<cole-h>
ldlework: Note: config.* will almost never do what you expect. AFAIU, it's supposed to be a read-only snapshot (or close enough) of your current configuration.
mauli has joined #nixos
<infinisil>
cole-h: Right and wrong. The `config` argument from the top of the module is the final config set. But assigning `config = ` in the module itself is actually what adds definitions for this final set
<infinisil>
Note that e.g. `environment.systemPackages = ...` is just a shorthand for `config.environment.systemPackages = ...`
Chiliparrot has quit [Ping timeout: 260 seconds]
<infinisil>
:)
<cole-h>
My whole world has been flipped upside down
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<superbaloo>
and I can't get carnix to work, anything I try fails
dxtr has quit [Ping timeout: 256 seconds]
Chiliparrot has joined #nixos
kvda has joined #nixos
<energizer>
infinisil: iirc it yells at me when i try to config.foo = bar. what's that about?
<superbaloo>
ha well, cargo changes the cargo.lock format in 1.41
<superbaloo>
which explains the failures I see
<infinisil>
energizer: Do you think this should be enough information for me to know the problem?
mauli has quit [Ping timeout: 260 seconds]
asbachb has joined #nixos
<infinisil>
(spoiler: it's not!)
<energizer>
infinisil: yes i did think that, since there's only ever one error message ive seen from that and iirc it happens every time. but ok i guessed wrong :-)
<infinisil>
config.foo = bar is literally the most generic module system assignment
<infinisil>
I could give you a dozen values for foo and bar that each give a different error
drakonis1 has joined #nixos
<asbachb>
Hi. Is there a way to prevent packages from beeing compiled. I want that rebuild switch fails in case the package is not available in cache.
<{^_^}>
[nixpkgs] @Infinisil pushed 2 commits to release-20.03: https://git.io/Jfarg
justanotheruser has joined #nixos
<cole-h>
adisbladis: When you get a nixops2 package up and running, let me know... I'm too lazy to figure out python packaging atm, but ready to hop on master.
pjan has quit [Quit: さようなら]
<asbachb>
energizer: Thanks.
gustavderdrache has quit [Quit: Leaving.]
<NobbZ[m]>
I have lorri as a demon + direnv. When in a "direnv+lorri" environment, lorri uses my systems nixpkgs channel, though when I do lorri shell then the environment is correctly built against my users nixpkgs channel.
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JfarK
<energizer>
infinisil: is that saying modules aren't allowed to have an attr named `config`?
orivej_ has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
markus1189 has quit [Ping timeout: 265 seconds]
markus1189 has joined #nixos
<quinn>
energizer: most modules have a config attr
<energizer>
so what's that error condition about?
<infinisil>
energizer: If you assign `options` or `config` at the top-level of your module, you can't assign any option values at the top-level, but have to move them to config.* instead
<quinn>
energizer: see line 174: Massage a module into canonical form, that is, a set consisting of ‘options’, ‘config’ and ‘imports’ attributes. */
<infinisil>
So you can't have `options = ...; environment.systemPackages = ...`, but instead you need to do `options = ...; config.environment.systemPackages = ...; `
<quinn>
infinisil: this is a better explanation than mine (cause you understand it better than me) nice
<energizer>
infinisil: i see, thanks
<infinisil>
I wish the error explained it a bit better, the current one is a bit confusing
<asbachb>
But I wonder why hydra build does not fail.
mauli has joined #nixos
<cole-h>
Have you tried updating your channels?
<asbachb>
cole-h_: Yes. That's latest.
<lovesegfault>
Trying to build a system optimized for skylake using nixpkgs.localSystem.platform and setting gcc. arch and gcc.tune. I want to build this on a non-skylake server, how can I do that?
<lovesegfault>
I think this counts as cross-compiling?
evanjs has quit [Read error: Connection reset by peer]
<mac10688>
So I've been trying the past few days to try to set a background for my xmonad environment. clever recently assisted in showing that the code should be loading from .xsession. I added code to load a custom background using feh which I know that works because I can run it in terminal and it changes the background. But it doesn't work when I load xmonad.
<mac10688>
How are people writing custom config files to load a background?
evanjs has joined #nixos
philr_ has joined #nixos
<quinn>
mac10688: i didn't think of this when we last spoke, but i think xsession is only triggered if you use a display manager. if you start xmonad from a tty it probably won't work.
day|flip has joined #nixos
<mac10688>
I think I am using a display manager
<mac10688>
I tried googling earlier
<{^_^}>
[nixops] @adisbladis merged pull request #1351 → Add typing extensions as an explicit dependency → https://git.io/JfaoM
<mac10688>
I think a display manager is basically the login page right? In which case I am using sddm
<mac10688>
I'm probably wrong so feel free to correct me
<Henson>
mac10688: I did a bit of a hack to run the xmonad window manager on top of KDE. So KDE would set the background for me, and instead of loading the kwin_x11 window manager would load xmonad instead.
<quinn>
mac10688: that's strange. try using .xsession instead of .xsessionrc, i think i used .xsession a year ago when i had that setup
iyzsong has joined #nixos
<Henson>
mac10688: that way I'd get all of the goodies of KDE along with a real tiling window manager.
<quinn>
Henson: why'd you do it that way?
<quinn>
oh, what are those goodies?
<mac10688>
yeah I am totally set on .xsession now. clever even showed me the code where it's looking for ~/.xsession
<mac10688>
yeah, I'm curious Henson
<Henson>
quinn: pretty much all of the nice graphical configuration stuff in the KDE system settings (multiple monitor config, bluetooth, network-manager, printers, etc.)
<day|flip>
trying to used nix from a non nixos. can't seem to get chromium started. it keep giving me this error. https://hastebin.com/dozamoxuse.shell
mauli has joined #nixos
<Henson>
I think I also had spectacle working for screen shots and krunner to launch apps. I had xmobar at the top of my screen.
<Henson>
I've been a long time KDE user and so I really like the system configuration options and didn't want to learn how to do it with command-line tools.
<quinn>
Henson: that's interesting and cool
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Henson>
I eventually found trying to configure xmonad to do what I wanted to be too complicated. I love Haskell, but don't want to spend 3 days reading about some crazy Haskell abstraction in order to make a really simple change to my window manager. I really liked xmonad, but it eventually became too frustrating to maintain.
<quinn>
mac10688: try adding &> /tmp/debug to the end of your feh command and then read that file
<mac10688>
xmonad doesn't seem big on abstractions to me
<mac10688>
well what's weird, is that I had xmobar and stalliontry and all kinds of stuff working. Then through some confusion I accidentally renamed the file
<mac10688>
now I've renamed it to .xsession but it won't read from it anymore
<mac10688>
it's driving me bonkers lol
<adisbladis>
cole-h: Nice! :)
<quinn>
mac10688: if you compare Xorg itself and the library to the way it's exposed by xmonad, it's definitely an abstraction
<adisbladis>
I was just going to move my personal machines over to nixops master =)
<Henson>
mac10688: I can't remember exactly what it was that had me stuck, but the effort to learn how to do it was greater than my desire to do it. That happened for a bunch of things, so eventually I got stuck in an uncomfortable place and couldn't get out.
<quinn>
mac10688: you can just sudo rm ~/.xsession and then remake it if it's acting up
<cole-h>
adisbladis: My main motivation was the --test flag for deploy :P
<Henson>
mac10688: do you have your xmonad stuff under source control?
<mac10688>
I do!
<Henson>
mac10688: but the thing that has gone wrong is outside of your source control?
