<clever>
infinisil: "${./.}/foo" will first copy all of ./. into the store, then append foo, resulting in /nix/store/hash-src/foo
<infinisil>
haha I guess i just have an example of it right here
<clever>
infinisil: while ./. + "/foo" will just do the exact same thing as ./foo
<tertleeltret>
wow its works
<samueldr>
jordanmackie: looks like your installer image is "not right"
<tertleeltret>
how old is this project?
<clever>
infinisil: and if you later "${./. + "/foo"}" then you get /nix/store/hash-foo/
<infinisil>
clever: Ahhh, righttt
<infinisil>
string expansion will import it to the store and use that path
<samueldr>
jordanmackie: have you verified the .iso using the sha-256 provided on the download page?
<samueldr>
and jordanmackie, could you tell me the full name of the iso image you have used?
<ottidmes>
clever: nice, I was looking for path interpolation, I just converted them to strings whenever I needed to do this
<shlevy>
thoughtpolice: time for #nixos-risc? :D
<samueldr>
(I want to verify too if the image itself is wrong or if you're mainly unlucky with your download)
<clever>
infinisil: ive also had related issues, where importing ./. hauls in several gig of junk
<clever>
while properly telling nix i just want ./foo, skips that
<shlevy>
I'm adding the riscv qemu variant to nixpkgs now :)
<thoughtpolice>
shlevy: :) I've been meaning to get QEMU up and running but, deadlines, etc
<infinisil>
clever: yeah, builtins.filterSource is also useful for this
<shlevy>
Can't decide if I want to focus on cross or native... Guess we'll see how slow compilation is on softmmu :D
<clever>
infinisil: another weird bug, ./. + "/" is a no-op
<clever>
infinisil: which means ./. + "/" + filename is broken
<thoughtpolice>
shlevy: Does PXE/NFS root mount work these days? I think so, right? While those machines are pretty nice in terms of compute, the storage will be a pain point until they release the FMC expander with SSD or whatever.
<clever>
after every append, it parses the path, and strips out elements that arent needed
<jordanmackie>
samueldr: image was downloaded earlier this evening, and hard to checksum it now with my laptop in its current state
<samueldr>
jordanmackie: minimal or graphical?
<infinisil>
clever: haha, left-associativity of + and /root// = /root/
<shlevy>
thoughtpolice: I haven't tried it lately but I don't see any reason why it wouldn't
<jordanmackie>
samueldr: minimal
<samueldr>
(I'm guessing 64 bits)
<clever>
infinisil: it turns into /root
<jordanmackie>
64 bits
<samueldr>
thanks jordanmackie
<clever>
infinisil: and adding a / gives you /root/, which is shortened to /root again
<infinisil>
clever: ah yeah
<srk>
thoughtpolice: I'm running netbooted hydra slaves with overlayfs, no nfs tho
<samueldr>
(I'm downloading and verifying whether a bad image is on the server or if you're unlucky)
<shlevy>
Hmm is there a better way to track work than github milestones? We need a trello or something.
<jordanmackie>
samueldr: nice one, thankyou
<thoughtpolice>
shlevy: So, that'd be useful. I'm also unclear on whether QEMU or hw will work better, but I imagine QEMU for now, HW when it's here, of course. I was thinking at first if NixOS isn't working reliably in a simulator, it'd at least be good to bootstrap Nix there and use it as a remote builder.
<infinisil>
Alright, now I can do a normal emacs build which puts (load "/nix/store/.../foo.el") stuff in init.el or a debug build which uses (load "/path/to/my/config/foo.el") instead
<thoughtpolice>
Sonarpulse is also doing all this cross compilation stuff with bgamari, which I'm not sure how that's shaken out in upstream lately. According to bgamari recently it's apparently in pretty good shape.
<thoughtpolice>
e.g. he can reliably and easily cross compile an ARM sdimage, even with GHC, etc
<clever>
thoughtpolice: ive also got nixos expressions that can PXE boot nixos, without relying on the hdd ay all
<shlevy>
Yeah, so my thought is a) get everything up and running with the pre-compiled kernel images etc., b) try to get at least nix-daemon --stdio working, c) try to cross-build a base system, d) try to native-build nixos
<thoughtpolice>
srk: That's good to know, thank you! You too clever
<clever>
thoughtpolice: this fully configures a netboot server for PXE, and serves a nixos installer that runs from ram
<thoughtpolice>
shlevy: Yeah, I think step 2), just bootstrapping and making sure Nix works, is probably pretty easy to do in QEMU. If nix itself works, the rest should be much easier.
<srk>
clever: the only thing that sucks about netboot is that you get installer env and you need to revert some stuff such as autologin or ssh target to get usable "appliances"
<clever>
thoughtpolice: and this expression boots similar kernel+initrd pairs under qemu
<thoughtpolice>
Well, more like it's small enough to not make it intolerable to do now, in terms of speed. Maybe I'm not giving it enough credit, I don't know how optimized their QEMU work is.
<tertleeltret>
how "complete:" is NixOS?
<clever>
srk: thats mostly a matter of just selecting which nixos profiles are included
<thoughtpolice>
shlevy: And I think just making sure nix works on their BSP or whatever can be done before the HW even ships
<shlevy>
Do we actually have a good story for cross-compiling NixOS itself?
<shlevy>
Yeah, agreed
<thoughtpolice>
I can't imagine there's much missing, except maybe a Boehm-gc?
<srk>
clever: I was looking at how is netboot structured and it looks like some reorganization would be required
<clever>
srk: i think using netboot.nix directly, and skipping the netboot-base will also skip the installation device config
<thoughtpolice>
shlevy: That's the work bgamari was doing; he's got NixOS/nixpkgs reliably cross compiling to ARM. We do this at work but it's complicated and my patches are horrid (one offs).
<srk>
clever: cool, will try
<thoughtpolice>
And Sonarpulse, ofc
<shlevy>
thoughtpolice: At the very least there are still macros in the code for the no-boehm case, no idea if it actually works
<shlevy>
thoughtpolice: But we don't need boehm for the store
<clever>
srk: netboot-minimal is just netboot-base + the minimal profile
<shlevy>
I think I'll modify nix to have a store-only base. So libnixstore, nix-store, and nix-daemon basically
<clever>
srk: and netboot-base is netboot + all-hardware + base + installation device + no password
<shlevy>
Would be good to just be able to easily pass a configure flag for that for future such efforts
<srk>
clever: I see, thanks
<thoughtpolice>
shlevy: Also, I tried to fully merge my NixOS sources into my locked Nixpkgs set, and build .system closures and switch-to-configuration, but I see what you mean now about things using 'import <nixpkgs>' :(
<thoughtpolice>
There aren't that many now, at least
<shlevy>
Yeah :/
<thoughtpolice>
(I think we talked about this a bit on Twitter)
<clever>
thoughtpolice: oh, about dealing with import <nixpkgs>, there is an expression that can bootstrap a fixed nixpkgs, without relying on <nixpkgs>
<thoughtpolice>
shlevy: That would probably be pretty useful, I have an expression here on my laptop to build a RISC-V GCC cross compiler using their toolchain set
<clever>
thoughtpolice: that then lets you juset `unset NIX_PATH`
<clever>
thoughtpolice: then anything you miss, just fails hard
<thoughtpolice>
But it's for unknown-elf, not linux-elf
<thoughtpolice>
i.e. I need to get the userspace cross-sdk working...
<thoughtpolice>
clever: oh, does that work for Nix too? I'm actually just outright using fetchTarball now since I'm using Nix 2.0
BlessJah has quit [Ping timeout: 256 seconds]
<thoughtpolice>
Err, work for NixOS
<thoughtpolice>
Oh, I see what you mean.
<thoughtpolice>
Yeah, it works. But it would be nice to purge it from NixOS itself, at least.
<ottidmes>
tertleeltret: Pretty complete, but you might still have to package some less commonly used packages yourself
<thoughtpolice>
Since it makes building real closures a bit finnicky. Failing outright is better, at least.
<clever>
thoughtpolice: yeah, fetchTarball is best
<thoughtpolice>
But IIRC there are only like,
<clever>
thoughtpolice: the bulk of the expression on the wiki i linked, is to deal with it working on 1.11 as well
<thoughtpolice>
less than 10 occurrences of it inside NixOS directly, I think
<clever>
thoughtpolice: i think nixos and nixpkgs primarily use the right number of ./../'s
<jordanmackie>
samueldr: what's the verdict boss?
<jordanmackie>
(it's gone midnight where I am, but I can't sleep while my laptop is a tty...
<samueldr>
jordanmackie: could you try `cat /nix/store/g58*iwlwifi*/lib/firmware/iwlwifi-3160-13.ucode > /dev/null` (or the proper full path instead of wildcards)
<samueldr>
it should (hopefully) print SQUASHFS errors for you
<thoughtpolice>
shlevy: Anyway, just making some package with nix-store/nix-daemon would be pretty cool. And 2.0 makes it easy to use --builders to do one offs!
<shlevy>
Yeah exactly
<thoughtpolice>
So if you try this do let me know, I might have a few of my own patches to Nix I need to write...
<shlevy>
With nix-daemon we can use ssh-ng easily
marusich has joined #nixos
<shlevy>
thoughtpolice: I'm going to start adding issues to the milestone, feel free to assign yourself and unassign me ;)
<thoughtpolice>
shlevy: Alright I gotta run for tonight now, I think -- good chatting as usual.
<thoughtpolice>
clever: And thanks for the pointers as usual!
<thoughtpolice>
shlevy: SGTM
<shlevy>
Oh, github has projects now, that's better than a milestone
<jordanmackie>
samueldr: laptop went to sleep so I'd need to reinit everything - would reloading an earlier image fix this?
<samueldr>
well, the image on my end looks fine, so either re-downloading the same (and verifying the sum) or another should work
<samueldr>
though, it may also be an issue with the usb drive
<samueldr>
which, no image can fix :(
<jordanmackie>
roger that, I'll change up both (Y)
<samueldr>
I'm waiting for the install to actually finish in the VM, but catting the file *should* have triggered the same issue (I think)
<tertleeltret>
i think ive found a reason to run linux again
<tertleeltret>
i was always a fan of functional programming paradigm. F# Haskell Scheme? etcc
<samueldr>
jordanmackie: the install went through in my VM, let's hope it works better for you this next time :)
BlessJah has joined #nixos
<samueldr>
in theory, you can skip some steps, you can mount your partitions that you already created, and then nixos-install, since the configuration files will be as you left them
<clever>
thats also a common way to repair a nixos install
Neo-- has quit [Ping timeout: 265 seconds]
<ottidmes>
the initrd.secrets should not change permissions, right? I am getting an error that my private key is 444, while it should be 600 (the source file has that)
<clever>
ottidmes: i suspect that if you use an unquoted path, it gets copied into the nix store
<clever>
ottidmes: so your secrets are no longer secret
<clever>
if you use a quoted absolute path, it will probably work better
<ottidmes>
clever: So I should use toString on the path, right?
<clever>
yeah, that should work
kakuzade has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vAltY
<kakuzade>
Someone knows how to enable hardware video acceleration with proprietary nvidia drivers. I tried installing libva, vaapiVdpau and libvdpau with opengl.extraPackages but did not install them.
ambro718 has quit [Quit: Konversation terminated!]
<ottidmes>
clever: That gives me the following error "cp: cannot stat '/wheel/proj/nixcfg/server/config/openssh/host/id_ed25519': No such file or directory", even though that file exists, probably the sandbox thing?
<clever>
ottidmes: that sounds like its trying to read it at nix build time, which it shouldnt be doing
<clever>
ottidmes: which nixos option did you set?
<ottidmes>
clever: Right, then I guess I know the cause, I am using the same option to generate the necessary config, while before, when it did work, I just had one hardcoded key, so it was not used in the generation of the config
<bhipple[m]>
When serving a store with nix-serve, does it have to be run as a specific user? And is there a way to get it to print more debug info?
<bhipple[m]>
I'm trying to figure out why it's not getting hit as a binary cache from my client
<ottidmes>
clever: My own initrd sshd module's hostKeys option: boot.initrd.sshd.hostKeys.ssh_host_ed25519_key = toString ./openssh/host/id_ed25519;
<samueldr>
kakuzade: services.xserver.videoDrivers = [ "nvidia" ]; AFAIK is the only setting needed
<clever>
ottidmes: and what does the expression do with that string?
<samueldr>
(looked at my own configuration file)
<ottidmes>
bhipple[m]: I recently setup nix-serve, but I found the nix option ssh-substituter-hosts much easier to setup
<clever>
bhipple[m]: to start with, try doing an http query against nix-serve, take the hash from /nix/store/HASH-name and query http://example.com/hash.narinfo
<infinisil>
tertleeltret: this won't update stuff you installed with nix-env though
<tertleeltret>
oh
<tertleeltret>
so how should i install stuff?
<infinisil>
tertleeltret: with nixos you can always rollback, so it almost certainly won't break anything
<bhipple[m]>
clever: Ah nice, curling example.org<hash>.narinfo returned a lot of metadata about it in the store. So I guess the problem is client-side, not ix-serve server side
<tertleeltret>
or shuld i just update manually each package?
<clever>
bhipple[m]: does that metadata also include a signature?
<tertleeltret>
how to check for new kernel?
<clever>
bhipple[m]: nix will reject unsigned things
<infinisil>
tertleeltret: i think you can update nix-env packages with nix-env --upgrade
civodul has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
<samueldr>
if they were installed using sudo nix-env -i, you will need to use sudo to upgrade them
<kakuzade>
samueldr: Not in my case. I had to install vdpauinfo in the system packages. It's strange. It's a nvidia gt 630. Maybe I need a specific version of the driver?
ma271 has quit [Ping timeout: 256 seconds]
<samueldr>
I'm on a GTX570, but I'm not 100% sure how to verify mpv/vlc are using them
<samueldr>
though, pretty sure vdpauinfo isn't needed for them to use video acceleration
jb554 has quit [Ping timeout: 240 seconds]
<tertleeltret>
infinisil: you seem unsure about that
<bhipple[m]>
Ah, it does not; this nix-serve is meant to be a mirror of cache.nixos.org plus internal pkgs, for access behind a company firewall. Is there a way I can have it propagate cache.nixos.org sigs?
