<{^_^}>
[nixops] @adisbladis closed pull request #1247 → WIP: Make it possible to use Nixops without automatic SSH key provisioning → https://git.io/JvKoc
<{^_^}>
[nixpkgs] @WhittlesJr opened pull request #82294 → Octoprint 1.4.0 with Python 3 (WIP; help needed) → https://git.io/JvKoX
<{^_^}>
[nixops] @adisbladis reopened pull request #1247 → WIP: Make it possible to use Nixops without automatic SSH key provisioning → https://git.io/JvKoc
<{^_^}>
[nixpkgs] @erictapen opened pull request #82295 → nixos/release-notes: mention that dhcpcd stopped giving IPv4 addresse to bridges → https://git.io/JvKoD
greghale has quit [Remote host closed the connection]
Ralith_ has joined #nixos
werner291 has quit [Quit: werner291]
jb55 has joined #nixos
lordcirth_ has quit [Remote host closed the connection]
<Ralith_>
anyone know how to get tc netem working? I'm getting: No distribution data for normal (/nix/store/djgbl8rvjs2h3jmq65182yahchws8s35-iproute2-5.2.0/lib/tc//normal.dist: No such file or directory)
Wizek has joined #nixos
switchy has quit [Remote host closed the connection]
fre has joined #nixos
vidbina has quit [Ping timeout: 240 seconds]
switchy has joined #nixos
<Ralith_>
ah, looks like a broken package
mcwitt has joined #nixos
mexisme_ has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @Ralith opened pull request #82298 → iproute: include netem tools → https://git.io/JvKKM
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
achw has joined #nixos
ottidmes has quit [Ping timeout: 255 seconds]
drakonis has quit [Ping timeout: 255 seconds]
ruip has joined #nixos
bhipple has quit [Read error: Connection reset by peer]
<abathur>
anyone have a good grasp of nix/locale/glibc and maybe python?
gustavderdrache has quit [Quit: Leaving.]
<abathur>
I'm working on a CI build for oil shell; nix-build works fine on macos and ubuntu, and nix-shell works fine on macos, but a locale issue is dogging me on nix-shell + ubuntu
HensonDerp has quit [Client Quit]
<abathur>
the shell script executing the test can still see LC_CTYPE=C.UTF-8
<abathur>
and LOCALE_ARCHIVE=/nix/store/nl67flma20ixa0x5jms4wk0yfbx4c9wb-glibc-locales-2.27/lib/locale/locale-archive
bhipple has joined #nixos
mcwitt has quit [Ping timeout: 255 seconds]
Synthetica has quit [Quit: Connection closed for inactivity]
justsomeguy has quit []
heatmiser has joined #nixos
<abathur>
(but the python script is throwing a SystemError: Invalid locale for LC_CTYPE)
wavirc22 has joined #nixos
<adisbladis>
abathur: Try adding `glibcLocales` as a build input in your shell.nix
<abathur>
does it matter if it's already composed into the env?
<cole-h>
When packaging something for Nixpkgs that doesn't have a version or tags or w/e, does the version become the date the package is created, or the date of the last commit?
<abathur>
i.e., my shellHook has ${if glibcLocales != null then "export LOCALE_ARCHIVE='${glibcLocales}/lib/locale/locale-archive' LC_CTYPE='C.UTF-8'" else ""}
<adisbladis>
cole-h: Last commit.
<cole-h>
adisbladis++ Thanks.
<{^_^}>
adisbladis's karma got increased to 35
<adisbladis>
abathur: Ah, that looks equivalent to just adding the input.
<abathur>
ok, I didn't think so, and had assumed, but didn't *know*
<abathur>
I think maybe a wrapper or something might be messing it up, but it's weird to troubleshoot multiple levels of indirection down
<abathur>
the main difference is that the out-of-tree normal build writes its own wrappers for the shell version can't
<abathur>
so I assume that's the cause, though I don't really see why it should be
<abathur>
it doesn't unset it, or use env -i
<abathur>
so the other possibility is that it's the python weirding
iqubic has quit [Read error: Connection reset by peer]
ghasshee has quit [Ping timeout: 268 seconds]
blaggacao has quit [Read error: Connection reset by peer]
blaggacao has joined #nixos
hmpffff_ has joined #nixos
justanotheruser has quit [Read error: Connection reset by peer]
hmpffff has quit [Ping timeout: 256 seconds]
<cole-h>
clever: The "hang while `home-manager build`ing" has resurfaced :( Just gonna let it run while I eat dinner...
blaggacao has quit [Ping timeout: 268 seconds]
blaggacao has joined #nixos
<pie_[bnc]>
how can i import a drv into the repl so that i can access its attributes?
<pie_[bnc]>
i want to modify a buildCommand
<clever>
pie_[bnc]: run import on the path to the drv
ddellacosta has quit [Ping timeout: 240 seconds]
<clever>
pie_[bnc]: note, that all functions like .overrideAttrs are lost
lovesegfault has quit [Quit: WeeChat 2.7.1]
doyougnu has quit [Ping timeout: 272 seconds]
achw has quit [Quit: WeeChat 1.9.1]
<pie_[bnc]>
clever: i tried that but it doesnt really have much in terms of attributes at all
* pie_[bnc]
checks if hes importing the right drv
<clever>
pie_[bnc]: why cant you import the original nix expr?
<pie_[bnc]>
yeah it only has xx.all xx.drvPath xx.name xx.out xx.outPath xx.outputName xx.type
<pie_[bnc]>
because i dont know my was around haskell.nix and it doesnt look like it lets me override what i need
<pie_[bnc]>
im trying to add set -x for debugging
<clever>
pie_[bnc]: ah, thats the same kind of digging ive been doing, to fix all of the IFD stuff
<clever>
pie_[bnc]: you can run nix-shell on a .drv file, to get that env loaded
<clever>
pie_[bnc]: then manually run each phase the .drv would have ran
<pie_[bnc]>
clever: im trying to do something like that too
<pie_[bnc]>
hmm actually i didnt try to full on do that
<pie_[bnc]>
i tried to do some stuf fin the project dir with the stuff provided by that shell
<pie_[bnc]>
ok ill try that
<pie_[bnc]>
the realization that i can just shell a drv could have hit me a long long time ago...(i think i could have used it?)
<pie_[bnc]>
clever: but yeah if you can tell me a way to do tihs where i dont have to run stuff in the shell...
<pie_[bnc]>
i think if i ever build anything complicated i might add a module system that lets you orthogonally specify phase overrides
<clever>
pie_[bnc]: about all you can really do is read `nix show-derivation`, and then manually call `builtins.derivation` with the same values, using ${/nix/store/foo` where possible to maintain context
<pie_[bnc]>
ok that sucks
<pie_[bnc]>
clever: can we get something built in for this
<clever>
haskell.nix does have phase overrides
gentauro has quit [Read error: Connection reset by peer]
<clever>
let me grab an example
<pie_[bnc]>
clever: so can i add set -x
gentauro has joined #nixos
<clever>
maybe
<pie_[bnc]>
clever: what is it with me having a hard time with nix :P or is everyone else just a friggin pro
<pie_[bnc]>
haskell.nix is not exactly transparent, but people seem to manage
<clever>
pie_[bnc]: haskell.nix will find the file on its own, in src
<pie_[bnc]>
clever: i think im still going to put it in the ./cabal subdir, but right
<pie_[bnc]>
clever: so any idea how to eval a driv not in the store?
<clever>
pie_[bnc]: isnt really possible
<clever>
pie_[bnc]: all drv files are named after a hash of their contents, there might be a way to add the drv back into the store, but i dont know it off-hand
mexisme_ has quit [Ping timeout: 240 seconds]
<pie_[bnc]>
grr >:)
<pie_[bnc]>
theres gotta be a way
<pie_[bnc]>
so yeah, its pretty annoying the builds dont even fail if you have a malformed cabal.project (i just made another mistake) but im not sure if anyhting can be done about it
<pie_[bnc]>
unless its viable to fail on unknown field errors
<clever>
pie_[bnc]: after you strip the hash off the drv file
<pie_[bnc]>
the reason i think its gotta be possible is i assume eelco debugged his drvs somehow :P
<pie_[bnc]>
(well, maybe not like that)
iqubic has joined #nixos
<pie_[bnc]>
clever: also haskell.nix doesnt seem to play well with relative paths
<clever>
pie_[bnc]: are they going up or down?
<pie_[bnc]>
can cabal.project be in a non-root location?
<pie_[bnc]>
theyre going up
kvda has joined #nixos
<clever>
pie_[bnc]: then you need `src = "${../.}/non-root";`
wiml has quit [Quit: wiml]
<clever>
pie_[bnc]: to force nix to copy the root dir, and then reference the non-root path, maybe
<clever>
but your better off moving the cabal.project to the root, so all paths go down
gila has joined #nixos
<clever>
pie_[bnc]: did the --add-fixed trick work?
<pie_[bnc]>
im about to try
<pie_[bnc]>
variant 0 works: https://bpaste.net/DYTQ but not very nice, about to try what you said
<pie_[bnc]>
tbh i should have prbably asked you about this earlier but i didnt want to be a bother v.v
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
<pie_[bnc]>
error: the string 'cabal-plan-to-nix-pkgs' is not allowed to refer to a store path (such as '/nix/store/ldh5zmc7mnrcsy9kbfcikxzwx92bc16h-bb-clean'), at /nix/store/sm740vh74gh7r8ww6q3wf067xkq1nb64-nix-2.3.1/share/nix/corepkgs/derivation.nix:8:12
<clever>
pie_[bnc]: to me, that says that qtah isnt in the cabal.project file, and/or the hackage snapshot
<cole-h>
Is it possible to disable `checkPhase` for every package via an env var or something?