<mac10688>
finally, someone lets me feel smart on this irc channel lol
<Henson>
:-)
<mac10688>
ok Henson you got me there. I've been adding more stuff to my source control
mauli has quit [Ping timeout: 272 seconds]
<mac10688>
I kind of just rushed past all this stuff a year ago to get it working as a proof of concept
<mac10688>
now I'm going back to clean it up and trying to make it look nicer
<mac10688>
quinn: hey I think i did it! I got the background to show with services.xserver.displayManager.sessionCommands. I didn't get .xsession to work but I finally got something!
<ldlework>
error: 'mkOption' at /nix/var/nix/profiles/per-user/root/channels/nixos/lib/options.nix:30:5 called with unexpected argument 'name', at /nixcfg/modules/nixos/services/irccat.nix:59:17
nix has joined #nixos
<ldlework>
how is that right? name is an argument to mkOption?
nix has quit [Remote host closed the connection]
<energizer>
no it isnt
meh` has quit [Ping timeout: 246 seconds]
<energizer>
i mean, you're calling mkOption {name = something} but mkOption doesnt have a name parameter
<hexa->
and I end up with error: value is a string while a list was expected, at /nix/store/vgyn5vl07sgqchsrmjxkhdw6wkv55ljd-991dk606y0dk4m9vpjyji8pkdpm15ic2-nixpkgs-src/lib/trivial.nix:80:18
<veleiro>
what can you do about not having your network interface setup properly in configuration.nix, but require network to parse your configuration.nix
<energizer>
you can imperatively set up network
<energizer>
in the installer
drakonis_ has joined #nixos
<hexa->
or rollback if you have had a working configuration
<veleiro>
hmm
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
<energizer>
or you can build an image with everythign set up on another computer and transfer it over
<veleiro>
true, i went with the rollback
<veleiro>
it worked, thanks hexa-!
jluttine has joined #nixos
<veleiro>
i just noticed times ive had to use 'nixos-rebuild boot' in place of switch because immediate switches without reboot would result in "cannot connect to daemon"
lordcirth has quit [Remote host closed the connection]
lordcirth has joined #nixos
cmk_zzz_ has joined #nixos
cmk_zzz has quit [Ping timeout: 260 seconds]
mauli has joined #nixos
<la-s>
does anybody here know what to do when the cached source for something you're trying to build is corrupted?
<spinlock[m]>
Is it possible to get a NIX_PATH entry, as given?
<spinlock[m]>
practise" for that
<spinlock[m]>
When I add an entry to the nix path (ex: `foo=/etc/nixos/bar`), either through `$NIX_PATH` or the `-I`, I can reference that path with the handy `<foo>` angle brackets. This is great, however I'm finding that there are times where I want the verbatim path, and not the store `/nix/...` path. Is there a common way to do this? I suspect I can write something with `builtins.getEnv`, but I'm curious if there's a "best
mauli has quit [Ping timeout: 240 seconds]
<la-s>
Maybe you should just make foo be a separate environment variable?
<la-s>
it to me sounds like you're misusing NIX_PATH
<spinlock[m]>
"abusing" is probably more accurate. that sounds like a good idea though actually. if it's a separate variable then I think I can access it directly and also push it into `nix.nixPath` 🤔
<{^_^}>
[nixpkgs] @davidak opened pull request #88744 → nixos/chromium: add example and update URL → https://git.io/Jfa1x
buckley310 has quit [Quit: Connection closed for inactivity]
mauli has quit [Ping timeout: 260 seconds]
<ldlework>
Looking for help maintaining Styx, the nix-based static content generator - if anyone's interested in that.
<la-s>
nora: the one that is cached by the build system, but I found out that it (alsa-lib source) wasn't corrupted, the contents in the git repo are just inherently different from the packaged versions on some sites, since it uses autoconf. It would still be useful what to do in that case though.
<la-s>
to know*
<puzzlewolf>
What is the problem you're having?
<la-s>
well, none now, I just added the autoreconfhook and fixed it
<puzzlewolf>
Then probably open a PR to fix it for everyone :D
{^_^} has quit [Remote host closed the connection]
ddellacosta has quit [Ping timeout: 272 seconds]
{^_^} has joined #nixos
rogue_koder has joined #nixos
mauli has joined #nixos
<puzzlewolf>
Or first open an issue ;)
mac10688 has quit [Quit: WeeChat 2.7.1]
mac10688 has joined #nixos
Maxdamantus has quit [Ping timeout: 265 seconds]
Maxdamantus has joined #nixos
mauli has quit [Ping timeout: 260 seconds]
drakonis has quit [Quit: WeeChat 2.8]
palo1 has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
<hyper_ch_>
nixos unstable still having build problems
tom39291 has joined #nixos
joebobjoe has joined #nixos
<{^_^}>
[nixpkgs] @moinessim opened pull request #88750 → dotnetCorePackages: Add support for Darwin → https://git.io/JfaD9
dan224 has joined #nixos
<joebobjoe>
I have been having a persistant issue with nix on darwin. I think that whenever mac updates it overwrites /etc/zshrc where nix sources itself
<joebobjoe>
so I lose access to nix in my shells all the time
<joebobjoe>
when I look in /etc, .zshrc and .bashrc are read-only. are they not supposed to be written to?
<energizer>
normally environment variables aren't supposed to be set in zshrc
o1lo01ol1o has joined #nixos
<energizer>
it's more typical to do that in ~/.zprofile
<joebobjoe>
whenever I run a nix-env operation like -u I get warning: name collision in input Nix expressions, skipping '/Users/andrewmckinlay/.nix-defexpr/channels_root/nixpkgs'
<joebobjoe>
fixed it. seems like the intstaller installed different ~/.nix-defexpr/channels and ~/.nix-defexpr/channels_root pointing to the same thing. maybe I ran the installer at different times, as one link is from march and the other may
<jakobrs>
Is there any way to make building the NixOS iso faster?
<kloenk>
srhb: I tried to build alacritty, becaus it fails on my hydra, but afer running this `curl https://hydra.nixos.org/build/118982115/reproduce | bash` it only downloaded nixpkgs and printed a cryptic line
<kloenk>
But that didn't help me with my hydra sha256 miss problem. but thanks
<srhb>
kloenk: And yes, you can run nix-build with the arguments it printed for you, but it might have cleaned up the nixpkgs path, so you'll have to reproduce that on the specific rev, (eg by providing the url to a tarball of that rev from github)
<srhb>
kloenk: Er, and with --check added to rebuild
o1lo01ol1o has quit [Remote host closed the connection]
<srhb>
kloenk: But I'm not sure what you're really trying to achieve
<srhb>
kloenk: At a guess, you want to know why your hydra build failed. Are you sure that's the same rev?
<s1341>
hi all
<s1341>
how do i specify a secret for initrd ssh?
<srhb>
kloenk: Easy check is to see whether it has the same outPath
<s1341>
i now get the following:
<s1341>
- boot.loader.initrd.secrets values must be unquoted paths when
<s1341>
using a bootloader that doesn't natively support initrd
<s1341>
boot.initrd.secrets = {
<s1341>
secrets, e.g.:
<s1341>
};
<s1341>
"/etc/secret" = /path/to/secret;
<s1341>
Note that this will result in all secrets being stored
<s1341>
world-readable in the Nix store!
<kloenk>
Yes, I want to know why my hydra fail, but I also want to know why hydra.nixos.org did not fail
<srhb>
s1341: You should use a pastebin for multi line pastes in the future :)
<s1341>
srhb woops sorry.
