<MichaelRaskin>
Does builder think that the store path is valid?
<icetan>
and running recreate locally doesn't fail on the slave machine
<icetan>
running as root though
<MichaelRaskin>
Can Hydra controller do a remote build on the builder directly, without using hydra?
<LnL>
icetan: what platform is that?
<LnL>
the job
<NixOS_GitHub>
[nixpkgs] qknight pushed 2 new commits to release-16.09: https://git.io/vy5x0
<NixOS_GitHub>
nixpkgs/release-16.09 e7039c3 Joachim Schiele: all-packages.nix: removing wmutils-opt (accidentally commited)
<NixOS_GitHub>
nixpkgs/release-16.09 ad55ae3 Joachim Schiele: Merge branch 'release-16.09' of github.com:NixOS/nixpkgs into release-16.09
<icetan>
LnL: do you mean like x86_64-linux?
<c74d>
Is there a standard way to have a package's man-pages or other documentation be globally available, while keeping the rest of the package out of the set of system packages? (E.g., to handle project dependencies with nix-shell rather than installing them globally, but still be able to read their docs outside the nix-shell)
<dhess>
Let's say I have a NixOS host that's running Docker via `virtualisation.docker.enable = true`. If I add an expression to its configuration that does dockerTools.pullImage on a Docker image, will it start that container automatically, or do I need to create a systemd.service for it?
<c74d>
justan0theruser: oh, looks like `nix-build --repair`
myndzi has joined #nixos
<myndzi>
where does nix-env -i save the fact that a package is installed? how does this differ from environment.systemPackages, and which is to be preferred?
<barrucadu>
myndzi: You can see everything you have installed with `nix-env` with `nix-env -q`
<barrucadu>
Things installed with `nix-env -i` are only available to your user, whereas things in `environment.systemPackages` are available to all users
<c74d>
myndzi: I think it saves it in something called a manifest file in the Nix store, which environment.systemPackages also does
<c74d>
`nix-env` is imperative and per-user, `environment.systemPackages` is declarative and system-wide
<NixOS_GitHub>
[nixpkgs] joachifm pushed 2 new commits to release-17.03: https://git.io/vy5pd
<NixOS_GitHub>
nixpkgs/release-17.03 07c2541 Joachim Fasting: grsecurity: 4.9.15-201703150049 -> 4.9.16-201703180820...
<NixOS_GitHub>
nixpkgs/release-17.03 35e6ef8 Joachim Fasting: linux_4_9: 4.9.15 -> 4.9.16...
<icetan>
now my Hydra job won't run, it's queued with no other jobs queued
<icetan>
I'm monitoring disk space to make sure it's not that, but it won't start
<icetan>
ok, now it started
acarrico has quit [(Ping timeout: 260 seconds)]
<ndowens08>
I hate qt apps, they take so long to compile
<ivan>
I hate Qt for dropping the native rendering, making qt5 basically unusable in x2go
<icetan>
OK, now I finally got a build log for ghc-7.4.2 and it just aborts after patching scripts getting stuck there for a while
<joepie91>
is there a way to use NixOps with a pre-existing set of remote Linux systems?
<joepie91>
(if that makes sense)
<neeasade>
I did it! that wrapper thing is very neat.
<spinus>
joepie91: yes, if the linux is NixOS :)
<joepie91>
spinus: it is :)
<joepie91>
spinus: I've just had a glance at the docs and it all seems to assume that you'll let NixOps do the provisioning... where should I look for "how to use NixOps *without* it handling provisioning"?
<joepie91>
:P
<spinus>
so you just set deployment.targetEnv = "none" and nixops is using ssh to deploy the machine
<joepie91>
spinus: is there a specific docs section on this? I imagine I still need to tell it which box is where
ebzzry has joined #nixos
<NixOS_GitHub>
[nixpkgs] neeasade opened pull request #24037: mpvc: init at unstable-2017-03-18 (master...add-mpvc) https://git.io/vy5hy
<spinus>
joepie91: I'm not sure if there is anything in docs (just did quick check)
<spinus>
joepie91: never happened to me, but if you break config badly, potentially yes :)
<joepie91>
heh, alright
<spinus>
remember to copy hardware section from target system to your local nix recipes (exactly just to avoid pitfail)
<joepie91>
right
mexisme2 has quit [(Read error: Connection reset by peer)]
mexisme has joined #nixos
marusich has quit [(Quit: Leaving)]
<joepie91>
spinus: hmmm. do you happen to have an example handy of how to deploy custom expressions/packages to a system? or is this covered somewhere in the manual?
elninja44 has quit [(Remote host closed the connection)]
<spinus>
joepie91: :) I added a manual page about that but was rejected, as using specific nixpkgs for nixos was said to be nixos issue/feature rather than nixops.
<spinus>
there is few ways
<icetan>
I can't find any reason why ghc-7.4.2-binary would cause a problem in hydra
<joepie91>
spinus: any references would be most welcome :P
<icetan>
most times it doesn't even give me a log
<spinus>
simple, before executing nixops, override NIX_PATH=nixpkgs=/path/to/your/nixpkgs
<spinus>
other one, recently someone added overlays, so you can have your own set of packages and then at the top of your nixops network file you do "import <nixpkgs> { overlays=/your/stuff}" (but you need to search for overlay documentation for nixos
<joepie91>
spinus: right, I'd want to use upstream nixpkgs with some custom stuff on top of that
<joepie91>
so overlays seems more appropriate
<joepie91>
hm, overlay docs appear lacking...
<spinus>
there is something similar to overlays, little older mechanism but works as well.
<spinus>
In you machine config you do
<spinus>
nixpkgs.config.overridePackages (or something like that)
<joepie91>
spinus: I'm kind of having trouble seeing how it all fits together - any chance you can create a small example config (even if the things it configures are totally artificial) just so I can understand it from a high-level POV?
<spinus>
I checked and in all my deployments I use fixed NIX_PATH and then I override just single packages when I need, I don't use overlays (which were mered pretty recently, but it looks like that's the way to go) nor nixpkgs.config.override
<joepie91>
orivej: where would the `nixpkgs.overlays = ...` go? the network specification file?
<joepie91>
does each machine have its own nixpkgs.config or is this shared between machines?
<joepie91>
idem for nixpkgs.overlays
dhess2`` has quit [(Ping timeout: 240 seconds)]
<spinus>
each has its own
<joepie91>
alright
<joepie91>
then I think I have an idea where to start - thanks orivej and spinus :)
cgdub has joined #nixos
<Troupal>
re!
<Rizy>
hi, when i am using nix-prefetch-git, it has an option --rev, which is revision. Does this same as commit number in github? like 09c3d04b0e53d65f60569fc01698decff3a65754 ??
<Troupal>
I've try six time to reinstall nixos into VirtualBox, and, evrery times, I cannot connect after rebooting…
<spinus>
Rizy: I think so
<mike543>
I can't get ssh-agent to work on NixOS, I added this line to configuration.nix: programs.ssh.startAgent = true; and rerun nixos-rebuild, but nothing is changing on my system. I can't find any sysstemd services etc.
