<boomshroom>
I'm trying to build Redox's relibc on NixOS (it supports linux) and I can completely build it in a pure nix-shell. That said, Trying to use nix-build fails at `make: ./include.sh: Command not found`. `include.sh` is present and runs just fine when running buildPhase in a pure nix-shell, so I don't know what's happening.
<boomshroom>
That doesn't seem entirely relavent since 1. It's concerned with nix-env rather than nix-build vs nix-shell, and 2. It's looking at headers (which shouldn't be installed anyways) rather than being able to run a bash script.
jluttine has joined #nixos
<adamantium>
nixos doesn't seem to use bash_profile from a gui terminal, only the bashrc file. Does that sound about right?
<adamantium>
(on other distros I always sourced .bashrc from .bash_profile) So this seems different.
<ottidmes>
boomshroom: agreed, but it was just a thought that could be related to the underlying problem of your, why OK in nix-shell and not OK in nix-build, question
<boomshroom>
My understanding is that nix-shell --pure should give an almost identical environment to nix-build.
<ottidmes>
boomshroom: keyword being "almost", it might just be your nix-build environment that has something bleeding into it causing problem, which is circumvented with nix-shell pure
najibpuchong has joined #nixos
<ottidmes>
boomshroom: just curious, what if you run nix-build from your nix-shell?
<boomshroom>
nix isn't installed in a pure nix-shell. :P
<boomshroom>
No difference is an unpure shell
Ariakenom_ has quit [Ping timeout: 240 seconds]
<najibpuchong>
Hi all, I am using nixos 18.09 with sddm as display manager. How can I have sddm with selectable keyboard layout 'us' and 'us drovak' option in login screen?
shoogz has quit [Ping timeout: 264 seconds]
justanotheruser has quit [Quit: WeeChat 2.2]
<Lisanna>
gonna try my hand at some nix hacking... is there an internal function I could call in place of copyPaths that would let me substitute a fake output for a derivation?
justanotheruser has joined #nixos
drakonis has quit [Ping timeout: 246 seconds]
<Lisanna>
e.g., if I've in the meantime constructed some local file/directory in /tmp that I want nix to insert as an output into the store in place of the real output
adamantium has quit [Remote host closed the connection]
<Lisanna>
addToStore is probably what I want
<{^_^}>
[nixpkgs] @yashrajbharti opened pull request #49203 → Ackerman function → https://git.io/fx9Pg
<aleph->
Secondary question when generating a package how would I create additional folders in the nix store path? Just run mkdir -p ./PATH/HERE in the install phase step?
cinimod has quit [Ping timeout: 272 seconds]
<simpson>
Yes.
<simpson>
Well, `mkdir -p $out/path/here` but yes.
nD5Xjz has quit [Ping timeout: 264 seconds]
<drakonis1>
can i fetch from github two different repositories?
<aleph->
I remember you telling me about doing it.
justanotheruser has quit [Ping timeout: 252 seconds]
<aleph->
What drakonis1 is asking is, during a deriv build, can you fetch two different src folders from github? A la, $src = fetchGit {}; $src2 = fetchGit{};
<drakonis1>
basically that
<aleph->
With each src being a different git repository.
<simpson>
Yes, should work.
justanotheruser has joined #nixos
<drakonis1>
noice
mayhewluke has quit [Ping timeout: 246 seconds]
justanotheruser has quit [Ping timeout: 252 seconds]
<colemickens>
I was watching some of the NixCon stream. I'm wondering if a proper system for modules+pkgs would make auto-updating efforts easier to organize and implement as well?
