<clever>
Nobabs27: everything looks fine, what OS is the host that virtualbox is running on?
<Nobabs27>
clever: Arch
<clever>
Nobabs27: in the virtualbox settings, you can configure COM1 to tcp mode, and set it up to connect to an existing socket, let me find some exact values
slabity has joined #nixos
<Nobabs27>
clever: is that where it out puts to the terminal or something?
<clever>
Nobabs27: the guest can be configured to print more debug info to the serial ports
<clever>
and then you gain the ability to scroll
<Nobabs27>
ah
<clever>
socat tcp-listen:4321,fork stdin
<clever>
if you run this on arch, it will listen on port 4321
<Nobabs27>
ok
<clever>
you can then configure virtualbox to connect to 127.0.0.1:4321
Rotaerk has joined #nixos
<clever>
Nobabs27: then add boot.kernelParams = [ "console=ttyS0,115200" ]; to the configuration.nix and rebuild
mudri has quit [(Ping timeout: 240 seconds)]
ebzzry has joined #nixos
<clever>
that tells linux to output everything to the serial port, instead of the virtual gpu
<Nobabs27>
clever: ah
<clever>
so you should now get the entire error, and have the ability to copy/paste
<Nobabs27>
clever: ok lemme configure the port
<clever>
and if your lucky enough to have a hardware serial port, you can also use the above on real machines for debugging things remotely
<Nobabs27>
clever: do I leave path/address blank?
<clever>
path/address has to be set to 127.0.0.1:4321
<clever>
and the socat has to be running when you launch the guest
<Nobabs27>
ye
<Nobabs27>
alright now I just gotta go to the config
mellowmaroon has quit [(Quit: Leaving)]
ibrahims has joined #nixos
<Nobabs27>
clever: rebuild switch ?
<clever>
i would use "nixos-rebuild boot"
<Nobabs27>
kk
<clever>
the change requires a reboot to apply, so switch doesnt really help any
<Nobabs27>
maybe that was my whole problem
<clever>
probably isnt
<Nobabs27>
even though the kernel was upgraded?
arianvp2_ has quit [(Quit: arianvp2_)]
<clever>
rebuild switch will restart services and update config files, but the kernel change will wait until you reboot
<Nobabs27>
ik but how will grub know to use the new kernel ?
<Nobabs27>
esp if I didnt rebuild boot?
<clever>
every time you do switch or boot, nixos updates the grub.cfg
<Nobabs27>
k then
<clever>
nixos-rebuild test will activate the changes without making them the default
<clever>
nixos-rebuild boot will make it the default in grub
<clever>
switch is just test+boot
<Nobabs27>
alright restarting with that new stuff in the config
<clever>
Nobabs27: inside the guest, can you run "nix-build '<nixpkgs/nixos>' -A config.system.build.bootStage1" and then gist the file that result is pointing to
fnljk has quit [(Quit: ?)]
aneeshusa has quit [(Quit: WeeChat 1.8)]
<Nobabs27>
clever: it just spams the screen ?
lukec has quit [(Quit: WeeChat 1.4)]
<clever>
it should build a copy of stage1, and then give a storepath at the end
<pop>
What's the best way to define a system group (in this case `docker`) in the nixos config?
ebzzry has quit [(Quit: WeeChat 1.7)]
<simpson>
nixops question: Is there a recipe for forwarding ports from production to my working machine, or do I need to hack it up with $(nixops ssh)?
<simpson>
iqubic: The Nix store is where Nix keeps packages and stuff that it's built.
<nixy>
iqubic: If you haven't read the Nix manual I would strongly recommend it before going all in on Nix. Nix brings a lot of benefits, but you kind of have to understand how it works to take advantage of it
<iqubic>
Do you have a link to the NixOS Manual
indi_ has quit [(Remote host closed the connection)]
<simpson>
Read up a few lines.
acarrico has quit [(Ping timeout: 255 seconds)]
indi_ has joined #nixos
<dave___>
I was thinking of making a gcp functions backend implementation for nixops as a learning aid. There's a PR for vultr I was intending to use as a guide: https://github.com/NixOS/nixops/pull/634/files but is there any other documentation for how to implement MachineDefinitions/MachineStates, etc?
<dave___>
(Note: I'm coming at this from the angle of nixops vs terraform. If my comparing nixops to terraform is wrong feel free to let me know now before I hurt myself)
<simpson>
dave___: No, documentation is sparse. :C
<dave___>
Is the terraform comparison apt? The documentation for terraform plugins is quite nice and they structure their plugins as provider/resource pairs, with Create/Read/Update/Destroy functions. It looks like nixops MachineStates only have create/destroy methods
<simpson>
You're not wrong, but nixops tries to get per-machine configuration completely into Nix, so there's not really any update mechanism outside of that.
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
<dave___>
oh, you mean the read/update is done via nix, nixops only creates/destroys machines?
<simpson>
Pretty much.
<dave___>
is that why there's a create step followed by a deploy step, whereas terraform has just an apply step?
<simpson>
That sounds about right.
<dave___>
hmm, that makes sense
indi_ has quit [(Ping timeout: 260 seconds)]
<iqubic>
So if learn nix and really like it, should I switch to NixOS?
<dave___>
Pardon my poor use of nix terminology, total nix noob, but the sqlite state db and the nix specification passed to nixops would contain enough information in theory to wrap everything up into a single "apply" command? Would there be reasons besides code effort to not do that?
<dave___>
iqubic: I get the feeling that if you learn nix and really like it you wont bother asking that question :)
indi_ has joined #nixos
<simpson>
dave___: Well, yes, but we call it $(nixops deploy).
<simpson>
iqubic: If you like. I'm still not comfortable recommending that you do fancy things with your only machine.
jgertm has quit [(Ping timeout: 246 seconds)]
<dave___>
o.O? Is that mentioned in the documentation somewhere? I don't recall reading about that
<dave___>
specifically the $( ) notation
<dave___>
I'm under the impression that "nixops deploy" doesn't automatically run create/destroy.
<simpson>
$() is just how I quote shell stuff.
<simpson>
Deploying definitely creates machines.
<dave___>
okay then, my mistake
<iqubic>
simpson: I'm fine messing with my machine.
<iqubic>
I think that switching Distros should be easy.
orivej has quit [(Ping timeout: 255 seconds)]
pxc has joined #nixos
jgertm has joined #nixos
pxc has quit [(Quit: WeeChat 1.8)]
fnljk has joined #nixos
<slabity>
Given a package name, how can I get its path in the nix store in bash?
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
ericsagnes has quit [(Ping timeout: 246 seconds)]
<simpson>
slabity: nix-build echos the store path on success, so you could use that.
indi_ has quit [(Ping timeout: 240 seconds)]
vaibhavsagar_ has joined #nixos
<iqubic>
So, If I am going to use Nix as a single user, what should I know?
<noffle>
hello. I installed nix on arch linux, and in running 'nix-env -qa' I notice that packages for platforms = [ "i686-linux" ] don't seem to be listed
vandenoever has quit [(Ping timeout: 255 seconds)]
<noffle>
is this intentional, or do I need to flip a bit somewhere in a config?
<noffle>
I haven't done any configuration yet other than installing nix itself
<nixy>
i686 would be 32bit Linux wouldn't it? Are you running a 64bit system?
<noffle>
nixy: "x86_64-linux" doesn't show either
<noffle>
sorry, bad example
<nixy>
It is possible that the package is under an attribute set and wouldn't show up under `nix-env -qa` I know most of the haskell infrastructure won't show up
<Ralith>
slabity: you can write a string like "${pkgs.powerlevel9k}/share/whatever"
<noffle>
I'm trying to install skype or skypeforlinux in particular
<clever>
noffle: nix-env -iA nixos.skype
<clever>
or nixpkgs if your not on nixos
<Ralith>
there's the unfree bitflip
vaibhavsagar has joined #nixos
<noffle>
clever: error: attribute ‘nixos’ in selection path ‘nixos.skype’ not found
<noffle>
oof, bad quotes
iqubic has quit [(Remote host closed the connection)]
<clever>
noffle: nix-env -iA nixpkgs.skype
<noffle>
clever: that did it; now I'm getting the unfree license wall that Ralith mentioned
<clever>
yep, it will tell you what to put into config.nix, and where config.nix should be
<noffle>
ah, so it does. time to do some reading comprehension
<noffle>
thank you
<noffle>
clever: is the "nixpkgs" scope necessary for me to add for any nixos package?
<noffle>
I used to run nixos and don't remember needing that (I think!)
justbeingglad has joined #nixos
justbeingglad has left #nixos []
<clever>
noffle: that tells nix-env which channel to look in, you can see the available names in nix-channel --list
* clever
heads off to bed
<noffle>
clever: thank you for the help!
<clever>
yep
ChongLi has quit [(Ping timeout: 255 seconds)]
ChongLi has joined #nixos
zeus_ has joined #nixos
<ison111>
slabity: I'm not sure about referring to your own packages, but it's usually something like ${pkgs.PACKAGE_NAME}
marsel has joined #nixos
<slabity>
ison111: Yea, I'm trying to figure out how to refer to a package I made myself
<slabity>
Ralith: It tells me that attribute does not exist. Are custom packages part of 'pkgs'?
<Ralith>
they are if you put them there
<Ralith>
they aren't if you don't
<Ralith>
you can handle your custom packages however you like
<Ralith>
simply sticking a .nix file somewhere does nothing at all
<Ralith>
(excepting magic files like /etc/nixos/configuration.nix)
<slabity>
Ralith: Then what do I do after I make the file?
<Ralith>
if the expression in your file looks like a typical package, then you'll probably want to do the standard `pkgs.callPackage /path/too/foo.nix {}` dance
jbgi has joined #nixos
indi_ has joined #nixos
<Ralith>
which you can stick in your overrides, or in a let binding, or wherever else
stallion has quit [(Ping timeout: 255 seconds)]
indi_ has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vHbZE
<NixOS_GitHub>
nixpkgs/master 4ef8313 Vladimír Čunát: aiccu: remove leftover from 3dcecf09fc1c
NixOS_GitHub has left #nixos []
<slabity>
Ralith: I am at a loss. I have "let powerlevel9k = pkgs.callPackage ./powerlevel9k.nix {}; in {...}", but when I try to use "${pkgs.powerlevel9k}" it says the attribute is missing.
stallion has joined #nixos
<Ralith>
slabity: you introduced a variable named `powerlevel9k`, you haven't touched `pkgs`
<Ralith>
if it's a local variable, just reference it directly
<Ralith>
"${powerlevel9k}/foo"
Mercuria1Alchemi has joined #nixos
<dave___>
how is data programmatically retrieved about created resources? "nixops info" will display IP addresses, but what about arbitrary data such as the name/id of an s3 bucket, which isn't an IP address? Terraform has a tfstate json file that can be queried but I'm getting the feeling that nixops is overly specialized to deal mostly with just servers, not arbitrary resources
<slabity>
Ralith: That gives me "attempt to call something which is not a function but a set"
<Ralith>
can't help you with that unless you share full code
<Ralith>
slabity: your powerlevel9k.nix file is not a nix package
<Ralith>
it returns a derivation, but you need to return a function that, when called with suitable arguments, produces a derivation.
<slabity>
Ralith: That's what I was told to do... What is it then? Just an expression?
marusich has quit [(Ping timeout: 240 seconds)]
<Ralith>
for example, replace line 1 with `{ fetchFromGitHub }:`
<Ralith>
er
<Ralith>
`{ stdenv, fetchFromGitHub }:`
marsel has quit [(Ping timeout: 240 seconds)]
<Ralith>
also, I don't know what "packages.nix" is; that looks like a configuration.nix to me
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edwtjo pushed 9 new commits to release-17.03: https://git.io/vHbnO
<NixOS_GitHub>
nixpkgs/release-17.03 d4b6398 Jonathan Rudenberg: libhdhomerun: init at revision 1efbcb2...
<NixOS_GitHub>
nixpkgs/release-17.03 c864dff Jonathan Rudenberg: kodi: init pvr-hdhomerun plugin at 2.4.7...
<NixOS_GitHub>
nixpkgs/release-17.03 3c66850 Edward Tjörnhammar: kodiPlugins.exodus: 3.0.5 -> 3.1.13...
