<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fAddi
Maxdamantus has quit [Quit: leaving]
Maxdamantus has joined #nixos
Fare has joined #nixos
haslersn has joined #nixos
<haslersn>
clever: awesome, your advice worked
<clever>
once it boots like that, you can try switching the vars back around, and confirm it worked with efibootmgr -v
jtojnar has joined #nixos
haslersn has quit [Ping timeout: 256 seconds]
haslersn has joined #nixos
<haslersn>
clever: I did that and it still works, but what effect does it have? (the switching back)
<clever>
its more proper i believe
<clever>
it will also give the OS a cleaner name in the firmware, when you are selecting between EFI based OS's
<clever>
if the firmware is even capable of that
<haslersn>
The name is still the same. I think the rebuild after switching back didn't change anything in my efi partition
<clever>
also double-check that /boot is mounted when running nixos-rebuild
<clever>
oh, and you may want nixos-rebuild switch --install-bootloader
<haslersn>
ah
<haslersn>
I missed that
Kelppo has joined #nixos
astronavt has quit [Ping timeout: 272 seconds]
haslersn_ has joined #nixos
<haslersn_>
clever: Sorry, I couldn't read your last message before I left.
<clever>
2018-09-22 21:55:28 < clever> oh, and you may want nixos-rebuild switch --install-bootloader
<clever>
2018-09-22 21:55:38 < haslersn> ah
<clever>
thats the last thing i said
<haslersn_>
clever: Anyway, I rebuilt with --install-bootloader and the new bootloader has been installed but my UEFI sees only the old one (which is called HDD1)
haslersn has quit [Ping timeout: 256 seconds]
<haslersn_>
Maybe my UEFI is too old
commander has quit [Ping timeout: 252 seconds]
commander has joined #nixos
haslersn_ has quit [Ping timeout: 256 seconds]
haslersn has joined #nixos
<haslersn>
Thanks for the awesome help
<clever>
yep
__lawlesseel__ has joined #nixos
lawlesseel has quit [Remote host closed the connection]
astronavt_ has joined #nixos
wpcarro has joined #nixos
EternalZenith has joined #nixos
astronavt__ has joined #nixos
<EternalZenith>
I seem to somehow have messed up my xorg
astronavt_ has quit [Read error: Connection reset by peer]
<EternalZenith>
.xorg.log is filled with "Failed to load module "libinput" (module does not exist, 0)"
<EternalZenith>
I have libinput.enable = true; in my configuration.nix, and I tried installing xf86-input-libinput and libinput just in case
<clever>
installing things wont fix it
<clever>
they have to be correctly added to the nixos module for xorg
<EternalZenith>
How should I go about that?
<EternalZenith>
I assume the display managers were performing that before I disabled them
jedahan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @andrew-d opened pull request #47202 → erlangR18: apply upstream patches to fix nondeterministic build failures → https://git.io/fAdNg
wpcarro has joined #nixos
abueide has joined #nixos
jedahan has joined #nixos
wpcarro has quit [Ping timeout: 240 seconds]
jedahan has quit [Ping timeout: 240 seconds]
bbarker has joined #nixos
hiratara has joined #nixos
<bbarker>
is it possible to get the directory of the nix expression being executed by nix-shell inside of a shellHook?
<samueldr>
not sure, but is there a specific goal in mind? in case there are other ways to achieve the goal
<Myrl-saki>
bbarker: `shellHook = FOO=${./.}` ?
<Myrl-saki>
I forgot some quotes there.
<bbarker>
samueldr; I have a few scripts that I'd like to be put on the PATH in the nix expression. I guess in my heart I know the right way to do this is to create a derivation, but there seems to currently be a bug with hashing src=file:// etc
<bbarker>
Myrl-saki, oh, I assumed that might be the directory of wherever `nix-shell` is being run from, which could be different - i'll check
<samueldr>
./ in a nix expression is always relative to the file itself
<bbarker>
oh, awesome
<samueldr>
definitely
<Myrl-saki>
I think if I'm working on nix, then one thing that I wouldn't want to be assigned to is file paths.
<Myrl-saki>
Not because they're confusing, but because they work.
<bbarker>
Myrl-saki, you mean, assigning to PATH is bad? I usually only do this in a nix-shell where it sometimes seems handy
<bbarker>
and doesn't affect the rest of the system outside the shell of course
<Myrl-saki>
bbarker: Noo, I mean file paths.
<Myrl-saki>
Not the envar PATH.
<Myrl-saki>
For example, `--arg foo ./.` with nix-shell will properly use your current directory.
<Myrl-saki>
How the hell does it do that?
silver has quit [Read error: Connection reset by peer]
<samueldr>
(not sure if the "how the hell" was rethorical, but digging in the source is fun)
<bbarker>
lol
<bbarker>
I'm curious if anyone can build plan9port, in order to maintain a semblance of the dream set forth by our forefathers: https://github.com/NixOS/nixpkgs/pull/47067 Could be this works on NixOS; I'm trying on Ubuntu.
<bbarker>
Seems to either be a case of "what once built now does not" or "I'm somehow doing it wrong", because binary fetched in normally from the cache works as desired
<samueldr>
bbarker: "if anyone can build" with your PR or on master?
<bbarker>
samueldr, either - I just verified the results are the same
<bbarker>
at least, on my system
<bbarker>
the build appears to complete but many binaries are missing and there are many worrisome messages of "wrong ELF type"
<samueldr>
I don't know about binaries missing and wrong ELF type yet
<samueldr>
but plan9port things I think need the `9 ...` invocations
<samueldr>
but I'm kinda expecting you already know
<bbarker>
interesting; in my shell I had just done a temporary hack: export PATH=$PATH:${plan9port}/plan9/bin:${./bin}
<bbarker>
which seemed to work as well
<bbarker>
but, one difference is on the cached binary you will find e.g. :${plan9port}/plan9/bin/acme exists, but does not exist in the build (at least not when I build it)
<samueldr>
9 seems to do that :)
<samueldr>
(building)
<samueldr>
I wouldn't be surprised that the "wrong ELF types" come from 9c, but I don't know enough about plan9port to be sure
<samueldr>
I know though a bit about plan9 and how it's definitely its own non-posix beast :)
<bbarker>
it takes a while, alas, I probably will be gone by the time it finished but will check back later
<samueldr>
hmm, too bad it isn't parallelized
<samueldr>
though, can you tell me one thing that fails with the new build, and doesn't with the old?
