<taeradan>
hi, I've got a problem with lib.recursiveUpdate : a function "myFunction = someOptions: someOptions" works, but "myFunction = someOptions: lib.recursiveUpdate {} someOptions" complains that "value is a function while a set was expected", where am I wrong ? I thought that recursiveUpdate output should be of same type as its inputs
clonesAnts has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to master: https://git.io/vHOEQ
<NixOS_GitHub>
nixpkgs/master 2d04e56 Franz Pletz: imagemagick7: 7.0.4-6 -> 7.0.5-7
<NixOS_GitHub>
nixpkgs/master e1b3f85 Franz Pletz: imagemagick: 6.9.8-4 -> 6.9.8-6
NixOS_GitHub has left #nixos []
justan0theruser has joined #nixos
mizu_no_oto has joined #nixos
<Nobabs27>
so, say I downloaded a theme, where would I put it?
<clever>
Nobabs27: the stuff in ~/.config/ should also work
<Nobabs27>
??
<Nobabs27>
so like, make a .coonfig/themes or?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin opened pull request #26075: [WIP] One way to do other types of services (master...service-generalization) https://git.io/vHOu3
NixOS_GitHub has left #nixos []
<clever>
there should be a path in $HOME that allows themes, just not sure where
<clever>
google probably knows
alx741 has joined #nixos
<Nobabs27>
duckduckgo for the win (unless images or converters) xd
dbmikus has joined #nixos
<Nobabs27>
.themes apparently
yegods has joined #nixos
<clonesAnts_>
I'm going to try installing again at a later time. It's failing in the same way on a wired connection with '-j 1'. I'm getting tired so I'll leave this istallation for later
<clonesAnts_>
Thanks for the help everyone :)
clonesAnts_ has quit [(Quit: Page closed)]
sheenobu has joined #nixos
<sheenobu>
i seem to be linking against two separate versions of glibc when building some binaries
<sheenobu>
pthread and libc are coming from glibc 2.24. the rest are coming from glibc 2.25
hellrazo1 has quit [(Read error: Connection reset by peer)]
hellrazor has joined #nixos
<clever>
sheenobu: can you gist your nix expression?
<sheenobu>
I'm just hoping into a nix-shell and running go build
fifo23 has quit [(Remote host closed the connection)]
fifo23 has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
mizu_no_oto has joined #nixos
fifo23 has quit [(Ping timeout: 240 seconds)]
indi_ has quit [(Ping timeout: 240 seconds)]
indi_ has joined #nixos
ertesx has joined #nixos
fifo23 has joined #nixos
indi_ has quit [(Ping timeout: 240 seconds)]
ertes has quit [(Ping timeout: 246 seconds)]
ertesx is now known as ertes
matthewbauer has joined #nixos
indi_ has joined #nixos
<dmj`>
nix nix nix nixnxixnixn
bhipple has joined #nixos
Nobabs27 has quit [(Quit: Leaving)]
johnsonav has joined #nixos
schoppenhauer has quit [(Ping timeout: 260 seconds)]
indi_ has quit [(Ping timeout: 240 seconds)]
indi_ has joined #nixos
schoppenhauer has joined #nixos
jgertm has quit [(Ping timeout: 240 seconds)]
hotfuzz_ has joined #nixos
hotfuzz has quit [(Ping timeout: 260 seconds)]
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
marsel has joined #nixos
marsel has quit [(Ping timeout: 240 seconds)]
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
systemfault has joined #nixos
systemfault has quit [(Client Quit)]
systemfault has joined #nixos
sheenobu has quit [(Ping timeout: 246 seconds)]
dbmikus has quit [(Quit: WeeChat 1.7)]
cpennington has quit [(Remote host closed the connection)]
abrar has quit [(Ping timeout: 260 seconds)]
takle has joined #nixos
takle has quit [(Ping timeout: 272 seconds)]
matthewbauer has quit [(Ping timeout: 240 seconds)]
stallion_ has quit [(Quit: stallion_)]
seanz has joined #nixos
jgertm has joined #nixos
seanz has quit [(Client Quit)]
seanz has joined #nixos
Tipping_Fedora has quit [(Ping timeout: 268 seconds)]
<benley>
nixynixynix
<benley>
Anyone else here use services.redshift.enable=true?
matthewbauer has joined #nixos
yegods has quit [(Remote host closed the connection)]
<benley>
I'm trying to use that, and I've ended up with redshift running both as my user account and as the sddm system account simultaneously, which doesn't exactly work well. The sddm instance is crashlooping.
<benley>
and I don't know how to fix that.
stallion has joined #nixos
Tipping_Fedora has joined #nixos
ericsagnes has quit [(Ping timeout: 268 seconds)]
stallion has quit [(Ping timeout: 268 seconds)]
systemfault has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] nocoolnametom opened pull request #26081: znc module: Fix error with bitlbee channel closing tag missing a newline (master...update-znc) https://git.io/vHO9C
NixOS_GitHub has left #nixos []
marsel has joined #nixos
jgertm has quit [(Ping timeout: 260 seconds)]
watersoul_ has quit [(Ping timeout: 260 seconds)]
jgertm has joined #nixos
seanz has quit [(Quit: Leaving.)]
ericsagnes has joined #nixos
watersoul has joined #nixos
systemfault has joined #nixos
watersoul has quit [(Read error: Connection reset by peer)]
alx741 has quit [(Quit: alx741)]
watersoul has joined #nixos
stallion has joined #nixos
eacameron has joined #nixos
stallion has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Ping timeout: 246 seconds)]
jgertm has quit [(Ping timeout: 260 seconds)]
jluttine has joined #nixos
betaboon has quit [(Quit: This computer has gone to sleep)]
fifo23 has quit [(Remote host closed the connection)]
fifo23 has joined #nixos
fifo23 has quit [(Ping timeout: 240 seconds)]
proteusguy has quit [(Remote host closed the connection)]
eacameron has joined #nixos
systemfault has quit [(Quit: Bye!)]
takle has joined #nixos
stallion has joined #nixos
eacameron has quit [(Ping timeout: 246 seconds)]
stallion has quit [(Ping timeout: 245 seconds)]
takle has quit [(Ping timeout: 245 seconds)]
matthewbauer has quit [(Ping timeout: 240 seconds)]
eacameron has joined #nixos
<evangeline>
hi, is there any view/chart to see the nixpkgs the developer is maintaining; assuming not grepping through nixpkgs?
eacameron has quit [(Ping timeout: 255 seconds)]
stallion has joined #nixos
thc202 has joined #nixos
stallion has quit [(Ping timeout: 240 seconds)]
takle has joined #nixos
<dalaing>
lukec1: I have just finished mucking about with curl to grab data from hydra and from gitlab - will break ground on the code tomorrow
eacameron has joined #nixos
<lukec1>
dalaing: nice.
eacameron has quit [(Ping timeout: 246 seconds)]
takle has quit [(Remote host closed the connection)]
justan0theruser has joined #nixos
jacob_ has joined #nixos
justanotheruser has quit [(Ping timeout: 240 seconds)]
jacob_ is now known as Guest93183
Bradyn2000 has joined #nixos
<ikwildrpepper>
copumpkin: will check, we had some network issues, perhaps that was related
<viric>
Hello nixers
<ikwildrpepper>
hiya viric
<viric>
here we face a need of a package manager... but we have one requirement that is not quite met with nix
<viric>
Once we build something, we need to have unique output binaries
<viric>
bit-to-bit
<viric>
So we likely have to ensure that they are never rebuilt again
stallion has joined #nixos
<viric>
ikwildrpepper: niksnut: have you dealt ever with an 'artifactory' or thing like that? Or hydra providing unique versioning? We need, basically, a version (major.minor.build or so) that will uniquely mean some binaries, and that potentially uploaded to an "artifactory server" or something like that
<viric>
ikwildrpepper: niksnut: I have to defend nix in front of TeamCity or Jenkins, regarding the "result binaries stored in an artifactory server"
<viric>
ikwildrpepper: niksnut: uniquely identified by a name-major.minor.build 3-integers thing
stallion has quit [(Ping timeout: 260 seconds)]
<ikwildrpepper>
viric: well, you can just write a hydra plugin to upload in whatever scheme you want, right?
<ikwildrpepper>
(or whatever CI you use with nix)
eacameron has joined #nixos
<ikwildrpepper>
the only issue you might encounter, if dependencies change which would produce a build, you might have 2 builds with the same x.y.z build
proteusguy has joined #nixos
<viric>
ikwildrpepper: I don't understand this casse. Can you ellaborate?
<viric>
case
<ikwildrpepper>
simplest case: you have a build that uses nixpkgs
<ikwildrpepper>
source of your app does not change, but nixpkgs does
<viric>
The CI should 1) increase the "build" integer of the version and 2) pass the version to the build recipe, so outputs can include this version in it.
<ikwildrpepper>
ah
<ikwildrpepper>
misread you version scheme
<ikwildrpepper>
sorry
<viric>
ah ok
<viric>
do you have anything like that somewhere?
<viric>
hydra managing some uniqueness version numbers
<makefu>
hey all, someone already has a hotpatch for samba as a nix expression?
<ikwildrpepper>
this will be a bit tricky for hydra, as you don't know build number at eval time
<viric>
It makes the CI have a 'state'.
<ikwildrpepper>
viric: does the source come from git?
<ikwildrpepper>
if so, then you can do what nixos does with its versions
eacameron has quit [(Ping timeout: 255 seconds)]
<ikwildrpepper>
although that still doesn't include build number, just a revision number
<viric>
ikwildrpepper: likely git. You mean "git describe"?
