<colemickens>
Do flake authors use `nix flake check`? It's too whiney about packages that don't eval on certain archs
<colemickens>
and it feels wrong to filter those out instead of letting the eval error surface to the user
itai has quit [Remote host closed the connection]
<infinisil>
Yeah same problem here. I'm just using `nix build .#checks` instead (assigning `packages.checks`)
Qwerky has joined #nixos
fuiltilt has joined #nixos
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos
Qwerky has quit [Ping timeout: 272 seconds]
<noonien>
are there any plans to make the `nix` command more pipe-friendly?
<{^_^}>
[nixpkgs] @primeos pushed to master « chromiumDev: Build without TFLite and apply a fix »: https://git.io/JtRIZ
<noonien>
for example, `nix search` is pretty annoying to use as opposed to the old version. it doesn't search all registries anymore, and it always assumes a terminal, which means it always outputs progress and colors, which makes it basically impossible to pass to grep/less
alex-v has quit [Remote host closed the connection]
alex-v has joined #nixos
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 52 commits to staging-next: https://git.io/JtRIR
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 53 commits to staging: https://git.io/JtRIE
rajivr has joined #nixos
andymandias has quit [Ping timeout: 265 seconds]
andymandias_ has joined #nixos
tsrt^ has quit []
<{^_^}>
[nixpkgs] @risicle opened pull request #111514 → [20.09] pythonPackages.bottle: add patch for CVE-2020-28473 → https://git.io/JtRIy
nicoo has quit [Ping timeout: 268 seconds]
nicoo has joined #nixos
steve-chavez has joined #nixos
<bennofs>
btw, is there any flag for nix build to return the final built store path? I often use $(nix-build something) in scripts, but that doesn't work with nix build since it doesn't show the final path?
<steve-chavez>
Hey guys. How can I negate the true value inside `lib.optionalAttrs !true {k = "val";}`? It gives a syntax error when using the `!` operator.
rb2k has quit [Ping timeout: 240 seconds]
<leo60228>
bennofs: i don't believe the text output of `nix build` is intended to be machine-readable, but passing `--json` will output JSON that contains the output path (among other things)
xps has quit [Ping timeout: 240 seconds]
<bennofs>
ah, but then I still have to pipe it to jq to extract the store path. Most of my use cases only want to know the final store path, without poluting the current directory with a GC root
redcedar[m] has joined #nixos
nuh^ has joined #nixos
rb2k has joined #nixos
<infinisil>
bennofs: I wouldn't recommend using the new cli in scripts, it's still experimental and will change
<samueldr>
well, it might just work without anyway if it's the only session?
<colemickens>
I think our Firefox perf slipped again. I think we're back to getting 'wayland' instead of 'wayland/drm'.
<colemickens>
I'm looking into it.
dxb[m] has joined #nixos
<lovesegfault>
do we have a generic updater yet for unstable pkgs with a github src?
<samueldr>
lovesegfault: do you mean something that r-ry//antm would run, and update to arbitrary latest revs, or something you as a user would run to work a bit faster?
<samueldr>
I wouldn't think automatically updating to arbitrary latest revs would be desirable tbh
<samueldr>
but the latter sounds good
<lovesegfault>
I mean, if the bot did it, that'd be cool
<lovesegfault>
but I meant something I can do easily
<lovesegfault>
I remember seeing a generic updater
m0rphism1 has quit [Ping timeout: 256 seconds]
<ccao001[m]>
<samueldr "well, it might just work without"> Yes, it is the only session.
<lovesegfault>
unstableGitUpdater is what I want I think
<ccao001[m]>
What is the minimum I need from the displayManager service?
<samueldr>
ccao001[m]: I was assuming either it would be the only session, or there would be a rescue session thing
<ccao001[m]>
* What is the minimum I need to enable from the displayManager service?
<ccao001[m]>
my rescue session is the tty ;)
<samueldr>
I meant, automatically added to the DM
<ccao001[m]>
ohh
steve-chavez has joined #nixos
<ccao001[m]>
samueldr: What was involved in getting nixos to run on the pinephone? I'd like to do the same for guix system
<ccao001[m]>
It looks like displaymanager just loops the login prompt after turning it on with dwm service
<samueldr>
ccao001[m]: it's an A64 platform, so basically it's just a fancy A64 device with fancy hardware
<samueldr>
ccao001[m]: so in theory, trivial if guix has good support for u-boot on Allwinner A64
<samueldr>
ccao001[m]: hmm... you'd need to get the logs of the X server to see
<ccao001[m]>
guix currently is able to boot on the pinebook pro
<ccao001[m]>
* guix system currently is able to boot on the pinebook pro
<samueldr>
with lightdm it's under /var/log/lightdm/... something like x-0.log
<samueldr>
ccao001[m]: not A64, but yeah, u-boot based, so depending on how the u-boot setup is, it's possible a lot cane be re-used
sigmundv__ has quit [Ping timeout: 264 seconds]
<samueldr>
can*
<matthewcroughan>
does anyone actually know how to solve the problem I proposed earlier with wireguard?
steve-chavez has quit [Ping timeout: 265 seconds]
<matthewcroughan>
if I do `networking.wg-quick.interfaces.dns = [ "192.168.4.1" ]` yeah, it'll use my server's dnsmasq, totally, great now I can have my remote server services available via wireguard. Problem, OOPS, can't get to nasa.astral now, because now the LAN DNS isn't in use anymore. Solution? Dnsmasq running on all devices. There must be a better way?