<pie_[bnc]>
clever: qtbase rather
<cole-h>
*for every package being built
<clever>
cole-h: only using an overlay
<cole-h>
:(
<clever>
cole-h: and that change will result in you needing to re-compile glibc and gcc from scratch
Lears has quit [Ping timeout: 265 seconds]
<kolby_>
if I've been the only contributor to a package for a while, is it ok for me to add myself as a maintainer or is there other requirements?
<clever>
kolby_: that sounds fine
<kolby_>
clever: Thanks I just wanted to make sure.
<cole-h>
I don't think there are any guidelines for being a maintainer than 1) actually maintaining it, and 2) not removing whoever else is maintaining it
<pie_[bnc]>
clever: the goal seems to be to get something like this to happen https://bpaste.net/ZD4A
doyougnu has quit [Quit: WeeChat 2.7.1]
<pie_[bnc]>
wait...
<kolby_>
another question real quick. How do I respect nix.buildCores when I have to declare buildPhase manually? I tried setting "make -jN" to "$NIX_BUILD_CORES" but sometimes it spins up way to many compiler processes.
<pie_[bnc]>
AHA! maybe i can cheat pkgs.buildPackages.qtah-generato
<clever>
kolby_: each nix job, will spin up its own $NIX_BUILD_CORES jobs, so if you `nix-build -j 5`, it will be up to 5 seperate derivations, each doing make -j $NIX_BUILD_CORES
<pie_[bnc]>
wait no that doesnt help because the prolem is the build tools list doesnt get generated to begin with
<pie_[bnc]>
i am very smart -_-
<pie_[bnc]>
so to recap, its generated, but cabal chokes
<pie_[bnc]>
for whatever reasob
<kolby_>
clever: I'm declaring "let jobs = "$NIX_BUILD_CORES";" and then declaring in buildPhase "make -j$jobs" in pkgs/misc/emulators/retroarch/cores.nix. When doing a nixos-rebuild with nix.maxJobs set to 1, it still spins up tons of compiler jobs ( fills the screen on top ).
kolbycrouch has quit [Remote host closed the connection]
<clever>
kolby_: if that variable is empty, you wind up doing `make -j` which tells make to spin up infinite jobs
<kolby_>
clever: I get a maximum of 6 compiler processes on 1 derivation for any other derivation. Which is expected ith my settings.
kolbycrouch has joined #nixos
<clever>
kolby_: what is wrong with the default buildPhase ?
<clever>
kolby_: you either want to set vars to make buildPhase() do what you want, or copy line 1020 to your custom buildPhase
<kolby_>
clever: Most of the libretro packages only contain a Makefile, or require you to run make against a makefile instead of configure.
<clever>
kolby_: nix will do that by default
<clever>
kolby_: and you can use `makeFlags = [ "foo" ];` to add extra flags to make
<kolby_>
clever: I forgot to mention, most of the makefiles are named "Makefile.libretro", and if there is a GNU configure script, it can't be used to build a libretro core.
<kolby_>
clever: I'm just wondering how I get the value of nix.buildCores that a user sets.
<clever>
kolby_: 1020 does bash magic to compute the right value, make sure its a bash var and not a nix var
<clever>
kolby_: try `dontConfigure = true; makefile = "Makefile.libretro";` first, and see what happens
<kolby_>
clever: I've just been doing "patchPhase = "rm configure" and buildPhase = "make" which at least works.
<clever>
kolby_: `dontConfigure = true;` is enough to make it ignore the configure script
<kolby_>
clever: Ahh let me try that. I didn't know about dontConfigure. I didn't know that setting makefile might work either.
<clever>
kolby_: line 1027 of the file i linked, if $makefile is set, it will do `make -f $makefile`
<kolby_>
clever: Yea I knew that, I had that set for a few of the cores, I didn't know about makeFlags either though. I think that's why I was declaring buildPhase manually. Because many of the cores need extra variables set.
<kolby_>
clever: it will take a little bit to test it because I'm using an aarch64 sbc, but looking through this I'm quite sure it will work. Thanks!
<clever>
pie_[bnc]: build-tools is exactly what my example was setting
<pie_[bnc]>
theres so many of your repos i should look at and no time ;_;
<pie_[bnc]>
i should also look at what you link properly :I
kleisli has joined #nixos
<danderson>
so, I installed nixos on my laptop. I'm still getting it properly set up, but `nix-env -qa` is horribly slow to do anything, and it's getting in the way. Is there some kind of caching I'm missing that's off by default or something?
<clever>
danderson: `nix search` is the new replacement, that does include caching
<danderson>
aah, much much better. Thank you!
wavirc22 has quit [Ping timeout: 258 seconds]
<danderson>
okay, now to find the right vdpau configuration for radeon...
<clever>
danderson: that will have to be done in configuration.nix
<danderson>
yup, I'm working on it. Exploring available packages
wavirc22 has joined #nixos
<{^_^}>
[nixpkgs] @sjmackenzie opened pull request #82303 → seabreeze: udev rules not installed → https://git.io/JvK1Z
<clever>
danderson: and its typically done by proper options, adding it to systemPackages wont work fully
orivej has joined #nixos
<danderson>
yup, I'm adding them to hardware.opengl.extraPackages
<danderson>
drat, it doesn't have much useful stuff, just things I knew already about needing a recent kernel. Ah well, the search is half the fun :)
cole-h has quit [Ping timeout: 265 seconds]
hmpffff_ has quit [Quit: Bye…]
<danderson>
hmm, is there a way to evaluate a nixos config and print a merged config, but only the bits that my snippets specify?
<danderson>
e.g. I added a nixos-hardware build, I'd love to see in a single diff all the changes. But my choices are either walk through the nixos-hardware settings by hand, or apply and pray.
<clever>
danderson: not easily
kahiru has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @srhb merged pull request #81941 → tests/kubernetes: remove unreferenced variable and import from inexisting file → https://git.io/Jvwyz
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
kahiru has quit [Ping timeout: 255 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
bvdw has quit [Read error: Connection reset by peer]
m0rphism has joined #nixos
kahiru has joined #nixos
bvdw has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-hardware] @danderson opened pull request #149 → lenovo/thinkpad/t495: init from e495. → https://git.io/JvKMI
<jluttine>
are emacs fonts broken for anyone else? some characters (at least # and ') are replaced by a rectangle that represents a missing character in the font. this happens only when running emacs as it's own gui window, everything works when running emacs with -nw inside a terminal window. i tried two different fonts and different font sizes, no help
knupfer has quit [Ping timeout: 240 seconds]
<jluttine>
i tried "JetBrains Mono" and "DejaVu Sans Mono". both have worked before. this happened after a recent upgrade on nixos-unstable i suppose..
alexherbo2 has joined #nixos
sgrunert has joined #nixos
jb55 has quit [Ping timeout: 240 seconds]
kleisli has quit [Remote host closed the connection]
Wizek has quit [Quit: Connection closed for inactivity]
kleisli has joined #nixos
jb55 has joined #nixos
zupo has joined #nixos
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
zdm has joined #nixos
alexherbo28 has joined #nixos
<jluttine>
should package derivation meta homepage value be with or without double quotes? the manual and nixpkgs itself has plenty of examples without double quotes. has something changed? i couldn't find any information on this
alexherbo2 has quit [Ping timeout: 255 seconds]
alexherbo28 is now known as alexherbo2
<clever>
jluttine: id prefer to have it quoted
<jluttine>
clever: why? how does it affect whether it's quoted or not?
<clever>
jluttine: only real effect is if you start having weird stuff like maybe a ? or spaces in the url, otherwise they are the same
<jluttine>
clever: ok, thanks!
c0c0 has quit [Ping timeout: 255 seconds]
kahiru has quit [Ping timeout: 260 seconds]
<DigitalKiwi>
jluttine: i just solved that problem the other day
<DigitalKiwi>
jluttine: let me try and remember what i did
<jluttine>
DigitalKiwi: ah, i removed symbola because it had became proprietary..
<fooker>
Is there a builder which allows me to write multiple files to a single derivation? Something like writeFile, but accepting a attrset or something?
<DigitalKiwi>
and dejavu_fonts
<clever>
fooker: either buildEnv a few writeFile's together, or just use runCommand and maybe passAsFile
maddo has joined #nixos
<fooker>
clever: ok, thanks
<clever>
fooker: if you do `foo = "bar"; passAsFile = [ "foo" ];` then youll get a $fooPath instead of a $foo at build time, which saves you all the escaping nightmares
<DigitalKiwi>
jluttine: if you disable prettify-symbols-mode or w/e it should be fine too if you don't care about that
<clever>
fooker: and this would create one file (thats basically just writeFile)
<lovesegfault>
eh, is toYAML not in lib?
<lovesegfault>
> lib.toYAML { foo = "bar" }
<{^_^}>
error: syntax error, unexpected '}', expecting ';', at (string):288:26
<lovesegfault>
> lib.toYAML { foo = "bar"; }
<clever>
lovesegfault: all yaml parsers accept json as well
<{^_^}>
attribute 'toYAML' missing, at (string):288:1
<DigitalKiwi>
jluttine: it's also possible to work around it by assigning your own but uh...that's beyond the level of effort i wanted to go to ;_;
LarryTheCow has left #nixos ["ERC (IRC client for Emacs 26.3)"]
<clever>
lovesegfault: just use builtins.toJSON
* lovesegfault
nods
<fooker>
clever: I see
zdm has quit [Quit: WeeChat 2.7.1]
<DigitalKiwi>
jluttine: and about the homepage urls...quoted at least for me helps make sure they actually open correctly from a terminal (vim) and also there's an initiative to remove that special syntax for allowing unquoted. so i agree with clever to quote them.
