simukis_ has quit [(Remote host closed the connection)]
stepcut has quit [(Remote host closed the connection)]
<disasm>
neonfuz1: I'm late to the party here, but I've been using EFI vfat for /boot ever since I got an EFI motherboard on my desktop (actually before that, when I was using refind/refit on a mac dual booting arch/osx). Made dual boots a whole lot easier imho, and at one time I had 5 or 6 different distros booting on my desktop. I have a retina mac and used https://github.com/mbbx6spp/mbp-nixos as a starting
<gchristensen>
freeman42x: I dunno, seems like an unfortunate move.
<disasm>
reference for my configuration.nix.
<freeman42x>
gchristensen, <I dunno, seems like an unfortunate move.> why "unfortunate"? Isn't Gitter better than IRC for most use-cases?
<gchristensen>
I don't think so
<disasm>
neonfuz1: two options for wifi, networking.wireless.enable and declaratively declar your wifi config, or enable network manager. My preference is networking.wireless, but forced to use network manager because networking.wireless doesn't support WPA enterprise auth.
<gchristensen>
I don't understand this move to highly centralized proprietary platforms, why a project would so closely binds itself to a company which wholly depends on another company
<disasm>
neonfuz1: also, what fs are you using? There's some flags you probably want to change in hardware-configuration to optimize for ssd.
<disasm>
gchristensen: 100% with you there... IRC works perfectly fine, why complicate things? So you can have emoji and embedded images??? :)
<gchristensen>
what if github (or git) is no longer cool tomorrow? gitter is gone?
WhittlesJr has joined #nixos
<gchristensen>
it seems so awkward to so tightly bind your project's chat with essentially an implementation detail of your community
<WhittlesJr>
Hey, brand new to nix... not sure if this is the right place to ask for advice on creating a somewhat non-standard package?
<disasm>
yeah, they'd have to completely rebrand
<globin>
WhittlesJr: it is the right place
<disasm>
WhittlesJr: absolutely, we love a challenge :)
bennofs has quit [(Ping timeout: 248 seconds)]
<neonfuz1>
disasm: ext4
<dash>
gchristensen: good thing like a third of this channel is on Matrix
<disasm>
neonfuz1: I think on ext4 you're good with the defaults.
<WhittlesJr>
Alright! I'll try to be brief. I'm coming from Arch, where I've got an AUR package and some elisp to import a bunch of useful utility functions from the Spacemacs project.
<neonfuz1>
gchristensen: >tfw working for people who use discord to communicate
<gchristensen>
dash: sure (1) decentralized, (2) open, (3) not bound to the VCS and VCS host you happen to use, (4) integrates with IRC
<neonfuz1>
disasm: alright. Would you suggest another fs?
takle has joined #nixos
<neonfuz1>
I don't even think there are discord self hosted servers yet
<neonfuz1>
I know there are custom clien
<neonfuz1>
jesus, what is with this typing it's horribl
<WhittlesJr>
Scratching my head on how to approach this with nix. Part of me thinks that it belongs in MELPA. Then I think about how it would have to be essentially dependent on Spacemacs, which isn't a MELPA package at all. I just want to pull down the Spacemacs source, pull out the utility functions, and arrange them as a package to be require'd
<disasm>
completely up to you. I'm running btrfs (but thinking about trying zsh), I know a bunch of people here use zfs. I'm no expert, but I've heard ext4 can quickly wear out the limited write-space on an SSD because of the journal constantly writing.
<neonfuz1>
so what parts aren't managed by the config file?
<WhittlesJr>
So then I think about just rolling it pretty similarly to the PKGBUILD. But I have no earthly idea how NixOS handles the path'ing for Emacs to find its installed packages.
<WhittlesJr>
So as for what to do with Emacs' path, I'm unsure.
<neonfuz1>
ah
<disasm>
neonfuz1: almost everything I have that's not in my home directory is managed by the config, or custom nix-shells.
<disasm>
and most of my home directory is directories with default.nix files :)
<disasm>
WhittlesJr: can you like the PKGBUILD?
takle has quit [(Ping timeout: 240 seconds)]
<disasm>
link :)
<neonfuz1>
disasm: so like, for example, your user's password
<disasm>
up to you again :) you can put the password hash in the config and have immutable users, or allow users to change passwords.
felipedvorak has quit [(Quit: felipedvorak)]
<disasm>
WhittlesJr: so I'd first look at pkgs/applications/editors/emacs-modes/melpa-generated.nix or any of the other modes in there. I'm not an emacs user, so I'm no expert here.
stepcut has joined #nixos
takle has joined #nixos
stepcut has quit [(Remote host closed the connection)]
<WhittlesJr>
disasm: !! You can just stick it in "$out/share/emacs/site-lisp" and it'll know where it lives??
<WhittlesJr>
disasm: Coming from Arch's KISS mentality, we just install things somewhere on the filesystem. I wasn't sure if any of that convention carried over here... There's so much automagic going on here and I'm not sure what's what
<disasm>
WhittlesJr: so $out is going to be /nix/store/<sha256 hash>/yourpackage-version. That sha256 is a hash of the derivation, and nix can actually determine the hash without even building it, it's pretty cool stuff.
<WhittlesJr>
disasm: Well if it's all somehow included in Emacs' search path then color me impressed.
<gchristensen>
WhittlesJr: in some ways, nixos is simpler but it seems complicated from all the complexity you're used to :) for now, I'd recommend just accept there is cool magic going on
<WhittlesJr>
disasm: Haha I'm getting that sense, yes. I fell in love with this distro as soon as I heard about it.
<WhittlesJr>
er, pinged linked the wrong person. New to IRC channels as well :P
<WhittlesJr>
I typically kept to the Arch forum
eacameron has quit [(Remote host closed the connection)]
<gchristensen>
:) well, welcome around! glad to have you!
eacameron has joined #nixos
<disasm>
WhittlesJr: I'm not sure how that works from an emacs perspective. gchristensen when you install a emacs mode, how is that picked up by emacs search path?
<WhittlesJr>
disasm, gchristensen: and from my perspective: how is anything picked up from anything's search path @_@
<gchristensen>
a wrapper sets the load path
<disasm>
oh, that's awesome :)
<WhittlesJr>
advanced alien technology. Damn this distro is cool. Well, I'll go see what I can do. Not sure this is something I'd want to pull-request but if I can get it to work for me, I'll be happy c: Thank you all!
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to openssl-1.1: https://git.io/v5UiJ
<NixOS_GitHub>
nixpkgs/openssl-1.1 135c0c8 Robin Gloster: openssl: add patches for 1.1, too
NixOS_GitHub has left #nixos []
felipedvorak has joined #nixos
erictapen has quit [(Ping timeout: 248 seconds)]
<neonfuz1>
how do i configure nixos so I can use wpa_cli
<disasm>
WhittlesJr: if you want to PR it, the only extra step is adding to pkgs/top-level/all-packages.nix: yourpackage = callPackage ../applications/editors/emacs-modes/yourpackage { }; That makes it available in nixpkgs for people to install.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] veprbl opened pull request #28455: hdfview: init at 2.14 (master...hdfview) https://git.io/v5UiG
NixOS_GitHub has left #nixos []
<WhittlesJr>
disasm: I guess if it's worth sharing.... It's kind of intimidating contributing to the main repo. The AUR feels much more, eh, "low-visibility," with every package being its own repository.
<neonfuz1>
is there anywhere that describes basic directory layout of nixos?
stepcut has joined #nixos
<neonfuz1>
like, I can't find anything
<neonfuz1>
I'd expect the manual to have a section on this
Fare has joined #nixos
<WhittlesJr>
oh and gchristensen: thanks for the welcome! Seems like a cool place to be. I'll always have a place in my hard drive for Arch, if for no other reason than for its raw educational merit. But NixOS seems like just the "right way" to do system administration. Although I reeeaaally tried to convince myself to go to GuixSD instead... I love me my lisp, but the distro as a whole just... just doesn't hold a candle to this
<disasm>
neonfuz1: everything is in the nix store /nix/store. All libraries, config files, binaries, profiles
<neonfuz1>
damn everything? how do you find anything?
<disasm>
typically, you don't mess around in the store on your own. There's scripts that get the paths correct, make sure your binaries are available, etc...
grw3 has quit [(Ping timeout: 240 seconds)]
<WhittlesJr>
neonfuz1: read: "magic."
<neonfuz1>
okay, so what script would I use?
<disasm>
for example, in this one you'll see systemd.services.wpa_supplicant. That'll generate a systemd service (which is thrown in the store), and then symlinked to where systemd can find the service. That service will then have hard-coded paths to the store for the wpa_supplicant package.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/v5UPU
<NixOS_GitHub>
nixpkgs/master 5404ac9 Samuel Leathers: atom: fixes #28436 - atom git integration
NixOS_GitHub has left #nixos []
<WhittlesJr>
Thanks again everyone for the help!
<gchristensen>
WhittlesJr: guixsd is based on nix's daemon anyway :) we're friendly!
WhittlesJr has quit [(Quit: Page closed)]
hamishmack has joined #nixos
<neonfuz1>
disasm: yeah i understand that
<neonfuz1>
disasm: what command would I use to find the real path
<neonfuz1>
gchristensen: I want to enable the interface so I can use wpa_cli
<neonfuz1>
I enabled that and did a rebuild, but I can't use wpa_cli
<gchristensen>
does `ip link` list a wireless nic?
cpennington has quit [(Ping timeout: 240 seconds)]
<neonfuz1>
yea
<gchristensen>
neonfuz1: you added networking.wireless.enable = true, ran `nixos-rebuild switch`?
<hodapp>
anyone have any clue why, if I'm in a nix-shell with particular Python packages present (like matplotlib), code that should be producing plot windows is not?
<neonfuz1>
how I fixed it in gentoo is ctrl_interface=/run/wpa_wpa_supplicant
<neonfuz1>
in /etc/wpa_sulicant.conf
<neonfuz1>
I guess I could do the same
cpennington has joined #nixos
<gchristensen>
sorry, what exactly isn't working?
eacameron has quit [(Remote host closed the connection)]
<disasm>
neonfuz1: it should just work. and you can't edit that file in nixos, it's a symlink to the nix store.
<neonfuz1>
oh, wpa_supplicant failed to start
stepcut has joined #nixos
<gchristensen>
please note you haven't actually said what isn't working or what you're trying
<neonfuz1>
disasm: that's the problem, there's no nixos equivalent to ctrl_interface that I can find
<disasm>
neonfuz1: you defined your networks in wireless.networks?
<neonfuz1>
gchristensen: wpa_cli won't connect to wpa_supplica
<neonfuz1>
that's whats wrong
<neonfuz1>
god I hate konsole
<neonfuz1>
it's literally just droping random letters...
<gchristensen>
neonfuz1: and you added networking.wireless.enable = true, ran `nixos-rebuild switch`?
eacameron has joined #nixos
<neonfuz1>
yes
<neonfuz1>
does nixos have urxvt? that's my preferred term emulator
<gchristensen>
wpa_supplicant -D nl80211,ext -i wlp2s0 -c <(wpa_passphrase YourWififNetwork "YourWifiPassword") is how I've done that in the past
<Fare>
neonfuz1, I used to use rxvt, but that was a decade or two ago.
<disasm>
neonfuz1: most likely, you won't really have to change anything in your config. Just update the channel, and do a nixos-rebuild switch --upgrade
freeman42x[m] has quit [(Ping timeout: 255 seconds)]
eacameron has quit [(Remote host closed the connection)]
freeman42x[m] has joined #nixos
ShalokShalom has joined #nixos
rauno has joined #nixos
<rauno>
Hi
<rauno>
Any idea why such error might occur:
<rauno>
Can't locate utf8.pm: /root/.nix-profile/lib/perl5/site_perl/utf8.pm: Permission denied at /run/current-system/sw/bin/nix-shell line 3.
<rauno>
BEGIN failed--compilation aborted at /run/current-system/sw/bin/nix-shell line 3.
<rauno>
It happens when trying to run nix-shell from regular user
eacameron has joined #nixos
<hyper_ch>
why is it looking at /root when running from regular user?
<rauno>
good question
<rauno>
this happend after i upgrade one library in my local nixpkgs
eacameron has quit [(Ping timeout: 248 seconds)]
eacamero_ has joined #nixos
<rauno>
ok this happened cause i was doing su <username> while being root, weird that i still thought im root..
<rauno>
seems like a bug
Ivanych has quit [(Ping timeout: 240 seconds)]
<rauno>
and now when trying to rebuild .nix with pypi2nix, warning: Nix search path entry ‘/etc/nixos/nixpkgs:nixos-config=/etc/nixos/configuration.nix’ does not exist, ignoring < might this also be related to permissions?
hamishmack has joined #nixos
<neonfuz1>
now I'm getting "warning: name collision in input Nix expressions, skipping ‘/home/sage/.nix-defexpr/channels_root/nixos’"
eacamero_ has quit [(Read error: Connection reset by peer)]
zeus__ has quit [(Remote host closed the connection)]
eacameron has joined #nixos
eacameron has quit [(Read error: Connection reset by peer)]
eacameron has joined #nixos
freusque has joined #nixos
krav_ has quit [(Ping timeout: 248 seconds)]
eacameron has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] mrobbetts opened pull request #28457: openldap: Fix build with LibreSSL. (master...openldap-fix-libressl-2) https://git.io/v5UdZ
<neonfuz1>
any idea why audio drivers wouldn't have been detected?
jbgi has quit [(Ping timeout: 248 seconds)]
eacameron has joined #nixos
<sphalerite>
neonfuz1: are you sure it doesn't recognise the audio device? Does alsamixer not show anything?
pxc has joined #nixos
<sphalerite>
chances are, the easiest solution is to enable pulseaudio (hardware.pulseaudio.enable = true)
eacameron has quit [(Ping timeout: 240 seconds)]
pxc has quit [(Ping timeout: 240 seconds)]
sellout- has quit [(Ping timeout: 248 seconds)]
ertes has quit [(Ping timeout: 240 seconds)]
canndrew has joined #nixos
<benkolera>
Oh my. Now that I am progressively becoming less of a nix noob, I am loving how well nix is meaning that I will never have to write another dockerfile ever again... The buildImage stuff is super rad for weaseling nix into a dockerized environment. Thanks to those responsible. :D
<bkchr>
srhb: can you restart the machine/server/whatever. Or does that need to be written on the mailinglist to inform someone?