<samueldr>
so I can at least have a known value to check?
wpcarro has quit [Ping timeout: 252 seconds]
<bbarker>
samueldr sure, and sorry if I'm repeating myself here, but plan9/bin/acme is missing when i try. it is a graphical editor, so you need X to actually try running it
<bbarker>
but as I said it is totally missing for me in the build
<samueldr>
oh right, I was thinking "other than acme" :)
<samueldr>
hmmm
<samueldr>
bbarker: with your WIP commit, rebased on nixos-unstable, I have acme up and running
<samueldr>
nix-build -A plan9port && ./result/bin/9 acme
slack1256 has quit [Ping timeout: 252 seconds]
<bbarker>
samueldr, that was fast ... well that's good news at least
* samueldr
thought the build took a while
<bbarker>
samueldr haha ... well, I guess I should just update the PR tomorrow and note that I haven't tested it but you have? I'm not sure I have ready access to a proper NixOS system. but, I should try your build command instead of directly invoking nix-env (though I can't imagine that should make a difference)
<bbarker>
i'm not sure what % of plan9port is used by acme, but, I think it is a lot
<bbarker>
I think Russ Cox made plan9port so he could use acme on Linux at Google
<samueldr>
yeah, nix-env -iA using -f like you did should be fine
mayhewluke has quit [Ping timeout: 260 seconds]
mayhewluke has joined #nixos
<bbarker>
samueldr, thanks again for giving that a go
<samueldr>
you're welcome, thanks for improving nixpkgs :)
<alex``>
> .rwxr-xr-x 11k alex 18 Sep 18:24 /home/alex/repositories/github.com/MarianArlt/kde-plasma-chili/Main.qml
<{^_^}>
error: syntax error, unexpected '.', at (string):195:1
<srhb>
alex``: Maybe something to do with the file:// prefix?
<alex``>
I don’t give any prefix
<alex``>
It’s a valid path (URL) I guess anyway
<tilpner>
My sddm.theme is an absolute path into the store
<srhb>
Right, but that error message looks like the error you'd get from something that interprets it as a unix path, not a URL.
brodul has joined #nixos
<{^_^}>
[nixpkgs] @andrew-d opened pull request #47209 → hdf4: fix build on aarch64-linux → https://git.io/fAFeS
vcunat has joined #nixos
<siers>
Is emacs fully configurable through nix? I want to try emacs, but I don't want to delve into the config much yet, so maybe I could try someone else's.
Fare has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed 8 commits to release-18.09: https://git.io/fAFvf
<{^_^}>
#34409 (by steveeJ, 33 weeks ago, open): qemu-vm: fix script syntax and VM args
init_6 has quit []
<{^_^}>
[nixpkgs] @vcunat pushed commit from @andrew-d to release-18.03 « libsndfile: Add patch for CVE-2018-13139 (#47160) »: https://git.io/fAFfM
yesbox has joined #nixos
periklis has joined #nixos
<steveeJ>
I somehow want to run the script which is produced by vmWithBootLoader within the test runner and control that VM as others can be controlled within tests
Xiro` has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
Xiro` has joined #nixos
<yesbox>
hi. i think it would be useful to use the hash of a package inside that package, specifically to add the hash to a programs "--version" and the like, in some cases. is it feasible to access the hash at that point or is this crazy talk? any existing examples?
<manveru>
yesbox: it's basically part of $out
thc202 has joined #nixos
periklis has quit [Ping timeout: 240 seconds]
<infinisil>
yesbox: If you have some src = fetchgit or so declared, it's probably better to take the hash from the git revision
<{^_^}>
[nixpkgs] @peti pushed 6 commits to haskell-updates: https://git.io/fAFUb
b has joined #nixos
Lears has quit [Remote host closed the connection]
Xiro` has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
<sphalerite>
if I'm booting nixos with a custom kernel, built independently of nixos, how can I get udev (if it's udev's responsibility? or whatever else is responsible) to autoload modules?
mayhewluke has quit [Ping timeout: 272 seconds]
<Dezgeg>
it's not udev, the kernel just launches modprobe when needed
perrier-jouet has joined #nixos
<perrier-jouet>
hi all
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fAFTI
<perrier-jouet>
I would like to know what framework or tools and templates is used to build the nixos website
mayhewluke has joined #nixos
<steveeJ>
where does 'system' within nixos get populated with 'build'? builtins.currentSystem is just a string
<steveeJ>
I'm trying to understand how to pass different modules and pkgs to a built VM than what's used for the build system
<steveeJ>
at first look it doesn't seem to be designed for it
<sphalerite>
Dezgeg: damn. Is there a way to tell it to pass a perticular -d to it? How does nixos do it? And is there a way to get it to do that again after it's booted?
<srhb>
steveeJ: Not sure what you're asking. <nixpkgs/nixos> takes a system argument which is just a string and has a system.build attribute
<srhb>
steveeJ: It also takes a configuration argument.
<infinisil>
steveeJ: system.build is just a general option of attrsets for the system build. It gets set in many modules and used in many modules
<{^_^}>
[nixpkgs] @symphorien opened pull request #47212 → openvpn plugins: use absolute paths to configuration gui → https://git.io/fAFTB
<Streetwalrus>
hi, what is the preferred way to enable systemd @ templates? (template@instance.service type units)
<srhb>
Streetwalrus: Generally we prefer not to use them at all, really. They don't make much sense in a declarative environment.
Xiro` has joined #nixos
<Streetwalrus>
fair enough
<Streetwalrus>
thanks
<perrier-jouet>
srhb: is that a bootstrap template ?