<slabity>
Is anyone able to build `vulkan-loader` on the any of the latest unstable channels?
<matthewcroughan>
nicolas[m]: how does that help?
<slabity>
For some reason I'm still getting pkg-config issues
<matthewcroughan>
That doesn't solve the problem I just proposed.
<matthewcroughan>
That doesn'\t even touch the problem I just asked about :D
<samueldr>
ccao001[m]: I don't know enough about Guix's organization to say for sure, but that pine64.scm file looks like it would be for an A64 device, so a good starting point
<nicolas[m]>
systemd-resolved handles the dns configuration so you don't have a dnsmasq running on all devices
<ccao001[m]>
samueldr: would I also need to package a new kernel?
<matthewcroughan>
nicolas[m]: except systemd-resolved doesn't run on Android
<ccao001[m]>
* samueldr: would I also need to package a new kernel for the pinephone?
<matthewcroughan>
it doesn't run on Windows or MacOS either.
<nicolas[m]>
matthewcroughan5: and can be configured to route domains through wireguard and certain packages
<matthewcroughan>
that's not the issue
<samueldr>
ccao001[m]: likely yes, megi's fork is basically the gold standard of things-not-yet-in-mainline for the pinephone... without it you won't go far
<matthewcroughan>
the issue is that if I use one DNS server, I can't use another.
<matthewcroughan>
And the only solution I can see to that is dnsmasq.
<matthewcroughan>
there has to have been a better solution made, it's 2021.
<nicolas[m]>
matthewcroughan: sorry, I assumed a Linux environment given this is the NixOS chat
<matthewcroughan>
The answer, probably, is to have everything on a public DNS server, yknow, cloudflare etc.
<matthewcroughan>
the internet.
<matthewcroughan>
But no, I don't want my services on the internet. I want them on my VPN. But this is very problematic.
<nicolas[m]>
can't you add extra entries to your resolvconf?
<matthewcroughan>
yeah, but that doesn't do anything, that's not how dns works.
<matthewcroughan>
If I add it to my resolvconf, nasa.astral (local machine on LAN) isn't going to magically respond.
<matthewcroughan>
You need dnsmasq, which will query BOTH dns servers and return the result from one based on configuration
K0kada has quit [Ping timeout: 246 seconds]
<matthewcroughan>
but again, I'm asking for a better way, because this is kind of upsetting.
ahmed_elgabri has quit [Ping timeout: 264 seconds]
nhs has joined #nixos
<lovesegfault>
What do I do if a package that's only distributed as a bin starts updating the same URL with new releases and thus changing the hash all the time
<lovesegfault>
Can we mirror it?
sss2 has joined #nixos
<lordcirth>
lovesegfault, what package? sure there's no permalink?
sss4 has quit [Ping timeout: 260 seconds]
<lovesegfault>
lordcirth: I'm positive. The package is roon-server. It's currently packaged (by me) with an older version that was the last one exposed on a proper versioned URL
<lovesegfault>
I opened an "issue" with the company but they haven't done anything about it
<lordcirth>
lovesegfault, Yeah, you probably won't have much luck getting proprietary software released sanely. And the license probably doesn't allow redistribution.
Darkmatter66 has quit [Ping timeout: 264 seconds]
<lovesegfault>
Well, guess I'll update it with the always-breaking-hash and try to stay on top of it
bingobango has quit [Ping timeout: 264 seconds]
sss2 has quit [Quit: Leaving]
sss2 has joined #nixos
sss2 has quit [Read error: Connection reset by peer]
<charukiewicz>
if I'm writing my own nixpkgs overlay for that produces a package with stdenv.mkDerivation, should I leave runHook preInstall/runHook postInstall in my custom installPhase?
freeman42x[m] has joined #nixos
cyphase has joined #nixos
<freeman42x[m]>
I am trying to play Factorio on NixOS and even though sound is working for stuff played in Chrome it is not playing any sound in the game. Any suggestion for what to try to fix it?
<freeman42x[m]>
Factorio is installed via Steam
alex-v has quit [Ping timeout: 256 seconds]
<notgne2[m]>
did you switch sound devices while the game was running? that screws it up sometimes in my experience
<freeman42x[m]>
no, I switched to bluetooth headset before starting it because my integrated speakers are dead (HW issue probably)
<freeman42x[m]>
don't really know what to do, I checked audio settings via right click of volume icon in tray and there is no option there that could fix it I think
knupfer has quit [Read error: Connection reset by peer]
<freeman42x[m]>
is padsp package on NixOS? can't find it
<notgne2[m]>
freeman42x: it looks like the `padsp` binary is in the pulseaudio package
<freeman42x[m]>
ah, so installed by default on NixOS?