<Troupal>
Please, help me ! Before I beacame crazy ! XD
ebzzry has quit [(Ping timeout: 240 seconds)]
<spinus>
Troupal: do you use nixops for that?
<neeasade>
LnL: if I remove the with, the package still fails with unexpected } second to the end and I'm not sure why -- can you spot it? http://sprunge.us/UeXL
<Troupal>
spinus: no, just NixOS graphical
<mike543>
how do I even start debugging?
<spinus>
Troupal: so, did you configure root password or ssh keys?
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 2 new commits to master: https://git.io/vy5jp
<NixOS_GitHub>
nixpkgs/master b9173e4 Peter Hoeg: libbluray: 0.9.2 -> 1.0.0
<NixOS_GitHub>
nixpkgs/master f4f29bc Peter Hoeg: handbrake: 0.10.5 -> 1.0.3
<Troupal>
[00:26] <Troupal> I have some problems after an install of NixOS, during the connection. I wrote this: https://gist.github.com/Troupal/0382f8ea1732e6d75fab72d6d7d8cadf in the file /mnt/etc/nixos/configuration.nix , I initiated nix-install , and I put "a" as the unix password. And my problem lies in the fact that I can't connect when kdm starts up, even by putting in login = "troupin" and password = "a" .
<Troupal>
[it's a repost of an IRC #nixos message !
<LnL>
neeasade: that snippet looks fine to me
<spinus>
Troupal: you don't have openssh there
mexisme2 has joined #nixos
<spinus>
ah sorry, are you trying to connect using "virtual screen"?
<Troupal>
no, I'm on my Manjaro's Computer, with VirtualBox 5.1.16
<mike543>
can anyone help?
<Troupal>
I've installed nixos on a virtual machine (after nixos-install and useradd and passwd and reboot)
<spinus>
Troupal: I would try to comment all shell definitions and check if that's not breaking. Other thing, you could force user password by defining it in the config
<spinus>
mike543: I have ssh-agent turned off in config, I start my ssh-agent when user session happens
shanemikel has joined #nixos
mexisme has quit [(Ping timeout: 240 seconds)]
<spinus>
I have systemd unit in ~/.config/systemd/user/
<shanemikel>
can we use url shortener for source code links?
<Troupal>
and when I relaunch nixos (installed), I have kdm asks me a login and a password
<ndowens08>
shanemikel: my opinion, no
<shanemikel>
so a really long line is okay? or use file path concat?
<Troupal>
spinus: How I can force user password?
cgdub has quit [(Ping timeout: 260 seconds)]
<mike543>
I need a global systemd unit for the agent, but there isn't one. shouldn't that be created with the line I pasted above?
<ndowens08>
idk how others feel, but i rather have a long one, versus something happening and that short url points to something else
<shanemikel>
nevermind, stupid question..
<spinus>
Troupal: password = "a"; in the user section, where you specify shell, createHome
<spinus>
mike543: I don't think so, if I remember correctly this setting was modifying user x shell script to start ssh agent at the start of user session (so it was no global systemd unit), but there were some changes to that I think
<mike543>
there is no package for ssh-agent, so I'm guessing it comes with nixos.openssh? where can I find definitions of "packages"?
<icetan>
LnL: do you know how ghc-7.4.2-binary could cause this problem in Hydra?
griff_ has joined #nixos
<NixOS_GitHub>
[nixpkgs] barrucadu opened pull request #24040: Expose the vsftpd no_anon_password flag. (master...vsftpd-no_anon_password) https://git.io/vyden
<mike543>
do you know where can I find definition of what that line does?
<joepie91>
spinus: curiosity question; if in my logical system specification I specify two separate configurations (eg. for two different applications), and in my physical network specification I specify the same hardcoded existing host... will it a) merge the configurations of the two apps into a single system, b) make one configuration override the other (meaning only one of them gets deployed), or c) scream at me and break?
<LnL>
icetan: no I only noticed that bootstrapping for darwin doesn't work properly if you don't have a linux job for it
<spinus>
joepie91: it can do any of them, depends how you specify it
<spinus>
if you use overlays, it should override I think
<icetan>
Looking at hydra.nixos.org the logs are almost identical
<icetan>
Oh that's strange
<joepie91>
spinus: when I say "two separate configurations" I should've said "two separate systems"
<neeasade>
LnL: alright I pushed it up with removal -- github syntax highlights it as well hm https://git.io/vydeu
<spinus>
you have them in separate files?
<neeasade>
oh it's there.
<joepie91>
spinus: assume that the configuration is exactly as described in those examples, just in the physical specification (example 3.4) I specify targetEnv none instead of targetEnv virtualbox
<joepie91>
(and add a single host)
<neeasade>
LnL: disreguard
Supersonic112 has quit [(Disconnected by services)]
derjohn_mob has quit [(Ping timeout: 258 seconds)]
Supersonic112_ has joined #nixos
<spinus>
joepie91: but example is explicitelly showing 3 hosts, how do you "merge" backend and proxy on one box?
<NixOS_GitHub>
[nixpkgs] neeasade opened pull request #24041: colort: init at unstable-2017-03-13 (master...add-colort) https://git.io/vyder
Supersonic112_ is now known as Supersonic112
<joepie91>
spinus: right, I don't know, my question is just how NixOps would deal with this :P
<shanemikel>
Oh It looks like I can't do this `http://acme.com + /path`
<joepie91>
by default
<joepie91>
it's really just a curiosity question
<shanemikel>
is a url a different type than a regular path?
<icetan>
Can I force Hydra to accept the aborted build any way? It actually succeeded once so it's really strange
<icetan>
But just trying to rebuild ghc
<spinus>
joepie91: but it really depends how you specify those options
<joepie91>
shanemikel: fairly sure that all URLs should be strings
<LnL>
neeasade: you also have some tabs in the file, that's why the indentation looks a little strange
<joepie91>
shanemikel: (and that "specifying them as not a string" is more a side-effect of how path syntax detection works, than an intentional feature)
<joepie91>
oh, that's actually detected as a string?
<joepie91>
odd
<joepie91>
[03:07] <spinus> joepie91: but it really depends how you specify those options
<shanemikel>
thanks
<joepie91>
can you elaborate on what it depends on?
<spinus>
joepie91: how you write expression.
<spinus>
machine = config1 // config2; # configs are merged (if they are sets)
<spinus>
but it's probably better to have just one config and have all services defined in there
<neeasade>
LnL: thanks for highlighting, I'm doing all the force pushes today :P
<joepie91>
spinus: right -- I feel like I'm not quite explaining myself well, which is probably because it's late here... but it doesn't matter as it's not an important question so nevermind :)
<icetan>
It's so irritating, I've finally managed to build our huge Maven project with Nix to try to convince my company to invest in reproducible builds but I'm stuck on this weird Hydra problem
fresheyeball has quit [(Quit: Konversation terminated!)]
mizu_no_oto has joined #nixos
dhess2`` has quit [(Ping timeout: 260 seconds)]
<c74d>
Am I right in thinking that lib.lowPrio only affects nix-env and doesn't affect whether files are considered to collide when a system generation is being built?