<DigitalKiwi>
unquoted urls try and open with the semi-colon as part of it
<clever>
DigitalKiwi: id blame that on the parser in vim not parsing the same way as nix
<lovesegfault>
> lib.toYAML { foo = "bar"; }
<{^_^}>
attribute 'toYAML' missing, at (string):288:1
<lovesegfault>
> builtins.toJSON { foo = "bar"; }
<{^_^}>
"{\"foo\":\"bar\"}"
<DigitalKiwi>
idk that it's vim i think it's the terminal?
* lovesegfault
squints
<clever>
DigitalKiwi: ah yeah, the terminal is also a major factor there
<DigitalKiwi>
i don't use any url opener from vim just whatever the terminal does
<clever>
DigitalKiwi: i use plain xterm, it doesnt auto-detect urls, i just select and paste
xelxebar has quit [Remote host closed the connection]
xelxebar has joined #nixos
tomturbo has joined #nixos
<tomturbo>
hello, I'm trying to manage my xmonad configuration in nixos via "services.xserver.windowManager.xmonad.config". I'm doing this on another machine, so I know that it should work. On this machine however, the key "config" does not exist. I've also confirmed this via the repl. According to the options online, the key should exist however. My stateVersion is 19.09 (same on the machine it works on) and I've
<tomturbo>
just run "nixos-rebuild switch --upgrade". Does anybody know what I have to to get the missing key?
kahiru has quit [Ping timeout: 256 seconds]
kahiru has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<lovesegfault>
tomturbo: reboot and see what happens (after the switch-upgrade)
<tomturbo>
@lovesegfault thanks, I will try it
tomturbo has quit [Quit: leaving]
tomturbo has joined #nixos
<tomturbo>
sadly, the reboot didn't help. the only options I see are "enable", "enableContribAndExtras", "extraPackages" and "haskellPackages"
<clever>
tomturbo: which channel are you on?
<tomturbo>
hm, how can I check this? I tried "nix-channel --list" but it didn't output anything. that being said, I don't recall switching channels since installing
<lovesegfault>
tomturbo: sudo
<clever>
tomturbo: --list as root
jb55 has quit [Ping timeout: 240 seconds]
cartwright has quit [Ping timeout: 240 seconds]
klntsky has quit [Ping timeout: 240 seconds]
chloekek has joined #nixos
<lovesegfault>
channels are the root of all evil
KeiraT has quit [Ping timeout: 240 seconds]
oida has quit [Ping timeout: 240 seconds]
fuwa has quit [Ping timeout: 240 seconds]
andreas303 has quit [Ping timeout: 240 seconds]
sdier has quit [Ping timeout: 260 seconds]
__monty__ has joined #nixos
xelxebar has quit [Ping timeout: 240 seconds]
<tomturbo>
ahh, thanks a lot guys. I'm on nixos-19.03, which might explain why the option is not there. should I try running "nix-channel --update"?
phreedom has quit [Ping timeout: 240 seconds]
EliotB has left #nixos [#nixos]
<clever>
tomturbo: `nixos-rebuild --upgrade` runs `nix-channel --update` for you, so that wouldnt help
goodwill has joined #nixos
<clever>
tomturbo: you must run `--add` and overwrite the nixos channel with a newer nixos
<lovesegfault>
You need to bump the channel itself
<{^_^}>
[nixpkgs] @flokli opened pull request #82310 → nixos/systemd: apply .link even when networkd is disabled → https://git.io/JvKyj
bahamas has joined #nixos
bahamas has joined #nixos
bahamas has quit [Changing host]
lovesegfault has joined #nixos
<lovesegfault>
Eh, how do I check if an attrset is empty?
<lovesegfault>
cc. clever
<bahamas>
when using nix as a script is there a way I can pass a flag to change the value of the interpreter (-i)? in my case, I want to switch between runghc and ghcid
<{^_^}>
nix#1361 (by alexeymuranov, 2 years ago, closed): Language feature proposal: exclusive 'with'
kolbycrouch has quit [Remote host closed the connection]
kolbycrouch has joined #nixos
veske has joined #nixos
veske2 has joined #nixos
o1lo01ol1o has joined #nixos
lovesegfault has quit [Quit: WeeChat 2.7.1]
lovesegfault has joined #nixos
<__monty__>
Ok, so the precedence isn't unreasonable.
o1lo01ol1o has quit [Remote host closed the connection]
<__monty__>
It's still very unfortunate though.
o1lo01ol1o has joined #nixos
<__monty__>
The nested withs are confusing imo. "With introduces names deferring to context rather than shadowing *unless* the context comes from another with."
<clever>
i try to avoid with when possible
<__monty__>
I can see what led to this choice but it seems inconsistent.
hyper_ch has quit [Read error: Connection reset by peer]
hyper_ch has joined #nixos
hyper_ch has quit [Read error: Connection reset by peer]
hyper_ch has joined #nixos
oida has quit [Ping timeout: 240 seconds]
kahiru has quit [Ping timeout: 240 seconds]
oida has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
kahiru has joined #nixos
kahiru has quit [Ping timeout: 258 seconds]
kahiru has joined #nixos
veske2 has quit [Quit: This computer has gone to sleep]
veske has quit [Quit: This computer has gone to sleep]
philr has quit [Ping timeout: 256 seconds]
werner291 has joined #nixos
v88m has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
veske has joined #nixos
veske2 has joined #nixos
kahiru has quit [Ping timeout: 265 seconds]
kahiru has joined #nixos
hyper_ch has quit [Read error: Connection reset by peer]
hyper_ch has joined #nixos
veske has quit [Quit: This computer has gone to sleep]
veske2 has quit [Quit: This computer has gone to sleep]
<infinisil>
Cadey: A simple way is to symlink /etc/nixos -> /path/to/your/config/<hostname>
<vxi___id>
Hello everyone. Is there a simple way to globally symlink 'python3' to python from a configuration.nix file? I don't want to use a nix environment unless there is an easy way to run an org-mode babel block in a nix env. Thanks!
<ottidmes>
Weird, my builds fail with a lib of 19.09 but succeed with lib of 19.03 or 20.03, it is difficult to debug, it seems to no longer import my /modules dir causing all kinds of option not defined errors. Happy it at least works again under 20.03, so I can just see it as another reason to update
<infinisil>
vxi___id: Not sure what you mean by this, you want python3 to actually be python2?
<vxi___id>
infinisil: No I want calls to "python" to default to "python3" instead of "python2"
<Taneb>
vxi___id: python the program or python the attribute of nixpkgs
<infinisil>
vxi___id: If you have both python2 and python3 in your systemPackages, which one does the python binary point to?
<vxi___id>
Taneb The program.
<ottidmes>
vxi___id: if you have both in your systemPackages, just use (lib.hiPrio python3) so that conflicts between the two version will prefer the python3 ones
bahamas has quit [Ping timeout: 240 seconds]
<infinisil>
^^
<vxi___id>
which python points to .nix-profile/bin/python -> /nix/store/v3b3x86cv9wfzwn2fcfwdfm11vq5s97p-python-2.7.17/bin/python
Neo-- has joined #nixos
<vxi___id>
ottidmes Interesting! How would I use this in my configuration?
<infinisil>
vxi___id: Oh, you have python installed in your user env with nix-env. You also need to uninstall that first before any change from configuration.nix applied
<infinisil>
applies*
<ottidmes>
vxi___id: { lib, pkgs, ... }: with lib; { environment.systemPackages = with pkgs; [ python2 (hiPrio python3) ]; } and keep in mind what infinisil said or else your nix-env would overshadow your system paths
<vxi___id>
infinisil: That was it, python was installed in my nix profile for some reason. That makes much more sense. Thanks.
<vxi___id>
Python now defaults to 3. Thanks everyone. I'll remember the hiPrio trick, it could be useful.
<infinisil>
Nice :)
<infinisil>
vxi___id: Although, you might want to consider using shell.nix files instead of installing python globally. That way each project can have its own python version
<infinisil>
Generally installing dependencies is not very good practice
<immae>
(If I understand the initial message maybe "fixing" org-mode would be even better than a shell.nix, no?)
<ottidmes>
Do others also get "[nginx_parser] WARNING File not found" when rebuilding NixOS? All those paths are symlinks and point to files that do exist
<infinisil>
immae: Oh yeah if possible that might be even better
<infinisil>
Although the python version you want to use in the blocks should maybe be project-specific anyways
veske2 has quit [Quit: This computer has gone to sleep]
veske has quit [Quit: This computer has gone to sleep]
<infinisil>
But it could fallback to a standard python at least
<vxi___id>
infinisil: There is a "org-babel-python-command" option in Emacs
<vxi___id>
Which didn't work in my case for some reason
<immae>
ah ok (I don’t know org-mode at all), then indeed it may not be the solution
<vxi___id>
Well it is very powerful but also very fragile. It can get quite frustrating when parts of my setup break for no apparent reason.
<adisbladis>
That may help you to understand the difference
orivej has quit [Ping timeout: 258 seconds]
bahamas has quit [Ping timeout: 256 seconds]
veske has joined #nixos
veske2 has joined #nixos
kolbycrouch has quit [Remote host closed the connection]
kolbycrouch has joined #nixos
<exarkun>
maybe I'll make it my desktop background
<exarkun>
I think it suggests that overrideAttrs is the right way to supply a new nativeBuildInputs value, though, so it doesn't explain why my build fails. :(
<exarkun>
Hm, except reading the crate2nix generated code, the first thing I see it call is lib.makeOverridable
delli3 has joined #nixos
delli3 has quit [Max SendQ exceeded]
orivej has joined #nixos
<exarkun>
Oh, that just means I can override the arguments to that function with override, but nativeBuildInputs is not one of its arguments
rogue_koder_ has joined #nixos
<genesis>
this lack the "super" when override
rogue_koder has quit [Remote host closed the connection]
fusion809 has quit [Remote host closed the connection]
<exarkun>
I see that buildRustCrate respects ... `EXTRA_LIB_SEARCH` environment variable ... but it seems unlikely that I'm just supposed to set random extra env vars to fix crate link errors?