<srhb>
perrier-jouet: Never looked at it. :)
<srhb>
Streetwalrus: If you're trying to work on some specific service, perhaps people can suggest how to set it up. :)
<sphalerite>
steveeJ: yeah I think your first look is correct. The only way to get different modules/pkgs is to import a different <nixpkgs/nixos> AFAIK
<srhb>
Streetwalrus: (as an example, the only one we have by default is getty@, I believe)
<yesbox>
manveru infinisil : thanks. i want it to match the nix store hash though, so that the hash tells you if anything is different in the build (extra patches for example, which wouldn't change the hash of the fetched git rev) and being able to compare it. so that would be $out, which makes perfect sense now that you're saying it.
<srhb>
Streetwalrus: A simpler example is the imperative container template unit in nixos/modules/virtualisation/containers.nix -- if you need similar functionality. But again, it's usually nicer and easier to generate multiple units declaratively, without using the templating system :)
<Streetwalrus>
srhb: it's something custom, we're moving some things from another distro where we had multiple scripts using a shared environment, so systemd templates allowed us to enable them all with a single unit file
<srhb>
Ah, yeah
<srhb>
Streetwalrus: Consider whether you can just generate all the units declaratively first. It's usually way nicer. :)
<Streetwalrus>
yeah that's what we're going to do
orivej has joined #nixos
smolboye has quit [Ping timeout: 252 seconds]
smolboye_ has joined #nixos
<sphalerite>
Dezgeg: ah, looks like I could set up a wrapper and tell the kernel to call it by writing to /proc/sys/kernel/modprobe. Is there also a way to get it to try and reload everything?
smolboye_ has quit [Remote host closed the connection]
Xiro` has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
zli has joined #nixos
sphalerite has quit [Quit: WeeChat 2.0]
<zli>
Hello all.
smolboye has quit [Client Quit]
Xiro` has joined #nixos
<zli>
How to build emacs with the m17n otf and xft support? Thanks in advance.
smolboye has joined #nixos
sphalerite has joined #nixos
<sphalerite>
um. Wtf. My tmux just crashed.
Xiro` has quit [Client Quit]
<zli>
sphalerite : that is certainly strange indeed.
<sphalerite>
it's also a real pain
orivej has quit [Ping timeout: 244 seconds]
patrl has quit [Ping timeout: 264 seconds]
jtojnar has quit [Remote host closed the connection]
<tilpner>
zli - Are you on Darwin/macOS?
<tilpner>
On Linux, m17n and xft should be enabled by default
<zli>
tilpner: I am on linux. I am trying to get some fonts to work with emacs. they work fine on gedit / kate etc.
<tilpner>
How do you know your emacs is built without those?
<etu>
Hmm, is it just me or have nix-mode stopped being autoloaded by emacs whenever opening a .nix file? I never had to specify this before. But recently on nixos-unstable it opens the files with fundamental-mode.
<steveeJ>
oh yeah that worked, thanks! at least I'm getting a different error now ;-)
<alex``>
Thanks
<{^_^}>
[nixpkgs] @mroi opened pull request #47214 → podofo: fix library linkage on Darwin → https://git.io/fAFIw
<sphalerite>
steveeJ: but yeah you don't need the { ... } @ bit, if you're not extracting any attrs from the set explicitly there's no point in it :)
<alex``>
tilpner: How do you add this file to your /etc/nixos/configuration.nix?
<gchristensen>
siers: if you find more, send a PR :) :D
<tilpner>
alex`` - No, I use an overlay
<alex``>
in the package list, you refer then to it as:
<alex``>
[
<alex``>
...
<alex``>
solarized-sddm-theme
<alex``>
?
<steveeJ>
sphalerite: gotcha. I didn't know if it would work like that. not having types makes me a bit insecure
<tilpner>
alex`` - I don't refer to it in my package list at all
<alex``>
tilpner: I don’t know about overlay
<tilpner>
,overlay
<{^_^}>
Overlays look like `self: super: { foo = ...; }`. Use the self argument to get dependencies, super for overriding things and library functions (including callPackage). More info: https://nixos.org/nixpkgs/manual/#sec-overlays-install
<tilpner>
But they're not necessary to get this working, just pretty
<tilpner>
I could also write theme = builtins.toString (callPackage ./theme.nix {});
<ixxie>
This is kinda off topicish but since this community is all into declarative configuration I must ask: does anybody know how to declaratively configure weechat?
<steveeJ>
sphalerite: wohoo, I wish I could say that about the errors :D
<steveeJ>
the assertion `assert args ? localSystem -> !(args ? system || args ? platform);` wouldn't this assertion also fail if none were passed explicitly?
<sphalerite>
steveeJ: no, since (args ? localSystem) would be false, leading to false -> foo which is always true
<ixxie>
gchristensen: good to know, but I was looking more configuration other than plugins
<{^_^}>
attribute 'foo' missing, at (string):195:20
<steveeJ>
sphalerite: gotcha, thanks. then I'm surprised as I'm not passing any of it.
<tilpner>
ixxie - Mine does configuration and scripts. It generates multiple files from one config attrset
<tilpner>
Which can then be passed to weechat at startup
<sphalerite>
steveeJ: use lots of builtins.trace? :D
<steveeJ>
it probably gets set somewhere along the way in an incompatible way
<gchristensen>
yep.
<gchristensen>
siers: almost all of them, but there are probably some stragglers.
<siers>
gchristensen, how come you haven't caught them all? :) couldn't you "catch 'em all" with grep '#idm'?
<steveeJ>
sphalerite: I have to read up on builtins.trace, haven't heard that before
<sphalerite>
steveeJ: print statements for nix! x)
<ixxie>
tilpner: I can see, looks neat!
silver has joined #nixos
<tilpner>
ixxie - I can provide makeBubblewrap to you if necessary, but it's not required for this approach. Disadvantages of doing it this way: painstaking initial conversion of existing config to config.nix, requires rebuild *AND* restart on every config change, interactive changes are lost (duh, but sometimes annoying)
<tilpner>
The rebuilds are okay-ish for me, as I had an imperatively found config that I was okay with. Now I only make changes to add/remove networks
<steveeJ>
sphalerite: it would also help if the assertion would print the results instead of saying "failed"
<sphalerite>
steveeJ: not really, since the result is "false" :p
<steveeJ>
but the details
<sphalerite>
as far as assert is concerned, there are no details ;)
<sphalerite>
but if you just put a builtins.trace around the equality test you can have the values printed out
__lawlesseel__ has quit [Remote host closed the connection]
__lawlesseel__ has joined #nixos
<ixxie>
tilpner: I don't have an existing configuration; just thought to switch now
<ixxie>
tilpner: maybe - if you have no problem with it - I could just snatch your configs and tweak them to my liking? :)
<tilpner>
ixxie - I suggest you start with mine then, and adapt it as you see fit... yes, exactly
<ixxie>
in that case I could probably use makeBubblewrap?