<c74d>
and if so, how (if at all) might I change a package's "priority" for collisions?
<gchristensen>
you have to specify that seemingly meaningless resource with the name "ssh-key"
<joepie91>
okay, but why...? :P
<gchristensen>
resources.sshKeyPairs automatically generates ssh key-pairs, which the D.O. support hard-coded a reference to the ssh-key sshKeyPair to send to the D.O. API
__monty__ has joined #nixos
<joepie91>
gchristensen: wait... so for *any* thing I specify in resources.sshKeyPairs.*, a new key is automatically created if one doesn't exist yet?
<joepie91>
in a NixOps spec, what is the `config` argument?
<__monty__>
I prefer Syslinux to grub, however since grub is mentioned in the default configuration.nix I'm assuming I'll lose some convenience? I'm following the arch wiki page on Syslinux and I'm stuck on the configuration. Will I need chainloading (/boot is separate ext2 partition, root is in a btrfs lvm partition)? Where do I point to for the kernel imag
<__monty__>
e(?), ../vmlinuz-linux on arch, and the initramfs image?
nschoe has joined #nixos
ixxie has quit [(Ping timeout: 260 seconds)]
<nschoe>
Hi everyone :) I have a small question: I'm trying to list the system generations (and delete some of them). I have sucessfully listed and deletes user generations with `nix-env --list-generations | --delete-generations`, but how cna I do the same for the system generations? The one which appear on the boot menu?
JagaJaga_ has joined #nixos
<nschoe>
Oh okay I must use the -p option: `nix-env -p /nix/var/nix/profiles/system --list-generations`, right?
JagaJaga has quit [(Ping timeout: 258 seconds)]
dhess2`` has quit [(Ping timeout: 260 seconds)]
<Yaniel>
or you can use the tool that works directly on the system profile
<nschoe>
Erf so I managed to do it, but it did not manage my problem: my /boot partition is full (EFI) and the majority of the space is taken up my nixos. How can I clean these?
<nschoe>
Yaniel, oh? such as?
<Yaniel>
nix-store and nix-collect-garbage AFAIK
<Yaniel>
the latter specifically with --delete-old or --delete-older-than
<MichaelRaskin>
I think the /boot is only cleaned on the next rebuild after removing system generations
<nschoe>
Yaniel, well I had used nix-collect-garbage, but it only removes the generations of the current proile (the user profile). So i had firs tto delete system generations.
<nschoe>
MichaelRaskin, ah? I'll try a rebuild then, to see.
<nschoe>
MichaelRaskin, no it doesn't change anything :/ `nixos-rebuild build` immediately returns because I had already rebuilt it before.
<MichaelRaskin>
you need to do the boot or switch
<DMZ__>
hello, how can we overcome this "mkdir: cannot create directory '/homeless-shelter': Permission denied" ?
<nschoe>
Hum ok
<MichaelRaskin>
build doesn't touch anything outside the store
<nschoe>
AH, good to know.
<MichaelRaskin>
At least it is not supposed to!
<MichaelRaskin>
DMZ__: The build tries to mkdir -p $HOME/something
<nschoe>
MichaelRaskin, awesome! It worked with switch
<MichaelRaskin>
Ideally, tell it to stop doing that
<nschoe>
Thanks MichaelRaskin & Yaniel :)
<MichaelRaskin>
Alternatively create a fake homedir inside the build directory and change HOME
<joepie91>
gchristensen: ah, ok. any idea who does have an idea?
<joepie91>
:)
<FRidh>
gchristensen: ok, then we'll wait till copumpkin shows up
<DMZ__>
Mic92: i have another problem :D nixpkgs/pkgs/desktops/gnome-3/extensions/default.nix’: No such file or directory when i do "nix-build -A dash-to-panel"
<DMZ__>
nix-build -K -E "with import <nixpkgs> {}; callPackage ./dash-to-panel.nix {}" this works ofc
<avn>
Mic92: fyi, if you touch go compilers by your iana-etc changes, can you also add `preferLocalBuild = true;` to go-bootstrap?
<gchristensen>
joepie91: I don't, sorry
spinus has quit [(Remote host closed the connection)]
<joepie91>
alright
<Mic92>
avn: you mean the runCommand expression?
<__monty__>
Any idea for the best place to look for help on nixos+Syslinux?
<avn>
Mic92: yep, resulting derivation still refer go-1.4, and they downloading both (so +~40mb download)
<rmrfroot>
suppose i have have the attribute set `interfaces = { enp1s0 = "blabla"; };` and the variable `lan = "enp1s0";`, how can I use the value of `lan` as the attribute key?
<__monty__>
To configuration.nix or to the modules? I'm still lost when it comes to nixos install/configuration.
<__monty__>
Also, what's not entirely clear to me is wether the uses booth Syslinux and grub or not?
<Mic92>
__monty__: there is not support to boot from syslinux except on the iso
<Mic92>
in nixos
<Mic92>
which means there is no module to configure nixos to boot from syslinux
<Mic92>
there is grub/systemd-boot though
<clever>
rmrfroot: { ${lan} = "blabla"; }
<__monty__>
Yeah I understand that but doing it semi-manually like what's done for the iso would work, right?
<rmrfroot>
clever: thanks! double quote not needed, huh?
<clever>
rmrfroot: yep, just works bare
<rmrfroot>
clever: awesome!
<Mic92>
__monty__: yes,
<Mic92>
__monty__: but you cannot add anything to your configuration.nix for that
<Mic92>
__monty__: boot.loader.grub.enable = false would disable grub
<__monty__>
So I need to define a Syslinux module first?
hotfuzz has joined #nixos
<ixxie>
hmm anybody got experience configuring R / rWrapper in NixOS configs?
<ixxie>
I could use an illustrative example
<Mic92>
__monty__: if you disable grub you would not get any bootloader updates. So if you want boot entries in the same style, you need a new module for syslinux
hotfuzz_ has quit [(Ping timeout: 260 seconds)]
<__monty__>
Does the iso get around this because it never needs to update the boot entries?
<NixOS_GitHub>
nixpkgs/master a4e7e0d Frederik Rietdijk: Merge pull request #24025 from ndowens/ahoviewer...
dhess2`` has joined #nixos
<countoren>
Hi, im starting with nixos (just installed 16.09 in vb) and i want to install the latest elm using nix-env -qa elm i noticed that the latest is 0.17.1
<joepie91>
uh... I have gotten myself into a situation where I can't destroy a droplet resource but I also can't create it because it doesn't see my DO API token...
<joepie91>
is there any way to tell NixOps "just forget that this resource exists"?