NixOS_GitHub has left #nixos []
<Ralith>
you should delete line 42 in it
<slabity>
Ralith: Yea, it's imported from configuration.nix. But your change seems to have fixed it
<Ralith>
it is at best doing nothing
ericsagnes has quit [(Ping timeout: 255 seconds)]
moet has joined #nixos
<Ralith>
(unless you genuinely need this package in your system environment for reasons other than setting the theme, in which case you should refer to the let-bound variable containing the package, rather than mucking about with import)
<Ralith>
(but I doubt that's the case)
<Ralith>
I'd also encourage you not to put user applications like firefox, steam, etc. in your global system environment
jbgi has quit [(Ping timeout: 240 seconds)]
<slabity>
Ralith: Why's that?
<Ralith>
if you install them in your user environment instead, you won't need to rebuild your system (or indeed acquire root privs at all) to update them
<Ralith>
and it will be considerably easier to update them piecemeal independent of the rest of your system should you want to
<slabity>
Well I just realized that the 'theme' option is looking for a relative path instead of absolute.
<Ralith>
the author probably intended you to copy your theme file into /etc/nixos manually
<slabity>
I mean it's looking for it relative to the ohMyZsh path
* Ralith
shrug
<Ralith>
maybe it's broken, I don't have any experience with that module in particular
<slabity>
Ehh, I think it'll work if I just do something disgusting like "../../../../${powerlevel9k}/powerlevel9k"
<Ralith>
that is pretty disgusting
<Ralith>
I would try to find the author and ping them, if nothing sane works
<Ralith>
and/or fix it and make a PR
<Ralith>
hm, maybe ohMyZsh comes with a variety of themes and they didn't think of people trying to use their own
<slabity>
Ralith: They come with a bunch of basic built-in ones, but there's a lot that aren't built-in
<Ralith>
that's probably the deal then
<Ralith>
I imagine an extension to the module to support third-party themes would be well received
<slabity>
oh-my-zsh expects themes to be installed in $OH_MY_ZSH/themes or something like that then
<Ralith>
I was talking about the module, not the upstream package, to be clear
<slabity>
It would be nice to just say, "theme = 'git://github.com/whatever/themething'"
<slabity>
I'll look into cleaning that
<Ralith>
the more idiomatic thing would be to take a derivation
<Ralith>
since that exposes full control over how it's obtained and packaged
<Ralith>
and you wouldn't want to break things for people who are already using the built-in themes
marusich has joined #nixos
moet has quit [(Ping timeout: 240 seconds)]
georges-duperon has joined #nixos
ericsagnes has joined #nixos
<sphalerite>
Or you could not use omz
ericsagnes has quit [(Ping timeout: 268 seconds)]
<slabity>
sphalerite: That doesn't sound like a good solution.
<sphalerite>
slabity: why not?
<sphalerite>
You can download a theme and source it directly
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edwtjo pushed 1 new commit to master: https://git.io/vHbcX
<NixOS_GitHub>
nixpkgs/master 0f9a228 Edward Tjörnhammar: ezquake: init at 3.0.1
NixOS_GitHub has left #nixos []
<sphalerite>
The only other features it has is wrapping some zsh built-in configuration functionality which you can use directly, and hacky update functionality
<sphalerite>
AFAIU
<slabity>
sphalerite: It's more of a collection of plugins that you can enable and disable. I don't really want to go through each plugin I want and install them separately
<sphalerite>
slabity: they're all just little snippets that would be better off managed by a proper package manager
<sphalerite>
like nix ;)
<slabity>
I don't disagree
zeus_ has quit [(Remote host closed the connection)]
ericsagnes has joined #nixos
bennofs has joined #nixos
slabity has quit [(Quit: leaving)]
mbrock has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 6 new commits to master: https://git.io/vHbCK
<mbrock>
I wonder if it's possible to produce a Haskell executable package that doesn't incur a Nix dependency for all its Hackage dependencies...
thc202 has joined #nixos
<mbrock>
like, it would have all its Hackage dependencies statically linked, and then installing it via Nix would require much fewer binary cache downloads
indi_ has joined #nixos
<sphalerite>
joachifm: any particular reason you reopened my PR for updating rigs of rods?
<sphalerite>
oh right I misread
<sphalerite>
nvm
mudri has quit [(Ping timeout: 255 seconds)]
indi_ has quit [(Ping timeout: 255 seconds)]
FRidh has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] pSub pushed 1 new commit to master: https://git.io/vHbBx
<NixOS_GitHub>
nixpkgs/master fe04169 Pascal Wittmann: git-standup: init at 2.1.8
NixOS_GitHub has left #nixos []
Itkovian has joined #nixos
zeus_ has joined #nixos
therp has joined #nixos
zeus_ has quit [(Ping timeout: 246 seconds)]
oida has joined #nixos
oida_ has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/vHb0T
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/vHb0E
<NixOS_GitHub>
nixpkgs/master 72f530b Domen Kožar: Fix #26441 second try: avoid infinite recursion
NixOS_GitHub has left #nixos []
kkourt has joined #nixos
<boj>
i'm trying to build a stack project with "nix-shell -p stack -p ncurses -p haskell.compiler.ghc802" and then "stack build". apparently it has a haskell dependency on terminfo, but fails to find ncurses headers. have looked at several related nix/terminfo/stack issues without any luck. anyone have any suggestions they could offer?
filterfish has quit [(Ping timeout: 240 seconds)]
<mbrock>
aha, I got rid of most of the runtime derivation dependencies for my Haskell package with postInstall = "rm -rf $out/{lib,share}";
jacob_ has joined #nixos
jacob_ is now known as Guest10403
mudri has joined #nixos
<MagneticDuck>
nix-store --register-validity --reregister seems to be stalling
<MagneticDuck>
I enter the store path (including /nix/store) into stdin, followed by \n0\n, and then terminate input with ctrl-D
<MagneticDuck>
if I don't have sufficient rights, I get an error when it tries to change the mode of the file to 1000
<MagneticDuck>
... oh wait, disregard this, it just takes a while
<tilpner>
Which will output "/nix/store/8ll9div7y2azpm3hsfs59lz57pyqx9y6-st.drv"
<tilpner>
But I want 8ll9div7y2azpm3hsfs59lz57pyqx9y6 without string manipulation
<tilpner>
And I can't use .drvAttrs.outputHash, because that's apparently not set in this case
m0rphism has quit [(Quit: WeeChat 1.8)]
<sphalerite>
But even string manipulation wouldn't get you the hash of what gets fetched into the store
<sphalerite>
Why do you even want it?
<tilpner>
Yes, I see that now. I was mistaken about the meaning of that hash, assuming it wasn't constructed from the rev but the hash of the repo to automatically maintain unique paths in the store
<tilpner>
Wait, what is it constructed from?
<sphalerite>
The derivation's hash? The derivation's contents
<tilpner>
Is 8ll9div7y2azpm3hsfs59lz57pyqx9y6 the derivation's hash?
pie_ has joined #nixos
ixxie has joined #nixos
<tilpner>
This may be an XY problem. I just want a suffciently unique name to pass to runCommand, from any derivation src that gets passed in
<sphalerite>
yes, 8ll9... is the derivation's hash. Why do you want a unique name for the runCommand?
<tilpner>
This runCommand would generate something based on that input derivation, and I want to make sure it doesn't collide with anything, so a name like rustRegistry-8ll9div7y2azpm3hsfs59lz57pyqx9y6 would be unique (the input is a rustRegistry src)
orivej has joined #nixos
<sphalerite>
Why do you not want the name to collide with anything? The fact that it gets its own output path with a hash in it is sufficient for that
<gchristensen>
tilpner: are you afraid of it colliding in the store?
<tilpner>
Huh, another misunderstanding then. What's the meaning of the name passed to runCommand then? And where do I find a list of all this, so that I don't bother you?
<gchristensen>
tilpner: /nix/store/bxr5lwdhhgfh1p28mh3w7kyc0spk7niz-bash-4.4-p5/bin/bash <- the name is bash-4.4-p5
filterfish has quit [(Ping timeout: 255 seconds)]
<gchristensen>
the bxr5lwdhhgfh1p28mh3w7kyc0spk7niz is the crypto sum of all of the inputs to bash
<tilpner>
I see. My worry was completely unfounded then, thank you
<gchristensen>
:) you're welcome
<gchristensen>
nixpkgs would be madness otherwise hehe
arianvp2_ has quit [(Quit: arianvp2_)]
<sphalerite>
tilpner: did you sort out the syntax error issue yourself?
<sphalerite>
tilpner: ? is only valid in a set-matching pattern, which is in a function definition
<sphalerite>
AFAIU
<gchristensen>
sphalerite: there is also:
<gchristensen>
nix-repl> f = { bar = "hi"; }
<gchristensen>
nix-repl> f ? foo
<gchristensen>
false
<gchristensen>
nix-repl> f ? bar
<tilpner>
sphalerite - null ? y returns false in nix-repl. What does it do then?
<gchristensen>
true
stallion has quit [(Ping timeout: 240 seconds)]
<sphalerite>
oooh yes, forgot about that
<tilpner>
Oh...
<sphalerite>
So just an infix version of builtins.hasAttr
<gchristensen>
? seems to always return false on non-sets
vaibhavsagar has quit [(Ping timeout: 255 seconds)]
<tilpner>
Well, that's not confusing at all :P
<gchristensen>
well it is asking if f has an attr named bar, and strings do not have an attr named bar
<gchristensen>
nor do nulls or bools
magnetophon has joined #nixos
<tilpner>
True, I'm just used to a type error instead
<magnetophon>
I updated the jackaudio pkg to 1.9.11-RC1. Does it make sense to do a PR, or should I wait for 1.9.11 proper? I've been using and updating the 'unstable' version for months now, and it has always been rock solid.
filterfish_ has joined #nixos
tmaekawa has joined #nixos
filterfish has quit [(Read error: Connection reset by peer)]
<gchristensen>
oh, 3 letter second level domain is premium
<ikwildrpepper>
2155 euro/year here
<pikajude>
i've gotten adverts for free
<gchristensen>
lol pikajude
<jophish>
hahahah
<jophish>
hmm, that addvert was a lie
<jophish>
7000$ from that site
<ikwildrpepper>
lame
<jophish>
perhaps there's a premium on three letter subdomains
<gchristensen>
that is what gandi says
<gchristensen>
nix.tools is cheap
<jophish>
not too bad
<jophish>
The suggestion after rejecting nix.store was nix.sucks
<gchristensen>
also expensive I think
<jophish>
can't imagine anyone buying that too soon
<gchristensen>
someone already got nix.sh
erictapen has joined #nixos
<nixy>
Ahhh vanity domains
<nixy>
What fun
ericsagnes has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fare opened pull request #26592: gerbil: 0.10 -> 0.11 (master...nixos-unstable) https://git.io/vHbP0
NixOS_GitHub has left #nixos []
<nixy>
Somebody already has nix.wang :(
<gchristensen>
hah! ".berlin" TLDs have a requirement that you have a contact _in_ berlin and they call out that if you don't have one, they will very gladly provide one for you
<gchristensen>
(they == Gandi)
reinzelmann has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] qknight pushed 1 new commit to master: https://git.io/vHbXq
<NixOS_GitHub>
nixpkgs/master c0df448 Ekaterina Vaartis: apache-httpd: fix mod_perl by refering to apacheHttpdPackages (#26579)
NixOS_GitHub has left #nixos []
ixxie_ has quit [(Quit: Lost terminal)]
iyzsong has joined #nixos
arianvp2_ has joined #nixos
katyucha has joined #nixos
tmaekawa has quit [(Ping timeout: 255 seconds)]
ericsagnes has quit [(Ping timeout: 268 seconds)]
magnetophon1 has quit [(Ping timeout: 255 seconds)]
filterfish_ has quit [(Ping timeout: 240 seconds)]
adamt has quit [(Ping timeout: 255 seconds)]
mudri has quit [(Ping timeout: 246 seconds)]
<Unode>
hi guys, few questions about nix and how it organizes the store.
<Unode>
For a while I've had a nix setup in a non nixos machine and a non standard /share/nix location that works via NFS.
<Unode>
This works 99% of the time but every now and then (with specific packages) things fail.
indi_ has quit [(Remote host closed the connection)]
<Unode>
one of the issues I've encountered is that the a library is being used when nix tried to remove/replace it.
indi_ has joined #nixos
<Unode>
Since this sits on NFS, the NFS system often leaves behind .nfs*** files to prevent resources from losing references to deleted 'shared' objects.
<gchristensen>
I imagine it might be b/c live gc root paths on nfs-mounted stores aren't possible to validate
<Unode>
this would all be great but for NFS but for nix this often results in "cannot remove directory" which then fails the compilation.