<tilpner>
You can use /fset to interactively test changes you make
abueide has quit [Ping timeout: 252 seconds]
<tilpner>
You could probably get away with a writeShellScriptBin "irc" ''${weechat} -c ${configDir}'' and installing that
<tilpner>
I use makeBubblewrap to sandbox weechat. It adds complexity to the whole setup that you may not find worth the effort
<steveeJ>
sphalerite: so somehow localSystem and system are both passed to the "impure.nix"
<tilpner>
Now that I've played with generating apparmor profiles, it might not even be necessary anymore
<tilpner>
sphalerite - Unimportant to me, I keep them on my ZNC. But it would be easy to link them out with bubblewrap too
<sphalerite>
right
<sphalerite>
steveeJ: is the culprit possiblt nixos/modules/misc/nixpkgs.nix ?
Xiro` has joined #nixos
<steveeJ>
sphalerite: I haven't included that in the modules list
<sphalerite>
steveeJ: but it's in baseModules
<steveeJ>
sphalerite: I'm struggling to understand what's different between importing nixpkgs/nixos/default.nix and the expression I wrote. it calls out to eval-config the same way
<steveeJ>
I could try to add my config in the tree and see if it's the same
<ixxie>
thanks tilpner!
<ixxie>
tilpner: is that a self-hosted pastebin?
<tilpner>
Yes, I wanted short links to pronounce on telephone calls/voice-chat
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « pgtap: 0.98.0 -> 0.99.0 (#46980) »: https://git.io/fAFtX
<ixxie>
tilpner: I will try setting it up this evening, thanks for the configs
iyzsong has quit [Ping timeout: 245 seconds]
iyzsong has joined #nixos
hamishmack has joined #nixos
<{^_^}>
[nixpkgs] @flokli opened pull request #47220 → php: set mysql socket path if mysql[i] or pdo_mysql support is enabled → https://git.io/fAFtp
<ixxie>
anybody got powerline-go running on ZSH in NixOS?
<jtojnar>
ixxie: regarding xmonad+gnome, you cannot use the normal GNOME, but it should work with Flashback
<ixxie>
I can't seem to get the configuration right
<ixxie>
jtojnar: I was mostly trying to find out how to select it in the GDM, but now I found it. I think I will try to switch to xmonad in the comings months but will wanna have gnome available as a fallback / for guests
tertl3 has quit [Quit: Connection closed for inactivity]
zli has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<{^_^}>
[nixpkgs] @Mic92 opened pull request #47222 → autoPatchelfHook: do not patch statically linked files → https://git.io/fAFYl
Thra11 has joined #nixos
jtojnar has quit [Quit: jtojnar]
jtojnar_ has joined #nixos
iyzsong has quit [Ping timeout: 252 seconds]
Lears has joined #nixos
rfold has joined #nixos
ixxie has quit [Ping timeout: 244 seconds]
<d1rewolf>
all, I have "(python2.withPackages(ps: with ps; [ dbus ]))" in my configuration.nix, but "import dbus" fails within the python repl. Any ideas what i'm missing?
<manveru>
really should change the php derivation to allow a custom php.ini without rebuilding the whole language :|
<Myrl-saki>
d1rewolf: Try [ps.dbus]
<Myrl-saki>
It won't solve your problem, but there's probably at least 3 packages named dbus in nixpkgs.
<Myrl-saki>
[myrl@myrl:~/Development/fbcli]$ nix run nixpkgs.pythonPackages.dbus-python
<Myrl-saki>
nixpkgs.pythonPackages.dbus-python
justanotheruser has joined #nixos
<Myrl-saki>
d1rewolf: Yep. Knew it. That's global dbus, not dbus-python.
<d1rewolf>
Myrl-saki: that tells me "error: attribute 'dbus' missing"
<Myrl-saki>
d1rewolf: Yeah, you were using pkgs.dbus.
<Myrl-saki>
d1rewolf: Use dbus-python in your [] instead.
<d1rewolf>
ps.dbus-python or just dbus-python?
<Myrl-saki>
dbus-python
<Myrl-saki>
Well either works.
mayhewluke has quit [Ping timeout: 240 seconds]
<d1rewolf>
Myrl-saki: hmmm...no error on rebuild, but "import dbus" still fails
<Myrl-saki>
Probably a wrong import.
<Myrl-saki>
Let me give it a try.
patrl has quit [Ping timeout: 260 seconds]
mayhewluke has joined #nixos
<Myrl-saki>
Works for python36
<Myrl-saki>
Trying again with python2
<Myrl-saki>
d1rewolf: Works for me.
<d1rewolf>
Myrl-saki: rly? with python2?
<sphalerite>
manveru: uh, does it not?
ronin7z has joined #nixos
<ronin7z>
hii there
<ronin7z>
was thinking of giving nixos a shot, just got one question
<Myrl-saki>
sphalerite: Maybe it started getting built recently?
<d1rewolf>
Myrl-saki: they're both 2.7.15, but "which python" shows a different path with external compared to nix-shell. However, isn't that the nature of running it with nix-shell (always a different path)?
<Myrl-saki>
Either way, wineWoW + Staing is still not in bincache, I believe.
endformationage has joined #nixos
<manveru>
ronin7z: using wine via steam atm, seems to work perfectly
<ronin7z>
you mean proton?
<sphalerite>
,library d1rewolf
<{^_^}>
d1rewolf: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<manveru>
yeah
<ronin7z>
yeee proton mostly works okay for me
<Myrl-saki>
sphalerite: I think he's doing it properly.