<tsmeets>
Hello, How would I go about putting nixos on my raspberry pi 3?
<tsmeets>
Can I download an aarch64 iso somewhere?
<Dezgeg>
then start from the configuration.nix snippet in https://nixos.org/wiki/NixOS_on_ARM except don't use linuxPackages_rpi and the binaryCaches isn't needed since aarch64 is on the official hydra
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to release-17.03: https://git.io/vyd89
<NixOS_GitHub>
nixpkgs/release-17.03 acb7958 Ben Darwin: mark old seg3d as broken for now since it no longer builds with stock gcc version and flags...
<joepie91>
..........
<joepie91>
there's a bug in the NixOps documentation
k2s has joined #nixos
<NixOS_GitHub>
[nixpkgs] lpenz opened pull request #24057: gnu global: set path of default gtags.conf in wrapper (master...gtagsconf) https://git.io/vyd8N
<joepie91>
"We're checking the environment for DIGITAL_OCEAN_API_TOKEN first and if that is not set we try this auth token."
<joepie91>
this is wrong, it's DIGITAL_OCEAN_AUTH_TOKEN :|
Rizy has joined #nixos
<joepie91>
there we go
<joepie91>
that's one half of the issue found...
mudri has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixops] joepie91 opened pull request #626: Fix incorrect environment variable name in DigitalOcean documentation (master...patch-1) https://git.io/vyd4L
<ikwildrpepper>
puffnfresh: ping
<ikwildrpepper>
joepie91: nixops delete --force ?
<ikwildrpepper>
that will remove the network completely, without touching the resources
justbeingglad1 has left #nixos []
<joepie91>
ikwildrpepper: hm. that's not in the --help?
<countoren>
running: nix-env -qa elm Returns: killed
<countoren>
any help?
tsmeets has quit [(Ping timeout: 260 seconds)]
<contrapumpkin>
FRidh: sure ,but why?
<NickHu>
countoren: You might be running out of memory by the sounds of it
<contrapumpkin>
FRidh: actually I have no power to pause
<FRidh>
contrapumpkin: the macs can't keep up
<contrapumpkin>
ikwildrpepper: did you see my email?
<contrapumpkin>
FRidh: I'm not sure how to pause :/
<NickHu>
countoren: Anyway, what you want is nix-env -iA nixos-unstable.elmPackages.elm
<ikwildrpepper>
contrapumpkin: haven't been at the office since thursday, so can only look at mac on monday. I know niksnut was looking at some issues with hydra.nixos.org before
griff_ has joined #nixos
<contrapumpkin>
ah, the main things I've been seeing are no logs on any builds
<contrapumpkin>
and mac4 being dead
<contrapumpkin>
oh and the 500s I guess
<FRidh>
most machines don't show any logs anymore is my impression
<countoren>
NickHu:getting back: error: attribute ‘nixos-unstable’ in selection path ‘nixos-unstable.elmPackages.elm’ not found
<NickHu>
yeah so you didn't add the channel
<countoren>
mm
<ikwildrpepper>
contrapumpkin: mac4 being dead, I can look into on monday, the other issues niksnut should look into those, as they are likely related to some changes he made
<contrapumpkin>
ikwildrpepper: anyway thanks. Do you think you can pause the job or does hydra not do that?
<ndowens08>
Hey guys
<countoren>
getting same error on nix-env -iA nixos-unstable.elmPackages.elm
<toppler>
Can anyone help me? I'd like to define a new service for use in my NixOS configuration. What's the basic way to go about it?
<ndowens08>
toppler: though creating modules
<ndowens08>
what kind of service
<toppler>
Cheers, I'll have a read of the wiki.
<toppler>
Actually, all I'm doing is fixing what appears to be a bug in ircdHybrid. Will submit an issue later.
rmrfroot has quit [(Remote host closed the connection)]
vcunat has joined #nixos
* ndowens08
feels like dancing; Now I am officially a member of the community, thanks to ikwildrpepper :)
<countoren>
NickHu: any idea?
<MichaelRaskin>
ndowens08: now you are officially a _trusted_ member of community
<joepie91>
teh[m]: around by any chance? I have some questions about your DO backend for NixOps, because I've found what seems like a really obvious bug, but I want to confirm that I'm not just missing something
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to master: https://git.io/vydBU
<NixOS_GitHub>
nixpkgs/master 9c7b771 Robin Gloster: rapidsvn: fix build
<ikwildrpepper>
contrapumpkin: I am not ure how, but you can just cancel the builds, I imagine?
chrishill has joined #nixos
arianvp2_ has joined #nixos
<arianvp2_>
How do I add a custom xkb keymapping to nixos?
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to release-17.03: https://git.io/vydBL
<NixOS_GitHub>
nixpkgs/release-17.03 39332aa Robin Gloster: rapidsvn: fix build...
<arianvp2_>
the only option I found was xserver.xkbDir
<joepie91>
teh[m]: specifically, in `<DigitalOceanState>.get_auth_token()` it attempts to access `self.auth_token`, but the `auth_token` property only seems defined on the `DigitalOceanDefinition`, not the `DigitalOceanState`, and I don't see any place where the property is copied over... similarly, there exists a `<DigitalOceanState>.region` but this is never set (instead the `defn.region` is used directly, passed into the DO library, and not stored anywhere)
<arianvp2_>
which doesn't tell me much, "path used for the xkbdir parameter"
<arianvp2_>
what is this parameter? how does it help me? how do I add new xkb models?
<arianvp2_>
as they're not in /usr/share as on normal linux distro's but in the nix store I assume
<ndowens08>
seems there is a issue with nox-review again
<ndowens08>
arianvp2_: yea everything tends to be in /nix/store
<NixOS_GitHub>
[nixpkgs] ndowens pushed 2 new commits to master: https://git.io/vydBn
<NixOS_GitHub>
[nixpkgs] ndowens closed pull request #24030: fontmatrix: Changed URL & homepage; they no longer exist (master...fontmatrix) https://git.io/vy5AG
<arianvp2_>
in nixos this seems to be some /nix/store path
<joepie91>
hold on... I think that what this DO backend is trying to do can't even work? to expose the key to the `destroy` function it'd have to store it in a state property, which would get persisted in the state file... which means that if you change your auth key inbetween deployment and un-deployment, the un-deployment will fail...
<joepie91>
to the issue tracker!
<arianvp2_>
you can override it wit "services.xserver.xkbDir" but this is too extreme. I don't want to fully change to a different xkbd path, but I want to add a new model to the existing xkb's
<arianvp2_>
=(
johbo has joined #nixos
<chrishill>
Hi. Im curious about how to package applications written in the Rust programming language. Anyone know about that?
griff_ has quit [(Quit: griff_)]
bkchr has joined #nixos
<Mic92>
chrishill: there are some example in nixpkgs. ripgrep for instance
<Mic92>
chrishill: I also add recently some documentation in nixpkgs
<bkchr>
Hi, I'm at installing Nixos, I wanted to ask, can I specify the channel in the configuration.nix?