<Unode>
(or perhaps the install).
<Unode>
... since compilation happens on a local folder.
<Unode>
my question is, is there any way to workaround this issue?
<LnL>
you can use NIX_OTHER_STORES for that
<Unode>
LnL: I'm not familiar with NIX_OTHER_STORES
<Unode>
can you elaborate a little (or point to documentation)
<Unode>
gchristensen: that might also be true. Last time I tried a gc run pretty much bricked the whole store.
arianvp2_ has quit [(Quit: arianvp2_)]
<Unode>
LnL: the current setup is, 1 machine has nix-daemon running and accepts requests via socket. Only this machine can 'write' to the store.
filterfish has joined #nixos
<LnL>
not sure if it's documented in the manual
indi_ has quit [(Ping timeout: 260 seconds)]
<Unode>
Other machines can read from it via NFS as mentioned above.
<Unode>
LnL: google doesn't give me much on it.
<Unode>
scattered bits here and there.
<LnL>
you can also do something with NIX_REMOTE I think but that's only for 1.12
<Unode>
the other missing bit of setup is that if users try to install things on other machines there's a socat tunnel that still talks to the first machine. This way even nix-shell works as expected on other machines.
filterfish has quit [(Remote host closed the connection)]
filterfish has joined #nixos
reinzelmann has quit [(Quit: Leaving)]
<Unode>
this socat part is something I see as benefiting from some work as it's quite non-standard and brittle but I also understand that most people simply copy /nix to different machines instead of sharing it (e.g via NFS).
tmaekawa has joined #nixos
<Unode>
LnL: so just to understand a bit better, what problem exactly would NIX_OTHER_STORES or NIX_REMOTE would solve?
<Unode>
NIX_REMOTE I already use. It's set to 'daemon'.
<Unode>
I'm not familiar with additional options other than set to void (NIX_REMOTE=)
<Dezgeg>
I think NIX_OTHER_STORES could be used to copy things from the NFS to the local disk
<Unode>
hum.. I don't think that's a target here.
<LnL>
yes if you have eg. NIX_OTHER_STORES=/run/nix/remote-stores/*/nix
<Unode>
I mean, the point of having the store on NFS is that one machine can install things and other machines with similar hardware can simply read and run from it.
<LnL>
and you put your nfs mount there nix will import the paths from there if they are available
<Unode>
perhaps one missing bit of information, I don't have root on the machine and /nix is not an option.
<Dezgeg>
so this .nfs* problem occurs during compilation or gc or what?
<Unode>
compilation currently happens on /dev/shm . The machine has enough RAM to do this. So far nothing failed due to this.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/vHbD0
<NixOS_GitHub>
nixpkgs/master 525c10e Domen Kožar: hackage2nix: purescript{,-bridge} compiles now
NixOS_GitHub has left #nixos []
<Unode>
The .nfs happens (I believe) when copying things gfrom /dev/shm to the target destination which lives on NFS.
<Unode>
I don't know enough of what nix does to understand what exactly is failing. During nix-store --repair-path this was definitely happening but there a store path is being gc'ed.
<sphalerite>
gchristensen: Just got my laptop's keyboard and motherboard replaced, that may have fixed the key repeat issue
<sphalerite>
Haven't seen it crop up yet in any case
<gchristensen>
whoa, really? how did you get it done? free?
<sphalerite>
yep, just asked @DellCares on twitter about it, answered a few questions, and they eventually sent an engineer to come do it
magnetophon1 has joined #nixos
<Dezgeg>
hm, I would definitely imagine those .nfs problems happen during gc but that happening when compiling things sounds odd
<gchristensen>
"sent an engineer" what does that mean? hmm
<sphalerite>
gchristensen: a guy came to my flat with a new keyboard and motherboard and installed them
<gchristensen>
no way
<sphalerite>
yes way
<gchristensen>
how big of a city do you live in? :P
<Unode>
Dezgeg: I'm currently installing/compiling a relatively big update. One package that showed an error was icu but the log was unusable since it's running multiple jobs in parallel and only one output stream.
<sphalerite>
Glasgow
<gchristensen>
nice
<gchristensen>
can you send me your convo? :)
<Dezgeg>
you could nix-store -l it if you can figure out the .drv path
<LnL>
Unode: and with 1.12 you can do something like this NIX_REMOTE="local?state=/home/alice/nix/var&real=/home/alice/nix/store"
<gchristensen>
sphalerite: I live in a much smaller city than glasgow (~40k people or fewer)
<gchristensen>
but I go to NYC all the time and would happily meet someone for a cup of coffee & a motherboard replacement...
<sphalerite>
gchristensen: I doubt the service is limited to bigger towns, but you can always ask
<Unode>
I'm waiting for this first compilation run to abort (running with -k) and then I can tackle the failing packages.
adamt has joined #nixos
<sphalerite>
I'll send you the conversation as soon as I find a convenient way to export it
<Unode>
LnL: what you that do exactly?
<gchristensen>
screenshots or whatever would be just fine
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 3 new commits to master: https://git.io/vHbye
<NixOS_GitHub>
nixpkgs/master 4ac6821 Franz Pletz: unbound: 1.6.2 -> 1.6.3...
<NixOS_GitHub>
nixpkgs/master cb59381 Franz Pletz: gnutls: remove dependency on unbound...
<NixOS_GitHub>
nixpkgs/master 72c36db Franz Pletz: bind: 9.10.5 -> 9.10.5-P1 for CVE-2017-3140
NixOS_GitHub has left #nixos []
<Unode>
Dezgeg: of the failing package?
<Dezgeg>
yes
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to release-17.03: https://git.io/vHbyo
<NixOS_GitHub>
nixpkgs/release-17.03 64323a9 Robin Gloster: bind: 9.10.4-P6 -> 9.10.5...
<NixOS_GitHub>
nixpkgs/release-17.03 9a45923 Franz Pletz: bind: 9.10.5 -> 9.10.5-P1 for CVE-2017-3140...
NixOS_GitHub has left #nixos []
<Unode>
not sure how to get the .drv though. Let me see if I can get one to fail and then I have something to show.
<Unode>
LnL: sorry I realized my english came out weird back there. I meant, "What does that specific setting do?" It uses sate and store from a different location while writing to the local one?
<Unode>
From what I understand I kind of have to use nix-daemon at all times. I've ran into DB corruption in the past before nix-daemon.
<Unode>
before *I started using nix-daemon
<raignarok>
could someone help me with a setuid wrapper problem? What does this mean: „wrapper.c:204: main: Assertion `!(st.st_mode & S_ISGID) || (st.st_gid == getegid())' failed.“ Context: I'm trying to send mail from a php application executed by php-fpm via postfix' sendmail and it fails with that error.
fpletz[m] has joined #nixos
seppellll has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jchildren opened pull request #26594: antlr4_7: init at 4.7 (master...master) https://git.io/vHbSC
NixOS_GitHub has left #nixos []
magnetophon1 has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] seppeljordan opened pull request #26595: guetzli: init at 1.0.1 (master...guetzli) https://git.io/vHbSP
NixOS_GitHub has left #nixos []
tmaekawa has quit [(Quit: tmaekawa)]
filterfish has quit [(Ping timeout: 240 seconds)]
cpennington has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] moretea opened pull request #26598: mysql module: add option to bind to an address (master...mysql-add-bind-option) https://git.io/vHbhl
NixOS_GitHub has left #nixos []
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<gchristensen>
MoreTea: might want to double-check your diff :)
<MoreTea>
oh lol
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rycee pushed 1 new commit to master: https://git.io/vHbh9
<NixOS_GitHub>
nixpkgs/master 19f8213 Robert Helgesson: perl-Perl-Critic: 1.126 -> 1.128
NixOS_GitHub has left #nixos []
<MoreTea>
I'm interested in declarative mysql permissions
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
indi_ has joined #nixos
bennofs has joined #nixos
Rotaerk has quit [(Quit: Leaving)]
<MoreTea>
I'm afraid that it would be a minefield of edge cases..
<gchristensen>
it would
<gchristensen>
heh
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
indi_ has quit [(Ping timeout: 260 seconds)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<MoreTea>
use mysql; alter table db add column nix_hash char(32);
<MoreTea>
moehaha
magnetophon1 has joined #nixos
<MoreTea>
if this is allowed, it should be fairly trivially actually!
<MoreTea>
Just some mundane work to convert nixos options into an sql statement
Mercuria1Alchemi has quit [(Ping timeout: 240 seconds)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
marsam has joined #nixos
ilyaigpetrov has joined #nixos
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
ericsagnes has joined #nixos
<dtzWill_>
aw man, new permissions security on Nix is rough re:evaluating old nixpkgs
<dtzWill_>
although hopefully the fixes are straightforward and/or can be cherry-picked from master O:)
goibhniu has joined #nixos
<dtzWill_>
"why did so many packages suddenly start failing" xD
<dtzWill_>
eep I have a tail
dtzWill_ is now known as dtzWill
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 opened pull request #26599: ocamlPackages.utop: allow utop to locate topfile (master...utop) https://git.io/vHNfb
NixOS_GitHub has left #nixos []
* dtzWill
wonders: Can the nixStable in release-17.03 necessarily build the channel? (AFAIK no explicit dependency on 'nix' from the packages...)
pie_ has quit [(Ping timeout: 240 seconds)]
erasmas has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
FRidh has quit [(Ping timeout: 240 seconds)]
* dtzWill
answers his own question: "no"
<dtzWill>
hmm
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Baughn closed pull request #21992: ipfs: Set IPFS_PATH globally when daemon is enabled (master...ipfs-env) https://git.io/vMys3
NixOS_GitHub has left #nixos []
freusque has quit [(Quit: WeeChat 1.7.1)]
amosbird has quit [(Quit: ZNC 1.6.4 - http://znc.in)]
amosbird has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
freusque has joined #nixos
* dtzWill
files issue and tries to gather list of failing packages
Wizek has joined #nixos
Wizek__ has joined #nixos
spinus has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
sitwon has quit [(Read error: Connection reset by peer)]
Itkovian has quit [(Ping timeout: 268 seconds)]
johnsonav has quit [(Ping timeout: 240 seconds)]
riclima has joined #nixos
ison111 has quit [(Quit: WeeChat 1.7.1)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #26602: shells-environment: do not let sudo make perl disfunctional (master...issue-25613) https://git.io/vHNOA
NixOS_GitHub has left #nixos []
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
jensens has joined #nixos
snikkers has quit [(Ping timeout: 255 seconds)]
trikl has joined #nixos
arianvp2_ has joined #nixos
goibhniu has joined #nixos
FRidh has joined #nixos
<trikl>
I'm trying to compile a KDE package that depends on Qt5's QtCore. Its CMakeLists.txt uses kdelibs' FindQt4 to try to find dependencies, but although I added qtbase as a buildInput (qtbase contains include/QtCore) QtCore is not found. Any suggestion?
<bennofs>
trikl: it uses FindQt4 to find Qt5?
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<trikl>
bennofs: Eh, good point, yes
<clever>
[6]- Segmentation fault chromium
<bennofs>
clever: o.o
<clever>
again??
magnetophon1 has quit [(Ping timeout: 240 seconds)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
k0001 has quit [(Remote host closed the connection)]
<sphalerite>
The constant dilemma... "This build has been running for a while now, should I restart it with more cores so it goes faster or will that end up taking more time overall?"
k0001 has joined #nixos
Itkovian has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
fnljk has quit [(Disconnected by services)]
fnljk_ has joined #nixos
arianvp2_ has quit [(Quit: arianvp2_)]
arianvp2_ has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
mudri has joined #nixos
<trikl>
bennofs: So I was checking a version of the package different of the version I was trying to build. Qt4 has QtCore in pkgs.qt4.out, which I added as a buildInput, but FindQt4 still doesn't find it
<clever>
trikl: did you also add pkgconfig to buildInputs?