<srhb>
kloenk: Sure. Try checking the outPath of your failed hydra build.
<s1341>
i think i'm using grub...
<s1341>
doesn't grub support initrd secrets
<s1341>
?
<kloenk>
s1341: the grub secret is not merged yet
<s1341>
kloenk: uh. i see.
<s1341>
so how do i do this without grub secrets support?
Vikingman has quit [Remote host closed the connection]
<s1341>
kloenk: how do i get that PR into my system?
<{^_^}>
[nixpkgs] @alexbiehl opened pull request #88762 → Properly add /nix/ and /nix/store/ first the the tarball → https://git.io/JfaHZ
<kloenk>
srhb: did `nix build` drop the `--check` flag?
<srhb>
kloenk: It's nix-build that has --check, it's called something else for nix build.. --repeat 1 or something.
<kloenk>
s1341: probaply clone you own nixpkgs, rebase the pr on it, and the set NIX_PATH=nixpkgs=/path/to/checkout
<srhb>
kloenk: But the reproduce script doesn't specify either by default
<kloenk>
srhb: thanks, for my own hydra im using flakes, so i have a flake uri to rebroduce
<s1341>
kloenk thanks
<srhb>
kloenk: In general, when you're confused why one build is failing and one you think is the same isn't, a good way to debug is to get the drv from each of them and nix-diff them to find the difference.
GiGa has joined #nixos
GiGa has quit [Changing host]
GiGa has joined #nixos
<srhb>
kloenk: If there isn't one, you've likely hit an impuriy.
<GiGa>
Morning all
<kloenk>
srhb: thanks, didn't know of nix-diff
<srhb>
kloenk: But in my experience that's pretty rare, and accidental differences in the drv is by far the most common.
<GiGa>
I'm trying to get pcov (PHP code coverage tool) running in a container, yet I'm always told the attribute is missing.
<GiGa>
environment.systemPackages = with pkgs; [ php74Packages.pcov ];
<GiGa>
I'm not sure why though
leah2 has quit [Ping timeout: 260 seconds]
<immae>
GiGa: on my side (latest unstable) it’s indeed not there
<maralorn>
c00w: I don‘t understand why you gave up on `regenGoMod`. Which property das `go mod tidy` have, that is problematic? My understanding is, that go mod tiday crawls the source code to find package references. That sounds very pure to me.
<maralorn>
c00w: I am not saying you're wrong. I have no clue about `go`. I am just very curious because it looked like such a useful option.
<immae>
I see it there though: php74Extensions.pcov
<GiGa>
Aha, let me try that.
noudle has joined #nixos
<immae>
there were recent changes in php packaging that are not in 20.03 yet
<GiGa>
immae: that seems to have got it, thanks
<immae>
cool
<GiGa>
Now I just need to work out how to tell configuration.nix to include that in the php.ini config
<GiGa>
Presumably just chuck the php.ini stuff in phpOptions?
<immae>
GiGa: use php74.withExtensions (p: [p.phpcov]) as "phpPackage" in phpfpm config
knupfer has joined #nixos
<GiGa>
immae: I've currently got this container set to use Apache, rather than Nginx. Do I still need that line?
<immae>
it will be elsewhere but yes you need to use this as "php package"
<immae>
you use apache "php module" right?
gnidorah has joined #nixos
__monty__ has joined #nixos
<immae>
in that case it’s services.httpd.phpPackage = php74.withExtensions (p: [p.phpcov])
<immae>
hmm actually you will get only this package, so you need to specify the default ones too it seems something like: php74.withExtensions (p: php74.enabledExtensions ++ [p.phpcov])
<immae>
well you need to prepend it with pkgs if you don’t use the "with pkgs;"
thc202 has joined #nixos
<immae>
pkgs.php74.withExtensions (..)
MmeQuignon has joined #nixos
marcusr has quit [Remote host closed the connection]
marcusr has joined #nixos
alp has joined #nixos
<la-s>
is there any optimal way to share overlays between the system configuration and the user environment?
<GiGa>
immae: That now gets error: "attribute 'phpcov' missing, at /var/lib/containers/evitabumgmt/etc/nixos/configuration.nix:29:49" on that line, so we're making progress :)
<GiGa>
immae: That now gets error: "attribute 'phpcov' missing, at /var/lib/containers/evitabumgmt/etc/nixos/configuration.nix:29:49" on that line, so we're making progress :)
<GiGa>
Sorry, didn't mean to send that twice
jgeerds_ has joined #nixos
<immae>
GiGa: yes it’s because I copied your name, but the real name is pcov :p
<GiGa>
immae: I did wonder, so I tried that too: error: attribute 'pcov' missing, at /var/lib/containers/evitabumgmt/etc/nixos/configuration.nix:29:49
Maxdamantus has quit [Ping timeout: 256 seconds]
<immae>
May I see what your configuration.nix looks like? I just tested it locally and it works fine for me
<GiGa>
Sure, do you want the whole lot?
<immae>
Just the line with phpPackage = ...
Maxdamantus has joined #nixos
<kloenk>
srhb: found the impuritie: my isp is shity, und files are corrupted while downloading
<kloenk>
la-s: I think not yet: you can use flakes, but they are experimental
<srhb>
la-s: Not optimal, but you can eg. set both <nixpkgs-overlays> and the nixos option to point at the same set of overlays.
<la-s>
thanks!
<srhb>
la-s: There isn't currently a way that I know of to do it with less than two options.
<immae>
GiGa: Ah, I have it working with php74.withExtensions (p: [php74.extensions.pcov])
<immae>
I don’t understand why it’s not in "p" though...
<immae>
note that as I said you should probably include "standard" extensions (like pdo & co), so it would be more like php74.withExtensions (p: php74.enabledExtensions ++ [php74.extensions.pcov])
alp has quit [Ping timeout: 252 seconds]
<immae>
(GiGa: and prepend pkgs. where necessary of course)
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
civodul has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<GiGa>
immae: that seems to have done it, thanks.
fendor has quit [Remote host closed the connection]
<immae>
cool
<GiGa>
Do I still need php74extensions.pcov in systemPackages?
fendor has joined #nixos
<immae>
it depends
<immae>
in systempackage it will permit you to run the pcov command from your shell (if there is such things). If you don’t need it then you may remove it
<GiGa>
Great, thank you
<GiGa>
I'll work that out then, as I'm still reading the docs :)
<GiGa>
I shall make notes so I don't have to ask again :) .
orivej_ has joined #nixos
<GiGa>
Got to get my head around Codeception and using pcov next
nixbitcoin has quit [Remote host closed the connection]
KeiraT has quit [Remote host closed the connection]
nixbitcoin has joined #nixos
o1lo01ol1o has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
KeiraT has joined #nixos
Valodim has left #nixos [#nixos]
leah2 has joined #nixos
o1lo01ol1o has quit [Ping timeout: 246 seconds]
nschoe has quit [Ping timeout: 252 seconds]
Gohla has quit [Quit: Bye.]
Gohla has joined #nixos
<la-s>
does anybody know of a way to evaluate configuration.nix so that you can use the attributes defined within it, also taking into account the imports within it?