<bhipple[m]>
Or set it up to re-sign anything it downloads (?)
<tertleeltret>
i ran it but got no output
<infinisil>
tertleeltret: Because I'm not really using nix-env
<tertleeltret>
oh nevermind it hink it worked
<tertleeltret>
what do you use infinisil, and whY/
<infinisil>
i use my configuration.nix for everything, or i try to anyways
<infinisil>
so it's a nixos-rebuild for me
<tertleeltret>
i must say, this is a very coherent distro
<clever>
bhipple[m]: nix-serve doesnt really act as a mirror, all it can do is serve what you have in /nix/store/
<infinisil>
but I'm using nixops for all my machines, so it's a nixops deploy for me
<clever>
bhipple[m]: and it always re-signs everything with the keys you configured it for
<infinisil>
but I've got a wrapper script around that, so it's `rb` for me
<tertleeltret>
rb?
<infinisil>
short for rebuild
<infinisil>
just the name i gave my scrip.t
<tertleeltret>
oh i see
<infinisil>
and i'm using a git checkout of nixpkgs, so it's a `git rebase` to upgrade
<kakuzade>
samueldr: vdpauinfo only provides information about libvdpau. What output do you get?
<ottidmes>
bhipple[m]: I had to dig to make it work, but this is what I had to do to make nix-serve work for me: https://pastebin.com/xCXCYZ5P
<bhipple[m]>
ottidmes: thank you. I didn't see anything in the manual about keys with nix-serve, so I will try this!
<clever>
bhipple[m]: oh, and nix has a cache for the cache, and will not request again after a cache miss
<clever>
ottidmes: oh, and also, your signing key is in the store now, so nix-serve is sharing it with the world
babyflakes has quit [Quit: Connection closed for inactivity]
<bhipple[m]>
Is there a recommended simplest way to do an internal mirror of everything on cache.nixos.org?
[0x4A6F]1 has joined #nixos
<ottidmes>
clever: Haha, I know... When you told me that about my host key a few minutes ago I realized I had to recheck all my config files to make sure I did not make those mistakes elsewhere. I should have changed that before posting, good catch
<clever>
bhipple[m]: it would need to be something like squid, that can proxy requests and cache the responses
<clever>
ottidmes: so if you posted something like this symlink, i could trivially track it down to the secret keys
<clever>
ottidmes: if i knew your public facing domain
<tertleeltret>
so is nix for Haskell deployments mostly?
<ottidmes>
clever: I am not freaking out right now because it is not public, only accessible from my local machines
<clever>
ottidmes: ah
<clever>
ottidmes: and as long as you keep your nixos store paths secret, it should be safe
<clever>
ottidmes: so just dont leak things like i did with mine above :P
<clever>
(that machine doesnt nix-serve, so its safe for me)
<kakuzade>
samueldr: In your case it seems that the proprietary driver automatically installed vdpau. In my case, no. It's strange. You only have the expression services.xserver.videoDrivers = ["nvidia"]? There's no more?
fragamus has joined #nixos
<ottidmes>
clever: I would never serve my nix store publicly, I have restricted them all to only access it via SSH tunneling
<clever>
ottidmes: also, hydra has a built in nix-serve like process, so any hydra server share it
<samueldr>
kakuzade: only references to nvidia in my config files are in that nvidia file I copied
<ottidmes>
clever: I do not forsee myself needing hydra in the near future, but good to know that it is somewhat similar
tgunb has quit [Quit: Leaving.]
<ottidmes>
clever: Well, I need it all the time, but I will not setup my own :P
<samueldr>
and only references to vaapi are in that nvidia file and the equivalent intel file *which is not included on this device*
<samueldr>
(let's hope I never include private bits!)
<ottidmes>
kakuzade: I only half read your messages, but could this help: hardware.firmware = with pkgs; [ firmwareLinuxNonfree ];
<ertes>
what's the equivalent of nix-prefetch-git with nix 1.12?
<samueldr>
I do have imports = [ <nixpkgs/nixos/modules/installer/scan/not-detected.nix> ]; in my hardware-configuration.nix, which *maybe* could be adding to firmwares
<samueldr>
(the relevant machine with nvidia hardware is duffman)
<infinisil>
ertes: why would it be different? Should still work no problem
<samueldr>
that was directed in a way that kakuzade could verify whether they have that import in their configuration
<samueldr>
but thanks for drilling in the configs clever :D
<tertleeltret>
infinisil: rolling release and up to date like fedora and arch
[0x4A6F] has quit [Ping timeout: 248 seconds]
<infinisil>
tertleeltret: the nixos-unstable channel is indeed
<samueldr>
tertleeltret: it depends on whether you follow a fixed release channel (e.g. 17.09) or nixos-unstable
shabius has quit [Quit: Leaving]
<tertleeltret>
default is 17.09 right?
<samueldr>
though, *some* updates (mostly browsers and monolithic behemots like browsers) are bleeding up through the latest release channels
<tertleeltret>
i just installed just now
<kakuzade>
samueldr: thanks! I tirie the nvidia.nix.
<samueldr>
tertleeltret: defaults to whichever is on the installer you used, if you used 17.09, defaults to 17.09
<tokudan[m]>
I'm having some trouble understanding npm and node2nix. manually running "npm install" works if I'm in a shell generated by node2nix and in a git clone directory of the source. letting nix build the actual package fails for some reason that I don't understand: https://gist.github.com/tokudan/5634ce6867541ced37d3c43a4025f90d
<ertes>
infinisil: just like i can still use nix-build… that's not the question though… the talk mentioned a way to calculate hashes of derivations and repos directly
<infinisil>
tertleeltret: you can run nixos-version to check
<tertleeltret>
cuz basic renoise install didnt work
<tertleeltret>
because*
<infinisil>
tertleeltret: why do you specifically want to use the file directly rather than just installing it as default?
<infinisil>
you can install stuff either by adding it to configuration.nix or via nix-env
<tertleeltret>
i just want to install it
davidak has quit [Ping timeout: 265 seconds]
<infinisil>
tertleeltret: Then you can just nix-env -iA nixos.renoise
<infinisil>
or really useful: `nix-shell -p renoise`, will give you a shell where renoise is installed, but won't actually install it, really useful for trying stuff out
<tertleeltret>
wow im dumb
<tertleeltret>
i didnt realize it was in the packages
<infinisil>
haha, yeah nixpkgs are nix' packages
<tertleeltret>
but it says i need to alter the path or something right?
<tertleeltret>
cuz thats just the demo
silver has quit [Read error: Connection reset by peer]
<infinisil>
tertleeltret: huh? what does it say?
justan0theruser has quit [Ping timeout: 240 seconds]
<tertleeltret>
it says to download your paid version first, then overide the releasePath attribute to point to the location of the newly downloaded bundle
<infinisil>
tertleeltret: Ah, it must be some kind of non-free proprietary software thing
<tertleeltret>
yes it is
<tertleeltret>
thats why I didnt think it would be in the repos
<tertleeltret>
my ass hurts from riding my bike
<infinisil>
you have a copy of all the nix files in the nixpkgs repo online on your machine
<lejonet>
srhb, infinisil: Now the service module has the last daemon required, now you can run and configure an entire ceph-cluster, with latest lts through nix :D
<tertleeltret>
infinisil: is that the store?
<infinisil>
lejonet: nice! gonna check it out
<infinisil>
tertleeltret: indeed, pretty much everything on nixos is in the store
<NixOS_GitHub>
nixpkgs/master d73aa60 Michael Raskin: Merge pull request #35090 from AndersonTorres/upload/palemoon...
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 2 new commits to master: https://git.io/vAlOJ
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
tertleeltret: note that it will probably say something like /nix/var/nix/profiles/per-user/root, which is a symlink to somewhere in /nix/store/...
<infinisil>
well .../root is not, but .../root/channels is
Ankhers_ has joined #nixos
<Ankhers_>
Does anyone have experience with using znc on NixOS? I'm getting a seg fault when znc is attempting to generate a pem file.
* infinisil
uses znc on nixos
<infinisil>
but I'm not doing anything with pem files
<infinisil>
nixpkgs tries to allow you to override everything, so you rarely have to change a nixpkgs file yourself
<infinisil>
and this would require a git checkout, since the nix store where your nixpkgs is, is read-only
justanotheruser has quit [Quit: WeeChat 1.9.1]
verite has quit [Ping timeout: 248 seconds]
justanotheruser has joined #nixos
<tertleeltret>
ohhhhhhhhhhhhhh makes sense now
Supersonic has quit [Disconnected by services]
<infinisil>
And with nix-env I guess `nix-env -iE '(import <nixpkgs>{}).renoise.override { releasePath = ./.; }'` would work, but that's kinda ugly
<infinisil>
in configuration.nix it's just `pkgs.renoise.override { ... }`
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<infinisil>
ah, for nix-env there's a config file in ~/.config/nixpkgs/config.nix which you can use to declare a packageOverride, which can do that and make it nicer, but eh
<infinisil>
there's lots of possibilities
<lejonet>
infinisil: its the directory where all the ceph daemons put their admin sockets
codygman has quit [Ping timeout: 260 seconds]
<infinisil>
lejonet: nice, i didn't know about tmpfiles!
<lejonet>
infinisil: oh? Well I guess its a semi-new thing that systemd added, for the /run "directory tree"
<infinisil>
Actually that's like exactly the thing I would've needed for one of my PR's
<lejonet>
Basically you can use it to get systemd to create all the ephemeral directories you need
<lejonet>
infinisil: look at man tmpfiles.d for a thorough description
<infinisil>
might take a closer look at some point
<lejonet>
Its a neat way of not having to do ugly hacks in preStart and such :P
<lejonet>
infinisil: peterhoeg has a point, however, you don't have to patch main.c at all, /var/run points to /run
<tertleeltret>
i will have to get with you tommorw infinisil, i need to get some rest
codygman has joined #nixos
<tertleeltret>
thanks for the help
<infinisil>
but it tries to create /var/run/usbmuxd.pid which it can't as non-root
<infinisil>
tertleeltret: not sure if i'm gonna be here tomorrow, just ask in here though, see ya!
<lejonet>
infinisil: that is where the RuntimeDirectory comes in
<shlevy>
\o/ gcc-7.3.0-riscv64-unknown-linux-gnu-stage-static build is chugging along
<lejonet>
infinisil: even tho I personally prefer having them all in one place, /etc/tmpfiles.d, there is a option to have them directly in the service files aka RuntimeDirectory
<codygman>
or only overlays? I heard package overrides might be disabled in a talk I think
<lejonet>
infinisil: the obvious win to have them in the service file is that you get to know that systemd handles that for you, directly when reading the service file
<infinisil>
lejonet: hmm, well it could be done, but it's not worth spending that much time on this
<lejonet>
infinisil: That is entirely up to you :P its 1 row in serviceConfig (and testing ofc)
<infinisil>
lejonet: well i tried to patch it but got another error i couldn't make sense of
<lejonet>
The dropping priv part of linux is quite well functioning and even if it was started by root, its damn near impossible to regain that after you've dropped privs (without an exploit that is)
<lejonet>
So the security implications are negligible imo, of having it run as root for start and then drop privs
<infinisil>
i know this is not an argument, but there's loads of other modules using root to run stuff
<lejonet>
You mean lazy devs?
<infinisil>
it's kind of a pain to add user and group and stuff
<infinisil>
maybe that's not always needed though, it doesn't need to be configurable all the time
<infinisil>
but still gonna be annoying
<lejonet>
on nixos, I would argue its a minor nuisance, on other oses, it can be a real pita
<lejonet>
There is always nobody/nogroup ^^
<infinisil>
I really want some systemd basic functions in nixpkgs
<infinisil>
e.g. a cron-like function that ends up as a systemd timer + service
<infinisil>
so you don't have to remember how to create a repeating service
<lejonet>
That would be nice
<lejonet>
One of my main gripes with systemd is the sheer amount of small, idiotic files you have to create for stuff
<lejonet>
systemd-networkd takes the cake
<lejonet>
"Oh, you have a vlan on a bonded device? Thats 4 files that contain less than 8 lines each"
<infinisil>
yeah, but with nix we have the power to abstract
<lejonet>
Mhm
<infinisil>
if we just had functions for common stuff
<lejonet>
I had a gateway, with systemd, that had 2 bonded interfaces, which one had 5 vlans attached to it, I think I in total had 20 or something files for that setup... which always lead to you opening atleast 3 files when trying to find where the hell to add or change an option
orivej has quit [Ping timeout: 256 seconds]
<infinisil>
ugh
<lejonet>
Whats the point of having ini style configs, when you're not using the full power of sections? :P
<infinisil>
aren't ini files like a subset of json?
<lejonet>
doubt it, ini files are as old as windows 95 if not older :P
<infinisil>
i mean that every ini file can be represented by a json one
<infinisil>
but not the other way around
<lejonet>
yeah, definitively
<lejonet>
sections are just attributes in a top-level hash, in json land
<lejonet>
that is why the toINI generator in nix is really small and neat
<lejonet>
attrsets is the perfect codification of INI files :P
<lejonet>
(okay, perfect is a stretch...)
<infinisil>
and json is a subset of nix
<infinisil>
so being able to use nix for configuration is bliss
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lopsided98 opened pull request #35091: kernel: fix cross compilation after switch to gcc 7 (master...kernel-gcc7-cross) https://git.io/vAl3l
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
infinisil: yeah, I really like the INI style config, most things can be described with it, with ease
<lejonet>
But its not that common, even tho you could argue that configs that are plain key=value configs, are INI-style with an implicit global section :P
<codygman>
Is it ~/.nixpkgs/config.nix or ~/.config/nixpkgs/config.nix for user configuration and overrides? I've seen both in various places.
<codygman>
Is there a command I can run to tell me what location nix is looking for my user config?