<srhb>
bkchr: I can't, some of the people that can frequent the channel, though I think they're all sleeping :-)
<bkchr>
okay :D
<srhb>
bkchr: You can add an issue for better visibility :)
arjen-jonathan has joined #nixos
<bkchr>
srhb: Yeah, I will add an issue in github. I'm often just unsure where to report those problems or I don't know if the people already aware of the problem.
bennofs has joined #nixos
gnuhurd has joined #nixos
<symphorien>
how can I export a store closure as a tar archive ?
<symphorien>
or how can I expand a nar archive without nix ?
<srhb>
bkchr: If they already are, it's cheap to just close the issue. :)
<srhb>
bkchr: No harm done.
<bkchr>
srhb: yeah, you are right!
mal_ has joined #nixos
Khetzal has quit [(Ping timeout: 246 seconds)]
snikkers has joined #nixos
<mal_>
hi, with following default.nix i get "error: undefined variable ‘buildEnv’ at /tmp/aaa/default.nix:29:9" when running nix-shell [https://paste.ccc.ac/?2fc0f8aed0223c46#ggSaI3TMvHXzlPM7A+MxzUeKaXAM4N29n5BF9mpRvh8=] any ideas why? :)
<srhb>
mal_: It'll be somewhere in pkgs, presumably
<srhb>
mal_: Oh, you have scoped that.
* srhb
looks
arianvp2 has quit [(Ping timeout: 240 seconds)]
arjen-jonathan has quit [(Quit: WeeChat 1.9)]
<mal_>
i mean the sha in the fetchfromgithub is wrong, could it be it? i'm a total beginner
Khetzal has joined #nixos
<srhb>
mal_: Well, it's not even evaluating that in order to conclude that buildEnv doesn't exist, so that's not the only issue at least.
<clever>
mal_: the pkgs you defined is a path to nixpkgs, not an instance of nixpkgs
<clever>
you need to import the result of fetchFromGitHub, and pass it {}
<clever>
the code at the top is basicaly just pkgs = <nixpkgs>;
<mal_>
clever: oh ok, i'll try to fix it, thanks!
<clever>
mal_: also, that buildEnv is useless, the buildInputs are already in the env
takle has joined #nixos
arianvp2 has joined #nixos
<mal_>
clever: sorry i shortened it a bit, there are buildinputs that are not in the env
<srhb>
mal_: On a stylistic note, I think it would be nicer to get rid of the double let .. in ..
<srhb>
mal_: You only need one there :)
<srhb>
mal_: ie let pkgs ...; libtins ...; in pkgs.stdenv.mkDerivation ...
arianvp2 has quit [(Client Quit)]
Tucky has quit [(Remote host closed the connection)]
arianvp2 has joined #nixos
<hyper_ch>
hmmm, is that bad:
<hyper_ch>
depmod: WARNING: could not open /nix/store/nr931gikcbbz7p89wi2mblkysx7jki9s-kernel-modules-shrunk/lib/modules/4.9.44/modules.order: No such file or directory
<hyper_ch>
depmod: WARNING: could not open /nix/store/nr931gikcbbz7p89wi2mblkysx7jki9s-kernel-modules-shrunk/lib/modules/4.9.44/modules.builtin: No such file or directory
<samae>
clever: oh, that's much simpler than I thought
<hyper_ch>
? everything else seems to have bult just fine
<samae>
thanks
<clever>
samae: as for super vs pkgs, pkgs refers to the final set, after all overlays are applied
<mal_>
srhb: oh yeah thanks, atm i don't really know what i'm doing :)
<clever>
samae: so trying to override foo = pkgs.foo.override would result in infinite recursion
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
<srhb>
mal_: That's OK, cargo culting can bring you incredibly far :-P
<clever>
samae: super is the pkgs set from the previous overlay (or the original un-altered one)
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
Tucky has joined #nixos
<clever>
so the overlays can chain together and apply changes in order
erictapen has joined #nixos
Khetzal has joined #nixos
<samae>
Ok, got it
ThatDocsLady has joined #nixos
<srhb>
was it possible to use buildEnv without defining a src at some point?
<clever>
buildEnv doesnt need a src
<srhb>
I mean, the surrounding mkDerivation
<srhb>
I see that littered on blogs around the webs.
ebzzry_ has joined #nixos
<clever>
srhb: and src = ./.; wont do?
<srhb>
clever: It probably will, I'm just wondering why it was previously possible to omit it.
<srhb>
And what the semantic difference is.
<clever>
the unpackPhase is what complains about it being missing
Arcaelyx has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<clever>
and nix-shell doesnt automatically run the unpackPhase
<srhb>
Ah.
<srhb>
So the difference is shell vs build.
<clever>
yeah
<clever>
nix-shell doesnt start a build
<clever>
but you can manually run genericBuild to start it up
<clever>
that will run all of the phases
<clever>
just dont do it as root
<srhb>
I guess I don't understand why people use this buildEnv method over the regular mkDerivations(...).env with nix-shell
<clever>
there is also another layer of caching and timeouts
eacameron has quit [(Ping timeout: 246 seconds)]
mpcsh has quit [(Quit: THE NUMERICONS! THEY'RE ATTACKING!)]
Fannar has quit [(Quit: leaving)]
<clever>
viric: ls -ltrh ~/.cache/nix/tarballs/
<viric>
clever: deleting the store path, it *unpacks* anew. It does not download, but unpacks it.
<clever>
viric: first layer is hash(url) + ".info" and hash(url) + "-file"
mpcsh has joined #nixos
<clever>
the .info contains timestamp and etag info, to cache downloading the tar.gz
<clever>
the -file symlink points to the result of the download
<clever>
then the second layer, is hash(tar.gz) + "-unpacked", which points to the unpacked version of the tar
<viric>
I have troubles in cygwin64. If TMPDIR is /tmp, it works ok. But if TMPDIR is /cygdrive/c/whatever, all files get +x, and all direct file references get a different hash on store.
<clever>
if you only deleted the unpacked directory, it will just re-unpack it
<clever>
but if you delete the .tar.gz itself, that -file points to, it has to re-download
<viric>
good ;thank you about the .cache
<clever>
any change to the url will also trigger a re-download
<clever>
because its using a hash of the url as a key
<clever>
but if it gives the same .tar.gz, it can reuse the unpacked directory, because it knows tar is pure
<viric>
So, basically, the value of TMPDIR determines whether nixpkgs will be good or not, and that result will be cached. :)
<clever>
and these symlinks arent gc roots
<clever>
so the garbage collector will just eat things without warning
<clever>
ahhh
<clever>
so the tar unpacking ISNT pure! lol
<viric>
Exactly.
<clever>
your fs is ruining things
<viric>
not pure on cygwin.
<clever>
the .tar.gz may also be +x'd, which impacts the path to the unpacked stuff
<clever>
but deleting the -unpacked variant would force it to re-unpack, which might be enough for this bug?
<viric>
pf.
<viric>
We are running nix inside teamcity agents, which like to redefine TMPDIR
<clever>
channel unpacking its its own derivation, that can run fully in a sandbox
<clever>
and is hardened against such things, though it relies on where libstore puts /tmp/nix-build-foo
<viric>
Well, both cases suffer the same problem
<viric>
isn't it?
<viric>
possible weird fs on TMPDIR.
<clever>
yeah
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
takle has quit [(Ping timeout: 240 seconds)]
erictapen has joined #nixos
<clever>
viric: how well does nix work under cygwin64, and is it possible to have a nix based toolchain that can create executables that dont depend on cygwin at runtime?
takle has joined #nixos
jellowj has joined #nixos
wizzup[m] has joined #nixos
<viric>
clever: we use it for the visual studio compiler
<viric>
clever: we have our kind of "mkDerivation"
<viric>
clever: and we don't use much of nixpkgs
<clever>
ah, thats similiar to what i'm thinking
<clever>
i was thinking of running ghc to build haskell packages
<clever>
is the install as simple as just doing ./configure && make && make install on a nix checkout, and then initializing the store as usual?
<clever>
viric: also, what does builtins.currentSystem eval to under cygwin64?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm pushed 1 new commit to master: https://git.io/v5Ttb
<NixOS_GitHub>
[nixpkgs] joachifm pushed 5 new commits to master: https://git.io/v5Tth
<NixOS_GitHub>
nixpkgs/master aa49e36 Michael Alan Dorman: elpa-packages: 2017-08-21
<NixOS_GitHub>
nixpkgs/master ec4f63c Michael Alan Dorman: org-packages: 2017-08-21
<NixOS_GitHub>
nixpkgs/master 2263a7a Michael Alan Dorman: melpa-stable-packages: 2017-08-21...
NixOS_GitHub has left #nixos []
<viric>
x86_64-cygwin
<clever>
ah, good
oahong has quit [(Ping timeout: 240 seconds)]
<clever>
so there wont be any confusion within hydra
<viric>
cygwin32 cannot work; don't even thin kabout it
<viric>
aka "cygwin" :)
<clever>
looks like i will need to start from setup-x86_64.exe
<viric>
right
oahong has joined #nixos
<clever>
lets see, where did i save that RDP info....
mal_ has left #nixos ["WeeChat 1.8"]
<adisbladis>
Does anyone have a good solution to this problem: My colleague would like to install nix on his machine but is running osx with fish as his shell
<adisbladis>
How do fish users on osx set up their environments?
<srhb>
Is anyone else experiencing really slow systemctl completion with zsh? completing a unit name after systemctl restart ... takes seconds for me
<makefu>
srhb: having the same issue, but i think it was worse a few months ago
<makefu>
now it 'only' takes a few seconds
<srhb>
makefu: You may be right. I still find it intolerable. Is it better for bash?
<makefu>
yes
<makefu>
(justed tested it)
<srhb>
Grr. :-)
<makefu>
srhb: i'd be most grateful if you could find the root cause and submit a PR to nixpkgs
<srhb>
makefu: I'll give it a stab, but I've no experience with completion modules.
<makefu>
me neither :D
<makefu>
but in nixOS i am doing stuff i've never touched before on a day-to-day basis
<srhb>
True, that happens :)
griff_ has joined #nixos
ShalokShalom_ has joined #nixos
<srhb>
Looks like they are bundled with systemd
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin closed pull request #27515: modules sks-keyserver and pgpkeyserver-lite web frontend (master...upstream-sks-keyserver) https://git.io/v7vCq
NixOS_GitHub has left #nixos []
<srhb>
So I guess it's not NixOS-specific
ShalokShalom has quit [(Ping timeout: 255 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/v5TOv
<NixOS_GitHub>
nixpkgs/master b2ae23a Vladimír Čunát: Merge branch 'staging' (older version)...