<V>
freeman42x[m]: there's very little that's "installed by default", but you'll have it if you set hardware.pulseaudio.enable or added pulse to one of the paths
ahmed_elgabri has joined #nixos
<freeman42x[m]>
nah, tons of stuff installed by default lol. but in this case it happened to be based on that setting
nhs has quit [Ping timeout: 264 seconds]
nolash has joined #nixos
<freeman42x[m]>
hmmm, so I tried: steam steam://rungameid/427520 and that does not fix it
<V>
freeman42x[m]: no, there's very little installed by default. the "tons of stuff" is largely based on the config options you set
<pie_>
does niv pin weird versions or something? when I try to add the graphic``` ( (import ./sources.nix {}).nixpkgs + "/nixos/modules/profiles/graphical.nix") ``` to my imports it wants to build half the world
ambro718 has joined #nixos
<pie_>
and im not sure if its my fault but i dont see why it would be
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to commercial-page: https://git.io/JtRal
EMayej has quit [Quit: Ping timeout (120 seconds)]
cptrbn has joined #nixos
sangoma has joined #nixos
vidbina has quit [Ping timeout: 264 seconds]
Qwerky has joined #nixos
<yorick>
pie_: is just pins master
<yorick>
pie_: or whatever branch you had
<yorick>
pie_: but master isn't nixos-unstable ;)
cosimone has joined #nixos
<pie_>
yorick: nah im specifically using -unstable and stuff
Qwerky has quit [Ping timeout: 256 seconds]
<pie_>
i _just_ figured out something might be going on related to my NIX_PATH
<pie_>
i thought id overridden it properly but maybe not
<pie_>
(i mean with the pinning)
<pie_>
(but even then it should be on unstable)
<{^_^}>
[nixos-homepage] @garbas pushed to commercial-page « display logo »: https://git.io/JtRVQ
Lord_of_Life_ has joined #nixos
Lord_of_Life has quit [Ping timeout: 265 seconds]
Lord_of_Life_ is now known as Lord_of_Life
nhs has joined #nixos
<freeman42x[m]>
I have services.zerotierone.enable = true; and services.zerotierone.joinNetworks = [ "0dccb453f7c87ab3" ]; in my configuration but my computer does not appear as connected in the zero tier one dashboard. any ideas why it is not connecting to the VPN?
<attila_lendvai>
i want to work on a program that compiles with -nostdlib, i.e. the makefile needs to access stuff like crt1.o when linking. could someone give me any hint what to look at for an example? or general guidance on how to approach at this on nixos? i cannot find anything helpful.
<dhess>
Therefore I'd like to run an Ubuntu/Debian VM on a NixOS host. What's the best way to declaratively manage that VM in NixOS? There are a lot of virtualization services in NixOS!
<dhess>
My current thinking is to use the libvirtd service, but does anyone prefer a different one?
<dhess>
(It cannot be a container, as the GitHub Actions Runner doesn't work well in a container.)
<yorick>
dhess: why not just package the github actions runner?
<__monty__>
,locate crt1.o
<{^_^}>
Found in packages: musl, glibc, uclibc, emscripten, glibc_multi, graalvm8-ce, graalvm11-ce, glibc_memusage
fendor_ has joined #nixos
<__monty__>
attila_lendvai: Pm {^_^} with such queries to figure out which packages provide the file and then add them to buildInputs one by one? Not sure whether that's enough though might need to set something like LIBDIR or somehow pass their locations to configure.
bukkitgerman8608 has joined #nixos
nhs has joined #nixos
<siraben>
How can I override doCheck for an individual Haskell package?
<yorick>
siraben: haskell.lib.dontCheck
schmim[m] has joined #nixos
<siraben>
yorick: thanks
<dhess>
yorick: It's packged as a big binary blob. It's not clear how long it would take to package it, let alone test and debug it, and I have finite time for this project. So this solution is less than ideal, but it should only take a few hours, at most.
<siraben>
didn't know that existed, I tried to use an overlay and set doCheck to false but it didn't work
<attila_lendvai>
__monty__, thanks, that helps moving forward. but is there a command line invocation that can get my a path from the nix-store? e.g. something along the lines of nix-store --query glibc.out/lib/crt1.o ?