<lejonet>
But time for me to head to bed, its waaay too late :P
<infinisil>
codygman: former is legacy
<infinisil>
lejonet: same
<lejonet>
infinisil: funny thing, this with time, I said I was gonna head to bed "soon" like 3 hours ago... xD
<codygman>
infinisil: Thanks
<codygman>
I've written my config to ~/.nixpkgs/config.nix and it seems to be having no effect on sudo nixos-rebuild switch
<lejonet>
stayed up and wrote the ceph-mgr part and deployed it on my cluster to make sure it works instead... :P
<infinisil>
lejonet: same man, and the worst is that I'll have to get up early again in a couple days when university starts again
<lejonet>
infinisil: haha, enjoy it while you can, I have to get up early already on monday ;)
<infinisil>
tuesday for me
<codygman>
I modified my ~/.config/nixpkgs/config.nix and added a package with overrideDerivation but nix-env -i says 'matches no derivations'. Full output: http://lpaste.net/5870952923327889408
mrkgnao has joined #nixos
kitemikaze has quit [Ping timeout: 260 seconds]
jb554 has joined #nixos
pie_ has quit [Ping timeout: 248 seconds]
<mrkgnao>
okay, I'm having a lot of trouble finding the IDE for the J language in nixpkgs
<mrkgnao>
the name doesn't help at all
Ankhers_ has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
nuncanada has quit [Quit: Leaving]
truthadjustr has joined #nixos
<infinisil>
mrkgnao: from what i can find there is a browser based IDE called JHS
<infinisil>
but i can't find jhs in nixpkgs
<infinisil>
mrkgnao: the interpreter seems to be available through the j attribute though
<truthadjustr>
is the containerization used in nixos using cgroups and namespaces?
dnovosel has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vAlsi
<NixOS_GitHub>
[nixpkgs] shlevy opened pull request #35092: Initial cross-compilation support for RISC-V (master...riscv-initial-cross-support) https://git.io/vAlZb
NixOS_GitHub has left #nixos [#nixos]
clhodapp has joined #nixos
mounty has joined #nixos
jrolfs has joined #nixos
<infinisil>
Probably not many people here, but does anybody just happen to know how to override emacs packages? specifically melpa's
<elvishjerricco>
Ralith: Ah, got it. How does steam know that's where it is?
alex`` has joined #nixos
<Ralith>
it's not steam's business how games find their libs
<Ralith>
the games find it by letting the dynamic linker do its thing, and the dynamic linker finds it by looking in $LD_LIBRARY_PATH
mrkgnao has quit [Ping timeout: 256 seconds]
<Ralith>
perhaps the game has some idiotic wrapper script that clobbers that
<Ralith>
you could easily hack around that, if so
<elvishjerricco>
Ralith: I tried another game. Same black screen on launch. But no error message about libGL.so.1. So I think the libGL.so.1 thing might be a red herring
<elvishjerricco>
Er, wait, I didn't check the right place for that message
<Ralith>
games are unlikely to work without libgl, so that is doubtful
<Ralith>
your system is fine, your games are janky :P
<Ralith>
check for a wrapper script
<elvishjerricco>
Ok there's no libGL error this time. So yea, I think the libGL thing is a red herring
<Ralith>
it is not a red herring
<Ralith>
you have two different games that are differently broken
<elvishjerricco>
Ralith: They are broken in the same way, yet only one of them shows the libGL error. Even if the libGL thing is a problem, it seems like that might come after solving whatever this common issue is
<Ralith>
you said "fails to find libGL," not "displays a black screen"
<elvishjerricco>
Ralith: Yea, my bad. At that point I had assumed the libGL error message burried in the console was definitely the culprit :P
robstr has joined #nixos
<Ralith>
if it is able to display anything at all, blackness included, it has not failed to find any of its immediately necessary dynamic libs
<elvishjerricco>
Well *something* definitely did :P
<Ralith>
it may be failing to dlopen stuff, but it would still be pretty weird for it to even get as far as opening a window in that state
<elvishjerricco>
But I'm guessing it's something unnecessary like the steam overlay
<elvishjerricco>
Er wait, I guess I don't know if that's a separate process or not...
<elvishjerricco>
Anyway
<elvishjerricco>
Point is: I currently just get black screens :/
<Ralith>
on what games?
mrkgnao has joined #nixos
<elvishjerricco>
Civ VI
<elvishjerricco>
And I downloaded a couple small free games to test and got the same issue
<Ralith>
did you try tf2?
<elvishjerricco>
Nope. Will try that
<nflores>
has anyone used fetchurl to successfully download artifacts from private GitHub releases?
<Ralith>
being a wildly popular and widely ported title it should be a good canary
<Ralith>
if that works I'd start googling for civ vi specific issues, digging through logs, stracing, etc
<elvishjerricco>
Ralith: Will take a while to download, and I'm not sure why it would work when three other games did not.
<Ralith>
well, you didn't say what the two other games are so I can't judge if they're credible tests or not
<Ralith>
a system-wide issue which somehow caused black screens for all steam games is implausibly although not impossibly specific
<elvishjerricco>
"Oldage" and "Dungeons Forever". I've never heard of either of them, but the fact that they exhibit the same issue is enough, right?
<Ralith>
not really; lots of stuff is broken out of the box on linux, especially if it's no-name free crap that hasn't been maintained in half a decade :P
<elvishjerricco>
Sure, I wouldn't be surprised for them to be broken out of the box. But for them all to be broken out of the box in the exact same way?
<elvishjerricco>
I'll try some other small ones
<Ralith>
similar symptoms is not the same thing as being broken in exactly the same way
<Ralith>
we currently know little to nothing about the precise manner in which any of your games are broken
<Ralith>
except that some component of one of them has something to say about libGL
<elvishjerricco>
Two more games broken with a black screen. But one that works! The only immediately obvious difference about the working one is that it does not attempt to go full screen
<Ralith>
interesting
<Ralith>
does it use OpenGL?
<Ralith>
(i.e. has obviously hardware-accelerated graphics)
<elvishjerricco>
Ralith: Not sure. It's pretty simple, graphically, so I think it could get by without. How can I check?
<Ralith>
running `ldd` on the binary might do
<Ralith>
(or it might not, depending on how it was built)
<Ralith>
generally speaking even stuff as simple as smooth animation entails hardware-accelerated graphics
<Ralith>
unless the resolution is very low
<elvishjerricco>
No dependency on libGL. I do see libgtk-x11 though
<Ralith>
gtk in a steam game? weird.
<Ralith>
I would hesitate to draw any strong conclusions then
<elvishjerricco>
yea
<elvishjerricco>
This will be an interesting test case: I *can* run Factorio when installed through nixpkgs. I'll see if I can run the demo as installed by Steam.
<srhb>
elvishjerricco: Have you tried killing your compositor if you have any? (I didn't scroll all the way up)
<elvishjerricco>
srhb: Don't even know what a compositor is :P
<srhb>
Oh, something like compton that allows for transparency etc. between windows
hiratara has joined #nixos
<elvishjerricco>
I just Xmonad. So I probably don't have anything like that
<srhb>
I'm able to play Factorio via Steam very well and Civ VI poorly.
<srhb>
Right, I use XMonad too, but with compton launched separately. Oh well, just an idea.
<elvishjerricco>
srhb: Have you had any weird troubles with steam games or Civ VI in particular? Or has it all mostly worked out of the box for you?
<srhb>
Lots of weird errors with ported games like Civ VI. For instance, I get a black screen on diplo screens if I turn the advisor settings all the way down, but not on the second lowest settings.
<srhb>
Most things are just slow though, not broken.
<srhb>
My steam also reports that error with pinned libs, so that appears to be a red herring
simukis has joined #nixos
<elvishjerricco>
srhb: Got it. Thanks.
<Havvy>
Can somebody do me a favor and take https://github.com/havvy/nixpkgs/tree/mdbook and get it working? I'm running into compilation issues on my end about running out of jobserver tokens or something, and I'm hoping that it doesn't happen on others' machines.
<elvishjerricco>
I think this might have something to do with it going full screen.
<srhb>
elvishjerricco: Oh.. Interesting.
<srhb>
I feel like I had something like this at one point, but I can't quite remember...
<elvishjerricco>
the other games that failed were mostly unity games, and their logs just show it going into full screen, and then nothing.
<NixOS_GitHub>
nix/master 690ac7c Shea Levy: configure: Add a flag to disable seccomp....
NixOS_GitHub has left #nixos [#nixos]
<srhb>
Oh, yes, lots of unity game failures here, too.
<elvishjerricco>
In fact, I can sometimes even see a non-fullscreen "Loading, please wait" screen when I launch Civ. But soon it goes away and I get the black full screen again
<srhb>
I think I never got Cities: Skylines to work
<srhb>
elvishjerricco: strace for great glory. :/
<sphalerite_>
Havvy: open a PR, mention clearly that it's still WIP so it doesn't get merged too hastily, and I'll get GrahamCOfBorg to build it for you :)
<elvishjerricco>
srhb: How can I strace a steam game?
<srhb>
elvishjerricco: You can right click and edit properties to alter the launch command in your library, or you could launch them without steam by finding them in ~/.steam/...
<srhb>
I wonder how much I have to build here... :-P
<Havvy>
I guess I shall have to wait until Rust 1.24 then.
<Havvy>
srhb: LLVM + rustc = ~2 hours
<srhb>
Wait what, rust isn't self-hosting?
<srhb>
Havvy: Ouch!
<sphalerite_>
is it not??
<sphalerite_>
I thought it was
<srhb>
I see lots of cpp going through right now at least
<Havvy>
srhb: That's LLVM.
<srhb>
Oh, okay!
<srhb>
I clearly need to take a look at rust one of these days. I know so little. :-)
<sphalerite_>
here goes, rebuilding the world for another update on my chromebook
<Havvy>
rustc 1.24 is bootstrapped with rustc 1.23. They both compile to LLVM instructions which get fed to a custom LLVM (and in rustc 1.25, that will be a v6.0.0 LLVM)
<srhb>
Havvy: Thanks!
<sphalerite_>
well not the *whole* world thanks to dezgeg but… a lot of stuff
<srhb>
sphalerite_: Oh, chromebooks are ARM?
<sphalerite_>
srhb: not all of them. Mine is though
<srhb>
Fun!
<Ralith>
srhb: note that LLVM is a huge build; be sure you passed --cores 0 and/or go have something to eat
<sphalerite_>
indeed. I need to investigate cross-compiling instead of building natively
<Ralith>
(rustc itself is no small thing either, for that matter)
ma271 has joined #nixos
<srhb>
Ralith: I bet it's not as bad as building Ceph from scratch... :P
<Havvy>
By the by, I'm having issues with compiling rustc from sources as well.
<Havvy>
It keeps complaining that g++ is not a real C (or was it C++?) compiler.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 18b41d6 to 03bd7de: https://git.io/v5aIf
NixOS_GitHub has left #nixos [#nixos]
<Havvy>
I'm getting it in a fresh NixOS install as well. ;.;
<Ralith>
Havvy: what sort of environment are you building in?
<Ralith>
NixOS installs are less sensitive to "freshness" than others, any issues are likely in your proximate workflow
robstr has quit [Remote host closed the connection]
<sphalerite_>
Havvy: yeah if it behaved differently between a fresh install and a preexisting one it would probably be a bug in the sandboxing or something
robstr has joined #nixos
<Havvy>
sphalerite: Well, I also have nix-env installed packages, and who knows what config files lying around.
<sphalerite_>
Havvy: they shouldn't affect nix builds though.
<sphalerite_>
If they do that's definitely a bug.
<sphalerite_>
at least the nix-env installed packages
<Havvy>
sphalerite: Well, when I call `nix-shell`, everything else stays available.
<Ralith>
Havvy: why do you have clang in nativeBuildInputs?
<Havvy>
Ralith: Was seeing if I could somehow get LLVM built with clang instead of g++.
<Ralith>
Havvy: you should use clangStdenv to accomplish that, rather than mucking about with build inputs
<Havvy>
I didn't know that was a thing.
kitemikaze has joined #nixos
<Ralith>
^^
marusich has quit [Quit: Leaving]
<Havvy>
See also: The standard library is a black hole of useful but poorly documented functions.
<Ralith>
fwiw, I build rustc fine with `stdenv.mkDerivation { name = "rust"; buildInputs = [ python2 cmake curl git ]; }`
robstr has quit [Ping timeout: 240 seconds]
<Ralith>
yes, nixpkgs relies to an unfortunate degree on tribal knowledge
robstr has joined #nixos
alex`` has quit [Ping timeout: 276 seconds]
<Havvy>
Well, let's see what happens with clangStdenv. But weird that it works for you like that, but not for me.
<Ralith>
note that I'm just letting it default to gcc
<Havvy>
I was too.
<Havvy>
The clang in there was from last time I tried debugging this.
kosmikus has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] thefloweringash opened pull request #35094: Restore linuxManualConfig as linux/kernel/manual-config.nix (master...restore-linux-manual-config) https://git.io/vAll0
NixOS_GitHub has left #nixos [#nixos]
jrolfs has quit [Ping timeout: 276 seconds]
klntsky has quit [Ping timeout: 255 seconds]
Acou_Bass has joined #nixos
<sphalerite_>
oh boy… can't deploy any more servers on scaleway because they're out of IPs…
<sphalerite_>
if only there were some protocol that allows more addresses than IPv4!
<srhb>
... ouch!
<sphalerite_>
well actually I can deploy another server, but I can't give it a public IP address
jensbin has joined #nixos
<typetetris>
Hi there, what can I do, so I don't have to include the `-I ...` param in nix-build all the time: `nix-build -I nixpkgs=~/.nix-defexpr/channels_root/nixos-unstable` ?
<sphalerite_>
typetetris: you can put it in NIX_PATH
<sphalerite_>
typetetris: -I just adds an entry to NIX_PATH for the purpose of the command that you're running
Shados has joined #nixos
jensbin has quit [Client Quit]
Acou_Bass has quit [Ping timeout: 260 seconds]
<hyper_ch>
ma271: online?
<typetetris>
sphalerite_: Thanks! Is there a way to do it with channels ?
<sphalerite_>
typetetris: not quite sure what you mean?
<ma271>
hyper_ch: yep
<Havvy>
Ralith: cargoStdenv worked.
<hyper_ch>
ma271: you're ma27 on github?
<Ralith>
Havvy: ...you mean clangStdenv?
<ma271>
correct, just screwed up my weechat atm
<Havvy>
Yes.