LysergicDreams has joined #nixos
<exarkun>
maybe this is what `defaultCrateOverrides` is for ...
iyzsong has quit [Ping timeout: 272 seconds]
mexisme_ has joined #nixos
<{^_^}>
[nixpkgs] @marsam pushed commit from @colemickens to release-20.03 « cfdyndns: v0.0.1 -> v0.0.3 »: https://git.io/JvKda
bvdw has quit [Read error: Connection reset by peer]
kim0 has joined #nixos
bvdw has joined #nixos
revtintin has joined #nixos
mexisme_ has quit [Ping timeout: 256 seconds]
blaggacao has quit [Read error: Connection reset by peer]
blaggacao has joined #nixos
<exarkun>
no obviously :/
<exarkun>
Too bad because I see an example in nixpkgs defaultCrateOverrides that does exactly what I'm trying to do ... provides the Security framework to a crate... but if I do exactly the same thing I still get build failures where the linker can't find Security.
magnetophon has joined #nixos
Rusty1 has joined #nixos
gustavderdrache has joined #nixos
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<magnetophon>
I'm packaging a plugin that uses zenity. I can wrap a binary to include it in it's path, but I don't think that will work with a plugin. Should I wrap the host?
cr4y1 has joined #nixos
zdm has quit [Quit: WeeChat 2.7.1]
<magnetophon>
It works if I install zenity
m0rphism has quit [Ping timeout: 240 seconds]
<exarkun>
aha. defaultCrateOverrides *does* help. I just have to pay attention to *which* crate is failing to build ... and add Security to *all* of the crates that require it, not just one of them. But the excitment just doesn't end ...
butterthebuddha has joined #nixos
<exarkun>
"error: failed to build archive: bad archive: truncated or malformed archive (offset to next archive member past the end of the archive after member sha2-12389d9af5.sha2.6rbaxvyb-cgu.1.rcgu.bc.z)" ...
<Orbstheorem>
I have a package with a setup.d directory. I would like to make a nixos-module that installs a derivation based on such package but adds a script to that directory. I have patched the program to read the setup.d directory location from an environment variable. Do you have any other ideas or an I going in the right direction ?
knupfer has quit [Read error: Connection reset by peer]
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
matthiaskrgr has joined #nixos
<exarkun>
I tried referencing $out in postInstall passed to overrideAttrs and ... it points to the wrong path?
Jackneill has joined #nixos
<exarkun>
The build fails because the output path "/nix/store/dd1692scymvj54wq1dbwq7nkrhm83sq1-rust_challenge-bypass-ristretto-ffi-1.0.0-pre.1" isn't created
nikivi has joined #nixos
<exarkun>
hmm nope that's not it
genesis has quit [Remote host closed the connection]
Jackneill has quit [Read error: Connection reset by peer]
butterthebuddha has joined #nixos
<exarkun>
more rust weirdness :/
vidbina has joined #nixos
vidbina has quit [Read error: Connection reset by peer]
<exarkun>
something is moving everything from $out to $out-lib after postInstall
<pinecamp>
ottidmes: thanks! I was curious about Nix-related channels. #nix-darwin especially is a great find :)
<exarkun>
but nothing that's part of the derivation, as far as I can tell
veske2 has quit [Quit: This computer has gone to sleep]
veske has quit [Quit: This computer has gone to sleep]
Jackneill has joined #nixos
mexisme_ has joined #nixos
<exarkun>
is it normal to have $lib set to a path not beneath $out?
zdm has joined #nixos
drewr has quit [Ping timeout: 260 seconds]
mexisme_ has quit [Ping timeout: 256 seconds]
veske has joined #nixos
veske2 has joined #nixos
<eyJhb>
So for some reason, the .flexget-wrapper does not include transmissionrpc, but I cannot figure out what generates this wrapper
rail has quit [Quit: gone...]
<ottidmes>
I am confused about permissions, in nginx I reference SSL certs generated by dehydrated (Let's Encrypt), those certs are owned by dehydrated:dehydrated with 600 as permissions, my nginx runs under www-data:www-data. On one server with the same config, I get no errors and it seems to read just fine. On my latest server I get a permission denied error on the cert. To be honest, I find the former to be
<ottidmes>
more surprising to work, since it should fail too
<ottidmes>
eyJhb: wrapProgram I believe
<eyJhb>
Yeah, but I cannot see it being run for flexget. Might be some default python thing
<eyJhb>
But it annoys me, because all of the sudden it doesn't work anymore...
Tucky has quit [Quit: WeeChat 2.7.1]
<eyJhb>
Seems like someone removed transmissionrpc at some point..
<exarkun>
Okay, that seems to be what's going on. buildRustCrate, or something near that, decides that the crate should have "out" and "lib" and then something, I don't know what, moves everything from $out to $lib and deletes $out :(
<Orbstheorem>
Does anybody have a nix expression that lists the files in a derivation directory? (e.g. "${pkgs.hello}/bin".)
<exarkun>
unfortunately $lib used to mean something different so .. ugh, hard to make something compatible across different nixpkgs revisions
<eyJhb>
If anybody is up for my Flexget request, then please do merge it. It is a very simple change :)
<ottidmes>
Orbstheorem: you could tweak this maybe: listFiles = dir: foldrWithName (file: type: files: let path = dir + "/${file}"; in if type == "directory" then (listFiles path) ++ files else [ path ] ++ files ) (readDir dir);
<philipp[m]>
ottidmes: Do you use the same nginx version on both hosts?
waleee-cl has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ingenieroariel has joined #nixos
<ottidmes>
philipp[m]: actually I am, 1.16.1 on both
<Valodim[m]>
exarkun: interesting. any place I can read up on that?
malSet has quit [Read error: Connection reset by peer]
<philipp[m]>
ottidmes: you and me both... maybe I'll just switch to the built in method in nixos...
cosimone has quit [Remote host closed the connection]
<Valodim[m]>
¯\_(ツ)_/¯
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
cosimone has joined #nixos
erasmas has joined #nixos
veske2 has quit [Quit: This computer has gone to sleep]
veske has quit [Quit: This computer has gone to sleep]
zupo has joined #nixos
<ottidmes>
philipp[m]: I used that initially but got fed up with it, so now I have a nice module for dehydrated and it has only failed on me once, which was due to some dependency updating that needed a small fix. I can confirm that fixing the user fixed the issue for me
fendor has joined #nixos
<Orbstheorem>
ottidmes: thanks
kenshinCH has quit [Quit: WeeChat 2.7.1]
m1sosoup has joined #nixos
<johnw>
clever: I'm on macOS :(
<exarkun>
darn, and all that was irrelevant, still fails to build on macOS
<johnw>
basically, what I want -- a Nix cache on an external HD that can grow to huge size while being effectively managed -- is proving impossible
<philipp[m]>
ottidmes: You sparked my interest! Would you share that module with me?
<johnw>
right now my cache is failing with: error: opening file '/Volumes/Backup/nix/store/15ydpqy0xnxad8qnamzj4cf4vmjifqyz-bootstrap-tools.drv': No such file or directory, and I have no way to correct this situation other than wiping the whole store and starting over
ingenieroariel has quit [Remote host closed the connection]
terlar has joined #nixos
<exarkun>
johnw: Why is it at /Volumes/Backup/nix instead of /nix
<exarkun>
johnw: That seems problematic.
<johnw>
the /nix volume is my active store on the SSD; /Volumes/Backup/nix is on an external HD where I have a ton more space and don't need to GC as much
ingenieroariel has joined #nixos
kenran has quit [Ping timeout: 256 seconds]
<johnw>
so what I do is nix copy from /nix to /Volumes/Backup/nix, and only GC often on /nix; I can let the backup store grow to 2 TB without worrying
<{^_^}>
[nixpkgs] @veprbl pushed commit from @xaverdh to master « wxmaxima: 18.02.0 -> 20.02.4 (#82323) »: https://git.io/JvKp6
pinecamp has quit [Remote host closed the connection]
<turion>
I have two systems, and checked out the same commit of the same repo containing the same nix & haskell code. I'm building with the same nix-channel. In my default.nix, I've added builtins.trace nixpkgs.lib.version, and on both systems it prints the same hash. Still, when I do nix-build, it will attempt to create derivations with different hashes, and one of the builds fails with a different error than the other. What could I be missing?
<{^_^}>
[nixpkgs] @vbgl pushed commit from @CohenCyril to master « coqPackages.hierarchy-builder: init at 0.9.0 »: https://git.io/JvKpb
<ottidmes>
turion: same nix-channel, do you also mean the same exact commit?
rail has joined #nixos
<ottidmes>
turion: other than that, try and reproduce it on the same system, by creating a clean environment that would be similar on the new system, if you get the same issue, than it would be easy to determine the cause
nschoe has joined #nixos
raingloom has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @hedning pushed 29 commits to gnome-3.36: https://git.io/JvKhe
<ottidmes>
turion: ah, then ignore what I just said, mine reports something differently, probably because my checkouts are made with fetchGit so they don't have a .git folder
<turion>
Arrrgh one git tree wasn't clean *facepalm*
<turion>
Ah interesting
<ottidmes>
turion: Ah, well at least you found the reason :)
<turion>
So you don't use nix-channel?
<turion>
Yes, phew. Thanks anyways
<ottidmes>
never have
genesis has joined #nixos
<turion>
Yeah, back then when I was new to all this, I couldn't imagine how it would work without, and now I'm sort of used to it
nschoe has quit [Ping timeout: 258 seconds]
<turion>
Do you do any kind of automatic upgrades
mounty has quit [Remote host closed the connection]
<ottidmes>
I do a lot of things automatic, but updates are one of the things I prefer to just do manually, but this comes down to just updating my config repo / calling a single command / running nixos-rebuild boot / reboot, not that painful
mounty has joined #nixos
<turion>
Fair enough. So you basically update the commit hash to the latest nixos-unstable or whatever you want?