<__monty__>
Looks like the proper way to me. You shouldn't hardcode any paths because any updates to the packages or their dependencies will change the path you need to use. So you should use `${pkgs.glibc}/lib/crt1.o`.
hiro98 has joined #nixos
* attila_lendvai
has realized he can put "cp ${pkgs.glibc}/lib/crt1.o ." into the shellHook in his shell.nix
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JtR6C
pheoxy has joined #nixos
<{^_^}>
[nixpkgs] @onsails opened pull request #111546 → vim: fixsplitjoin-vim and sideways-vim branches, add nvim-hlslens, vim-apm, train-nvim, neogit → https://github.com/NixOS/nixpkgs/pull/111546
<s1341>
has anyone succesfully built a cross-compilation nix-shell for android aarch64?
nasirkhan has quit [Client Quit]
<pheoxy>
hello there
<illustris>
/run/keys on nixos LXCs appear to be mounted ro, causing nixops deploy to fail. Does anyone know why it's ro, or what can be done to make it rw by default?
<pie_>
yorick: yeah i dont know. for example, its buildin glib.
<pheoxy>
on cross-compilation s1341 anyone know what I'm doing wrong with nix.buildMachines.sshKey that I keep getting permission denied?
<pheoxy>
what do you mean by android aarch64? they are the same thing as aarch64. The only difference is if you dealing with existing libraries and trying to bring something into that
<s1341>
Mic92: any chance you'd be willing to take a quick look?
<Mic92>
s1341: maybe give a link in the evening
<Mic92>
doing other stuff right now
<s1341>
Mic92: ok....
<clever>
Mic92: do you know much about gcc and multilib?
hiro98 has quit [Ping timeout: 256 seconds]
<Mic92>
clever: let's say I know stuff about libc and assembly: https://github.com/Mic92/nix-ld most of the time I work on x86_64 so.
<clever>
Mic92: ive patched the arm-embedded cross stuff, to create a multilib build (thumb+arm, soft+hard), and newlib automatically went multilib as well
<clever>
but the ld wrappers in nixpkgs arent searching the right dirs for newlib
<Mic92>
maybe it could be hacked into the cc-wrapper
<clever>
Mic92: the gcc is already able to produce both arm and thumb code, its just a matter of pre-compiled bits (newlib and crtbegin) being pre-built in both forms
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JtRMb
oldsk00l has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
<Mic92>
clever: so maybe it is just a matter of our binutils wrapper to add those directories by default.
ahmed_elgabri has quit [Ping timeout: 260 seconds]
<clever>
Mic92: the complex bit, is that where the linker should search, depends on the -march and --float? options
philr has quit [Ping timeout: 240 seconds]
<clever>
gcc already has some internal logic to pick the right one, and to do things right, ld needs to respect the same logic
<Mic92>
clever: maybe based on some value in lib/systems/platforms.nix ?
<clever>
Mic92: my guess, is that ld and gcc search something like /usr/lib/<suffix> automatically, and all of the buildInputs and -L magic on nix is the problem
<clever>
when you -Lfoo, it wont search foo/<suffix>, i think
<enteee_>
was something bigger changed in how fonts are handled that I missed? After an upgrade of unsable my urxvt no longer picks up any fonts I have installed.
<clever>
immae: is there a way to set the nixpkgs input to a given rev?
<angerman>
can I somehow run "just" the postInstall phase?
<immae>
clever: if you call it from the flake then you have the "follows" syntax
<angerman>
something is busted in there and I'm trying to debug what exactly is wrong.
<immae>
s/the/a/
<immae>
I’m not sure about the question though
<angerman>
$postInstall sadly won't work :-/
<clever>
immae: what if i'm converting a project from niv to flake, and nixpkgs master is broken, so i want flake.lock to start on a specific rev, and then figure it out later?
aveltras has joined #nixos
<immae>
clever: Ah, then you need to specify the rev you want in your flake.nix to "fix" it
<DigitalKiwi>
angerman: why won't that work
<angerman>
DigitalKiwi: because there are "#" markers in there :-/
<immae>
clever: do you need an example or you figured out?
<clever>
immae: i think i'll just keep editing flake.lock by hand, thats worked so far
<clever>
i need it to start on a given rev, but still allow updating the rev like normal
<immae>
ok this works, but it won’t survive an accidental "nix flake udpate" :)
<immae>
ok
<clever>
thats what git is for
sangoma has quit [Quit: WeeChat 3.0]
<immae>
It depends on the information you want to keep indeed. If you put it in flake.nix you basically say "I need this flake to be bound to that commit of that repo", but if you put it in flake.lock then it becomes ephemeral
<DigitalKiwi>
angerman: is it a flake? i think nix develop has a flag
<angerman>
I'm afraid not.
<Nickli>
how have you setup your git with nix configs symlinks, bare or stow?
sangoma has joined #nixos
mmohammadi9812 has quit [Ping timeout: 264 seconds]
<{^_^}>
[nix] @edolstra pushed to debugger « Rename to --debugger »: https://git.io/JtRAo
<s1341>
Ericson2314: can we chat in private? I'm not really comfortable with nix derivations to do this on my own and could use your guidance...
<Ericson2314>
s1341: sure
nhs has joined #nixos
Lord_of_Life_ has joined #nixos
Lord_of_Life has quit [Ping timeout: 256 seconds]
lemsip has joined #nixos
<lemsip>
has anyone experienced intermittent issues with their mouse's middle button? It seems to work intermitently for me, where I can see input events via evtest for the middle click when I first connect it, and then a while later they just don't come through. All other buttons just work, nothing funny in the logs besdies libinput complaining about events arriving late every once in a while.