<drakonis1>
okay so
<drakonis1>
i have a issue here
<drakonis1>
builder for '/nix/store/wnmrfnzx1bp5g49si49phj5x5mgm2q5q-lobsters.drv' failed to produce output path '/nix/store/q56vrr9rkjr7pnmc6mmr3kc8chnvs2ir-lobsters'
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
oldandwise has joined #nixos
aleph- has quit [Ping timeout: 252 seconds]
<oldandwise>
after i sync my cloned repo's master from upstream, how do I see the exact files (modified, added, deleted) ? I have done almost known attempts, but I am unable to do it. All I see is one filed being modified. However, during the `git merge upstream/master` there were 20+ files that scrolled, so I know there is more
<boomshroom>
Just pushed my first derivation to my nur repo.
<boomshroom>
Now to determine how I should access a function defined in mozilla-nixpkgs (which is also in nur).
init_6 has joined #nixos
<oldandwise>
Myrl-saki: thanks it worked. `git reflog` to get the beforehash and then do: `git diff --name-status $beforehash..$afterhash` will exactly show the detailed changes (added,modified,deleted)
<{^_^}>
[nixpkgs] @calbrecht opened pull request #49220 → vagrant: 2.1.2 -> 2.2.0 + vagrant-libvirt: init at 2.2.0 → https://git.io/fx95U
<Ericson2314>
jophish: platform is a grab bag of random stuff, system is a pair of CPU and OS, and config is a gnu/llvm style 3 or 4 tuple. At the top level platform and system should go away, instead leaving {cross,local}System.bothOfThem
<boomshroom>
Hmm... glibc is sneaking into my build even though I'm overriding libc in wrapCCWith.
<Myrl-saki>
boomshroom: How are you testing?
siers has joined #nixos
<boomshroom>
Myrl-saki: `nix-build -A relibcStdenv.hello` That wouldn't work in the current state since I'm changing the api to just give the stdenv rather than a package set built on it, but that's the idea.
<Myrl-saki>
boomshroom: And how do you know glibc is sneaking into your builds?
<boomshroom>
Myrl-saki: `/nix/store/kvdxajnlyisifi506ppbdpfycmcmsp6d-glibc-2.27-dev/include/bits/stdio2.h:97: undefined reference to `__fprintf_chk'`
<Myrl-saki>
Oh, lol
<Myrl-saki>
Hm.
<Myrl-saki>
boomshroom: How is relibcStdenv defined?
<siers>
gchristensen, can I use another xi:include in doc/functions.xml
<gchristensen>
siers: sure, what for?
<Myrl-saki>
boomshroom: Also, you might not want to use wrapCCWith.
<siers>
gchristensen, nix-gitignore section
<gchristensen>
sure, by all means!
<Myrl-saki>
How do I specify a file on IRC?
<boomshroom>
The interface changed a bit since the build, but the logic is the same.
<siers>
gchristensen, the only include in that file has kind of a weird name — shell.section.xml, which makes no sense considering how file listings are sorted
<gchristensen>
siers: ignore that one, it is because of a silly thing.
<gchristensen>
siers: name it like the other files :)
<siers>
gchristensen, no prefixes of "functions." or something? :)
<gchristensen>
siers: put it in ./functions/nix-gitignore.xml or whatever
<boomshroom>
Also, should I use LLVM instead of GCC, since relibc is in rust and compiles with LLVM?
<Taneb>
Well, if I can live without X I have a usable laptop
<Myrl-saki>
boomshroom: Give me a few.
<Taneb>
...I wonder if it'd be easier to get Wayland working
<Myrl-saki>
boomshroom: Actually, Idk what to do.
<boomshroom>
:P
<Myrl-saki>
boomshroom: Messing with stdenv is always weird for me.
<siers>
gchristensen, what's the shell.section.md for?
<Taneb>
Hmm, I there doesn't seem to be a way to use PRINTMODE:TEXT in the dwarf fortress in nixpkgs
<Myrl-saki>
boomshroom: I'd rather use overrideCC.
<Myrl-saki>
boomshroom: Than call stdenv directly.
<gchristensen>
siers: a limited number of docs sections are written in .md, and they need a certain naming convention to be built... but I'm wanting to get rid of them. actually, I thought I already did ....