<viric>
ikwildrpepper: that won't change if a dependency changes.
<ikwildrpepper>
no, but that might lead to two builds with the same version
<viric>
well, we can use another CI, not hydra, to get that unique integer per build.
<ikwildrpepper>
not sure if that would be a problem for you
<viric>
You have that integer; it's the identifier in the hydra db :)
<viric>
two builds with the same version is the worst case :)
<viric>
Meaning... it's a very bad case. Should never happen
<ikwildrpepper>
yeah, but adding that would make builds be dependent on the build id
<viric>
ikwildrpepper: I'd need a build id as impurity or so.
<ikwildrpepper>
I guess we could make something in hydra that would add that by overriding the derivation
ambro718 has joined #nixos
<viric>
ikwildrpepper: did you ever hear of connan.io?
<ikwildrpepper>
conan.io?
<ambro718>
How can I refer to a nix store path in a nix expression (for testing/debugging purposes)?
<ikwildrpepper>
jfrog
<ambro718>
e.g. as one of the build inputs
<viric>
ikwildrpepper: conan.io, yes
<viric>
Another thing that we would need in nix is some ACL in hydra/nix cache serving.
kampfschlaefer has joined #nixos
<viric>
Is all ACL delegated to an http server in between?
matthewbauer has joined #nixos
<ikwildrpepper>
viric: yes
justbeingglad has joined #nixos
<viric>
How can we authorize certain users to access a binary cache? Did anyone try that? the nix substituter won't ask for a user/password
<ikwildrpepper>
you'd need to implement acl yourself. currently it is basically access to all or nothing
<ikwildrpepper>
not sure if the substituter uses the /etc/nix/netrc file
<ikwildrpepper>
if it does, you could use that for http authentication
<viric>
/etc/nix/netrc? I didn't know of this. I'll check.
<ikwildrpepper>
viric: might need 1.12
<viric>
ok
<ikwildrpepper>
we use the netrc stuff to download files from artifactory
<viric>
ikwildrpepper: so do you know conan.io? I'll have to counterargue conan.io
<viric>
Ah ok, great
<ikwildrpepper>
viric: no, haven't seen it. looks like it is c++ focused
<viric>
ikwildrpepper: well, distribution of precompiled binaries
matthewbauer has quit [(Ping timeout: 240 seconds)]
<ikwildrpepper>
viric: we also distribute binaries, but it mostly a pain :D
<viric>
ikwildrpepper: right. We want something focused in distribution of fixed binaries
<ikwildrpepper>
i mean, just the packaging is trivial
<ikwildrpepper>
but the consequences are bad, so many old glibc's out there
<viric>
ikwildrpepper: we do Micros~1 Windows
<ikwildrpepper>
isn't windows better? :)
<viric>
:)
yegods has joined #nixos
<viric>
ikwildrpepper: so, one common counterview is "why do you want so much flexibility? We need clear unique unchanging binaries distributed version-tagged, not flexibility."
justbeingglad has left #nixos []
<viric>
ikwildrpepper: maybe having a cache server which disallows listing is enough as ACL, because the proper GET request will be done only by someone owning a recipe.
mudri has joined #nixos
<viric>
ikwildrpepper: on the other hand, knowing the hash would be enough for someone to get it.
<ikwildrpepper>
viric: to me, counterview is weird
<ikwildrpepper>
they mix up 2 different things
<viric>
ikwildrpepper: distribution and building?
trikl has joined #nixos
<ikwildrpepper>
the flexibility doesn't have anything to do with *archiving* binaries
stallion has joined #nixos
<viric>
ikwildrpepper: well, there is also the idea of enforcing the people to use the archived binaries, and not their own.
eacameron has joined #nixos
<ikwildrpepper>
viric: we do that too at LB
edoput has joined #nixos
<viric>
ikwildrpepper: you do what?
<ikwildrpepper>
viric: we just package those 'fixed' binaries in nix
<ikwildrpepper>
and make people use that
<viric>
ikwildrpepper: fixed-output pieces?
<ikwildrpepper>
yeah, binary tarballs basically
<viric>
ikwildrpepper: ah ok. Are the binary tarballs expressions generated manually or automatically by some CI?
<ikwildrpepper>
no, we add it semi-automatically, when we release (monthly)
<viric>
ikwildrpepper: ok, fine
<ikwildrpepper>
but seems easy enough to automate
<ikwildrpepper>
btw, not saying here that you should use nix, as I don't know the specifics
<viric>
ikwildrpepper: I want to use nix because it's a knowledge I can broadly apply
indi_ has quit [(Remote host closed the connection)]
<ikwildrpepper>
I just know that I use nix as possible, as it seems to be the only sane package manager / build tool that allows me to track everything
<viric>
Exactly. Same here.
<viric>
Easy to understand, tweak, etc.
indi_ has joined #nixos
<ikwildrpepper>
also, if you have trivial builds, other CI's are fine
<viric>
Exactly. We might do that. We have preassure to use TeamCity
stallion has quit [(Ping timeout: 260 seconds)]
<ikwildrpepper>
composing builds is another thing that other CI's are not so good at imho
eacameron has quit [(Ping timeout: 246 seconds)]
<ikwildrpepper>
viric: well, you could use other CI's than hydra, and still use nix
<ikwildrpepper>
that might ease them into it a bit
<viric>
exactly, we head towards that
<ikwildrpepper>
the nice things about other CI's is that the UI's are less confusing to new people, and they usually have a lot more plugins
<viric>
but the company is also looking for a package manager for windows+linux
<viric>
(they are used to "nuget")
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
<viric>
right, plugins for the artifactory storages for example
indi_ has quit [(Client Quit)]
johnramsden has joined #nixos
<viric>
ikwildrpepper: how would you program in Hydra something like the upload somewhere after a build? Using a nix expression for the upload?
<ikwildrpepper>
I think shlevy wrote that for s3 backup once
pie___ has quit [(Remote host closed the connection)]
eacameron has joined #nixos
<johnramsden>
Does anyone know anything about how to set typelibs? I've been unable to figure out how I should properly set them and if tried everything I could find. This is what I'm trying to package: https://github.com/NixOS/nixpkgs/pull/24986
<viric>
ikwildrpepper: ah ok
<viric>
ikwildrpepper: thank you for your consulting :)
<viric>
service
<ikwildrpepper>
viric: personally, I would do everything in steps
<ikwildrpepper>
first get nix more familiar
eacameron has quit [(Ping timeout: 240 seconds)]
<ikwildrpepper>
then perhaps later, experiment with hydra and show off the use-cases needed
<ikwildrpepper>
and make sure nix is actually a good fit
<ikwildrpepper>
otherwise you'll keep on struggling
simukis has joined #nixos
takle has joined #nixos
<viric>
ikwildrpepper: well, it's a struggle between tracked flexibility and centralised rigideness, and I'm asked "Why do you need flexibility? Drop it"
eacameron has joined #nixos
ambro718 has quit [(Ping timeout: 260 seconds)]
eacameron has quit [(Ping timeout: 260 seconds)]
simendsjo has joined #nixos
<viric>
ikwildrpepper: so whatever benefit I want (that nix provides) I'm asked "but why do you need this?", in the sense of "you are overcomplicating"
<ikwildrpepper>
viric: reasonable question, I guess
<viric>
They come from a world that committed produced binaries to vcs repositories.
stallion has joined #nixos
trikl has quit [(Ping timeout: 246 seconds)]
goibhniu has joined #nixos
trikl has joined #nixos
stallion has quit [(Ping timeout: 255 seconds)]
Guest93183 has quit [(Quit: Lost terminal)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] MP2E pushed 1 new commit to master: https://git.io/vHOjq
<NixOS_GitHub>
[nixpkgs] Ma27 opened pull request #26083: programs.zsh: rename oh-my-zsh to ohMyZsh (master...rename/oh-my-zsh) https://git.io/vHOjS
NixOS_GitHub has left #nixos []
<MichaelRaskin>
viric: if people are OK doing whole-image builds OpenWRT-style and committing the result, maybe Nix is not a good fit. Otherwise, the point of Nix is probably deduplication while shipping all the dependencies
<viric>
MichaelRaskin: but somehow there is the desire to ensure that people are using that-specific-binary as dependency, built by a controlled CI and not your own compiler
yegods has quit [(Remote host closed the connection)]
<MichaelRaskin>
Well, if you don't give people source Nix expressions by default, just store paths for nix-store -r …
<viric>
MichaelRaskin: and then using a package manager well integrated. Imagine Microsoft "nuget". You have your nuget server and projects just refer to "nuget dependencies" (name-version match)
<viric>
and Visual Studio kind of "just works"
<MichaelRaskin>
Is the software Windows-only?
<viric>
of course, "name-version" does not allow for much specification of compiler, CRT libs, achitecture, debug/release, ...
<viric>
MichaelRaskin: not windows-only, but these are the main users.
<MichaelRaskin>
Horrible truth: include build timestamp into build ID, and store the flags in the database, and your build version will technically be able to specify all the needed stuff
<viric>
MichaelRaskin: quite that.
<MichaelRaskin>
We-ell, if the code assumes that Windows codepaths are compiled with VS versions of compilers, Nix may be a truly hard sell. You may be in for more fun than macOS people…
<viric>
I have a full set of nix expressions that build pieces with VS compilers, from vs2008 to vs2017
<viric>
out of nixpkgs
<viric>
I have nix working for our projects
<viric>
Full dependencies
<viric>
but they think that something less flexible and "more popular" will be better
<ikwildrpepper>
viric: if you ever figure out a way to compile c/c++ binaries in newer nixpkgs that will run on older glibc's, please let me knwo
<viric>
And they are ready to accept not having the nice features of using nix.
yegods has joined #nixos
<viric>
ikwildrpepper: where will those old glibcs come from? other distros?
