<haslersn>
samueldr: should that CPPFLAGS be a string (with spaces) or a set of strings? Both works
<samueldr>
both incidentally work AFAICT
<samueldr>
the string will be set as-is to the environment variable
<samueldr>
and the list will be joined with spaces when set to the environment variable
<haslersn>
Yeah both work but I wonder what's recommended
<rawtaz>
ok, i finally got both fstab mounting of shared folders under Fusion to work, as well as shared folders mounting when you just enable it in the VMs settings, to work. it's all just about problems with paths to the relevant open-vm-tools binaries in nixos, and systemd units not adhering to path settings. will sum it up in an issue later.
<rawtaz>
but dang, this was progress, now things are actually working :)
<{^_^}>
[nixpkgs] @grahamc merged pull request #47017 → dockerTools.buildImage: test that created=now makes an unstable date → https://git.io/fAQd6
<Enzime>
is there an easy way to use my nixpkgs.config.packageOverrides over a custom nixpkgs to see what will get built by nixos-rebuild
* rawtaz
&
<haslersn>
What should be the commit message if I add changes that were requested by a pull request review? (I suppose squashing that commit can be done later)
<samueldr>
is you intend to squash, then it doesn't matter much :)
<samueldr>
make an intelligent commit message then
<samueldr>
and make it obvious you intend to squash before the merge, possibl
<samueldr>
y
<haslersn>
Ok, thanks
<haslersn>
Does fetchFromGitHub have something similar to the deepClone option of fetchgit?
<haslersn>
(In order to init submodules)
Zer000 has joined #nixos
<haslersn>
I found it. It's called fetchSubmodules
<{^_^}>
[nixpkgs] @grahamc pushed to master « Clarfy the binary reproducibility problems of created=now with dockerTools.buildImage. »: https://git.io/fA72a
camsbury has quit [Ping timeout: 240 seconds]
<Enzime>
if you're planning to squash it anywhere
<Enzime>
you can make a fixup commit
<Enzime>
git commit --fixup=<commit>
<Enzime>
then it can easily be detected by git rebase --autosquash
<haslersn>
slabity: you don't need the store path for packaging: patchelf --set-rpath "${lib.makeLibraryPath [ glib ] }" # I don't know about libgthread
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hakujin has joined #nixos
<jackdk>
in recent nixpkgs, ghc821Binary has fixes to work on a musl system, but ghc7103Binary does not. I am trying to backport those fixes, but the configure script is failing with a "not found" while trying to run a binary. This is usually a sign of a failure when dynamic loading, but ldd is not showing any missing files. how do I dig further?
<jackdk>
bah, and after all that it segfaults. I give up.
silver has quit [Read error: Connection reset by peer]
Supersonic has joined #nixos
hakujin has joined #nixos
sir_guy_carleton has joined #nixos
hamishmack has joined #nixos
haslersn has quit [Quit: Page closed]
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #47096 → haskellPackages.x509-system: Override based on the host, not target platform → https://git.io/fA7wH
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #47097 → haskellPackages.x509-system: Override based on the host, not target platform for 18.09 → https://git.io/fA7w5
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #47096 → haskellPackages.x509-system: Override based on the host, not target platform → https://git.io/fA7wH
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #47097 → haskellPackages.x509-system: Override based on the host, not target platform for 18.09 → https://git.io/fA7w5
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-18.09: https://git.io/fA7re
<kyren>
hey, so I'm trying to set up a local development derivation usable with nix-shell that uses rustup and I'm running into a problem I think with some kind of linker wrapper
<kyren>
I have a shellHook that runs cargo, and when building I get an error due to an impure path being used while linking
<kyren>
I know it's impure, I just literally want to do the build right in that directory, my goal is not really to make a real derivation just to set up a dev environment
<jackdk>
I don't know rust. In the haskell world, the nix files produced by cabal2nix often end with `if pkgs.lib.inNixShell then drv.env else drv`, where `drv` is the derivation you've constructedd
<kyren>
I either want to disable whatever the linker wrapper is, or I'm not doing things right at all and should just do this in some vastly different way, I *have* actually figured out how to build this purely, but for dumb reasons there's a completely different problem when building that way that has to do with mozilla's rust overlay
<kyren>
the rust equivalent is "carnix", and I'm not using it, it's a *very* simple shell.nix file
<kyren>
the problem, I think, is just that the shellHook hits a purity test in a wrapper and I basically don't care about that at all, I intend for this to be impure
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Ralith>
kyren: there's not much point in using rustup when you have nix
<kyren>
holy cow I actually found it, the solution is 1) set NIX_ENFORCE_PURITY=0, and 2) use mkShell instead of mkDerivation just to be totally clear that this is only to set up an impure dev environment
<kyren>
I actually agree, and spent a loooong time trying to get the pure version to work
<kyren>
that's more or less what I'm doing actually, I don't know what the bug is actually
<kyren>
I wasn't able to use carnix so I ended up vendoring all of my dependencies and that seemed to work, though it's not ideal
<ekleog>
oh if you're compiling via nix then it's maybe different, I'm just using a shell.nix and compiling with regular cargo
<kyren>
I'm also attempting to use cargo-web which may have something to do with it, it's bad because I'm new at both wasm32-unknown-unknown *and* nix
<kyren>
oh right right
<kyren>
yeah that definitely works, it was working before
<ekleog>
(rustUnstable being pkgs.rustChannelOf from 2018-08-01 ; ISTR I had issues when I last tried updating)
<kyren>
I just couldn't put a command in shellHook because something was freaking out about impurity
<kyren>
oh, well sorry no I got confused, you're still using the overlay you're just not doing the build purely, I haven't tried that combination
<kyren>
sorry it's late and I'm tired and confused haha
<ekleog>
no problem :)
<kyren>
I'll try that now because I'd rather not use rustup
<ekleog>
if you do that, also try going to a nightly from 2018-08-01, I remember weird problems with both more recent and older nightlies :)
<kyren>
yeah I jumped around versions trying to see if the thing I was running into was related to a specific rust version, so I know how to do that and have hard coded the version you recommended :)
<kyren>
huh, weird, using the overlay and building impurely results in the same LinkError, I'm going to try using the specific exact overlay archive version that you are...