<ndowens08>
Mic92: createrepo_c is causing issues with nox-review, something about patches
<ndowens08>
in which i see nothing wrong with it
<arianvp2_>
I think I need to override the xorg nix package for this ...
<arianvp2_>
T_T
<Mic92>
ndowens08: I forgot to add them
<ndowens08>
ah :)
<NixOS_GitHub>
[nixops] domenkozar closed pull request #626: Fix incorrect environment variable name in DigitalOcean documentation (master...patch-1) https://git.io/vyd4L
<joepie91>
gchristensen: I actually found the code to be refreshingly easy to understand
<joepie91>
gchristensen: it has the usual issue of hunting through layers of inheritance
<joepie91>
but besides that, it's quite okay
<joepie91>
I've scratched my head *much* more over some of the stuff in nixpkgs... :)
<gchristensen>
:)
<joepie91>
anyhow
<joepie91>
env var it is for now...
gbbrt has joined #nixos
<clever>
arianvp2_: some programs arent patched to read from a configurable location, so those read from /etc instead
<gbbrt>
hi! I'm reading this blogpost http://anderspapitto.com/posts/2015-02-28-deb-installation-nixos.html about buildFHSUserEnv, but somehow the contents of the `hello` derivation don't end up in the fhs environment. Its from 2015 so I suspect the workflow has changed, but couldn't find any hints (even in nixpkgs) how to do it
<clever>
gbbrt: 99% of the time, buildFHSUserEnv is the wrong way to package something
<clever>
gbbrt: its usually better to use patchelf first
<clever>
gbbrt: and also setting builder in a derivation disables a lot of support code that is normaly active
spacekitteh has joined #nixos
arianvp2_ has quit [(Ping timeout: 260 seconds)]
<gbbrt>
clever: hm, I see.. I'm trying to get some multi-lib stuff for this audio software working (bitwig) and couldn't get patchelf on the 32bit parts to work for me
<clever>
gbbrt: you need to use the 32bit ld.so and the 32bit libs when patching a 32bit program
<clever>
gbbrt: can you throw what you already have on gist.github.com?
<gbbrt>
clever: so I need to use callPackage from stdenv_32bit?
<clever>
gbbrt: callPackage_i686 also works
<joepie91>
argh. is there any work scheduled on improving the stacktrace output of Nix?
<joepie91>
formatting mostly
<gbbrt>
clever: the other part why I thought that it could be useful to use FHS is that I could use that for 32bit shared libraries as well (VSTs) without having to patch each and everyone of them
<clever>
gbbrt: if they are dyn libs loaded by bitwig, then it should also use the rpath of bitwig
<clever>
gbbrt: but if the VST starts to execute its own stuff, yeah, an FHS might help more
<ndowens08>
i forget, how do i specify to use my local nixpkg repo when doing nixos-rebuild build-vm, need to test awesome update before committing
<clever>
gbbrt: and with the strict licensing stuff of VST, i can see how it could be hard to fix
<clever>
gbbrt: here is how steam sets up its FHS env
<ndowens08>
clever: it says no url specified when there is
<clever>
ndowens08: can you gist the nix expression?
<gbbrt>
clever: any idea how I can amend that basic `hello` example from that blog to work? I saw that the `hello` binary *does* and up in the `hello-usr-target` output
<clever>
jophish: all setuid programs must be configured within security.wrappers from a nixos module, which will make a setuid wrapper within /run/wrappers/bin at runtime
<clever>
joepie91: hmmm, yeah, i can see the issue
lezed1 has quit [(Remote host closed the connection)]
<clever>
joepie91: it looks like every callFunction in the entire stack will catch the exception, print a line of nix stack, then re-throw the exception
<joepie91>
clever: even if just "you're totally wrong and you don't understand how NixOps works" :)
<clever>
joepie91: not sure what the right fix would be, sounds like there needs to be a global variable or some state somewhere, to remember which token to use at destroy
<clever>
joepie91: but its not clear how much time there can be between create and destroy
<NixOS_GitHub>
nixpkgs/master 575cf2e Joachim F: Merge pull request #24035 from ndowens/leocad...
pSub has quit [(Remote host closed the connection)]
<joepie91>
clever: well, as I understand it, that time could be months
ryantrinkle has joined #nixos
pSub has joined #nixos
pSub has quit [(Changing host)]
pSub has joined #nixos
<clever>
joepie91: so i would either file a bug with digital ocean for needing a token from months ago, or store the token create used into the nixops database, so it knows how to destroy an instance
<ndowens08>
clever: it didnt give an error on fsck but didnt help
<ndowens08>
ah nvm, dangling commit
<joepie91>
clever: huh? the token is basically an API key
<joepie91>
clever: so the problem is when eg. your token (key) leaks and you invalidate it and use a new one
<ryantrinkle>
has anyone run into a situation where xscreensaver can't unlock, even with the right password?
<joepie91>
clever: but all the existing VMs still remember the old, now invalid token (key)
<ryantrinkle>
(but a VTY unlocks just fine)
<joepie91>
clever: they're not tokens in the ephemeral sense
<clever>
joepie91: but usualy when tokens get rotated, the old token should become invalid
<joepie91>
yes
<joepie91>
but that means that nixops will just scream at you about the token being invalid
<joepie91>
with no way for you rectify that
<joepie91>
:)
<joepie91>
(other than the env var hack)
<ryantrinkle>
i saw that perhaps that can be caused by /etc/shadow having group 'root' instead of 'shadow'
<clever>
joepie91: is it being remembered within a file nixops baked into the VM?
cfricke has joined #nixos
<ryantrinkle>
and mine does indeed have group 'root'
<joepie91>
clever: as I understand it, it's stored in the state file on a per-machine basis
<clever>
joepie91: or is it that you need to inform nixops about the new token?