<LnL>
nix repl '<nixos/nixos>'
<LnL>
you can inspect config, etc. with that
<{^_^}>
[nixpkgs] @chrbauer opened pull request #88764 → Fix dependencies issues of postgresql-syntax and re+base → https://git.io/Jfa7Y
<la-s>
Thanks! That is surprisingly simple
markus1189 has quit [Ping timeout: 246 seconds]
<LnL>
even works on non nixos systems, you just have to also add -I nixos-config=./configuration.nix there
<GiGa>
in fact, everything in services.httpd.phpOptions seems to be getting ignored
sigmundv has joined #nixos
chloekek has joined #nixos
<GiGa>
immae: Should adding the extension like we did also add it in php.ini?
tobeportable has joined #nixos
<GiGa>
Interesting, the PHP ini for Apache _is_ being updated, just not the command line one
<GiGa>
Is there a way to configure php.ini for the command line PHP?
artistsvoid has quit [Quit: leaving]
<srhb>
GiGa: Looks like php itself takes arguments related to extensions. Check out the comments and the links to docs in the php/defaut.nix ?
FRidh has quit [Quit: Konversation terminated!]
<srhb>
GiGa: Apologies if I'm repeating something, I didn't read the full backlog.
<GiGa>
srhb: I'll try to find that file in the repo, one sec.
<GiGa>
srhb: not repeating at all
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<GiGa>
OK, got the file
<srhb>
GiGa: So what skimming it implies to me is that you can use php.withExtensions as your php package of choice to get whatever extensions composed into the php.ini
<GiGa>
srhb: That seems to work for Apache / Nginx but doesn't apply to the command line version.
<srhb>
GiGa: You're just putting a bunch of php stuff in systemPackages, not a composed withExtensions php :)
<srhb>
GiGa: I think that's the key you're missing.
<GiGa>
So that's added pcov to the Apache called version of PHP, but I need to add it to the command line PHP so I can work out code coverage with codeception
<srhb>
If that's the "command" you're referring to, at least.
o1lo01ol1o has joined #nixos
<srhb>
GiGa: iow you want something like environment.systempackages = [ (php.withExtensions ... ) ];
<srhb>
GiGa: Does that make sense?
<GiGa>
srhb: not sure
<GiGa>
Aha, possibly
<srhb>
GiGa: So, right now you have php74 in your systemPackages
<GiGa>
srhb: Yep, I've replaced it with (php74.withExtensions (p: [pkgs.php74.extensions.pcov]))
<srhb>
GiGa: I think you can essentially consider that equivalent to (php.withExtensions (p: []))
<srhb>
GiGa: Right, something like that :)
<srhb>
Ah, turns out there's some default extensions, but you get he basic idea.
<{^_^}>
[nixpkgs] @umazalakain opened pull request #88770 → agda: fix typo in library management documentation → https://git.io/Jfa5V
<GiGa>
srhb: that's certainly heading the right way. How do I now customise the php.ini for that though?
<GiGa>
srhb: that config has overwritten the default extensions too, so I need to work out how to add to the list, rather than replace it
orivej has quit [Quit: No Ping reply in 180 seconds.]
mallox has quit [Quit: WeeChat 2.8]
<GiGa>
Think I just found how to do that, just testing
o1lo01ol1o has quit [Remote host closed the connection]
<aanderse>
GiGa: you should consider phpfpm instead of httpd.enablePHP
orivej has joined #nixos
<asymmetric>
is anyone around to help me with updating the iosevka font? i'm having trouble with node packaging
<GiGa>
And then viewing the php.ini additional config as stated via "php -i" shows pcov is being loaded (it's right at the bottom of the file, not in the alphabetical list).
<GiGa>
So that's good
<srhb>
GiGa: Ah, so it was right.
<srhb>
Good :)
<GiGa>
So I now need to work out how to add the extra config ("pcov.enabled=1" etc.) to php.ini
<GiGa>
srhb: yes, sorry, I expected it to still be sorted alphabetically, not appended.
<GiGa>
My bad
<srhb>
GiGa: the extraConfig parameter from buildEnv
<GiGa>
srhb: I think so, I'm just not sure where to put php.buildEnv in the config
<srhb>
freeman42x[m]1: What do you mean exactly? I mean, given that we can import nix file from wherever, the number of ways to specify even just systemPackages is innumerable.
<aanderse>
GiGa: we all start as noobs :D
Darkmatter66_ has quit [Ping timeout: 265 seconds]
<srhb>
freeman42x[m]1: In other words, none of my packages are installed from any of the paths you list.
<GiGa>
srhb: Is your solution in the systemPackages section or outside of it?
<freeman42x[m]1>
srhb: I am obviously only asking about the main ways to install packages
<GiGa>
aanderse: true
<srhb>
freeman42x[m]1: Okay, that was not obvious to me. I don't have a good gauge of what is more common in the community.
<srhb>
GiGa: It doesn't matter. I prefer let... in; like aanderse proposed.
<srhb>
GiGa: Keeping my systemPackages free of longer expressions.
<aanderse>
and of course you can reuse the expression too then
<GiGa>
srhb: I think "let" is probably cleaner. Apparently I just don't know how to do that properly!
<GiGa>
aanderse: yep
<GiGa>
I'm just letting in the wrong place
<aanderse>
currently php.buildEnv + apache httpd is broke, so looking at a workaround...
<srhb>
GiGa: Yes, you can't `let ... in ...` in an attrset nakedly
<srhb>
GiGa: eg this is illegal
<srhb>
> { let foo = 42; in foo; }
<{^_^}>
error: syntax error, unexpected LET, at (string):312:3
<aanderse>
GiGa: i haven't tried to build this yet, but i think this gets you a fair bit closer: https://pastebin.com/FDaTq0ex
<srhb>
GiGa: It's short for argument set, the set of arguments at the top of many nixos modules, `{ ... }:`
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos
<aanderse>
oh maybe Jan Tojnar already fixed services.httpd.phpPackage... nice :D
<srhb>
asbachb: It looks like you're trying to pass in coreutils in the NixOS module args
<srhb>
asbachb: As if it were an argset passed to callPackage
<srhb>
asbachb: It's not.
<srhb>
asbachb: The function that composes nixos modules is completely different
<srhb>
asbachb: So arbitrary packages are not available like that. Only config, pkgs, options, lib, basically.
<srhb>
asbachb: You want to grab coreutils from the pkgs argument.
<GiGa>
aanderse: that's much closer now. I can see the php.ini has the extra config now too.
<GiGa>
aanderse: so now I just have a codeception config issue, but that's fine
<GiGa>
srhb: noted, thanks for the definition
o1lo01ol1o has joined #nixos
<GiGa>
aanderse: I had to take out phpCust.composer as that wasn't valid apparently
<aanderse>
GiGa: oh... phpCust.packages.composer
<aanderse>
sorry
<GiGa>
(attribute 'composer' missing(
<GiGa>
aanderse: ah, yes, I should have spotted that
<asbachb>
srhb: Thanks. Got it fixed.
<GiGa>
That now builds
<GiGa>
I think I even understand what's going on
<GiGa>
Thanks immae, srhb and aanderse
<GiGa>
Now to fix codeception :)
meh` has quit [Ping timeout: 264 seconds]
<GiGa>
Codeception still says "no code coverage driver available", so I need to work out why that is
<GiGa>
but I really appreciate all the help with this folks.
<GiGa>
I think there's just one hurdle left
<aanderse>
:)
meh` has joined #nixos
<GiGa>
I need to take a break, been sat here all morning.
<GiGa>
Thank you for your help folks. I'll try to type this up later so I can blog about it once I've got everything working. Hopefully help someone else that way.