NixOS_GitHub has left #nixos []
leat has joined #nixos
arianvp2 has quit [(Quit: arianvp2)]
eacameron has joined #nixos
joehh has joined #nixos
leat has quit [(Quit: leat)]
arianvp2 has joined #nixos
leat has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
joehh has quit [(Ping timeout: 240 seconds)]
takle has quit [(Ping timeout: 240 seconds)]
FRidh has quit [(Ping timeout: 246 seconds)]
FRidh has joined #nixos
griff_ has quit [(Quit: griff_)]
eacameron has joined #nixos
griff_ has joined #nixos
pxc has joined #nixos
eacameron has quit [(Ping timeout: 246 seconds)]
dredozubov has quit [(Ping timeout: 248 seconds)]
evangeline has quit [(Ping timeout: 248 seconds)]
darlan has joined #nixos
griff_ has quit [(Client Quit)]
musicmatze_ has quit [(Ping timeout: 248 seconds)]
musicmatze_ has joined #nixos
takle has joined #nixos
schoppenhauer has quit [(Ping timeout: 248 seconds)]
pxc has quit [(Ping timeout: 240 seconds)]
dredozubov has joined #nixos
evangeline has joined #nixos
arianvp2 has quit [(Quit: arianvp2)]
leat has quit [(Quit: leat)]
pie_ has quit [(Ping timeout: 240 seconds)]
leat has joined #nixos
Ross has quit [(Ping timeout: 248 seconds)]
baconicsynergy[m has quit [(Ping timeout: 246 seconds)]
kitten[m] has quit [(Ping timeout: 246 seconds)]
hedning[m] has quit [(Ping timeout: 246 seconds)]
reactormonk[m] has quit [(Ping timeout: 246 seconds)]
Guest7639 has quit [(Ping timeout: 246 seconds)]
TimePath has quit [(Ping timeout: 246 seconds)]
systmkor has quit [(Ping timeout: 240 seconds)]
sphalerite has quit [(Ping timeout: 246 seconds)]
aniketd[m] has quit [(Ping timeout: 246 seconds)]
JosW has quit [(Ping timeout: 248 seconds)]
catern has quit [(Ping timeout: 248 seconds)]
kyren has quit [(Ping timeout: 248 seconds)]
iMatejC has quit [(Ping timeout: 248 seconds)]
andrewrk has quit [(Ping timeout: 248 seconds)]
joko has quit [(Ping timeout: 248 seconds)]
primeos has quit [(Ping timeout: 248 seconds)]
qewzee has quit [(Ping timeout: 248 seconds)]
Profpatsch has quit [(Ping timeout: 248 seconds)]
Wysteriary[m] has quit [(Ping timeout: 246 seconds)]
ArdaXi[m] has quit [(Ping timeout: 246 seconds)]
AmineChikhaoui[m has quit [(Ping timeout: 246 seconds)]
a123123123[m] has quit [(Ping timeout: 246 seconds)]
dtz has quit [(Ping timeout: 246 seconds)]
zimbatm has quit [(Ping timeout: 246 seconds)]
betaboon has quit [(Ping timeout: 248 seconds)]
brynedwards has quit [(Ping timeout: 248 seconds)]
mbrgm has quit [(Ping timeout: 248 seconds)]
aminechikhaoui has quit [(Ping timeout: 248 seconds)]
pSub has quit [(Ping timeout: 248 seconds)]
slyfox has quit [(Ping timeout: 248 seconds)]
Ross has joined #nixos
samtoday has quit [(Ping timeout: 255 seconds)]
Mic92 has quit [(Ping timeout: 246 seconds)]
nikivi has quit [(Ping timeout: 260 seconds)]
danyc0 has quit [(Ping timeout: 248 seconds)]
Acou_Bass has quit [(Ping timeout: 248 seconds)]
ekleog has quit [(Ping timeout: 248 seconds)]
hodapp has quit [(Ping timeout: 248 seconds)]
systmkor has joined #nixos
kier has quit [(Ping timeout: 248 seconds)]
ekleog_ has joined #nixos
JosW has joined #nixos
siel has quit [(Ping timeout: 240 seconds)]
samtoday has joined #nixos
joko has joined #nixos
Twey has quit [(Ping timeout: 260 seconds)]
slyfox has joined #nixos
pSub has joined #nixos
aminechikhaoui has joined #nixos
pSub has quit [(Changing host)]
pSub has joined #nixos
brynedwards has joined #nixos
danyc0 has joined #nixos
Profpatsch has joined #nixos
Acou_Bass has joined #nixos
iMatejC has joined #nixos
nikivi has joined #nixos
primeos has joined #nixos
Twey has joined #nixos
catern has joined #nixos
hodapp has joined #nixos
kier has joined #nixos
joehh has joined #nixos
mbrgm has joined #nixos
andrewrk has joined #nixos
kyren has joined #nixos
samtoday has quit [(Ping timeout: 255 seconds)]
pie_ has joined #nixos
qewzee has joined #nixos
schoppenhauer has joined #nixos
oahong has quit [(Ping timeout: 248 seconds)]
Mic92 has joined #nixos
zimbatm has joined #nixos
<viaken>
1) Does Nix have an equivalent of /etc/skel? 2) Is there a reason that "ssh nixserver ls" tells me command not found? My default shell is zsh if it matters. 'ssh nixserver "echo $PATH"' looks right.
kitten[m] has joined #nixos
aniketd[m] has joined #nixos
<clever>
viaken: did you set isNormalUser on the user when you made it in configuration.nix?
<viaken>
Not the first time, but I did on my second rebuild.
<viaken>
Is that one of the staetful things?
<viaken>
stateful*
<clever>
maybe, does the home directory exist?
samtoday has joined #nixos
a123123123[m] has joined #nixos
hedning[m] has joined #nixos
ArdaXi[m] has joined #nixos
Guest7639 has joined #nixos
TimePath has joined #nixos
sphalerite has joined #nixos
Wysteriary[m] has joined #nixos
<viaken>
It does now, but I think I created it by hand.
AmineChikhaoui[m has joined #nixos
dtz has joined #nixos
baconicsynergy[m has joined #nixos
reactormonk[m] has joined #nixos
<clever>
try commenting out the user in configuration.nix and do a switch, so nixos deletes the user
<clever>
then uncomment it, and switch again, so nixos creates it properly
<viric>
Does nixos have any virtualbox guest option so its xorg screen is mapped to the virtualbox window containing it?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm pushed 4 new commits to release-17.03: https://git.io/v5TBi
<NixOS_GitHub>
nixpkgs/release-17.03 ffabae3 Maximilian Bosch: geogebra: 5-0-369-0 -> 5-0-377-0...
<NixOS_GitHub>
nixpkgs/release-17.03 54644b3 Maximilian Bosch: geogebra: 5-0-377-0 -> 5-0-382-0...
<NixOS_GitHub>
nixpkgs/release-17.03 171094a Maximilian Bosch: geogebra: fallback to web.archive.org...
NixOS_GitHub has left #nixos []
takle has quit [(Remote host closed the connection)]
eacameron has joined #nixos
<viric>
ah damn it, kms
stepcut has joined #nixos
<Rajigo>
Hi, libsigsegv-2.11 does not compile here (opened https://github.com/NixOS/nixpkgs/issues/28464). Yet that exact lib is also provided by the vendor (Scientific Linux) and there is no mention of failing tests with that kernel version. I don't really know what to do from here but I'll do as suggested if anyone has an idea.
eacameron has quit [(Ping timeout: 248 seconds)]
arianvp2 has quit [(Quit: arianvp2)]
takle has joined #nixos
eacameron has joined #nixos
<symphorien>
you can do `nix-shell '<nixpkgs>' -A libsigsegv --pure` then `genericBuild`, it will leave you in a shell where you can debug why tests fail
arianvp2 has joined #nixos
freeman42x[m] has joined #nixos
ShalokShalom_ is now known as ShalokShalom
eacameron has quit [(Ping timeout: 248 seconds)]
eacameron has joined #nixos
freeman42y[m] has quit [(Ping timeout: 260 seconds)]
eacameron has quit [(Ping timeout: 255 seconds)]
<viric>
Mpf. virtualbox is too old in nixpkgs 17.03
<viric>
far too old.
FRidh has quit [(Quit: Konversation terminated!)]
<viric>
ls
jellowj has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 5 new commits to master: https://git.io/v5TEE
<NixOS_GitHub>
nixpkgs/master 4c07f40 Peter Simons: hackage2nix: keep old base-orphans around for liquidhaskell
<romildo>
I want to use sed in a nix expression, but the sed argument needs to see the literal text ${efl_version} without any variable substitution.
<romildo>
postPatch = ''
<romildo>
sed -i "/ethumb_client >= ${efl_version}/a emile >= ${efl_version}" configure.ac
<romildo>
'';
<romildo>
How should it be escaped? It seems that escaping $ with '' is not enough because than the shell seems to expand the variable ${efl_version} before running sed.
teozkr_ is now known as teozkr
nico202 has joined #nixos
<Enzime>
why was the Nix Wiki killed? was it just getting too out of date too quickly?
<danbst1>
romildo: ''{ should be enough to escape
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NickHu opened pull request #28467: scaleway-cli: init at v1.14 (master...scaleway-cli) https://git.io/v5Tgy
NixOS_GitHub has left #nixos []
freeman42y[m] has quit [(Ping timeout: 255 seconds)]
<nico202>
Hi, anybody using emacs+pdf-tools can help? It used to work 'till some update ago, now it can't find (and build) the server (epdfinfo), even if the binary gets compiled upon install (and is in /nix/store/
<danbst1>
if you want to escape $ inside bash, do ''$\{
<gchristensen>
Enzime: a persistent wave of spam-bots continuously ruined the content, it was made read-only because we couldn't fend them off, then the content became too out of date.
stepcut has quit [(Remote host closed the connection)]
stepcut has joined #nixos
<romildo>
danbst1, sed -i "/ethumb_client >= $''{efl_version}/a emile >= $''{efl_version}" configure.ac gives a syntax error: unexpected '}', expecting '.' or '='
<gchristensen>
just replace the "s with 's
<__monty__>
danbst1: I'm not using nixos. Can I just create ~/.nixpkgs/configuration.nix with contents { allowUnfree = true }?
<gchristensen>
sed -i '/ethumb_client >= ''${efl_version}/a emile >= ''${efl_version}' configure.ac
<clever>
__monty__: the name must be config.nix, but otherwise, yes
<gchristensen>
__monty__: ~/.config/nixpkgs/config.nix is the preferred location these days
<romildo>
but sed does not accept it. How is the right way for putting this literal \ char in the sed expresson?
<makefu>
goibhniu: exactly, nixos.wiki is the future
freeman42y[m] has joined #nixos
<gchristensen>
romildo: \/ tells sed to ignore the / and treat it like a literal "/"
freeman42y[m] has quit [(Client Quit)]
<gchristensen>
romildo: you'll have to do \\/ which says a literal \, then the / separator
<clever>
gchristensen: sed can also use a character other then /
<gchristensen>
clever: yes but that doesn't address the issue, because \# will be treated as a literal #
<clever>
i usualy just throw \'s at the problem until it goes away, lol
<makefu>
gchristensen: i've added a clarification to the front-page
<gchristensen>
makefu: cool, thanks!
freeman42y[m] has joined #nixos
freeman42x[m] has quit [(Ping timeout: 246 seconds)]
<gchristensen>
as of this morning, the nix pills series has been moved to https://nixos.org/nixos/nix-pills/ -- thank you ixxie, disasm, Ankhers, sphalerite for helping port them, and obviously Lethalman for writing so much great content. if you've ever wished something was edited or updated in the nix pill series, open an issue or send a PR: github.com/nixos/nix-pills
<disasm>
thanks gchristensen for getting that live and organizing the effort, we couldn't have done it without your leadership!
<gchristensen>
aw shucks :) glad to help
Anon-kun has joined #nixos
<Ankhers>
It was a good excuse to dip my toes into docbook.
<Anon-kun>
Hi guys
<goibhniu>
Awesome job! Thanks a million to everyone involved!
<Anon-kun>
I've been distro hopping for a while now and want to try NixOS next
<Ankhers>
That makes a little more sense. I didn't think I saw it there when I wrote it.
<disasm>
ah :)
<gchristensen>
a bit awkward that the semantics of the CSS on the homepage apply directly to the (different) semantics of the docs.
<sphalerite>
gchristensen: wouldn't <application> be more appropriate than <command> in that case though?
<gchristensen>
could be, but the bug still exists. <command> in the nix pills means something different from <span class="command"> in the rest of the website
<gchristensen>
(I can't do that now, but perhaps someone else could)
goibhniu1 is now known as goibhniu
Anon-kun has joined #nixos
Anon-kun has quit [(Client Quit)]
simukis_ has joined #nixos
gnuhurd has quit [(Ping timeout: 240 seconds)]
WhittlesJr has joined #nixos
<WhittlesJr>
Hey, sorry to bother... what's the preferred method of doing pull requests? Clone upstream and branch, fork and commit into master, what?
MercurialAlchemi has quit [(Ping timeout: 255 seconds)]
sauyon has joined #nixos
gnuhurd has joined #nixos
<gchristensen>
send a PR to master, whatever branch name you use on your end doesn't matter
<gchristensen>
but a branch per PR is a good idea
<viric>
Hm "impureEnvVars" when building with daemon... come from the daemon process, or the nix-store -r process?
<WhittlesJr>
k, thanks
<clever>
viric: i believe they come from the nix-store -r
<clever>
viric: not 100% sure though
<viric>
clever: I kind of experience the opposite
<disasm>
WhittlesJr: I usually create a separate branch per PR to make it easier to make changes going forward. If you PR master, then it's just difficult to tease apart which commits are for what PR.
<disasm>
WhittlesJr: did you have any problems getting your emacs package to build?
<clever>
viric, gchristensen: looks like it just does getEnv as the process initiating the build (the daemon)
<WhittlesJr>
disasm: Haven't gotten that far yet! I went to bed pretty much right after our conversation yesterday. I've got quite a few packages to submit to get my emacs setup fully ported, so there should be a couple PRs heading your way today if all goes well.
<disasm>
nice!
takle_ has joined #nixos
<gchristensen>
WhittlesJr: how many packages? are they from melpa?
<WhittlesJr>
disasm: The things that are missing right now are mostly some fonts and then that Spacemacs utility function thing I mentioned.
<WhittlesJr>
disasm: all-the-icons-ttf, file-icons-ttf, material-design-icons-ttf, weathericons-ttf namely. I figure each should be a separate package?
stepcut has quit [(Remote host closed the connection)]
<viric>
I'm trying to get a fixed-output-derivation to use a user launched ssh-agent
<viric>
It seems that it requires same UID, beyond the unix socket permissions
stepcut has joined #nixos
<viric>
grmbl.
<clever>
viric: i already have a gist that solves this
stepcut has quit [(Remote host closed the connection)]
takle has quit [(Ping timeout: 246 seconds)]
<viric>
clever: oh, show it
stepcut has joined #nixos
sellout- has joined #nixos
michiel_l has quit [(Remote host closed the connection)]
<clever>
viric: the socat running as root is the "client" for ssh-agent, and root is an exception to the uid lockout (so you can use things under sudo)
<viric>
Of course.
<clever>
it then acts as a proxy
<clever>
also, this doesnt use env vars at all
<disasm>
WhittlesJr: material is there, I'd follow that as an example for fonts: pkgs/data/fonts/material-icons
<clever>
it uses the NIX_PATH
<viric>
clever: great!!
<viric>
clever: doesn't it require /tmp in the sandbox paths?