<acker>
Hello, I'm trying to start gitlab by manual, but for some reason it doesn't start. I see errors with gitlab-sidekiq.service in journal, but gitlab logs directory is empty. I didn't install nginx, but on CentOS all works even without web server.
<acker>
Can someone explain me what is my problem?
<kaba>
Hi, I have some trouble installing cxxfilt , a python package. pypi2nix fails with TypeError: 'PrefetchedRepository' object is not subscriptable. Am I doing something wrong?
domogled has joined #nixos
jonringer has quit [Remote host closed the connection]
<simonpe^^>
How do you generally expand "$@" inside a string? Say for example I want to run (the properly escaped version of) `nix-shell --run "app "$@""`?
rb2k has quit [Ping timeout: 260 seconds]
jonatanb has joined #nixos
rb2k has joined #nixos
sumner has joined #nixos
<simonpe^^>
before I've used something like `printf "%s\n" arg1 arg2 | nix-shell --run "xargs app"` but that is a bit nasty and if I want to `exec` it I'm not sure how the pipe impacs `exec`
<lordcirth>
simonpe^^, I think you want nix-shell --run 'app "$@"'
waleee-cl has joined #nixos
nhs has quit [Ping timeout: 240 seconds]
<simonpe^^>
lordcirth: won't the single quotes make the $ be passed literally?
cptrbn has joined #nixos
<lordcirth>
literally to nix-shell, but then they should be stripped by nix-shell
<{^_^}>
[nixpkgs] @dotlambda pushed to release-20.09 « libsixel: mark as insecure »: https://git.io/Jt0vJ
<simonpe^^>
try making a file test.sh with the contents `nix-shell --run 'echo args "$@"'`. The expected behavior if I run `./test.sh hello` would be `args hello`, but it is simply `args`
<mikroskeem>
i got rid of mine some time ago, it was more pain to manage than it was useful
pinage404[m]1 has joined #nixos
pinage404[m]1 has quit [Client Quit]
<Mic92>
siraben: well. just my own server. More for getting to know with kubernetes
<Mic92>
kubenix is quite nice
pinage404[m]1 has joined #nixos
est31 has quit [Ping timeout: 268 seconds]
est31 has joined #nixos
<mikroskeem>
but yeah kubernetes needs to force 'systemd.enableUnifiedCgroupHierarchy = false;' as well
Compilador has joined #nixos
<Mic92>
mikroskeem: are you sure? It seems to work fine
<mikroskeem>
did you configure your system to use cgroups v2 explicitly?
<Mic92>
/sys/fs/cgroup is cgroup v2
<mikroskeem>
hmm
<Mic92>
It seems to rely on systemd now to manage cgroups
<Mic92>
Just like docker
hyper_ch has joined #nixos
<mikroskeem>
i see
<Compilador>
Using buildRustPackage seems to break on any project using jemalloc, and I'm totally stumped as to why. I had a nix expression that already worked
<Compilador>
lordcirth You're right, only missing -x, though I wonder why this worked before, I've been building/deploying this nix expression for a while
cptrbn has quit [Quit: Gooooooood niiiiiiiiiight]
<lordcirth>
Compilador, how is the source fetched? fetchTarball? fetchFromGithub?
vidbina has quit [Ping timeout: 265 seconds]
<Compilador>
lordcirth, fetchFromGithub, but the dependency itself must be being pulled by cargo as part of buildRustPackage I think
<elux>
ive recently upgraded my motherboard's bios, and the system feels snappy and great.. except, im seeing every few hours in my system logs: [Hardware Error]: Corrected error, no action required., [Hardware Error]: CPU:0 (17:71:0) MC25_STATUS[-|CE|MiscV|-|-|-|-|CECC|-|-|-]: 0x98004000003e0000, [Hardware Error]: IPID: 0x000100ff03830400, Hardware Error]: Platform Security Processor Ext. Error Code: 62 ... and
<elux>
then the system goes into a full reboot.. any thoughts? :/
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<elux>
i have `hardware.cpu.amd.updateMicrocode = true;` and `hardware.enableRedistributableFirmware = true;` in my nix config... maybe i should remove these..?
vidbina has joined #nixos
<Compilador>
lordcirth, so it's weirder than I thought, those files are produced by `configure` during the build step, so there should be no reason whatsoever they are not +x
<lordcirth>
Compilador, does ./configure change umask?