<ryantrinkle>
but i'm not sure what may have changed
<joepie91>
clever: yes, you need to inform nixops about the new token, but there is no facility for that now
<joepie91>
as far as I can tell
<clever>
joepie91: ah, sounds like it needs to be a single global token for all of nixops, and ignore the per-machine ones
<joepie91>
clever: yeah, basically
<joepie91>
clever: or at least an indirect reference
<joepie91>
clever: eg. I might have two DO tokens (one for myself, one for my employer, whatever), and the per-machine state stores the identifier of the token, not the token itself
<joepie91>
clever: and then I can use multiple tokens for different networks
<joepie91>
but still replace a token for every machine if it ever has to be invalidated
Fare has joined #nixos
<joepie91>
for every affected machine *
<joepie91>
that's my idea anyway, but I don't know if it makes sense
<ndowens08>
looked how to remove them but it wont do it b/c bad head
<joepie91>
because I don't have a complete knowledge of how NixOps works :P
<ndowens08>
though maybe deleting .git will work, from .nox dir
<ndowens08>
yup seems so :)
<clever>
ndowens08: i never really bothered with nox review
<ndowens08>
clever: ah, i do to make sure it isnt gonna effect deps of the package and see if it is a sorta-mass rebuild or is a mass rebuild
<clever>
ndowens08: ah, i avoid editing those kinds of files, and i just know all of the mass-rebuild like things off the top of my head
<clever>
ndowens08: i get a bit OCD about knowing how things work, and have read a large chunk of the source behind the stdenv
<ndowens08>
clever: i dont :); I dont want to push bad things, so going through double checking before merging mine
Rizy has quit [(Quit: Rizy)]
<ndowens08>
Found one that is a mass-rebuild, so not comforting to merge myself
<arianvp2_>
there is this one package called "xkeyboardlayout" that contains _all_ nixos keymaps and a lot of packages depend on it
<arianvp2_>
only thing it does it populare <nix-store-hash>/etc/X11/xkb with layout files
<arianvp2_>
and there is really no way to add new keymaps to it, which seems to be a shame
<arianvp2_>
there is an option services.xserver.xkbDir that lets you point to your own "xkb" directory, but that means you lose all the predefined keymaps
<joepie91>
arianvp2_: seems like the solution might be to submit a patch to the package/service/whatever-it-is to allow adding custom items?
<arianvp2_>
yes that is my conclusion as well
desktop has joined #nixos
<arianvp2_>
I'll see what I can come up with :)
<MichaelRaskin>
By the way, passing xkbDir a union of default and your added layouts is an option
<MichaelRaskin>
arianvp2_: just use buildEnv {name="…"; paths = [xkeyboard_config additions]; }
<MichaelRaskin>
That should help about «only one directory»
<ndowens08>
that was one of the packages that travis said no CC was found, so had to add gcc as a darwin dep but dunno if it changed anything or not; may have to disable darwin support if it doesnt work
<erictapen>
hello everyone, I seem to experience a ux problem. Want to add a patch via fetchpatch and I got it's sha256 from nix-prefetch-url. When I rebuild my system with the patch, the hashes don't match...
<erictapen>
Is there anything like nix-prefetch-patch?
<erictapen>
sha256 seems to be default, already tried that
<joepie91>
hm, odd.
<joepie91>
not sure then
<erictapen>
output path ‘/nix/store/sm0mrwvs0437x26a856fvj6q5jclq453-abook_vcard_import.patch’ has sha256 hash ‘1vxpj6n38dhjli2kq07ds99wilibjmy17h034a9ci7pqzm289pwl’ when ‘1hg5gkp361szl01xhi8lqlv9gwz5n0id685a6xwlvmlarxs48d22’ was expected
<joepie91>
erictapen: personally I always just shove in a random hash, wait for it to error out, and copy the hash it claims to have
<joepie91>
:p
<erictapen>
:D
<clever>
ndowens08: i have ssh to a darwin box
<erictapen>
Ok thats also a way
<ndowens08>
ok this is weird guys, I was reviewing someone's PR and it said guile was missing, and now I am looking at another and it gives guile is missings; sounds as a dep missed a dependency :)
<erictapen>
thanks
<clever>
joepie91: oh, correction to the previous stack trace stuff
<ndowens08>
autogen seems to be at fault
<arianvp2_>
MichaelRaskin: so `additions` would be my package that has the custom xkb files correct?
<clever>
joepie91: i just noticed, its calling .addPrefix on the error object that it keeps throwing to itself
<MichaelRaskin>
Yes
<ndowens08>
looked at autogen and it has guile listed but maybe missing wrapper now?
<clever>
joepie91: currently, it will prefix the prefix_ variable with the line of stack, you could then either parse the whole thing, or store the data in a more structured manner
<ndowens08>
weird, doesnt show a recent update to autogen and now it is failing
cfricke has quit [(Quit: WeeChat 1.7)]
<joepie91>
clever: heh, this is mostly Chinese to me :)
mudri has joined #nixos
<kthnnlg>
Hi, Has anyone experienced their xwin freezing on a blank screen on a regular basis? I'm using the intel driver with hdmi output. When it occurs, xwin starts only after reboot, so it could be a kernel issue.
<clever>
joepie91: the most complex part i can see is where Error gets defined, on line 124 of types.hh
<clever>
joepie91: its usingthe macro on 117 to programaticaly generate classes
<arianvp2_>
Is there a way to make a derivation with multiple files using pkgs.writeTextFile ?
<clever>
arianvp2_: probably simpler to use runCommand
<arianvp2_>
what does that do / where do I find what that does?
<joepie91>
clever: right, but my understanding of C++ is *extremely* limited, so I'm missing the high-level understanding of the code's structure
erictapen has joined #nixos
<joepie91>
totally unrelated; what would be the cleanest approach to specifying a custom nixpkgs branch / channel in my NixOps configurations, ideally without having to manually `git pull` a repo constantly?
<joepie91>
err, to clarify: the custom branch only being used for specific packages
<joepie91>
not for everything
<joepie91>
(duplication of libraries is not important)
<clever>
joepie91: one min
Fare has quit [(Ping timeout: 268 seconds)]
eacameron has quit [(Remote host closed the connection)]
<copumpkin>
joepie91, clever: unfortunately that won't really affect your modules, only your packages
<copumpkin>
I think
<clever>
joepie91: only issue with that form, is that it will checkout the latest 16.09 every time you eval, and may update things without you noticing
<copumpkin>
unless nixops is magicking things
<joepie91>
copumpkin: most modules allow you to specify a package, don't they?
<pikajude>
_deepfire: it's building the shell environment
<pikajude>
not the package
<_deepfire>
joachifm, yes, thank you for looking!
<pikajude>
the shell environment doesn't have $src it just has some shell setup stuff and build inputs
<_deepfire>
pikajude, why do you say so?
<_deepfire>
oh, because .env in haskell derivations works like this
<pikajude>
because it says "builder for ‘/nix/store/9yw3grpj5d49xrsr5c5nc9zsdj34bdc4-interactive-reflex-glfw-0.1.0.0-environment.drv’ failed with exit code 1"
<_deepfire>
the plain derivations have no difference between shell and non-shell variants, AIUI
<pikajude>
well they wouldn't
<pikajude>
that wouldn't make sense
<_deepfire>
but in the haskell package world there are two kinds
<pikajude>
a lot of generated files end in `if lib.inNixShell then drv.env else drv`
Guest48360 is now known as katyucha
<_deepfire>
yeah
<_deepfire>
thank you a lot!
tsmeets has joined #nixos
k2s has quit [(Ping timeout: 260 seconds)]
<_deepfire>
I keep rediscovering this detail over and over, sadly..