<kyren>
okay, I'm using exactly the same setup as you, and I'm basically just trying to run the contents of the stdweb "echo" example, nothing more
<kyren>
that's really weird!
rauno has joined #nixos
<rauno>
how can i run fsck on boot partition? tried changing default grub entry's init=/bin/sh but that fails somewhy, sais /mnt-boot/bin/sh does not exist or smth
<clever>
rauno: tell it to continue even though it doesnt exist
<ekleog>
kyren: hmm, I'm not compiling stdweb examples, just custom-built examples, so it looks like I can't help you on that, sorry :/
<clever>
rauno: its a bug in the detection code, not dealing with symlinks under a chroot
<rauno>
ahaa! thx will try again :)
<ekleog>
kyren: if you try compiling just `#[no_mangle] pub extern "C" fn test() { }` with crate-type=cdylib, does it work out?
<rauno>
even thought to try that option but then just went for reboot and searching more infomration
Thra11 has quit [Ping timeout: 272 seconds]
<ekleog>
(I also have lto=true, incremental=false, opt-level="z" and panic="abort", for what it's worth, but don't think it should impact anything)
<kyren>
hmmm, lto=true and incremental=false might help
<kyren>
well, I guess I don't know because I don't actually know how wasm linking works and I guess it uses rust-lld
<kyren>
here is the example I'm compiling, if you're interested, but I'm going to try the things you suggested as well: https://github.com/kyren/wasm_test
ambro718 has joined #nixos
<kyren>
well, it changed behavior at least, "import object field 'memcmp' is not a Function"
Ariakenom has joined #nixos
<ekleog>
you see that… in the javascript console when trying to actually run it?
<kyren>
yeah
brejoc has joined #nixos
<ekleog>
hmm, so my example will work for you
<kyren>
it's in WebAssembly.instantiate actually
<ekleog>
I'd be curious what wasm-objdump -x [your .wasm file] gives for memcmp
<ekleog>
,locate wasm-objdump
<{^_^}>
Couldn't find any packages
<kyren>
I'm guessing it's in wasm? that's what I'm trying now
<kyren>
well, nope
<ekleog>
huh? is {^_^} running on 18.03? it's in the `wabt` package
<ekleog>
or maybe the PR never made it to nixpkgs and I'm just getting it off my fork, idk :°
<kyren>
no, it's definitely there
<ekleog>
nah, merged a month ago
<kyren>
I'm on unstable so I got it
<ekleog>
:)
<kyren>
man, I am *so* not a web developer, okay
<ekleog>
infinisil: suggestion: have {^_^} on unstable? that should have at least as many packages as 18.03 :)
<kyren>
it appears that memcmp.... refers to env.memcmp, which there is no other mention of
<kyren>
maybe, I have no idea what I'm doing
<ekleog>
hmm, so it's likely llvm emitting a dependency on a stdlib function that isn't provided by wasm indeed
* ekleog
doesn't do web programming, but does wasm for… other reasons :p
<kyren>
I'll compare it with one built with rustup
Kelppo has quit []
<ekleog>
worst case, you can implement memcmp yourself as a #[no_mangle] pub unsafe extern "C" fn memcpy(...) { ... }
<ekleog>
(I think)
<kyren>
that might just be the first of several problems, I have a feeling it's just giving me the *first* error like that?
<ekleog>
you can get the list of imported functions from wasm-objdump -x
<ekleog>
(the `env.*` stuff is imported from the runtime)
<ekleog>
but yeah, there should be a way to tell llvm not to emit function calls, but I can't remember it :/
<kyren>
yeah okay, now memcmp appears to not be coming from env.
<ekleog>
you can consider looking at https://stackoverflow.com/a/34288242 too, I guess :) there really should be a streamlined solution for wasm, though
<kyren>
yeah it's not quite worth it just to not use rustup, clearly there's some kind of bug in the mozilla overlay
ambro718 has quit [Ping timeout: 260 seconds]
<kyren>
I *could* reimplement things that cause linker errors, or I could add to the bug reports for the mozilla overlay haha
Zer000 has quit [Ping timeout: 246 seconds]
<kyren>
it looks like it's (in the very simple example) only memcmp, memcpy, memset, and memmove so that's actually not *terrible*
<kyren>
anyway, thanks for going through that with me :D
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
periklis has joined #nixos
<ekleog>
well, good you figured it out :)
andreabedini has joined #nixos
hyper_ch2 has joined #nixos
rprije has quit [Ping timeout: 260 seconds]
rprije has joined #nixos
rotaerk has quit [Ping timeout: 240 seconds]
goibhniu has joined #nixos
rprije has quit [Ping timeout: 260 seconds]
Anton-Latukha has joined #nixos
hamishmack has joined #nixos
<{^_^}>
[nixpkgs] @xeji opened pull request #47102 → nixos/tests/atd: wait for atd to start → https://git.io/fA7SN
rotaerk has joined #nixos
georges-duperon has joined #nixos
ixxie has joined #nixos
<ixxie>
Hey folks
orivej has joined #nixos
<ixxie>
I've been greenlighted to setup NixOS on personal work machine
<ixxie>
unfortunately I still need to use some windows stuff, so I need to decide on a way to have them work in tandem
<ixxie>
Ideally I want to have a desktop manager
hyper_ch2 has quit [Ping timeout: 256 seconds]
<ixxie>
so I am split between the options of virtualizing NixOS on Windows or making the desktop serve a remote desktop protocol to my windows laptop
<ixxie>
so I am wondering if there are reliable solutions for remoting
civodul has joined #nixos
sb0 has quit [Quit: Leaving]
Izorkin has quit [Disconnected by services]
mikeplus64 has joined #nixos
<mikeplus64>
`nix copy` stalls for me ~250/330MB (it varies each time how far it gets) and it isn't an OOM issue. nix-copy-closure works OK. any ideas?