<boomshroom>
Myrl-saki: That doesn't look like it would replace wrapCCWith, but would eliminate much of the boilerplate around it.
<Myrl-saki>
boomshroom: Yep.
<Myrl-saki>
boomshroom: I thought there was an stdenv version, but I was wrong lol
<Myrl-saki>
boomshroom: But yes, you might want to start from that.
<Myrl-saki>
boomshroom: I was working on a libc project, but I've since deleted the code. :/
<Myrl-saki>
boomshroom: lol, that's apparently exactly what I did.
jsgrant has quit [Remote host closed the connection]
<CMCDragonkai>
I'm trying to run `nix-build` on my haskell project. I always get this same error: `package.yaml: Yaml file not found: /build/haskell-demo-0.1.0.0/package.yaml`
<{^_^}>
[nixpkgs] @disassembler pushed commit from @r-ryantm to master « lighttpd: 1.4.50 -> 1.4.51 (#49238) »: https://git.io/fx9jl
<{^_^}>
[nixpkgs] @cleverca22 opened pull request #49242 → wireshark: include headers in wireshark.dev, to support building wireshark plugins → https://git.io/fx9j4
<symphorien>
<schopp0r> why is nix-build *ignoring* --arg doInstallCheck false << you want nix-build -E "with import <nixpkgs> {}; nix.overrideAttrs(old: { doInstallCheck = false; })" or so. Please read https://nixos.org/nixpkgs/manual/#sec-overrides
<{^_^}>
#49242 (by cleverca22, 9 minutes ago, open): wireshark: include headers in wireshark.dev, to support building wireshark plugins
<hyper_ch>
symphorien: ah.... I remember... I'm on the server back to pre-18.03 again because if I upgrade to 18.03 or newer, then networking doesn't work with the brdige for some reason
<snajpa>
guys, how can I test for existence of an attribute in another attrset, when defining the first said attrset? I have a file https://paste.vpsfree.cz/5DNyQUGl/ - documentation.nixos.enable didn't exist in 18.03, but it does in 18.09, so I'd like to set it if it's avail (I want to do that with documentation.enable as well, which is were I started)
<snajpa>
I think it's due to the test, which references pkgs...
<snajpa>
can I make a copy of pkgs just for the test of attribute existence, or is there another solution?
<{^_^}>
[nixpkgs] @cleverca22 merged pull request #49242 → wireshark: include headers in wireshark.dev, to support building wireshark plugins → https://git.io/fx9j4
<aki_k>
I've just setup NixOS and nixos-rebuild is not working as expected. Is there a place where users/newbies can ask I've they're having issues with Nix(OS), something like archlinux forums or ubuntu forums?
<gchristensen>
a LOT of NixOS people are at NixCon right now so aaren't paying as much attention, but here is still good :)
<clever>
some of us are at nixcon and paying attention!
<aki_k>
Well if I'm adding an application to configuration.nix it does not appear in the application menu. I'm only able to run them via the command line.
<aki_k>
I've first installed java 8 and have now completely removed it. But Sbt (a buil tool for jvm languages) is still using java 8. I've restarted several times.
<colemickens>
And the sbt would be because sbt needs the jvm so it pulls it in?
<symphorien>
aki_k: are you using a "unusual" shell like fish ?
<colemickens>
And I think `nixos-rebuild switch` doesn't drop a result link, but `build` does? and/or checkout result-N
vidbina has quit [Ping timeout: 276 seconds]
<colemickens>
Actually most of the time in GNOME and KDE I had to restart to get new apps in the menus.
<aki_k>
I'm using zsh but it occures on bash as well.
<aki_k>
Sorry, i was a little imprecise. I've set the property "programes.java.package = pkgs.jdk10;" and "java(c) -version" return that it's using java 10 but sbt is still using java 8 eventough it should no longer be installed on my system.
<symphorien>
aki_k: is your user declared in /etc/nixos/configuration.nix ? or did you create it manually with useradd ?