<trikl>
I did not
<clever>
try doing that and see if anything changes
erictapen has quit [(Ping timeout: 240 seconds)]
<trikl>
Same
<bennofs>
trikl: you probably need to add pkgs.qt4.dev
<trikl>
bennofs: I checked that. include/QtCore is in pkgs.qt4.out
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
Itkovian has quit [(Ping timeout: 258 seconds)]
<bennofs>
trikl: yes but the pkg-config files are in .dev
<clever>
bennofs: its also usually better to just add pkgs.qt4, and then let the stdenv figure out which outputs to use
<bennofs>
probably
<trikl>
I added both qt4.out and qt4.dev to buildInputs, but it still doesn't find it
indi_ has joined #nixos
<clever>
cmake often uses pkgconfig to find things, and it needs pkgconfig in the buildInputs to do so
<trikl>
clever: I added pkgconfig to buildInputs too
ryanartecona has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
indi_ has quit [(Ping timeout: 268 seconds)]
Guest10403 has quit [(Ping timeout: 246 seconds)]
trikl has quit [(Ping timeout: 240 seconds)]
Itkovian has joined #nixos
jacob_ has joined #nixos
jacob_ is now known as Guest33244
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
freusque has quit [(Quit: WeeChat 1.7.1)]
myguidingstar has joined #nixos
magnetophon1 has joined #nixos
<myguidingstar>
why there is no "programs.ssh.enable" option in configuration.nix?
jensens has quit [(Ping timeout: 240 seconds)]
<myguidingstar>
how do I use ssh with configured values in "programs.ssh.*" ?
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
freusque has joined #nixos
arianvp2_ has quit [(Quit: arianvp2_)]
civodul has quit [(Quit: ERC (IRC client for Emacs 25.2.1))]
teodorlu has joined #nixos
<teodorlu>
Hello! I have a naïve question. Why don't distributions like Ubuntu just adopt
<teodorlu>
Nix?
<teodorlu>
<teodorlu>
To me, the choice between basing your product on something that may fail and
<teodorlu>
something that may not is easy. Is there something I'm missing?
<ToxicFrog>
That's not the choice, though; it's a choice between two things that can both fail but in mostly different ways
marsel has quit [(Ping timeout: 255 seconds)]
<disasm>
myguidingstar: what are you trying to do? enable/disable ssh is in services.openssh, not program. Programs is more things like starting the agent for users, etc...
<ToxicFrog>
And nix has some sharp failure modes, e.g. if any one installed package is broken in the channel(s) you use you can no longer install/update anything until it's fixed.
<teodorlu>
Has that been a problem for Nix?
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<teodorlu>
I've never experienced the channels not working.
<disasm>
teodorlu: someone could if they wanted. I think they have so much invested in their existing systems, they don't want to make a switch.
<ToxicFrog>
Getting software that expects FHS to work right is full of land mines; Steam works by just providing it with an FHS chroot, AIUI, and KDE5/Plasma5 have recurring issues with following symlinks into /nix/store and then baking references to them into the config files which then break when updates are installed.
<myguidingstar>
disasm: I want to use ssh client. I find intersting default values of programs.ssh.*
<ToxicFrog>
teodorlu: lucky you! Chrome was broken for a few days in 17.03 earlier this month, for example.
<disasm>
myguidingstar: define what you mean by ssh client. You want to ssh from one host to another?
<teodorlu>
FHS?
<sphalerite>
Filesystem Hierarchy Standard
<ToxicFrog>
Filesystem Heirarchy Standard, i.e. the filesystem layout used by "normal" linux distros
<teodorlu>
Thanks
<myguidingstar>
from my local machine to some remote host
<disasm>
err myguidingstar ^^
arianvp2_ has joined #nixos
<ToxicFrog>
myguidingstar: you probably just want openssh, then?
athan_ has quit [(Ping timeout: 240 seconds)]
<ToxicFrog>
Like, add it to systemPackages or install it with nix-env
<ToxicFrog>
programs.ssh.* is for setting system-wide defaults for ssh configuration.
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<myguidingstar>
I installed openssh
arianvp2_ has quit [(Client Quit)]
<ToxicFrog>
Ok, you should now be able to run `ssh` and do the thing.
<myguidingstar>
programs.ssh.startAgent is true by default
<myguidingstar>
and programs.ssh.askPassword is some x11 program
<ToxicFrog>
Yes?
<ToxicFrog>
This is all normal
<ToxicFrog>
What is your question/problem?
<myguidingstar>
so I guess that program should ask me for the first time I tried to use ssh
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<MagneticDuck>
how do I find the hash that requireFile expects, given a file in the store that I want it to refer to?
<teodorlu>
Disgarding invenstment and switching costs, do you think the current issues with Nix are fixable to the extent that Nix can become a strictly better (in every regard) package manager than apt-get and others?
<myguidingstar>
which is not happening and I am having to run ssh-add manually
<disasm>
myguidingstar: ah, I see what you're saying now :)
<sphalerite>
teodorlu: well, in what regards is nix not a strictly better package manager yet? :)
Itkovian has quit [(Read error: Connection reset by peer)]
<myguidingstar>
:)
<sphalerite>
myguidingstar: look at the AddKeysToAgent option in man ssh_config
<teodorlu>
Some issues were mentioned above! I use Nix for Ubuntu for everything I can, but I stick to apt-get for packages that have to integrate a lot with the rest of the system.
ryanartecona has quit [(Quit: ryanartecona)]
<sphalerite>
teodorlu: nix will never be a better package manager than the native one on distros that aren't based on it
<teodorlu>
(And I guess we don't have to argue that Nix is great on this channel)
<sphalerite>
If you want nix to integrate with stuff, you want NixOS :)
<teodorlu>
What if it was adopted to replace apt-get?
<teodorlu>
Yeah, I know. Still, not as ... easily usable as Ubuntu. Yet.
<sphalerite>
myguidingstar: you can add a definition of the setting using programs.ssh.extraConfig
<MagneticDuck>
hey, are we discussing problems with nix the package manager?
acertain has quit [(Ping timeout: 268 seconds)]
<disasm>
I agree, most people that want nix are going to want the declarative configuration.nix you get with nixos.
snikkers has joined #nixos
<disasm>
It's one of those things, once you get used to it, you never go back :)
<ToxicFrog>
I'm totally sold on NixOS for servers, but I'm considering switching my laptop back to SUSE. It would mean switching away from ZFS as well, but on the other hand, no longer needing to deal with nix-related issues with steam, or dbgl, or the like.
<MagneticDuck>
how about the problem I'm having right now -- I can't use the Mathematica installer to install Mathematica, but there's a derivation that uses it to create a Mathematica derivation
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<myguidingstar>
fyi, I thought I need something like programs.ssh.enable because of the case I've had with fish shell
<ToxicFrog>
And yeah, the prospect of using nix-on-suse is completely unappealing; it feels like it'd get me the worst of both worlds.
<MagneticDuck>
the problem is, the whole thing is about 4G and nix-store --add literally puts the whole file in memory (why?)
<sphalerite>
teodorlu: I don't think apt (apt-get is only one tool that's part of apt as a whole) can be replaced by a package manager that breaks FHS as part of its model at all
<ToxicFrog>
configuration.nix is what originally attracted me to nixos, since my ansible configuration was getting increasingly large and fragile.
<MagneticDuck>
so now I'm jumping through hoops involving remounting /nix/store and trying to figure out what requireFile wants
<myguidingstar>
first, I just added the fish to system packages and not enable programs.fish, and no shell aliases available to my fish
<MagneticDuck>
as usual, there's no documentation, so I'll just read the definition of requireFile and work from there
<myguidingstar>
until I enable programs.fish, all the shell aliases work
MichaelRaskin has joined #nixos
<myguidingstar>
that's why I thought of programs.ssh.enable :)
<sphalerite>
myguidingstar: yes, I was a little confused by the absence of programs.ssh.enable while looking that stuff up just now too...
<teodorlu>
What about all the people that just use Ubuntu because "it's easy and works"? I agree that configuration.nix is a better solution than scattering system config all around. Way better. My feeling, though, is that what Nix needs now is more developer time, and Ubuntu (for example) has plenty. How can we make a compelling argument for them that allows them to keep what they have created?
JagaJaga has quit [(Ping timeout: 240 seconds)]
<teodorlu>
I just want people to give higher priority to Nix packages for distribution.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] k0001 opened pull request #26603: ghc821: init at 8.2.1-rc2 (a.k.a., 8.2.0.20170507) (master...ghc821rc2) https://git.io/vHN4q
NixOS_GitHub has left #nixos []
<ToxicFrog>
myguidingstar: yeah, that's a recurring issue with shells in general in nix; I had the same experience with zsh
snikkers has quit [(Ping timeout: 260 seconds)]
<ToxicFrog>
teodorlu: before you can start competing with Ubuntu on those grounds, IMO, you need good documentation and a foolproof, user-friendly installer; Nix has neither. So that's a lot of groundwork.
<sphalerite>
teodorlu: I don't think that will happen. Canonical seem to be pretty invested in their "snaps" thing
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
arianvp2_ has joined #nixos
<teodorlu>
Ubuntu 18.04 drops their "Unity" thing for Gnome :)
<disasm>
teodorlu: the biggest problem is you'd have to retrain all the power users that use ubuntu as well. With nix, now they have to specify dependencies in the nix store for building their applications instead of just having everything already in /usr/lib. It's rather fundamentally different way of looking at packaging. And not a lot of folks are using nix-shell to develop their apps.
<clever>
disasm: and i often see people new to nix trying to nix-env -i every library, then wonder why it doesnt work
<teodorlu>
Yeah, that's right. Developer overhead as well. I don't presume it's possible to do some kind of static analysis to find /what/ packages in /usr/lib that are needed, and suggest a starting point for a Nix configuration?
arianvp2_ has quit [(Client Quit)]
<disasm>
clever: exactly... I never let my nix-env get more than 5 or 6 packages in it at a time. If it's specific to a project, those packages go in a nix-shell. If it's something I want all the time, it eventually makes it into configuration.nix.
<teodorlu>
clever: I'm installing things with nix-env -iA. Kinda wondering when -i is more useful, and why it's default.
<clever>
i always say to use -iA
<clever>
its much faster
nschoe has quit [(Quit: Program. Terminated.)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<sphalerite>
-i is never more useful iiuc, and it's default because of legacy. The new CLI defaults to it I believe
<teodorlu>
Nice to hear
snikkers has joined #nixos
pxc has joined #nixos
<sphalerite>
ugh, qemu takes ages to compile
q6AA4FD has quit [(Ping timeout: 246 seconds)]
riclima has joined #nixos
<teodorlu>
If we only could go back and give Nix to K&R
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<gchristensen>
it is too powerful for that time
arjen-jonathan has quit [(Ping timeout: 240 seconds)]
<teodorlu>
I think they were pretty smart
<gchristensen>
I think nix can only exist when it is okay to waste processing power and disk space in the name of sanity
<teodorlu>
Good point
<sphalerite>
Can I pause a nix build? Sending SIGSTOP to nix-build doesn't stop the actual job, and my laptop's running out of battery but I don't want to lose the progress it's made on the build
<teodorlu>
When there were only one version of a program, why use different ones under limited resources.
<gchristensen>
sphalerite: brutal ...!
<clever>
sphalerite: when using nix-daemon, you have to stop the child of the daemon, not the nix-build that is remotely triggering it
<sphalerite>
and iirc the nix daemon will kill a build if it doesn't produce any output for too long, so I don't want to trigger that by just stopping the build process either...
ryanartecona has joined #nixos
<clever>
sphalerite: yeah, that seems like a tricky thing
<sphalerite>
I could stop the whole daemon maybe? :D
<clever>
sphalerite: maybe
q6AA4FD has joined #nixos
<sphalerite>
what's the worst that could happen...
<clever>
no other builds can take place, oh right one other thing
<clever>
sphalerite: nix-daemon will fork out a child for each client, and i believe that child is responsible for the timers
<clever>
sphalerite: if you check "pstree", you should see a child with the pid of nix-build in its arguments
<clever>
that sub nix-daemon is the one you want to sigstop
<clever>
then the root nix-daemon can still do new builds
<sphalerite>
I just did sudo pkill -STOP nix-daemon :p
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<sphalerite>
I don't actually need to do any new builds in the meantime so whatever
<clever>
and make sure it goes to every child of that as well
magnetophon1 has quit [(Ping timeout: 240 seconds)]
<sphalerite>
I stopped the make processes as well, I think that should be enough?
<clever>
yeah
<clever>
once the gcc's under it finish, it will hang
nevermind has joined #nixos
<sphalerite>
Yeah the gccs have gone zombie
<clever>
yep, waiting for make to waitpid() them
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<sphalerite>
Sorry gcc, your parents are asleep. You'll have to wait a little.