<aanderse>
GiGa: awesome! good luck. and don't be shy to ask for help if you get stuck again :)
<{^_^}>
undefined variable 'nixos-20' at (string):312:1
<bqv>
> nixos-20_03.openssl.version
<{^_^}>
"1.1.1g"
<bqv>
asbachb: ^
<AndersonTorres>
Hey! Do we have a nix interpreterb bot??
<bqv>
> "No, definitely not"
<{^_^}>
"No, definitely not"
<asbachb>
Mh. I wonder why `nix-shell -p sslscan` results into `Package ‘openssl-1.0.2u’ in /nix/store/vibyf5af9v3qxlfsj9j25khi0h3c38bd-nixos-20.03.1950.48723f48ab9/nixos/pkgs/development/libraries/openssl/default.nix:132 is marked as insecure, refusing to evaluate.`.
<nbp>
Has anybody done any domotique hub with Nix/NixOS. I see that we have an home-assistant service, but it lacks addons, and added components are not listed anywhere (?)
<etu>
nbp: home-assistant in nixos works, but it's a bit flaky when it comes to components.
orivej has quit [Read error: Connection reset by peer]
<nbp>
I tried to set it up in a nixos-container but it was completely useless, nix-shell is the best result I've got so far.
orivej has joined #nixos
<etu>
nbp: I'm not sure why, but the package doesn't care about your extraComponents unless you set an option for the service. It doesn't make any sense.
<etu>
nbp: And the entire packaging is horrible
asbachb has quit [Ping timeout: 245 seconds]
<bqv>
Damn, I was about to set it up
<nbp>
etu: unless you set an option for the service? What do you mean?
<etu>
nbp: Check my second link
<nbp>
etu: you mean the explicit listing of components?
<srhb>
jakobrs: Again I think it's safe to do this manipulation. Every store path you have right now is broken, so I doubt you're _worse_ off, but I don't know for sure. :-)
scoates has quit [Ping timeout: 260 seconds]
davidtwco has quit [Ping timeout: 260 seconds]
orivej has quit [Ping timeout: 258 seconds]
eddyb[legacy] has quit [Ping timeout: 260 seconds]
<srhb>
jakobrs: But you can test it out on a subset and check with nix-store --verify --check-contents
orivej has joined #nixos
midchildan has quit [Ping timeout: 260 seconds]
pittma has quit [Ping timeout: 260 seconds]
steveeJ has quit [Ping timeout: 260 seconds]
geekthattweaks has quit [Ping timeout: 272 seconds]
<jakobrs>
But would it theoretically be possible to have something like `with-path` above?
<gchristensen>
less nixos/default.nix -> { configuration ? import ./lib/from-env.nix "NIXOS_CONFIG" <nixos-config>, system ? builtins.currentSystem }:
<gchristensen>
only with scopedImport, and it is not advisable as it is very slow, and it is much better to be explicit about passing inputs
<talyz>
energizer: Well, I'm using bind mounts in the module since it's generally more transparent than symlinks and was easy to implement. It likely doesn't make any big difference in most cases, though.
<jakobrs>
I mean as a new primitive
<jakobrs>
Not that there's any reason for it, just theoretically.
<gchristensen>
it is already possible, just not a good idea :)
orivej has quit [Quit: No Ping reply in 180 seconds.]
<raghavsood>
I'm trying to package a program that requires a db set up command (`app --install`) to be run every time the db config is changed. What would be the most idiomatic way to do that in a nix module? Looking for something like `if cfg.db has changed`
<bqv>
Is it not idempotent?
dermetfan has joined #nixos
<bqv>
Theres an onChanged hook, somewhere
<raghavsood>
It is not idempotent - if you run it on an already set up db, it will drop the entire database and recreate a new, empty one
<infinisil>
bqv: That's a home-manager thing
<infinisil>
raghavsood: Can you give more info?
<bqv>
Ah
agsdheidjd has quit [Ping timeout: 260 seconds]
waleee-cl has joined #nixos
calbrecht has quit [Ping timeout: 272 seconds]
iyzsong has quit [Ping timeout: 260 seconds]
calbrecht has joined #nixos
<raghavsood>
infinisil: The app in question is https://github.com/knadh/listmonk - I wasn't looking towards home manager since it is something you'd usually put on a server.
<raghavsood>
to set it up, you give it credentials to postgres, then run `listmonk --install`, which sets up the schema
agsdheidjd has joined #nixos
<raghavsood>
Which also brings me to a follow up question - is it idiomatic to have the module for `listmonk` auto enable postgres? listmonk cannot run without postgres, so that feels necessary. The gitea module also enables postgres if you set the postgres options, so I feel like it is idiomatic
linarcx has joined #nixos
<infinisil>
Yeah that's idiomatic
<infinisil>
raghavsood: Now what does "the db config is changed" mean? How do you know it changed?
<jakobrs>
Is there a function to get the name of a package from its outPath?
<raghavsood>
infinisil: I'm here because I can't find a way to know if it is changed :D. By `changed`, I mean that if the postgres connection settings (namely host, database, user) have been updated (do not match their previous settings). Alternatively, I'm considering running a simple bash script in `ExecStartPre` that connects, checks to see if the db has the
<raghavsood>
listmonk tables already, and runs the install step if it does not
<raghavsood>
The whole `do not match their previous settings` is a bit iffy since I don't necessarily want to clobber the db if someone just wants to change the user and still use their old db
<raghavsood>
`ExecPreStart` scripts might be the only useful option
<infinisil>
Checking if it exists already sounds like the most reasonable solution
opthomasprime1 has quit [Remote host closed the connection]
<raghavsood>
Right then, seems like that's the route I'll be taking
<raghavsood>
infinisil: thanks!
opthomasprime1 has joined #nixos
<infinisil>
You might want to take a look at postgres' ensureUser and ensureDatabase options
<jakobrs>
It uses rot13 from the pkgs.bsdgames package
<raghavsood>
Looks like ensureUser and ensureDatabase will help make sure listmonk's user and database are good to go, but can't actually verify the schema
<dragonker>
hello, everybody! I have read everything i could find to enable virtualization with qemu/kvm but still says that it could not detect a default hypervisor
<dragonker>
weirder, the service starts active after boot but soon stops
<dragonker>
Ah, sorry. Tried to create newlines between messages
<jakobrs>
Is it considered better to do "${from}" or `$from` and `inherit from;`?
<bqv>
those are not equivalent
<jakobrs>
I know they're not equivalent; that's why I'm asking if one is preferred
<bqv>
to do what?
<commandocrypto[m>
I would be shocked if a db setup command wasnt idempotent (for a particular checkout of the project). How could it be different each time??
<jakobrs>
See the rot13.nix file from the gist I just created
<bqv>
although, at least if you use {} then it'll be typechecked
<bqv>
well, not typechecked, but, syntax checed
<adisbladis>
In this case I'd actually prefer `inherit`
<raghavsood>
commandocrypto[m: It's not so much so that it is different (the resulting schema will always be the same). It's just that it will clobber any data you have in your db, you always end up with empty tables: https://github.com/knadh/listmonk/blob/master/install.go#L26
<bqv>
that would too, yeah
<adisbladis>
Because it will cause less rebuilds for the builder script
<jakobrs>
yeah right because it's the same script, just with different values of `$from` each time
<jakobrs>
I mean, in the real world I should probably have written the builder script in a separate file, and then I'd have to use `inherit from;` anyway.
cr4y1 has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
jakobrs has left #nixos ["WeeChat 2.7.1"]
kaliumxyz has quit [Remote host closed the connection]
<LnL>
only scary part there is partitioning but never had any problems with it
<bqv>
excellent, ok
<jluttine[m]>
Are instructions about Python development mode in section "15.17.1.2.3. Develop local package" in https://nixos.org/nixpkgs/manual/ up to date? It seems to me that my package gets actually installed to store instead of symlinked to my working directoy.. I might be doing something else wrong too..
