in my queue xD
<dtz>
Mic92: nope sorry :(. Would be interested as well!
<dtz>
(for any version)
sanscoeur has joined #nixos
<betaboon>
anyone has a hint how to build uwsgi with additional plugins ?
johanot has joined #nixos
turion has quit [Ping timeout: 252 seconds]
Xyliton has joined #nixos
jperras_ has joined #nixos
jperras has quit [Ping timeout: 264 seconds]
<Xyliton>
How do I add patches to st? I tried adding ` st { patches = ["/etc/nixos/st_patches/some-patch.diff"]; }` to my systemPackages but nix then claims that it isn't of type package
<Xyliton>
Oh, thanks! Also, no quotation marks around the path?
jperras_ has quit [Client Quit]
<symphorien>
yes, this will make nix copy the file in the store
<symphorien>
otherwise, when building, the patch won't be found because of the sandbox
<Xyliton>
ohh, I see
revtintin has joined #nixos
<Xyliton>
hm
sanscoeur has quit [Remote host closed the connection]
<Xyliton>
I tried adding the dracula patch from the suckless site as-is, but that gives me "Hunk #1 FAILED at 84". Is the patch just outdated or do I have to rewrite parts of it to work with nix?
sanscoeur has joined #nixos
<acowley>
Do people use nixos-unstable-small rather than nixos-unstable? I've always used nixpkgs-unstable or nixos-unstable, but feel like maybe -small is a smarter choice at this point. Has anyone had things break on them because -small was small?
<Baughn>
acowley: -small is small because it only compiles and tests a subset of the OS. If you use anything that's not in -small, you'll have to compile it yourself; and -unstable will be behind -unstable-small specifically in the cases where those parts *don't work*.
<Baughn>
acowley: So no, you probably shouldn't use it.
peacememories has quit [Client Quit]
johanot has joined #nixos
alex``` has quit [Ping timeout: 260 seconds]
ubert has quit [Remote host closed the connection]
<ddellacosta>
is zfs a good choice for filesystem on a daily use work laptop? Seems like both it and btrfs have some benefits above and beyond ext4 but seems like zfs is more mature and well supported on NixOS? (been a while since I've used linux as my desktop, tips and suggestions appreciated)
Xyliton has quit [Quit: Lost terminal]
<Synthetica>
Could someone review #49595? Mostly relevant to Go.
<Baughn>
ddellacosta: ZFS works great on NixOS. BtrFS... um, I can't really recommend it anywhere.
<Baughn>
You have to explicitly enable it, and all mountpoints (that matter for anything started on boot, anyway) need to use mountpoint=legacy, but that's all. Otherwise it works just like any other filesystem.
<ddellacosta>
Baughn: gotcha, that seems to be the general consensus as I've poked around. Thanks for the tips.
<Baughn>
`boot.supportedFileSystems = ["zfs"]` will do. You can edit and nixos-rebuild in the installation image to enable it there.
<ddellacosta>
okay cool, good thing I asked because I didn't realize I had to do anything other than format the partition appropriately
<Baughn>
Adding that line is what gives you access to the ZFS tools and kernel driver. You won't be able to format the partition without it.
<{^_^}>
[nixpkgs] @domenkozar pushed to master « chromium: use jumbo builds to speedup compilation »: https://git.io/fxAzT
hyperfekt has joined #nixos
<Synthetica>
Jumbo builds 🎉
<ddellacosta>
Baughn: I'm guessing maybe the live cd has that baked in then, as I was (seemingly) able to format without it
<Baughn>
Maybe. That's new, if so.
<ddellacosta>
well, I'm still a NixOS newb so I'll report back once I actually have a clue, haha
sanscoeur has joined #nixos
<drakonis>
Baughn: facebook runs it apparently
<drakonis>
they're betting money on btrfs
<Baughn>
Doesn't change my opinion.
<Baughn>
Btrfs is immature. RAID modes other than mirror don't work, they don't have a thorough test suite, and I've repeatedly lost data to it.
Dedalo has joined #nixos
<ddellacosta>
Baughn: btw I was wrong, I was confusing xfs and zfs (in terms of what is possible to format from live cd)
<Baughn>
Facebook has the manpower to avoid its sharp edges, they have the scale to use clustered filesystems that put the redundancy above the btrfs layer, and they *can't use ZFS* due to the licensing.
<Baughn>
None of those are true for common users.
<teser>
srhb, symphorien hi, I've been testing what you suggested with bacula, but I get the following: http://dpaste.com/176SH7N ; it seems that the $out is not expanded when the "make install" is being run. I can share the current package?
<drakonis>
perhaps if they upstreamed code harder
<symphorien>
in the example I gave, it was $(out)
<symphorien>
not $out
sigmundv has quit [Ping timeout: 252 seconds]
<hyperfekt>
Does anyone know the reason config.boot.initrd.luks.devices is an attrset instead of a list?