<symphorien>
the NIXPATH this is not necessary if you don't mind nix-env not seeing your overlay
elgoosy has joined #nixos
<symphorien>
s/this/thing/
<feep>
I do mind that; nix-env is the primary method I use for installing packages..
<jluttine>
in python package definition, how can i request a build input that is from top level packages instead of python packages? if i add "redis" to function arguments, the python one is passed..
<feep>
why would I want an overlay that nix-env can't see
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @r-ryantm to master « python36Packages.grip: 4.4.0 -> 4.5.2 (#49050) »: https://git.io/fxHJQ
<ekleog>
gchristensen: pong?
<symphorien>
a simpler example: glibc 2.27 broke locale format. if you rebuild your system to 18.09 from 18.03 but don't update things installed with nix-env they will complain that locale is broken. Putting everything in /etc/nixos/configruration.nix ensures you only use one glibc at a time and have no such problem
Itkovian has joined #nixos
<feep>
okay, let me ask differently
<aki_k>
clever: It seem to be working now. Isn't it possible to set that property by default?
<feep>
why does nix-env -i not put things in a file loaded from configuration.nix?
<symphorien>
I don't understand
<feep>
like, with gentoo (my previous system) when you try to install things what it does is it adds the expression you gave it to a world list of installed root files
<feep>
this sounds similar to environment.systemPackages
elgoosy has quit [Remote host closed the connection]
<feep>
I mean, how do I make it work period, would be good
<clacke[m]>
feep: nix-env as root modifies the packages installed for root, nix-env as user modifies the packages installed for user, systemPackages modifies the packages installed for the whole system. there is no conflict or confusion.
<symphorien>
unpack this in /etc/nixos/waterfox, and add to your /etc/nixos/configuration.nix `imports = [ ./waterfox ]`
<aki_k>
I've removed the "programs.java.package = jdk10;" property so I'm should be using java 8. And taht's also what "java -version" said, but intelliJ was still using java 8 I ran "nixos-rebuild switch" as root *and* as regular user. Is this intended behavior or a bug?
<aki_k>
*Intellij was using java 10
<feep>
clacke[m]: alrighty
dramforever has quit [Ping timeout: 256 seconds]
<feep>
clacke[m]: semi-relatedly, how do I list nix-env installed packages?
<{^_^}>
[nixpkgs] @disassembler pushed commit from @r-ryantm to master « mysql57: 5.7.23 -> 5.7.24 (#49221) »: https://git.io/fxHTV
noam has joined #nixos
<feep>
the problem with environment variables is
aki_k has quit [Quit: Leaving]
<feep>
you set nix.maxJobs/nix.buildCores and also add a source package, you have to comment it back out, rebuild, comment it in, and THEN and only then does it actually build with -j8
elgoosy has quit [Ping timeout: 245 seconds]
<feep>
anyway let's see if 32gb is enough for firefox at j8
<feep>
also how do I add a patch to a package?
<feep>
I have a bunch of patches leftover from gentoo
<{^_^}>
[nixpkgs] @nthorne opened pull request #49264 → cppclean: init at 2018-05-12 → https://git.io/fxHTH
<feep>
oh, duplicate the .nix file from upstream and edit? can do
aki_k has joined #nixos
<__monty__>
feep: I think you could do that with an overlay.
<__monty__>
feep: If you reread you'll find no less than 4 ways of specifying it in the section above the one I linked.
fendor has joined #nixos
mayhewluke has quit [Ping timeout: 246 seconds]
mayhewluke has joined #nixos
<{^_^}>
[nixpkgs] @marsam opened pull request #49267 → pgcenter: init at 0.5.0 → https://git.io/fxHk8
<symphorien>
feep: for a patch, instead of adding foo to environment.systemPackages, add (foo.overrideAttrs(old: { patches = (old.patches ? []) ++ [ ./patch ];))
<aki_k>
Is it possible to change the current time on NixOS e.g. via timedatectl?