<Compilador>
lordcirth, Ah no I'm wrong, it doesn't, if I pull out that specific dependency from the dir at -K, and run configure manually, I get the same errors
<Compilador>
Still begs the question why `cargo build` in a shell works fine
<lordcirth>
Compilador, I wonder if something is changing the umask and preventing +x
fendor__ has quit [Remote host closed the connection]
<Compilador>
lordcirth, I'm not skilled enough in nix tooling to debug it
<Compilador>
But I did realise the package in the workspace depending on jemalloc-sys is one I don't need, so I'll just see if I can work around this by patching out that item
<Compilador>
Fingers crossed
<lordcirth>
elux, you can try, but generally those should improve stability
eeva has quit [Remote host closed the connection]
<lordcirth>
elux, what's your CPU and motherboard? How long ago was the bios update published? Sure it's not a BIOS bug?
<Ericson2314>
gchristensen: we made a #nix-android, what is that magic thing you do with the channel namespace again? :)
<Ericson2314>
(oh right, you need ops in it, and the person that made it is going to sleep, so no rush)
<gchristensen>
it is automatically part of the NixOS group registration since it is within the reservedprefix, so if there is an issue we can contact freenode staff and get access :)
<gchristensen>
so, no worries, all set
simonpe^^ has quit [Remote host closed the connection]
joesventek has joined #nixos
lsix has quit [Ping timeout: 240 seconds]
packer has joined #nixos
malook has joined #nixos
roebuck[m] has quit [Ping timeout: 244 seconds]
notgne2[m] has quit [Ping timeout: 244 seconds]
littlebenlittle[ has quit [Ping timeout: 244 seconds]
lytedev[m] has quit [Ping timeout: 244 seconds]
mcbits[m] has quit [Ping timeout: 244 seconds]
ejpcmac has quit [Ping timeout: 244 seconds]
govanify has quit [Ping timeout: 244 seconds]
veleiro has quit [Ping timeout: 244 seconds]
nikivi[m] has quit [Ping timeout: 244 seconds]
dunc4n has quit [Ping timeout: 244 seconds]
NaN-irc_ has quit [Ping timeout: 244 seconds]
woobilicious[m] has quit [Ping timeout: 244 seconds]
McSinyx[m] has quit [Ping timeout: 244 seconds]
willghatch has quit [Ping timeout: 244 seconds]
mindtree[m] has quit [Ping timeout: 244 seconds]
tcakoi has quit [Ping timeout: 244 seconds]
thefloweringash has quit [Ping timeout: 244 seconds]
lambdaclan has quit [Ping timeout: 244 seconds]
yurb has quit [Ping timeout: 244 seconds]
nightmared has joined #nixos
NaN-irc has joined #nixos
pheoxy has quit [Ping timeout: 244 seconds]
cwNovusordoseclo has quit [Ping timeout: 244 seconds]
test1[m] has quit [Ping timeout: 244 seconds]
spinlock[m] has quit [Ping timeout: 244 seconds]
totoroot[m] has quit [Ping timeout: 244 seconds]
nickbublik[m] has quit [Ping timeout: 244 seconds]
miangraham[m] has quit [Ping timeout: 244 seconds]
iuiiik[m] has quit [Ping timeout: 244 seconds]
jbal[m] has quit [Ping timeout: 244 seconds]
codyopel has quit [Ping timeout: 244 seconds]
samir420[m] has quit [Ping timeout: 244 seconds]
a7p[m] has quit [Ping timeout: 244 seconds]
ragge42[m] has quit [Ping timeout: 244 seconds]
gahha[m] has quit [Ping timeout: 244 seconds]
andrey1978[m] has quit [Ping timeout: 244 seconds]
CodeBitCookie[m] has quit [Ping timeout: 244 seconds]
figsoda[m] has quit [Ping timeout: 244 seconds]
maximumtix[m] has quit [Ping timeout: 244 seconds]
rb2k has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @prusnak pushed to master « cjson: fix build on darwin »: https://git.io/Jt0tn
malook has quit [Quit: malook]
<elux>
lordcirth: im on a Gigabye Aorus Master x570, and ive been having this issue since i upgraded to F33a bios yesterday.. so its definitely related to that.. and ive found reports of other users..
nahamu has joined #nixos
<elux>
the system feels very stable and fast.. but it happens everytime my system goes to sleep i find.. when i leave it from activity.. is there a way to prevent it from reducing power or whatever its doing when going to sleep..?
malook has joined #nixos
<lordcirth>
elux, you can probably disable various kinds of powersave in BIOS
malook has quit [Client Quit]
malook has joined #nixos
<elux>
ok nice. i will try to look those up.. any kinds of words i should look for?
<niksnut>
I think our CI still tests the ISO on i686-linux
<bennofs>
usually you could look it up on hydra but seems hydra's web interface is having some trouble right now
<niksnut>
(onAllSupported "nixos.iso_minimal")
<niksnut>
in nixos/release-combined.nix
<bennofs>
ugh I don't remember if "all supported" includes i686
<bennofs>
i remember the names of those sets being not exactly obvious
jlv has joined #nixos
<SomeoneSerge>
Eh, I'm rebuilding my homeenv (from the same nixpkgs I built it with yesterday, I believe) and suddenly it starts to build cachix from scratch. And that build fails
<SomeoneSerge>
Is that even supposed to happen?