<srhb>
Is there a good way to renice a running nix build?
<hyper_ch>
ma271: so you tried to fix flexget... what didn't work?
* srhb
wants to play EUIV while rustc is building >>)
<Ralith>
Havvy: interesting; maybe having clang in your buildInputs was messing things up somehow
<Ralith>
glad you're set, anyway
jrolfs has joined #nixos
<ma271>
I fixed the actual build (you have to patch requirements.txt as it pins versions explicitly), but several tests were broken
<Havvy>
Ralith: It was happening without clang in my buildInputs though.
chrisbarrett has joined #nixos
<hyper_ch>
ma271: did it work?
<hyper_ch>
or can the tests be neglected?
<typetetris>
sphalerite_: Maybe my understanding of channels is wrong. I thought I do `nix-channel --add <stuff>` and bam it uses that instead of the normal one for each command. (Will use NIX_PATH for now, but I am curious about channels)
<Ralith>
Havvy: weird; can't really investigate further right now, anyway
<Havvy>
I should be heading to bed myself.
<ma271>
hyper_ch: well, the tests are the easiest way to ensure that the package actually works
<sphalerite_>
typetetris: nix-channel basically just downloads some stuff and puts it in a channel profile like /nix/var/nix/profiles/per-user/root/channels
<ma271>
however I'm not sure anymore what went wrong after that^^
<sphalerite_>
typetetris: (which ~/.nix-defexpr/channels_root is a symlink to)
<hyper_ch>
ma271: but tests failing doesn't necessarily mean the package fails, right?
<sphalerite_>
typetetris: I think just doing -I ~/.nix-defexpr/channels_root, or adding ~/.nix-defexpr/channels_root to your NIX_PATH, will probably do what you want. I'm not 100% sure I've understood what you want correctly though.
<typetetris>
sphalerite_: If I change my nix path, will nix-env -iA nixos.<something> still use the stable channel or do I have to put `-I` in there for that?
<sphalerite_>
typetetris: changing NIX_PATH and using -I is equivalent
<sphalerite_>
nix-env's default behaviour is stupid though and uses ~/.nix-defexpr instead of NIX_PATH
<sphalerite_>
which is why I recommend always using it with -f (I have it aliased to nix-env -f '<nixpkgs>')
<typetetris>
sphalerite_: Actually, that would be cool. So `nix-build` would use nixos-unstable by default and `nix-env` nixos-17.09. Thats what I would like.
<sphalerite_>
typetetris: well not really since you need to tell nix-build what to use explicitly
<ma271>
hyper_ch: correct, I wanted to point out that tests are helpful to confirm the functionality of a package, otherwise we'd have to check each package to be rebuilt manually when changing e.g. a python library
<typetetris>
sphalerite_: Hmm I thought `NIX_PATH` would have taken care of `nix-build` ?
<ma271>
the thing is: IIRC there was another issue, but I can't remeber which ATM, sorry :(
<sphalerite_>
typetetris: no, nix-build builds stuff from the cwd
<hyper_ch>
ma271: well, I'd love to move another server to nixos but need flexget for it :(
<sphalerite_>
or the path that you pass to it
<typetetris>
sphalerite_: Ah ok, I think I get it: `{ nixpkgs ? import <nixpkgs> {} , <stuff> }:` means it will use dependencies from whatever `nixpkgs` is set to on NIX_PATH. But as `nix-env` doesn't use `NIX_PATH` but `~/.nix-defexpr` I would get the behaviour I want (with this type of `default.nix`)? `nixpkgs` in `nix-build` usages would be `nixos-unstable` and `nix-env` would use `nixos-17.09` because of the contents of
<typetetris>
`~/.nix-defexpr`?
* sphalerite_
doesn't know anymore
<sphalerite_>
:po
<ma271>
hyper_ch: understood, the actual build succeeded when you patch requirements.txt accordingly, the only build issue were the tests
<ma271>
as I mentioned, IIRC there was an issue when trying to use the package, but I don't remember which...
asuryawanshi has joined #nixos
ma271 has quit [Quit: WeeChat 2.0]
ma27 has joined #nixos
<shlevy>
thoughtpolice: So I'm able to run nix under qemu riscv user mode emulation... Going to try getting a native toolchain building and see if the times are unbearable
mrkgnao has quit [Ping timeout: 248 seconds]
Acou_Bass has joined #nixos
chrisbarrett has quit [Quit: chrisbarrett]
<hyper_ch>
ma27: :(
i-am-the-slime has joined #nixos
<i-am-the-slime>
Hi. For some reason xmonad doesn't honour my "xkbOptions". Where can I set those instead?
vcunat has joined #nixos
robstr has quit [Remote host closed the connection]
robstr has joined #nixos
alex`` has joined #nixos
ma27 has quit [Ping timeout: 240 seconds]
robstr has quit [Ping timeout: 276 seconds]
aarvar has quit [Ping timeout: 256 seconds]
asuryawanshi has quit [Remote host closed the connection]
<clever>
/nix/store/i75qkyhlvjy5li19rf65jcg8hxnvzbv0-xkeyboard-config-2.22/share/X11/xkb/symbols/us: include "eurosign(5)"
<clever>
i-am-the-slime: and it contains things that match up with the xkb options
<clever>
./share/X11/xkb/symbols/eurosign
<clever>
it also has directories that match the options
<i-am-the-slime>
clever how did you start this?
<clever>
i-am-the-slime: i just did 'ps aux | grep X'
<i-am-the-slime>
Ah okay
<clever>
[clever@amd-nixos:/nix/store/i75qkyhlvjy5li19rf65jcg8hxnvzbv0-xkeyboard-config-2.22/share/X11/xkb/symbols]$ ls
<clever>
also, if you run `nix-store -r /nix/store/i75qkyhlvjy5li19rf65jcg8hxnvzbv0-xkeyboard-config-2.22`
ma27 has joined #nixos
<clever>
it will download that exact path from the binary cache
alex`` has quit [Ping timeout: 276 seconds]
<clever>
so you can look at the exact same files i have (which may be newer or older then your own variant)
<typetetris>
clever, i-am-the-slime: ps aux | grep X reveals the used config file (argument to the -config switch) why not just take a peek into it, if it was generated ok? So one could make sure, wether that is an X or nix issue?
<i-am-the-slime>
I am looking right now.
<clever>
typetetris: yeah, that also works
<clever>
no harm in reading what nix generates
<i-am-the-slime>
At that xkb folder
<clever>
eurosign lists 4 different configs, to have the euro sign on the e, 2, 4, or 5
MP2E has quit [Remote host closed the connection]
<i-am-the-slime>
Okay I ran that grep too now (man, you're fast)
<i-am-the-slime>
I see ;"include "eurosign(5)"' twice
ixxie has joined #nixos
<i-am-the-slime>
I don't really understand what this tells me to be honest.
Arcaelyx_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
i-am-the-slime: i can see options to map shift to capslock, swap control and caps, make escape capslock and more
jrolfs has quit [Ping timeout: 240 seconds]
<hyper_ch>
so how can I import a git repo master? download only allows zip
<clever>
hyper_ch: github allows tar
<i-am-the-slime>
I saw one nice option which has tab act as hyper (meta? what's the difference) when being held down. I could not for the life of me google what little string this corresponds to.
<hyper_ch>
clever: where?
<clever>
hyper_ch: just change the .zip to .tar.gz
<clever>
i-am-the-slime: meta is alt, hyper is the windows key
<hyper_ch>
clever: why don't they indicate that :(
<sphalerite_>
huh, I thought meta is alt and *super* is the windows key
<sphalerite_>
and hyper isn't present on most keyboards
jrolfs has joined #nixos
<hyper_ch>
I'm present on all keyboards :)
<clever>
sphalerite_: i might be mixing hyper and super up
<clever>
sphalerite_: also, i have a stop and help button on my keyboard, so things are a bit different here, lol
<clever>
its even got brightness adjustment keys
<clever>
altwin:xkb_symbols "hyper_win" {
<clever>
capslock:xkb_symbols "hyper" {
<clever>
i-am-the-slime: i see 2 files that name hyper
<sphalerite_>
is it a sun keyboard?
<clever>
sphalerite_: yep
<sphalerite_>
I have one of those sunwhere
<sphalerite_>
I like it
<clever>
on normal keyboards, i have trouble setting up global keybinds, everything is in use in some app :P
<clever>
but now i have a whole 11 buttons that nobody expects you to have
<clever>
so they are always free
<sphalerite_>
exactly
<clever>
63 // Hyper is mapped to the Super.
<clever>
65 xkb_symbols "hyper_win" {
<sphalerite_>
meanwhile on my chromebook I don't have a bunch of keys that people do expect you to have...
<clever>
i-am-the-slime: i think this is altwin:hyper_win
<sphalerite_>
(not even via fn like on my big laptop)
<clever>
sphalerite_: that reminds me of the day i tried to type c code on a treo 650, it doesnt have { or } keys, or ;
<clever>
so i had to go 3 levels deep into a menu every time
vidbina has joined #nixos
<clever>
that compiler also lacked #include, so i had to retype every single header file
<i-am-the-slime>
Okay, I'm happy that it's not completely obvious to everybody else at least.
<sphalerite_>
clever: don't you know about C's digraphs and trigraphs?
justanotheruser has quit [Ping timeout: 265 seconds]
<clever>
ok, so -option prepends things, and i have no way to undo...
<clever>
and it accepts invalid options silently
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 3957b6b Nikolay Amiantov: moinmoin: refactor and fix build
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] abbradar pushed 1 new commit to master: https://git.io/vAlRE
jrolfs has quit [Ping timeout: 256 seconds]
<clever>
(as saved in root window properties). If you want to replace all previously specified options, use the -option flag with an empty argument first.
<clever>
so this wipes the list, then restores the original value
<clever>
without the wipe flag, it always prepends
<clever>
oh, i also see a man directory in this storepath
<clever>
[clever@amd-nixos:~]$ grep eurosign /nix/store/cj8sj64sbbj950hd00wz9q4ccq3ssdfj-setxkbmap-1.3.1/share/X11/xkb/rules/base.lst eurosign:e Euro on E
<clever>
sphalerite_: aha, and this actually has the full option string, and a description!
jrolfs has joined #nixos
omger has joined #nixos
robstr has quit [Remote host closed the connection]
<hyper_ch>
like when I'm working on the government project.... I have to use their hardware
<sphalerite_>
and you're willing to put your email credentials in those? :o
<bachp>
What is the recommended way to apply an additional patch to a package (matrix-synapse) in my configuration.nix? I found overrideDerivation or .override but I'm not sure how to use it.
<hyper_ch>
sphalerite_: need it for rss :)
<sphalerite_>
bachp: overrideAttrs is what you'd want
<bachp>
sphalerite_: Can you point me to an example?
<sphalerite_>
bachp: the function you pass to overrideAttrs is called with the original values of what's passed to mkDerivation, then hwatever it returns is merged (//) into it
orivej has joined #nixos
dan_b has joined #nixos
<i-am-the-slime>
I think I just realised my mistake
<i-am-the-slime>
I had it as caps:hyper or something and naively changed it to caps:ctrl
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/gcc-7 9a2914a Vladimír Čunát: icu: build with gcc6 on i686
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to gcc-7: https://git.io/vAl04
<bachp>
sphalerite_: So as far as I understand now .override can't be used to add patches, but overrideAttrs can
<sphalerite_>
bachp: yes. override changes what's passed into the package function, at the top of the file. overrideAttrs changes what's passed into stdenv.mkDerivation
jrolfs has joined #nixos
<bachp>
sphalerite_: Ok I will give it a try. Thanks
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] mredaelli opened pull request #35096: Some QuantStack libraries (master...master) https://git.io/vAl0i
NixOS_GitHub has left #nixos [#nixos]
asuryawanshi has joined #nixos
<i-am-the-slime>
Hmm, is there another name for the tab key?
pkill9 has joined #nixos
jrolfs has quit [Remote host closed the connection]
asuryawanshi has quit [Ping timeout: 256 seconds]
ottidmes has joined #nixos
jrolfs has joined #nixos
buovjaga has joined #nixos
ma27 has quit [Ping timeout: 256 seconds]
ma27 has joined #nixos
<clever>
sphalerite_: i'm doing some cleanup on my machine, and i'm finding result links everywhere, just found a /etc/nix/result left-over from nixos-rebuild build
jrolfs has quit [Ping timeout: 264 seconds]
<buovjaga>
I'm trying to do this inside a qemulated aarch64 img: nix-build '<nixpkgs/nixos>' -I nixos-config=nixos/modules/installer/cd-dvd/sd-image-aarch64.nix -A config.system.build.sdImage
<buovjaga>
but I get: warning: Nix search path entry ‘nixos/modules/installer/cd-dvd/sd-image-aarch64.nix’ does not exist, ignoring
<buovjaga>
I've never done it before, how is it supposed to work?
<clever>
buovjaga: you need to give it the path to the nixos directory inside the nixpkgs folder
<clever>
buovjaga: you need to insert the full path that --find-file finds
<buovjaga>
ok sorry about that
<sphalerite_>
hyper_ch: good to know
<clever>
2822 store paths deleted, 3856.34 MiB freed
<clever>
sphalerite_: wow, got 3.8gig just from /etc/nix/result
<buovjaga>
maybe I can write a howto if I succeed with this :)
<buovjaga>
I already have notes on everything up to this
robstr has joined #nixos
silver has joined #nixos
<hyper_ch>
sphalerite_: works fine with sqlite.. still not sure about the permissions though
<buovjaga>
I want to produce my own custom aarch64 images and now it is possible to do inside qemu, so I thought it is the most convient way... fingers crossed
i-am-the-slime has quit [Remote host closed the connection]
<sphalerite_>
just a random question, how likely is executing random data on a modern CPU (various architectures) to cause any damage? I'm guessing pretty unlikely because it'll try to access invalid memory or try to execute an invalid instruction first
<hyper_ch>
(what user nginx runs under and is needed for accessing files/folders)
<sphalerite_>
hyper_ch: ps -f $(pgrep nginx)
<sphalerite_>
:)
<clever>
sphalerite_: there are programs designed to do just that on purpose, one min
<buovjaga>
sphalerite_: you are looking into cpu fuzzing?