<ronin7z>
but i use wine-staging for world of warcraft and Overwatc h
<Myrl-saki>
sphalerite: d1rewolf | all, I have "(python2.withPackages(ps: with ps; [ dbus ]))" in my configuration.nix
<d1rewolf>
sphalerite: Myrl-saki: right. and I get no errors upon rebuild
<sphalerite>
oh
<ronin7z>
hmm do i go with minimal install cd
<manveru>
ronin7z: fair enough :) i think that works too... i used to play HotS in wine a few years ago
<ronin7z>
or with the kde thingy
<sphalerite>
ronin7z: yes. The graphical install CD is kind of silly, because there's no graphical installer :p
<ronin7z>
does compiling/installing everything take long?
<sphalerite>
no, it doesn't take long. Not much compiling to do usually.
<manveru>
depends on your bandwidth
<sphalerite>
oh yeah true
<Myrl-saki>
Well.
<sphalerite>
your bandwidth and which software you've got in your config
<Myrl-saki>
TBF, the squashing takes 15 mins easily.
<ronin7z>
so whats the point of the graphical install cd?
<Myrl-saki>
On my i5-5005U.
<ronin7z>
just that you have a functional enviroment
<ronin7z>
while installing?
<sphalerite>
ronin7z: having a graphical partition editor I suppose
<Myrl-saki>
Oh wait.
<Myrl-saki>
I thought he was asking about building minimal install CD. Disregard. :P
<srhb>
ronin7z: A proper browser can also be nice.
<srhb>
Personally, I prefer a graphical browser but a CLI partition editor.
<ronin7z>
yeee
<sphalerite>
oh yeah, if you've only got the one computer handy
<Myrl-saki>
srhb: You mean w3m is not a proper browser???
<ronin7z>
would be nice for reading installation guides etc
rfold has quit [Quit: leaving]
<srhb>
Myrl-saki: Sorry, i should have used "modern new-fangled"
<Myrl-saki>
:P
mkoenig has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @erictapen closed pull request #45889 → buildRustPackage, fetchcargo: optionally use the real config from cargo vendor → https://git.io/fAWa0
<ronin7z>
oh right
<sphalerite>
ronin7z: well the nixos manual is included in the CD and accessible on tty6 so you won't need the graphical stuff for that really
<ronin7z>
what init does nixos use?
<sphalerite>
systemd
<ronin7z>
that was a quick answer
<sphalerite>
easy question, quick answer ;)
Izorkin has joined #nixos
mkoenig has joined #nixos
<sphalerite>
manveru: seems like a limitation on the phpfpm module, not so much the package
<manveru>
sphalerite: right now i have a postInstall overrideAttrs for php that replaces it...
<manveru>
what's a better option there? :)
<Myrl-saki>
I wonder how many result files I have lol
<Myrl-saki>
d1rewolf: Something might be putting python in your PATH.
acarrico has joined #nixos
ronin7z has quit [Ping timeout: 256 seconds]
<d1rewolf>
Myrl-saki: you mean make that line look like "(ps: with ps; [ dbus-python ])"?
<Myrl-saki>
d1rewolf: Yes, delete the whole line.
<d1rewolf>
Myrl-saki: wait....delete the whole line or just the python2.withPackages part? Or are you saying you're trying to hunt down what other .nix config might be adding python?
<Myrl-saki>
d1rewolf: The latter.
<d1rewolf>
k. one sec
<d1rewolf>
it's still there after rebuild. Let me pastebin a grep of python of all my .nix files
<{^_^}>
[nixpkgs] @pvgoran opened pull request #47224 → nixos/tomcat: add aliases sub-option for virtual hosts → https://git.io/fAF3W
alex`` has joined #nixos
<Myrl-saki>
d1rewolf: You can use nix why-depends..
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-18.09: https://git.io/fAF3A
klntsky has joined #nixos
<nh2>
domenkozar: after `cachix push`, should the pushed bits be immediately available for downloaders or are there caches at play that may need to be refreshed (either at cachix itself or the user's nix)?
<nh2>
e.g. does nix "cache" cache info temporarily?
<Myrl-saki>
Yes.
<srhb>
nh2: Assuming this is using regular substitution mechanics, yes, nix does cache the results
<srhb>
nh2: ~/.nix/cache/binary-cache* for instance
<Myrl-saki>
[myrl@myrl:~]$ ls ~/.cache/nix/binary-cache-v5.sqlite
<alex``>
Who use custom shells are their main shell here?
<alex``>
like Elvish
<alex``>
I have some trouble having it as main shell
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @flokli to release-18.09 « php: set mysql socket path if mysql[i] or pdo_mysql support is enabled »: https://git.io/fAFGR
<alex``>
when logging without display manager, PATH is not set like it is for Bash
<gchristensen>
Ericson2314: looks like just a missed patch?
<Ericson2314>
the release notes are out of sync
<Ericson2314>
missed revert
<srhb>
Ah, yes.
<srhb>
Looks like just one commit.
<Ericson2314>
the 2.0 change is only on master
<srhb>
Yeah, it was reverted for 18.09
<Ericson2314>
I was going to make a PR to move release notes for 19.03 for 2.0
<Ericson2314>
that seem correct?
patrl has joined #nixos
<srhb>
The previous discussion does seem to imply that we can certainly do the bump in 19.03
<Ericson2314>
indeed
<{^_^}>
[nixpkgs] @grahamc pushed commit from @matthewbauer to master « Revert "nixos docs: add release notes for nix 2.0 requiremnt bump" »: https://git.io/fAFGr
<gchristensen>
nbd :)
<Ericson2314>
:)
<Ericson2314>
thanks
<srhb>
Ericson2314: Since you're here, what on earth is going on with 344e6e229c29347db5a7fab07ee8392387adfc17 and 45f40f5fe02d71f9863e54277a50fe7e1dc46afa ? :P
<Ericson2314>
srhb: that's the downside of my merge-base trick
<Ericson2314>
there were two most-common ancestors, so I merged them and then merged that into each branch
<Ericson2314>
making one common ancestor
<srhb>
Ah, I see. I was super mystified. :P
<Ericson2314>
(the trees were not changed with those merges)
<Ericson2314>
I've had to do so much immediate-to-18.09-and-master small changes for reflex-platform
<srhb>
nh2: Though the default cache is just one hour for negative lookups
<srhb>
nh2: One month for positive, iirc.