<ikwildrpepper>
yeah
yegods has quit [(Remote host closed the connection)]
<viric>
ikwildrpepper: I'm not likely to play that game anytime soon
<ikwildrpepper>
:D
<ikwildrpepper>
bummer, thought I almost tricked you into that :D
<tomjaguarpaw>
In particular is it known that essentially all Lua packages are broken?
<tomjaguarpaw>
I'm rather confused
yegods has quit [(Ping timeout: 240 seconds)]
Wizek_ has joined #nixos
<johnramsden>
Does anyone know anything about how to set typelibs? I've been unable to figure out how I should properly set them and if tried everything I could find. This is what I'm trying to package: https://github.com/NixOS/nixpkgs/pull/24986
<tomjaguarpaw>
How can I find the path than an expression is installed at?
<nixy>
tomjaguarpaw: If the package is in your environment, you should be able to check either /run/current-system/sw or ~/.nix-profile and follow the symlinks to the nix store
peacememories has joined #nixos
<simpson>
Okay, I've reproduced my Docker problem on vanilla NixOS.
<simpson>
So I guess that I just am not grokking how to Docker with Nix. Oh well!
sziszi has quit [(Read error: Connection reset by peer)]
<jophish>
tomjaguarpaw: another thing you could do is run `nix-instantiate expr` which will produce the .drv file for that expression. You can look for the "out" attribute in that file
<jophish>
pretty-show acts as a decent formatter for .drv files I've found
<tomjaguarpaw>
lua is doing my head in
<jophish>
tell me about it, dealing with torch again at the moment
<tomjaguarpaw>
Do you find that Lua stuff works at all in Nix? I don't understand how to get it to see modules.
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
jacob_ has joined #nixos
jacob_ is now known as Guest38005
mpcsh has quit [(Quit: THE NUMERICONS! THEY'RE ATTACKING!)]
mpcsh has joined #nixos
<LnL>
tomjaguarpaw: I think somebody was working on that, you have to set LUA_PATH or whatever it is yourself
<tomjaguarpaw>
I'm trying ...
<tomjaguarpaw>
Not having much success ...
bkchr has quit [(Ping timeout: 240 seconds)]
<tomjaguarpaw>
Trouble is I'm going to visit Docker in 10 minutes and if I can't get Nix to work properly I'm going to feel like a failure :(
<tomjaguarpaw>
Yeah that's linked from the issue I li
<tomjaguarpaw>
nked earlier
<tomjaguarpaw>
Does LUA_PATH hide LUA_CPATH? I'm so confused
<avn>
tomjaguarpaw: no, one for lua sources and one for .so
matthewbauer has quit [(Remote host closed the connection)]
<tomjaguarpaw>
If I set LUA_CPATH then it can find my .so. If I then *also* set LUA_PATH it looks for the .so in the LUA_PATH and seems to ignore LUA_CPATH.
<tomjaguarpaw>
So confused
arian_ has joined #nixos
yegods has quit [(Remote host closed the connection)]
phreedom has joined #nixos
<Mateon1>
Am I misunderstanding something?: nix-shell -p python27Full -p python27Packages.numpy; python -c 'import numpy' # fails with ImportError
<xificurC>
since #nix seems rather empty I'm not sure where to ask but here goes - I used the curl | sh way to install nix. Today I did `nix-channel --update` and `nix-env -u` which created the 2nd generation. Here `nix-env --help` doesn't work anymore. Switching back to generation 1 I see the man page was at ~/.nix-profile/share/man/man1/nix-env.1.gz and it'
<xificurC>
s not there anymore. What goes?
johnramsden has quit [(Quit: Leaving)]
<LnL>
I don't think #nix is a thing, this is the channel for all nix related things
<MichaelRaskin>
Hm. Interesting. I guess nix-env -iA nix.man should help…
<jophish>
tomjaguarpaw: in my lua shell (with the cjson package working) I have: LUA_CPATH=/nix/store/zmryx9cjdlbdy894zi7ykz3w7d7jl0q8-lua5.1-cjson-2.1.0/lib/lua/5.1/?.so and LUA_PATH=/nix/store/zmryx9cjdlbdy894zi7ykz3w7d7jl0q8-lua5.1-cjson-2.1.0/lib/lua/5.1/?.lua
<MichaelRaskin>
This is an argument contra, not pro, right?
<clever>
ah, this is where the extra <para> comes from
bennofs has quit [(Quit: WeeChat 1.8)]
<gchristensen>
MichaelRaskin: IMO, PHP has some of the best, most thorough, and approachable docs of any programming language
<clever>
which just helps keep the plague alive :P
<nixy>
One of the things that got brought up when people were lamenting docbooks was that you can use pandoc, but how well does that work?
<danbst>
why not state explicitly, that nixpkgs source _is a_ documentation? It rarely contains loops, so much of source theoretically can be described in triple store. Probably, "Literal Nix" would be a nice addition
<clever>
yeah, i often feel that docs can be out of date or just wrong, and the source cant lie
<MichaelRaskin>
«Can't» is a strong word when you trust the trust
<MichaelRaskin>
And some of our overrides come close
<clever>
just recently, i saw a blog post claiming grub master supported nvme, but a quick grep of the grub source showed no trace of nvme support
<clever>
and #grub confirmed, grub has no real nvme support, it expects the EFI to provide nvme drivers
edwtjo has joined #nixos
darlan has joined #nixos
<danbst>
clever, ? I have a hetzner server on NVMe with grub
darlan has quit [(Client Quit)]
<clever>
danbst: the EFI must support nvme then, for it to boot
<clever>
but if you slap an NVME->PCIE adapter into an older desktop, your out of luck
<danbst>
actually, I have 2 NVMe's in raid, and grub on top of raid
<danbst>
so maybe you're right
<nixy>
I believe source is useful for finding answers, but source and documentation are fundamentally separate
<clever>
if the EFI doesnt support it, you must put /boot on a different media
<clever>
then the linux nvme driver lives in the initrd on /boot, and grub wont care
<nixy>
I don't think Nix's documentation issues are with references for experienced users, it is getting people over the hump and into the ecosystem
stanibanani has joined #nixos
eacameron has joined #nixos
<clever>
nixy: part of it is just convincing people that you cant just install gcc and openssl.dev, and expect it to work
<nixy>
Yeah that too. Thats particularly troublesome for python libraries
<gchristensen>
nixy: agreed
<gchristensen>
re: "the hump"
<clever>
somebody else recently was confused as to why pkgs.writeTextFile couldnt write to his home directory
<nixy>
The best thing I have seen to get over the hump has been the "Nix pills" series, but those seem to target Nix itself rather than the whole Nix ecosystem
<clever>
yeah, and there are 3 fairly seperate chunks, nix, nixpkgs, nixos
eacameron has quit [(Ping timeout: 240 seconds)]
<clever>
i have made an example of what can be done with just nix alone, if you somehow provide it pre-compiled binaries (i cheated and used nixpkgs)
<MichaelRaskin>
I dunno, for me The Thesis was just right to get the entry
<gchristensen>
MichaelRaskin: you know you're exceptional <3
<nixy>
I would like to see the manual split into tutorial reference and how-tos, and be made a little more approachable
<MichaelRaskin>
And regardless of the documentation, I am not sure all our tools have stabilized enough for use by people for whom Nix pills are not enough
luigy has left #nixos []
<MichaelRaskin>
gchristensen: not really special, there are a lot of CS PhDs
luigy has joined #nixos
<gchristensen>
you have a PhD? I didn't know that
<nixy>
MichaelRaskin: What do you mean by "stabilized"? It seems to me that most things in Nix are preserving backwards compatibility and aren't likely to change that
<xificurC>
nixy: sorry I was away for a while, thanks for the PR. I see it got merged but I guess it's not in the channel yet?
<xificurC>
nixy: the man pages fix, to give context
<MichaelRaskin>
Not in the channel
<nixy>
xificurC: I am not 100% on the release cycle, but I think it has to sit for a while before it goes in channel
<LnL>
no that takes some time, depending on how stable master is
<MichaelRaskin>
nixy: well, you think on the level where this stability is true, so do I, but for example remote-building and channel manifest -> binary cache things have changed
<MichaelRaskin>
And there is nix-env -i but you really want to use nix-env -iA
<LnL>
but it's very easy to install a single package from master with nix-env
<MichaelRaskin>
And we consider deprecating nix-env etc to have just nix binary to call with subcommands, because nix-env still needs subcommands
<nixy>
Yeah I would hope that the whole symbolic vs attribute name changes in a way that would probably break things
<LnL>
I think the new nix command always uses attribute names
<MichaelRaskin>
I actually agree with your claim, but I think it is not the whole story
<nixy>
Probably not. I am motivated by the hope that Nix can be a tool for everyone though
<MichaelRaskin>
People who understand Nix in terms where Nix behaviour is really stable and stays compatible are the same people who can (maybe with some complaining) read the manual and the man pages
takle has quit [(Remote host closed the connection)]
<clever>
and some things, like setup-hook.sh, i havent seen mentioned anywhere in the documentation
<MichaelRaskin>
I feel the things that are needed for the next level of Nix users to come and use Nix will still change a few times, so trying to make documentation accessible to these users without explicitly admitting they will have to relearn is a bit of trickery towards them
<clever>
MichaelRaskin: cmake and unzip are missing, and it doesnt really say how setup hooks work
abrar has joined #nixos
eacameron has quit [(Ping timeout: 272 seconds)]
<MichaelRaskin>
gchristensen: Yes I do, and I have some publications of pure-math variety and some of quite applied «summarising practical experience» variety. Actually getting the PhD is just a part of choices to make, but knowing quite a bit of math is a part of me for sure.