<fps>
in combination they work to install my little package..
EatThem has joined #nixos
evils has quit [Quit: Lost terminal]
<fps>
now i wonder how to use nix-env to get a development shell. i tried this: nix-shell -E 'with import <nixpkgs> {}; (python37Packages.callPackage ../ogfx-ui.nix {})'
<flokli>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 88
<fps>
but i don't get e.g. jalv or ogfx-tools build outputs in my path
<{^_^}>
[nixpkgs] @NickHu opened pull request #88783 → libayatana-appindicator: init at 0.5.4 libayatana-indicator: init at 0.6.3 ayatana-ido: init at 0.4.90 → https://git.io/JfapM
<{^_^}>
[nixpkgs] @NickHu opened pull request #88784 → redshift: enable appindicator support → https://git.io/Jfap9
<{^_^}>
[nixpkgs] @NickHu opened pull request #88785 → parcellite: enable appindicator support → https://git.io/Jfapb
chiastre has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @NickHu opened pull request #88786 → clipit: enable appindicator support → https://git.io/Jfaph
rembo10 has joined #nixos
rembo10 has quit [Client Quit]
rembo10 has joined #nixos
iwq has joined #nixos
evils has joined #nixos
<raghavsood>
Is there a normal pattern for ConditionPathExists in systemd units? Regarding my previous "Single run init command" for a module, I'm currently looking to set up a oneshot service. A lockfile would help ensure it isn't run on reboots. Where should such a lock file go?
<cole-h>
lovesegfault: Reason I pinged you yesterday was to talk about the nixpkgsPath nixops PR
<cole-h>
lovesegfault: I was wondering if you ever experimented with actually running it, because I'm not seeing how it would work (unless it's just framework for the Python wizards to come along and actually implement)
<{^_^}>
[nixpkgs] @matthewbauer pushed commit from @oxalica to master « systembus-notify: init at 1.0 (#88441) »: https://git.io/JfahO
agsdheidjd has joined #nixos
<cole-h>
(I started to write some more docs for it but then realized that `nixpkgsPath` isn't an argument to the default.nix, and eval-machine-info.nix isn't exposed anywhere)
<{^_^}>
[nixpkgs] @Mic92 closed pull request #60996 → Warn on deprecated /var/run in shell paths → https://git.io/JfajD
pamplemousse has quit [Ping timeout: 272 seconds]
numkem has quit [Ping timeout: 258 seconds]
<NickHu>
I'd like to get more involved with nixpkgs, and I've gotten a fair few number of prs merged over the years. I've just been reading https://github.com/NixOS/nixpkgs/issues/50105, and I was just wondering if anyone closer to the project could give me feedback as to whether or not I meet the criteria. If not, I'd also welcome suggestions on what I could be doing better to become more involved :)
<{^_^}>
#50105 (by Infinisil, 1 year ago, open): New nixpkgs committers requirements/process
nschoe has quit [Ping timeout: 256 seconds]
user_0x58 has joined #nixos
<cole-h>
NickHu: I don't think it's just PR count, but whether somebody else (dunno if it has to be a current committer or not) thinks you're doing an exemplary job helping out NixOS/nixpkgs.
<asymmetric>
is it ok to use yarn2nix for a packge in nixpkgs? or should i use the npm workflow?
ddellacosta has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
justanotheruser has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @romildo opened pull request #88791 → nixos/lumina: use xsession provided → https://git.io/JfVeL
linarcx has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
gustavderdrache has joined #nixos
linarcx has joined #nixos
<NickHu>
cole-h: Yeah, I know it's more than PR count. I guess I'm really asking for feedback on how I'm doing with respect to contributing to nixpkgs
<cole-h>
Well, I'm not an authoritative source on this (not a committer either), but I think you could do more with respect to *other people's* PRs. `is:pr reviewed-by:nickhu -author:nickhu` shows 3 total, whereas `reviewed-by:cole-h -author:cole-h` shows 79 total!
<cole-h>
(Granted, many of these are me restarting borg evals because of GitHub API weirdness, but I hope you get my point.)
<chrisaw>
Folks - I have a really strange issue - trying to use the Vivaldi browser. The package for it has 2 options: proprietaryCodecs and enableWidevine - both of which I've set via nixpkgs.config.vivaldi = { }; but they're being completely ignored (on 20.09) - any ideas?
drakonis1 has joined #nixos
tobeportable has quit [Ping timeout: 272 seconds]
<cole-h>
chrisaw: I don't think that's how you set those options -- you need to override the package itself to enable them.
<cole-h>
Without crafting an overlay, you could do: `environment.systemPackages = [ pkgs.asdf (pkgs.vivaldi.override { proprietaryCodecs = true; enableWidevine = true; }) pkgs.jkl ];`
<cole-h>
Note the parentheses: otherwise it'll look for a package `pkgs.vivaldi.override` and a package `{ proprietaryCodex = true; enableWidevine = true; }`
<cole-h>
(or something to that effect)
<chrisaw>
cole-h: +1000 beers for you! That's EXACTLY what I did wrong and exactly the problem I faced. :D
<cole-h>
:^)
<NickHu>
cole-h: That's a good point. Seeing as most of the PRs are about bumping versions, it seems like a review is a bit redundant. Other than that it's difficult to find PRs that I'm interested in
<cole-h>
NickHu: Even reviews for version bumps are invaluable. Otherwise, it falls to somebody else, and that person may not come by in a long while.
drakonis has joined #nixos
<NickHu>
But for a version bump, are you suggesting that we should test the application works extensively or that just checking that it builds is fine? Because in the case of the latter, a review wouldn't be doing anything OfBorg isn't
<cole-h>
Well, ofborg is only a bot :) Human reviews >>> bot reviews.
nschoe has joined #nixos
<LnL>
having somebody who actually uses a project test it is very useful to check for issues that a build wouldn't catch
commandocrypto[m is now known as comrandroxaos[m]
<cole-h>
The way I usually test PRs is `nix-shell -p nixpkgs-review --run 'nixpkgs-review pr #####'`, then test the produced binar(y/ies). If it's a graphical application, I'll run it and click around. If it's a CLI application, I'll `binary --help` and then run it and make sure it doesn't segfault or something
<cole-h>
The only times I drop a review and don't test /that/ in-depth are when they require some sort of infrastructure (e.g. if I wanted to test a Hashicorp Vault update -- I don't have a vault setup, so I can't really do much there)
orivej has quit [Quit: No Ping reply in 180 seconds.]
<LnL>
that has a nixos test :)
<cole-h>
Just an example :P
<LnL>
yeah I know, we can't feasibly do that for every package
<imalsogreg>
I'm having some trouble learning to use hydra. When I add a project that links to the github.com/shlevy/declarative-hydra-example, the evaluation of `.jobsets` fails with a database error. (`DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed: ERROR: null value in column "type" violates not-null constraint`). Anyone know how I can debug?
justanotheruser has joined #nixos
<{^_^}>
[nixpkgs] @veehaitch opened pull request #88793 → nodePackages.redoc-cli: init at 0.9.8 → https://git.io/JfVeS
<emily>
cole-h: you probably want to include commenter: not just reviewed-by:
<emily>
since the latter only shows up if you remember to click the specific github review ui
<cole-h>
Which brings the totals to 8 and 124, respectively.