<sphalerite_>
clever: no I don't mean fuzzing the instruction set
<sphalerite_>
I mean (accidentally) executing just plain random data
<clever>
sphalerite_: do you mean damage to the cpu, or damage via executing a syscall that does harm?
<sphalerite_>
both I guess
<sphalerite_>
and the same if the code is running in kernel mode
<clever>
sphalerite_: if that fuzzer hasnt broken the chip, then the cpu likely cant be damaged
<sphalerite_>
fair enough
<clever>
as for just running random garbage, chances are, its going to try to address something outside the mappings and cause a page fault
<clever>
if the opcode is even valid
<sphalerite_>
fair enough
<sphalerite_>
I guess that's less likely on a 32-bit ARM device in kernel mode though for instance?
<clever>
yeah
ckopo has joined #nixos
<clever>
in kernel mode under linux, the 0-3gig range is mapped to the current process, and 3-4gig kernel land
<sphalerite_>
although I don't really have any idea how MMUs work
<clever>
and depending on how much ram you have, the entire 3-4 range may be valid
<sphalerite_>
right
<sphalerite_>
can an ARMv7 device have more than 4G of physical ram?
<sphalerite_>
Also, unrelated, are there tools for working out whether (and if not, why not) a derivation is reproducible with nix? Beyond nix-build --check of course
<sphalerite_>
say I've built a tree of derivations on two different machines and want to compare the results
<buovjaga>
when booting with qemu to the arrch64 img, I had to use -m 2048 as 4096 or higher did not work
<clever>
LPAE is an optional feature that supports a 40 bit address bus, while operating in 32bit mode
<clever>
sphalerite_: there is an --option build-repeat 2
<sphalerite_>
ah neat
<clever>
sphalerite_: which will build it several times, then tell you if it differed
<sphalerite_>
clever: but I want to compare the results from different machines
<clever>
sphalerite_: only thing i can think of then is to nix-store --query --hash each storepath after building, and diff them
<sphalerite_>
something that could diff multiple nix-store --exports would be neat
<sphalerite_>
hm yeah I could do that
<clever>
The Large Physical Address Extension (LPAE), which extends the physical address size from 32 bits to 40 bits, was added to the ARMv7-A architecture in 2011.[104]
<sphalerite_>
buovjaga: huh. but aarch64 is 64-bit so it should support more physical RAM anyway?
<sphalerite_>
well it definitely does, like the community box has 256GB iirc
<clever>
buovjaga: there is also qemu-user that may be of interest to you
<clever>
buovjaga: the qemu-arm64 attribute in here builds a variant of qemu-user that supports aarch64
<sphalerite_>
so why would qemu not support 4G of RAM?
<clever>
if you run the ./result/bin/register within it, then the linux host magically gains the ability to just run aarch64 ELF files
<clever>
without having to boot an arm kernel under qemu
<buovjaga>
ok, thanks
<clever>
buovjaga: and the nix patch in the same file, will add a build-extra-platforms field to /etc/nix/nix.conf, so you can convince nix-daemon to run binaries that shouldnt work
kosmikus has joined #nixos
<ij>
Does hydra build the cache for all the haskellPackages or just some?
<sphalerite_>
all of them iirc
<ij>
I couldn't get telegram-api to come from cache, it kept being built.
<ckopo>
hello, need a little help with virtualbox. there's a mismatch of SHAs in package and on site, so I can't upgrade.
<ottidmes>
Are the nix-* tools interpreting <nixpkgs-overlays> differently from nixos? I am getting: error: opening directory ‘/wheel/proj/nixcfg/shared/config/nixpkgs/overlays.nix’: Not a directory
<ottidmes>
While if I check the source code that uses <nixpkgs-overlays> in nixos, it accepts both a directory and a file
<peel[m]>
C
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] tohl opened pull request #35098: sbcl: 1.4.3 -> 1.4.4, tested on nixos (master...master) https://git.io/vA8fE
<sphalerite_>
ij: nix-instantiate --eval '<nixpkgs>' -A haskellPackages.telegram-api.meta.hydraPlatforms
<sphalerite_>
idk why or the details of how this stuff is done
jrolfs has quit [Ping timeout: 268 seconds]
klntsky has joined #nixos
monotux has joined #nixos
spear2 has quit [Remote host closed the connection]
<sphalerite_>
ij: I think it's because it depends on servant-client-core which is also set not to build on hydra, but again I don't know why
<sphalerite_>
maybe someone who knows about the haskell infrastructure ( fpletz ?) can enlighten us
<clever>
sphalerite_: i think all of the haskell packages default to an empty hydra platforms, too much to build
<clever>
sphalerite_: so anything that is indirectly referenced, via something in the pkgs. set has to pull it in
<WilliamHamilton>
hi, I'd like a pointer on overlays: let's say I want to update this expression https://github.com/NixOS/nixpkgs/blob/master/pkgs/development/tools/minizinc/default.nix ; I think just changing the git commit and the other details should suffice, and I'd like to test it in an overlay; how would I write the overlay in my `.config/nixpkgs/overlays/` folder?
<sphalerite_>
clever: no… folgerhs for instance, which I doubt anyone but me and my friend who wrote it has ever used, is in hte binary cache
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dywedir opened pull request #35101: hack-font: 2.020 -> 3.002 (master...hack) https://git.io/vA8Jt
NixOS_GitHub has left #nixos [#nixos]
<clever>
sphalerite_: ah, i see, in hackage-packages.nix, some have no hydra platforms, some dont
<sphalerite_>
clever: exactly
<clever>
maybe its only those that declare themselves to have an executable
<clever>
libraries are of no use on their own
<buovjaga>
ahh, image built and I was able to boot into it in qemu :) now I should ask you silly questions about options
<sphalerite_>
clever: halma-telegram-bot seems to have an executable but depends on telegram-api
<sphalerite_>
I think
thc202 has joined #nixos
<sphalerite_>
yep
<clever>
sphalerite_: halma doesnt define itself as having executables
<sphalerite_>
clever: it's not marked as dontDistribute either though
<sphalerite_>
clever: I think it's because servant-client-core's dependencies aren't satisfied
<sphalerite_>
it wants newer versions of http-client, servant, and text
<sphalerite_>
oh wait "isInDefaultPackageSet" as well
<sphalerite_>
yeah no I still think it's because of the failing dependencies
<clever>
sphalerite_: its also possible that the hydra platforms is allowing hydra to build it, but hydra just fails
<clever>
so it doesnt get into the cache
<sphalerite_>
WilliamHamilton: for changes that you plan to PR to nixpkgs using an overlay isn't that useful though
<sphalerite_>
clever: the hydraPlatforms is set to [] for it though
ambro718 has joined #nixos
<WilliamHamilton>
ottidmes: eh, unfortunaly I had already read those two and watched the related presentation. I'm afraid I don't understand a more basic thing: the overlay takes `super` and `self`, while the original expression `default.nix` takes `{ stdenv, fetchFromGitHub, cmake, flex, bison }:` and calls `mkDerivation`. How to mix those two things?
<sphalerite_>
nix-instantiate --eval '<nixpkgs>' -A haskellPackages.servant-client-core.meta.hydraPlatforms
<clever>
sphalerite_: not sure then
<ambro718>
I have an issue with cupsd in nixos-unstable, it logs a message "Expiring subscriptions..." each second.
<sphalerite_>
clever: I think it's the dependencies thing
ThatDocsLady has quit [Read error: Connection reset by peer]
<WilliamHamilton>
sphalerite_: I could do a PR, but I don't know how to test that the expression is valid
<sphalerite_>
WilliamHamilton: make your change in a checkout, nix-build path/to/checkout -A minizinc
<WilliamHamilton>
sphalerite_: in the past I tried to base my installation of a local copy of nixpkgs, and it was a mess (undoubtely due to my inexperience)
<WilliamHamilton>
sphalerite_: will do that, thanks
<sphalerite_>
WilliamHamilton: to base a full nixos generation on it, just pass -I nixpkgs=path/to/checkout to nixos-rebuild
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] hamishmack opened pull request #35102: cc-wrapper: Use response files for cc (master...response-files) https://git.io/vA8Ja
NixOS_GitHub has left #nixos [#nixos]
jrolfs has joined #nixos
<ottidmes>
WilliamHamilton: This should suffice to load your package (stdenv.mkDerivation), in your overlay do: sedutil = super.callPackage ./sedutil { }; (where in my case it is for sedutil, which you of course needs to change to reflect your package
<ij>
sphalerite_: I'm not sure which one of those, but couldn't build because of a lack of ram, so maybe that's why.
<sphalerite_>
ij: well telegram-cli won't build on current nixos-unstable because some of the versions won't match up
<ottidmes>
WilliamHamilton: Looking at nixpkgs/pkgs/top-level/all-packages.nix was insightful for me at the time
<ij>
sphalerite_: Ah right, I had to override servant, so I guess that's why it doesn't build.
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to gcc-7: https://git.io/vA8TP
<NixOS_GitHub>
nixpkgs/gcc-7 088d7c9 Vladimír Čunát: libffado: 2.3.0 -> 2.4.0 (maintenance)...
NixOS_GitHub has left #nixos [#nixos]
niklob has joined #nixos
cfricke has quit [Quit: WeeChat 2.0]
zzamboni has joined #nixos
zzamboni has quit [Client Quit]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bjornfor opened pull request #35107: qt5X.full: append "-full" to store path name (master...qt5full-add-full-to-name) https://git.io/vA8kJ
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 2 new commits to master: https://git.io/vA8kT
<NixOS_GitHub>
nixpkgs/master 0552147 Michal Rus: catdocx: init at 20170102
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 57062da Michael Raskin: Merge pull request #35105 from michalrus/catdocx...
<clever>
buovjaga: create a configuration.nix file, and add the original nixos-config= path to the imports list
<clever>
buovjaga: then switch to nixos-config=./configuration.nix
<clever>
you are now free to add any nixos option you want to the image
<buovjaga>
ok thanks, maybe it's so simple I don't get it :)
<clever>
but imports entries like that cant be overriden
<clever>
so youll want to copy that .nix file out of nixpkgs, modify the copy (and fix the relative paths), then use the copy
kelleyNif has joined #nixos
<buovjaga>
yeah I tried it like that actually, but I couldn't get it to work. I put absolute paths like /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/profiles/base.nix but it still said error: getting status of ‘/profiles/base.nix’: No such file or directory - Maybe I have to try again
<clever>
it should give line numbers telling you where the issue is
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to gcc-7: https://git.io/vA8kr
<NixOS_GitHub>
nixpkgs/gcc-7 0a52c59 Vladimír Čunát: dbus_cplusplus: fix build with gcc7 via Fedra patches
NixOS_GitHub has left #nixos [#nixos]
kelleyNif has quit [Client Quit]
Neo-- has joined #nixos
sternenseemann has left #nixos [#nixos]
<buovjaga>
duh, I had not updated the .nix files inside the img, trying harder now
<hyper_ch>
so, rss2emails have a problem with a feed that seems to work just fine... but for the rest it runs really well... in addition I have setup the nixos-mailserver... works like a charm and and also locally deployed roundcube for when I don't have my email client available
niklob has quit [Ping timeout: 248 seconds]
Izorkin has joined #nixos
<srk>
hyper_ch: nixos-mailerver? tell me more
<srk>
found it, cool
rauno has quit [Ping timeout: 248 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] michalrus opened pull request #35108: arping: init at 2.19 (master...arping) https://git.io/vA8Is
NixOS_GitHub has left #nixos [#nixos]
<steveeJ>
loading Docker images from nix store lists them as created 48 years ago, awesome :-D
<genesis>
EPOCH :)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to gcc-7: https://git.io/vA8IR
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/gcc-7 bf25148 Vladimír Čunát: ntopng: fixup build with gcc7...
<srk>
hyper_ch: full of win. exactly what I need to get rid of the rest of non-nixos zombies
<hyper_ch>
then download full roundcube package, unpack, set permissions and run through installer using sqlite
<hyper_ch>
will try to make a PR to add that to nixos-mailserver as well
Itkovian has joined #nixos
<srk>
yeah, would be nice if there's no need to run the installer but I can try to work on that
<hyper_ch>
since if you opt to use letsencrypt on nixos mailsrver it also uses nginx there and might lead to problems.... not so sure about how nixos/nginx handles all that
<hyper_ch>
I get my LE certs through DNS-01 and use them for syncthing, now nixos-mailserver, ngingx, shellinabox etc
netixx has joined #nixos
jrolfs has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<netixx>
Hello, I am trying to make evolutions to the openvswitch vswitch configuration (https://github.com/NixOS/nixpkgs/issues/34851). I have performed the upgrade to the latest openvswitch version (as suggested by @volth) which went well. Now I am trying to add more advanced feature (such as vlans), but I am not sure about the format to use for configuration.
<srk>
hyper_ch: tried enableACME few days ago and it worked like a charm
jrolfs has quit [Ping timeout: 240 seconds]
<steveeJ>
what would be the simplest way to declare a script at say /bin/preCmd.sh in with dockerTools.buildImage? do I need a derivation that has $out/bin/preCmd.sh?
<steveeJ>
otherwise it would also be interesting to just write something to /any/path/script.sh
<NixOS_GitHub>
[nixpkgs] michalrus opened pull request #35109: arpoison: init at 0.7 (master...arpoison) https://git.io/vA8L6
NixOS_GitHub has left #nixos [#nixos]
<ottidmes>
I just tried nix 2.0 to see if it fixed an issue I had, but I got: warning: in configuration file '/etc/nix/nix.conf': unknown setting 'ssh-substituter-hosts'
tertleeltret has quit [Quit: Connection closed for inactivity]
ckopo has quit [Quit: Page closed]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ambrop72 opened pull request #35113: cups: Add patch for bug where CUPS fails to save files to disk (master...cupsd-fix) https://git.io/vA8t0
NixOS_GitHub has left #nixos [#nixos]
pie_ has joined #nixos
coot has joined #nixos
fendor has joined #nixos
ryanartecona has joined #nixos
Itkovian has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/gcc-7 837de09 Vladimír Čunát: mp4v2: fix build with gcc7 via Fedora patch...