<srhb>
Which is tons of fun if your cache is garbage collected often :P
<srhb>
Though I think there was a PR to recheck if there's a failure on a positive cache hit, which used to be a hard build error.
<srhb>
Might already be in 2.1
Thra11 has quit [Ping timeout: 252 seconds]
<Ericson2314>
hmm now the 19.03 release notes are missing the breaking changes section
<Ericson2314>
there are also formatting differences
<gchristensen>
run `make format` on both
* Ericson2314
finds makefile :)
<nh2>
srhb: OK, I improved the wording to reflect that
<gchristensen>
Ericson2314: did it work?
<Ericson2314>
gchristensen: yes, thanks!
<gchristensen>
good :)
<Ericson2314>
making PR for nix 2.0 notes for 19.03 now
<{^_^}>
[nixpkgs] @flokli opened pull request #47227 → php: set mysql socket path if mysql[i] or pdo_mysql support is enabled → https://git.io/fAFZM
<gchristensen>
we should add a builtins.trace warning for nix1 on unstable, probably, so it isn't a surprise
<Ericson2314>
gchristensen: agreed
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #47228 → release-notes/19.03: add release notes for nix 2.0 requirement bump → https://git.io/fAFZp
rfold has joined #nixos
<rfold>
It seems like nix-copy-closure always uses SSH. I want to copy the closure into a tar. I can do it in Nix with closureInfo, but is there a command-line utility for this?
<Ericson2314>
so one coming from 18.03 will see a breaking change
<Ericson2314>
and it already hit master
<Ericson2314>
so release-18.09 right now is the anomaly
xeji has quit [Quit: WeeChat 2.1]
notlar has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 pushed 21 commits to staging-18.09: https://git.io/fAFnw
<samueldr>
Ericson2314: when you say "[users] coming from 18.03 will see a breaking change", that is before or after this fix, and what kind of breaking?
<perrier-jouet>
manveru: this template seems popular because it has a clean and minimalist feeling
<perrier-jouet>
manveru: are you sure they dont use a framework ?
<Ericson2314>
samueldr: sorry that's backwards. they *won't* seee a breaking change
<Ericson2314>
it's a new service
<samueldr>
perrier-jouet: do you mean CSS?
<manveru>
perrier-jouet: verify for yourself man :)
<Ericson2314>
original author condones the change that PR made
michalrus has joined #nixos
<Ericson2314>
the only breaking change is for people with custom mutable stuff coming from before-now release-18.03
<samueldr>
Ericson2314: no worries, I was worrided about breaking things, but couldn't see anything described in the PR :)
<{^_^}>
[nixpkgs] @bhipple closed pull request #47230 → Remove dead code from stdenv check-meta license logic → https://git.io/fAFcP
<gchristensen>
I <3 nixops' dry-activate
<gchristensen>
it is so good
<juhe>
gchristensen: is it also in 18.03 release (I didn't dare to update to 18.09 yet)?
<{^_^}>
[nixpkgs] @bhipple reopened pull request #47230 → Remove dead code from stdenv check-meta license logic → https://git.io/fAFcP
<gchristensen>
it has been there for a long time I think
<juhe>
oh, ok. I never used it (yet), though will try next time when updating cluster of machines being maintained by nixops.
<gchristensen>
I have my deployment automatically done, and it automatically does a dry-activate, and then rquires a push-button confirmation to do the real activation
<{^_^}>
[nixpkgs] @oxij opened pull request #47233 → treewide: mass rebuild noop cleanups → https://git.io/fAFWw
<steveeJ>
sphalerite: any epiphanies since earlier today? :D
<srhb>
Ericson2314: I've begun to only open backport PRs after the master one is merged, seems to help avoid fracturing discussion :P
<Ericson2314>
makes sense
<{^_^}>
[nixpkgs] @Ericson2314 closed pull request #47239 → lib: Make `overrideScope` take arguments in the conventional order for 18.09 → https://git.io/fAF4B
<srhb>
During release it's especially fun if you end up having a three-way conversation :D
<samueldr>
seems to be the last super:.*self: in the codebase
<nDuff>
Are patches to nixpkgs that make systemd dependencies optional (to make more of the tree buildable with musl) likely to be accepted, or does that fall into the unwanted-maintenance-complexity category?
<Ericson2314>
samueldr: good catch!
fendor has quit [Ping timeout: 252 seconds]
<steveeJ>
welp
chrisaw has joined #nixos
<chrisaw>
Can anyone tell me what I'm doing wrong in my config here? I'm trying to remove an interface from being managed by networkd. My config line is: systemd.network.links."eno2".linkConfig = { Unmanaged = true; }; which is based on info from https://github.com/NixOS/nixos/blob/master/modules/tasks/network-interfaces.nix but I am getting: error: The option value `systemd.network.links.eno2.linkConfig' in `/etc/nixos/server/network.nix' is n
<chrisaw>
... of type `attribute set of systemd options'.
<samueldr>
the general sentiment I've seen is "it would be hard to keep maintained, so without the right people maintaining it on top of all the maintaining it can't really happen"; so such contributions can't be pump, dump and go away; they'd need care, thought and collaboration to stay relevant :/
<samueldr>
(still, no official word, just one person talking)
Ralith_ has joined #nixos
<sphalerite>
nDuff: oh yeah, the SLNOS thing ("suckless nixos") is probably worth mentioning. Idk if it's dead or just happening in secret
<nDuff>
The particular dependency I'm looking at fixing is for a tool under a Free license that gets commercial use, so I could promise to take care of it for as long as I'm remaining in my current job position, still bundling that tool in our product, &c. Couldn't make that commitment across more of the repo than the parts I can justify spending day-job time on.
notlar has quit [Quit: notlar]
<sphalerite>
nDuff: oh yeah, not-os is also worth mentioning — it's not so much for general-purpose stuff as nixos AFAIK, but it runs without systemd https://github.com/cleverca22/not-os/
<sphalerite>
and is built on top of nixpkgs
<gchristensen>
nDuff: what package?