<mikeplus64>
on nix 2.0.4
hyper_ch2 has joined #nixos
<goibhniu>
ixxie: how about running windows in a VM?
<mikeplus64>
(copying over ssh)
<ixxie>
goibhniu: hmmm nice idea :)
<ixxie>
after all its only for the office suite
<goibhniu>
perfect! it could be more tricky if it was for some dodgy VPN tools etc.
glasserc has quit [Ping timeout: 244 seconds]
Izorkin has joined #nixos
<goibhniu>
ixxie: BTW congrats on being able to use NixOS at work \o/ #NixOSAllDay
<srhb>
ixxie: I've had to use Windows in a VM before and it's much more tolerable than the other way around, so I second that
<emily>
you might also be able to use wine
<srhb>
ixxie: (I use a tiling WM, so it was a bit funny that workspace 5 was essentially just Windows. I've had colleagues use my PC in that workspace not knowing that it was a *nix machine, simply because the only "tell" is the shortcut to detach focus from the VM)
<sphalerite>
ixxie: yaaaaay! I'd suggest wine if it works for the applications you want to use, otherwise yeah a VM (windows does work in qemu, which is a nice way to avoid virtualbox which I kind of dislike)
<hyper_ch2>
sphalerite: but then, vbox is much less pain to work with attaching usb devices IMHO :) but once you have qemu up and running it's nice (also because of SPICE html5)
<{^_^}>
[nixpkgs] @luke-clifton opened pull request #47103 → dvtm: refactor (fix on darwin) → https://git.io/fA7QD
<sphalerite>
hyper_ch2: how so? I haven't really tried it myself but there are monitor commands for it
reinzelmann has quit [Ping timeout: 260 seconds]
<sphalerite>
attaching a virtual USB stick definitely works like a charm, I can't say much about passthrough from the host
<{^_^}>
[nixpkgs] @xeji pushed commit from @marsam to master « bat: install man page (#47099) »: https://git.io/fA75k
<{^_^}>
[nixpkgs] @xeji pushed commit from @marsam to release-18.09 « bat: install man page (#47099) »: https://git.io/fA75q
<ixxie>
I will trust the wisdom of the crowd on this one
<ixxie>
xD
<ixxie>
so besides vbox there is qemu and kvm as the main options or?
<sphalerite>
qemu uses kvm
<sphalerite>
it's basically virtualbox, qemu, maybe vmware stuff? idk if it runs on nixos, but if you're not required to use it I'd suggest not because it's proprietary binary-only stuff and we all know how much fun that is
<ixxie>
so qemu it is
hyper_ch2 has quit [Quit: Page closed]
* etu
use vbox of old habits and vagrant reasons
<sphalerite>
oh or maybe you could use xen or something like that? I don't know things about running a hypervisor "above" linux
<etu>
sphalerite: xen for graphical stuff? Not sure, maybe?
<goibhniu>
I switched from vbox to qemu after I had some issues with VMs failing after an upgrade of vbox ... quite possibly a user error, but I've never had a problem with qemu
<sphalerite>
etu: many years ago a friend of mine was doing that I think
<tazjin>
setting `doCheck = false;` on the generic Haskell builder skips running the test suite, but still builds it - I'd like to avoid that, does anyone know if there is a toggle for it?
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @xeji to release-18.09 « nixos/tests/ferm: disable dhcpcd »: https://git.io/fA7Fg
<srhb>
tazjin: You probably need a configure flag for that
NightTrain has quit [Ping timeout: 246 seconds]
<srhb>
tazjin: Got a good package to test on?
<srhb>
tazjin: I think appendConfigureFlag pkg "--disable-tests" might do it
<tazjin>
srhb: doing it on a non-public project atm, but I'll try that - thank you!
<{^_^}>
[nixpkgs] @vcunat pushed 4 commits to release-18.09: https://git.io/fA7xh
<booglewoogle>
hey, currently reinstalling 32-bit nixos on an old machine. while it works, it gave me "warning: unable to download 'https://cache.nixos.org/nix-cache-info': Peer certificate cannot be authenticated with given CA certificates (60);" a few times, it did actually start the DL after about a minute. don't know if this is normal/important?
<realrokka>
"No manual entry for cc" what is the package to get cc manpages? I already installed manpages, but can't seem to find the pkg for man cc to work
<{^_^}>
[nixpkgs] @xeji pushed commit from @7c6f434c to release-18.09 « acl2: 8.0 -> 8.0.post.2018.09.20; fixes build with fresher SBCL »: https://git.io/fA7pi
<Myrl-saki>
realrokka: cc is a symlink to whatever compiler you're using, isn't it?
<Myrl-saki>
booglewoogle: I don't know what that means, but if it's an old version of NixOS, it might mean that the certificate is outdated.
<realrokka>
realrokka: i see, thx just expected it to work like on other distros :)
<Myrl-saki>
realrokka: FWIW, you can try man $(basename $(realpath $(which cc)))
<Myrl-saki>
realrokka: Just don't tell #bash. :P
<srhb>
tazjin: Great :)
<realrokka>
Myrl-saki: /prism (つ ͡° ͜ʖ ͡°)つ
<booglewoogle>
Myrl-saki: the nixos ISO is freshly downloaded. It also gave me the "warning: The option `networking.enableIntel3945ABGFirmware' defined in `/mnt/etc/nixos/hardware-configuration.nix' has been renamed to `hardware.enableRedistributableFirmware'." --- maybe the 32 bit ISO from the homepage is outdated?