<jasom>
I have kvm_amd in my boot modules, and it loads, but there is no /dev/kvm; any ideas?
<ottidmes>
I have a list of channels to commits in my lib, and the lib is part of my nixos config git repo, and I have my NIX_PATH variable include nixpkgs=/etc/nixpkgs-channel, and then I have a command that updates /etc/nixpkgs-channel to a checkout of the commit according to my lib
cr4y1 has quit [Ping timeout: 256 seconds]
lord| has joined #nixos
magnetophon has quit [Ping timeout: 258 seconds]
<turion>
Makes sense, thanks!
turion has quit [Quit: Leaving.]
<jasom>
Mar 10 20:00:16 mutsumi kernel: kernel BUG at arch/x86/kvm/mmu.c:296! <-- that's probably why
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
m1sosoup has quit [Remote host closed the connection]
m1sosoup has joined #nixos
<jasom>
Looks like it was a regression in v4.19.108 so I could just downgrade my kernel to v4.19.107 instead; not sure which is easier: 1 line patch vs kernel downgrade
domogled has joined #nixos
<ottidmes>
jasom: if it is short term and will be fixed soonish, I would downgrade, if it is longer term, I would patch, you don't want to stick to an old kernel
<symphorien>
Or use linux_latest maybe ?
chagra_ has joined #nixos
lord| has quit [Read error: Connection reset by peer]
<jasom>
ottidmes: won't be fixed upstream until v4.19.110 since it's not fixed in .109 either.
<clever>
exarkun: it has both a stripDebugList and a stripAllList
hoijui has quit [Ping timeout: 272 seconds]
<bela>
Hi, I'm sorry to bother you, but should i directly try to get the attention of some maintainer to merge the package, seeing there is a massive backlog or should i just wait?
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pikajude>
does nix-serve support `list` or does it only return results for specific paths?
<cole-h>
clever: It works very well for me. When I'm in my Windows VM, I ssh into my host for access to my password store and GPG indeed prompts with ncurses.
bvdw has quit [Read error: Connection reset by peer]
<exarkun>
Oh hm, because it's actually set to nothing and I have been misreading rustc.nix, thinking that stripDebugList there affects crate builds when it only affects the build of rustc itself
<clever>
exarkun: they can also be set by the stdenv and any setup hooks, try running nix-shell on the .drv file, and check the real env of that shell
<exarkun>
indeed, not set at all
justan0theruser is now known as justanotheruser
<{^_^}>
[nix] @roberth opened pull request #3403 → Issue 3398 path info cache ttls → https://git.io/Jv6JU
Jackneill has quit [Remote host closed the connection]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
acarrico has quit [Ping timeout: 256 seconds]
Neo-- has quit [Ping timeout: 256 seconds]
<exarkun>
Maybe just setting them in all of the crate sets that get generated ... it looks like buildRustCrate takes extra attributes and makes them derivation attributes
<qyliss>
You can globally override buildRustCrate
bela has quit [Remote host closed the connection]
<exarkun>
Maybe. I did try that, stuff kept exploding.
m1sosoup has quit [Quit: Leaving]
<qyliss>
e.g. for the thing I'm working on at the moment I do buildRustCrate = args: pkgs.buildRustCrate ({ inherit release; } // args);
<qyliss>
Which lets me globally enable/disable release mode
<exarkun>
That's pretty close to what I tried. The crate2nix generated code said "got a function where a set was expected"
<exarkun>
it uses buildRustCrate.override
<qyliss>
then it must have been critically different in some small way :)
<exarkun>
so I tried lib.makeOverridable but it didn't change anything in any way I could detect
<andi->
exarkun: IIRC you can provide a buildRustCrate function when calling the Cargo.toml
<exarkun>
andi-: On the command line?
<exarkun>
I didn't see any options for controlling what got generated like that
<andi->
exarkun: no, just wrap that with another file (default.nix or whatever)
<exarkun>
yea... that's what I was saying to qyliss
<exarkun>
but my alternate buildRustCrate was never good enough for it
<andi->
exarkun: mind creating a reproducer? Happy to take a look and make changes to crate2nix if required
<{^_^}>
[nixpkgs] @andir pushed 9 commits to release-19.09: https://git.io/Jv6Jl
<bqv[m]>
Whats buildRustCrate vs buildRustPackage?
heatmiser has quit [Max SendQ exceeded]
ingenieroariel has joined #nixos
<andi->
buildRustPackage is a huge fixed output hack that just uses cargo vendor in the fetch phase and packages one project as whole. buildRustCrate build a single crate (both (r|dy)lib's and binaries). You can stack many buildRustCrate instances to build a project.
<andi->
Usually you use something like carnix or crate2nix to do that
heatmiser has joined #nixos
<exarkun>
andi-: I don't know Rust or Cargo very well. I have a rather _large_ reproducer that I can share ... ie, the project I am actually working on now. :) Not sure I will have much success creating a smaller example, at least not in a reasonable timeframe.
raingloom has joined #nixos
<andi->
exarkun: mh, maybe just show me your alterntive buildRustCrate functions that didn't work?
o1lo01ol1o has quit [Remote host closed the connection]
kenran has quit [Ping timeout: 258 seconds]
<bqv[m]>
hmm. Is nix recursive yet?
<andi->
exarkun: I don't think you need the makeOverrideable there at all.
NobbZ[m] has joined #nixos
<exarkun>
andi-: if I just delete that and leave the function the same, I get the same error in the same place
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<exarkun>
"value is a function while a set was expected" on `else buildRustCrate.override {´ in the definition of buildRustCrateWithFeatures
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<andi->
exarkun: let me hack on an example
<exarkun>
Separately, I'm not sure the change I am hacking in there makes any difference anyway
<exarkun>
After re-reading the bash man page I'm not convinced `stripDebugList = null;` does what I meant. but `dontStrip = true;` should, but if I hack that into the crate attrsets, none of them get rebuilt, so I guess it doesn't actually do anything there
<exarkun>
Maybe the derivation that gets created from the `internal.crates` set doesn't actually contain the libraries, instead it just depends on other derivations that do
<lovesegfault>
andi-: I found a _really_ annoying thing with morph :(
<andi->
lovesegfault: write a bugreport? :)
<lovesegfault>
it will add the healthchecks to your system drv even if you don't set any, leading to a different output than building the system "raw"
<lovesegfault>
andi-: I'm trying to fix it instead
malSet has quit [Read error: Connection reset by peer]
iqubic` has joined #nixos
<zeta_0>
i tried nodePackages.angular-cli but it does not work?
Church- has quit [Ping timeout: 260 seconds]
<taika>
exarkun: well, I have my pythonXXPackages.package in system-wide pkgs, i assume it's impossible to expand inside the list, is there a beautiful way to do that.
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<lovesegfault>
but I don't know how to `nix-shell` this even as the @ makes nix really confused
<exarkun>
taika: I'm not sure (a) why you have python packages in system-wide pkgs - I assume you mean environment.systemPackages or (b) why it would be impossible to expand that list or (c) why you would want to expand it or (d) what you might consider beautiful
malSet has joined #nixos
<exarkun>
taika: Can you give a specific example of what you want to accomplish and perhaps describe the difficulty you have encountered in accomplishing it?
malSet has quit [Read error: Connection reset by peer]
iqubic` has quit [Remote host closed the connection]
<zeta_0>
lovesegfault: ya, the syntax highlighting for nodePackages.@angular/cli is really weird, but i'll give it a try
<lovesegfault>
zeta_0: `@` is an operator, so watch out
<exarkun>
andi-: I guess my solution for reaching inside the generated `internal.crates` is doomed to fail because the rootCrate is computed by that point and nothing cares about my changes anymore?
<lovesegfault>
you'll need to escape it or something
<exarkun>
if so maybe I really have to figure out how to override buildRustCrate after all
<{^_^}>
[nixpkgs] @Ma27 opened pull request #82339 → nixos/captive-browser: set chromium's data-dir to a XDG-compliant location → https://git.io/Jv6Uv
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<andi->
exarkun: yeah, the override story isn't exactly great right now.. It probably takes another attempt.
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<zeta_0>
lovesegfault: remind me, how do i escape the @ character, with a \ ?
<taika>
exarkun: (a) because its the place for all the packages? (b,c,d) I've tried pythonXX with pkgs { ... }; - that I consider *beautiful*. Well, Isn't it better to use { ... } list other than lost of packageXXPadckages.1 etc.
<lovesegfault>
zeta_0: I don't think I've had to do it before?
<exarkun>
taika: It's only the place for all system-wide packages, for what it's worth.
<lovesegfault>
andi-: did the ff-bin changes land in master yet?
<taika>
I am relatively new to NixoS so pls dont judge harshly :^)
<exarkun>
taika: I still don't really know what you're trying to accomplish or why. Why do you have Python packages installed system-wide and why do you want to have more of them?
<lovesegfault>
> let florb."@foo" = "bar"; in florb
<{^_^}>
{ "@foo" = "bar"; }
<lovesegfault>
> let florb."@foo" = "bar"; in florb."@foo"
<{^_^}>
"bar"
<lovesegfault>
zeta_0: ^
<andi->
lovesegfault: no, someone has to review them and we are still waiting on staging.. You are encouraged to use a *stable* channel for you need stable packages ;)
<exarkun>
taika: Local effects are better than remote effects. It's easier to reason about things that are nearby. System-wide configuration is more like remote than local. Something in your system-wide configuration controls whether something you're doing in an entirely different context will work or not.
mexisme_ has joined #nixos
<zeta_0>
lovesegfault: i placed ^ before the @ character, but a syntax error is still being thrown?