<gchristensen>
joepie91: other than wanting the features that file provides inside a nix-shell?
barrucadu has quit [(Quit: Lost terminal)]
<joepie91>
gchristensen: that would be the answer then :)
adamCS has quit [(Ping timeout: 240 seconds)]
barrucadu has joined #nixos
barrucadu has quit [(Changing host)]
barrucadu has joined #nixos
<joepie91>
gchristensen: I think you just accidentally solved another problem I had :P
<gchristensen>
oh?
bkchr has quit [(Quit: Konversation terminated!)]
<joepie91>
gchristensen: I want a way to carry around my nixops setup (state file + deployment) with me on multiple systems (desktop + laptop) while using Git for managing the actual network config
<joepie91>
gchristensen: this kind of shell config + hooks seems like the right approach for that
adamCS has joined #nixos
mudri has joined #nixos
<gchristensen>
the state file is tricky but doable
<gchristensen>
you need to encrypt it on dump and decrypt it on load
<joepie91>
gchristensen: I don't intend to track it in the repo
<joepie91>
rather out of band
<gchristensen>
cool
<joepie91>
I mean, I *could* track it in the repo with encryption
<gchristensen>
same recommendation though :)
<joepie91>
but this seems messy
<gchristensen>
it includes a root ssh key for all your boxes
<joepie91>
yeah, I know
<joepie91>
I'd probably just end up using something a la syncthing or even just a local-network rsync
<joepie91>
so it'd never leave my network
<gchristensen>
aye
<joepie91>
and if rsync, then likely over SSH :)
<joepie91>
so the problem would solve itself
<joepie91>
the biggest issue is preventing conflicts really
tsmeets has quit [(Ping timeout: 240 seconds)]
dhess2`` has joined #nixos
k2s has joined #nixos
vcunat has quit [(Read error: Connection reset by peer)]
eacameron has quit [(Remote host closed the connection)]
<ndowens08>
LnL: on highlight package, do i need to remove optional gcc?
<LnL>
I just pushed my changes
<ndowens08>
LnL: ok to merge?
Fare has joined #nixos
slack1256 has joined #nixos
dhess2`` has quit [(Ping timeout: 261 seconds)]
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vydMs
<NixOS_GitHub>
nixpkgs/master 6ea3eef Dan Peebles: gcc5: fix for libc++ 3.8 and above...
ilyaigpetrov has quit [(Quit: Connection closed for inactivity)]
chrishill has quit [(Quit: Leaving)]
odi_ has quit [(Ping timeout: 240 seconds)]
<joepie91>
hmmm. if I want to set up deployment for my Node.js application, and I want my setup to pull whatever is the latest version in the repo and generate its dependencies using node2nix prior to building it... how would I best approach this? (context: NixOps)
<joepie91>
the hash of the application changing on every new version is fine in this case
stepcut has joined #nixos
tsmeets has joined #nixos
<tsmeets>
how do i merge multiple configurations just like using 'imports'?
<tsmeets>
the function 'lib.recursiveUpdate' is almost what I need, but it doesnt merge lists. (only attributes)
<NixOS_GitHub>
[nixpkgs] MasseGuillaume opened pull request #24077: cre2: c wrapper for google's re2 regex library (master...cre2) https://git.io/vydSq
dhess2`` has joined #nixos
eacameron has joined #nixos
ambro718 has joined #nixos
ryanartecona has joined #nixos
* contrapumpkin
unsticks joepie91
<contrapumpkin>
you're welcome
<pikajude>
put some wd-40 on that
<joepie91>
haha
eacameron has quit [(Ping timeout: 240 seconds)]
<joepie91>
contrapumpkin: if you have a pointer for me on my issue, that'd really unstick me :P
<contrapumpkin>
what issue is that?
periklis has quit [(Ping timeout: 258 seconds)]
<joepie91>
[21:59] <joepie91> hmmm. if I want to set up deployment for my Node.js application, and I want my setup to pull whatever is the latest version in the repo and generate its dependencies using node2nix prior to building it... how would I best approach this? (context: NixOps)
<joepie91>
[22:00] <joepie91> the hash of the application changing on every new version is fine in this case
<joepie91>
constraints: no .nix files in project repo, no extra manual generation step before deployment
<contrapumpkin>
joepie91: you'd probably want something like import-from-derivation, which is frowned upon in nixpkgs but generally works. You'd do something like
<contrapumpkin>
let outer = import <nixpkgs> {}; in import (outer.runCommand "foo" { buildInputs = [ outer.node2nix ]; } ''node2nix blahblah'')
<joepie91>
contrapumpkin: hmm. how would I integrate that with something like a fetchgit?
<contrapumpkin>
fetchgit against what?
<contrapumpkin>
oh, you want a latest fetch?
<joepie91>
against the repo that contains the software I wish to deploy, as well as the package.json from which the node2nix stuff is generated :)
<joepie91>
yeah
<joepie91>
my intention is to, when I do a nixops deploy, just "grab whatever the latest version is in the repo and build if needed, using node2nix and Nix"
<joepie91>
the repo is controlled by me so versioning isn't a big issue
MP2E has quit [(Ping timeout: 260 seconds)]
<contrapumpkin>
there's a builtins.fetchgit in recent nix versions but I'm not sure it's in a stable release
<joepie91>
(and I'll probably generalize the implementation over time for more strict constraints)
shanemikel has joined #nixos
<contrapumpkin>
otherwise your best bet is builtins.fetchurl and builtins.fetchTarball
dhess2`` has quit [(Ping timeout: 256 seconds)]
<contrapumpkin>
which are in current releases
<contrapumpkin>
you'd need your git server to serve up tarballs
<pikajude>
how do I add vim packages to knownPlugins
<pikajude>
it used to be configure.vam.knownPlugins = {...}
<contrapumpkin>
we don't have a good generic way to fetch latest things outside of those
<pikajude>
but that doesn't seem to work anymore
<joepie91>
contrapumpkin: assuming I have access to nixpkgs, would it be possible to use the fetchgit from there (from build-support) and grab that prior to node2nix'ing?
<contrapumpkin>
not really, since it doesn't have a notion of fetching latest
<joepie91>
contrapumpkin: could you fork/modify that Gist with how you would implement it there? a simple real-world example will probably be much faster to understand than going back-and-forth about stuff I barely understand for 30 minutes :D
<joepie91>
to get an understanding of how everything ties together at a high level
<joepie91>
s/at/on/
alx741 has quit [(Quit: alx741)]
<FRidh>
11287 jobs in the Hydra queue of which 11135 are darwin. If only we could have separate channels
<joepie91>
contrapumpkin: oh, and just assume that it's a tarball at some random URL
<joepie91>
once I have a high-level understanding I probably won't have any issues switching out and reorganizing specific bits anyway :P
<contrapumpkin>
okay
<contrapumpkin>
sure
mkoenig has joined #nixos
<_deepfire>
was "the qt5 breakage" on master fixed?