<Myrl-saki>
booglewoogle: I believe that it's nixos-generate-config that's outdated in this case.
asymmetric has joined #nixos
<booglewoogle>
myrl-saki: wouldn't that still mean the ISO is outdated as the script comes with it?
<srhb>
Arguably.
<booglewoogle>
oh, and if it's just the script: is there a way I can update this inside the installer?
<srhb>
booglewoogle: I don't think anyone has fixed the script yet :)
<{^_^}>
[nixpkgs] @fpletz pushed commit from @r-ryantm to master « earlyoom: 0.11 -> 1.1 »: https://git.io/fA5ef
<rawtaz>
infinisil: can you merge #46921 so i can backport it?
<tilpner>
asymmetric - nixpkgs doesn't allow IFD, so you can't import (fetch...) it, but NUR does. It's not quite as nice as being in nixpkgs, but better than being your own repo. https://github.com/nix-community/NUR
bara has quit [Ping timeout: 272 seconds]
orivej has quit [Ping timeout: 252 seconds]
orivej_ has joined #nixos
booglewoogle has quit [Remote host closed the connection]
silver has joined #nixos
waleee has joined #nixos
<asymmetric>
tilpner: what does IFD stand for?
<tilpner>
,IDF
<tilpner>
,IFD
<tilpner>
Come on
<tilpner>
It's import-from-derivation
<shreyansh_k>
Hi, can you confirm that switching to 18.09 only requires updating the channel and not the configuration file? I am thinking of switching to 18.09 .
<symphorien>
(be aware that 18.09 is not released yet)
<tilpner>
shreyansh_k - Generally yes, though there may be deprecation warnings you need to fix. For more safety, it doesn't hurt to skim the release notes
<asymmetric>
tilpner: doesn't NUR kind of solve the same problem as overlays?
<tilpner>
asymmetric - Not quite. If a user sets up the NUR once, they can use any NUR packages, while they would have to individually setup each overlay they use
<asymmetric>
yeah, modulo these differences. i'm going to look into NUR, thx
<tilpner>
asymmetric - Additionally, an overlay is much unsafer than the package sets used by the NUR. An overlay has much more power, to e.g. replace systemd. Users would have to audit all used overlays
<asymmetric>
that's a good point. so the main reason why someone would add a package to NUR instead of nixpkgs is IFD?
<tokudan[m]>
Mic92: i saw that you merged one of the automatic update PR requests for burp, which is actually not good. I've just made a new PR (#47115) to revert it. could you merge that to avoid someone updates to that old version?
<gchristensen>
ryantm: ^
<srhb>
tokudan[m]: Can you explain that a bit more in the PR? According to github, 2.2.4 is newer than 2.1.32
<{^_^}>
[nixpkgs] @nthorne opened pull request #47117 → uftrace: init at 0.9 → https://git.io/fA5qb
<tokudan[m]>
srhb: burp has to "branches". the stable branch is still 2.1.32, while the latest branch is updated roughly once a month and is currently at 2.2.10
<tokudan[m]>
we could add another package burpLatest or something like that, but having it somewhere besides master would make no sense. especially in a release-* would mean the time it's released, there's already one newer version
<gchristensen>
I think your revert is the right thing to do, no need to complicate it
<srhb>
Yeah, agreed.
<tokudan[m]>
last time the automatic update process was running, i caught it in time, but this time you're just working too fast ;)
<srhb>
tokudan[m]: Let's put a comment near the version attribute
<sphalerite>
Phillemann: if you just want to use it one-off (as opposed to using it as a dependency of something else in nixpkgs) you can just use python3Packages.pysdl2.override { SDL2_ttf = (your SDL2_ttf); }
<betaboon>
a question on nixops+aws: do i realy need to `associatePublicIpAddress = true;` ?
<gchristensen>
did somebody actually implement the idea of having each store path be a docker layer?
iyzsong-x has joined #nixos
<sphalerite>
gchristensen: what would the point be?
<gchristensen>
more sharing between dockerTools.buildImage results
<Phillemann>
Mic92: No quite I think. SDL2_ttf is a (C) dependency of pysdl2.
iyzsong has quit [Ping timeout: 252 seconds]
<Mic92>
stty columns 212; stty rows 59
<Mic92>
h30:
<Mic92>
oh man a tmux short going wrong
<Mic92>
*shortcut
<sphalerite>
gchristensen: not really I think? Because layers can't be mixed and matched AFAIK?
<Mic92>
Phillemann: yes, but you want to override a python one.
<gchristensen>
sphalerite: they can, layers are content-addressed and order is not fixed at the layer level
<sphalerite>
gchristensen: oh sweet
<gchristensen>
sphalerite: the problem is there is a maximum of (iirc) 127 layers, so it is only a sometimes-possible optimization.
<Mic92>
gchristensen: having too many layers breaks overlayfs so
<Mic92>
and make it slow
<gchristensen>
also that, so it isn't always possible or desirable.
<Mic92>
I still think bind mounts would be the better option
<sphalerite>
just bind-mount /nix from the host and set a-rw,a+x on /nix/store
<sphalerite>
:D
<Mic92>
or the individual dependencies.
<Mic92>
/nix/store/<foo>
<gchristensen>
I don't disagree, but if we can produce pretty good OCI containers it is a good answer for teams who can't install Nix on their OCI-compatible clusters.