<emily>
sure
<cole-h>
(I only used reviewed-by because that's what is used in the "give this person committer because they're really cool" thread)
<emily>
tbh I don't think comparing/worrying about specific numeric totals is really productive, people can contribute to nixpkgs in varied ways
<emily>
I do wish there were more concrete standards/procedures for commit access but I'm not sure ">N PR reviews" is really the right way to decide that
<cole-h>
I agree. I was just trying to demonstrate some areas in which they could feasibly do more.
<emily>
especially since, frankly, I've seen some pretty bad reviews sometimes too; I get the desire to established sustained participation but I think focusing on quantity is a mistake
<numkem>
clever: for posterity, since I deleted my local nix config, my hydra binary cache is being used by just having nix.binaryCaches and nix.binaryCachePublicKeys defined. The whole problem was because of the weird merge with the local config.
<NickHu>
emily: Yeah, I think you make valid points. In general I wouldn't think to look at other people's PRs because it seems kind of artificial for me to test something just because a PR exists, rather than actually using the software - for instance, I might miss something that is blindingly obvious to someone who knows what the program is supposed to do
<NickHu>
That being said thanks for your feedback cole-h, I've tried to do a few more reviews today to add value :)
fusion809 has quit [Remote host closed the connection]
<craftyguy>
or maybe that's what ccacheStdenv is supposed to provide?
<nbp>
etu: Ok, I finally managed to get tasmota devices appearing in home-assistant, many configuration issues in mosquitto to get messages going through the MQTT server.
orivej has quit [Ping timeout: 260 seconds]
<morgrimm>
Anyone find a workaround for nginx read permissions on acme certs?
marsh has joined #nixos
fendor has quit [Read error: Connection reset by peer]
fendor has joined #nixos
orivej has joined #nixos
<emily>
morgrimm: you should probably just chmod manually after setting allowKeysForGroup
<energizer>
my nixos has lately been failing to wake from suspend. screen turns on but black. how to debug?
<emily>
morgrimm: I recommend making an acme group, having certificates owned by acme:acme, turning on allowKeysForGroup, and adding nginx to the acme group, but the defaults of the enableACME are to just set the certificate user/group to nginx
<etu>
nbp: Hooray 🎉
<nk3>
i'm happy that i don't need to fiddle with certbot ever again after trying nixos :)
drakonis1 has quit [Ping timeout: 258 seconds]
<morgrimm>
emily: After reading the roundup of issues atm, I did that for now - I think I missed a step though. The cert group is acme, and has allowKeys set to acme as well - nginx has acme as an extra group
<morgrimm>
permissions look right on the cert files themselves
<morgrimm>
But the nginx service still doesn't have perms to load them
<morgrimm>
I don't have an acme user though - the certs are still root owned
user_0x58 has quit [Remote host closed the connection]
medvid has quit [Quit: WeeChat 2.3]
user_0x58 has joined #nixos
slack1256 has joined #nixos
medvid has joined #nixos
<symphorien>
I have a slow connection to some machine to which nix copy some derivation
<symphorien>
is it possible to have the receiving machine use the binary cache to fetch some of the dependencies ?
<emily>
morgrimm: are you sure the running service has picked up the group change?
<emily>
morgrimm: I strongly suggest against having certs owned by root because it means lego runs as root.
<morgrimm>
I hope so - I've rebuilt nixos, and rebooted the machine, so I hope it picked it up on restart
<pie_>
Profpatsch: ldlework and styx; <ldlework> The thing I like about using Nix for this, is that you can just whip up secondary tools and bring in anything you want to process your data however you'd like.
<morgrimm>
emily: Do you recommend chowning each cert dir based on the user the service using it is running?
<morgrimm>
i.e. my domain is run by nginx, so I should just manually chown the cert dir for nginx?
<emily>
morgrimm: I personally think the best setup is to have an acme user/group used for all cert users/groups and add the users of services to the acme group where relevant... if you want to split it up further then I guess you could do something like nginx-used certs are owned by (acme:nginx)
<morgrimm>
Makes sense, I don't think it's particularly useful to get more granular with it than that
<emily>
these options are kind of in flux, we've discussed making allowKeysForGroup mandatory and getting rid of at least the user setting
<emily>
since the default enableACME setup isn't really ideal (nginx gets write perms to certificates, harder to share certs across services, ...)
<morgrimm>
Yeah, it seems much more flexible to enforce direct setup through acme, and then letting users plug those in whereever they need
<emily>
yeah. the reason for potentially wanting multiple users/groups is to keep certain certificates isolated from each other, but nixos containers or fancy ACL setups are probably a better way of achieving that
<morgrimm>
emily: Seems the service is trying to write a new file in place of the cert that already exists - which would explain the failure, since I've only given it r perms
<morgrimm>
a vhost with useACMEHost shouldn't be expecting write perms, right?
<emily>
nginx shouldn't expect to write there, but lego certainly will
<srid>
how good is the AMD Ryzen 7 3700 PRO processor? (as for as compiling and running stuff on dedicated servers is concerned)
<srid>
Ashy: ^
evanjs has joined #nixos
<{^_^}>
[mobile-nixos] @samueldr merged pull request #149 → boot/gui: Use libffi rather than bespoke bindings → https://git.io/JfafB
<{^_^}>
[mobile-nixos] @samueldr pushed 12 commits to master: https://git.io/JfVTO
<morgrimm>
emily: Lego is the acme service though, right? That's working fine, it can set up the cert fine
<quinn>
so i'm trying to perform antiquotation on a string i'm importing, but the problem is the syntax for the string (asterisk dialplan) includes ${FOO} already. any thoughts?
<symphorien>
,escape'
<{^_^}>
Escape this in " and '' strings with: '
<symphorien>
,escape
<{^_^}>
Usage: ,escape <text> to show how to escape the given text in Nix
<symphorien>
,escape ${FOO}
<{^_^}>
Escape this in '' strings with: ''${FOO}
<{^_^}>
Escape this in " strings with: \${FOO}
<symphorien>
^ quinn
<{^_^}>
[nixpkgs] @msfjarvis opened pull request #88809 → rtl8821ce: bump to 69765eb288a8dfad3b055b906760b53e02ab1dea → https://git.io/JfVTR
<quinn>
,escape }
<{^_^}>
Escape this in " and '' strings with: }
<quinn>
symphorien: i gotta level with you dog, escaping it did not occur to me. thank you
<quinn>
symphorien++
<{^_^}>
symphorien's karma got increased to 28
orivej has quit [Read error: Connection reset by peer]
<symphorien>
(btw, the new escape function of the bot is really nice)
orivej has joined #nixos
<quinn>
that is veryy nice yes
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<emily>
morgrimm: right. not sure what's wrong, but I'd make sure the permissiosn on all the directories make sense as well as the files
benob has quit [Remote host closed the connection]
meh` has quit [Ping timeout: 246 seconds]
remirol has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
nixbitcoin has quit [Quit: nixbitcoin out]
lorimer has quit [Ping timeout: 258 seconds]
orivej has joined #nixos
nixbitcoin has joined #nixos
<{^_^}>
[nixpkgs] @flokli opened pull request #88810 → fetchgitlab: fix passing tags as rev → https://git.io/JfVT1
morgrimm has quit [Ping timeout: 265 seconds]
<calbrecht>
energizer: i have no clue. but the same problem. :) i blame it on a bios update and disabling the powermgmnt of it because of strange crackling in sound. but that makes me suspicious, are you on master and is it by any chanve a lenovo?