<NixOS_GitHub>
[nixpkgs] vcunat pushed 2 new commits to gcc-7: https://git.io/vA8tQ
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/gcc-7 5beae83 Vladimír Čunát: blender, openimageio: build with gcc6 again...
<shlevy>
vcunat: Are you working on general stdenv stuff or just gcc7?
<vcunat>
just gcc7 right now
<shlevy>
OK
zarel has joined #nixos
Itkovian has quit [Client Quit]
simendsjo has joined #nixos
<zarel>
I'm installing another nixosbox (yay!), I booted in the graphical live but the backspace doesn't work somehow, it keeps the characted printed on screen
<zarel>
and prints a space instead
<lassulus>
try export TERM=vt100
<zarel>
it works! thank you lassulus
<lassulus>
well it works for now, but other software could break
<lassulus>
there was a way to repair the correct path, give me a sec
<zarel>
if it's only an issue of the installer I can live with it for the next hour or so
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] oxij opened pull request #35115: libva: make `libva-full` the default (staging...pkgs/libva-full-by-default) https://git.io/vA8qt
NixOS_GitHub has left #nixos [#nixos]
<zarel>
I'm gonna try anyway
robstr has quit [Remote host closed the connection]
robstr has joined #nixos
<zarel>
fun fact, if you add a trailing slash to the nix-store --repair-path it errors with "cannot unlink... directory is not empty"
<buovjaga>
as a little test in my aarch64 image building, I tried services.xserver = { enable = true; but it failed with error: Package ‘xf86-video-intel-2017-10-19’ in /nix/store/xg4a61cxwp57s76a5qhvr01k1vchk85z-nixos-18.03.git.e2d25035155/nixos/pkgs/servers/x11/xorg/overrides.nix:600 is not supported on ‘aarch64-linux’, refusing to evaluate.
<buovjaga>
for now I'm building without xserver, but would be interesting to learn what is the way to declare I want X...
<clever>
buovjaga: for the above error, i believe you need to set services.xserver.videoDrivers
<clever>
buovjaga: vesa is the simplest driver i believe
<buovjaga>
ok great, thanks again :) for now my more minimal build is chugging away
<buovjaga>
I am surprised I managed to get this far :)
robstr has quit [Ping timeout: 268 seconds]
<buovjaga>
ah, now that I peeked at Dezgeg's nixconfs I see videoDrivers = [ "modesetting" ];
simendsjo has quit [Remote host closed the connection]
alex`` has joined #nixos
<aristid>
buovjaga: you could probably send in a pull request that changes the default not to include "intel" on non-x86 machines
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vA8qp
<hask_bee_3>
They give this example of how to use Nix in stack.yaml: packages: [glpk, pcre]
<hask_bee_3>
Why would anyone want to do that? That doesn't specify the versions of glpk and pcre. Isn't the whole point of Nix to clearly define versions?
zzamboni has quit [Quit: Leaving.]
<clever>
hask_bee_3: you can only clearly define the versions if you provide a shell.nix file
<hask_bee_3>
clever ok, so that's what i should *always* do then right?
<hask_bee_3>
because otherwise why even bother with Nix...
<hask_bee_3>
if i haven't defined versions properly
<clever>
hask_bee_3: you could also look into stack2nix, which will build the entire project using nix
<hask_bee_3>
just manually look at the docs for every single haskell library in the dependency graph?
<clever>
hask_bee_3: build it in the nix sandbox, so things you forget are just not there, then look at the errors
<hask_bee_3>
:)
<zarel>
do you see any evident trouble in sharing /home with another distro (Arch, specifically)?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar opened pull request #35120: meson: print failed test logs (staging...meson-fail-logs) https://git.io/vA83q
NixOS_GitHub has left #nixos [#nixos]
<clever>
zarel: if you mount the nixos /nix/store within arch, and configure nix-daemon to run, you could share your entire nix-env profile between both with little issues
<zarel>
clever: thanks for the tip, I may try. what about uids and gids not matching?
<clever>
zarel: youll want to ensure they match up between all users in /home
<clever>
zarel: users.extraUsers.clever.uid = 1000; for example, pins the uid on the nixos side
coot has quit [Quit: coot]
<zarel>
clever: old files created with the old ID must be "migrated" manually, right?
<clever>
zarel: yeah
ryanartecona has quit [Quit: ryanartecona]
<zarel>
clever: thank you a lot :-) I'm definitely gonna do it
<hask_bee_3>
clever that's a great idea. Would you agree with me that "packages: [glpk, pcre]" (in stack.yaml) is great to use *during development* to determine the latest dependencies, and that when it's time to deploy, you take the versions Nix happened to use and bake them into shell.nix?
<clever>
hask_bee_3: you could also just take the nixpkgs revision your using, and put that into the shell.nix
<clever>
hask_bee_3: that will then just lock the versions of everything
<hask_bee_3>
ah, even better!
<hask_bee_3>
your nickname is accurate
<zarel>
I'm trying to install two distros alongside with a couple of LUKS containers, some with LVM and some not. Ubuntu screwed from the beginning, Arch worked after figuring out grub values... NixOS autogenerated the configuration correctly almost out of the box
<hask_bee_3>
thanks
<clever>
zarel: :D
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] xeji opened pull request #35121: firehol: 3.1.5: fix errors when running firehol command (master...firehol-35114) https://git.io/vA83K
NixOS_GitHub has left #nixos [#nixos]
<ixxie>
is there an intro to Nix 2 somewhere?
<makefu>
it would be great to have an overview "function | old command | new command"
<ixxie>
hmm weird
<ixxie>
I thought all the new commands are nix X Y Z
<ixxie>
I tried nix --help but it says I don't have nix installed
Neo-- has quit [Ping timeout: 265 seconds]
<ixxie>
srhb: I would still need to explicitly install nixUnstable?
robstr has quit [Remote host closed the connection]
<cinimod_>
clever: that is great but sadly my nix fu is not up to using it without a few hours work - there's a tutorial here: https://github.com/Gabriel439/haskell-nix/blob/master/project4/README.md but maybe I will come back to it in a few days (if byorgey hasn't fixed his package by then)
magnetophon has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 47f25fa Vladimír Čunát: redshift, liferea: remove the extra cache file...
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vA8ZS
NixOS_GitHub has left #nixos [#nixos]
<magnetophon>
I noticed the hash of the unstable channel was reverted back to a previous version. Can anybody tell me why that happened?
<magnetophon>
Should I get rid of the generation that was reverted?
<clever>
magnetophon: depends on what was reverted, and ive never seen a channel go backwards like that
<samueldr>
nixos-unstable or nixpkgs-unstable?
<magnetophon>
samueldr: nixos
<obadz>
when are we forking 18.03 btw?
<clever>
obadz: probably in the 3rd month
<obadz>
hehe
<obadz>
never too much time in beta huh
<samueldr>
using the irc logs I don't see anything like a channel going back... though nixos-unstable-small advanced, then nixos-unstable advanced to an older rev, an finally nixos-unstable advanced to the same revision than nixos-unstable-small
<magnetophon>
samueldr: it went from 9b25b9347d9 to something else and back
<sphalerite_>
how can I get a nix-shell for a part of stdenv that's failing to build?
<hask_bee_3>
What is the Nix equivalent of Stack resolvers, and where can I see my current Nix "resolver"?
humanoyd has joined #nixos
<simpson>
hask_bee_3: I *think* that Nix does not have any such thing. What are you trying to do?
<magnetophon>
samueldr: that hash from 20 hours ago I don't understand. https://howoldis.herokuapp.com/ claims the latest change is from 22 hours ago, to the hash mentioned above.
<samueldr>
magnetophon: how did you notice it was "reverted back to a previous version"?
<samueldr>
the "XX hours ago" from the IRC bot is in relation to when the irc bot posted
<sphalerite_>
is there a way to try building one specific package that's part of the stdenv bootstrap?
<magnetophon>
samueldr: my grub has twice the same hash with a different one inbetween. Also I seem to remeber an update only a few hours after the last, which I think was the revert.
<hask_bee_3>
simpson I'm seeing a lot of Nix examples where people are specifying package names (ex: buildInputs = with pkgs; [ sassc zlib ]; ) without any mention of *which versions* to pull. This has me all confused. I thought the whole point of Nix was to clearly define package versions you want to use.
<hask_bee_3>
That's why I wondered whether it's actually using a "resolver" to determine the version.
<clever>
hask_bee_3: it uses whatever version is defined inside the nixpkgs files
<simpson>
hask_bee_3: Nope, it's merely nixpkgs having versions of stuff available already. If, for example, zlib 4.2 were the version that everybody wanted, then likely nixpkgs would have a `zlib_42` which people could use instead of `zlib`.
<samueldr>
weird, the channel is at 9b25b9347d9, but the nixpkgs-channels repo shows it later, and the channel history at nix.gsc.io https://channels.nix.gsc.io/nixos-unstable/history shows two further updates :/
<hask_bee_3>
simpson But isn't it possible that "zlib" could tomorrow suddenly refer to zlib_43 instead of zlib_42?
<hask_bee_3>
that could break things
<sphalerite_>
hask_bee_3: that's what we have CI for
<simpson>
hask_bee_3: Sure, and many Haskell packages are broken because their version specifications are overly-restrictive. Happens.
silver has quit [Read error: Connection reset by peer]
<sphalerite_>
hask_bee_3: and no, it won't just spontaneously change to zlib 4.3. It'll change when nixpkgs is updated to contain that newer zlib
guibou has joined #nixos
<sphalerite_>
hask_bee_3: but if it does break, the older version is still around and there's nothing stopping you from going back to it
<magnetophon>
samueldr: would it help to know if the "in between" hash in my grub is one of the two further updates? If yes, can I find out without rebooting?
<sphalerite_>
I have a range of commits and I want to find the most recent one that changed a particular derivation. How can I do this?
<sphalerite_>
That is, the derivation and not the expression. I've found several changes to the expression in the range, but none of them affected the derivation
<i-am-the-slime>
I see there are some options there.
<hask_bee_3>
sphalerite_ I'm still wrapping my mind around this. Suppose a Linux server running a webapp through Nix reboots. Couldn't it suddenly fetch say a newer zlib on next launch (after reboot) if zlib version wasn't specified and a newer zlib was in the meantime made available in nixpkgs?
<sphalerite_>
hask_bee_3: no, it doesn't fetch anything automatically
<clever>
hask_bee_3: the version can only change when you update the nixpkgs version, which is usually at nix-channel --update
<sphalerite_>
hask_bee_3: unless you're using nix-shell with the channel URL or something
<clever>
hask_bee_3: but if you lock the nixpkgs version in your .nix files, it will only update when you change that version in the file
<hask_bee_3>
clever Ah. Is there a way I can see my current global nixpkgs version? It's gotta be somwhere in ~ I guess
<clever>
and also nix-instantiate --find-file nixpkgs
<sphalerite_>
nix-instantiate --eval '<nixpkgs>' -A lib.nixpkgsVersion
<sphalerite_>
:p
<hask_bee_3>
Would you say that Nix's nixpkgs version is analogous to Stack's "resolvers"?
<yonk_>
hey, i wanted to try to run something in wine quickly and dropped into a nix-shell with it (`nix-shell -p wine`), but for some reason the wine I got was 32 bit and I cannot execute it. I am not sure what to do, this never happened to me before. It seems to me that all hydra builds for wine are 32 bit (https://hydra.nixos.org/job/nixos/release-17.09/nixpkgs.wine.x86_64-linux), but is there some way to just
<yonk_>
force nix-shell to build the packages locally?
<clever>
hask_bee_3: but better, since it covers every single system library and executable
<hask_bee_3>
yeah
<hask_bee_3>
Everything makes sense now.
<hask_bee_3>
Thank you!
<clever>
yonk_: wine defaults to 32bit mode, not much uses 64bit wine
<clever>
yonk_: wine.override { wineBuild = "wineWow"; }; will give you a 64bit capable wine
<yonk_>
clever: I see, but I cannot execute it when it is 32bit, I get an exec format error
i-am-the-slime has quit [Remote host closed the connection]
<clever>
yonk_: what distro are you on?
<sphalerite_>
yonk_: that's very strange, what kind of machine are you on and how are you trying to run stuff?
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
<yonk_>
clever: nixos 17.09
<hask_bee_3>
Is there a way to browse all packages in say "18.03pre125026.f607771d0f5" ?
<hask_bee_3>
On some website online
<hask_bee_3>
(similar to stackage.org allowing browsing of stack resolvers)
<yonk_>
sphalerite_: just a thinkpad, I ran `nix-shell -p wine` and inside the shell just `wine`
<clever>
yonk_: can you run nix-store --verify --check-contents ?
<sphalerite_>
hask_bee_3: not really. There's a list of packages for the latest version of nixos 17.09 at https://nixos.org/nixos/packages.html (iirc)
Baughn has joined #nixos
<sphalerite_>
hask_bee_3: but it doesn't include nonfree packages and stuff
<sphalerite_>
hask_bee_3: you can use nix-env -qaP to list packages in your local package set though
simukis has quit [Remote host closed the connection]
<clever>
yonk_: and if you run the file command on it?
<yonk_>
clever: /nix/store/6lzbj0al7jshsa96zl1q2fqbw89v8wyi-wine-2.0.2/bin/wine: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /nix/store/32cv6nxi3bk9zpx184nzx58vfp2zspnc-glibc-2.25-123/lib/ld-linux.so.2, for GNU/Linux 2.6.32, with debug_info, not stripped
<clever>
yonk_: hmmm, that should work
nuncanada has quit [Read error: Connection reset by peer]
orivej has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dotlambda opened pull request #35126: pythonPackages.pynacl: use system libsodium (master...pynacl) https://git.io/vA8Wt
NixOS_GitHub has left #nixos [#nixos]
<yonk_>
clever: should I open an issue about it?
<clever>
yonk_: can you gist your configuration.nix file?