<allenj12>
Hey, I tried nix a while ago and am considering it for a work os between manjaro. How is the project coming along? I remember before there was a learning curve and had trouble finding some docs is the doc situation better now?
reinzelmann has quit [Quit: Leaving]
<goibhniu>
allenj12: it's getting better every day :D
<Acou_Bass>
there has been a bit of splitting with the docs, theres now https://nixos.wiki/ which is unofficial and (right now) fairly small, theres now a discourse forum as well though which is cool
<etu>
allenj12: And changes and suggestions are welcome :)
<sphalerite>
allenj12: it's still far from perfect. There's lots of good stuff on the wiki, but I often find myself asking questions here (and I've been using nixos for 2 years now) — people here know lots of things and are very helpful :)
<Acou_Bass>
the only thing i would say arch/an arch spin like manjaro has over nixos is, a lot more of the software is more updated (for example theres a lot of music production software that even in Master is horrendously outdated), but if thats not a big deal, come on over ;D
<sphalerite>
and nowadays our software tends to be alot fresher thanks to ryantm and his bot
<sphalerite>
a lot
<goibhniu>
Acou_Bass: ah, I thought the audio stuff was being kept up to date very well by ryantm
<Acou_Bass>
most of it is to be fair
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
<allenj12>
Thank guys! ill check out the wiki page and poke around. I think at the very least ill use it as a home os I really did enjoy what I experienced before
<Acou_Bass>
my only real gripe is tuxguitar which is like, yearssss out of date
<gchristensen>
mog: _finish_? I've only just seen it! :)
<Acou_Bass>
ive botched together an absolutely atrocious nixpkg for the updated version but i dont really want to share it because its a bit crap and some things dont work within it
<mog>
gchristensen, i hit discourse and github first
<goibhniu>
ah, perhaps that's not easy to update that automatically
<Acou_Bass>
yeah just bumping the version number didnt work as the nwer versions have a lot of different features and are built against different stuff to the older one
<Acou_Bass>
so it probably needs a clean new re-write really hehe
<goibhniu>
hrm, maybe it's better to remove it then
<niksnut>
viric: to be clear, I have no actual reason to suspect your crash is caused by coroutines
lawlesseel has quit [Ping timeout: 256 seconds]
<Acou_Bass>
hey mog just a question - are there any legality issues in using the nixos logo for non-commercial use, ie. if i print it onto a shirt myself? XD
<{^_^}>
[nixos-artwork] @mogorman opened pull request #36 → adding artwork for 18.09 release → https://git.io/fA5Zz
<mog>
Acou_Bass, you can but let me pull up where it was written as i dont have it at my finger tips
<mog>
The logo is made available under a CC-BY license.
<Acou_Bass>
ahh col
<Acou_Bass>
i mean im not trying to setup a rival shop to yours or anything
<mog>
totally fine if you do
<Acou_Bass>
but i have a habit of printing logos onto shirts and stuff so y'know its good to know i can do so without issue :P
<mog>
and my modifications of the logo fall under same stuff. i dont care if people print them externally
<Acou_Bass>
a while ago i made a cycling kit with the nix logo on it, ive not had a chance to wear it in public though so no ones pointed it out yet :(
<Acou_Bass>
i had one with the arch logo too but it was very shoddily done so eh
<mog>
i know that about arch, but how was the clothing... ^__^
<Acou_Bass>
well it wasnt done with a sublimation machine so the printing is sort of 'on top of' the fabric... so it doesnt really look any good, it just looks like a sticker hehe
Phillemann has left #nixos ["WeeChat 2.2"]
<{^_^}>
[nixpkgs] @ckauhaus opened pull request #47127 → lua: remove lua4 and lua5_0 → https://git.io/fA5nk
<Taneb>
Is there a "This is what needs to be done before nixos 18.09" anywhere?
<Myrl-saki>
Apparently 3.8 MB of drvs.
<Myrl-saki>
I was expecting 10 MB.
<ckauhaus>
Taneb: #45960
<hyper_ch2>
:)
<Myrl-saki>
With --apparent-size, 24M.
<Myrl-saki>
2.4M*
<Myrl-saki>
This is on ext4, so no compression magic.
<hyper_ch2>
you still haven't upgraded to zfs? :)
<Myrl-saki>
hyper_ch2: I've upgraded my /home/$USER to zfs.
<romildo>
I am building a package (enlightenment.rage) which is failing with the message
<romildo>
Package gbm was not found in the pkg-config search path.
<romildo>
Which dependency should I use to fix this?
<romildo>
to the PKG_CONFIG_PATH environment variable
<romildo>
Perhaps you should add the directory containing `gbm.pc'
romildo has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<Myrl-saki>
Sad.
<Myrl-saki>
It's happening again. :(
<Taneb>
ckauhaus:
<Taneb>
Thanks
<samrose>
when working with nixos containers declaritively, what is the typical way people are starting the container in the nix expression (for instance in configuration.nix)?
<Myrl-saki>
Oh yeah.
<Myrl-saki>
Er, nothing.
<Myrl-saki>
Oh wait, I guess this is a valid question. Is there a way to pass a user argument to configuration.nix?
<Myrl-saki>
I'm thinking of making making my own `nixos-rebuild` where certain services are activated and deactivated.
<hyper_ch2>
there probably is - but I have no idea what it would be
<Myrl-saki>
hyper_ch2: My guess is something along the line of runlevels.
<Myrl-saki>
Or in the case of systemd, targets.
<Myrl-saki>
But... EW SYSTEMD
<aanderse>
just filed this one: https://github.com/NixOS/nixpkgs/issues/47130 scratching my head as nothing obvious jumps out to me as why not working. would really appreciate anyone able to provide some feedback
romildo has joined #nixos
<samrose>
or maybe the only way to do it right now is nixos-container start <container> even if you've defined your container in configuration.nix?
<Myrl-saki>
aanderse: What does systemctl status say?
orivej has quit [Ping timeout: 252 seconds]
<sphalerite>
infinisil: I miss {^_^} 's issue info :(
<romildo>
Which dependency should I use so that pkg-config can find *mesa-noglu-18.1.7-dev/lib/pkgconfig/gbm.pc ?
<Myrl-saki>
sphalerite: Why was it removed or something?
<Myrl-saki>
Me neither. I'm just basing this on my knowledge of how *NIX processes should work.