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej_ has joined #nixos
<calbrecht>
it's just that it doesn't bother me so much, that i prefered to wait until it solves by time. because this is another mind bending adventure, i guess.
<bigvalen>
Do people use mail relays on Nixos ? I was hoping to use gmail as an outbound relay with Postfix..but it seems the base config doesn't support sasl auth.
<energizer>
for me it's pretty annoying that i can't wake from suspend and have to reboot
seku has joined #nixos
<calbrecht>
i just do not suspend :)
<nk3>
i do suspend but i do not hibernate
<calbrecht>
i have lidEventCommands = "systemctl suspend" in my configuration.nix, so i guess i do neither.
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
andreas303 has quit [Remote host closed the connection]
<nk3>
calbrecht: i don't have that and my system does suspend when i close the lid, no idea why
chloekek has joined #nixos
andreas303 has joined #nixos
orivej_ has quit [Ping timeout: 272 seconds]
orivej has joined #nixos
morgrimm has joined #nixos
<calbrecht>
so you are lucky :) most hits on a quick search complain about not suspending ;)
<Profpatsch>
> In Greek mythology, Styx (/ˈstɪks/; Ancient Greek: Στύξ [stýks][citation needed]) is a deity and a river that forms the boundary between Earth and the Underworld.
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):312:19
<Profpatsch>
I see
<nk3>
o.O
smatting has quit [Ping timeout: 256 seconds]
<numkem>
I'm working on an option module and I'd like to see an example of the case where I'd want to add something to a set only if a variable is true
<superbaloo>
now I have to fix an issue in either crate2nix or nixpkgs :D but tests run :D
cr4y1 has joined #nixos
<cole-h>
And so the yak gets shaven
<superbaloo>
yup
<superbaloo>
yak hair all the way down
mac10688 has joined #nixos
dan224 has quit [Ping timeout: 260 seconds]
<superbaloo>
that project started as me trying things for a wifi setup at home, then "how nice would that be to run a software defined router" -> "now I need a test harness" -> laurn -> and now I'm fixing rust tests compilation ...
<cole-h>
Hehe
<cole-h>
What's broken about the test compilation?
<superbaloo>
it doesn't provide the CARGO_BIN_EXE_ variables like cargo does
<hexagoxel>
is there a construct that matches haskell's exception-handling `try` for derivations?
<hexagoxel>
for a derivation `d`, `f d` would be a derivation that produces either a file containing the output of trying to build `d` (if it fails) or a folder "success" that contains the output of `d` ?
<hexagoxel>
"output of trying to build `d`" as in: the stdout/stderr of `nix-build d`
<samueldr>
hmm, with the nixos modules system, I think there is no way to have an option A depend on B, and B depend on A, only when one is set, right?
<samueldr>
when both are unset, they will cause infinite recursion since it would try to set one according to the "unknown if set" other one, which depends on the other one "unknown if set" still
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #nixos
<{^_^}>
[nixpkgs] @IvarWithoutBones opened pull request #88821 → sm64pc: init at unstable-2020-05-19 → https://git.io/JfVtj
<ldlework>
If I runCommand and pipe the output of some command to $out
<ldlework>
how can I get that content?
<ldlework>
I tried `foo = runCommand ...; bar = readFile foo` but it doesn't work
knupfer has quit [Ping timeout: 260 seconds]
xd1le has quit [Read error: Connection reset by peer]
drakonis_ has joined #nixos
<{^_^}>
[nixpkgs] @mweinelt opened pull request #88822 → pim6sd: init at unstable-2019-05-31 → https://git.io/JfVqU
drakonis1 has quit [Read error: Connection reset by peer]
<ldlework>
oh
<Gaelan>
ldlework, I believe the feature you're describing is called import-from-derivation (or IFD for short). It's supported, but experimentally, and some projects (including nixpkgs) avoid it for now
drakonis1 has joined #nixos
<clever>
ldlework: the nix code you gave should work
<clever>
cole-h: just run ls -lh on the path, what does it say?
<bbigras>
What is the command in this channel to find which package a binary belongs to?
<cole-h>
It's a 0 size, nice.
<clever>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<clever>
cole-h: then your download failed successfully, and it gave you the hash of an empty file
<bbigras>
clever: thank you very much
<cole-h>
Then the question becomes: why does `curl`ing that link work?
teto has quit [Quit: WeeChat 2.8]
<asbachb>
Actually it seems that prefetch downloads something
<clever>
cole-h: no idea
<cole-h>
Well, I'm not invested in this :P Hope you figure out what's going on asbachb
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos
<asbachb>
clever: Can you paste the command you executed?
<{^_^}>
asbachb: To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected. See: tofu-vim
<maralorn>
Is it encouraged/allowed to make a PR to enter myself as package maintainer for Haskell packages for which I already feel like the maintainer?
alexherbo2 has quit [Ping timeout: 240 seconds]
<cole-h>
I mean, if you're planning to maintain then, I feel that puts you in a pretty good position to be a maintainer... ;^)
<asbachb>
clever: Yeah. I'm aware of that. I just wondered why I could not reproduce the hash.
<cole-h>
Something may have changed. Sounds like a good bug report to NixOS/nix
<maralorn>
cole-h: Yeah, that‘s clear. But I found the maintainer list in configuration-hackage2nix.yaml and it‘s very short. So I wonder if this should only be done for very important packages or something.
<cole-h>
Ah, we don't have individual packages for haskell stuff do we
<cole-h>
nh2: Knowing you're involved with our Haskell stuff, how should this be done? ^
<nh2>
maralorn cole-h: Peti explained his idea it in one of the last Haskell-maintenance-in-nix sessions that happen on Fridays (you are very welcome to join! It is quite fun), he puts things into the maintainer list under `peti:` that he specifically careas about it. I'm sure everybody would be happy if you put yourself in there and help out!
<maralorn>
But I wonder if entering myself there might e.g. lead to highlights when someone opens an issue against one of the packages.
<nh2>
I don't know if the Haskell packages have automatic clever github integration to ping you on changes
<maralorn>
nh2: Yeah, wanted to join for weeks, but every friday something else comes up.
<cole-h>
nh2: I'm there almost every week ;) Must not have been paying attention though
cr4y1 has quit [Ping timeout: 265 seconds]
<nh2>
maralorn: I also couldn't join the last 2 weeks due to business
kvda has joined #nixos
<maralorn>
nh: Okay, then I guess I’ll make a PR for the packages that are already in my personal monitoring.
<maralorn>
So if the haskell pr get‘s merged first, which is realistic, because Haskell maintenance works far above nixpkgs average , that might actual break.^^
<emily>
maralorn: a PR that references maintainer-list entries that don't exist won't get merged
<emily>
you should just roll that commit into the PR(s)
agsdheidjd has quit [Ping timeout: 246 seconds]
quinn has joined #nixos
<maralorn>
emily: Won‘t that create merge conflicts?
<emily>
maybe if multiple PRs are merged (though I think it'd just work as long as the commit has the same hash in both?)
<emily>
though I'm not actually wholly sure the hackage2nix package-maintainers list actually maps to a real maintainers list, so it might not matter
<emily>
but e.g. ofborg will refuse to test your PRs if you reference maintainers that don't exist