<disasm>
WhittlesJr: also all-the-icons is there: pkgs/data/fonts/emacs-all-the-icons-fonts/default.nix
<clever>
viric: that leaks in i think, let me check
<WhittlesJr>
disasm: Ohh, I was just searching https://nixos.org/nixos/packages.html to see if they were there... Didn't realize there was a separate fonts/ directory
Rotaerk has quit [(Quit: Leaving)]
<WhittlesJr>
disasm: please bear with my greenness
<disasm>
WhittlesJr: no problem :) I tend to just `git grep` for packages myself most of the time.
<romildo>
I need some help debugging pkg-config not listing some requirements when building EFL applications.
pakettiale has quit [(Read error: Connection reset by peer)]
<disasm>
WhittlesJr: they might not have a top-level and might only be included by other packages as dependencies
darlan has joined #nixos
<WhittlesJr>
disasm: Gotcha. Well then with the all-the-icons font package, I should be set on the font front.
pakettiale has joined #nixos
<disasm>
romildo: if you do a nix-build -K it'll keep the build in the temp directory, and you can manually run the commands to configure/build to debug.
<clever>
viric: my first guess is that because it came from NIX_PATH, maybe nix considers it a legal input to the build, and tries to copy it in
<romildo>
pkg-config in NixOS has somehow been patched for disabling private requires? EFL applications use them.
darlan has quit [(Client Quit)]
<clever>
viric: it might even copy the unix socket intact into the store? not sure
sellout- has quit [(Quit: Leaving.)]
<romildo>
disasm, I have asked on #e and raster said the eet.pc files content is right: http://dpaste.com/3V2FDT8
<romildo>
disasm, nonetheless pkg-config does not list the private requirements:
<yorick>
they're asking for a docker image that would build a nix package that people can then install
<clever>
viric: i'm not sure how that worked either, now that i think of it
<simpson>
FROM nixos/nix
<viric>
clever: :D
<clever>
viric: i just remember that it worked with a zero byte sshc-onfig in the NIX_PATH
leat has joined #nixos
<yorick>
I think maybe the best they can do is put a default.nix somewhere and have people do a fetchurl in their config
<simpson>
yorick: No, the best that they could do is to relax and let downstreams exist.
<clever>
viric: __impureHostDeps is another thing, its in build.cc again
stepcut has quit [(Remote host closed the connection)]
<yorick>
simpson: to be fair, we're still packaging 0.4.8 or something terrible that nobody should use and they had to come yell at us to add a warning to it
<yorick>
simpson: so we have proven to be terrible already
stepcut has joined #nixos
nevermin_ has joined #nixos
<simpson>
yorick: The slap-fight between upstream and downstream always ends in downstream's favor unless upstream is more productive than downstream or closed-source. The reasoning is hopefully obvious.
<disasm>
romildo: hmmm, I would think that would work...
<simpson>
Sure, we should have packaged the less-buggy version, but CLV.
ryantrinkle has joined #nixos
erasmas has joined #nixos
<yorick>
simpson: they didn't want us to package anything after that
<viric>
clever: so, I used a path for the ssh-config-file, and the recipe cannot access it (not in sandbox)
<simpson>
yorick: "That's nice" is my maintainer-hat opinion.
nevermin_ is now known as nevermind
<yorick>
simpson: fine, but people want to use mailpile. there is a service. maybe ask people to supply the package themselves
Wizek_ has joined #nixos
stepcut has quit [(Ping timeout: 240 seconds)]
<yorick>
domenkozar: what did you have in mind for the mailpile people?
Wizek_ has quit [(Client Quit)]
WhittlesJr has quit [(Quit: Page closed)]
<romildo>
disasm, according to raster from #e, explicitly listing these requirements should not be needed. pkg-config HAS to find them, because they are already specified on eet.pc. And they are successfully found on other systems, but not on NixOS.
<viric>
clever: /tmp/hax doesn't get into the sandbox eithe
<viric>
clever: either
<viric>
clever: are you sure you had useSandbox enabled there? :)
takle has quit [(Ping timeout: 248 seconds)]
<romildo>
disasm, and he asks whether pkg-config has been patched somehow in a way that would affect this behaviour.
<disasm>
clever: you have any suggestions for romildo? I'm not familiar enough with pkgconfig to understand what's going on here.
<clever>
viric: it might have been off back then
<clever>
disasm: is the problem happening under nix-build?
<romildo>
clever, disasm, yes, under nix-build.
<disasm>
clever: that's my understanding, right romildo?
takle has joined #nixos
<yorick>
simpson: I suppose they could make a channel that people add?
Sonarpulse has quit [(Ping timeout: 246 seconds)]
<clever>
romildo: i remember something about pkg-config silently dropping things, if the dependencies listed in the .pc file cant be found
<clever>
romildo: and you have to add the deps the .pc wants to the buildInputs
<romildo>
disasm, clever, all requirements that pkg-config is failing to list belongs the the single efl package, which is already in buildInputs.
<__monty__>
clever, simpson: Is there a way I could find that out without asking here every time? I appreciate the help but I don't want to rely on support just to select the right packages ; )
<__monty__>
Is the only way to read the source?
stepcut has joined #nixos
<disasm>
__monty__: if you don't care about the logo, go with the source build. otherwise, go with the binary. To my knowledge the trademarked logo is the only difference between the two.
<simpson>
__monty__: TBH I guessed that thunderbird-bin was a binary installer, but checked nixpkgs before saying for sure. The -bin pattern is pretty common.
leat has quit [(Ping timeout: 248 seconds)]
<simpson>
__monty__: If you want to be independent about it, it's not hard to do $(git grep thunderbird-bin) inside nixpkgs.
<__monty__>
Oh, so thunderbird-bin is unfree. Does nox indicate free/unfree packages?
<yorick>
it would be nice if we could have an auto-updater for fetchgit/fetchFromGitHub packages that checks pgp keys
<simpson>
gchristensen: Yes, but it's always good to revisit once in a while. Daira's imagination is an interesting place.
<__monty__>
simpson: It's not hard to read the nix expression for a package per se but it's quite a bit of redundant information so I figured there might be a better way.
erictapen has joined #nixos
freusque has quit [(Quit: WeeChat 1.7.1)]
<romildo>
clever, disasm, it seems that pkg-config has been patched to deal with the bug https://bugs.freedesktop.org/show_bug.cgi?id=4738 from FEDORA, which dates from 2005 and has not been close. It seems to change the pkg-config behaviour regarding including private requirements. With the patch private requirements are ignored. Considering how old the bug is, and that has not been closed yet, maybe the patch is not desirable. Is it important for NixOS?
eacameron has quit [(Ping timeout: 276 seconds)]
eacameron has joined #nixos
<clever>
romildo: i havent seen that bug before, but there is the optional on !vanilla in there
<clever>
romildo: so if you add (pkgconfig.override { vanilla = true; }) to your buildInputs, you will get an unpacked version
<disasm>
clever: that's what I was wondering, if vanilla = true would solve it :)
<romildo>
clever, disasm, with vanilla = true some other requirements are not found: mount
<disasm>
romildo: but they were found with vanilla false?
miefda has joined #nixos
eacameron has quit [(Ping timeout: 276 seconds)]
eacameron has joined #nixos
mrkgnao has joined #nixos
<romildo>
disasm, with vanilla false mount is not mentioned. With vanilla true it gives the message:
<romildo>
Package 'mount', required by 'eeze', not found
<disasm>
romildo: you may need utillinux in buildInputs
eacameron has quit [(Ping timeout: 248 seconds)]
stepcut has quit [(Remote host closed the connection)]
stepcut has joined #nixos
<romildo>
clever, disasm: raster observers that back in 2013 (from https://bugs.freedesktop.org/show_bug.cgi?id=4738) it was noted that bug cannot be fixed in the current fields without breaking the expected pkg-config behavior. And it criticizes NixOS for following that route. And he asks: nixos now expects everyone to patch their builds, pkgconfig files and so on for nixos's patches that broke compatibility for pkg-config? In summary it considers it was bad
<romildo>
disasm, clever: I would like to have clean nix expressions for EFL applications, without the need to explicitly adding the private requirements from .pc files. Do you think I should open an issue for pkg-config regarding that? Or should forget that and keep the status quo?
<dash>
Error I'm getting is: "error: infinite recursion encountered, at /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/lib/modules.nix:151:28"
<disasm>
__monty__: it's an alias: pkgs/top-level/aliases.nix
stepcut has joined #nixos
<nwspk>
Is there a way to download more than one file as src for a package? Like calling fetchurl twice.
eacameron has joined #nixos
ryantrinkle has quit [(Ping timeout: 246 seconds)]
<__monty__>
disasm: How about nixpkgs.youtube-dl? It's not an alias so why does it exist in addition to the ones under pythonpackages?
<disasm>
romildo: if it's something needed for all efl it could be propogated in pkgs/desktops/enlightenment/efl.nix
<LnL>
nwspk: srcs = [ ... ];
<nwspk>
LnL: Oh thanks. Didn't know this one.
<LnL>
if you use that you'll have to set sourceRoot yourself in postUnpack
bennofs has quit [(Ping timeout: 276 seconds)]
eacamero_ has joined #nixos
eacameron has quit [(Ping timeout: 255 seconds)]
<romildo>
disasm, I will try the other packages that uses efl and see what is needed. If they are needed for all of them, I will propagate them and switch to vanilla=true for pkg-config. I will do that later on. Is that reasonable, right?
bennofs has joined #nixos
eacamero_ has quit [(Ping timeout: 240 seconds)]
nevermin_ has joined #nixos
erictapen has quit [(Ping timeout: 248 seconds)]
himmAllRight has quit [(Remote host closed the connection)]
<__monty__>
I get this message for inkscape, file-roller, thunderbird: Gtk-Message: Failed to load module "canberra-gtk-module" I've tried installing all three available expressions for libcanberra but neither of them helped. Do the expressions for the programs themselves have to be changed?
sellout- has joined #nixos
himmAllRight has joined #nixos
iyzsong has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
erictapen has joined #nixos
<page>
LS
takle_ has joined #nixos
<neonfuz1>
sphalerite: alsamixer has the devices, and I can mute and unmute, but no volume bars
<neonfuz1>
and no audio comes through
ryantrinkle has joined #nixos
arianvp2 has quit [(Quit: arianvp2)]
<simpson>
gchristensen: Yeah, every few years, somebody proposes a cap-safe or cap-aware processor. Sometimes they even get taped out. I'm not sure why they always fail.
takle has quit [(Ping timeout: 246 seconds)]
<dash>
simpson: power law of marketing and capital :(
<simpson>
dash: Well, sure, for everything after the Pentium. But this Intel caper and also the Burroughs were both before that, right?
<dash>
anyhow. where do I start looking when I've got an infinite recursion in a module definition
erictapen has quit [(Ping timeout: 248 seconds)]
<dash>
well, the intel one failed because it was too ambitious for what was feasible at the time
<dash>
I don't think the Burroughs machine can be called a failure.
eacameron has joined #nixos
miefda has quit [(Quit: Leaving)]
stepcut has quit [(Remote host closed the connection)]
<dash>
nixos-rebuild doesn't seem to take a --show-trace flag :(
Infinisil has joined #nixos
<disasm>
dash: not going to fix the problem, but first thing I'd do is change enable option to mkEnableOption
<clever>
dash: i think it does, but only for the nixos side, not for the nix side
<clever>
dash: try nixos-rebuild --fast instead
stepcut has joined #nixos
<clever>
dash: behind the scenes, nixos-rebuild will first try to build nix, then use that nix to build nixos
<clever>
dash: and it may not pass all of the args to the first nix build
<clever>
dash: but nixos-rebuild --fast both skips that, and does --show-trace
stepcut has quit [(Remote host closed the connection)]
vagrant- has joined #nixos
stepcut has joined #nixos
<disasm>
dash: I think you want lib.mkIf, not pkgs.mkIf
<vagrant->
hi all, as anyone else noticed an error message "This rsync lacks old-style --compress due to its external zlib" when using the nix-build rsync?
<vagrant->
apparently there's some compile-time option which causes this?
<vagrant->
not sure if this is a known thing or if I should look into tweaking the rsync nix definition to fix it
<disasm>
dash: I think ndnConfig needs to be in a let {
nixo202 has joined #nixos
nico202 has quit [(Read error: Connection reset by peer)]
nixo202 has quit [(Client Quit)]
<disasm>
dash: I'd probably move ndnConfig below let cfg =
nico202 has joined #nixos
<vagrant->
looks like we're using a `--without-included-zlib` option which might be causing it...?
DrWaste has quit [(Quit: Run away!)]
snikkers has quit [(Ping timeout: 240 seconds)]
peacememories has joined #nixos
<Infinisil>
vagrant-: Well it's easy to check yourself :)
stepcut has quit [(Remote host closed the connection)]
<vagrant->
Infinisil: yes, I'm building it. The reason I asked is if there's a reason we're building it this way then it might not be worth a PR.
stepcut has joined #nixos
supercynic has quit [(Ping timeout: 246 seconds)]
stepcut has quit [(Ping timeout: 248 seconds)]
<dash>
disasm: thanks for having a look.
<dash>
yeah I made ndnConfig an attribute since I planned to interpolate some stuff. clearly I have not yet.