<nDuff>
gchristensen, nwipe. The problematic dependency chain is parted -> lvm2 -> systemd
<gchristensen>
ah yeah
ixxie has quit [Remote host closed the connection]
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
reinzelmann has quit [Quit: Leaving]
Thra11 has joined #nixos
Thra11 has quit [Ping timeout: 252 seconds]
jD91mZM2 has quit [Quit: WeeChat 2.0]
Ericson2314 has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @veprbl opened pull request #47243 → pyjet: init at 1.3.0 → https://git.io/fAFEm
<{^_^}>
[nixpkgs] @aszlig opened pull request #47244 → stdenv: Improve ELF detection for isELF → https://git.io/fAFEO
<steveeJ>
aszlig: ideally I'd want the packages in the VM to be built from a different nixpkgs set than the VM tools which assemble and run the image
<nDuff>
Hmm. I was under the impression that using overrideAttributes to remove references to systemd would implicitly remove it from buildInputs, but that seems not to be the case.
<sphalerite>
nDuff: no — you'd need to pass in a new buildInputs without it.
<sphalerite>
nDuff: if you need both, you need some extra parens, like (foo.override {systemd = null;}).overrideAttrs (o: {mkDerivation arg overrides here;})
<{^_^}>
[nixpkgs] @basvandijk pushed 2 commits to release-18.09: https://git.io/fAFE6
<asymmetric>
i don't understand the difference between nix-build and nix-instantiate
<asymmetric>
don't they both save derivations in the store?
<manveru>
asymmetric: that's because nix-build is a wrapper for nix-instantiate
<manveru>
as the manpage says: "nix-build is essentially a wrapper around nix-instantiate (to translate a high-level Nix expression to a low-level store derivation) and nix-store --realise (to build the store derivation)."
<asymmetric>
ah, i see
<asymmetric>
but the man page for nix-instantiate says "The paths of the resulting store derivations are printed on standard output."
<asymmetric>
and the man page for `nix-build` says that it's nix-instantiate + nix-store --realize
<manveru>
derivations aren't derivation outputs
<manveru>
you have to realize them to actually do the build
<manveru>
a derivation is basically the blueprint
<asymmetric>
ah, got it
<nDuff>
Are there tools for pretty-printing a derivation?
<manveru>
nDuff: `nix show-derivation` is one
<manveru>
also there's `:p` in `nix shell`, but not sure how well it works for derivations
<steveeJ>
sphalerite: it's working \o/
<sphalerite>
yaaaaay!
rfold has quit [Quit: leaving]
<sphalerite>
What was the key?
juhe has left #nixos ["PART #nixos :PING 1537733745"]
<{^_^}>
[nixpkgs] @manveru opened pull request #47246 → terraform-provider-gandi: init at 0.0.1 → https://git.io/fAFuq
<nDuff>
okay, so the changes I've made successfully changed systemd from ["dev", "out"] to just ["dev"]
<nDuff>
however, it's not buildable at all with musl, so having it a build dependency isn't workable either.
<steveeJ>
sphalerite: I updated the expressions on gitlab
patrl has quit [Ping timeout: 264 seconds]
<sphalerite>
nDuff: you could always use an overlay that replaces systemd with null for the entirety of nixpkgs. Of course that means building everything from source and likely a large amount of failures, but this sounds like a cost you'll have been anticipating anyway :)
<steveeJ>
in short, it's only importing the qemu-vm module from a different path, everything else is from the pkgs from which the VM packages should come from
<steveeJ>
the 'pkgs' argument to the qemu-vm module is what ultimately gets into the image
tobiasBora has quit [Quit: WeeChat 1.6]
<steveeJ>
but it also uses the qemu from those pkgs which possibly breaks compatibility between the module and the qemu version, so it's very prone to breakage
orivej has joined #nixos
<{^_^}>
[nixpkgs] @dtzWill pushed to master « 19.03: fix section id »: https://git.io/fAFuR
<nDuff>
sphalerite, ...hmm. Putting in an overlay would at least make it easier to track where that build-time dependency is coming from. And "building everything from source" already happened last night when I attempted to do my first musl-based package build, so eh, why not do it again? :)
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #47250 → raspberrypi-bootloader: allow custom target directory and support 64-bit Raspberry Pi 3 → https://git.io/fAFgm
<alex``>
emily: it could be with ii :p
simukis has quit [Quit: simukis]
hakujin has joined #nixos
stanibanani has quit [Ping timeout: 252 seconds]
georges-duperon has quit [Ping timeout: 252 seconds]
catid has joined #nixos
<catid>
good afternoon all. i'm looking into switching to linux kernel 4.18+ as there are several improvements there for my laptop, but i don't see a package for it in "nix search" - any hints on how to upgrade?
<sphalerite>
catid: boot.kernelPackages = pkgs.linuxPackages_latest; generally speaking. However, "latest" in 18.03 (or whatever release happens to be the current stable) may well not be the actual latest kernel release
<steveeJ>
catid: your best bet is ` boot.kernelPackages = pkgs.linuxPackages_latest;` in you rconfig
<sphalerite>
alex``: nix-shell, generally speaking, enters an approximation of the environment used for building a derivation
<sphalerite>
your mkShell thing evaluates to such a derivation (one that doesn't actually build)
<sphalerite>
but since we have kakoune packaged, we can just reuse the environment for that
<alex``>
what does <nixpkgs>?