<exarkun>
taika: You can have Python packages installed for a certain application instead
<lovesegfault>
andi-: unstable-small or nothing
<exarkun>
taika: Or you can have a shell.nix that lets you "install" them for a single shell lifetime
malSet has joined #nixos
<andi->
lovesegfault: no
malSet has quit [Read error: Connection reset by peer]
<symphorien>
,libraries taika
<{^_^}>
taika: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
Makaveli7 has joined #nixos
<taika>
exarkun: Well, is't it weird to reinstall every imports every single time i want to code in python?
<symphorien>
taika: ^ this applies to python packages. Installing a python module system wide will have no effect
malSet has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<symphorien>
if you need to have some modules installed globally, you can install (python3.withPackages(ps: with ps; [ numpy ]))
<lovesegfault>
taika: IMHO one of the first steps towards understanding the NixOS way is to rethink the meaning of "installing" things
<lovesegfault>
and when exactly that is necessary
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<taika>
lovesegfault: Ok then, lets talk scope. If I am a python dev, why not install pythonPackages like [ numpy ] system-wide?
<exarkun>
Does your whole system need the development version of numpy that you realize you need one morning for some of your ongoing development work?
v88m has quit [Ping timeout: 255 seconds]
<zeta_0>
so how would i escape the @ and / characters in nodePackages.@angular/cli ?
<exarkun>
(no)
<taika>
To be clear, I am not using Nixos reproducibility for deploying, It's just my immortal workstation
<symphorien>
taika: you can start with globally installed stuff as a gateway drug. But one day you will have a project which requires a specific version of your globally installed module, and you will switch to nothing installed globally
<lovesegfault>
taika: do you have time? I think I can help explain things but it's a bit of a deep dive
<exarkun>
taika: You're using NixOS to have a more reproducible immortal workstation. :)
<taika>
lovesegfault: Yes, I have some
<exarkun>
taika: And to have more reproducible experiences on it
malSet has joined #nixos
<lovesegfault>
taika: Alright, let's go backwards a little bit then. Why do you normally install things system-wide and what does that _mean_? i.e. when is something "installed"
<taika>
exarkun: Well, as i said earlier, I'm still very new. 1.5 months in.
<lovesegfault>
and by "normally" I mean not on NixOS
<zeta_0>
i remembered having to use special syntax to escape certain characters a couple months back, but i can't seem to remember
<zeta_0>
hmm
absoluutely_ has joined #nixos
<taika>
lovesegfault: So, if its binary, it is copied in the appropriate place and linked as needed, automatically, as far as packet manager goes.
<lovesegfault>
taika: once upon a time disks were tiny so people wanted to put all their big stuff on a separate disk (and thus partition) from their root. that's how the / and /usr split was born
hoijui has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<lovesegfault>
taika: look into /bin, /lib, /usr/bin, /usr/lib in NixOS and tell me what you see
<lovesegfault>
red[evilred]: there's a workaround in the GH issu
<red[evilred]>
I'm looking at that now - I think
<red[evilred]>
thanks
jco has quit [Quit: WeeChat 2.7]
<taika>
lovesegfault: not much. /usr/bin has env
<Ankhers>
Would anyone be able to explain to me what an FHS environment is? I see something about it for android studio, but I have not idea what it means.
<taika>
lovesegfault: /lib /usr/lib i cannot access
<adisbladis>
Ankhers: You know how "regular" linux distros have /bin, /usr/bin, /lib and whatnot ?
<lovesegfault>
taika: Yeah, so these "places where things usually go" is what is called the FHS, the filesystem hierarchy standard
<lovesegfault>
taika: NixOS, by design, does not follow the FHS
<adisbladis>
Ankhers: Some packages are so painful to package that we provide a chroot to "emulate" the standard hierarchy
<{^_^}>
[nixpkgs] @amurrayc opened pull request #82341 → blender: use PYTHONPATH (#82337) → https://git.io/Jv6UC
<lovesegfault>
taika: What NixOS does, instead, is to manage dependencies separately. All your system's pkgs live in /nix/store and the PATH is patched to include the appropriate dep's contents
<lovesegfault>
(this is a simplification)
philr has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<lovesegfault>
so you wouldn't install numpy systemwide, where would it even go? Instead what you do is let Nix handle making python see the numpy in /nix/store
<Ankhers>
adisbladis: Does this have any affect on binaries downloaded by the application?
<lovesegfault>
for a quick dev environment that looks like this: nix-shell -p 'python3Full.withPackages (ps : with ps; [ numpy ])'
<adisbladis>
Ankhers: Yeah, in the sense that they are more likely to work out of the box.
raingloom has quit [Ping timeout: 272 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<lovesegfault>
taika: run the above command and then do `python -c "import numpy"` to verify it works
<taika>
lovesegfault: And when does that env cease to exist? On system exit or on quit?
<lovesegfault>
taika: when you exit the shell, but the contents live on in the nix store
malSet has joined #nixos
<lovesegfault>
so if you open that shell again it gets created again without having to redownload anything
<lovesegfault>
unless, of course, you run nix-collect-garbage which deletes all "unused" elements in the store
malSet has quit [Read error: Connection reset by peer]
<lovesegfault>
such as dependencies to a no-longer-active nix-shell
<Ankhers>
adisbladis: So if, in my case, Android Studio downloads something, it _should_ work?
waleee-cl has quit [Quit: Connection closed for inactivity]
<taika>
lovesegfault: Oh, that's convenient. But i tend to constantly collect garbage. And still see old gen entires on boot despite commands from manuals used.
malSet has joined #nixos
<{^_^}>
[nixpkgs] @Ma27 opened pull request #82342 → rustPlatform: increase build-speed of `checkPhase` for rust-packages → https://git.io/Jv6Ug
malSet has quit [Read error: Connection reset by peer]
<cole-h>
Correct me if I'm wrong, but I believe all (or most) generations are kept for rollback purposes
<lovesegfault>
taika: to nuke old generations you need `-d`
<lovesegfault>
`sudo nix-collect-garbage -d` will clean up all old gen
malSet has quit [Read error: Connection reset by peer]
alexherbo24 has quit [Ping timeout: 265 seconds]
<sjuberman>
infinisil: It's not possible to create a nix expression for what I'm trying to do. I'm trying to use Bazel in NixOS and Bazel is build system that bootstraps various binaries and executes them.
<infinisil>
Did you try pkgs.buildBazelPackage?
teto has quit [Quit: WeeChat 2.7.1]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<infinisil>
It's pretty much always possible to use Nix to package something or to use it for development. It's just a matter of how much effort is required to do it :)
malSet has joined #nixos
malSet has joined #nixos
<gamble>
Huzzah `nix-shell -p appimage-run --run "appimage-run ./navicat15-pgsql-en.AppImage"`. Works
<sjuberman>
Yeah, but sometimes the effort unfortunately does not make much sense. E.g. I'm working in a monorepo where we use Go/Scala/TypeScript/Python and creating some workaround for all these toolchains is just a massive pain :(
malSet has quit [Read error: Connection reset by peer]
alexherbo2 has quit [Ping timeout: 255 seconds]
o1lo01ol1o has joined #nixos
Makaveli7 has quit [Quit: WeeChat 2.4]
<sjuberman>
But I think I might try to use the modules in the PRs I linked to see if I can make it work with Bazel.
<sjuberman>
This is the only blocker for me so far. Otherwise I'm pretty much loving NixOS.
<bqv[m]>
sjuberman: well surely you could just build a fhs environment with pacman in it or something lol
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<sjuberman>
I tried using buildUserFHSenv and I got segmentation faults :(
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<bqv[m]>
that seems like something that could be solved
<bqv[m]>
what segfaulted?
gamble has quit [Remote host closed the connection]
<sjuberman>
Don't remember, was a while ago. If I remember correctly I installed the Bazel package in the FHS env and then it segfaulted when I tried to run it.
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<Ankhers>
adisbladis: In my case, I am trying to download the android NDK. Unfortunately all of the binaries need to be patchelf'd because of an incorrect interpreter. Is this something that is expected, or should it Just Work™?
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
malSet has quit [Read error: Connection reset by peer]
raingloom has quit [Ping timeout: 272 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<Ankhers>
adisbladis: Did you have to compile things manually using the NDK, or did you somehow build it through android studio?
heatmiser has quit [Ping timeout: 240 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<adisbladis>
Ankhers: I did it through android studio
<{^_^}>
[nixpkgs] @Pattag opened pull request #82343 → <!-- Nixpkgs has a lot of new incoming Pull Requests, but not enough people to review this constant stream. Even if you … → https://git.io/Jv6TB
<sjuberman>
jared-w: Yes, but this is not an option in my case since I can't change the WORKSPACE dependencies/configuration since it affects all of my coworkers.
raingloom has joined #nixos
<{^_^}>
[nixpkgs] @Pattag closed pull request #82343 → haskellPackages.HSH build on my Machine → https://git.io/Jv6TB
zupo has joined #nixos
lovesegfault has quit [Quit: WeeChat 2.7.1]
reactormonk has quit [Ping timeout: 256 seconds]
zupo_ has joined #nixos
wfranzini has quit [Remote host closed the connection]
absoluutely__ has quit [Read error: Connection reset by peer]
absoluutely__ has joined #nixos
absoluutely___ has quit [Ping timeout: 255 seconds]
<Ankhers>
If an application in an FHS environment downloads a binary, should I be able to run that binary from my shell, or does it only work through the application that downloaded it?