<vagrant->
is there a reason why equality doesn't seem to work on derivations?
michiel_l has joined #nixos
<vagrant->
I have two derivations both with the same hash, and yet `==` returns false
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
peacememories has joined #nixos
MinceR has quit [(Ping timeout: 240 seconds)]
<fpletz>
disasm: see my comment :)
<disasm>
thanks fpletz :)
<rauno>
does anyone know if i enable postgresql service, how can i access it? It doesn't work with sudo -u postgres psql
MinceR has joined #nixos
freeman42x[m] has joined #nixos
<domenkozar>
yorick: I was wondering how hard it is
<domenkozar>
and provide instructions how to package
freeman42z[m] has joined #nixos
snikkers has joined #nixos
bennofs has joined #nixos
freeman42y[m] has quit [(Ping timeout: 240 seconds)]
<yorick>
domenkozar: they could make a channel with a corresponding cache. they could also put a .nix file somewhere. they could also take down .tars if they don't want people using them and then we don't cache them anywhere
eacameron has quit [(Remote host closed the connection)]
<yorick>
that'd satisfy their updating needs maybe
<disasm>
rauno: sudo psql postgres
freeman42x[m] has quit [(Ping timeout: 255 seconds)]
Myrl-saki has quit [(Ping timeout: 255 seconds)]
miefda has joined #nixos
<disasm>
thinking of tackling this: https://github.com/NixOS/nixpkgs/issues/28014 The best way to handle that would be to build from source right, or should I download the war and put it in the store?
elastic_dog has quit [(Ping timeout: 240 seconds)]
[0x4A6F] has joined #nixos
jellowj has quit [(Ping timeout: 240 seconds)]
elastic_dog has joined #nixos
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gchristensen>
manveru: sorry my internet died right after I sent that :P
<disasm>
gchristensen: do you have an opinion on the above issue, if I should build from source, or get the standalone package (existing subsonic package is standalone).
<gchristensen>
disasm: I have no idea, sorry
<disasm>
it looks like jenkins uses a pre-built war file, I think I'll go the pre-built war direction.
<yochai[m]>
Not sure how to see when staging was last merged.
Arcaelyx has joined #nixos
<gchristensen>
manveru: I'd like to use bundix where my gemfile depends on 2 separate gems, both local directories, how ... how would I construct a gemfile to have bundix do this? (I know how to convince bundlerEnv to handle this later on)
takle has quit [(Remote host closed the connection)]
<clever>
plopix: the wrapper will generate a bash script, that prefixes PATH, then runs the old emacsWithPackages
<srhb>
There's a typo in line 37, builtInputs
<clever>
fixed
bkchr has quit [(Quit: Konversation terminated!)]
phreedom has joined #nixos
bennofs has quit [(Ping timeout: 240 seconds)]
plopix has quit [(Ping timeout: 240 seconds)]
reinzelmann has quit [(Quit: Leaving)]
Filystyn has joined #nixos
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
nwuensche has joined #nixos
nwuensche has quit [(Client Quit)]
<Infinisil>
I guess not :P
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] disassembler opened pull request #28476: airsonic: init at 10.0.0 (master...airsonic) https://git.io/v5kY8
NixOS_GitHub has left #nixos []
yegods has quit [(Remote host closed the connection)]
<srhb>
Something fishy is going on after switching to zfs, I always seem to be out of RAM, it doesn't show as just cached in free and no application appears to be using it; is this normal?
Guest89902 has quit [(Read error: Connection reset by peer)]
<srhb>
Don't I risk running out?
<clever>
srhb: and /sys/module/zfs/parameters/zfs_arc_max is the max number of bytes
<gchristensen>
disasm: sure, can you add any more comments or data to it?
<srhb>
I mean, I only have 16GiB
<gchristensen>
srhb: it'll release ram as needed
<srhb>
"only"
<clever>
srhb: zfs will automatically shrink the arc based on how much ram is used
<srhb>
Okay :)
<clever>
srhb: and it defaults to a max of 50% of your ram
zeus_ has joined #nixos
leat has joined #nixos
<srhb>
Thanks. It's a bit shame "legacy" metrics don't work though..
<disasm>
yeah, might as well rename it too. I'll clean it up first, one sec.
<disasm>
oh, I can't rename it, lol :)
<gchristensen>
I can, add a comment with your suggestion
<srhb>
But I must say the tooling for zfs seems much cleaner than btrfs so far, and more trustworthy, even if it subjectively feels somewhat slower.
takle has quit [(Remote host closed the connection)]
<gchristensen>
srhb: ZFS has corrupted at least 1,000 servers fewer than btrfs :)
<gchristensen>
(for me)
<srhb>
gchristensen: That's a lot ;-)
<globin>
disasm: are you coming to nixcon?
<gchristensen>
disasm: say yes!
<Infinisil>
srhb: How is the tooling for zfs cleaner? I never checked out btrfs
<disasm>
globin: I would love to, still seeing if I can get funding from my company :)
<gchristensen>
disasm: if you fly from BOS, its only like $550 flight
Mateon3 has joined #nixos
bennofs has joined #nixos
<disasm>
wow!
<gchristensen>
that is what I said!
<srhb>
Infinisil: It's just easier to get at the information in a lot of cases, the btrfs subcommands are convoluted, I think.
<Infinisil>
I see
<disasm>
gchristensen: what airline you taking?
sauyon has joined #nixos
<srhb>
Infinisil: I do miss being able to arbitrarily mess with the layout of my disks, raidz seems to be very "set once, don't touch again"
jellowj_ has joined #nixos
<gchristensen>
I haven't booked but probably Lufthansa
<srhb>
Infinisil: I ended up just mirroring instead.
Mateon1 has quit [(Ping timeout: 240 seconds)]
Mateon3 is now known as Mateon1
<srhb>
Infinisil: I hope btrfs does get more stable, it seems both faster and more flexible (but scary) :-P
<clever>
srhb: i think you can also dynamicaly change a raidz into a mirror of anything + old_raidz
<gchristensen>
I've not really ever flown internationally. any recommendations or counter-recommendations on airlines?
<srhb>
clever: Yes, that appears to be possible :)
<clever>
srhb: so if you have enough drives, you could mutate your 4 drive raidz into a mirror between a 10 drive raidz and the 4drive raidz
<clever>
and half the space will go to "waste"
<srhb>
clever: I definitely need that now ;-)
<clever>
then drop the smaller half
jellowj has quit [(Ping timeout: 248 seconds)]
jedai has joined #nixos
<srhb>
I just like hot-unplugging my disk and watching things fix themself. >_>
<clever>
lol
<Infinisil>
I very much would like that..
<srhb>
I know, hardware people tend to keep me at arms length.
<globin>
gchristensen: only ever flown muc<->bln with lufthansa but that was perfectly fine and they have a good reputation here
<Infinisil>
Does btrfs allow that?
<srhb>
Infinisil: zfs does too.
<clever>
zfs also has hot spares, but the fully automatic replacement logic needs the bays to map to stable device nodes
<srhb>
Infinisil: (I know, I tested it!)
<gchristensen>
globin: cool. do they pack people in like sardines (more or less than other airlines)
<Infinisil>
It does?? I remember zfs still thinking it was connected and I could do nothing to get it to work again
<srhb>
Well, at least for a drive suddenly gone awol and returning
<globin>
gchristensen: generally AFAIK less than other airlines as they seem to be trying to keep a "high quality" image
<gchristensen>
awesome :D
<Infinisil>
srhb: I am using ZFS with an external disk connected via USB, but I heard that's not recommended, maybe that's why I got that problem
<Infinisil>
I just make sure to always `zpool export` before unplugging now
<srhb>
Infinisil: That seems sane.
aloiscochard has quit [(Quit: Connection closed for inactivity)]
mrkgnao has quit [(Quit: WeeChat 1.7.1)]
<clever>
srhb, Infinisil: there is also the offline/online commands
<clever>
offline lets you temporarily suspend all activity to a drive in the array
<clever>
i think in the case of raidz members, it will operate in a degraded state and track what blocks are out of sync
zarel has quit [(Ping timeout: 240 seconds)]
<clever>
and when you online them, it will restore the missing blocks
<clever>
but i dont think you can offline enough drives that data becomes unreachable
<disasm>
gchristensen: so is there some way I can contribute in some way if I can make it? Doesn't need to be a full talk slot, Could be something as simple as a Q&A on new contributions or maybe talk about documentation efforts.
nevermind has joined #nixos
nevermind is now known as Guest32249
<Infinisil>
clever: Nice to know
<vagrant->
is there a nix-store command that can tell me the references closure of a store path? not just immediate references?
<globin>
disasm: <3 for more doc efforts
<vagrant->
sorry, referrers not references
<vagrant->
(--requisites is the answer to the first question)
<clever>
vagrant-: nix-store -qR and --tree both go down from a package to all of its deps
<vagrant->
clever: yeah sorry, I meant the other direction
<clever>
vagrant-: nix-store -q --roots goes up, to the GC roots that holds something in the store
<vagrant->
clever: so that will tell me why something is considered "alive"?
<globin>
we don't really have anyone who really has the time and motivation to do much work on that, I'd be really, really happy if someone would start on coordinating the work on docs!
<clever>
vagrant-: yeah
<globin>
disasm: ^
<vagrant->
clever: how about non-alive paths though?
<clever>
vagrant-: i dont think there is any way to see the full closure of what depends on X when those things arent alive
<vagrant->
as in, what if I just want to answer the question "what paths currently in the store depend on this path directly or indirectly"
<vagrant->
oh I see
<vagrant->
well, I could write a script to do it without too much effort I suppose
<clever>
yeah
<gchristensen>
disasm: I'd be happy to help with organizing on docs if you're interested in taking a role on that :)
<disasm>
sure thing :)
<gchristensen>
disasm for doc czar!
<Infinisil>
gchristensen: disasm: globin: What do you have in mind for how to organize the docs?
plopix has joined #nixos
<gchristensen>
by organizing I mean organizing people :P
takle has joined #nixos
<Infinisil>
Ah
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin closed pull request #28475: More repology.org problem fixes (master...more-problem-fixes) https://git.io/v5kqI
NixOS_GitHub has left #nixos []
<disasm>
I think we have pretty good technical docs with the nix/nixos/nixpkgs pages on the site. I think what we need is a lot more tutorial example type docs like we did with the nix pill porting. Expanding on the pills would be a goal, another one might be some user level tutorials. We tend to get a lot of people in IRC for the first time and don't understand how to install stuff, why when they install a package
<disasm>
using nix-env, there's no systemd service, etc...
<Infinisil>
disasm: You mean we need like.. a wiki?
<gchristensen>
Infinisil: please don't
<disasm>
I hate wiki's...
<gchristensen>
I agree, things like use-case oriented docs
<gchristensen>
let's not discuss the wiki right now
<Infinisil>
Nah, I'm totally fine with the new user-wiki, it's working great
<globin>
gchristensen, disasm: I think we should have a general "teams discussion" at nixcon, docs team, infrastructure team, whatever...
plopix_ has joined #nixos
<plopix>
clever: thx for your help. when i use nix-build emacs.nix i have "error: undefined variable ‘wrapProgram’ at /home/alex/src/emacs.nix/emacs.nix:37:63
<plopix>
"
<disasm>
I'd love to help with infrastructure too :) that's kinda my day job, lol :)
<gchristensen>
disasm: that one is _much_ harder because it takes a long time of being an active member to develop the trust to help admin it
<disasm>
yeah, figured that :)
<disasm>
do we mostly use nixops for deploying stuff?
<gchristensen>
that said (as demonstrated) contributors can make PRs to it and get them merged (I have no such access, but managed to get HSTS deployed)
yegods has quit [(Remote host closed the connection)]
<clever>
plopix: try pkgs.wrapProgram ?
<lejonet>
Where are the terminfo "packages" defined? I can't find rxvt_unicode.terminfo in nixpkgs repo and I'm suspecting it is doing something stupid and wanna try and correct that :P
yegods_ has joined #nixos
<plopix>
clever: I tried but it does the same
<clever>
plopix: ah, its called pkgs.makeWrapper
<clever>
always getting those 2 mixed up
<Infinisil>
gchristensen: How long have you been active in the nix community?