<SomeoneSerge>
I.e. cachix not being fetched from cache?
rb2k has quit [Ping timeout: 260 seconds]
<matthewcroughan>
bennofs: what are the -small images for?
simba1 has joined #nixos
<bennofs>
matthewcroughan: you mean like https://channels.nixos.org/nixos-20.03-small? They require smaller sets of packages to pass tests and be built before they are published
<SomeoneSerge>
I think a tool visualizing cache misses could be super useful - smth, hmmm, smth that would know which versions of derivations _are_ in the cache and what is the least change in deps required to hit some of those versions
<vrinek>
Hello! One build of qutebrowser on 20.09 works fine, but a subsequent build of it (same version, different hash) does not. I have gathered enough info about the exact nature of the problem, but I do not have the skills to fix it myself. Should I open a Github issue about this?
sigmundv has quit [Read error: Connection reset by peer]
Qwerky has quit [Remote host closed the connection]
Acou_Bass has quit [Ping timeout: 264 seconds]
derpadmin has joined #nixos
sangoma has quit [Read error: Connection reset by peer]
<jgart[m]>
<ehmry "I avoid singleton if I don't nee"> Ok, so unless you need to reuse it or pass it somewhere else then you avoid the sugar :) I'll keep that in mind
fresheyeball has joined #nixos
<fresheyeball>
hey out there
<fresheyeball>
I am using nix to generate an nginx.conf file
<fresheyeball>
and I need to include mime.types
upendra has quit [Ping timeout: 248 seconds]
<aveltras>
im having an awful network bandwith and have encountered what looks like data corruption when trying to enter a nix-shell on a project, is there a way to "clean" it ?
<aveltras>
the thing to keep in mind is that you should first add the binary cache and rebuild before adding the packages depending on it or the cache won t be used
marcusr has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @prusnak pushed to master « pythonPackages.hwi: drop dependencies that are bundled »: https://git.io/Jt0le
dandart has joined #nixos
fuiltilt has joined #nixos
marcusr has joined #nixos
ahmed_elgabri has joined #nixos
Acou_Bass has quit [Ping timeout: 258 seconds]
logand``` has quit [Ping timeout: 246 seconds]
ahmed_elgabri has quit [Ping timeout: 258 seconds]
attila_lendvai has quit [Ping timeout: 264 seconds]
dnlkrgr has quit [Ping timeout: 246 seconds]
<bingobango>
morning, if i have a postgres database, and i want to completely clear it back to normal (such that my initialScripts will run again next build from fresh), whats the correct procedure for doing so?
<aanderse>
infinisil: gchristensen reasonably convinced me that `ensure*` options are "bad". the more time that passes the stronger i agree with his position. i'm trying to remove the `services.mysql\postgresql.ensureDatabases` and `services.mysql\postgresql.ensureUsers` options and replace them with something a little bit less damaging
<aanderse>
the PRs i have made so far haven't had much traction
<infinisil>
Hmm I see
<infinisil>
aanderse: What's the replacement?
<aanderse>
had a couple options so far
<aanderse>
most recently i'm thinking a `services.mysql.activationScripts` is probably the "best" (least damaging?) idea, wher this option somewhat mimics user activation scripts... except run by the `mysql` option
<{^_^}>
#107342 (by aanderse, 6 weeks ago, open): nixos/mysql: replace initialDatabases, ensureDatabases, and ensureUsers options with activationScripts option
<fresheyeball>
how does one flush dns on nixos?
<clever>
fresheyeball: restart nscd.service
Acou_Bass has quit [Ping timeout: 256 seconds]
<fresheyeball>
clever: got it
<infinisil>
aanderse: Hmm..
<infinisil>
This is similar to the problem of upgrading postgresql's version
<aanderse>
feel free to say terrible things about it
<aanderse>
at this point i'm just looking for a way forward... any way forward :)
thelounge52 has joined #nixos
<infinisil>
I'm not a fan of having more activation script-like things, but I also don't like the ensure* options really
zebrag has quit [Quit: Konversation terminated!]
<infinisil>
What I think would be a good solution is to have the concept of state migration be more fundamental to NixOS
K0kada has joined #nixos
zebrag has joined #nixos
<aanderse>
this isn't specifically for state migration
<aanderse>
its generic
<aanderse>
the user can do what they want
<infinisil>
I previously talked about having a per-module state version, where module authors could then include migration scripts or walk-throughs
<aanderse>
its a shotgun...
<aanderse>
but we need basic database provisioning in nixos modules or we end up having a ton of manual steps
<aanderse>
and ensure*, aside from being a bad idea, doesn't work
<aanderse>
there are a ton of modules which use pgsql but have to specify a different encoding
<infinisil>
Hmm though I guess this doesn't quite fit into the same state migration thing
<infinisil>
Yeah, lemme think..
<aanderse>
so they write really minimal systemd services
<aanderse>
which lack the sandboxing the postgresql module has
<aanderse>
(or could have? should have?)