<mdash>
Anybody been seeing weird stuff with intel video drivers in the past few weeks on unstable? alacritty is locking up, getting transient video glitches in firefox, etc
rb2k has joined #nixos
<{^_^}>
[nixpkgs] @etu opened pull request #82348 → [wip] php extensions as packages → https://git.io/Jv6kR
<manveru>
exarkun: `--trace-function-calls` can help sometimes
<drakonis>
see also the source code for the flakes branch
<bqv[m]>
also, i've had plenty of people say it's cool but it's not very useful unless someone actually bothers to merge it, and i'm sick of trying to casually mention it here to get attention
<bqv[m]>
this PR hell is a mess
<manveru>
yeah... sorry, but i don't have merge rights :P
<rb2k>
Hey! A possibly somewhat horrible question... I am trying to build a quick prototype that includes running nix on macOS. So far so good. I would love to compile packages from source. "--option substitute false" works fine for me with that. The problem: coreutils refuses to compile when the build process runs as root. Sadly, mine does. The derivation has support for setting the env var on solaris: https://github.com/NixOS/nixpkgs/blob/c195ef6aab79b819
<rb2k>
4158b0b2fadbf96f986b14c8/pkgs/tools/misc/coreutils/default.nix#L116 Now the question: Is there any way for me to inject that into the build environment?
<rb2k>
I tried export, but that didn't seem to arrive where I wanted it to
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
<rb2k>
(I'm a nix noob in case that wasn't obvious)
bhipple has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @Mic92 opened pull request #82350 → nixos/haproxy: Revive the haproxy user and group → https://git.io/Jv6kP
o1lo01ol1o has quit [Remote host closed the connection]
fendor has quit [Read error: Connection reset by peer]
<immae>
bqv[m]: I understand your frustration, it’s the same reason I keep my softwares in NUR rather than taking the risk of making a PR that would be ignored forever...
feepo has joined #nixos
<immae>
(you might consider publishing your code in NUR though, if you like)
<bqv[m]>
NUR is a much nicer idea
newhoggy has joined #nixos
cbarrett has joined #nixos
<bqv[m]>
i'm holding off for flakes to stabilize so this problem is solved once and forever
<danderson>
NUR is going to have the same problem as AUR, which is vastly varying quality bars
rodarmor has joined #nixos
elvishjerricco has joined #nixos
parseval has joined #nixos
<bqv[m]>
and i'm fine with that because at least it'll be possible to contribute to it without having to go through what nixpkgs is like at the moment
<danderson>
I'd love to help with the PR backlog, but I don't know how :/ I can review simple stuff, but without the power to merge anything it just ends up being more "LGTM" comments on the PR that also end up languishing for weeks
<bqv[m]>
the fact that there's even a separate forum for people to post PRs that have been open and reviewed for ages is just such an indictment
bgupta has joined #nixos
jfhbrook has joined #nixos
<bqv[m]>
that really shouldn't have to be the case, ever
nh2 has joined #nixos
marcinkuzminski has joined #nixos
diamondman has joined #nixos
mudri has joined #nixos
lightbulbjim has joined #nixos
TallerGhostWalt_ has joined #nixos
johanot has joined #nixos
noonien has quit [Write error: Connection reset by peer]
lukego has quit [Remote host closed the connection]
sdier has quit [Remote host closed the connection]
techtangents has quit [Remote host closed the connection]
sevanspowell has quit [Remote host closed the connection]
hoek has quit [Write error: Connection reset by peer]
kitemikaze has quit [Remote host closed the connection]
d10n-work has quit [Remote host closed the connection]
adamse has quit [Remote host closed the connection]
evck has joined #nixos
rajivr___ has quit [Remote host closed the connection]
<danderson>
I agree. How do we fix it? I want to volunteer time to help, but afaict there's just no way for me to do anthing useful
<danderson>
... wow.
ixxie has quit [Ping timeout: 265 seconds]
tnks has joined #nixos
zertox has joined #nixos
lstanley has joined #nixos
teehemkay has joined #nixos
<bqv[m]>
well, we can't fix it, that's why i closed it, waiting for flakes to stabilize so we can just stop relying on nixpkgs and community repos are first class citizens
<qyliss>
danderson: I can assure you that those LGTMs are extremely helpful
NemesisD has joined #nixos
<infinisil>
bqv[m]: Mind reopening the PR so I can merge it?
tazjin has joined #nixos
johs has joined #nixos
pasukon has joined #nixos
d10n-work has joined #nixos
zupo has joined #nixos
adamse has joined #nixos
zupo_ has quit [Ping timeout: 268 seconds]
christiaanb has joined #nixos
<danderson>
I'm fine with the general idea of NUR and adopting an Arch-ish model of "core vs. contrib", but I'm worried about putting too much stuff in NUR and not enough in core.
<infinisil>
It's looking good to me, especially with nloomans approval comment
<danderson>
I don't want my nixos to become npm :/
<qyliss>
me too
d1rewolf has joined #nixos
waleee-cl has joined #nixos
<qyliss>
not sure what we can do about it, though...
chessai has joined #nixos
nlofaro has joined #nixos
ajmcmiddlin has joined #nixos
sdier has joined #nixos
kitemikaze has joined #nixos
techtangents has joined #nixos
<rb2k>
infinisil that's fine, we have thousands of machines
kalbasit has joined #nixos
coot has joined #nixos
hoek has joined #nixos
<danderson>
well, using the factory throughput model, the bottleneck right now is reviewers to merge stuff. Two solutions: reduce the amount of incoming stuff (make it harder to submit PRs, or push people to NUR), or scale up the bottleneck (more submitters)
lukego has joined #nixos
<bqv[m]>
seriously?~
<bqv[m]>
no
<bqv[m]>
i don't want it merged because then i become a maintainer of it
sevanspowell has joined #nixos
alunduil has joined #nixos
c00w has joined #nixos
blackriversoftwa has joined #nixos
<bqv[m]>
and then i'm gonna have to throw a hissy fit to get any edit added
bhipple has quit [Ping timeout: 265 seconds]
<infinisil>
Would it be okay if I removed you as the maintainer for it?
aria has joined #nixos
mexisme_ has quit [Ping timeout: 256 seconds]
fnords has joined #nixos
<qyliss>
please don't merge code with no maintainer
<infinisil>
I can add myself as a maintainer then, it does work nicely for slack :o
<bqv[m]>
do what you want with it. i just don't want to have to deal with any nixpkgs bs
cvlad- has quit [Remote host closed the connection]
joshmeredith has quit [Remote host closed the connection]
sorear has quit [Remote host closed the connection]
teozkr has quit [Remote host closed the connection]
philipcristiano has quit [Remote host closed the connection]
alanz has quit [Remote host closed the connection]
<danderson>
good grief freenode, chill out would ya
<qyliss>
looks like it's irccloud's problem, not freenode's
<immae>
infinisil: Did you notice that the ripcord license says that slack use requires a paid license after some time?
etrepum has joined #nixos
dukedave has joined #nixos
aristid has joined #nixos
teozkr has joined #nixos
<infinisil>
Oh haha nope
<immae>
(just in case, it works nicely now but it may change later :p )
philipcristiano has joined #nixos
alanz has joined #nixos
r0bby has joined #nixos
ctp has joined #nixos
<infinisil>
bqv[m]: Alright, I'll open a new PR with it, removing you as a maintainer. Thanks for all the work on it though <3. Want me to also rewrite the commit as mine such that you're not in the commit log?
kvda has joined #nixos
vdemeester has joined #nixos
mitsuhiko has joined #nixos
<bqv[m]>
im not fussy on that, up to you
sorear has joined #nixos
<infinisil>
Alright cool, I'll leave you as the commit author then
ryjm has joined #nixos
joshmeredith has joined #nixos
lally has joined #nixos
claudiii has quit [Read error: Connection reset by peer]
peel has quit [Remote host closed the connection]
jchw has quit [Remote host closed the connection]
bitonic has quit [Remote host closed the connection]
digitalgrease has quit [Remote host closed the connection]
zupo has quit [Ping timeout: 240 seconds]
cvlad- has joined #nixos
<manveru>
hopefully by the time it needs a license we'll have migrated to mattermost :)
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #nixos
<{^_^}>
[nixpkgs] @Infinisil opened pull request #82352 → ripcord: init at 0.4.23 → https://git.io/Jv6IC
emilazy has joined #nixos
zdm has joined #nixos
<infinisil>
Maybe we should have an IRC channel like #nixos-prs where people can ask for PR reviews or merges
peel has joined #nixos
kapil_ has quit [Remote host closed the connection]
Tritlo has quit [Remote host closed the connection]
dmj` has quit [Remote host closed the connection]
thoughtpolice has quit [Remote host closed the connection]
nick_h has quit [Remote host closed the connection]
<immae>
infinisil: if you have the contacts and time you really should start a discussion about the state of nixpkgs for "non-official" maintainers, bqv is not the only one frustrated by the current state and it’s only chance if you were there when he expressed it...
pkral has joined #nixos
<infinisil>
immae: Hm what's the thing to discuss?
TheNumb has joined #nixos
claudiii has joined #nixos
<immae>
improving the workflow for contribution to nixpkgs
<immae>
(it can be by making it impossible, it’s kind of a solution)
sethetter_ has joined #nixos
zimbatm has joined #nixos
<immae>
(but giving a false hope of contributing and then ignoring ccontributions forever is clearly a bad thing)
Tritlo has joined #nixos
<{^_^}>
[nixpkgs] @Ma27 opened pull request #82353 → nixos/nextcloud: fix upgrade path from 19.09 to 20.03 → https://git.io/Jv6I8
thoughtpolice has joined #nixos
mpickering has joined #nixos
nand0p_ has joined #nixos
dmj` has joined #nixos
<infinisil>
immae: I think rfcs#50 might help a lot
<immae>
(I have to admit that I’m not very serene with the fact that the PR is closed and that the original poster himself decided that it wasn’t a priority anymore)
kvda has joined #nixos
justache has joined #nixos
<infinisil>
Another thing that might help the situation is to encourage (privileged) reviewers to update the PR on their own
<infinisil>
Unless the author unchecked "Allow edits from maintainers" for the PR this is possible
<immae>
It could help indeed, but I’m not sure it would be sufficient (as an example the only PR I posted in nixpkgs took many weeks in spite of the allow-edits checkbox and the fact that it was a trivial change in documentation)
<immae>
infinisil: do you have an idea of the number of privileged reviewers of nixpkgs?