<feep>
so this is what I'm trying: nixpkgs.overlays = [ (import ./overlays/gwenview) ];
<feep>
where overlays/gwenview has a default.nix defining an overlay that overrides gwenview to set a patches variable, which is probably wrong
<aki_k>
If I want to change the time, I must disable ntp. But "timedatectl set-ntp 0" yields "Failed to set ntp: Changing system settings via systemd is not supported on NixOS."
<{^_^}>
error: syntax error, unexpected ':', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):205:15
aki_k has quit [Remote host closed the connection]
<feep>
darn
<Shados>
feep: Yes, I think the wrong hash was probably left over from me testing with a local checkout. What I linked you is a way to get it in your system config though. Or, well, the way I'm doing it.
<feep>
Shados: thanks~ I already figured out how though
<typetetris>
For my multi package haskell projects, I use nix expressions like: https://pastebin.com/U7XNRLjj . Could I somehow merge my config from ~/.config/nixpkgs/config.nix with the one provided therein, with the project one taking precedence in case of conflicting stuff?
<Mo0O>
thanks a lot clever :)
<feep>
hm. trying to su -l into a user I created, and I get "ls not found"
<{^_^}>
[nixpkgs] @c0bw3b closed pull request #36261 → nixos/networking: add hostname to /etc/hosts by default → https://git.io/vA9oU
<feep>
gonna reboot and see if that fixes it
<feep>
it happens on every kde login now
<feep>
even root, which was a fresh acct
<symphorien>
does LD_LIBRARY_PATH contain /run/opengl-driver/lib ?
disasm has quit [Quit: WeeChat 2.0]
<Profpatsch>
ekleog: Uuuh, I just saw the rss2email PR. Nice.
<feep>
symphorien: will check in a sec
disasm has joined #nixos
<feep>
symphorien: yes
<feep>
yyep, I can no longer log into kde. :/ blegh
<feep>
it gives me a mouse cursor but no desktop
zarel has quit [Remote host closed the connection]
<feep>
owait hang on I may know what up
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ekleog>
Profpatsch: well, I rarely install a service without a nixos module :° unfortunately I used the after-upgrade OpenSMTPD, so it's not ready-to-merge yet… but obadz said it looked good, so I don't really know why it's not merged? (obadz is a committer, right?)
<ekleog>
maybe waiting for confirmation from rickynils?
<feep>
symphorien: I was playing around with nix-env, and I switched to configuration.nix, but I forgot to uninstall the crap from nix-env
Itkovian has joined #nixos
<infinisil>
> 1 + 1
<{^_^}>
2
<typetetris>
does nixos do errata to its release notes?
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
_kwstas has quit [Quit: _kwstas]
<{^_^}>
[nixpkgs] @aanderse opened pull request #49283 → WIP: solr: 4.10.3 -> 7.5.0, refactor service to reflect major changes in version bump → https://git.io/fxH3e
seang961 has quit [Quit: Page closed]
<{^_^}>
[nixpkgs] @LnL7 opened pull request #49284 → watchexec: fix darwin build → https://git.io/fxH3Z
jdnavarro has quit [Remote host closed the connection]
<Profpatsch>
ekleog: I *might* have sent the guy an email
<ekleog>
Profpatsch: yeah, it appears to have made the guy move again :D
<Profpatsch>
He’s not active on GH because of work and only reach-able by mail
<Profpatsch>
Hard to understand when most of your dayjob involves interacting with Github :Ä
<ekleog>
indeed ^^
<ekleog>
I guess now things will be totally calm until someone feels allowed to trigger a release / handle the waiting PRs :)
<Profpatsch>
ekleog: Feel free. I’ve watched the repo, so I’ll get notified when something changes
* ekleog
personally prefers to stay aside for the time being, having literally 0loc in the code, and come only if need arises ^^'
<ekleog>
like, if in a week nothing has moved I'll try to move things
<Profpatsch>
I think we can go forward and change stuff. The original contributor is inactive, jsbackus only had about 20 lines of changes as well.