<sphalerite>
nix-shell '<nixpkgs>' -A kakoune means get the shell for the derivation that's at the "kakoune" attribute of the expression you get when importing "<nixpkgs>"
<alex``>
and attr kakoune
<sphalerite>
and '<nixpkgs>' in turn means "the entry named nixpkgs on NIX_PATH"
<sphalerite>
it's the same thing that's used for other nix commands in other ways. You might have done `nix-env -iA nixpkgs.foo` at some point, which is related but suffers from nix-env weirdness
<sphalerite>
but in each of the cases, at some point you're calling into nixpkgs, the nix package collection, which contains several thousand packages, and selecting something within it
<sphalerite>
There are lots of inconsistencies in the UI which make this less obvious, but one day the new `nix` command will cover all the functionality of the "legacy" nix-* commands and everything will be wonderful. At least that's how I imagine it ;)
<{^_^}>
[nixpkgs] @Mic92 closed pull request #47249 → stdenv: make isELF more precise → https://git.io/fAFu7
<sphalerite>
alex``: yes exactly that file plays a key role
<sphalerite>
it's called by pkgs/top-level/all-packages.nix (unfortunately github chokes on that file because it's too big), which "fills in" the arguments that the kakoune expression takes
<alex``>
Mic92: {project-name}.gemspec is for installing in ~/.gem/ user directory the dependencies, and the Gemfile takes that gemspec to bundle the dependencies?
<Mic92>
infinisil: I would be interested in the source
Ariakenom_ has quit [Ping timeout: 252 seconds]
<srhb>
palo: Why not use the already-packaged version?
<infinisil>
(it's a lot of functions, I'd prefer not to spam this chat)
<Mic92>
infinisil: A link is enough
<alex``>
Mic92: You know where the dependencies are stored using Bundler?
<infinisil>
There is no link, the implementation is in the bot only
<palo>
srhb: because I'm running `cabal install --only-dependencies`
Ariakenom_ has joined #nixos
<Mic92>
alex``: they get also installed into ~/.gem when you use the bundle command
<infinisil>
(unless somebody writes it out)
<srhb>
palo: We can do that too, that's essentially nix-shell '<nixpkgs>' -A haskellPackages.wai-app-static.env
<palo>
srhb: I ran a nix-shell -p haskellPackages.wai-app-static but same issue
abueide has joined #nixos
<srhb>
Oh really.
<alex``>
Mic92: :OO
<srhb>
palo: Oh, you mean using cabal install again? That's expected, still.
<palo>
srhb: I'll try your command, just a sec
<alex``>
I was thinking Bundler was designed to avoid the mixing
<Mic92>
infinisil: so somebody also hacked the whole thing into a chat?
<samueldr>
alex``: they do the right thing to not mix and match
<infinisil>
I did lol, was just experimenting with it
<srhb>
palo: If you configure it from within the above shell, it'll work.
<samueldr>
(outside of nixos)
<samueldr>
the gems are installed in a per-ruby-version folder, and bundler will refer to the right versions
<Mic92>
alex``: it does on a project level, they are seperated. but the storage is shared in ~/.gem
<samueldr>
what I did: chmod a-rwx ~/.gem
<samueldr>
(whilc clearing it first)
<justanotheruser>
what font pack do I need if I'm missing a ton of unicode fonts?
<srhb>
palo: I'd recommend not using cabal for installing haskell packages at all though. :)
<Mic92>
justanotheruser: dejavu_fonts is a good start
Ariakenom has quit [Ping timeout: 245 seconds]
<srhb>
(cabal-install, that is)
<palo>
srhb: the -A and -p commands cant be mix easily, ...
<srhb>
palo: What are you actually trying to do?
Ariakenom has joined #nixos
<palo>
srhb: want to code a servant-server
<Mic92>
justanotheruser: unifont, if you need more then that
<palo>
I have no experiance with stack
<srhb>
palo: Why not skip cabal-install completely and just use nix?
<palo>
I don't understand
<srhb>
palo: you don't have to ever run cabal install (or any cabal command) using Nix to manage your project.
<srhb>
palo: Or install a haskell executable.
Ariakenom_ has quit [Ping timeout: 272 seconds]
<palo>
srhb: so what is the alternative for cabal run ?
<justanotheruser>
looks like dejavu_fonts gave me what I know was missing
<justanotheruser>
I'll comment unifont in case I need it, thanks!
<srhb>
palo: You _can_ use that in a nix project just fine, fwiw
<srhb>
palo: Generally, if you have a cabal file, just use cabal2nix . --shell > shell.nix
<srhb>
Then enter the shell with nix-shell
<srhb>
palo: Inside, cabal commands should work for you. (or just plain ghc commands)
<palo>
ah I see
<palo>
I'll try that
Ariakenom_ has joined #nixos
<srhb>
palo: It also has the benefit of not needing to compile any of the dependencies when you do it like that :)
<palo>
srhb: ok the -A haskellPackages....env thing worked so far
<srhb>
palo: Yup, the shell.nix method essentially does the same thing (but with your project's dependencies rather than just one package from haskellPackages)
<palo>
srhb: yeah
Ariakenom has quit [Ping timeout: 260 seconds]
Ariakenom has joined #nixos
<palo>
the cabal2nix . --shell way works also perfectly fine
Ariakenom_ has quit [Ping timeout: 260 seconds]
<srhb>
palo: I recommend sticking with that. :) It scales really well once your project grows.
notlar has joined #nixos
<alex``>
emily: Keeping Bash as user shell, you could add in your ~/.profile `exec elvish`
<palo>
hmm but I can't find the executable anymore with the cabal2nix version
<palo>
srhb: I thought the executables are installed this way
<alex``>
I think it would work (not tested yet though)
<srhb>
palo: Not unless you explicitly do so
<srhb>
palo: Generally I'd just avoid installing anything.
<emily>
alex``: yeah but that feels uglier somehow.
<alex``>
There is no solution unless writing a parser
<palo>
srhb: I don't mean that I can run the executables i defined in the project.cabal file
Ariakenom has quit [Ping timeout: 240 seconds]
<palo>
s/don't/do/
<{^_^}>
[systemd] @flokli opened pull request #23 → nspawn: chown() the legacy hierarchy when it's used in a container → https://git.io/fAFVp
Rovanion has quit [Ping timeout: 244 seconds]
<alex``>
sphalerite: Is there a way to fil an entry-point to nix-shell?
<alex``>
I would like it to run `make debug=no`
<alex``>
when entering in the shell
<sphalerit>
justanotheruser: unifont iirc is a bitmap font which has great coverage but is also really ugly. To close that gap, the various parts of noto are nice
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<{^_^}>
[nixpkgs] @andrew-d opened pull request #47255 → boost: limit concurrent jobs to the maximum supported number → https://git.io/fAFwm