<Myrl-saki>
Let me give it another look.
<aanderse>
Myrl-saki: thanks. appreciate it. i'm at a complete loss here
<Myrl-saki>
aanderse: Oh yeah, I see what you mean now.
<Myrl-saki>
aanderse: Seems like it should work, TBH.
<aanderse>
Myrl-saki: that was my thought too :\
<Myrl-saki>
By default, the log file is synchronized every time a syslog message is generated. Use a dash mark (-) as a prefix of the file path you specified to omit syncing:
<manveru>
is the nixos cache down?
<Myrl-saki>
Oh, that's `sync` like disk sync.
<Myrl-saki>
Huh.
<Myrl-saki>
aanderse: I'm just as lost as you are! :P
<Myrl-saki>
aanderse: Sorry, can't help any further.
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
haitlah has joined #nixos
Xyliton has joined #nixos
Xyliton has quit [Client Quit]
<{^_^}>
[nixpkgs] @xeji pushed commit from @7c6f434c to release-18.09 « acl2: 8.1.post.2018.09.20 -> 8.1; turns out there is a release »: https://git.io/fA522
romildo has joined #nixos
<romildo>
How can I easily list the files installed by a given package?
<haitlah>
tree the store ? :D
<betaboon>
clever: any suggestion how to resize the default ebs using nixops post instance-creation ?
<sphalerite>
romildo: find $(nix-build '<nixpkgs>' -A foo)
Thra11 has quit [Ping timeout: 244 seconds]
Fare has quit [Ping timeout: 240 seconds]
jackjennings has joined #nixos
jackjennings has quit [Client Quit]
<gchristensen>
infinisil: {^_^} seems to have gone quiet in #nixos-unregistered
<infinisil>
Yeah sorry for that, the bot is on a short.. vacation
<sphalerite>
betaboon: I'd say at least 4GiB for almost all use cases
<sphalerite>
but 4GiB is already tight
<betaboon>
the 3G-root-storage has worked so far. just ran into it being to tight now when trying to do the 17.09->18.09 upgrade. currently thinking about sizing the root 5G.
<gchristensen>
you can define the size of the root volume at startup with ec2
<gchristensen>
and set boot.growPartition = true to tell the FS to increase its size to fit the root disk on boot.
<betaboon>
gchristensen: by "startup" you mean initial creation of the resource ?
<gchristensen>
yeah
Fare has joined #nixos
<betaboon>
gchristensen: you mean "deployment.ec2.ensInitialRootDiskSize" ?
<betaboon>
*ebs
<gchristensen>
sure, that should do it too
<betaboon>
or are you talking about another option that i missed ?
orivej has quit [Ping timeout: 240 seconds]
<gchristensen>
well I'm more talking about how AWS lets you do it
<betaboon>
ah ok :D
<betaboon>
or would it be a better idea to attach a different ebs-volume just for /nix ?
<{^_^}>
[nixpkgs] @basvandijk pushed to master « haskellPackages.mandrill: fix build for ghc843 »: https://git.io/fA56C
sanscoeur has joined #nixos
ixxie has quit []
<{^_^}>
[nixpkgs] @basvandijk pushed to release-18.09 « haskellPackages.mandrill: fix build for ghc843 »: https://git.io/fA56r
<betaboon>
I'm glad i tried the 17.09->18.09 upgrade path on our development-deployment first xD
jtojnar_ has joined #nixos
<betaboon>
gchristensen: i just saw your nixops-PR about plugin-ifying nixops earlier today. what happened to the teraform approach? would that solve all the problems ? oO
<catern>
is there something like steam-run for arbitrary random proprietary packages?
<catern>
i.e., I want to run Sublime Merge on NixOS, what do I do?
<symphorien>
steam-run may work
<symphorien>
iirc it is not really steam specific
<catern>
yeah, I would try steam-run
<catern>
but I guess my real question is: How do I run Sublime Merge on NixOS? Is the answer steam-run?
<tilpner>
catern - steam-run is using buildFHSUserEnv, which can be used to build arbitrary FHS environments
<tilpner>
catern - You could define an FHS env capable of running just Sublime
<tilpner>
Or look at the sublime derivation, maybe you can adapt it
<sphalerite>
betaboon: oh hm I'm not sure this would work for ec2 since the filesystem needs to contain the initial contents of the store
<catern>
./sublime_merge: relocation error: /usr/lib/libc.so.6: symbol _dl_exception_create, version GLIBC_PRIVATE not defined in file ld-linux-x86-64.so.2 with link time reference
<catern>
that's what I get from steam-run ./sublime_merge
<sphalerite>
betaboon: I think you need to keep the nix store in the root fs :/
<catern>
what does this crazy error message even mean?
<symphorien>
probably that you have to use patchelf :)
<sphalerite>
betaboon: that, or find a way to also convince nixops to image the nix store volume before starting the machine
<betaboon>
sphalerite: yeah i feel so. i was hoping that by moving it onto a different ebs-volume i can prevent those problems i had today. being able to just resize it later-on if a problem occures
<sphalerite>
betaboon: the root one can't be resized?
<catern>
symphorien: yeah but, what do I do with patchelf?
<betaboon>
sphalerite: seems like it can, but just from the aws-console. haven't tried
<exarkun1>
/proc/pid/status still says the process only belongs to one group.
<sphalerite>
for some reason /run/keys is root:root, not root:keys, for me..?
<gchristensen>
I dunno, it works for me X)
<exarkun1>
dunno about that. I don't think I did anything special to configure the ownership/permissions I have.
<sphalerite>
is it root:keys for you?
<exarkun1>
the /etc/group file was rewritten to say tor belongs to keys but `systemctl restart tor` doesn't give me a tor process in the `keys` group... Possibly tor is doing something to throw out the extra groups... for security or whatever
<gchristensen>
and you did restart the process?