[0x4A6F] has quit [(Ping timeout: 248 seconds)]
<lejonet>
Aha! As suspected, rxvt_unicode.terminfo turns the symlinked terminfo/r in /run/current-system/sw/share into a normal dir and symlinks its stuff into that, which is why it is dropped every reboot and I need to constantly put rxvt_unicode.terminfo back -.-
<gchristensen>
Infinisil: since 2016-01-15
hc_ has joined #nixos
leat has quit [(Quit: leat)]
<plopix>
clever: cool ! it works ! thx
<Infinisil>
gchristensen: Nice! Also, ISO8601 :)
hc has quit [(Ping timeout: 240 seconds)]
<gchristensen>
:) I fetched it from my IRC logs
leat has joined #nixos
<yochai[m]>
QT fixed on master?
hc_ has quit [(Ping timeout: 248 seconds)]
hc has joined #nixos
snikkers has quit [(Remote host closed the connection)]
freeman42z[m] has quit [(Ping timeout: 240 seconds)]
yegods has joined #nixos
phinxy has joined #nixos
yegods__ has joined #nixos
hc has quit [(Ping timeout: 240 seconds)]
yegods_ has quit [(Ping timeout: 248 seconds)]
hc has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/v5knk
<NixOS_GitHub>
nixpkgs/master fcd8950 Nick Hu: scaleway-cli: init at v1.14
<disasm>
gchristensen: that's right, you're on the security team, I can see how you'd want HSTS deployed :)
peacememories has joined #nixos
yegods has quit [(Ping timeout: 276 seconds)]
yegods has joined #nixos
yegods_ has joined #nixos
mgdelacroix has quit [(Remote host closed the connection)]
yegods_ has quit [(Remote host closed the connection)]
yegods has quit [(Read error: Connection reset by peer)]
ertes has joined #nixos
yegods has joined #nixos
<gchristensen>
disasm: iirc it was Infinisil who pointed out the issue and nerd-sniped me in to fixing it :P
freeman42x[m] has joined #nixos
yegods__ has quit [(Ping timeout: 248 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] romildo opened pull request #28477: enlightenment fixes and updates (master...fix.efl) https://git.io/v5kny
NixOS_GitHub has left #nixos []
tokudan[m] has joined #nixos
yegods has quit [(Ping timeout: 276 seconds)]
eacameron has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 opened pull request #28478: haskell infra: Make sure packages get custom stdenv (master...haskell-custom-stdenv) https://git.io/v5kcs
NixOS_GitHub has left #nixos []
<lejonet>
Anyone with knowledge with how the terminfo "packages" in nixos are made, exist, w/e? :P
<clever>
lejonet: any package you add to environment.systemPackages gets merged together, and the merged result lands at /run/current-system/sw
<clever>
lejonet: so if you create a package containign share/terminfo/foo and add that to systemPackages, you will get a /run/current-system/sw/share/terminfo/foo
plopix_ has quit [(Ping timeout: 248 seconds)]
plopix has quit [(Ping timeout: 240 seconds)]
<tokudan[m]>
Could someone check the Firefox PR 28450 to see if there is anything left that would prevent merging?
eacameron has quit [(Ping timeout: 255 seconds)]
<Infinisil>
gchristensen: Heh yes that was me :P
<Infinisil>
I have my last exam in two days, I'll hope to contribute a bit more then :D
<lejonet>
clever: the thing is, rxvt_unicode.terminfo doesn't seem to be a package in the real sense... I can't find it in nixpkgs, and it seems to be doing stupid things :P
<gchristensen>
tokudan[m]: hot dog you did a lot of work
<lejonet>
clever: it converts the symlinked directory in the /run/current-system/sw/share/terminfo directory into a real directory, and symlinks its stuff into it, which means every, frikin, reboot, the results of installing rxvt_unicode.terminfo is wiped (due to /run being a tmpfs)...
<lejonet>
clever: that would install the terminal, which is kindof pointless on a headless server :/
<clever>
lejonet: the only thing in /run is the symlink current-system
<clever>
lejonet: everything else is in /nix/store/
<lejonet>
clever: unless your package makes something not a symlink in /run/current-system anymore... :P
<clever>
lejonet: what program are you running that does the conversion?
<lejonet>
I'm trying to install rxvt_unicode.terminfo, so that my shell isn't borked when I ssh in from my desktop
<clever>
lejonet: how are you trying to install it?
<clever>
what have you tried so far?
<lejonet>
clever: by having it in environment.systemPackages, any other way seems to fail (nix-env doesn't believe its a valid derivation, rxvt_unicode.terminfo that is, do I need to prefix it or something?)
<clever>
brb
miefda has quit [(Quit: Leaving)]
<globin>
gchristensen: go ahead :)
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gchristensen>
globin: can you do it? I can't press green buttons that aren't for work right now
<Ankhers>
I'm a little surprised I didn't hear about that.
<gchristensen>
Ankhers: do not share a profile between nightly and regular firefox
<Ankhers>
gchristensen: Okay. Can I ask why?
<clever>
Ankhers: and the strange part, is that my telephone goes over fiberoptic, the analog terminates in the basement, how am i hearing static on the line?
<gchristensen>
Ankhers: it screws it up somehow and leads to unhappy users. I don't think it lets you by default, you'd have to go work for it
<clever>
do they turn it back into analog somewhere upstream?, and the failure was beyond that?
<Ankhers>
clever: Sounds like it.
<Ankhers>
gchristensen: Thanks.
<clever>
Ankhers: and internet access was fully working, and ran over the same fiber
<clever>
Ankhers: so if they had just kept the phone as packets longer, it could have routed around the hole on its own
<gchristensen>
Ankhers: it'll ask you to upgrade once a day, but it is very painless -- takes about 2s to restart nightly
<Ankhers>
Easy enough.
yegods has quit [(Ping timeout: 248 seconds)]
<clever>
except, it has to be patchelf'd to run on nixos
<gchristensen>
macos :)
<clever>
and every time it updates itself, it unpatches itself
<clever>
ahh, then the kernel actively stops us from ever having such features :P
<gchristensen>
oh?
<clever>
the mach-o files have a field containing /usr/bin/dyld, which acts like ld.so
<gchristensen>
ahh
<clever>
but the kernel rejects any mach-o with a different dyld path
<gchristensen>
right right
<aristid>
gchristensen: i am worried about tree-style tabs breaking completely with ff 57
<gchristensen>
aristid: you are worried
<aristid>
it's already half-broken in 55
<gchristensen>
it is done for, unfortunately
<aristid>
other than that, it's great that firefox is becoming faster
<gchristensen>
it is an unfortunate reality that they had to kill the tightly integrated addons to become better
<aristid>
they also killed their own vertical tabs experiment recently?
<aristid>
rather confusing.
<gchristensen>
also yes, I loved that one
<aristid>
the whole point why i switched from chromium to firefox is that horizontal tabs are useless
<aristid>
tilpner: that sounds like it's similar to the "solutions" that exist for chrome
Fare has joined #nixos
acarrico has quit [(Ping timeout: 240 seconds)]
jellowj_ has quit [(Ping timeout: 240 seconds)]
<Infinisil>
Oh no
<Infinisil>
I just started using tree-style tabs and love them :(
acarrico has joined #nixos
<dash>
they're working on making the new addons model support them
<tilpner>
aristid - Why is it not a true solution?
<gchristensen>
tilpner: one thing is the tabs still show up at the top
<aristid>
tilpner: the reviews mention all the ways it is inferior to tree-style tabs
<aristid>
the developer even says: "It's a ported version from chromium"
<dash>
Can overlays provide NixOS modules?
<lejonet>
Hmm, how is directory merging done for stuff in the /run/current-system tree? I think the terminfo conflict I'm seeing might be due how directory merging is failed to be done (all terminfo but the rxvt_unicode.terminfo one comes from ncurses, and they are done properly)
<clever>
dash: you can just add custom modules to the imports list
<clever>
lejonet: is it giving a warning at nixos-rebuild?
<maurer>
tbh I swapped off firefox when I found out their new addons model was going to break pentadactyl
<maurer>
Made it unusable
<gchristensen>
maurer: where did you go?
<dash>
clever: Yeah. I want to provide both some new packages and a service module that runs one of them though
<maurer>
gchristensen: chromium + vimium
<lejonet>
clever: nah, not at all, that is why I think the rxvt_unicode terminfos aren't being put in correctly, because that would mean that it symlinks them into the ncurses directory, in the current-system profile atleast
<maurer>
gchristensen: It's inferior to pentadactyl
<gchristensen>
aye
<dash>
clever: So I currently did both an overlay and a separate module file
<tilpner>
gchristensen - Oh, that's not good :/
<gchristensen>
maurer: the same level of support exists for ff
<maurer>
gchristensen: but I can't use a version of FF old enough to support pentadactyl for security reasons
<clever>
dash: you could create a 2nd module, that sets nixpkgs.overlays and imports the 1st module
<clever>
dash: then it does both things at once
<dash>
clever: ha ok
<maurer>
gchristensen: Sure, but the old addon ecosystem was literally the one thing firefox did better than chrome
<lejonet>
clever: so my guess is that there is no directory merging being done, and the ncurses terminfo just takes precedence or something, knocking the rxvt-unicode ones off, unless forced back in
<dash>
clever: that's the sort of thing i figured could be done but it wasn't obvious to me
<maurer>
gchristensen: And they killed it in an effort to be more like chrome, so now I don't know why I'd use it :P
<aristid>
maurer: and it was also what prevented firefox from adopting a lot of improvements
<clever>
lejonet: it uses buildEnv, so it should create a directory of symlinks when it needs to merge things
<gchristensen>
yeah but now they're going to be better than chrome :)
<maurer>
That seems.. unlikely
<dash>
lejonet: yes, unfortunately not compatible with the modern internet :(
<maurer>
but good luck to them
arianvp2 has joined #nixos
<maurer>
I hope they succeed, not the least of which because I really like Rust, and if Firefox dies, the money behind Rust will die too
<dash>
firefox's addon APIs were a mistake
<Infinisil>
Web extensions?
<Infinisil>
Or the old ones?
<aristid>
Infinisil: no, xul
<Infinisil>
ah
<gchristensen>
rust is what will save them
<dash>
waiting this long to correct them is causing more pain than it had to
<lejonet>
clever: Yeah, that is my understanding of how it is supposed to be done, but it doesn't seem like anything got merged at all
<maurer>
Their previous API was probably not correct, but killing the APIs before they had a replacement was a mistake
<maurer>
Anyone who was using firefox for a set of specialized extensions left
<gchristensen>
you can't just wait for everything to be perfect before switching
<gchristensen>
I don't believe it maurer
<dash>
maurer: If they'd done it around firefox 3 it would have been much better
<maurer>
I mean, I left, I know another person who left
<gchristensen>
left to go where? to another browser which doesn't have them?
<maurer>
Yeah
<clever>
lejonet: can you make a gist containing the config you are using?
<lejonet>
clever: sure
<aristid>
maurer: while i might disagree about what they prioritised, if you insist on replacing ALL add-on apis before you move, you never will
acarrico has quit [(Ping timeout: 248 seconds)]
<gchristensen>
sure
ertes has quit [(Ping timeout: 276 seconds)]
<maurer>
If you've got FF and Chrome, and FF has some super good extensions, and chrome doesn't, you might keep using FF
<maurer>
if FF then drops those extensions, you'll decide on other points
<gchristensen>
ah fair, yes
<maurer>
and chrome/chromium imo has a superior security and memory usage records
<gchristensen>
hopefully, firefox will win the users back :) that is their bet
<clever>
and mozilla has ears everywhere
<maurer>
I was giving those up for pentadactyl
<aristid>
firefox has much better memory usage than chromium in my experience...
<clever>
they just sent me an email, "how are you?"
<clever>
lol
<Infinisil>
creepy
<gchristensen>
the users using firefox before (that they lost) weren't enough to survive, so they are making a gamble that they can make a better browser to make up for it
<Infinisil>
they are probably in this channel
<dash>
the fundamental problem is, web browsers were developed for many years before anyone knew what they were
<dash>
(a secure distributed application runtime)
<maurer>
gchristensen: I'm not really sure why you think rust will save them though
Arcaelyx has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gchristensen>
maurer: well it already is
<maurer>
gchristensen: It might make them more productive, or more secure
acarrico has joined #nixos
<dash>
the problem is, you can't make a good secure distributed application runtime by accident
<maurer>
but it's not going to give them a user-facing advantage over chrome/chromium
<gchristensen>
their css engine is way way way fast and can be highly parallelized, and safely
<maurer>
The browsers do have dick measuring contests over page load speed, but IME all modern browsers render "fast enough"
<Infinisil>
Idris needs more <3, thanks srhb :)
<srhb>
Whenever I claim to fix something there's prooobably a hitch :-P
<gchristensen>
oh! lol
<maurer>
and boosts at this point isn't going to really improve user experience
<srhb>
Infinisil: {please test if you can! :-)
<Ankhers>
Infinisil: Want to start writing a browser in Idris?
<Infinisil>
maurer: My firefox is currently super slow..
<gchristensen>
maurer: paralelization is more than just "fast" performance, but ability to use many slower cores
<Infinisil>
I hope the new rust stuff improves this..
magnetophon has quit [(Ping timeout: 248 seconds)]
<gchristensen>
or use all the cores briefly and then turn them off (mobile power consumption)
<maurer>
gchristensen: Sure, there's an argument for mobile stuff
simendsjo has quit [(Quit: ERC (IRC client for Emacs 25.1.1))]
<Infinisil>
srhb: I might :)
<maurer>
gchristensen: but for desktop web, I just don't experience a meaningful speed differential
<Infinisil>
Ankhers: Great idea! And we mathematically prove that it doesn't do anything malicious :P
giusef has joined #nixos
<joepie91>
maurer: fwiw, I'm not so concerned about Rust funding
<Infinisil>
That is with Idris' type system of course
<joepie91>
it would suck if it goes away, but probably not be fatal
<cocreature>
Infinisil: that’s easy, just prove that it doesn’t do anything at all :)
<gchristensen>
maurer: they're able to dramatically and effectively target power consumption and memory usage, something that benefits mobile and laptop users alike
<joepie91>
the Rust ecosystem has an unusual amount of collaboration going on
<cocreature>
that’s also easy to implement
kiloreux has joined #nixos
<maurer>
joepie91: There are some pieces that wouldn't have happened if funding wasn't still coming
<lejonet>
clever: so basically it seems like the /share/terminfo from ncurses-6.0 and rxvt-unicode-with-perl-unicode3-9.22-terminfo (which seems to be the package that rxvt_unicode.terminfo pulls in) get merged, at all
<maurer>
joepie91: For example, the whole tokio infrastructure probably would not have shown up nearly as quickly
<Infinisil>
cocreature: Sound? Sounds good to me :)
<Ankhers>
cocreature: That reminded me of an XKCD about Haskell being lazy.
<joepie91>
maurer: sure, but that doesn't mean it wouldn't. there's a big difference between "this will kill the project" and "this will get in the way"
<clever>
lejonet: checking the gist ...
<joepie91>
for many newer languages and platforms, it's the former
<joepie91>
for Rust, it's almost certainly the latter
<gchristensen>
maurer is just being contrarian :P
<joepie91>
(which is a good thing)(
<pstn>
Should a declarative nixos container with private Interface allow outgoing UDP traffic?
<maurer>
gchristensen: ?
<giusef>
Hi folks, anyone of you knows about NixOS with MIPS support ?