<clever>
and a quick search says there are 64 setup-hook.sh's in nixpkgs
<MichaelRaskin>
clever: ah, you mean that these are not documented one-by-one
<clever>
it would be better to make them easy to find, and explain how they work, then to try and document the side-effects each one does
<gchristensen>
MichaelRaskin: I think I knew about your interest in / knowledge of math. cool :)
<MichaelRaskin>
Although also too briefly
<MichaelRaskin>
I have some completely abstract math publications (and some completely applied stuff at European Lisp Symposium, too)
<clever>
MichaelRaskin: yeah, no mention that putting your package into buildInputs makes the hook get sourced
<clever>
which can apply mutations to the stdenv
* gchristensen
should go read up on setup hooks
eacameron has joined #nixos
<clever>
gchristensen: as the stdenv goes over everything in buildInputs, it will check for a $foo/nix-support/setup-hook, and if found, it will source it into tbe bash doing the build
<gchristensen>
hey that is really cool
<clever>
it can then modify bash variables, and add new bash functions
<MichaelRaskin>
Didn't you know how cmake support works??
* gchristensen
thinks through what horrible things he can do
<MichaelRaskin>
And even override the existing bash functions!
<danbst>
the problem with nix-env I often had, is unability to reproduce original arguments to nix-env. Say you install some stuff like LnL showed above, after a month or so you revisit your profile and can't remember what expression have produced that "nix" binary. Or, you've written some expression and nix-env -i'd it, and in a month you have a working program, but you forgot where have you written the expression
<clever>
in the case of cmake, it will replace the configurePhase with a cmakeConfigurePhase
<gchristensen>
MichaelRaskin: :) I knew adding it to buildInputs caused stdenv to behave differently, but didn't know the exact mechanics behind it :)
<MichaelRaskin>
danbst: which is why I use nix-build for my profile expressions
<clever>
in the case of unzip, it will magicaly give the default unpackPhase zip support
<danbst>
in my case, it was written in .config/nixpkgs/config.nix
takle has joined #nixos
<MichaelRaskin>
This is acutally a part of what I hope to stabilize with time
<gchristensen>
same
<danbst>
MichaelRaskin: as for me, saving the source of expression alongside the derivation is a way to go
<clever>
danbst: yeah, thats why i always embed such things in config.nix
<MichaelRaskin>
We teach nix-env to newcomers, then at the next stage of initiation it turns out everyone who is in for real uses nix-build and nix-shell and avoids nix-env
<clever>
i also see people using nix-env to test new packages
<gchristensen>
MichaelRaskin: I think newcomers are willing to do it the real way anyway
<clever>
which just spams the profile and leaves 100's of GC roots behind
eacameron has quit [(Ping timeout: 245 seconds)]
<gchristensen>
I also hope we can push past single user installs
<MichaelRaskin>
Some are willing, some are not, I guess
<MichaelRaskin>
But if anyone wanted to internalize how Masons or Scientology keep straight faces while teaching the newcomers beliefs they have abandoned for themselves, you are welcome
<gchristensen>
:D
<nixy>
So I hear nix-env catch a lot of flak, but the issues with it aren't entirely clear to me
<clever>
there is also the weird part about how nix-env will obey $NIX_PATH, but never actualy try to use it by default
<LnL>
I don't see a problem with nix-env unless you start installing libraries there
<clever>
nix-env is the only tool that directly understands channels
<MichaelRaskin>
It is easy to have stuff pinning different glibc versions and not notice it, it doesn't save the exact expression installed…
<LnL>
the fact that I think python and perl packages look in NIX_PROFILES should be removed IMHO
<MichaelRaskin>
I really hope that the current (3rd?) iteration of channels is either not hte last one, or just will fall out of use
<clever>
i currently have 7 glibc's in my ~/.nix-profile/
<gchristensen>
MichaelRaskin: say more?
<clever>
LnL: that reminds me, kdenlive refuses to run via ./result/bin/kdenlive
<MichaelRaskin>
Well, the fact that it is tied together with nix-env (and nix-env needs a rethinking) should be enough, no?
<clever>
LnL: it relies on propagatedUserEnv stuff, and also conflicts with older kde stuff from another version of itself
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edwtjo pushed 2 new commits to master: https://git.io/vH3n7
<NixOS_GitHub>
nixpkgs/master 0c61530 Edward Tjörnhammar: kodi: 17.2 -> 17.3
<NixOS_GitHub>
nixpkgs/master 1f98e47 Edward Tjörnhammar: kodiPlugins.exodus: 3.0.5 -> 3.1.13
NixOS_GitHub has left #nixos []
<clever>
LnL: i tried to run the latest kdenlive in a nix-shell one day, and tracked the problems down to an older kdenlive already being in nix-env
<LnL>
clever: yeah :/ alltho that's a bit more explicit
<clever>
that defeats the entire point of nix
<MichaelRaskin>
Also, I think that currently adding a channel just doesn't have a hook to prompt you to add a binary cache, too
justelex has joined #nixos
<clever>
MichaelRaskin: channels can specify a binary cache
<MichaelRaskin>
What happens when you add one as a non-root user?
<clever>
MichaelRaskin: a channel is just an http dir with 2 files, binary-cache-url and nixexprs.tar.xz
revtintin has quit [(Ping timeout: 260 seconds)]
<clever>
yeah, if you lack root, then nix will probably either ignore the extra binary cache, or complain about its signature
<MichaelRaskin>
I think you are wrong, when I was hand-mirroring channels I needed 5 files or so
<MichaelRaskin>
(I do know it is just a small amount of static files)
<clever>
i'm going off info i got back in 2015
<clever>
when i setup my own channel for the rpi
hellrazo1 has joined #nixos
hellrazor has quit [(Ping timeout: 246 seconds)]
<gchristensen>
erm yeah you specify the binary cache in the channel
<LnL>
clever: but I think NIX_PROFILES is a lot worse, since global libraries will get loaded implicitly
<gchristensen>
man nix-channel -> "CHANNEL FORMAT" -> "binary-cache-url"
<clever>
LnL: yeah, i feel that a well written package will use absolute paths for everything, and should run with ./result/bin/foo with nothing above the base-system installed
<LnL>
exactly
<MichaelRaskin>
gchristensen: it can specify the binary cache, but there is also a question of what nix-channel will say to the user…
<clever>
gchristensen: for example, nix-env uses a copy of buildEnv, running on its own perl (the one nix was built against, not the one in the current nixpkgs)
sziszi has joined #nixos
<clever>
and that buildEnv can be out of sync with the buildEnv in nixpkgs
<clever>
luigy: what part of it do you want to override?
<luigy>
I tried something like this but no good (darwin.apple_sdk.override (drv: { sdk = my_sdk })).frameworks.WebKit
<clever>
ah
goibhniu has joined #nixos
<clever>
because frameworks and sdk are in the same let block, its imposible to override sdk
<clever>
you have to directly modify that file
<luigy>
ohhhhhhhhhhhhhhhh
<clever>
it would have to be rewritten some to allow such an override to work
<luigy>
I see
<luigy>
thans clever
<clever>
the only thing .override can affect, are the arguments passed in on line 1
<clever>
and while you could in theory use .overrideDerivation to change the installPhase on frameworks, you would have to copy/pasta the entire installPhase string, and change every ${sdk} in it
<clever>
simpler to just change the file directly
<clever>
but you could copy the whole file to your project, then change it, if you want to avoid modifying nixpkgs
<gchristensen>
the `${}` bit? I'm looking for a wide variety of nix syntax to test against (and I'm not sure I can fix the `${...}` bit here :P
proteusguy has quit [(Remote host closed the connection)]
mudri has joined #nixos
<gchristensen>
_ideally_ I'd be able to actually syntax highlight _all_ of nixpkgs and diff it to after my changes but I can't figure out how to do that with https://github.com/wmertens/sublime-nix
<clever>
gchristensen: but it all seems to highlight right
<gchristensen>
yeah, now I need to make sure I don't break it :P
<gchristensen>
github's syntax highlighter is regex based
<xificurC>
nix-env -qaP | grep will do for now, thanks
orivej has quit [(Read error: Connection reset by peer)]
<LnL>
xificurC: no, I generally use the tab completion of nix-repl, if you want a search tool check out nox
orivej has joined #nixos
<nixy>
Nox does caching too, so the searches should be faster
<LnL>
nox hello => hello-2.10 (nixos.hello) A program that produces a familiar, friendly greeting
<gchristensen>
xificurC: `-P` takes a regex
<gchristensen>
xificurC: `nix-env -qaP '.*foo.*'
<xificurC>
gchristensen: that's way worse than `nix-env -qaP | grep foo` or defining `function nixgrep { grep -E "${1}" <(nix-env -qaP); }`
<gchristensen>
ok
<nixy>
gchristensen: Takes a basic regex, which is an important distinction
<gchristensen>
sure
<gchristensen>
unix utilities are good yes :)
Sufixx_ has joined #nixos
<xificurC>
gchristensen: but thanks :)
Sufixx_ is now known as Sufixx
<Sufixx>
hi, I've just configured postfix with very basic settings (enable = true, setSendmail = true;), but attempt of sending mail root -> root ends up with error: Cannot open mailer: No such file or directory
<xificurC>
so the manual describes how to build and test a package when adding it to all-packages.nix. What if I have a local git repository that has a default.nix and I want to try to build that?