<infinisil>
The number of the people that can merge PRs is over 100 I think
<infinisil>
Though many of which aren't active anymore
<infinisil>
This issue has since been used to nominate people
<immae>
ok
<immae>
I’m only starting to read it, but the first thing that comes to mind is "50 PR’s is impossible to reach currently without getting crazy"
<infinisil>
Hehe yeah
<infinisil>
It was just a starting point
<gentauro>
I just "upgraded" to the newest stable NixOS channel (19.09) and due to this, I got the latest FF (73.0.1 (64-bit)). The problem is that I have no audio
<gentauro>
I started Chromium and there I have audio
<gentauro>
anybody know where to `start digging`?
<immae>
(I honestly think I would have reached that point if the reviewing process didn’t seem impossible to reach and so give up before even starting)
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
<danderson>
same. I added a nixos module for pppd, and that took so long I basically stopped trying :/. I want to try again, but I honestly don't even know how to maintain long-lived PRs like this without going mad
<immae>
Archlinux looks at personnal packages (equivalent of NUR), and how much the person interacts with the community (IRC/forum/...), would that be a sensible thing to look at?
<infinisil>
Yeah, so with a completely random PR all of them have equal chance of being looked at, which should decrease the maximum waiting time for the same amount of review time spent
<immae>
infinisil: you could filter PR’s that are "mergeable without conflict" :)
<manveru>
i'm mostly just making PRs and then forget about them... sometimes they get merged, but i'm not depending on it anyway
<immae>
(I don’þ know how hard it is)
<infinisil>
immae: That might be easy actually! Lemme see..
<colemickens>
subdirs, working groups with authority delegated to them, something. Or some tooling so that users can nominate PRs as ready to review and we can crowd-source manual verification to build up more of a signal for those blessed with merge power to evaluate.
drakonis has quit [Ping timeout: 255 seconds]
mexisme_ has joined #nixos
<qyliss>
All maintainers have access to labelling, I believe.
<qyliss>
And anybody can do an approval
<immae>
qyliss: I have 58 original programs in my NUR (clearly not all of which are worth going in nixpkgs of course, and I know there is some "beginner errors" there), and I have about twenty modules (some that corresponds to the packages), that is without counting the overlays (fixes/improvements). I guess I could make all the PRs as required, but I would probably ragequit before I would have a chance to reach half
<immae>
of the requirements. Why would looking at NUR (rather/in addition to PRs) be a bad thing to evaluate the possibility of someone becoming privileged maintainer, for instance?
<qyliss>
As a committer, I can filter PRs by ones with an approval.
<qyliss>
I didn't say it would be?
<immae>
Oups right
<colemickens>
qyliss: alright, well maybe I need to try harder with the tooling I have. and/or review what sort of documentation we have on this for community members.
<immae>
(bad wording, I didn’t mean the phrasing as an accusation)
<qyliss>
immae: But I think mostly, if I were evaluating these things, I'd be interested in seeing that you could work within Nixpkgs.
<peelz>
how long after a PR getting merged can I expect to see it appear on nixpkgs-unstable?
<zeta_0>
if i install home-manager separately from my system configuration will i have to reboot every time i install something with home-manager? right now i have home-manager integrated with configuration.nix, so every time i install a package in home.nix i need to run nixos-rebuild switch, then reboot(which is a hassle)
<qyliss>
zeta_0: You shouldn't need to reboot.
<immae>
qyliss: yes but for that there is the barrier of "is that feasible"? (and clearly I lack knowledge in that, I completely reckon, in particular because it seems impossible to do)
<qyliss>
nobody is expecting new committers to have 50 prs
<Ankhers>
bqv[m]: That makes sense. So in that case, I should just be writing a derivation for the thing I need? Or is there some magic I could do in order to get it to work correctly in my make script.
CMCDragonkai1 has quit [Ping timeout: 258 seconds]
<qyliss>
those were proposed requirements that were imo unreasonably high
<gentauro>
vk3wtf this should do the trick right? `nixpkgs.config.pulseaudio = true;`
<immae>
ok
<gentauro>
I will still have ALSA, but thorugh the pulseaudio API
rb2k has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixos-homepage] @garbas opened pull request #335 → include nix install script in the website and remove the redirect (fo… → https://git.io/Jv6IA
<bqv[m]>
i actually like the subdir idea as well. anything to reduce the fact that all PRs are bottlenecked by a tiny cabal of people who give the appearance of ignoring PRs as a habit. i like the idea of the nur or flakes more, because then the delegation is even further, but at the basic fact that an "average" open PR in nixpkgs is 2 years old is ridiculous
<qyliss>
bqv[m]: you are being quite rude about our volunteer committers and I am suggesting that you be careful of that
<{^_^}>
cole-h: I'll think about it..
<bqv[m]>
Ankhers: i can't remember what I responded to, sorry
<qyliss>
The "tiny cabal" has 132 members
<gentauro>
vk3wtf: thx :)
<{^_^}>
cole-h: I'll think about it..
<bqv[m]>
qyliss: i've phrased that very carefully to not be rude, note the word "give the appearance of". I'm not saying they aren't making efforts, but if you can give me any proof that appearance isn't given, i'd be very interested.
<qyliss>
That is a double digit percentage of all Nixpkgs maintainers.
<gentauro>
I'll give it a try !!!
zdm has quit [Quit: WeeChat 2.7.1]
<maralorn>
The problem with nixpkgs is that it is one big repo written as one turing complete program. Any maintainer even if they had only write access to a certain directories needs a quite large amount of trust.
<infinisil>
Haha whoops
<infinisil>
{^_^}: Stop that
<qyliss>
bqv[m]: Your "careful phrasing" didn't avoid it being rude. That's not how language works.
<peelz>
how long after a PR getting merged can I expect to see it appear on nixpkgs-unstable?
<peelz>
qyliss: what's the process for updating the unstable channel?
<peelz>
is it automated?
<qyliss>
yes
<Cadey>
how do i use my own nixpkgs repo in a nixos config?
<peelz>
qyliss: okay thanks :)
<infinisil>
bqv[m]: Hold on, that was a *random* PR. The age of a random single sample doesn't say anything about the average age
<infinisil>
In fact I think the average PR age is probably rather low
<bqv[m]>
i don't understand how what I said is incorrect. quite clearly the common opinion among non-maintainers here is that completing a PR in nixpkgs seems like a neverending uphill struggle, thus the appearance given is that nobody cares. I again am not saying that that is the case, but the appearance is given
<qyliss>
439 pull requests have been merged THIS WEEK
zdm has joined #nixos
<immae>
qyliss: it still takes weeks to have even trivial PR’s merged, so I completely understand bqv[m]’s impression...
kleisli has joined #nixos
<qyliss>
I'm not saying there isn't a problem.
<immae>
does 439 include automatically merged PR’s by the update bot ?
<bqv[m]>
all i'm saying is that there is a problem.
<infinisil>
Nothing gets merged automatically
<qyliss>
no PRs are automatically merged
<immae>
ah ok
<Valodim[m]>
I'm a non-maintainer, started with nix in january this year, got a couple of PRs merged. Perhaps I'm just lucky, but not everyone has a terrible experience :)
<qyliss>
I strongly reject the implication that our committers don't care or are ignoring PRs.
<bqv[m]>
infinisil: good point, though it looks like at least the median is 4 or so months
<Valodim[m]>
(my girlfriend got a pr merged, too :D)
__monty__ has quit [Quit: leaving]
<gentauro>
vk3wtf: it worked !!! Thx again :)
kleisli has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @alyssais pushed commit from @jansol to master « renderdoc: 1.6 -> 1.7 »: https://git.io/Jv6LT
<zeta_0>
qyliss: well then, i guess i'll keep my current setup where i have home.nix integrated as a module to configuration.nix, that way i don't need to remember the home-manager commands as well
<vk3wtf>
gentauro: fantastic :)
kleisli has quit [Ping timeout: 255 seconds]
<maralorn>
I think the most marvelous advantage of nixos over arch is that I never again have the "you can‘t mount that filesystem because your installed kernel modules are to new for the running kernel" problem.
<immae>
puzzlewolf: if you’re fine with it, then it’s good for you, but I cannot hold a finished work for so long in my mind without getting crazy (anod not caring is not an option for me)
<infinisil>
I shall now calculate the average PR age
<{^_^}>
[nixpkgs] @alyssais merged pull request #82190 → rainicorn: remove broken package that has been abandoned upstream → https://git.io/JvoHr
<{^_^}>
[nixpkgs] @alyssais pushed commit from @bhipple to master « rainicorn: remove broken package that has been abandoned upstream »: https://git.io/Jv6Ln
<immae>
anyway, it was just an attempt to start a discussion, I’m fine with NUR as far as I am concerned, I’m just a bit sad not to be able to contribute correctly
<{^_^}>
[nix] @edolstra pushed commit from @dtzWill to master « local.mk: fix user-env.cc dep on buildenv.nix.gen.hh, resolve occasional build failure »: https://git.io/Jv6LC
<bqv[m]>
honestly so am i, i'm quite happy to not contribute to nixpkgs because i don't see a confident solution to this beyond nur or flakes, cause i'm not gonna sit around waiting for a PR for 9 weeks either