<clever>
pstn: if it has privateNetworking = true; then it has no internet access, until you configure it seperately
<lejonet>
clever: if you want I could give you a gist of the result of ls -lah of /run/current-system/sw/share/terminfo directly after reboot and after running nixos-rebuild switch to pull in the rxvt_unicode.terminfo package again
<clever>
lejonet: this sounds like a different problem entirely
<joepie91>
maurer: as for the main reasons I see Rust being beneficial to Firefox' future: 1) easier to do concurrency (which roughly translates into "much better concurrency), 2) zero-overhead abstractions (greatly desirable for a browser), 3) resource safety (no resource leaks), and 4) easier to spread out the development to be way more collaborative -> more contributions -> moves faster
<clever>
lejonet: do you have a /boot partition?
<mog>
giusef, i don't think nixos exists for mips but no reason it wouldnt work with a lot of work though...
<lejonet>
clever: I do indeed
<clever>
lejonet: is it mounted on bootup?
<joepie91>
right now Chrome has a pretty terrible reputation for resource usage and Firefox can definitely beat it on that, given the right tools
<lejonet>
clever: Yep
<tilpner>
joepie91 - Do we count memory as a resource? Memory leaks are Safe™
<joepie91>
tilpner: yeah; I'm counting that under 'resources' because so does Rust :)
<clever>
lejonet: can you gist the output of "fdisk -l /dev/sda ; blkid /dev/sd*" ?
<gchristensen>
tilpner: if you leak enough your buffer overruns won't clobber anything!
<Infinisil>
There is a story about a guy who made a server restart like 50 times a second to prevent excessive memory leaks
<joepie91>
tilpner: a memory allocation isn't that different from eg. a file handle, if you think about it
<lejonet>
clever: you mean gdisk -l of /dev/sde (this is a future filserver, so it has a few disks... :P )
<joepie91>
not any more different than a file handle is from a network socket anyway :P
<clever>
lejonet: preferably, the -l of every disk
<joepie91>
so it makes sense to just consider them all 'resources' and have comparable ownership and disposal mechanisms
<joepie91>
which... Rust pretty much does, afaikm
elastic_dog has joined #nixos
<clever>
lejonet: and the contents of /etc/nixos/hardware-configuration.nix
<joepie91>
afaik*
<lejonet>
funny thing is, I can't find the /boot entry in fstab, tho I do know it was mounted, at /boot, when I did generate the inital conf
hamishmack has quit [(Quit: hamishmack)]
<aristid>
gchristensen: parallelization is more than just that, but i also want to see a browser take advantage of a 18-core cpu just for the coolness factor :D
<lejonet>
(with at /boot, I ofc meant at /boot of the root of the new install, /mnt/boot)
<clever>
lejonet: does "mount | grep boot" show it as mounted right now?
<aristid>
gchristensen: those tend to be weak-sauce </intel fanboi>
jbgi has quit [(Ping timeout: 240 seconds)]
<lejonet>
clever: does indeed not
<lejonet>
so the problem could actually be that it continues to boot into old config?
<Infinisil>
(we should really have offtopic discussions in like #nixos-offtopic or so)
<clever>
lejonet: exactly
<joepie91>
maurer: fwiw, I'm not treating Rust as a magical fairy dust here that magically makes everything work better, but there are some pretty real benefits to be gained from it in the case of Firefox specifically; for the user-facing stuff, point 4 is probably the most critical
<clever>
lejonet: you need to manualy mount the boot fs to /boot, run nixos-generate-config, and then nixos-rebuild switch
jbgi has joined #nixos
<lejonet>
clever: I did find it funny that /boot didn't show up in hardware-configuration.nix
<aristid>
Infinisil: the eternal offtopic discussion struggle
<clever>
lejonet: this is the rollback feature in nixos, allowing you to undo almost any change in the system
<clever>
lejonet: and because /boot wasnt mounted, it couldnt update what the "latest" version was
<clever>
so it rolled back every time you rebooted
<joepie91>
the Rust community as a whole is uncannily collaborative, even on small bits and pieces, and the support for small dependencies means that it's much easier to have people independently working on the underlying mechanics for new browser features, without needing to constantly coordinate with the Firefox team on everything they do
<lejonet>
clever: yeah, so far I've understod, I'm still wrapping my head around nixos and the symlink madness, but so far I'm liking it (I'm a long time gentoo user :) )
<joepie91>
this *considerably* lowers the bar for contributing to Firefox
<clever>
i also switched to nixos from gentoo
<tilpner>
joepie91 - As a result of the Leakpocalypse, memory leaks were declared safe
<mog>
giusef, i cant seem to find anything about that in some time. if you are serious about trying it id take a distro better supported for mips like debian and then apply nix packae manager on top of it
<tilpner>
joepie91 - (That's what I was referring to)
<lejonet>
clever: is there anyway to get nixos to handle 2 /boot partitions and keep em in sync? (as this is a server, it has 2 system disks, which btrfs keeps in a RAID1 for rootfs and home, but esp has to be vfat, so can't really do that)
<mog>
giusef, what hardware do you have/
<lejonet>
clever: or will I have to resort to some type of clever hack of my own to do that? :P
<joepie91>
tilpner: unsure how that relates to what you said, have something more specific / with more context?
<pstn>
clever: But isn't that all tcp?
<clever>
lejonet: the old zfs guide on the wiki showed how to make an mdadm mirror for /boot
<clever>
pstn: that covers all protocols
<Infinisil>
srhb: In your PR description, shouldn't it be `with-packages` instead?
<lejonet>
clever: that is what I was pondering about doing
krizoek has joined #nixos
<srhb>
Infinisil: Oh oops, thanks for catching that.
<clever>
lejonet: but i dont think the UEFI can handle vfat inside madm
<Infinisil>
srhb: :)
<tilpner>
joepie91 - You said Rust would prevent resource leaks in Firefox, when resource/memory leaks aren't unsafe under Rust terminology
<lejonet>
clever: got a handy link, or should I just google it? :)
<lejonet>
clever: well, a friend of mine thinks it might work with the old layout, where metadat is at the end of the partition, instead of the beginning and end, which will make them look like any other vfat partition individually
<pstn>
clever then the documentation of networking.nat.forwardPorts.*.destination is wrong, I think
<lejonet>
metadata*
<joepie91>
tilpner: but concretely, under what conditions can memory be leaked? afaik the ownership system prevents this
<clever>
pstn: the forwarding part is only for incoming traffic, and only does tcp
<clever>
pstn: the privateNetwork part does all protocols
<tilpner>
joepie91 - It might prevent some, even with them being safe. But most people assume resource leaks are covered under Rusts guarantees, when it's just a best-effort thing
<joepie91>
right, I'm not considering it a guarantee
<clever>
pstn: and privateNetwork also cuts off outgoing, you need nat to re-enable outgoing traffic
<giusef>
mog I need to run a Linux distribution MIPS based. Debian has it, but ABI is not N32 as I need. I've found someone who created a Debian MIPS port with N32 binaries for GSoC 2013 but I can't find anything more
<joepie91>
if somebody really wants to leak a resource then they can do so; but under normal operation, afaik it won't happen
<joepie91>
that is, unless you do very weird stuff it's not going to occur
<giusef>
mog I only need to run in QEMU
<tilpner>
joepie91 - Yes, it will catch a lot in trivial usage, but browsers don't really qualify as trivial, right?
ambro718 has joined #nixos
<Infinisil>
srhb: Didn't build :(
<pstn>
clever: So, I'd only need a raw nat without any rules in my case, I guess. Thanks.
<joepie91>
tilpner: I more mean it in the sense that "you have to intentionally carry around resources that you no longer need to produce something that can be considered a resource leak"
<joepie91>
is that not the case?
justbeingglad has joined #nixos
jbgi has quit [(Ping timeout: 248 seconds)]
<lejonet>
clever: that would also explain why it didn't accept my changing of the shell lol :P
ertes has joined #nixos
<clever>
lejonet: yep
magnetophon has joined #nixos
<joepie91>
tilpner: so to clarify, I distinguish between "complex things that are done in a sensible way", and "things that are done in a more complex/unusual/non-idiomatic manner than the requirements warrant"
<tilpner>
joepie91 - That's almost a tautology, resources leaks are defined as things you carry around that you no longer need. And adding intention to it doesn't make it any easier to reason about. Most bugs aren't intentional, they still happen
<joepie91>
it's my understanding that in the latter case, you can produce resource leaks; but in the former case, that's negligibly unlikely to happen
<joepie91>
tilpner: that's not true; resource leaks typically occur because somebody forgets to explicitly free a resource
<joepie91>
regardless of whether they keep around any references to it
<joepie91>
(this is different in a garbage-collected language, of course)
anton___ has quit [(Remote host closed the connection)]
<tilpner>
joepie91 - I interpreted "carry around" to mean "don't free"
ona has quit [(Ping timeout: 240 seconds)]
<joepie91>
tilpner: no, I mean it more specifically; in the case of Rust, I mean it in the sense of "explicitly move ownership elsewhere to keep it around, and prevent it from being disposed of"
tokudan has joined #nixos
<joepie91>
in the case of a garbage-collected language, it would be "keep around a reference to it somewhere"
<joepie91>
the exact meaning of the term is necessarily going to vary by language :P
bfrog has joined #nixos
<Infinisil>
Umm, `error: Nix daemon out of memory`
<Infinisil>
I should maybe close some FF tabs
<joepie91>
good thing they aren't Chrome tabs, har har :p
justbeingglad has left #nixos []
<clever>
Infinisil: did you use nix-env -i ?
<Infinisil>
clever: nope, nix-build
<clever>
ah
<clever>
any possible infinite recursion?
<Infinisil>
closed some stuff, hope it works now
<clever>
like default.nix with import ./.
<Infinisil>
Nah
<Infinisil>
I only have 8GB of memory and had quite a lot of stuff open
<Infinisil>
and no swap (currently)
<tilpner>
joepie91 - To choose a different example, Drop is not guaranteed to run, e.g. with panics inside Drop code. Because most Drop implementations will free up resources, not running them will constitute a resource leak (which is rarely a problem). This is not some contrived example, it's quite easy to make a mistake while implementing Drop
<gchristensen>
I know I've done it
<Infinisil>
But, don't leaks not matter anyways on a panic?
<Infinisil>
The process's gonna terminate anyways
<ndash>
Infinisil: if it's a resource the OS will cleanup on exit, yes.
<joepie91>
tilpner: ah, right. I was only thinking ephemeral resources, not persistent ones
<joepie91>
for persistent resources I can see the problem :)
<Infinisil>
Mic92: At least I can still login with GitHub. While I liked it being on GitHub, using MediaWiki is probably a better choice in the long run
<Infinisil>
How do I `nix-shell -p foo` while using the current directory's nixpkgs?
anton__ has joined #nixos
<clever>
Infinisil: nix-shell -p foo -I nixpkgs=.
<Infinisil>
Ahh right, thanks
magnetophon has quit [(Ping timeout: 240 seconds)]
<Infinisil>
How about having an `example` package in nixpkgs
<gchristensen>
have you seen `hello`? :)
<Infinisil>
Hold on I'm not done
<gchristensen>
oh ok
<Infinisil>
This example package would be implemented in all kinds of module systems
<Infinisil>
e.g. as an idris package / node package / go package / etc.
ona has joined #nixos
<Infinisil>
It should not be trivial, have some dependencies, and showcase how packages work for that module system
<Infinisil>
And it could be used to verify that a module system works
cpennington has quit [(Remote host closed the connection)]
<tilpner>
I don't really see the need. If I need an example, I search for build*Package, and have a lot of choice of existing, and usually fairly complex packages
<Infinisil>
I'm also thinking of being able to test e.g. all options of a cabal file, to make sure cabal2nix (or so) works
<Infinisil>
But fair enough, for most module systems there are lots of examples already
<tilpner>
But that wouldn't test cabal2nix's behaviour in case an option is absent, right?
<Infinisil>
Heh right
<Infinisil>
But I think that's not as crucial
nand0p has joined #nixos
<Infinisil>
Also such an example package could be used throughout the documentation, for all different languages, maybe making inconsistencies more obvious
<Infinisil>
Eh, just an idea
<pierron>
Mic92: is it wise to split the wiki effort?
cement has quit [(Ping timeout: 248 seconds)]
Filystyn has quit [(Remote host closed the connection)]
Guest32249 has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<Infinisil>
srhb: Oh regarding your PR description (again :P): It should be `:module Lightyear` instead of `import Lightyear`
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jwiegley pushed 1 new commit to master: https://git.io/v5kai
<NixOS_GitHub>
nixpkgs/master d9b27fa John Wiegley: pythonPackages.logfury: New expression 0.1.2...
NixOS_GitHub has left #nixos []
ona has quit [(Read error: Connection reset by peer)]
ona has joined #nixos
<Mic92>
pierron: the old github-based one is deprecated and the content is moved to the new one
freeman42y[m] has joined #nixos
<pierron>
garbas: ^ is that true?
<Infinisil>
It's been known for a while
magnetophon has joined #nixos
<Infinisil>
Mic92: What's #krebs exactly?
<Mic92>
irc channel
<Infinisil>
I got that much :P
<Mic92>
let's say a Nix user group.
<garbas>
pierron: yup, there are ppl that want to write wiki. that is what we wanted :)
freeman42x[m] has quit [(Ping timeout: 240 seconds)]
<Infinisil>
Mic92: Ah
<pierron>
garbas: so we transitions from wiki, to github to the wiki again? I hope we did not lost much in the process.