<LnL>
nix-build looks for a default.nix so you should be able to just build it
<gchristensen>
LnL: cool, I'll try that
<xificurC>
$ nix-build
Itkovian has joined #nixos
eacameron has joined #nixos
<xificurC>
I guess I want to use stdenv from <nixpkgs>?
ertes has quit [(Ping timeout: 246 seconds)]
nslqqq has quit [(Ping timeout: 260 seconds)]
Infinisil has joined #nixos
<xificurC>
is this a valid request? Me having a package somewhere locally and wanting to use it with packages in <nixpkgs>? Building and installing it from a local checkout of a single package
<LnL>
what does the expression look like?
<dash>
xificurC: the magic words are: "with (import <nixpkgs> {});"
<dash>
xificurC: then you can use stuff from nixpkgs in your .nix file after that
<xificurC>
think of an in-house business project that won't go to the official channel but I don't have a separate channel where I could put that
<xificurC>
dash: will try that
<dash>
xificurC: the other way to go is "let pkgs = import <nixpkgs> {}; in ..."
<dash>
xificurC: and refer to pkgs.foo for stuff in nixpkgs
<xificurC>
phone
<Infinisil>
How am I supposed to install Emacs packages? Should I use nix for that or can I use package-install within emacs?
<LnL>
an expression can also return an expression, that makes it a bit more flexible since could then can override the arguments with, eg. nix-build --arg nixpkgs /path/to/nixpkgs
takle has quit [(Remote host closed the connection)]
<dash>
Infinisil: yes to both I think :)
takle has joined #nixos
<Infinisil>
I'm asking because I'm trying to use rust+racer in spacemacs, but I can't get any completion whatever I do
proteusguy has joined #nixos
<Infinisil>
Even though `racer complete std::` outputs stuff, paths set up correctly, company mode enabled, just no pop up with completions
earldouglas has quit [(Quit: leaving)]
<Infinisil>
So there aren't any problems to using emacs in nixos? Then it would be purely an emacs problem and asking in an emacs chat would be better :P
<dash>
Infinisil: how's racer being inoked? flycheck?
<dash>
er invoked
greymalkin has quit [(Ping timeout: 260 seconds)]
<Infinisil>
I don't know about flycheck
<Infinisil>
Is it required?
<dash>
i'm asking you
cpennington has joined #nixos
<Infinisil>
Ah I'm only trying to get the stdlib completion working for now
<dash>
oh it does completion
takle has quit [(Ping timeout: 240 seconds)]
<Infinisil>
racer does completion yeah
<dash>
well, find out how emacs is invoking it, see if its the right path etc
<dash>
well, find out how emacs is invoking it, see if its the right path etc
<gchristensen>
LnL: I couldn't make that work
stanibanani has left #nixos []
<Infinisil>
dash: I can even see the output of the racer command in the racer-debug window, which outputs what it should, but no pop up inline
<Infinisil>
Oh well I'll ask in #emacs
matthewbauer has joined #nixos
<LnL>
gchristensen: just a hunch, pretty sure I've never even even opened a tmbundle before
cpennington has quit [(Ping timeout: 240 seconds)]
<dash>
Infinisil: ok yes at this point i doubt it's a nix-specific problem.
nslqqq has joined #nixos
<Infinisil>
Yeah
Infinisil has quit [(Quit: brb)]
ilyaigpetrov has joined #nixos
Infinisil has joined #nixos
<gchristensen>
I love the "attrset-for-sure" thing papering over how the format is lacking
matthewbauer has quit [(Ping timeout: 240 seconds)]
<clever>
dash: for example, if the editor knows that something is an attrset, can it select the entire attrset and cut/paste without ever causing a syntax error?
<gchristensen>
github uses ^
<gchristensen>
and I'm specifically trying to fix github's rendering. emacs handles that gist just fine
<LnL>
oh I didn't know that was also an issue
<trikl>
If a derivation created with mkDerivation has a curl and curl.out as inputs and dontStrip and dontPatchELF are set, should the generated binary not somewhere have the path to curl in it?
<gchristensen>
xificurC: hmmm I would recommend not using the `builder` option, but instead overwriting the `buildPhase` and `installPhase`
<LnL>
you probably don't want to override the builder from the stdenv
<rvolosatovs>
hi. Where can I find any guidelines regarding multihead setup? I want to use a 4k screen with my x260 and `bspwm`. Tried putting `xrandr` command in services.xserver.sessionCommands, but that does not seem to work too good with the WM.(I only get 1 desktop and have to manually add new ones)
Raimondii has joined #nixos
<LnL>
that sets up a bunch of stuff for the build environment (like ld) so it works properly
<xificurC>
I'm using the manual, which does override the builder
<gchristensen>
heh, we were just talking this morning about how the manual is wrong to do that
<xificurC>
so the only official documentation is teaching bad habits :(
<gchristensen>
evidently so
<LnL>
that's not a good example unless it's specifically about the builder
<clever>
xificurC: also, your cd'ing into $src (which is read-only) and then doing gcc -o as-table, which writes to the current dir
Raimondi has quit [(Ping timeout: 268 seconds)]
Raimondii is now known as Raimondi
<clever>
xificurC: and -static doesnt work out of the box, because nix disables a lot of things needed for static linking
<xificurC>
-static I forgot to remove, the cd thing, ok thanks
<clever>
it just compiles a single c file directly to $out/bin/nix-repl
<trikl>
clever: What I don't understand is how they didn't find the issue of Xonotic not downloading maps earlier, after all there seems to be a stable userbase of nixos xonotic players
<rvolosatovs>
is anyone here using multihead on nixos? can some maybe give an example configuration?
<clever>
trikl: either somebody did something impure and installed libcurl globally, or they didnt test things well enough when making the latest update
<trikl>
Also, any harm in using patchself --add-needed ${curl.out}/lib/libcurl.so instead of resetting the rpath?
<clever>
rvolosatovs: i just left xorg all on defaults, and the login page is a bit messed up, but xfce does all the multi-monitor setup after login
<trikl>
From patchelf's documentation, it's not clear to me whether --set-rpath adds to the rpath or resets it, and if it does reset it, how to refer to the previous existent rpath
<clever>
trikl: --print-rpath i believe
<rvolosatovs>
clever: so I should go for xfce+bspwm?
<arian_>
I have a build step in a package I'm trying to package that requires sudo
<arian_>
can i do sudo in a derivation?
<gchristensen>
no
<arian_>
:/
<clever>
arian_: no, all builds must be done without root
<arian_>
darnit
<gchristensen>
why does it want to be sudo'ing?
eacameron has quit [(Remote host closed the connection)]
<arian_>
okay I guess I need to do the same hack as dockerTools then
<arian_>
I am writing a helper script to create ACI containers using acbuild
eacameron has joined #nixos
<arian_>
but the "Run in container" step (that for example can set up users) requires sudo
<clever>
is it modifying metadata on the ext4 image?
erasmas has joined #nixos
<arian_>
there is nothing with an ext4 image
<clever>
where is it storing the ACL's?
<arian_>
ACI*
<arian_>
not ACL
rvolosatovs has quit [(Quit: WeeChat 1.7.1)]
<arian_>
"App Container Interface"
<arian_>
it is
<arian_>
:P
<clever>
where does it store that data after gaining root?
<gchristensen>
arian_: are you running nixos? are you using sandboxing?
<arian_>
no special location. just the current directory where the build script is executing
<arian_>
no I'm on Arch linux
<clever>
then it shouldnt need root to do it
eacameron has quit [(Ping timeout: 246 seconds)]
sitwon has quit [(Ping timeout: 240 seconds)]
eacameron has joined #nixos
sitwon has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] unaizalakain opened pull request #26089: xonotic: fix xonotic not finding libcurl at runtime (master...fix-xonotic-libcurl) https://git.io/vH32l
NixOS_GitHub has left #nixos []
bkchr has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<xificurC>
LnL: I now get install: cannot create regular file '/nix/store/mgpbh0x0s8w0k069cxmc2ykk56vp8nyz-as-table-0.1/bin': No such file or directory
<clever>
xificurC: did you mkdir -pv $out/bin/ first?
<LnL>
^ I thought install created the basedir for you
<xificurC>
I just copied the gist :)
eacameron has joined #nixos
Rotaerk has quit [(Quit: Leaving)]
<clever>
-D create all leading components of DEST except the last, then copy SOURCE to DEST
<clever>
LnL: i think thats an optional feature
<LnL>
yeah -D or something?
<clever>
thats what the man page says
<LnL>
I usually just use cp :p
<clever>
same
<arian_>
anybody familiar here with the dockerTools in nixpkgs?
<xificurC>
install -D -m ... dies with the same error, that's interesting
<xificurC>
actually this is the first time I'm using install :D
<luigy>
I remember using at some point something like going from a local path give to the import statement to something that referenced the nixpkgs... so instead of ./there.nix -> <nixpgks>/path/to/there.nix
<xificurC>
ok now that I built, I can `nix-env -i -f.` or something?
derchris has quit [(Ping timeout: 246 seconds)]
hellrazo1 has quit [(Quit: WeeChat 1.8)]
bennofs has quit [(Quit: WeeChat 1.8)]
eacameron has quit [(Ping timeout: 246 seconds)]
<LnL>
yep
<LnL>
you can also look at the result symlink that nix-build created, that will point to $out of the package
<luigy>
Oh I see! the closing bracked should be at the end of the path and not nixpkgs
<UrsidaeJohn>
I've hit one of those issues which seems so basic that I'm sure I must be doing something wrong. When I install a package with nix (ex: `nix-env --install vim`), I can't use it as root or via sudo. So, what am I supposed to do if I need to use elevated permissions?