Sonarpulse has quit [(Ping timeout: 260 seconds)]
<clever>
ive also see a bug, only on arm i believe
<clever>
where make doesnt waitpid
<clever>
so it hangs just like above, with no sigstop to blame
<clever>
but if i sigchld the make, the build resumes like everything is ok
<sphalerite>
huh
<sphalerite>
kernel bug? Or make?
<clever>
i'm guessing its a race condition in make, where the child finishes too quickly, before the sigchld handler is setup
<sphalerite>
ah
<MagneticDuck>
who wants to guess how long it takes for nix-prefetch-url to pull a 4 Gb file from a local path?
<clever>
irssi had a similiar problem in /exec at one time
<sphalerite>
MagneticDuck: many
<clever>
if the child exited, irssi stopped reading the stdout pipe
<MagneticDuck>
... on a machine with 3 Gb free RAM
<clever>
so a command that does a ton of output and quits asap, like df -h, will get truncated
<clever>
irssi would stop at child exit, not at eof
k0001 has quit [(Remote host closed the connection)]
<sphalerite>
oh, that's convenient
<sphalerite>
Can't make set up the SIGCHLD handler before forking?
<clever>
it would often truncate the output, id have to /exec -o df -h ; sleep 2
k0001 has joined #nixos
<sphalerite>
Or the child wait for the SIGCHLD handler to be set up before execing?
<clever>
sphalerite: i'm purely guessing on how i would have written it and could have messed up, havent checked the make source
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
but it could do those things
<sphalerite>
ah right
stallion has joined #nixos
goibhniu has quit [(Ping timeout: 240 seconds)]
zeus_ has joined #nixos
MichaelRaskin has quit [(Ping timeout: 240 seconds)]
chpatrick has joined #nixos
<chpatrick>
is it possible to access the running nixos config from a nix-shell script?
myguidingstar has quit [(Quit: ChatZilla 0.9.93 [Firefox 53.0.3/20170608225630])]
<MagneticDuck>
but yeah, does anyone have a clue what hash requireFile is expecting? I have a thing in the nix store, how do I convince requireFile to refer to it?
<chpatrick>
specifically I would like to make a buildenv that sets ld_library_path to include "${config.boot.kernelPackages.nvidia_x11}/lib"
<MagneticDuck>
it's not `nix-hash --base32 --type sha256 <file>`
<clever>
MagneticDuck: i believe its the hash of the NAR file
endformationage has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
MagneticDuck: if you ran nix-store --dump on the path, then pipe that nar into nix-hash, i think
k0001 has quit [(Remote host closed the connection)]
<clever>
MagneticDuck: and also, if you just manualy copy something into the store, db.sqlite wont say its valid, and nix will just delete it
k0001 has joined #nixos
<MagneticDuck>
clever: I already did the magical reregistration dance
<clever>
ah
<clever>
let me check some figures on this end
* MagneticDuck
waits patiently for things to stop using swap memory on the hard disk
<teodorlu>
Question: when installing Nix on Ubuntu, do I get a declarative file I can put my list of installed packages in (so that I can version control it and copy it to other systems)?
<clever>
teodorlu: line 12- makes a new package called mystuff, which i can then nix-env -iA nixpkgs.mystuff
* MagneticDuck
sure is curious why nix likes to read *entire* files into memory before writing to the store
<MagneticDuck>
I hope it has a pretty good reason
<teodorlu>
Cool
<clever>
MagneticDuck: it needs to know the hash of the file before it knows where to write it, and it doesnt want you being naughty and changing the file between the hash and copy
<MagneticDuck>
clever: apparently the hash in requireFile matches the hash I get when passing --flat, so the issue is that I didn't copy to the store correctly
<teodorlu>
It seems I can't write to ~/.nix-profile/config.nix
<clever>
MagneticDuck: what name did you give it in the store?
<clever>
teodorlu: oh, it has to go in ~/.config/nixpkgs/config.nix
<MagneticDuck>
well, the name doesn't use the --flat hash
<clever>
teodorlu: nix-env manages all of .nix-profile, which must never be modified
<clever>
MagneticDuck: the hash in this path has no visible connection to the --flat hash
<clever>
and if that path is wrong, nix wont find the file
<MagneticDuck>
I don't suppose I can refer to a particular store path in a derivation?
marsel has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
MagneticDuck: you may want to check the .drv file thats failing, to see the right storepath
<clever>
then copy it there, and do the registration thing again
<MagneticDuck>
the drv file that's failing?
<clever>
ah, that was in step 1, maybe something got messed up then
<clever>
MagneticDuck: can you gist the entire console output from the failing nix-build?
<MagneticDuck>
clever: well there's not much to see
<clever>
the paths and hashes are important
<MagneticDuck>
it says it's going to produce the file with a hash that's not my hash, and then it gives me the requireFile message
<clever>
i need to see those exact paths
<MagneticDuck>
under "these derivations will be built:" in the nix-build output, I have /nix/store/y83prin1dixpzp78mqg9lnzmf1jldxgw-Mathematica_11.0.1_LINUX.sh.drv
<clever>
yeah, thats the hash of the drv. not the output
<MagneticDuck>
and the file in my store is /nix/store/dchih04sii8rrvklkjvkfwbmvfs1anks-Mathematica_11.0.1_LINUX.sh
<clever>
run cat on that .drv file, and youll see a second path near the top
<MagneticDuck>
oh I selected the wrong thing
<clever>
"cat /nix/store/y83prin1dixpzp78mqg9lnzmf1jldxgw-Mathematica_11.0.1_LINUX.sh.drv" and the first path should be the path its expecting to find in /nix/store/
<MagneticDuck>
which is /nix/store/j413aqsnbcmj4dz1bm816ih446sv91r2-Mathematica_11.0.1_LINUX.sh
<clever>
jake_: thats exactly what nixos-install does
<jake_>
and that won't mess with any grub/bootloader stuff in the existing system?
<ToxicFrog>
To actually use it live you'll probably have to bind-mount /dev /proc /sys into it but nix-install should handle the rest.
<ToxicFrog>
It *can* but it doesn't have to.
<clever>
jake_: it will only mess with the device listed in boot.loader.grub.device within the configuration.nix file
Shados has quit [(Quit: No Ping reply in 180 seconds.)]
<clever>
you can set that to "nodev" to turn that off
<clever>
ToxicFrog: nixos-install handles all of those bind-mounts for you
<jake_>
so it's totally usable as a subdirectory within another operating system? permanently, not just as a temporary way to install it as the main os?
mkoenig has quit [(Ping timeout: 240 seconds)]
a123123123[m] has joined #nixos
<ToxicFrog>
clever: Don't forget to set boot.loader.efi.canTouchEfiVariables to false as well if you're on a UEFI system.
<ToxicFrog>
er
<MagneticDuck>
clever: ahahah
<ToxicFrog>
jake_: ^
Shados has joined #nixos
<MagneticDuck>
and then after all that, it fails at runtime
<clever>
jake_: maybe, but the systemd stuff will want to be pid 1, so you would want to look into containers
<ToxicFrog>
clever: it does? Is there anything special needed to get it to do that, say, on every boot so you can just `chroot` into it?
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
ToxicFrog: nixos-install also has a --chroot flag, that will just run bash instead of nix-build
<clever>
ToxicFrog: so you can get a shell inside a nixos
<ToxicFrog>
Oh sweet.
<a123123123[m]>
can I verify somehow that executables in /nix/store/* weren't patched, by comparing to the hash value?
<jake_>
wait so i'm a little confused. what exactly are the set of steps i should do to get a subdirectory i can chroot into that has nix-os in it but doesn't affect my larger os
<simpson>
clever: Is that basically $(guix challenge) for Nix?
<clever>
simpson: not familiar with guix commands
<a123123123[m]>
clever: cool, that was fast
<sphalerite>
clever: No, why would I be able to?
<simpson>
clever: Me neither, but it rebuilds a package from scratch to check the hashes.
q6AA4FD has quit [(Ping timeout: 255 seconds)]
<clever>
simpson: ah, thats --check in nix
<clever>
simpson: --verify --check-contents only checks that the hash matches what it originaly was at install time
<clever>
sphalerite: ah, thought you might, its not that big of an issue
<clever>
jake_: what is your end goal with the nixos chroot?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz opened pull request #26605: ssh module: disable agent by default (master...disable-ssh-agent) https://git.io/vHN2y
NixOS_GitHub has left #nixos []
<sphalerite>
clever: heh, well I feel flattered that you think that much of my involvement in nixos :D
<clever>
:D
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<disasm>
jake_: why not use systemd-nspawn?
<clever>
ah, make-system-tarball may help more now that i think of it
stanibanani has quit [(Remote host closed the connection)]
hellrazor has joined #nixos
<clever>
jake_: is the host systemd based?
<jake_>
clever: to use on a chromebook. it has a linux system with limited permissions in the root. i want to have the chrome os but also be able to chroot into a nixos system within that for developement
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vHNaJ
<NixOS_GitHub>
nixpkgs/master 1b6176e Roger Qiu: gnupg module: Added extra and browser sockets (#26295)...
<jake_>
clever: i'm not quite sure what being sytemd based means
afics has quit [(Ping timeout: 240 seconds)]
<clever>
jake_: does systemd-nspawn exist in $PATH as root?
arianvp2_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] robx opened pull request #26606: python-modules/protobuf: fix darwin build by using $CXX instead of $CC (master...fix-python-protobuf) https://git.io/vHNa4
NixOS_GitHub has left #nixos []
<jake_>
clever: no
arianvp2_ has quit [(Client Quit)]
<clever>
ah, then a pid namespace may be tricky
<clever>
jake_: i can still write an expression that will get you the root for chroot, give me a few minutes
<jake_>
clever: what is the importance of the pid namespace
<jake_>
clever: thank you!
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
the systemd inside the chroot wants to be pid 1
<clever>
and a pid namespace gives the guest its own pid table
<jake_>
clever: oh and nixos relies on systemd
<clever>
yeah, so you wont have any services
<clever>
just a raw shell
<jake_>
clever: okay so getting the pid namespace would be nice, but a raw shell with nixos inside would still be better than nothing. how do i set that up at all?
<clever>
jake_: but you could get most of the same things if you just install normal nix
<clever>
(this keyboard doesnt have a working pipe key)
<jake_>
clever: I don't have ownership of the root directory, so i'm going to need to chroot anyway
<clever>
the chroot command needs root
<jake_>
clever: my backup plan is using debootstrap to create a debian chroot and installing nix inside there, but i figured it'd be better
<jake_>
clever: sorry ownership is the wrong term
<jake_>
clever: i can become root, but the file system is mounted read only in the root
<clever>
ahhh
<clever>
mount / -o remount,rw
<clever>
you can just make it writable
<clever>
then put a symlink at /nix pointing to somewhere that will remain writeable later
<jake_>
clever: interesting. i still want to try and get this chroot thing working though because it leaves the whole chromeos system alone and i want to try and have it integrate with crouton, which creates chroot, because it provides integration with the chrome gui
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<jake_>
clever: so is there a way to get it working from a chroot, with or without the pid namespace?
<clever>
jake_: ok, let me write the tarball thing
<jake_>
clever: thank you! also the remount fails with an error that it is write protected
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #26607: vivaldi: fix space in sha256 (master...vivaldi-sha256) https://git.io/vHNVB
NixOS_GitHub has left #nixos []
<jake_>
clever: i do think there is a known way to get around that if i really tried though
<clever>
jake_: ah, then the block device itself might be read-only
stanibanani has quit [(Quit: Leaving.)]
<clever>
the nixos-install wouldnt have worked either then
<clever>
since that needs nix on the host
<sphalerite>
I'm writing some nixos tests for some changes I'm working on. I need multiple machines with different configurations for them, but these machines never need to interact with each other. Should I create a single test with multiple machines, or create multiple tests? In the latter case, how do I put multiple tests in one file?
freusque has quit [(Quit: WeeChat 1.7.1)]
<clever>
sphalerite: i think you can set which vlan each machine is on
stanibanani has joined #nixos
<sphalerite>
clever: but given that they're essentially completely independent, wouldn't it be neater to have separate tests?
<clever>
sphalerite: yeah, that might be better
<clever>
then you can see which part failed
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<stanibanani>
I bought just bought a wifi dongle (ASUS Wireless-AC1300) does anyone here know how to make it work?
<clever>
stanibanani: does it appear in "ip link"?