<gchristensen>
pierron: well, all the stuff on the old wiki
<gchristensen>
(I still have it backed up, though)
<betaboon>
yeah i saw those in nixpkgs. but i am running into weird problems. i have it working on another service, but can't figure out what the problem is right now
<srhb>
betaboon: Gonna need more than that to go on..
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Gerschtli opened pull request #28484: php pcs: init at 1.3.3 (master...add/php-pcs) https://git.io/v5kom
<betaboon>
but i run into "createuser: could not connect to database postgres: FATAL: role "mypackage" does not exist"
<srhb>
betaboon: So the service is not running as the user you expect
<betaboon>
well it is running as the user i expect, which is "mypackage". but the database "postgresql" is owned by "root" therefore only root should be able to add new users.
<betaboon>
for some odd reason it works like that in another service that i implemented.
<srhb>
I don't think it normally drops privileges that early, does it?
<srhb>
That should ensure privileges are only dropped for ExecStart
yegods has joined #nixos
silver_hook has quit [(Ping timeout: 240 seconds)]
NickSeagull has joined #nixos
<betaboon>
srhb: i set that option ... now that you said to set it i noticed a typo i wrote "Pre" ... those values don't get evaluated, i missed the typo, and therefore it didnt work as expected. correctly using PERMISSIONS works XD
<srhb>
betaboon: Aha! :-)
<srhb>
I didn't notice either... Bad skimming powers.
hiratara_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm closed pull request #28358: (legacy) gstreamer: fetch archive with httpS instead of http (master...pr-gstreamer) https://git.io/v7pD1
NixOS_GitHub has left #nixos []
<NickSeagull>
Hello, has anyone managed to install R on OSX using Nix? I get an old gfortran error...
hiratara has quit [(Ping timeout: 276 seconds)]
hiratara_ is now known as hiratara
yegods has quit [(Ping timeout: 248 seconds)]
yegods has joined #nixos
simukis_ has quit [(Ping timeout: 248 seconds)]
stepcut has quit [(Remote host closed the connection)]
94KABANZJ has joined #nixos
<94KABANZJ>
[nixpkgs] joachifm pushed 2 new commits to master: https://git.io/v5kKd
<94KABANZJ>
nixpkgs/master 518fb60 James D. Trotter: fenics: init at 2017.1.0
<94KABANZJ>
nixpkgs/master 1a7d6f6 Joachim F: Merge pull request #28298 from jamtrott/fenics...
<srhb>
It's an attribute set, and you cannot in general define attributes again.
<srhb>
Try it, define foo twice somewhere, it should error out.
hiratara has joined #nixos
<ashkitten>
how do i set the channel in the configuration?
<ashkitten>
I can't find anything about it, only about using nix-channel
<srhb>
ashkitten: In general you either use channels _or_ set the nixpkgs source in configuration.nix
<ashkitten>
Oh
<srhb>
ashkitten: There are various ways to do so, I have a local git checkout of nixpkgs that I point my configuration.nix at.
<ashkitten>
Well how do I set my system to staging?
<Infinisil>
nixpkgs checkout
<srhb>
ashkitten: If you're comfortable with checkout out nixpkgs somewhere, simply checkout staging and point nix.nixPath = [ "nixpkgs=/your/checkout" ]; at it
elurin has joined #nixos
<Infinisil>
srhb: But that removes the default paths from it
<neonfuz1>
alright I got it
<neonfuz1>
I just knew that you could do the same statement a few different ways
<srhb>
Infinisil: Sorry, how?
<neonfuz1>
like nixos.config.allowUnfree = true; vs nixos.config = { allowUnfree = true; };
<srhb>
neonfuz1: Yeah, same thing. :)
<neonfuz1>
so I was wonderg if the right was syntactical sugar for the left, or no
<neonfuz1>
no it's not
<neonfuz1>
well it is
<neonfuz1>
this one example is
<Infinisil>
srhb: NIX_PATh normally contains also a `nixos-config=/etc/nixos/configuration.nix` and `/nix/var/nix/profiles/per-user/root/channels`, doesn't setting nixPath like this not preserve these?
<srhb>
Infinisil: Oh, you're probably right about that!
<neonfuz1>
but it's not just syntactcal sugar, as you could not have another of the right with a different body
<neonfuz1>
because nixos.config was already defined
<neonfuz1>
so it doesn't "merge" them, because the right is not just syntactical sugar for the left
phinxy has quit [(Read error: Connection reset by peer)]
<neonfuz1>
unless I still misunderstand
<srhb>
neonfuz1: No, that sounds right.
<Infinisil>
srhb: ashkitten: nix.nixPath = [ "nixpkgs=path/to/nixpkgs" ] ++ options.nix.nixPath.default; would work (with `options` in the arguments list at the beginning of the file)
<neonfuz1>
alright cool, thanks
<srhb>
I'm not sure you want to preserve the root channels in this case though
<srhb>
But you probably want to preserve nixos-config
yegods has joined #nixos
<srhb>
I'm honestly not sure what it does when it's the default value anyway.
Fare has quit [(Read error: Connection reset by peer)]
<srhb>
Infinisil: Oh, I missed that in nixos-option nix.nixPath
<srhb>
Still don't think it does anything interesting!
* srhb
proceeds to shave random bits off configuration
<srhb>
This is like unplugging the harddrive for fun!
<Infinisil>
srhb: I actually believe this is the only thing that sets NIX_PATH
<Infinisil>
:P
<srhb>
Yes, I think so too.
<Infinisil>
Btw, a better way to check an option: nix-instantiate --eval -E '(import <nixpkgs/nixos> {}).options.nix.nixPath.default'
<srhb>
Why is it better?
<srhb>
I was assuming that was what nixos-option did.
<Infinisil>
Well, nixos-option doesn't work for me often
<srhb>
Oh.
mizu_no_oto has joined #nixos
<srhb>
I'll keep that in mind, thanks.
<Infinisil>
Not sure why really
<srhb>
Magic, probably.
<Infinisil>
Ah, what's cool with nix-instantiate, is that you can also check the resulting config
<manveru>
nixos-option depends on your nixos config to get the current values, so if there's an eval failure there it won't output anything
yegods_ has joined #nixos
yegods__ has joined #nixos
<Infinisil>
manveru: nix-instantiate also needs to evaluate it though
yegods has quit [(Ping timeout: 248 seconds)]
<manveru>
i don't think it does with the way above
mizu_no_oto has quit [(Ping timeout: 240 seconds)]
<Infinisil>
manveru: I mean it needs to evaluate it to determine the values
yegods__ has quit [(Client Quit)]
<Infinisil>
It wouldn't be evaluated if --parse was used
<manveru>
you query nixpkgs, not <nixos-config>
yegods_ has quit [(Read error: Connection reset by peer)]
<Infinisil>
But nixpkgs/nixos uses <nixos-config>
<manveru>
oh :)
<Infinisil>
:P
<manveru>
maybe lack of --strict?
<Infinisil>
--strict is useful for e.g. this: nix-instantiate --eval --strict -E 'map (x: x.name) (import <nixpkgs/nixos> {}).config.environment.systemPackages'
<Infinisil>
which checks the installed systemPackages
<Infinisil>
although its output is not very readable admittedly
<Infinisil>
[ "xterm-327" "git-mini...
<Infinisil>
it's okay I guess
<ilpianista>
how do I get make output when using "nix-build"?
zennist has joined #nixos
<Infinisil>
ilpianista: You probably wanna use nix-shell instead
nix-gsc-io`bot has joined #nixos
<zennist>
newbie question: I have a shell.nix that sets up the desired env of my project, but no default.nix for the final product; how can i make the default.nix such that it just installs everything needed as in shell.nix?
<zennist>
if I issue nix-env -f shell.nix -i it tells me I have to have a src specified in the derivation
<ilpianista>
Infinisil: I did `nix-build -A pkgs.qoauth`, which failed with "1 dependencies couldn't be built". How should I use nix-shell to do the same?
<srhb>
zennist: This is sort of specific to the kind of derivation.
<srhb>
zennist: For instance, some languages have specific frameworks for building packaegs.
<srhb>
zennist: Like python with its own buildPythonPackage
<srhb>
zennist: The other way around is usually easier.
<Infinisil>
ilpianista: I'm not 100% sure, but `nix-shell -A pkgs.qoauth` then `genericBuild` should leave you with the build artifacts
<zennist>
so do you know how e.g., 'buildPythonPackage' defines the env to be installed as package?
<srhb>
zennist: I don't understand that question.
<zennist>
i probably should check it out myself
<ilpianista>
Infinisil: nope, it does the same of nix-build
<zennist>
for example, say I'm developing a specific build environment for python
<srhb>
zennist: Usually you don't install python packages directly, but use something like python.withPackages
<Infinisil>
ilpianista: Then you should do `nix-shell -A pkgs.<package that doesn't build>`
<srhb>
(Installing it directly won't teach Python how to find it)
<ilpianista>
Infinisil: it's qoauth :-)
<Infinisil>
ilpianista: When nix-shell fails like this, it means that a dependency failed to build
<zennist>
if I'm distributing python as a tool that's easy and i have no trouble defining a default.nix for that; but want i want is to define the specific environment I have (a set of dependencies on a set of programs; but no actual program of its own) as a default.nix to be installed
<Infinisil>
ilpianista: nix-shell doesn't build qoauth, it only sets up an environment to build it by building all dependencies
<ilpianista>
Infinisil: "output path ‘/nix/store/asnb75yr3lygkw770ylz2267s687k74h-qoauth-2.0.0.tar.gz’ has sha256 hash ‘06jrmj52v195j738nibiyzxlql33mbw48anj1aa5z6jrw5ivdikr’ when ‘0cz3hm1bgc4z1zf6fa1l6ddfq165yv6pjxh77ph37lqhcjc0b052’ was expected"
<srhb>
zennist: OK, let's say it just is a collection of python packages you want to make python know, yes?
<Infinisil>
ilpianista: Ah, well that's a hash mismatch
<zennist>
something like that i guess
<zennist>
think of it as shell.nix
<ilpianista>
Infinisil: hash of what? of the source tarball?
<Infinisil>
ilpianista: Looks like it
<zennist>
maybe what i need is just an empty derivation format
<srhb>
zennist: That would be a valid default.nix file for nix-build
<srhb>
zennist: It gives you a python interpreter that knows about numpy.
<Infinisil>
ilpianista: If you changed anything of the fetcher, you might get a different hash
Fare has joined #nixos
bennofs has quit [(Ping timeout: 240 seconds)]
<bendlas>
An Idea: What do you think about introducing build snapshots to Nix? Think Docker - like snapshots of a build directory in progress. By default, this could be done after configurePhase and buildPhase, but larger builds, like browser engines could do this for various submodules. Just think of the possible speedups when working on the later stages of a build ...
<zennist>
srhb: okay thanks will give it a try
<zennist>
<ilpianista>
Infinisil: ok, now I got it, since I read that strange hash I thought it wasn't related to that. Thank you!
<srhb>
bendlas: That's already doable, just carve up packages into several stages
<srhb>
bendlas: Have one depend on the "configurePhase" derivation of the other, etc.
<Dezgeg>
it's for C/C++ code where you don't know the exact dependencies before actually compiling the file the firt time
<Infinisil>
Dezgeg: Sooo, how do these dependencies get resolved then?
jellowj has joined #nixos
<Dezgeg>
it only matters for incremental builds
jellowj_ has quit [(Ping timeout: 248 seconds)]
<Dezgeg>
say, foo.c depends on bar.h but not baz.h. on the first ninja run, you don't care because you need to compile foo.c anyway. but after that, you can edit baz.h and ninja knows that it doesn't have to recompile foo.c
<Infinisil>
Dezgeg: Hmmmmm
<Dezgeg>
and by "foo.c depends on bar." i mean it has '#include "bar.h"'
<Infinisil>
AH just read that from the ninja manual
<Infinisil>
Okay I got it
<ashkitten>
Hey, how can I install firefox nightly?
<clever>
the major difference from the old buildEnv, is that these remain as seperate entities within nix-env, and you can just nix-env -iA nixos.firefox to upgrade something imperatively
<clever>
and nix-env -e firefox to imperatively remove something
<clever>
so you can add it to the config.nix, and then nix-env -iA it, to skip updating everything
<bendlas>
I've been thinking about keeping most of my installs in a bin folder at ~, that would consist of wrappers for `nix-shell -p`. With some cleverness, this could even be mostly done by pure softlinking
<Infinisil>
Also maybe nix will eventually be able to update a (human readable) nix file with the installed packages, so you can `nix-env -iA`, but it's more declarative
<clever>
bendlas: that is basicaly what nix-env does, it manages a buildEnv in ~/.nix-profile/
<bendlas>
yep
<bendlas>
but i can't easily check nix-env into a git repo ...
<clever>
just set a rule that you only ever nix-env -ieA nixos.mystuff
<clever>
then check config.nix into git
<bendlas>
also, nix-env keeps everything installed, while nix-shell -p loads on demand
<Infinisil>
I'd prefer the former if I'm using it regularily
joehh has joined #nixos
<bendlas>
right, and those I keep in my system profile. but my system profile is still full of crap that I picked up while i didn't know about nix-shell -p
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
markus1189 has joined #nixos
<Infinisil>
Same :P
<Infinisil>
I generally don't use nix-shell a lot though
<bendlas>
I do
mizu_no_oto has joined #nixos
markus1199 has quit [(Ping timeout: 240 seconds)]
<bendlas>
it's really useful to bootstrap into repl-driven workflows
<Ankhers>
Does anyone know an approximate timeframe for when the gcc 5.4.0 fix will reach nixpkgs unstable?