<dash>
xificurC: there was a more recent one whose link i don't have on hand
<dash>
UrsidaeJohn: install it as root too
eacameron has joined #nixos
<dash>
UrsidaeJohn: or use emacs and have it open files as root for you... ;-)
<UrsidaeJohn>
alright, here's a variant of that, what if I'm in an environment that I created with `nix-shell` and I have only one thing that I need to sudo?
<UrsidaeJohn>
just kill the shell and try again as root?
<clever>
UrsidaeJohn: there is also "sudoedit" which will copy the file to /tmp owned as the current user, and run $EDITOR on it
<UrsidaeJohn>
vim was just an example
<clever>
UrsidaeJohn: then when the editor quits, sudo will copy it back to the original place
<clever>
so the editor itself never gains root
<UrsidaeJohn>
I'm actually trying to do something more complicated
derchris has joined #nixos
<clever>
sudo on some distros will reset $PATH for security reasons
<UrsidaeJohn>
I'm processing live camera feeds, and to manage them I need elevated permissions
<UrsidaeJohn>
so is the issue with sudo? I should find a way to tell it to keep $PATH?
<clever>
for example, on gentoo, sudo resets $PATH to "sane defaults"
<UrsidaeJohn>
I think I know how to do that....
<clever>
but on nixos, it doesnt
<UrsidaeJohn>
this is Ubuntu
iclanzan has joined #nixos
justan0theruser has joined #nixos
ij has joined #nixos
<clever>
Defaults env_keep+=SSH_AUTH_SOCK
<ij>
What does cabal2nix's "--shell" do?
<clever>
UrsidaeJohn: i see this in nixos's sudoers file
<clever>
UrsidaeJohn: a similiar line targeting PATH would probably fix your issue
<UrsidaeJohn>
that what I was just thinking, clever
<UrsidaeJohn>
thanks
bennofs has joined #nixos
<UrsidaeJohn>
I had to do something similar so that sudo wouldn't nuke $PYTHONPATH
bennofs has quit [(Client Quit)]
derchris has quit [(Ping timeout: 240 seconds)]
bkchr has quit [(Ping timeout: 240 seconds)]
bennofs has joined #nixos
bennofs has quit [(Client Quit)]
derchris has joined #nixos
<xificurC>
I have nix and shellcheck installed, how come `du -hs /nix` returns 779M? Is nix or shellcheck that big? Is there some commands I can find that out?
<xificurC>
dash: thanks for the link, will check out
bennofs has joined #nixos
<xificurC>
dash: eh, last post almost 2 years ago, are the posts still valid?
bennofs has quit [(Client Quit)]
<gchristensen>
xificurC: nix pills are very good
<dash>
xificurC: nix language has not changed in that time, some minor features have been added
<xificurC>
ok thanks
bennofs has joined #nixos
derchris has quit [(Ping timeout: 240 seconds)]
derchris has joined #nixos
<pikajude>
i don't think nix has ever had breaking syntax changes
<sphalerite>
How can I find out where the "Dumping a very large path" warning comes from?
<pikajude>
manually
<pikajude>
would be nice if nix printed out the path
<sphalerite>
yeah
<LnL>
node stuff often has that
<pikajude>
one time, i got that warning because a package was importing all of nixpkgs into the store
bennofs has quit [(Quit: WeeChat 1.8)]
<sphalerite>
xificurC: nix, shellcheck and all their dependencies are probably that big.
<pikajude>
"${nixpkgs.path}/foo"
<clever>
sphalerite: simplest thing i can think of is to peek in the /proc/<nix-build/store pid>/fd/ while its dumping it
<pikajude>
terrible
bennofs has joined #nixos
<clever>
sphalerite: that will show all files its opening
<sphalerite>
clever: haha nice
<sphalerite>
or strace it I suppose
<clever>
sphalerite: those symlinks also bend the rules, the symlink can point to a deleted file, but still be opened for reading
derchris has quit [(Ping timeout: 260 seconds)]
<clever>
sphalerite: and /proc/<pid>/root can even cross chroot borders, in both directions
<clever>
need a pid namespace to prevent that from being abused
<ekleog>
hmm, what's the policy of nixos wrt version numbering and security patches? like, if there's a patch for a security vulnerability that didn't trigger a release upstream, should we ship a patched version? if so, with a separate version number from the original?
<clever>
sphalerite: and /proc/<pid>/fdinfo/ shows the seek position within every file, so you could monitor how far along something is in reading a file, without the overhead of strace or pv
<sphalerite>
clever: and exe. That's a hack I saw quite a lot for people getting locked out of their tmux sessions when tmux was upgraded to a newer version (on silly distros) that was incompatible with the previous one
<clever>
sphalerite: ah yeah, ive had to use that for screen on nixos
<clever>
with screen, it will partialy connect, but in a write-only fasion
<clever>
so you can ctrl+c your programs, but your doing so blindly, and you cant disconnect
bennofs has joined #nixos
<sphalerite>
nice...
<clever>
yeah
<sphalerite>
but at least on nixos you still have the old version available unless you've gc'd it
<sphalerite>
idea: /proc should be a GC root
<clever>
yeah
<clever>
it is
<sphalerite>
oh nice
<clever>
nix will scan maps, fd, exe, and environ
<sphalerite>
so you can't even gc it
<clever>
which also means, "sudo nix-store --delete /nix/store/foo" will never work
<bennofs>
clever: how do you use nix-store --delete then?
<clever>
the path is in the arguments of sudo, so its "in use"
<sphalerite>
oh, heh
<sphalerite>
bennofs: without sudo
<clever>
bennofs: without sudo
<bennofs>
oh
<simpson>
Or, if you really insist, $(sudo -i) first.
<clever>
any user on the system is free to delete any path thats not in use
Itkovian has joined #nixos
<simpson>
(I guess on a single-user install?)
<clever>
on a single-user install, you have direct write access to the store, so sudo isnt needed
<clever>
and using sudo in a single-user setup can break things, leading to root owning parts of the store, and now your single-user cant manage it right
<clever>
sudo git can do similiar things
<simpson>
I have a setup at home (which I want to deprecate!) where I have to $(sudo -i -u user_who_owns_nix_store), but that's obviously not common.
<clever>
whenever that comes up, i just convert it into a multi-user setup
<clever>
make a nixbld group, add nixbld{1..5} to the group, chown most of /nix to root, run nix-daemon as root, set NIX_REMOTE=daemon for all users
mizu_no_oto has quit [(Quit: ["Textual IRC Client: www.textualapp.com"])]
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
<bennofs>
clever: what parts of /nix shouldn't be chowned to root?
<gchristensen>
some bits in /nix/var, it is a bit complicated
<bennofs>
I guess per-user profiles?
<clever>
each user should own the directory named after them, in /nix/var/nix/profiles/per-user/
romildo has joined #nixos
<clever>
and the user original installing it needs +w to the default profile
romildo has quit [(Client Quit)]
<clever>
i think nixos has things in pam that manage that
<gchristensen>
profiles/ can be written to by anyone and the profile creates the dir and chowns it
<clever>
drwxr-xr-x 3 root root 52 May 23 09:01 /nix/var/nix/profiles/
<clever>
not on this nixos system, and that would allow anybody to mess with default (roots nix-env) and system (the nixos build)
ThatDocsLady has quit [(Quit: Arma-geddin-outta-here!)]
derchris has joined #nixos
savage_engineer has joined #nixos
endformationage has joined #nixos
jgertm has quit [(Ping timeout: 255 seconds)]
eacameron has joined #nixos
<savage_engineer>
Hi guys, new to nixos here. Is there a way to perform a build action in a package only if on NixOS?
<MichaelRaskin>
Makes no sense per se, as NixOS and Nix-on-other-Linux can share store. So just add a flag nixosSupport to the expression
simukis has joined #nixos
<LnL>
jeeze, the cryptography tests literally take forever on this arm machine
<MichaelRaskin>
(Well, share store as in use each other as binary cache)
<LnL>
why would you want that?
derchris has quit [(Ping timeout: 240 seconds)]
<savage_engineer>
Yeah, this is resultant from the package being a shell script which checks for a program in hardcoded directories. I'm currently replacing a string of the form `/usr/lib/<binpath>` to the location of the package on nixos
<savage_engineer>
I have thought about doing a pull request to upstream for it's existence in .nix-profile, but that would be redundant for the majority of users
<clever>
savage_engineer: you should apply the same path replacement even outside of nixos
<clever>
savage_engineer: it should use absolute paths, like ${curl.out}/bin/curl
<savage_engineer>
I currently am using ${pkgs.w3m}, would that not break dependency if they don't have w3m installed through nix or is that not a concern?
<gchristensen>
it would depend on w3m and install it at the same time
<savage_engineer>
ah
<clever>
savage_engineer: nix will automaticaly download w3m
<clever>
without directly installing it in anything
<savage_engineer>
Ah
<clever>
so only your bash script can make use of it
<savage_engineer>
It's an optional dependency for the script
<savage_engineer>
Hmm will have a bit of think and come back to it - just wanted to contribute back to the package so others wouldn't have to use a custom package for the bit of functionality :D
phreedom has quit [(Remote host closed the connection)]
hellrazor has joined #nixos
<savage_engineer>
Thanks!
derchris has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] abbradar pushed 1 new commit to staging: https://git.io/vH3F4
<NixOS_GitHub>
nixpkgs/staging e7fa622 Nikolay Amiantov: polkit: add patches from Fedora...