<stanibanani>
no it doesnt
<sphalerite>
Yeah. Just wondering how to put them in a single file, I know I could just make the file evaluate to a list containing the tests and that would work with nix-build but I'm not sure about the machinery that aggregates test results
<clever>
sphalerite: if you return an attribute set of tests, then it should just list each of them seperately on hydra
<sphalerite>
oh right, that sounds good
<sphalerite>
thanks
arianvp2_ has joined #nixos
<jake_>
clever: there are directories that i can write to: /usr/local and Downloads basically
<clever>
jake_: the nix file i'm writting will generate a .tar file containing a full nixos, similiar to a gentoo stage3
<clever>
jake_: you can then unpack that to anything, and chroot into it
<jake_>
clever: gotcha. amazing.
<sphalerite>
Vaguely related, how feasible (performance-wise) is using an ARM VM on x86 to build ARM stuff, thereby avoiding all the complications of cross-compilation?
hellrazor has quit [(Ping timeout: 255 seconds)]
<clever>
sphalerite: i have used qemu-user-arm to do similiar, with less overhead
acertain has quit [(Ping timeout: 240 seconds)]
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<sphalerite>
clever: how does that work? Emulating the CPU but using the host kernel and translating syscalls?
<clever>
sphalerite: exactly
chpatrick has left #nixos []
<clever>
sphalerite: it uses binfmt-misc to detect the arm fields in an ELF file, and runs qemu-user as an interpreter
<sphalerite>
What a time to be alive
<clever>
so the kernel treats it like a bash script
<clever>
sphalerite: the register script in this derivation will add it to binfmt-misc, and the nix patch will convince nix-daemon that the host can run arm binaries
<sphalerite>
wow
<stanibanani>
I have done some research and it needs the rtl8812au driver. there is a linuxPackages.rtl8812au package in the nix repo, but I have no idea how to load it. Is it enough to just add it to "hardware.firmare" option?
<clever>
jake_: if you run nix-build in this directory, it will generate a tar at result/tarball/nixos-system-x86_64-linux.tar.xz
stanibanani has quit [(Remote host closed the connection)]
<clever>
jake_: inside that tar, is a full nixos container, with a /init and /bash symlink to help you enter it
arianvp2_ has quit [(Quit: arianvp2_)]
<clever>
jake_: unpack it anywhere, and then either run /bash under a chroot, or /init as pid1, ah, and the activate stuff may need init to work
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh closed pull request #26607: vivaldi: fix space in sha256 (master...vivaldi-sha256) https://git.io/vHNVB
NixOS_GitHub has left #nixos []
<sphalerite>
How can I see what caused a systemd unit to start?
stanibanani has joined #nixos
<sphalerite>
jake_: and for getting init to be pid1, unshare -p could be helpful
drakonis has joined #nixos
<clever>
sphalerite: ah, would it be possible to recreate systemd-nspawn with unshare?
<sphalerite>
clever: not quite, I think it supplies a bunch of extra functionality
<clever>
jake_: oh right, is the chromebook x86 or arm?
guillaum1 has joined #nixos
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
turion has joined #nixos
teodorlu has left #nixos ["ERC (IRC client for Emacs 25.1.1)"]
<stanibanani>
clever: no, networking.enableRTL8192cFirmware = true; didn't do it. the device is still not showing up in ip link :/
<sphalerite>
clever: actually maybe. It would be a lot of work though
<clever>
stanibanani: what does dmesg say after you unplug and replug the device?
<gchristensen>
LnL: thanks :D
<clever>
sphalerite: dont need the full feature set, just enough to run a nixos container without any nix shared to the host
<stanibanani>
clever:
<stanibanani>
[ 334.175655] usb 1-8: USB disconnect, device number 4
<stanibanani>
[ 339.340123] usb 1-8: new high-speed USB device number 5 using xhci_hcd
<stanibanani>
[ 339.508393] usb 1-8: New USB device found, idVendor=0b05, idProduct=17d2
<stanibanani>
[ 339.508396] usb 1-8: New USB device strings: Mfr=1, Product=2, SerialNumber=3
<stanibanani>
[ 339.508399] usb 1-8: Product: 802.11n NIC
<stanibanani>
[ 339.508401] usb 1-8: Manufacturer: Realtek
<stanibanani>
[ 339.508404] usb 1-8: SerialNumber: 123456
<turion>
I was about to install nixos with the minimal console installer, but it didn't have the b43 broadcom firmware and/or driver, which I need. Are my chances better with the graphical installer?
<guillaum1>
Something I'm unsure of. I'm using nixos, I have the nixos-17.03 channel activated. However I can see some "<nixpkgs>" or "<nixos>" derivation. Where does the "<nixpkgs>" ones comes from?
<clever>
stanibanani: have you rebooted after doing nixos-rebuild switch?
<stanibanani>
turion: afaik there is no difference between the minimal and the graphical installer except the fact that graphical installer just opens a KDE session
<stanibanani>
clever: yeah I have
<jake_>
clever: i will give this a try. its x86
<clever>
jake_: ah, then it should be fairly simple
digitalmentat has joined #nixos
<turion>
stanibanani, ok, makes sense. How would I get the drivers on the installer USB stick then, such that I can install them from there without internet?
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
<clever>
turion: you would need to either build a custom ISO that has the drivers, or just get an ethernet cord
<turion>
clever, Can I somehow put the drivers on a second stick to install? But in what format?
<clever>
turion: not easily, nix doesnt allow you to just install random drivers like that
<turion>
Ah... does the kernel in the live system support USB tethering?
<turion>
I see
<clever>
not sure, never tried that
stallion has joined #nixos
<turion>
Hmm... how do I install custom firmware like for b43 anyways? I remember it being a hassle on other distributions
stallion has quit [(Read error: Connection reset by peer)]
bennofs has quit [(Ping timeout: 240 seconds)]
stallion has joined #nixos
<clever>
turion: networking.enableB43Firmware = true; in the nixos configuration.nix and nixos-rebuild switch
<clever>
but you need working internet to complete that
sitwon has joined #nixos
sitwon has quit [(Read error: Connection reset by peer)]
MichaelRaskin has joined #nixos
<turion>
clever, oh wow, that's easy, thanks! So let me try with USB tethering as soon as I have downloaded the graphical installer
<jake_>
clever: tried init and you're right, its unhappy about systemd. this sytem doesn't have unshare -p on it. I guess I could write my own using the unshare c function or install it using nix in my other chroot and fiddle with the rpaths so it links to the host glibc etc
<jake_>
clever: does the new tarball just include the unshare -p or does it fix the memory issue
<sphalerite>
What if you unshare -p exec bash ?
<clever>
it just has a bash script with the correct path to unshare
<sphalerite>
It's not actually a memory issue
bennofs1 has joined #nixos
<clever>
sphalerite: yeah, it appears to be failing when the stage2 tries to launch a child
<sphalerite>
I remember this issue
<sphalerite>
OH yeah that's right
<sphalerite>
hang on let me see if I can find the docs
<jake_>
sphalerite: unshare -p exec bash complains about no exec
ashu has quit [(Ping timeout: 260 seconds)]
<sphalerite>
jake_: what about unshare -p bash and then in the resulting bash type init?
<jake_>
sphalerite: what do you mean bash type init?
<jake_>
sphalerite: init is /run/current-system/sw/bin/init
<sphalerite>
sorry a bit confusing there.
Guest83227 has quit [(Ping timeout: 246 seconds)]
<stanibanani>
I noticed something strange. I have added the "linuxPackages.rtl8812au" package to my stystem packages, built, switched, but the corresponing kernel module doesn't appear in its kernel-modules folder
<jake_>
sphalerite: it was my bad. is that a good thing?
<sphalerite>
see pid_namespaces(7): If the "init" process of a PID namespace terminates, the kernel terminates all of the processes in the namespace via a SIGKILL signal. This behavior reflects the fact that the "init" process is essential for the correct operation of a PID namespace. In this case, a subsequent fork(2) into this PID namespace will fail with the error ENOMEM; it is not possible to create a new pro
<sphalerite>
cesses in a PID namespace whose "init" process has terminated.
<clever>
jake_: there are 2 init's
<sphalerite>
So you want to 1. unshare -p bash — the first child of that bash will be PID 1; so in that bash you run init
ndrei_ has joined #nixos
<jake_>
i can't even run ls
<clever>
stanibanani: kernel modules have to be added to boot.kernelPackages
<clever>
jake_: read the enter script to find the correct value of $PATH
<sphalerite>
unshare -p bash -c '/init'
<sphalerite>
I think that should work
<jake_>
what i've done is from the original chromos ran sudo chroot /usr/local/nixos-container/ /bash
<clever>
sphalerite: oh, the first child of the unshare child is 1?, not the bash itself?
<ndrei_>
Hello, I'm having trouble using Nginx with SSL-enabled virtualhost through LetsEncrypt, I'm getting an error on .well-known verification attempt
<sphalerite>
clever: exactly
<clever>
thats weird
<jake_>
running unshare -p bash -c '/init' has the same fork problem
<ndrei_>
anybody here using this feature?
<clever>
sphalerite: yeah, chroot . bash -> unshare -p bash -> /init, appears to work
<jake_>
sphalerite: running unshare -p bash gives me a shell where every command fails because of forking
<clever>
sphalerite: how do i make that systemd die?, lol
<jake_>
sphalerite: running unshare --map-root -p bash after the first chroot doesn't work
<sphalerite>
clever: systemctl poweroff I think?
<jake_>
sphalerite: operation not permitted
<clever>
sphalerite: but i have no shell inside that root, need to -9 it and redesign
<sphalerite>
jake_: the --map-root is just for my case because I ran it as an unprivileged user on my laptop
<sphalerite>
clever: ah, that's a bummer of course :D
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<jake_>
sphalerite: OH BOY
<sphalerite>
jake_: ?
<ndrei_>
oh nevermind its working now!! This. Is. Awesome.
athan_ has quit [(Ping timeout: 260 seconds)]
<jake_>
sphalerite: it's running init! seems to be stuck on starting systemd
<sphalerite>
yay! aww.
<ndrei_>
I'm guessing some DNS resolving issues, having just switched servers.
<clever>
sphalerite: i think i also want a mount namespace, systemd left all kinds of mounts in /sys, and now i cant rm -rf
<sphalerite>
clever: ouch
<sphalerite>
yes, you do probably :D
<clever>
lol, it even bind-mounted /nix/store read-only!
<clever>
rm: cannot remove 't/var/empty': Operation not permitted
<clever>
yep, immutable bit, lsattr!
<sphalerite>
heh
<jake_>
sphalerite: i can't even stop the systemd with C-c
<sphalerite>
jake_: yeah you're probably seeing the exact same thing as clever
stanibanani has left #nixos []
<clever>
jake_: i had to -9 it, then clean up nearly a dozen mounts
riclima has joined #nixos
<sphalerite>
Also, there's got to be a Better Way™ of running nixos tests when modifying systemd that doesn't involve rebuilding qemu every time systemd is changed :/
<sphalerite>
anyone know such a way?
turion has quit [(Ping timeout: 260 seconds)]
<clever>
jake_: i'm writing a new script now, with the improvements
ertes has joined #nixos
<niksnut>
sphalerite: yes, set systemd.package to the new systemd
<jake_>
clever: okay. i'm just going to nuke the directory and wait for the new tarball
<clever>
jake_: and you need heavy use of umount before rm -rf will be happy
stanibanani has joined #nixos
<jake_>
clever: i have discovered this. okay
<clever>
jake_: double-check the output of "mount"
<niksnut>
sphalerite: so I always make a copy of the systemd expression and add an attribute systemd_new to all-packages.nix
<niksnut>
and then set systemd.package = pkgs.systemd_new;
<jake_>
clever: just a question, i saw /dev/sda1 on /usr/local/nixos-container/nix/store type ext4 (ro,nodev,relatime,commit=600,data=ordered) in mount. does that mean my /dev/sda1 has been overwritten with the nix store?
<clever>
jake_: thats a bind mount, a subdir of sda1 (the old holding nix/store) was mounted again to nix/store in read-only mode
<clever>
jake_: that forcibly sets /nix/store to be read-only
nschoe has joined #nixos
<jake_>
clever: gotcha.
ndrei_ has quit [(Quit: leaving)]
<jake_>
clever: and what is this about the immutable bit? i'm like 5 minutes behind you exactly
<clever>
chattr -i
<clever>
on the var/lib/empty path i think
<jake_>
clever: it was /var/empty. there we go all clean
ris has joined #nixos
<clever>
oh
<clever>
i think i know why my system is misbehaving now
<clever>
i ran rm -rf t while things like /proc and /dev where mounted
stanibanani has quit [(Ping timeout: 255 seconds)]
<clever>
now links dont work in teamspeak, and my hotkey to launch new terminals is broken
reinzelmann has quit [(Quit: Leaving)]
<clever>
yeah, xterm just doesnt launch, period
<jake_>
clever: will a restart fix that?