<symphorien>
ah no sorry
<symphorien>
I am wrong
<symphorien>
you can use ${placeholder "out"} instead, then
<tilpner>
hyperfekt - It is actually
<tilpner>
"Type: list or attribute set of submodules"
<symphorien>
teser: I really doubt with-logdir is the right variable
<symphorien>
read the makefile of your software
<symphorien>
you cannot guess the name without reading this
betaboon has quit [Quit: WeeChat 2.2]
<teser>
symphorien, ah yes, will do.
<teser>
symphorien, one more thing, after updating the package from 9.0.6 version to 9.2.1, the nix-env / nix-shell are still pulling a 9.0.6 version from somewhere even through I've specified NIXPKGS with "export NIXPKGS=/home/user/nixpkgs" (where I'm editing bacula) and then "nix-env -f $NIXPKGS -i bacula". Any ideas why?
<symphorien>
did you change the hash ?
<teser>
symphorien, yes
<teser>
symphorien, the sha256 of the downloaded binary, yes
<hyperfekt>
tilpner: Yeah, I saw that, but the default is an attrset. I'm a bit confused on whether that means a list entry would be allowed as well (and merged?) and why an attribute is necessary at all, given that the name is repeated inside the submodule.
<tilpner>
hyperfekt - If it wasn't an attribute, how would you set options for a device from two different modules?
<ddellacosta>
folks if I want to enable zfs support at the time I'm installing, what do I have to do? Do I actually have to modify the live image I'm booting from somehow, or can I do it at the point I boot into the live image?
<ddellacosta>
I understand the basic instructions to be clear--I don't need help with that (setting boot.supportedFilesystems, etc.)
<ddellacosta>
oh jeez I loaded that up but failed to scroll lower in the page where it actually explains how to do that at install time, d'oh thanks again gchristensen
<gchristensen>
no worries, that page could use a bit of cleaning up I think
<ddellacosta>
also tips on setting up encryption are just what I wanted
<gchristensen>
like the advice about ashift advice is not necessarily good. also, ZFS encryption is not recommened by the ZFS team!
<ddellacosta>
oh really?
<gchristensen>
"Native encryption is not production ready, keep backups (but it works great)" from #zfsonlinux's /topic
<ddellacosta>
huh okay. Makes me wonder if I should use something else (ext4?) in that case
<ddellacosta>
not that I particularly mind living dangerously (i.e. keeping backups)
<gchristensen>
I use zfs on top of luks on one system
nschoe has joined #nixos
<ddellacosta>
okay, good to know. I guess I'll start by installing this and then figure it out from there
<teser>
symphorien, I have the new code that builds correctly, thanks for your help. One more thing: since the runtime directory will still be /var/log/bacula, do I need to create a wrapper around the bacula binary that actually creates the directory if it doesn't exists?
<symphorien>
teser: if it is meant to be a service, let the systemd unit file take care of this
<nschoe>
Hi everyone, I'm a bit confused by an error while trying to build a derivation. What's weird is that it used to work, and I haven't updated the git commit for the nix pkgs it build against. The log is here: https://pastecode.xyz/view/550e01b3 (this is the log for a `nix-shell` command, for a very simple `shell.nix` that only have `catkin` as buildInputs.
<nschoe>
I don't understand this issue with the "env.sh". Is there smth obvious I'm missing?
<teser>
symphorien, just curios: if this was a normal program, then we would need a wrapper where we would create the directory with "mkdir -p" right?
<symphorien>
yes, but mind that "program --help" would create the directory :/
<symphorien>
this is not great, so you need to parse the arguments to the program in the wrapper, and this is brittle
<symphorien>
so, better avoid it :)
<teser>
symphorien, so the program itself can create the necessary directory, but that happens only if it has the written functionality for it (or why would program --help create the directories)
MichaelRaskin has joined #nixos
jperras has quit [Read error: Connection reset by peer]
<symphorien>
teser: so I mean "if the wrapper run mkdir, then program --help would create the dir"
<symphorien>
s/so/no/
<{^_^}>
[nixpkgs] @vcunat pushed to master « chromium tests: inherit timeout from the package »: https://git.io/fxA66
<{^_^}>
#49347 (by sjau, 4 days ago, open): easysnap: init at 2018-10-28
<gchristensen>
why is sjau resolving unresolved things
<gchristensen>
also seems weird to package a 10 day old project in nixkgs
<gchristensen>
not that that hasn't stopped us before
<cransom>
this reminds me that i was going to update my cold hard drive backup while i didn't have to be in the same room for the heavy duty hard drive clicking. oh well.
<Synthetica>
gchristensen: s/hasn't/has/?
<gchristensen>
right
<drakonis>
packaging his own thing for backing up zfs
<hyper_ch>
gchristensen: do be precies, it's two shell scripts ;)
<drakonis>
two shell scripts :|
<gchristensen>
:')
<hyper_ch>
and on the weekend I have to figure out why my office server installation, that's nixos pre-18.03 unstable doesn't properly work anymore when I go to 18.09 or latest unstabled... it's a kvm host and after update it can't connect to anything except the router while the VMs still all work properly
<teser>
symphorien, I've modified the bacula service in my local nixpkgs, how can I now rebuild the system (I can't use nixos-rebuild build-vm, since I'm aready running a VM and can't turn it off now, but KVM conflicts with the current hypervisor). Let's say that I run the "nixos-rebuild switch", how can I ensure that the overwritten service from local nixpkgs is taken into account?