<Profpatsch>
At least merge PRs that look sensible.
<ekleog>
well, Yannik appears to have quite a few changes
<{^_^}>
rss2email/rss2email#13 (by leezu, 39 weeks ago, open): Transfer some PRs from wking/rss2email/
<Profpatsch>
are we in sync with wking’s repo at least?
<ekleog>
yeah, upstream:master is a descendent of wking:master
<ekleog>
an upstream:dev too
<ekleog>
Profpatsch: feel like joining #rss2email? planning to add it to the readme in my cleanup for next week if no-one more “legitimate” made a move first :)
<bpye>
Does buildGoPackage let you pass in an -ldflags arg?
<Church->
Any articles on patching services to add options?
justanotheruser has joined #nixos
<simpson>
Church-: You'll probably want to read the Nix expression defining the service, and alter the expression. Some services support patching in extra configuration under names like `extraConfig`.
<Mic92>
Izorkin: I put it on my list, I am about to get into a plane.
<Izorkin>
ok
<catern>
hey #nixos, if I want to write a Nix-specific appplication that knows how to automatically deploy itself onto a remote host, should I have the (built) application store its own store path and use nix-copy-closure to the remote host, or should I have the application know the Nix expression it was built from and use nix-build onto the remote host?
<catern>
or, third option, should I instead make sure that the closure is already present on the remote host, and pass the remote store path in to my application?
<simpson>
Could you transfer only the Nix expression, or are you trying to keep that private?
<catern>
no, no need to keep the Nix expression private, but I don't understand what you mean - that was the second option I presented?
<simpson>
Oh, sorry, somehow I misread your second option. Yeah, that's what I was thinking of.
<simpson>
I guess it depends on where you want to do the build and bandwidth costs.
adamantium has joined #nixos
<catern>
I was thinking that that option was definitely inferior to just using nix-copy-closure, since it requires I keep around the Nix expression/a full description of how I was built
<catern>
I guess that's a fair point about bandwidth costs vs CPU cost
<simpson>
There *should* be a tool/technique for just keeping around the derivation, but I don't know it offhand.
<catern>
Even just keeping around the low-level derivation seems very distasteful
<catern>
Because it gives my application a bunch more knowledge about how it was created that it doesn't really need to just copy itself blindly
<catern>
Why would you keep around the derivation rather than use nix-copy-closure on yourself?
adamantium has quit [Read error: Connection reset by peer]
<catern>
(the concern isn't about distributing source instead of a binary - I don't care about that - the concern is about excessive authority)
adamantium has joined #nixos
Lisanna has quit [Quit: Lisanna]
<simpson>
I mean, if the app already knows that it's on Nix...
<simpson>
But sure, I see your point.
<catern>
(In reality I think I would have to be somewhat pluggable here, and not just hardcode usage of Nix, though I don't think that would stop any of these approaches, including keeping-around-the-Nix-expression - the Nix expression would just be kept around behind an interface)
<{^_^}>
[nixpkgs] @periklis opened pull request #49310 → googletest: Add pkgconfig support → https://git.io/fxHR0
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @Izorkin to master « znc: add configure options (#48683) »: https://git.io/fxHRi
<catern>
Hmm, thinking about the interface more, I think ideally I'd have something like a path-augmented-with-dependency-information object,
tg has quit [Quit: Leaving]
<catern>
just like I do in the Nix language. But the pathw ould have a "host" baked into it, and if I wanted to use it on a different host, then I'd have to run it through some translator to get a path on the different host.
<catern>
And that would do the deployment/copying/whatever
<catern>
Does that sound reasonable/make sense as an interface?
<catern>
At this point I feel sad that the Nix language is not general-purpose, so I can't just write this in the Nix language. And also I feel jealous of Guix, which has a good language, so they could definitely write this stuff in Guile :)
<drakonis1>
guix on the other hand has a lot less growth potential
<drakonis1>
why be jealous of guix for being written in guile when it has far less growth potential due to the way things are handled?