<joepie91>
with `package` being an attribute in the node2nix-generated default.nix
<contrapumpkin>
ah yes
<contrapumpkin>
updated again
<contrapumpkin>
basically, the inner mkDerivation I wrote in there should produce a default.nix
<contrapumpkin>
then you import that, which gives you the contents of that default.nix
<contrapumpkin>
then you ask for the attribute in it and use it as you normally would
shanemikel has joined #nixos
<joepie91>
contrapumpkin: maybe I'm misunderstanding - I thought mkDerivation only produced a *definition* of a derivation that would then be used by nix-build to produce an actual build, but this seems to suggest that it's built immediately?
<contrapumpkin>
if you import it, it'll force it to get built
<contrapumpkin>
which is why it's kind of weird
<joepie91>
aha!
<joepie91>
yeah, I think that was the missing piece in the puzzle for me :)
<contrapumpkin>
I'm not 100% sure if it'll work with the weird nixops remote build shenanigans but import-from-derivation generally works
<joepie91>
contrapumpkin: thanks, I'm going to have a try with this and see if I can get it to work
<contrapumpkin>
and is mostly supported, if discouraged
<joepie91>
yeah, I can see why
<joepie91>
it's definitely not something I'd use in another scenario :P
<contrapumpkin>
what I'd really like is for nix to get a more complete notion of staged evaluation
<contrapumpkin>
so it has a first-class way of interleaving builds with evaluation
<joepie91>
contrapumpkin: in the sense that "produce a Nix expression and then do the next step" would be a fully supported thing to do?
<contrapumpkin>
but right now, IFD just means it does this weird build inside the evaluation
<joepie91>
like here
<contrapumpkin>
yeah
mkoenig has quit [(Ping timeout: 240 seconds)]
<joepie91>
yeah, that'd definitely be a nice addition
<joepie91>
would make packaging of eg. Node stuff a lot easier :)
<ndowens08>
How do I push changes to a PR that isnt mine?
<ndowens08>
through git?
<simpson>
ndowens08: You need to be able to push to the underlying branch.
FRidh has quit [(Remote host closed the connection)]
<ndowens08>
ah darn lol
<LnL>
ndowens08: if they enabled the allow edits, you can push with: git push git@github.com:<user>/nixpkgs.git <branch>
<dtzWill>
did 17.03 (channel) get a big bump recently? haha
<MichaelRaskin>
ndowens08: some people do allow pushes from committers of NixPkgs, though
tsmeets has quit [(Quit: leaving)]
hiratara has quit [(Ping timeout: 246 seconds)]
<ndowens08>
so this one i eidted manually, comes from MasseGuillaume:cre2, so I can do git push git@github.com:MasseGuillaume/nixpkgs.git/cre2 LnL ?
mkoenig has joined #nixos
hiratara has joined #nixos
<LnL>
ndowens08: use a space between the remote and the branch
<_deepfire>
ndowens08: what is the qtbase5 issue id, by the way?
<ndowens08>
_deepfire: it was an qt5base build issue, unsure it has been fixed?
<ndowens08>
in which your pkg did depend on, so w/o qt5base building fine, yours wouldnt either
eacameron has quit [(Ping timeout: 246 seconds)]
slack1256 has quit [(Remote host closed the connection)]
<_deepfire>
ndowens08: I get that yes, but was there a qt5base issue filed in nixpkgs?
<ndowens08>
_deepfire: i believe so
ryanartecona has quit [(Quit: ryanartecona)]
<_deepfire>
ndowens08: I have updated my nixpkgs checkout, and now I get "#error qdatastream.h must be included before any header file that defines Status"
<_deepfire>
..which seems like it comes from qt5base, but I wanted to make sure
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #24082: neo4j: 3.1.1 -> 3.1.2, minor fixes (master...fix/neo4j-and-update) https://git.io/vydHG
<ndowens08>
I updated nano, though there are i think 8 or so pkgs that uses it. some were the nix-prefetch-* and some others
mudri has quit [(Ping timeout: 240 seconds)]
<LnL>
the prefetch scripts depend on nano?
<ndowens08>
when i did nox-review it built them with nano
<ndowens08>
yea surprised me a lil :)
<ndowens08>
I think it is fine, but I will let someone else look at it, since it affects several pkgs :)
stepcut has quit [(Remote host closed the connection)]
abbradar has joined #nixos
<abbradar>
Dezgeg: hi! I'm trying to set up NixOS on ODROID XU4. First thought of bootstrapping it from ArchARM, set up Nix with a sandbox and tried to build "hello". Noticed that there is a hardcoded domain with your nickname for bootstrap tools and came to ask if there is any channel for armv7h available :D
<dhess>
Does anyone develop with NixOps, i.e., using a version you've checked out somewhere and are actively hacking? I'm a bit confused about how I might use my own dev version in my Nix environment
wkennington has quit [(Ping timeout: 256 seconds)]
stepcut has quit [(Remote host closed the connection)]
wkennington has joined #nixos
<dhess>
neeasade: yeah, I know how to do that in general when the package includes a default.nix file in the distribution. NixOps has a release.nix, which I can't callPackage directly, it seems
<dhess>
The default expression for NixOps in nixpkgs uses a generic.nix wrapper and then overrides a few things. I *think* I can use that directly and just override src, but I'm not entirely sure yet.
<disasm>
so I forked nixops and cloned it. I'm working on adding a vultr backend using the vultr python api. I've added nix/vultr.nix and nixops/backends/vultr.py. How do I go about testing this locally? I'm doing dev on my mac (which has nix and nixops currently installed) but also have a nixos server I can ssh into in the other room if it's easier to do from there.
<dhess>
disasm: If you're new to this you might want to do it on the NixOS machine, rather than the Mac. I'm not an expert but I'm not exactly a newb, either, and I haven't been able to get NixOps to deploy reliably from my Mac.
<disasm>
nice! thanks dhess! What are you hacking on with the d.o backend? I was looking at that and the ec2 for inspiration.
<dhess>
disasm: currently just a bug, but I'd like to get IPv6 support working.
<Mic92>
I mean creating real vms can be expensive
<abbradar>
Awesome, I've succeeded booting NixOS on ODROID XU4 without _any_ knowledge of what I'm doing (u-boot? what's that? :D)
<abbradar>
replaced half of files on boot partition with those from ArchARM
<abbradar>
somehow it works
<abbradar>
(with nixos kernel and initrd)
<dhess>
abbradar: what's the Ethernet chipset on that thing?
<abbradar>
dhess: something gigabit
<abbradar>
i haven't tested it extensively yet, just got the board today
<abbradar>
so no idea if it delivers what it promises (but ethernet works at least)
Fare has joined #nixos
<abbradar>
now on to properly adding nixos support for this, it will require actually understanding how boot sequence on those boards work...
<dhess>
abbradar: I have a Jetson TK1 with a built-in Realtek NIC. It uses the rl8169 driver in the kernel and hangs reliably after just a few minutes :(
eacameron has joined #nixos
<dhess>
I have to tell the switch to lock it at 100mbits in order to use it.
<abbradar>
dhess: I would have said you what the chipset is but it seems that my awe was too early -- UART is one-way only :D