<exarkun1>
that's what `systemctl restart tor`'s job is right? I can verify the process went away and a new one took its place (several times now).
<gchristensen>
something is fishy. maybe try a reboot. something is holding on to that old group info for some reason.
<emily>
maybe you need to get systemd to reload?
<realrokka>
someone else is using electrum wallet and the build fails ? https://pastebin.com/q3wrEKse :( should I open up an issue on gh or something?
<exarkun1>
Okay ... Tor calls `setgroups(1, &pw->pw_gid)`
<exarkun1>
So whatever groups you started with, you'll only have the primary gid of the user it selects
abueide has joined #nixos
Sigma has quit [Ping timeout: 264 seconds]
<exarkun1>
Can I get past "The unique option `users.users.tor.group' is defined multiple times," somehow? Or does it come down to patching tor.nix?
<gchristensen>
lib.mkForce maybe :/ but aoeunthaoeunth probably better to patch it out
<exarkun1>
I tried the `filesystem."/run/keys"` approach. It changes the mode in /etc/fstab but it doesn't actually change the permissions of the mounted filesystem.
<d1rewolf>
is there a relatively easy way to use synergy (mouse/keyboard sharing) commercial version with nixos?
* exarkun1
reads about mkForce
<exarkun1>
messing with run/keys via `fileSystems` does seem to have screwed up something and now I can't deploy keys to the node anymore 🙂
<MarcWeber>
PHP -> rewrite -> causing trouble to me. http://dpaste.com/1QXQ1CY why does cfg have sha nad version only? Shouldn't cfg have all optional keys with default value thus also include tidySupport?
<{^_^}>
[nixpkgs] @ryantm pushed 0 commits to revert-47093-auto-update/burp: https://git.io/fA5FA
<exarkun1>
rebooted, set group to "keys" w/ mkForce, success
<exarkun1>
if I were to fix this upstream, I wonder what the fix would be.
<exarkun1>
a nixos option to _not_ set `User tor` in the tor config file might fix it... I think that would prevent Tor from messing with its uid/gid/groups at all. Then I could just tell systemd how I want those configured.
<exarkun1>
I don't fully grasp the motivation for tor's setgroups(1, ...) call. Maximally limited permissions, I guess? Defense in depth?
<d1rewolf>
if I "nix-shell -p python27Packages.dbus-python", then run python, and then import dbus, it works. however, if I put python27Packages.dbus-python in my configuration.nix and rebuild, even though rebuild works, "import dbus" fails. Any ideas why?
<exarkun1>
d1rewolf: aren't you supposed to use `python27.withPackages`?
<d1rewolf>
exarkun1: I think you're right. The nix wiki isn't as full as an example...it cuts off. but now that you mention it, I think there's something in the nixpkgs manual. thanks ;)
hotfuzz_ is now known as hotfuzz
<symphorien>
,library d1rewolf
<symphorien>
hum the bot is down
<symphorien>
in short, putting libraries/modules in systemPackages is unlikely to work, by design
<symphorien>
the bad value must have the right length (52 chars)
<symphorien>
infinisil++
<{^_^}>
infinisil's karma got increased to 26
<alex``>
symphorien: I never do that, but that information is where in the output when building?
<symphorien>
you won't miss it. it is not hidden
<alex``>
Cool :p
<Mic92>
gchristensen: basically anything that will do a inode lookup (checking if a file exists or not) will be lineary decrease the performance. It will not affect the performance of open files or inodes that were previously accessed. So you probably don't want to compile stuff or unpack tarballs on the overlayfs if you have too many layers.
<alex``>
Is there a nix command to generate a basic skeleton for nix shell for example, shell.nix?
<Mic92>
gchristensen: I personally have not used overlayfs too much, so not in a production environment so I cannot give you raw numbers. I wrote the zfs docker driver and end up using that. I recently benchmarked a fuse intensivly and read a good part of the overlay kernel implementation. you can try a benchmark called compilebench to get the worst-case performance.
<Mic92>
alex``: it also generates the direnv part, which you may or not may want to use.
<alex``>
symphorien: What is the difference?
<dhess>
question about something that's been bothering me for awhile: in my experience, when a NixOS service has keys.target in "wants" and "after", it will always wait for that target to fire if I'm using NixOps to deploy that host.
<symphorien>
alex``: it is equivalent but shorter :)
<dhess>
so then things like NSD get stuck if I've rebooted the host but not yet had a chance to `send-keys` to it
<Mic92>
symphorien: is it really the same?
<alex``>
symphorien: It is also used in nix build? (mkShell)?
<dhess>
but I had thought that "wants" means that it only waits for some timeout, and then fires anyway, vs. "requires" which will wait indefinitely.
<dhess>
Is that not the case?
<alex``>
default.nix
<Mic92>
symphorien: ah I see
<Mic92>
alex``: as I ready this is not meant for nix-build
<Mic92>
*I read
<Mic92>
it explicitly disable this part
work_ has quit [Quit: Connection closed for inactivity]
<alex``>
I saw that configuration.nix could handle environment variables, is it a good idea to move all env from my shell to it?
<symphorien>
it is a matter of taste
<Mic92>
alex``: Personnally I try to put as little as possible there, because then I can be sure that my default.nix is complete. you may also see conflicts between projects.
<Mic92>
sometimes build systems also pick up the wrong stuff that I did not intend to use.
<alex``>
What is the purpose of "name" I see the the FAQ, for nix-shell
<alex``>
so for nix-shell, stdenv.mkDerivation will yield if I do not give him "name"?
<alex``>
even if it not use it
<symphorien>
no mkDerivaton requires a name
<Mic92>
yes and nix-shell still evaluates it
<alex``>
Got it
<alex``>
symphorien: Thanks for pointing me the source of mkshell
<jasom>
So, there's a program I use that isn't in nix; I wrote an expression for it and currently use it in an overlay. I don't currently have the time to commit to maintaining it. Should I make a PR?