NixOS_GitHub has left #nixos []
<Infinisil>
What would be the best way to implement like a power-save mode in nixos? It would turn off bluetooth, turn down the backlight, disable certain daemons, lengthen refresh rates, etc.
<clever>
Infinisil: i think that can be done with a systemd target
<Infinisil>
But I'd have to make all kinds of things in my configuration dynamic right?
savage_engineer has quit [(Ping timeout: 246 seconds)]
<Infinisil>
Wouldn't it be possible to always build 2 configs, one with (powersave = true) and one with false, then switch to them to make them active
hiberno has quit [(Quit: leaving)]
<MichaelRaskin>
nesting.children I think
derchris has quit [(Ping timeout: 240 seconds)]
<Infinisil>
Ohh interesting, didn't know about that
<MichaelRaskin>
Well, noone who uses it wants to draw _too_ much attention to it, lest it be declared designed wrong and redesigned in a way that requires to change the invocations
samvher has joined #nixos
<Infinisil>
What's the actual use case for this option?
<MichaelRaskin>
Close enough to what you describe
hellrazor has quit [(Ping timeout: 272 seconds)]
<MichaelRaskin>
I don't remember why I initially pushed for something like that, I think it was something about two incompatible sets of modules for the same hardware
<Infinisil>
Are you using it still?
<samvher>
when I search through nixos packages at https://nixos.org/nixos/packages.html I see a package llvm_4, but when I add it to a nix file in buildInputs I get "undefined variable 'llvm_4'" - why could that be?
<MichaelRaskin>
I am not using mainline NixOS anymore
<MichaelRaskin>
samvher: did you also add it to parameters of thje file?
hellrazor has joined #nixos
<Infinisil>
MichaelRaskin: Care to elaborate?
<MichaelRaskin>
Infinisil: my PID1 is sinit, and the process that spawns BIND and CUPS and nix-daemon is actually Steel Bank Common Lisp instance
<samvher>
MichaelRaskin: I don't understand what that means :) I'm not experienced with writing nix expressions
<samvher>
I tried adding it to my /etc/nixos/configuration.nix as well but that also didn't work
<Infinisil>
MichaelRaskin: I see, well if it works for you :)
derchris has joined #nixos
<MichaelRaskin>
samvher: Are you familiar with any programming language?
<samvher>
MichaelRaskin:
<samvher>
yes
<MichaelRaskin>
Infinisil: well, systemd super-does-not work for me
<MichaelRaskin>
{stdenv}: stdenv.mkDerivation {…} is a function definition
sellout- has quit [(Quit: Leaving.)]
<MichaelRaskin>
all-packages.nix inspects the function and passes it the matching parameters from the package list
sellout- has joined #nixos
<MichaelRaskin>
But the body of the function doesn't see the full package list, only the parameters
<samvher>
hm ok. I added ghc to buildInputs as well and that *did* work
<MichaelRaskin>
Maybe it was already among parameters?
<samvher>
I don't see it anywhere, where are the parameters in the expression?
<clever>
samvher: usualy on line 1, can you post the file in a gist?
<Infinisil>
MichaelRaskin: Why doesn't systemd work for you?
derchris has quit [(Ping timeout: 260 seconds)]
<MichaelRaskin>
Because it is written by people who think Apple produces usable software
<MichaelRaskin>
Infinisil: and I have a list that it just does plain wrong, but most of these things I can ignore. Like, it fails logging forever, it fails session lifetime forever etc
<clever>
samvher: ah, then your getting stuff from sept of 2016, and llvm_4 was added in march of 2017
<samvher>
ah ok. I ran nixos-rebuild switch --upgrade just now, I assumed that brought me to the latest version
eacameron has quit [(Remote host closed the connection)]
<clever>
you need to manualy change the channel to update to the next stable release cycle
<clever>
nixos-rebuild --upgrade only updates within the channel
<Infinisil>
MichaelRaskin: That's hard to believe, you mean systemd does something you don't want because it isn't properly configured or it actually has relevant bugs?
<samvher>
alright, that makes sense. I will do that
pie_ has quit [(Changing host)]
pie_ has joined #nixos
<MichaelRaskin>
Infinisil: systemd has a few actual bugs that require a minor redesign here and there
<MichaelRaskin>
They don't care much about fixing them, but these are obviously bugs
<clever>
MichaelRaskin: i have also managed to crash pid1 once, by trying to systemd-nspawn against a fuse rootfs, and the fuse daemon crashed
<samvher>
clever: thanks
eacameron has joined #nixos
<MichaelRaskin>
clever: in this case I can believe they could eventually fix it
<MichaelRaskin>
Crashing systemd is not as hard as it should be, but I think they have to fix that
derchris has joined #nixos
<Infinisil>
Oh hey clever, there is the option networking.usePredictableInterfaceNames, which is enabled by default, this might be equivalent t the boot.kernelParams = [ "net.ifnames=0" ]
<danbst>
savage_engineer: actually, there is a way to build only on nixos
<danbst>
with import <nixpkgs> {};
<danbst>
stdenv.mkDerivation {
<danbst>
name = "build-only-on-nixos";
<danbst>
buildCommand = ''
<danbst>
grep -iq nixos /proc/sys/kernel/version || (echo "Can be built only on NixOS!" && exit 1)
<danbst>
touch $out
<danbst>
'';
<danbst>
}
<MichaelRaskin>
This will interact with Nix sandboxing and binary caches in a really interesting way
<clever>
danbst: but the instant a binary cache comes into play, it just downloads the pre-built copy and ignores that rule
sellout- has joined #nixos
<danbst>
MichaelRaskin: I specifically tested this works with sandboxing
<clever>
then you can insert things at 15, to squeeze between 2 things that would have been too close previously
<Infinisil>
clever: Swift 1 and 2 used this model, from 0 to 200 or something, but as you can see they abandonned it
<clever>
danbst: http-proxy depends on http-proxy', and http-proxy' depends on http-proxy, you need to use the super more somewhere, but this expression looks a bit off, so i'm not sure where
<MichaelRaskin>
danbst: but then you can evaluate on NixOS and copy the resulting derivation name to non-NixOS and run nix-store -r
<clever>
Infinisil: i first learned that method of priority numbering when i was programming basic on the c64
<Infinisil>
But then again, Haskell, the language of the many operators, has survived with these 9 priorities just fine for decades
wak-work has quit [(Ping timeout: 240 seconds)]
<clever>
Infinisil: the c64 used those numbers just to tell the editor where to insert a line within the program, and it was completely incapable of shifting lines up/down at a later point
<Infinisil>
clever: Ah yes I've seen that before
<clever>
so you had to leave large gaps in the numbers, for future planning
georges-duperon has quit [(Ping timeout: 240 seconds)]
<srhb>
super.pkgs ?
<ben>
it's not even a direct dependency D:
<clever>
probably super.haskell.lib.dontCheck
Acou_Bass has left #nixos []
<clever>
both super and self are variants of the pkgs set
<clever>
so you dont need a .pkgs on them
<srhb>
Funnily, both things work
<srhb>
Different hashes though
samvher has quit [(Ping timeout: 260 seconds)]
peacememories has joined #nixos
<srhb>
Does the self: super: idiom always mean the same thing? Like, in overlays supposedly super is the package set "so far" (whatever that means) and self is the final package set. Is that true in the haskellPackages override as well?
<clever>
yeah
<clever>
and i believe haskell did it first, and nixpkgs copied it
<ben>
haha, that does not surprise me
<ben>
I'm starting to feel like people might have been too clever at some point th
<clever>
the haskellPackages part
<srhb>
so I couldn't use selfHS.http-proxy because that would recurse
<clever>
yeah
<srhb>
But self.haskell.lib.dontCheck is fine, because I'm not defining that.
<clever>
correct
<srhb>
Phew, finally that clicked. :-)
Infinisil has quit [(Quit: leaving)]
<srhb>
Thanks!
<clever>
and self.haskell might even be better, if you wanted to allow other overlays to change it
<srhb>
Indeed.
<bennofs>
note that `super` is not *quite* the package set so far
<clever>
another problem though
<srhb>
bennofs: That always did confuse me somewhat.
<clever>
each call to haskellPackages.override, will overwrite the previous
<clever>
so only one set of haskell overrides will ever apply
<srhb>
clever: That sounds.. Problematic.
wak-work has joined #nixos
<bennofs>
we should really add a haskellPackage.extends
<clever>
there is only a single overrides attribute
<bennofs>
that merges the override functions
<ben>
is super.pkgs == super?
arian_ has quit [(Quit: arian_)]
<srhb>
clever: And this breaks silently as well, doesn't it?
<ben>
hmm, can you add overlays on nixpkgs on the fly
<clever>
srhb: you would probably need to use a foldl' to merge a list of haskell overrides (each in their own file) and then make a single overrides to pass to haskellPackages
<srhb>
clever: Yeah
<danbst>
MichaelRaskin: sure, but you still need a NixOS to make the build. Hard requirement
<srhb>
bennofs: I don't understand what that does.