<clever>
probably, one min
<clever>
so many open windows!
<disasm>
clever: haha, I know the feeling :) although I've found since switching to i3, I'm much more likely to clean up the unused terminals a lot quicker.
<clever>
there, last one, lol, brb
riclima has joined #nixos
<clever>
(facepalm)
<clever>
systemd refuses to even shutdown, lol
<clever>
note to self, be carefull with sudo rm -rf
<disasm>
ctrl-alt-del over and over? :)
<clever>
disasm: no response
<jake_>
clever: can you reach the power cord
<clever>
jake_: holding the power button down also works
riclima has quit [(Client Quit)]
<catern>
hey #nixos
riclima has joined #nixos
<catern>
I want to set up my own channel based on my own fork of nixpkgs
<clever>
sphalerite: aha, unshare also has a --fork
<catern>
LnL: but, right, I'm not sure I understand
<catern>
LnL: does that preclude having a binary cache?
<clever>
sphalerite: i see how it works now, unshare creates a new pid namespace, and then execve's the given child, but still has the original pid
<clever>
sphalerite: and the first child it (or the child) makes, becomes pid 1
<sphalerite>
catern: no. Binary caches and channels are basically independent
<sphalerite>
clever: exactly
<clever>
and --fork saves you a step
<sphalerite>
clever: didn't know about the --fork flag though, that's good to know
<catern>
LnL: like, how do you point to a binary cache with that? just commit binary-cache-url in the root of the repo?
<catern>
sphalerite: what? no... "A Nix channel is just a URL that points to a place containing both a set of Nix expressions and a pointer to a binary cache."
<LnL>
you don't have to if you use the nixos.org cache
<sphalerite>
catern: oh whoops never mind me then
<clever>
unshare: cannot stat /proc/16719/ns/mnt: No such file or directory
<catern>
wait so how does this work in a multi-user setup? is there some way to have a system-wide blessed channel, whose binary-cache-url is trusted?
Judson is now known as Guest14983
<catern>
or do I just need to customize in nix.conf, binary-caches?
<LnL>
you can put your cache in nix.conf
<catern>
right
<catern>
i *can*
<catern>
but should I?
<catern>
I guess there's no notion of a system-wide channel
<catern>
is that right?
riclima has joined #nixos
<LnL>
not sure, maybe nix uses it from the channel
<catern>
or a trusted channel, or anything like that
<clever>
LnL: the nix-channel man page explains it
<clever>
the channel should be a directory, containing nixexprs.tar.gz, and binary-cache-url
<catern>
clever: but... LnL's example above was adding a tarball directly
<LnL>
yeah, but I would assume you still have to add the signing key to your config
<catern>
rather than a URL
<clever>
catern: yeah, then it probably cant auto-detect the binary cache
athan_ has joined #nixos
turion has quit [(Ping timeout: 260 seconds)]
<LnL>
or is that not needed for channels from root or something
<clever>
sphalerite: i see, unshare is trying to access /proc from inside the chroot
jbgi has quit [(Ping timeout: 240 seconds)]
<clever>
sphalerite: i may want to run the unshare before the chroot then
MP2E has quit [(Read error: Connection reset by peer)]
MP2E has joined #nixos
stanibanani has joined #nixos
<clever>
sphalerite: but then i will need to be available on the host, hmmm
nschoe has quit [(Quit: Program. Terminated.)]
ma9e has joined #nixos
<clever>
sphalerite: rm -rf --one-file-system !!
<sphalerite>
clever: can't you unshare twice, once for mount namespace and again later for /proc?
<sphalerite>
err again later for pid namespace
<clever>
playing with it a bit more...
siddharthist has joined #nixos
<ma9e>
Hello, can anyone help me with setting up gitlab (or gogs) on 17.03? I have a lot of questions
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lheckemann opened pull request #26608: systemd: allow setting timezone via timedatectl (staging...timezone-pr) https://git.io/vHNQZ
NixOS_GitHub has left #nixos []
<sphalerite>
^ reviews from people who know about systemd and stuff would be highly appreciated!
<sphalerite>
Also about the nixos test suite
acertain has quit [(Ping timeout: 268 seconds)]
<siddharthist>
Is anyone else getting a "Curl error 35" when trying to download from the binary cache?
<clever>
sphalerite: ah, i see, "unshare --fork --pid --mount-proc" will create a mount namespace, and mount a new proc over /proc, but / is still the host
<siddharthist>
When I navigate to https://cache.nixos.org in Firefox, I get an error that says "SSL_ERROR_NO_CYPHER_OVERLAP"
acertain has joined #nixos
vaibhavsagar has joined #nixos
<sphalerite>
siddharthist: I just get a 404
<catern>
hey #nixos
<catern>
would it be a useful patch to the nix-channel logic, to handle using git to grab channels?
<catern>
or would this be bad because it would add a new dependency on git?
<spinus>
is it possible to run command inside "shell.nix" somehow? (I tried to 'build' shell with nix-build shell.nix (which is doing (callPackage ./bla.nix {}).env) but in that case I'm getting $src should point to the source error)
cpennington has quit [(Ping timeout: 255 seconds)]
civodul has joined #nixos
<sphalerite>
spinus: it is possible to run commands in various ways, but it depends on your usecase
<spinus>
sphalerite: works like a charm, thank you
<sphalerite>
oh so that's the one you wanted, good :D
<spinus>
I wanted to build the env and run the command manually
<spinus>
but going with --run kind of works as well
<spinus>
sphalerite: what would be the other way?
<spinus>
(just out of curiosity)
<sphalerite>
siddharthist: I have no issues downloading stuff from the cache. Probably a regional cloudfront issue or something
athan_ has quit [(Ping timeout: 260 seconds)]
<sphalerite>
spinus: if you want to run a command using nix-build, you can do something like pkgs.runCommand "name" { buildInputs = [...]; } ''some shell script which puts a file or directory in $out goes here''
<sphalerite>
If you use nix-shell on that expression you get the environment that that shell script would run it
<sphalerite>
in*
erictapen has joined #nixos
Guest14983 is now known as judson
<siddharthist>
sphalerite: thanks for checking, I appreciate it :)
<spinus>
sphalerite: I think I'm looking for the second option you described. basically, ghc compilation is slow, so I'm trying to implement alternative workflow with incremental compilation, but in the same env as original one
<spinus>
the second one you mentioned is nix-shell --run?
ryanartecona has quit [(Quit: ryanartecona)]
<sphalerite>
spinus: typical development workflow with nix, I believe, is to have a default.nix which can be built entirely using nix-build. You can also use nix-shell and then use stateful compilation tools that keep object files around so they don't need to be recompiled every time
<sphalerite>
I'm not sure about the exact details of how that works with haskell unfortunately, but presumably there's some sort of stack build command or so that you'd use in a nix-shell when developing actively
<sphalerite>
then you'd use nix-build/nix-env/nixos-rebuild/nixops for deployments/packaging
<spinus>
exactly what I'm trying to do now
cpennington has joined #nixos
<spinus>
sorry, it's not haskell specific I think. I just need to run command in the same env as nix-build does. so for now with nix-shell --run, I think I'm sorted, thanks again :)
<pxc>
what do I have to do to get an installable tarball (for a system that doesn't have Nix on it yet) when building Nix from source? `nix-build release.nix -A build.x86_64-linux` appears to me to just produce the Nix package for Nix
simukis has quit [(Quit: simukis)]
<pxc>
should I look at some Hydra job to recover that informatioN?
<clever>
sphalerite: cant seem to get nsenter to work right
<clever>
oh, maybe i also need -r
Filystyn has quit [(Quit: Konversation terminated!)]
<clever>
yeah, -r definitely did something good
<clever>
sphalerite: and *boom*, "systemctl poweroff" killed the host networking, and ctrl+c to systemd killed display-manager, lol
<sphalerite>
hahaha fun
<clever>
this is why i only ever play with containers in vm's
<sphalerite>
Probably makes sense to combine it with chroot x)
<clever>
which defeats the entire point of containers!
<clever>
[16401:16451:0615/142845.850724:ERROR:bus.cc(427)] Failed to connect to the bus: Failed to connect to socket /tmp/dbus-0MkGeEjdoV: Connection refused
<clever>
yeah, it borked dbus
<clever>
probably needs an ipc namespace also
<sphalerite>
I don't think so
<clever>
the container firewall also failed to come up right, because i didnt give it a network namespace
<avn>
Folks, I have small but complex problem -- I have some software, with absolute non-deterministic source (sftp server+dir). I already have script, which switch it to git repo, but I need to propagate it as nix derivation, with generations. What a best way to do it?
<clever>
but that would complicate getting internet
<avn>
Highly opinionated ideas welcomed ;)
<sphalerite>
ipc namespaces are for this weird system V ipc thing
<Dezgeg>
I think if you just disable dhcpcd you can rely on the host network namespace
<clever>
Dezgeg: yeah, just need to make the guest not mess with the network stack any
<Dezgeg>
also you could drop the net_admin priviledge
<sphalerite>
the relevant namespaces are the mount namespace (and chroot) for sockets that live in the filesystem, and the network namespace for abstract sockets
<clever>
sphalerite: i think a lack of ipc namespacing is what caused the crosstalk in the poweroff command
<sphalerite>
oh, that works via system V IPC?
<Dezgeg>
what about abstract unix sockets? what namespace they live in if not ipc?
georges-duperon has quit [(Ping timeout: 255 seconds)]
<sphalerite>
Dezgeg: network
<sphalerite>
22:34 < sphalerite> the relevant namespaces are the mount namespace (and chroot) for sockets that live in the filesystem, and the network namespace for abstract sockets
<Dezgeg>
hmm, when I tried lxc with the host networking it worked just fine, so probably something else is borked then
<sphalerite>
"IPC namespaces isolate certain IPC resources, namely, System V IPC objects (see svipc(7)) and (since Linux 2.6.30) POSIX message queues (see mq_overview(7))."
<clever>
browser is still crippling the machine as it starts up
<sphalerite>
nh2: siddharthist had the same issue earlier
<ekleog>
gchristensen: thanks :) btw, do you plan to also make the "Security fixes from" series numbered and published? (like DSA's or SSA's on oss-security) (well, when it resumes, of course -- I seem to remember you stopped it due to LWN's CVE database shutting down)
<sphalerite>
I couldn't reproduce it, I presume it's on cloudfront's end
<gchristensen>
uhoh nh2
<clever>
sphalerite: it might be spreading, check your end again?
<sphalerite>
nope, still works for me
<gchristensen>
CloudFront error perhapsp
<gchristensen>
ekleog: that would be ideal yes :)
<pxc>
sphalerite: I'm silly; I think I just needed to do binaryTarball.x86_64-linux instead of build.x86_64-linux
<clever>
[clever@amd-nixos:~/nix-tests/container-generator]$ nix-build -A test-guest --option binary-caches "" -Q
<ekleog>
:)
<sphalerite>
Is it just me or does cloudfront have a lot of issues?
<clever>
nh2: with the extra --option at the end, you can disable the binary cache temporarily
<nh2>
clever gchristensen: From my Singapore server it's working, from my European servers they all get the error
<ekleog>
(just noticed it'd be named NSA-2017-00xx ; this name is just great 8D)
<clever>
gchristensen: and i'm in canada
<gchristensen>
from my NY,USA IP it works from my MA,USA IP it doesn't
<mellowmaroon>
clever: I see. Related to my questions from yesterday, would it possibly be easier for me to use nix-shell to create a development evironment with both IDE and OpenGL packages?
<clever>
mellowmaroon: maybe, the mesa_noglu package should also let you compile/link against opengl
<mellowmaroon>
clever: Since I'm able to run my OpenGL test code from outside of the IDE, but not in it
<mellowmaroon>
clever: Ok, I'll look into it then. Thanks so much for the constant help :)
<clever>
ah, it might be that the presense of mesa_noglu is breaking the runtime stuff
<mellowmaroon>
clever: Should I make sure that's uninstalled first, you think?