<teser>
symphorien, I'm not sure that would very well, since I'm not sure exactly what will be rebuild (probably every stale program in my local nixpkgs - even if I update to the 18.03, will it be out-of-sync and will rebuild some portion of the system). In any case, I think the best option is to overwrite the service similarly to packageOverrides, is that possible?
Church_ has quit [Ping timeout: 252 seconds]
<teser>
s/would/would work/
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<symphorien>
hum. maybe, but I don't know how.
<symphorien>
if you rebase your changes over the commit of your current system (see the output of nixos-version) then changes will be minimal
<clever>
teser: build-vm can work without kvm, but will be a lot slower without it
alienpirate5 has quit [Ping timeout: 252 seconds]
jD91mZM2 has quit [Quit: WeeChat 2.2]
<teser>
clever, how can I run it without KVM?
<clever>
teser: if /dev/kvm doesnt exist, it will silently not use kvm
worldofpeace has joined #nixos
erictapen has joined #nixos
<teser>
symphorien, the nixos-version reports "18.09.932.09195057114", is the "09195057114" a commit?
jsgrant has quit [Remote host closed the connection]
nschoe has quit [Ping timeout: 252 seconds]
<tokudan[m]>
trying to use docker on nixos: is there a way in configuration.nix to let docker run specific containers from specific images that I have gotten through pkgs.dockerTools.pullImage?
<chris|>
has someone managed to get dfhack working for dwarf-fortress on 18.09? On my fresh install using the full pack, it just complains about not being able to detect the df version (xml version missmatch) and just dies
alex```` has quit [Ping timeout: 252 seconds]
drakonis has quit [Quit: WeeChat 2.2]
random_yanek has joined #nixos
peacememories has joined #nixos
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos
vk3wtf has quit [Ping timeout: 252 seconds]
drakonis has joined #nixos
drakonis_ has quit [Ping timeout: 272 seconds]
o1lo01ol1o has joined #nixos
<o1lo01ol1o>
I have a couple haskell depedencies in an expression that generate multiple executables when built. They don't know anything about nix, but expect that those executables are somewhere in the PATH. What's the best way to have nix make these available to my top-level derivation?
<pareidolia>
But I get "error: attribute 'extend' missing, at /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/system/boot/kernel.nix:39:31
<{^_^}>
[nixpkgs] @dywedir pushed commit from @screendriver to master « yarn: 1.10.1 -> 1.12.1 (#49616) »: https://git.io/fxxe7
Ariakenom has quit [Read error: Connection reset by peer]
<thblt>
Hey, I have a strange (DNS?) issue on my NixOS laptop. The WiFi had a hiccup, and it seems the DNS cache somehow remembered some NXDomains, eg twitter: curl https://twitter.com complains that "curl: (6) Could not resolve host" but "host twitter.com" behaves correctly ("twitter.com has address 104.244.42.1" etc). Any ideas? Thanks!
zack_moe_ has quit [Ping timeout: 252 seconds]
booglewoogle has joined #nixos
<infinisil>
thblt: I think those 2 programs use different ways to resolve DNS, there is a recent blog post series about this, (trying to find it...)
<thblt>
infinisil: Sorry, I should I specified I'm aware host doesn't use the system's DNS cache, but I can't find a way to reset this cache.
<infinisil>
(Doesn't seem to have a second part :()
<thblt>
dig adds a strange warning only on broken domains ";; Warning: Message parser reports malformed message packet."
<infinisil>
thblt: No idea either, but this is generally nothing specific to NixOS, so probably people in #networking or ##linux could help better
<thblt>
infinisil: thanks, will try!
<{^_^}>
[nixpkgs] @obadz merged pull request #49423 → nixos/zerotier: binds to network-online.target to avoid the 1m30s … → https://git.io/fx7Fb
<{^_^}>
[nixpkgs] @obadz pushed to master « nixos/zerotier: binds to network-online.target to avoid the 1m30s timeout before kill on shutdown »: https://git.io/fxxvr
lostman has quit [Quit: Connection closed for inactivity]
booglewoogle has quit [Quit: Leaving]
c0bw3b_ has left #nixos [#nixos]
<ddellacosta>
folks, I'm trying to install nixos in a dual-boot configuration with an OEM Windows 10 install. I keep getting the error 'File system "/boot" is not on a GPT partition table'. The main difference is that I didn't create a new boot partition but used the already present one created by Windows ("/dev/disk/by-label/EFIBOOT"). Does anyone know what obvious thing I'm screwing up here? Let me know if this is not enough info
<ddellacosta>
oh that error is when I run `nixos-install` if that wasn't obvious
<srid>
what's a good software for setting up monitoring graphs just for local nixos machine (my laptop), over long-term?
<emily>
how did you partition the hard drive?
<emily>
you need to do it with a gpt-supporting partitioning tool, at least