<alex``>
where mkShell is defined?
sanscoeur has quit [Ping timeout: 260 seconds]
<alex``>
I don’t see mkShell reference in the source
<Mic92>
alex``: nix edit nixpkgs.mkShell
<alex``>
I was thinking it were mkshell
<alex``>
not mkShell
sanscoeu_ has quit [Ping timeout: 245 seconds]
<alex``>
as it’s lowercase in the path mkshell/default.nix
<alex``>
> error: cannot auto-call a function that has an argument without a default value ('stdenv')
<{^_^}>
error: syntax error, unexpected $undefined, at (string):195:82
<elvishjerricco>
alex``: There is not, but you can kinda simulate it by making your file look like `{ pkgs ? import <nixpkgs> {}; stdenv ? pkgs.stdenv, ... }: ...`. That'll work with both `callPackage` and a plain `nix-build`
abueide has quit [Remote host closed the connection]
abueide has joined #nixos
hamishmack has quit [Read error: Connection reset by peer]
funkmybios has joined #nixos
thc202 has quit [Ping timeout: 260 seconds]
Fare has quit [Ping timeout: 246 seconds]
funkmybios has quit [Ping timeout: 256 seconds]
Anton-Latukha has quit [Quit: Leaving.]
erasmas has quit [Quit: leaving]
NightTrain has joined #nixos
booglewoogle has left #nixos ["Leaving"]
<catern>
heh
<catern>
I'm currently working on replacing a proprietary package manager/build system/deployment system at $DAYJOB with Nix, and someone had an idea which I think is pretty good, if totally weird from a Nix point of view
<catern>
to support incremental rollout, start off with only providing Nix dependencies at build time
<catern>
then you don't have to deploy Nix to all your production hosts, just to your build servers and developer machines
<catern>
now this obviously severely limits what you can use through Nix, but it at least lets you get, e.g., patchelf and gcc through Nix
<clever>
catern: the ELF files nix will produce refer to /nix/store/ and will break when copied to another machine manually
<catern>
(I think, anyway, about gcc?)
<catern>
clever: no, not if you don't take any library dependencies
brejoc has quit [Ping timeout: 252 seconds]
<clever>
even the ld.so in dynamic elf files is in the nix store
<catern>
sorry, to be clear, these builds won't take place in a Nix sandbox, they'll be in nix-shell
kyren has quit [Ping timeout: 240 seconds]
<clever>
nix-shell products still refer to the store in the same way
<clever>
but lack GC roots, so they can even break locally if you garbage collect
<catern>
clever: okay but you can use the patchelf from Nix to patch ELF files and it doesn't create any dependency on the Nix store in the ELF file
<clever>
yeah, you can use patchelf to un-nix the files
<catern>
I understand what you're saying
<clever>
the nixos initrd is even doing a form of that
<clever>
rather then depend on all of glibc in the initrd, it copies libc.so to a new derivation
<catern>
But what I'm saying is that if you avoid getting your runtime deps from Nix, and instead get only deps from Nix that are fine to disappear at runtime
kyren has joined #nixos
<clever>
and runs patchelf on all files, to point them to that new derivation
<catern>
Then everything is fine
<clever>
which discards a lot of the trash not needed in an initrd
<catern>
Don't you agree? (it's kind of axiomatically true)
<clever>
the nix provided gcc has been modified to not look in /usr/lib and friends
<clever>
so it wont use non-nix libraries
<catern>
Sure but you can override that with command line arguments, right?
<clever>
probably
<catern>
So this will completely work
<catern>
And it's very convenient for an incremental migration from this crappy system
mizu_no_oto has joined #nixos
<catern>
(which currently provides patchelf and gcc itself, but will now use patchelf and gcc from Nix instead)
DigitalKiwi has quit [Ping timeout: 246 seconds]
lassulus has quit [Ping timeout: 246 seconds]
<catern>
like essentially what I'm talking about is making a nix-shell that has gcc and patchelf on the PATH, and then running the completely normal build and deploy process from there
<catern>
replacing the gcc and patchelf that the build system itself usually provides
echel0n_ has quit [Quit: WeeChat 2.0]
<catern>
Does that make sense? Doesn't it sound like a good incremental step, if an unintuitive one?
georges-duperon has quit [Ping timeout: 252 seconds]
<clever>
catern: it may work
jluttine has quit [Ping timeout: 252 seconds]
<samueldr>
catern: make sure to take plenty of notes, and transform that into an interesting blog posts series or a talk :)
lassulus has joined #nixos
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
jluttine has joined #nixos
<{^_^}>
[nixpkgs] @orivej-nixos pushed 5 commits to master: https://git.io/fAdLE
jperras has joined #nixos
jedahan_ has joined #nixos
nullp0tr has joined #nixos
brodul has quit [Ping timeout: 240 seconds]
jtojnar has quit [Remote host closed the connection]
Fare has joined #nixos
jtojnar has joined #nixos
mayhewluke has quit [Ping timeout: 260 seconds]
<avn>
catern: is a lot of variants to compose prodiction system from nix store (on build host) -- look how dockerTools or other similiar stuff works.
jedahan has quit [Ping timeout: 252 seconds]
<infinisil>
clever: Do you happen to know how to get a debug build of ZFS?
<infinisil>
My pool is hosed for some inexplicable reason..
mayhewluke has joined #nixos
<clever>
infinisil: nope, but i have used zdb a bit, one min
<clever>
infinisil: zdb -l /dev/mapper/root
<clever>
infinisil: that shows metadata about a vdev, even if you dont import it
<clever>
ive even used it to inspect vdev's i have entirely removed from the pool
<infinisil>
clever: Yeah, checked that out already: http://ix.io/1nee
<infinisil>
Don't think there's anything wrong with that