<Ralith>
because it's written in a good language, of course
<Ralith>
didn't he say that
<drakonis1>
a good language?
<simpson>
catern: AISI it's a good thing that Nix isn't general-purpose. There's also nixops, if you want to take this idea way too far.
<catern>
okay that was really just a side comment, I'm more interested in discussing this as an interface in some-language-which-is-not-Nix for working with paths/dependencies that can be deployed to a remote host and used there
<drakonis1>
fair enough, it is doable
worldofpeace has joined #nixos
justanotheruser has quit [Ping timeout: 252 seconds]
tg has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<__monty__>
drakonis1: Why does guix lack growth potential?
<drakonis1>
guile, being affiliated under the gnu project
<drakonis1>
it doesn't exactly lend to a environment equivalent to what nix has
<__monty__>
I don't see the problem.
<__monty__>
It's not like someone couldn't provide an external nixpkgs-esque repo for guix.
<drakonis1>
guile is the bigger issue
<drakonis1>
everything is guile
timokau[m]1 has joined #nixos
<__monty__>
No Idea why that's an issue.
<__monty__>
Also don't see why it has to be. Afaik they still use the same drv format?
zarel has quit [Quit: Leaving]
mayhewluke has quit [Ping timeout: 240 seconds]
mayhewluke has joined #nixos
<timokau[m]1>
Is there some way to run the regular checkPhase but then execute a hook if it fails?
<timokau[m]1>
I've found `failureHook` but that doesn't discriminate between phases
bts- has joined #nixos
wykurz has quit [Read error: Connection reset by peer]
bigvalen has quit [Read error: Connection reset by peer]
<arianvp>
anyone here using Yubikey PIV on NixOS for ssh?
oida has joined #nixos
<arianvp>
I can log into ssh with the yubikey, but I get "agent refused operation" if I try to add the key to my ssh-agent
<arianvp>
and no i'm not running gnome-keyr
<clever>
arianvp: that action would likely require copying the secret out of the yubikey and into the ssh-agent, and then all security is gone and somebody could just clone the yubikey
<arianvp>
but yeh... PKCS11 support has always been .. flimsy at best
<andi->
clever: ssh -J $jumpHost $actualHost
<arianvp>
I don't get how people work with HSMs in the real world
<arianvp>
it's all so extremely buggy. it's scary
<clever>
arianvp: ah
<arianvp>
TIL -J
<andi->
you can ofc also configure ssh so it does that automagically
<clever>
andi-: that would likely simplify this, lol: [clever@system76:~]$ ssh ext.earthtools.ca -A -t 'ssh c2d -t "screen -xRRU"'
<andi->
you can even do that conditional with a match and exec
<{^_^}>
[nixpkgs] @tathougies opened pull request #49311 → nixos/networking: do not copy /etc/rpc on non-glibc systems → https://git.io/fxHEd
silver_ has joined #nixos
<arianvp>
oh well. typing in my pin every time I ssh isnt that bad
<arianvp>
better than a passphrase :P
<clever>
arianvp: an ssh master connection can also help
<andi->
at least you are using a protected ssh key
<clever>
arianvp: you can rig it up so ssh shares a single tcp socket (and auth session) between many ssh shells
<arianvp>
I'll open up an issue with yubikey perhaps they know whats up
<andi->
clever: I managed to reproduce the timeout scenario in a vm test with hydra in nixpkgs.. It just doesn't quiet bug the same way yet.. How does the /etc/nix/machines file look like in you hydra? Do you use any special "transport" (ssh-ng, …)
<arianvp>
googling the erro doesn't really help
<arianvp>
just people saying "Dont use macos"
<arianvp>
but I dont use macos
<clever>
andi-: nope, one sec
<arianvp>
oooh no. prometheus module still needs to be updated to v2
<clever>
andi-: its just things like builder@amd.localnet x86_64-linux /root/.ssh/nix 8 1
<arianvp>
=(
<arianvp>
sounds like a chore
<clever>
andi-: though i currently lack access to the hydra i reproduced it on, the power blipped at home, and that box never came back online
<andi->
awh :/ I've set the timeout to 3600 now, setting it to 10 seconds made it stop within the expected timeframe..