<clever>
you shouldnt ever install libraries like that
<clever>
thats just not how nix is meant to be used
<mellowmaroon>
clever: Understood
mbrgm has joined #nixos
spinus has quit [(Read error: Connection reset by peer)]
magnetophon1 has quit [(Ping timeout: 260 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
jake_ has quit [(Quit: Page closed)]
jbaum98 has joined #nixos
<catern>
clever: anyway, thank you! you solved my problem :)
<kugelblitz2>
how can I use nixos.python35Packages.powerline within ~/.tmux.conf as source [...]/powerline/bindings/tmux/powerline.conf without the /nix/stopre prefix?
jbaum98 is now known as jake_
spinus has joined #nixos
<digitalmentat>
clever, a few paths were not able to be repaired with the binary-cache method but all of the other paths were, I believe
<clever>
digitalmentat: yeah, it will only be able to repair things that are on the nixos cache
<clever>
digitalmentat: next thing id check is nix-store --query --roots and nix-store --delete, see if you can delete any of the corrupt things
<clever>
and dont force --delete
<digitalmentat>
clever, yeah that's what I was going to try next
<digitalmentat>
the corrupted files appear related to the nix-env profile, which I rolled back
ambro718 has quit [(Ping timeout: 255 seconds)]
<clever>
digitalmentat: if --roots says that the only roots are in your profile, you can safely delete the unused generations (make sure not to delete the active one)
<digitalmentat>
yeah, it's not, I think this is fine as-is
<digitalmentat>
that was definitely helpful, thank you
<mellowmaroon>
Silly question: Is there a way to reset the generations counter?
<clever>
mellowmaroon: no proper way to do it, only by manualy messing with key symlinks
<mellowmaroon>
clever: That's fine, I don't plan on messing with it. Just curious
<clever>
mellowmaroon: you can also avoid creating generations in a few days
<clever>
ways*
<clever>
for nixos, there is "nixos-rebuild test"
hiratara has joined #nixos
<clever>
which will build and activate, but not create a generation, and the changes are undone upon reboot
<clever>
and for testing normal programs, i prefer nix-build over nix-env
<mellowmaroon>
Interesting. What's nix-build do?
<mellowmaroon>
I've only used nix-env to install new packages interactively, and delete older generations
pie_ has quit [(Ping timeout: 240 seconds)]
<clever>
nix-build '<nixpkgs>' -A firefox
<clever>
this will create a symlink called result, pointing to the firefox storepath
stallion has quit [(Quit: stallion)]
<clever>
so you can then ./result/bin/firefox to run it
<clever>
and when your done, delete the result symlink
<clever>
no generations, no firefox "installed" in anything
<mellowmaroon>
clever: neat :)
<clever>
and if your re-building the same package 200 times to work out a bug, no spam in your generation list
<mellowmaroon>
If I do nix-shell, it'll reuse the packages from my /nix/store, right?
<clever>
yeah
magnetophon1 has joined #nixos
<mellowmaroon>
I thought so, I was curious bc I tried something with nix-shell last week and was surprised that it was downloading a lot
<mellowmaroon>
more than I was expecting, at least
<clever>
-p or -A?
<mellowmaroon>
-A
kkourt has quit [(Quit: leaving)]
<clever>
nix-shell '<nixpkgs>' -A firefox, will give you a shell suitable for compiling firefox from source
<clever>
but nix-shell -p firefox, will give a shell with firefox pre-compiled
<mellowmaroon>
Ah ok, so I'll probably want -p?
<clever>
yeah
<mellowmaroon>
I guess it depends on the situation/software, but that's very useful info, thanks
jtojnar_ has joined #nixos
<clever>
yep
stallion has joined #nixos
jtojnar has quit [(Read error: Connection reset by peer)]
acertain has joined #nixos
jtojnar has joined #nixos
magnetophon1 has quit [(Ping timeout: 240 seconds)]
jtojnar_ has quit [(Ping timeout: 240 seconds)]
obadz has quit [(Ping timeout: 255 seconds)]
pie_ has joined #nixos
obadz has joined #nixos
<mellowmaroon>
Is there a way to have Nox show unfree packages? I don't think it shows the Oracle JDK
<clever>
mellowmaroon: have you already enabled unfree packages in config.nix?
mudri has quit [(Ping timeout: 240 seconds)]
<mellowmaroon>
In my /etc/nixos/configuration.nix, yep
<clever>
that doesnt effect nix-env and nox
<clever>
you also need to enabled it in the config.nix
<mellowmaroon>
Ahhh, got it. Is there a user-level customization I need?
<clever>
you need to set it in ~/.config/nixpkgs/config.nix
<clever>
if you try to build something unfree, it will tell you: nix-build '<nixpkgs>' -A google-chrome
<mellowmaroon>
Perfect, thanks
<clever>
once enabled, nox will probably find it
mudri has joined #nixos
<mellowmaroon>
is there something I need to refresh it? or do I only use `allowUnfree = true`, instead of `nixpkgs.config.allowUnfree...`
<mellowmaroon>
Tried reopening my terminal
<clever>
all nix tools will read it on start, no need to refresh them
<clever>
but nox will cache the package list
<jake_>
mellowmaroon: you only use allowUnfree = true;
<mellowmaroon>
thanks :)
<mellowmaroon>
is there a way to invalidate nox cache? it's no biggy, I just want to make sure it's good while I'm here
<jake_>
clever: is there anything i can do to help with the container thing
<clever>
jake_: taking a break right now, but i can commit what i have if you want to play with it
<clever>
jake_: its on my github now
<jake_>
clever: thanks. what is the general status of it at this point
<clever>
jake_: systemd starts, but i cant get nsenter to work, so no shells
<jake_>
gotcha
<clever>
jake_: you would need to setup sshd before you make the tarball, and they it might work
<clever>
then*
<clever>
also, every time i try to stop the systemd, it stops the wrong systemd
zeus_ has quit [(Read error: Connection reset by peer)]
<clever>
and breaks the host
<clever>
but that will likely behave differently if the host lacks systemd
kugelblitz2 has quit [(Quit: Leaving)]
zeus_ has joined #nixos
fnljk has quit [(Disconnected by services)]
fnljk_ has joined #nixos
fnljk has joined #nixos
magnetophon1 has joined #nixos
eacameron has joined #nixos
fnljk_ has quit [(Ping timeout: 240 seconds)]
markus1199 has joined #nixos
markus1219 has joined #nixos
<mellowmaroon>
For the nix-shell derivation for my development env, do I need to put the packages in `buildDepends`?
<mellowmaroon>
Tried making one, and was able to use nix-shell, but it didn't find libXxf86vm again, even though I listed it
<clever>
mellowmaroon: it has to go into buildInputs
<mellowmaroon>
Ok, I'll retry
<mellowmaroon>
Hmmm, same issue... is there anything else I need in the derivation? The only thing is that list inside a mkderivation
markus1189 has quit [(Ping timeout: 240 seconds)]
markus1209 has quit [(Ping timeout: 260 seconds)]
<clever>
mellowmaroon: can you gist the exact contents, and the command you ran?
<mellowmaroon>
Sure, exact command was `nix-shell -p` (file name is default.nix)
<mellowmaroon>
and I'll need a sec to post the file
orivej has joined #nixos
eacameron has quit [(Ping timeout: 260 seconds)]
<clever>
mellowmaroon: -p doesnt load default.nix
<clever>
you need to just run "nix-shell default.nix"
<mellowmaroon>
Oh, I see
<mellowmaroon>
Got it
<mellowmaroon>
do I need something before mkDerivation? That's what nix-shell's telling me
<clever>
with import <nixpkgs>{}; stdenv.mkDerivation { ...
marsam has quit [(Remote host closed the connection)]
MP2E has quit [(Ping timeout: 240 seconds)]
<mellowmaroon>
The error remains, when I try running the gradle task in IDEA, it can't find libXxf86vm
<mellowmaroon>
I was able to make the derivation, at least
<clever>
not sure what gradle would be doing differently to mess things up
<mellowmaroon>
all the packages are under buildInputs...
<mellowmaroon>
It may be easier for me to try to get things working globally, since I'm even more of a novice with nix-shell
<clever>
installing the libraries wont help anything
<clever>
nix goes out of its way to not include headers and such when you install a package
<mellowmaroon>
I just need to figure out what isn't visible within IDEA, since I'm able to do `gradle run` in a normal shell and have it work
<clever>
was idea started from inside the nix-shell?
<mellowmaroon>
yep
<clever>
not sure then
<mellowmaroon>
I put in every package I could think of related to buildInputs, and didn't work...
<clever>
its possible that idea is trying to clean the env up before running gradle
<clever>
and thats breaking everything
<mellowmaroon>
Any other ideas on what I could try? Otherwise, I'll probably just run it outside of the IDE
justelex_ has quit [(Remote host closed the connection)]
<clever>
cant think of anything else at the moment
<mellowmaroon>
That's fine, thanks again for the help :)
<mellowmaroon>
With something like IDEA, I wouldn't be too surprised if it's just something hard-coded that shouldn't be
MP2E has joined #nixos
<judson>
fetchFromGitHub fetchSubmodules doesn't appear to be working for me.
jsgrant has joined #nixos
erasmas has quit [(Quit: leaving)]
<jsgrant>
I was trying to install NixOS to my "final respite" from full-on, on all my boxes; An ultrabook I've had GNOME on. Attempted to do a similar install via 17.03 & unstable -- and I'm getting some weird s-trace (assuming) kernel spaz; Is there any guide how to debug this anywhere?
<jsgrant>
Or should I just wait around for a release on Fedora, until I better know what I'm doing?
newhoggy has joined #nixos
<jsgrant>
Right now, think latter is more likely.
<clever>
jsgrant: what kind of weirdness is it doing?
<jsgrant>
But have an ant up my butt to go "all in".
<jsgrant>
clever: Very unreliable booting. Sometimes, not a problem (sans display-mananger refuses to start); Sometimes a near-instant kernel-panic.
<clever>
jsgrant: what does nix-channel --list say?
<clever>
when ran as root
<jsgrant>
'/nixos-unstable'
<clever>
jsgrant: have you ran memtest86 on the hardware recently?
<jsgrant>
Relatively new ultrabook, so I assumed something in upstream kernel just wasn't supported (remember having various little issues last release on Fedora (24) (namely servies and not kernel)) -- but have no idea.
fnljk has quit [(Disconnected by services)]
fnljk_ has joined #nixos
<jsgrant>
clever: I know pathetically little atm, about lower-level debug, even though it's an area I'm very interested in getting into.
<jsgrant>
Yeah, I really should/"need to" look into this a lot more. Sans my phone/tablet, pretty much everything else installable-on has NixOS on at this point. This laptop, desktop, htpc, homeserver. Might as-well go "all in"...
<jsgrant>
clever: What I wonder though, is why is it working "perfectly" on Fedora25? I'm guessing there's some inconsistent kernel option/flag. :^P
mellowmaroon has quit [(Quit: Leaving)]
<clever>
jsgrant: would need to compare the kernel versions
<jsgrant>
clever: Is NixOS using upstream? I know Fedora rolls there own.
<jsgrant>
their own*
mudri has quit [(Ping timeout: 240 seconds)]
* jsgrant
is just going to have to look into it more; Just think I have the aformentioned "ant up my butt" after seeing Fedora 26's Beta just dropped & I had a loose goal of "Converting this box over, by the next major release".
roconnor has joined #nixos
phinxy has quit [(Read error: Connection reset by peer)]
zeus_ has quit [(Remote host closed the connection)]
pie_ has quit [(Ping timeout: 255 seconds)]
zeus_ has joined #nixos
pie_ has joined #nixos
zeus_ has quit [(Ping timeout: 258 seconds)]
Rotaerk has joined #nixos
<catern>
hey #nixos
<catern>
do I need to use channels at all?
<catern>
can I just make ~/.nix-defexpr a regular folder, and put nixpkgs in there?
<clever>
you can also just modify $NIX_PATH
<catern>
and git clone nixpkgs in there*
<catern>
clever: huh? but nix-env uses ~/.nix-defexpr by default
<clever>
yeah, nix-env is the only tool that uses nix-defexpr
<clever>
but you can still do nix-env -f '<nixpkgs>' -iA hello
<catern>
ah
<catern>
I see
<catern>
so all nix-channel really does, is affect the behavior of nix-env?
<clever>
yeah
<clever>
and the default $NIX_PATH is set to a dir that nix-channel outputs to
mudri has joined #nixos
<catern>
hmm, you can't even use it to keep up your NIX_PATH up to date, right?