<aanderse>
and one module had to do the same for mysql now
<bennofs>
i think some stage migration helpers based on systemd units would benice
vifon has left #nixos ["WeeChat 2.9"]
<aanderse>
`services.mysql.activationScripts` is also "nice" because then users can (ab)use it in ways they see fit, given they have been warned
thelounge52 has quit [Ping timeout: 240 seconds]
<infinisil>
Actually this does fit into a migration scheme
<Siyo>
i think it does fit into state migration. you could potentially think about it like database migrations in those database libraries. Migrating state and nix store would be separate things. Writing migrations should be append only, each migration having a version number
<infinisil>
Yeah ^
<infinisil>
The main problem with the ensure* options is that once it's created, any change to the option can't be applied anymore
<infinisil>
So the nice solution is to also encode migration paths into it
<aanderse>
yeah well you're smarter than i am so please do think about this problem 😃
<infinisil>
Nah I'm not smarter, we just have difference specializations :)
<infinisil>
Unfortunately I don't have a lot of time to work on nixpkgs, I wish I had..
<infinisil>
But I think having a generic way to encode state migrations is the way to go
<aanderse>
awesome
<aanderse>
let me know if you come up with anything
<infinisil>
In fact I think that's a problem that pops up almost everywhere with Nix
<infinisil>
Updating a nix profile is just a migration from generation n to n + 1
<infinisil>
Rollbacks are just the reverse (if possible)
<infinisil>
Anything that is managed statefully can have a notion of a version
<infinisil>
The main NixOS activation script is really just a state migration tool
<infinisil>
Deployment tools too
<Siyo>
I think you'd want them separate in practice though. You might want to rollback a generation, but leave the state alone
Acou_Bass has joined #nixos
<infinisil>
The state of NixOS generations is just "which generation is the active one"
<infinisil>
Oh, and I guess once it calls the switch script, the activation script has the state of the whole system
dps910 has joined #nixos
<Siyo>
i meant the database state for instance
<infinisil>
Siyo: But yeah, that's the thing about state, we can't just recreate it, we need to deal with the existing one, which is where migration stuff comes into play
<dps910>
anyone use macOS?
fendor__ has quit [Read error: Connection reset by peer]
<dps910>
anyone use macOS? I can't build iterm2. Idk how to do code blocks since I'm new to irc but hopefully this is fine
<lovesegfault>
ashkitten: /nix/store/m2nb3lhhsmn1ddpszf4db3rhgbwl7mp4-binutils-2.35.1/bin/ld: /build/build/./gcc/liblto_plugin.so: error loading plugin: /nix/store/xfd2dkxg0ha8z3a517ag2jvf6gwnplim-bootstrap-stage0-glibc/lib/libc.so.6: version `GLIBC_2.33' not found (required by /build/build/./gcc/liblto_plugin.so)
<lovesegfault>
that's the current error
<lovesegfault>
when building gcc10
<ashkitten>
this is with your updated glibc?
<lovesegfault>
yup, on the branch I pushed with no other changes
<lovesegfault>
in partnership with the ghost army from lord of the rings
<lovesegfault>
and the army of the dead from game of thrones
<lovesegfault>
more seriously: no idea
<matthewcroughan>
lovesegfault: lol
<Yaniel>
that's a lot of maintainers
<matthewcroughan>
but yes, it really doesn't work
<matthewcroughan>
18.03 works, anything later hangs at systemd
<matthewcroughan>
20.09 hard crashes the system at boot with: Assertion 'clock_gettime(map_clock_id(clock_id), &ts) == 0' failed at src/basic/time-util.c:53, function now(). Aborting.
<matthewcroughan>
20.03 just hangs and never finishes initialisation of all system services
<lovesegfault>
matthewcroughan: AFAIK the state of it is largely unmaintained
<lovesegfault>
I have advocated to just getting rid of it for a while
<lovesegfault>
b/c having a broken thing is worse than not having it at all IMHO
simba1 has quit [Ping timeout: 264 seconds]
<lovesegfault>
niksnut Ericson2314 andi- flokli: how do I regenerate the bootstrap binaries?
<lovesegfault>
I think it will be necessary to bump glibc
<andi->
Isn't that usually done after the bump once hydra has built it?
<lovesegfault>
maybe? I have no idea! :D
<lovesegfault>
I'm attempting bootstrap again to see
<lovesegfault>
Yeah, same issue above
* lovesegfault
scratches head
<lovesegfault>
Trying to interpret the error:
<lovesegfault>
/nix/store/xfd2dkxg0ha8z3a517ag2jvf6gwnplim-bootstrap-stage0-glibc/lib/libc.so.6: version `GLIBC_2.33' not found (required by /build/build/./gcc/liblto_plugin.so)
<lovesegfault>
it seems to be looking for GLIBC_2.33 in the boostrap binary
<lovesegfault>
but that won't be the new glibc, it will be the old one AIUI?