silver has quit [Ping timeout: 252 seconds]
<clever>
andi-: it may also be possible to rig up a nixos-test using the existing qemu framework, to spin up a hydra and build slave, with the sortaSlow derivation as a job
<andi->
that is what I did
<clever>
then you could test it in a pure manner, and hydra can keep testing it in the future
<clever>
nice
<andi->
I have a build doing while true; echo -n; done all day...
<andi->
didn't want to deal with importing anything external to just make it hang there
acarrico has quit [Ping timeout: 246 seconds]
<clever>
andi-: does that test reveal any behaviour differences between user@host and ssh-ng://user@host in the machines file?
<andi->
clever: haven't tried ssh-ng yet. I will check if the 1h timeout works as expected.. I guess it will
<andi->
Might be that just ectending the timeout doesn't work and my initial assumption might be wrong
<schmittlauch[m]>
That's probably a bad thing as the latest firefox-63 release, including security fixes, is not included
<schmittlauch[m]>
clever (IRC): So updating channels usually doesn't take that long, right?
ryantrinkle has joined #nixos
<clever>
correct, only happens if something within the channel's tested set fails
<andi->
Model> Advantage2
<andi->
Firmware> 1.0.431.us (4MB), 11/14/2017
<andi->
grml, hand rolling on the keyboard /o\
<schmittlauch[m]>
clever (IRC): Who needs to be notified about that? Not having security updates for 10 days isn't nice
<schmittlauch[m]>
Or are people aware of this already?
<clever>
sounds like a task for #nixos-dev
<clever>
i'm also looking into the plasma failure
<schmittlauch[m]>
Ok. Will you bring this up in #nixos-dev or shall I?
drakonis_ has joined #nixos
<clever>
you can
<schmittlauch[m]>
k, will do
jasongrossman has joined #nixos
<__red__>
What's the conceptual difference between nixos and nixos-dev? General user stuff vs Distro engineering?
<simpson>
More or less. Not everybody in #nixos-dev is a committer, but generally only stuff relevant to maintaining Nix, nixpkgs or NixOS happens in there.
<Lisanna>
Where's the spot in the Nix code that handles the informational updates that the stdenv builder can send to the "nix" process to update the status bar?
__monty__ has quit [Quit: leaving]
drakonis has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @andrew-d opened pull request #49315 → vlc: shrink runtime closure size by removing CONFIGURE_LINE → https://git.io/fxHgW
<clever>
Lisanna: was just reading that ~5 hours ago
<{^_^}>
[nixpkgs] @Infinisil pushed 2 commits to release-18.03:
<pareidolia>
clever: Is the syntax for nix copy in the documentation?
<clever>
not really, but you can `nix copy --help` to get most of the args
<clever>
the store-uri for --from and --to supports several schemes
<clever>
local tells it to just open /nix/store/ directly
<clever>
daemon is for connecting to nix-daemon via /nix/var/nix/daemon-socket/socket
<clever>
local?root=/mnt/ will make it emulate a chroot before working on a localstore (it just prefixes all paths with /mnt/)
<clever>
ssh://host and ssh-ng://host will work over ssh to a remote nix
<clever>
s3://bucket will work for amazon S3 buckets
<clever>
file:///path/to/dir i believe will generate a pile of .nar.xz and .narinfo, that could then be rsync'd to an http server and act as a binary cache
<pareidolia>
It's not documented presently?
<clever>
i dont think the above is documented fully, but ive not re-read the docs in a while
sigmundv has quit [Read error: Connection reset by peer]