<clever>
yonk_: and also nix-shell -p gist followed by `cat /proc/config.gz | gunzip | gist -p -`
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/gcc-7 13ec57e Vladimír Čunát: yap: -fpermissive to fix build with gcc7
<NixOS_GitHub>
[nixpkgs] vcunat pushed 2 new commits to gcc-7: https://git.io/vA8Wu
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/gcc-7 666e7ca Vladimír Čunát: k3d: fix by building with gcc6
<hask_bee_3>
clever, simpson, sphalerite_ I think I get it now. I have written down a few bullet points based on your insights. Are they an accurate depiction of how Nix should be used in practice? https://pastebin.com/ccSa3e64
<simpson>
hask_bee_3: Most packages *are* versioned inside nixpkgs, but the version numbers aren't used by Nix, just by users.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] netixx opened pull request #35127: nixos/openvswitch : upgrade 2.5.4 -> 2.8.1 and vlans (master...better-openvswitch) https://git.io/vA8Wg
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
hask_bee_3: the "they can opt specifically for an older version of a package" only applies to a few packages which are explicitly packaged in multiple versions
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] netixx closed pull request #34586: nixos/vswitches : Fix vswitch-netdev.service going down just after startup (master...fix-networking.vswitches-#34336) https://git.io/vNxFX
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
hask_bee_3: but yes, typically people pin nixpkgs when using it with a project that isn't inside
<sphalerite_>
ugh, so used to writing bash that I never know if the code that I write doesn't use bash-specific features that won't work in busybox's ash
<hask_bee_3>
sphalerite_ by "inside" you mean "inside nixpkgs"? as in an external project, such as my great webapp
<sphalerite_>
hask_bee_3: yep
<hask_bee_3>
ok cool
<clever>
yonk_: nothing looks abnormal in that config
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rvl opened pull request #35129: buildkite-agent service: declarative hooks and extraConfig option (master...buildkite-agent-hooks) https://git.io/vA8WQ
NixOS_GitHub has left #nixos [#nixos]
dan_b has joined #nixos
alex`` has quit [Ping timeout: 255 seconds]
roblabla has joined #nixos
<sphalerite_>
does anyone know of a simple wrapper for nix-build that will cache build failures by derivation? Would make this bisect a bit quicker :)
rauno has quit [Ping timeout: 252 seconds]
<clever>
sphalerite_: sorta, but the change i have in mind also breaks the binary cache
<clever>
sphalerite_: if you set succeedOnFailure=true; on a derivation, it will still succeed, even upon failure, and create a $out/nix-support/failed file
<clever>
that allows nix to remember it
<sphalerite_>
ooh
<clever>
but adding that to a derivation, changes the hash, so you get zero help from the binary cache
<clever>
rebuild the world, at least once
<clever>
sphalerite_: oh
<clever>
sphalerite_: but if you know which derivation is failing a lot, you could config.nix package-override just that
<simpson>
sphalerite_: I'd go the other way and write a bisection wrapper which checks and memoizes derivation paths.
<sphalerite_>
I was thinking more along the lines of drv=$(basename $(nix-instantiate "$@")) ; if [[ -e /tmp/failed/$drv ]] ; then echo "cached failure" ; exit -1 ; fi ; nix-build "$@" || touch /tmp/failed/$drv
<sphalerite_>
simpson: like ^ ?
<simpson>
sphalerite_: Yeah, along those lines. If you're careful, you can actually wire this up to git-bisect directly and auto-bisect.
<NixOS_GitHub>
nixpkgs/master 6952a2b Joachim F: Merge pull request #35101 from dywedir/hack...
yonk_ has quit [Remote host closed the connection]
klntsky has joined #nixos
cnidario has joined #nixos
niklob has joined #nixos
rauno has joined #nixos
<sphalerite_>
hyper_ch: how do you remember which out of ${foo#bar} and ${foo%bar} is which? :p
<samueldr>
(I personally always use the bash repl because I can't remember % %% # ## which is the one I want at the moment)
<sphalerite_>
I check using man bash\n/##\n
<sphalerite_>
but it would be nice to be able to actually remember :p
MP2E has quit [Remote host closed the connection]
<sphalerite_>
Hm. With parallelism of some sort, it should be possible to do n-section rather than bisection with git, shouldn't it?
<sphalerite_>
I wonder how much faster it would be
<clever>
sphalerite_: that reminds me, i had an idea for a new builtin, par
Arcaelyx has joined #nixos
MP2E has joined #nixos
<sphalerite_>
meaning the same as in haskell?
<clever>
i think so
<clever>
just forceValue() each item in the list, in seperate threads
<sphalerite_>
do it! :p
<sphalerite_>
That n-section thing would be really nice though for finding failure points in nixpkgs when you can spread the building across multiple machines
ambro718 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vA88I
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to gcc-7: https://git.io/vA88t
<NixOS_GitHub>
nixpkgs/gcc-7 ba5563b Vladimír Čunát: hhvm: fix build with gcc7
NixOS_GitHub has left #nixos [#nixos]
<simpson>
shlevy: Ooh, Nix-configurable binfmt sounds fun. What are y'all using it for, if you can share?
<samueldr>
ooh nice
<sphalerite_>
simpson: shlevy is porting nix stuff to RISC-V so I'm guessing that's what it's for
<shlevy>
simpson: I'm working on riscv support in nixpkgs, I have cross-compiled nix itself so I'm using qemu-user to run riscv builds without needing a full riscv system
<samueldr>
IIRC some distros use/used them for .net exe and/or windows exec with wine
<clever>
sphalerite_: youve probably seen my qemu-user stuff?
<sphalerite_>
shlevy: do you already know about clever's hack?
<shlevy>
Yeah, I'm using qemu-user with binfmt stuff
<simpson>
shlevy: I want what you're doing but for ARM, to combine with nix-bundle, so that I can build ARM stuff for a few Monte community members without waiting weeks for native builds. Thoughts?
<clever>
sphalerite_: this generates qemu-user's for armv7/8 and aarch64
<clever>
shlevy: oops, ^
<clever>
shlevy: and it also includes a nix patch to allow the x86 nix-daemon to run arm binaries
<sphalerite_>
simpson: clever's thing does that
<samueldr>
6/7?
<shlevy>
simpson: seems reasonable
<clever>
samueldr: 6/7 i mean
<simpson>
sphalerite_: I didn't think that clever's structure could be used with nix-bundle, which only works with nixpkgs.
<shlevy>
I also wrote a wrapper around qemu-user to support the C flag to binfmt_misc
<shlevy>
So it will work with e.g. sudo binaries
<clever>
simpson: that gives me an idea, if nix-bundle has root, it can configure binfmt-misc, and run things under a mount namespace
<simpson>
Like, I had to package (and will be upstreaming) a cut-down Monte interpreter into nixpkgs in order to nix-bundle it.
<sphalerite_>
simpson: clever's structure is for nixpkgs..?
* sphalerite_
is confused now
self has joined #nixos
<simpson>
sphalerite_: nix-bundle can only bundle derivations *in* a nixpkgs.
self is now known as Guest85664
<clever>
simpson: ive solved that
<simpson>
clever: That could be interesting for fat-binary-style things...
<simpson>
How have you solved it, and how can others use it?
<NixOS_GitHub>
nixpkgs/python-unstable 1237395 Frederik Rietdijk: pathpy fixup
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable a4e5c6d Frederik Rietdijk: python: pendulum: 1.4.0 -> 1.4.1
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from b33b38d to 999f10d: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 7c22371 Frederik Rietdijk: python: pexpect: 4.3.1 -> 4.4.0
<hyper_ch>
sphalerite_: can't get sieve to work properly in roundcube /me is a sad panda
<sphalerite_>
aww
<sphalerite_>
clever: symphorien: if you're on nixUnstable you can also use channel:nixos-unstable instead of the full URL
<clever>
sphalerite_: oh yeah, i saw that in the nix/release.nix, it got in my way when i tried to build nix 2.0 with 1.11 a few days ago
<sphalerite_>
it's very convenient though :p
humanoyd has quit [Quit: WeeChat 2.0.1]
zzamboni has joined #nixos
jacob_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/python-unstable 97229f1 Frederik Rietdijk: python.pkgs.numpydoc: 0.6.0 -> 0.7.0
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vA8B7
jacob_ is now known as Guest51236
zzamboni has quit [Client Quit]
newbie__ has quit [Ping timeout: 260 seconds]
robstr has quit [Remote host closed the connection]
robstr has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vA8Bx
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable dfe7856 Frederik Rietdijk: awscli: 1.14.32 -> 1.14.41
day|flip has quit [Remote host closed the connection]
lxpup_24904 has joined #nixos
jb554 has joined #nixos
robstr has quit [Ping timeout: 248 seconds]
ixxie has quit [Ping timeout: 256 seconds]
zzamboni has joined #nixos
jsgrant has joined #nixos
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable 21f95f4 Frederik Rietdijk: python.pkgs.h5py: fix for numpy 1.14
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vA8RY
niklob has joined #nixos
<jsgrant>
Okay, finally got up the courage and yeah -- am sticking with NixOS on my main boxen; Reinstalled to my laptop that has the kworker bug I thought I previously bypassed (apci interrupt, gpe13 causing a maxout on one core) and now on new 17.09 install same problem ...
<clever>
jsgrant: does /proc/interrupts blame that irq on any piece of hardware?
<jsgrant>
Anyone see anything wrong with 'services.cron.systemCronJobs = ["@reboot echo 'disable' > /sys/firmware/acpi/interrupts/gpe13"];' ?
<clever>
jsgrant: id use a one-shot systemd service to do the same thing
<clever>
and the other answer depends heavily on what /proc/interrupts blames it on
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] xeji opened pull request #35134: mfcl2720dwlpr / mfcl2720dwcupswrapper: init at 3.2.0-1 (master...mfcl2720dw) https://git.io/vA8Rc
NixOS_GitHub has left #nixos [#nixos]
fragamus has joined #nixos
<fragamus>
why is haskell stack so poorly supported in nixos
<clever>
fragamus: what part isnt working right?
bas_ has joined #nixos
<jsgrant>
clever: Hm, I don't even see 13 listed anymore when I cat /proc/interrupts
<clever>
jsgrant: how do you know its gpe13 causing the fault?
<fragamus>
install requires stack upgrade and that makes a second executable
<fragamus>
which is not under nixos control
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vA8R6
<NixOS_GitHub>
nixpkgs/master 655446c Vladimír Čunát: libav*: maintenance updates...
NixOS_GitHub has left #nixos [#nixos]
<clever>
fragamus: you need to work with the nix tools, not against them
<jsgrant>
clever: Because it was in Fedora 27 & even in NixOS awhile ago; And once disabled it at boot -- back to normal.
<clever>
jsgrant: id leave it disabled then, and see what happens
zzamboni has quit [Quit: Leaving.]
<fragamus>
please tell me what to do to work with nixos and stack
<jsgrant>
clever: Well, cron job claims it disables it but since I reinstalled it no-longer respects it; After it did for the week or-two I initially had it running here. :^I
<clever>
fragamus: nix-env -iA nixos.stack and stack --nix
Risty1_ is now known as Rusty1
mrkgnao has quit [Ping timeout: 240 seconds]
<fragamus>
nice i'll try that
<jsgrant>
Like /sys/firmware/acpi/interrupts/gpe14 should have 'disabled' subsituted in there on boot & did for sometime.
<fragamus>
it seems to require a stack upgrade but that makes the other executable that noxos doesn't control
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/python-unstable de88dd2 Frederik Rietdijk: python: aenum: 2.0.9 -> 2.1.0
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 21f95f4 to 7e2a46e: https://git.io/v5aIf
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable 6d7b06f Frederik Rietdijk: python: alembic: 0.9.7 -> 0.9.8
<NixOS_GitHub>
nixpkgs/python-unstable 8310a2b Frederik Rietdijk: python: argcomplete: 1.9.3 -> 1.9.4
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/staging d8c2bac Frederik Rietdijk: Merge pull request #35123 from NixOS/python-unstable...
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vA80U
NixOS_GitHub has left #nixos [#nixos]
<clever>
fragamus: what does nix-channel --list say?
tnks has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to release-17.09: https://git.io/vA80Y
<NixOS_GitHub>
nixpkgs/release-17.09 23e418f Vladimír Čunát: libav*: maintenance updates...
NixOS_GitHub has left #nixos [#nixos]
hiroshi- has joined #nixos
<fragamus>
clever: i updated the gist with the results of nix-channel --list
<clever>
fragamus: try doing `nix-channel --update` then `nix-env -iA nixos.stack` again, does it give a newer version?
niklob has quit [Ping timeout: 240 seconds]
tmaekawa has joined #nixos
<fragamus>
clever: i updated the gist with the results of nix-channel --list
hiroshi has quit [Ping timeout: 260 seconds]
hiroshi- is now known as hiroshi
<fragamus>
clever: i updated the gist with the results of doing `nix-channel --update` then `nix-env -iA nixos.stack` again
tmaekawa has quit [Client Quit]
katyucha has joined #nixos
jb554 has quit [Ping timeout: 276 seconds]
<clever>
fragamus: if you just want ghci, you dont even need stack, nix-shell -p haskellPackages.ghc --run ghci
<jsgrant>
clever: Okay - made a multiline note in my config; Tyvm!
<fragamus>
I want stack - i am going to deploy on stack
<clever>
fragamus: try using a proper stack.yaml then, it might only be the ghci command that is broken
<fragamus>
oh
<fragamus>
perhaps...
* jsgrant
is hoping to get his other laptop (the one that fails 75% of the time with NixOS installed) installed with NixOS by end of day today as-well.
<jsgrant>
Really I just think I need to enable systemd.coredump on that, hopefully find log-file & that'll be super obivious though.
<clever>
jsgrant: i have that on with a few machines, helps debug random things
<clever>
[root@amd-nixos:~]# ls -ltrh /var/lib/systemd/coredump/
<clever>
jsgrant: they land here
<clever>
and coredumpctl lets you list and gdb them
<fragamus>
clever: i tried something else like creating a project from a template but no workie; i updated the gist to show you
<jsgrant>
clever: Oh, that seems pretty easy actually; Yeah! Ty!
* jsgrant
really needs to get more comfortable with systemd and related tooling.
<jsgrant>
Looks like some actually pretty nice stuff there.
<clever>
fragamus: also, id avoid running that kind of stuff as root, just out of habbit
<fragamus>
ok
* jsgrant
threw another note in his config; Very good/nice stuff.
marusich has joined #nixos
jrolfs has quit [Ping timeout: 252 seconds]
* jsgrant
will brb; Rebooting host-machine.
<steveeJ>
where does /run/wrappers/bin/fusermount come from and how can I install it in a Docker container built with dockerTools?
rauno has joined #nixos
<fragamus>
clever this stack thing is really broken what would you suggest
jrolfs has joined #nixos
<clever>
fragamus: check the ~/.stack directory and see where that string from the error is held
<LnL>
steveeJ: the wrappers are created by the module system
<LnL>
and setuid is forbidden in nix builds
raynold has joined #nixos
<clever>
steveeJ: it gets created on bootup, by a process that has root
<shlevy>
OK, needed to disable sandboxing, but successfully ran a derivation with system = riscv64-linux through qemu-user
<clever>
steveeJ: so the docker scripts would need to recreate similar files before starting the non-root stuff
rauno has quit [Ping timeout: 256 seconds]
<clever>
shlevy: there is a nix.conf to inject things into the sandbox
<NixOS_GitHub>
[nixpkgs] abbradar opened pull request #35141: dhcpcd service: always want but set service order (master...dhcpcd-order) https://git.io/vA8E6
<NixOS_GitHub>
[nixpkgs] Kaali opened pull request #35142: rpm: Add Darwin support (master...master) https://git.io/vA8ES
NixOS_GitHub has left #nixos [#nixos]
earldouglas has joined #nixos
<earldouglas>
Does anyone know how to get gnuradio-companion to find the osmocom block? I have gnuradio-with-packages and gnuradio-osmosdr installed.
ertes has quit [Quit: Bye!]
Odays has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<earldouglas>
Solution: add the following to the [grc] section in ~/gnuradio/config.conf: local_blocks_path=/nix/store/dbn1y4ds2sczlxd3n5wz1jdpi8g1ppls-gnuradio-osmosdr-0.1.4/share/gnuradio/grc/blocks
<earldouglas>
This seems fragile, since it depends on the path of the installed gnuradio-osmosdr package.
<clever>
earldouglas: you could put a path to ~/.nix-profile/share/gnuradio/grc/blocks, and then install it with nix-env
webrover has joined #nixos
webrover has left #nixos [#nixos]
<earldouglas>
When I nix-env -i gnuradio-osmosdr, I don't see a ~/.nix-profile/share
suvash has joined #nixos
<symphorien>
how did you install gnuradio-with-package ?
suvash has left #nixos [#nixos]
<earldouglas>
I tried both via configuration.nix and via nix-env -i
webrover has joined #nixos
webrover has left #nixos [#nixos]
<steveeJ>
LnL: could I somehow set the setuid flag in the the runAsRoot in the Docker steps?
<earldouglas>
Oh, the missing share dir was a red herring. I needed to cd .. ; cd ~/.nix-profile to see it
<earldouglas>
Almost there: I can see the osmocom blocks, but on running I get "ImportError: No module named osmosdr"
<earldouglas>
symphorien: Thanks for the help. I think something is broken with the gnuradio-with-packages derivation. I see no results with grep osmo $(which gnuradio-companion)
<symphorien>
this commit is only two weeks old
<earldouglas>
I am indeed, and this was working with an older version of nixos.
<symphorien>
maybe the nixos-unstable version works with the patch above
<earldouglas>
I actually am on unstable on this machine. I'll update and see.
<earldouglas>
Thanks for the help.
zarel has quit [Quit: Leaving]
fragamus has joined #nixos
<fragamus>
OK I HAVE A GOAL - I aim to find someone who uses stack on nixos and adopt their ways of working
<clever>
fragamus: i prefer using stack2nix
<clever>
fragamus: it converts your stack.yaml into a default.nix, then nix does everything
<earldouglas>
symphorien: Success! I owe you a beer.
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jmackie has joined #nixos
KABA has joined #nixos
klntsky has quit [Ping timeout: 255 seconds]
<jmackie>
I'm trying to add vim plugins to my vimrc, but names prefixed with "github: ..." in the nixpkgs/vim-plugin seem to cause attributes error. What am I missing? Example error: `error: attribute ‘github:flazz/vim-colorschemes’`
<KABA>
Hi, is there any work being done on updating libreoffice to 6.0?
katyucha has quit [Ping timeout: 255 seconds]
fragamus has joined #nixos
<KABA>
I managed to get it to work with flatpak on NixOS which is fantastic but native version would be perfect
jmackie has quit [Client Quit]
KABA2 has joined #nixos
klntsky has joined #nixos
KABA2 has quit [Client Quit]
<Odays>
Hello , just wondering if anyone try working with custom kernel and patches on nixos? It the only thing I need to figure out. get my laptop working like normal. most everything work but. few weird issue with supend and hibernate.
<KABA>
Odays: boot.kernelPatches is the option you should be looking into, but I don't have any experience with it.
rihards has quit [Quit: rihards]
<Odays>
sweet. any other lead way will be awesome. trying to make my own kernel.nix package. may get a little overwilling
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
Itkovian has quit [Client Quit]
zzamboni has quit [Quit: Leaving.]
knupfer has quit [Remote host closed the connection]
alex`` has quit [Ping timeout: 255 seconds]
<ivanivan>
I'm considering switching channels from nixos-17.09 to nixos-unstable, because there are a number of bug-fixes that have merged into master but haven't made it into the stable channel.
<ivanivan>
How unstable is unstable? Would I be trading one problem for another?
hask_bee_3 has joined #nixos
<vcunat>
It's unstable mainly in the sense that it's changing.
<samueldr>
many users here are daily-driving (and thus dogfooding) unstable
<hask_bee_3>
How can I specify a particular nixpkgs revision in my Haskell Stack project's shell.nix file?
<shlevy>
ivanivan: Also 18.03 is just around the corner
<vcunat>
Yes, planned to be forked from master around the end of the month.
<vcunat>
"unstable" has the same channel-update criteria as stable, so various tests have to pass.
Itkovian has joined #nixos
<ivanivan>
Oh cool, i thought that might be the case. i'm tempted to stick with stable, but I don't know how rational the temptation is
zzamboni has joined #nixos
<disasm>
vcunat: is there a hit list of things that need resolved before 18.03 is stable? Hoping I'll have some time to do more contributions in March.
<sphalerite_>
ivanivan: the main stuff I'd worry about is newly introduced bugs (but they're not that likely to stop you from working since you have rollbacks) and the sort of changes that you'd usually go through on a major version upgrade being possible at any time (see previous release notes at https://nixos.org/nixos/manual/release-notes.html for examples)
<sphalerite_>
oh hey disasm haven't seen you around in a while
<vcunat>
disasm: I'm not aware of any real list. There's the milestone in github, but that's like some wishlist plans, and things get removed from there.
<disasm>
ivanivan: I run off a local nixpkgs I rebase off of 17.09 once a week or so and cherry pick commits from master that I want. Gives me the best of both worlds.
<disasm>
sphalerite_: hey!
<vcunat>
In March we'll probably start the usual Zero Hydra Failures effort.
<vcunat>
And there will now be aarch64-linux as well.
<disasm>
sphalerite_: yeah, my company wasn't too keen on my suggestion that I get company time to work on nixpkgs, and my son has been a lot more demanding this last few months (He's two).
hiratara has quit [Ping timeout: 265 seconds]
<sphalerite_>
aw, that's a shame
<sphalerite_>
and the son thing is understandable ^^ I imagine I was pretty demanding when I was two as well
<sphalerite_>
:D
<disasm>
sphalerite_: but... I'm switching jobs March 1st :) So I'll get to at least play with nixos on a daily basis again :)
jb554 has joined #nixos
<shlevy>
vcunat: is gcc-7 going through or straight to master?
<sphalerite_>
disasm: ooh, to a nixos-y job??
<disasm>
sphalerite_: you have no idea how excited I am to not have to touch puppet or ansible :)
<sphalerite_>
awesome
<vcunat>
shlevy: I'd hope to go straight, because we have binaries for it now
<ivanivan>
vcunat: samueldr: sphalerite_: disasm: thanks for the channel feedback. i'm going to switch to unstable and see how it goes
<vcunat>
we don't have binaries for current staging
vidbina has quit [Ping timeout: 256 seconds]
<sphalerite_>
ivanivan: :) and of course you can roll back from that too :D
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy created glibc-2.27 (+1 new commit): https://git.io/vA8w0
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/glibc-2.27 d471a5c Shea Levy: glibc: Make 2.27 the default.
hiratara has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Odays>
I am not getting firefox to show up on plasma 5.12. make me wonder if firefox.desktop got deleted after reinstall firefox. and notice that opera is still showing up even after removing it from my config
kim0 has joined #nixos
<ivanivan>
sphalerite_: how do you control which channel is used by `nixos-rebuild`? meaning, if I add the unstable channel and don't remove stable...
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to staging: https://git.io/vA8wN
<vcunat>
ivanivan: it uses root's channel that you called "nixos", by default
<ivanivan>
vcunat: makes sense. thanks!
<vcunat>
you may e.g. remove it and re-add with the new URL as root with the nix-channel command
<vcunat>
(or hand-edit the text file :)
hamishmack has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy opened pull request #35150: linuxHeaders: Bump default to 4.15. (staging...linux-headers-4.15) https://git.io/vA8rs
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
or pass -I nixpkgs=/path/to/alternative to build from another version on a one-off basis
<jsgrant>
clever: Welp, somehow this ultrabook did start the second-time I tried booting after install; Coredump enabled, but nothing is in /var/lib/systemd/coredump from the first-time failed.
<jsgrant>
Unless it clears it out on sucessful boot. :^P
<jsgrant>
Might just have to wait for it to fail then check in a chroot or similar.
<clever>
jsgrant: it should persist between reboots
<ivanivan>
sphalerite_: good to know - thanks
ivanivan has quit [Quit: WeeChat 1.9.1]
<jsgrant>
samueldr: I was for a lot of when I was first-dipping in (in unstable); It's pretty nice when you are using a lighter environment -- nothing founds this, but assume it's more of an issue on a 'proper desktop environment'.
<NixOS_GitHub>
[nixpkgs] bendlas pushed 1 new commit to master: https://git.io/vA8ru
NixOS_GitHub has left #nixos [#nixos]
hamishmack has quit [Quit: hamishmack]
ivanivan has joined #nixos
ivanivan has quit [Client Quit]
<jsgrant>
clever: It's hard to crash, because on ssd but type in luks pass, some services start, then huge kernel panic that fills at least 3 screen's worth.
zzamboni has quit [Quit: Leaving.]
<jsgrant>
Not instantly, systemd obviously starts and if it makes it to dbus I know it's going to boot up right -- but like 1/4th of the time it seems generally.
<jsgrant>
samueldr: About dogfooding unstable.
<clever>
jsgrant: kernel panics cant be captured by coredumpd
<clever>
jsgrant: if you have grub available, try enabling memtest86, and giving it a spin
* jsgrant
is running systemd-boot, but seems trivial to switch.
<jsgrant>
Also, ah okay.
hamishmack has joined #nixos
<samueldr>
clever: never really followed on that for "normal computers", but do you know if there's a way to do like on android(?) or ARM platforms and have a section of ram where the panics are written and can be logged on reboot?
jb554 has quit [Ping timeout: 256 seconds]
kelleyNif has quit [Ping timeout: 256 seconds]
<clever>
samueldr: thats usually handled by a crash kernel, which nixos partially supports
<clever>
in the event of a panic, it will kexec itself over to this new kernel
<clever>
where you can then use some special tools to dump the ram
<clever>
in a proper setup, it would save a dump and reboot itself
<clever>
but with nixos, it just boots the normal system from that 2nd kernel
<jsgrant>
It is weird, becuase I've also tried Archlinux on here besides the much earlier aforementioned Fedora and never had this issue.
schoppenhauer has quit [Ping timeout: 248 seconds]
<samueldr>
and now that I have asked the question, I'm searching a bit, android uses pstore for /sys/fs/pstore/console-ramoops, which is platform-dependent
<clever>
i think thats very limited in how much storage it has
<clever>
while the crashdump setup basically just boots a 2nd kernel+initrd that was kept in ram the entire time
<samueldr>
yes, and probably very-platform-dependent, I was just curiout what mechanism was used
<clever>
and keeps the original failed os fully in ram
<KABA>
Anyone started to work on libreoffice 6?
<clever>
the rtc block on x86 machines is sometimes abused for some of that
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vA8iB
<NixOS_GitHub>
nixpkgs/master 7f3c97a Shea Levy: riscv-pk: Init at 01.pre438_e5846a2.
NixOS_GitHub has left #nixos [#nixos]
astrofog has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 2 new commits to master: https://git.io/vA8iH
<NixOS_GitHub>
nixpkgs/master c7439b4 Shea Levy: Merge branch 'kernel-gcc7-cross' of git://github.com/lopsided98/nixpkgs
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 643ba82 Ben Wolsieffer: kernel: fix cross compilation with gcc 7
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy closed pull request #35091: kernel: fix cross compilation after switch to gcc 7 (master...kernel-gcc7-cross) https://git.io/vAl3l
NixOS_GitHub has left #nixos [#nixos]
asuryawanshi has joined #nixos
ma27 has quit [Ping timeout: 265 seconds]
asuryawanshi has quit [Ping timeout: 264 seconds]
fragamus has joined #nixos
<dgpratt>
if I'm looking at a default.nix file and it starts with e.g. "{ pkgs ? <something>..." I'm pretty sure that `<something>` is the default value when the `pkgs` parameter is not supplied; what I am less certain about is when I start nix-shell in the directory containing that default.nix file if there isn't some mechanism that may automatically provide a value for `pkgs` or whether it will have its default value unless
<dgpratt>
e.g. some additional parameter(s) was passed to nix-shell...is my question clear as mud?
contrapumpkin has joined #nixos
cnidario has quit [Remote host closed the connection]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vA8PK
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master c03f1ed Peter Hoeg: zeal: 0.5.0 -> 0.6.0
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bjornfor pushed 1 new commit to gcc-7: https://git.io/vA8P1