<danbst>
MichaelRaskin: kinda
matthewbauer has quit [(Ping timeout: 260 seconds)]
<bennofs>
srhb: if you use .override (previous: ...), then previous will be an attrset with the arguments that were previously passed to the function. So you can look at previous.overrids, and if it exists, merge the previous overrides with your new override
<pmade>
I'm writing a derivation for a closed-source tool I need to use. Most of the binaries are working after patching them, but one of them dies with "JLinkGDBServer: relocation error: JLinkGDBServer: symbol , version GLIBC_2.2.5 not defined in file libc.so.6 with link time reference" which makes me think it was linked against another version of glibc. Any suggestions?
erasmas has quit [(Quit: leaving)]
<srhb>
bennofs: Ah, okay.
erasmas has joined #nixos
Filystyn has quit [(Ping timeout: 272 seconds)]
<MichaelRaskin>
danbst: well, you can spin a qemu
<clever>
bennofs: oh, never noticed that variant of override
<MichaelRaskin>
You can also build without sandboxing with a bind-mount over parts of /proc
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<ben>
honestly this .override (old: {overrides = self: oldsuper: let super = oldsuper // (old.overrides or (_:_:{})) self oldsuper; in super // { ... }; }); situation makes me wonder where we have strayed from the path
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 closed pull request #26078: 3to2: refactor to use fetchpypi (master...3to2-fetchpypi) https://git.io/vHOKJ
NixOS_GitHub has left #nixos []
Guest95708 has quit [(Ping timeout: 268 seconds)]
Filystyn has joined #nixos
Acou_Bass has joined #nixos
orivej has quit [(Quit: No Ping reply in 180 seconds.)]
eacameron has quit [(Remote host closed the connection)]
justelex has quit [(Quit: Konversation terminated!)]
justelex has joined #nixos
leat has quit [(Ping timeout: 246 seconds)]
eacameron has quit [(Remote host closed the connection)]
justelex has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz opened pull request #26094: nixos/xsession: use graphical systemd user target (master...feature/graphical-user-session) https://git.io/vHst5
<cement>
if I enable virtualbox through my configuration.nix, how do I start virtualbox from the terminal?
<clever>
cement: Virt<tab><tab> in a shell
justelex has joined #nixos
alphor has quit [(Quit: Bye!)]
zraexy has quit [(Quit: Leaving.)]
kampfschlaefer has quit [(Ping timeout: 255 seconds)]
alphor has joined #nixos
kampfschlaefer has joined #nixos
betaboon has quit [(Quit: This computer has gone to sleep)]
eacameron has quit [(Ping timeout: 272 seconds)]
sellout- has joined #nixos
betaboon has joined #nixos
betaboon has quit [(Changing host)]
betaboon has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to master: https://git.io/vHsnH
<NixOS_GitHub>
nixpkgs/master a549596 Maximilian Bosch: programs.zsh: rename oh-my-zsh to ohMyZsh...
<NixOS_GitHub>
nixpkgs/master e1239fd Franz Pletz: Merge pull request #26083 from Ma27/rename/oh-my-zsh...
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Fuuzetsu pushed 1 new commit to master: https://git.io/vHslT
<NixOS_GitHub>
nixpkgs/master 8091869 Mateusz Kowalczyk: SystemdJournal2Gelf.service: new service
NixOS_GitHub has left #nixos []
takle has quit [(Remote host closed the connection)]
takle has joined #nixos
justelex has quit [(Ping timeout: 240 seconds)]
cement has quit [(Remote host closed the connection)]
justelex has joined #nixos
takle has quit [(Ping timeout: 246 seconds)]
eacameron has joined #nixos
GiGa|Laptop has joined #nixos
sellout- has quit [(Quit: Leaving.)]
<GiGa|Laptop>
hello all
<GiGa|Laptop>
I'm trying (again) to upgrade to 17.03 from 16.09 (there are problems with Gnome3 it seems). Do I want to change "system.stateVersion" to 17.03, or is that irreversible?
<clever>
stateVersion shouldnt be changed
<ekleog>
wait, wasn't stateVersion supposed to be bumped up once you were sure you wouldn't go back to an older revision?
<clever>
no
<GiGa|Laptop>
OK, so leave stateVersion at 16.09, OK
<clever>
stateVersion tells nixos what version the state was originaly created in
<clever>
so it can deal with things that break horribly in an upgrade
<clever>
like postgresql
<clever>
it needs to know what the on-disk format of the db is, to pick the right postgres version
<ekleog>
oh, I thought that was for the format of the database nix used
<clever>
nope
<ekleog>
then there's no way to upgrade at some point?
eacameron has quit [(Ping timeout: 255 seconds)]
<clever>
the nix database auto-updates itself and has no undo
<GiGa|Laptop>
Second question, how would I tell SLiM to use the UK dvorak keymap on the login screen?
<clever>
for postgres, you would need to dump the entire db as sql, delete the db files, update the stateVersion, and re-import it again
<LnL>
you can configure the package of postgres if you upgrade the data on disk first
<GiGa|Laptop>
(Figured I'd give SLiM a go, in place of GDM)
<ekleog>
hmm'k, is that mentioned in upgrade notes or...?
<clever>
ekleog: and i believe only 2 things even use stateVersion, postgres, and sshd (it enables some older less secure ssh host key stuff)
<clever>
fixing that ssh config would change the ssh host pubkey, and set off warnings for all ssh clients
<ekleog>
not even having basic docs on how to upgrade looks like an issue to me :/ (that said I don't use pgsql, so just complaining for the sake of complaining)
<clever>
personaly, i run nixos-unstable (so i never have to change the channel) and i just delete stateVersion from the config (so such breaking changes happen sooner)
<GiGa|Laptop>
clever, how often does unstable break for you?
simukis has quit [(Quit: simukis)]
<clever>
GiGa|Laptop: hasnt happened in a long while
<GiGa|Laptop>
clever, I've got that set to dvorak-uk
<GiGa|Laptop>
presumably I specify the same as I wold with loadkeys?
<clever>
not sure, i havent tried changing the layout on nixos
leat has joined #nixos
<betaboon>
hello, is there a way to apply nixos-switch behavior inside of a container without restarting it ?
simukis has joined #nixos
<LnL>
I don't think so
<GiGa|Laptop>
clever, OK thanks.
<GiGa|Laptop>
Going for a bounce, wish me luck!
GiGa|Laptop has quit [(Quit: Leaving)]
pbogdan has quit [(Remote host closed the connection)]
mudri has joined #nixos
bkchr has joined #nixos
georges-duperon has joined #nixos
simukis has quit [(Ping timeout: 240 seconds)]
justelex has quit [(Ping timeout: 258 seconds)]
carlosda1 has quit [(Ping timeout: 255 seconds)]
pbogdan has joined #nixos
alx741 has joined #nixos
alx741 has quit [(Client Quit)]
alx741 has joined #nixos
zraexy has joined #nixos
matthewbauer has joined #nixos
hellrazor has quit [(Ping timeout: 255 seconds)]
hellrazor has joined #nixos
matthewbauer has quit [(Ping timeout: 260 seconds)]
watersoul has quit [(Read error: Connection reset by peer)]
bkchr has quit [(Remote host closed the connection)]
watersoul has joined #nixos
pbogdan has quit [(Quit: ERC (IRC client for Emacs 25.2.1))]
<clever>
gchristensen: took one of my older examples from 2016 and merged it with my kexec trick
<stallion>
i'll just run off of unstable for now - 17.09 isn't too far off
k2s has joined #nixos
<clever>
so i can now create a zfs based nixos install on a remote machine after running ~5 commands, assuming the remote server has dhcp available
<clever>
potentialy in any datacentr
<dash>
clever: nice
<dash>
clever: now you just need a 0day or two and nixos market share can experience a sharp uptick
<clever>
lol
<clever>
dash: as for how its used now, edit both target-config.nix and configuration.nix to include your ssh auth keys, build the tarball, unpack the tar to / on the server (any linux distro works)
<clever>
then /kexec_nixos, ssh back in, run "justdoit", give nixos-install a root pw, and reboot
justelex has joined #nixos
GiGa|Laptop has quit [(Remote host closed the connection)]
marsel has joined #nixos
k2s has quit [(Remote host closed the connection)]
k2s has joined #nixos
Nobabs27 has joined #nixos
k2s has quit [(Client Quit)]
acarrico has quit [(Ping timeout: 268 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ptival opened pull request #26095: proofgeneral: revert bad modification from previous commit (master...proofgeneral-fix) https://git.io/vHs6G
NixOS_GitHub has left #nixos []
phinxy has joined #nixos
phinxy has quit [(Changing host)]
phinxy has joined #nixos
goibhniu has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ambrop72 opened pull request #26096: teensy-loader-cli: Update to latest git. (master...teensy-fix) https://git.io/vHs6j
NixOS_GitHub has left #nixos []
Filystyn has quit [(Quit: Konversation terminated!)]
Filystyn has joined #nixos
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
pie_ has joined #nixos
pie_ has quit [(Changing host)]
pie_ has joined #nixos
GiGa|Laptop has joined #nixos
<GiGa|Laptop>
Seems the best combination I can attain at the moment is SLiM plus XFCE
<GiGa|Laptop>
If I try to use Gnome3 it locks up after login
<GiGa|Laptop>
Are there still problems with gnome3 in 17.09?
<iclanzan>
Has anyone managed to get mkchromecast working in NixOS?
justbeingglad has joined #nixos
GiGa|Laptop has quit [(Remote host closed the connection)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] gebner pushed 2 new commits to master: https://git.io/vHsPa
<clever>
its following nixos-unstable-small, and set to do an armv7 build
<viric>
clever: can you run the https://test.webrtc.org? What do you get at "reflexive", Network?
<clever>
viric: triangle
<clever>
viric: my router is a nixos machine with nat+dhcp+bind
<viric>
clever: It says triangle, so... success. Right?
<clever>
[ WARN ] Could not connect using reflexive candidates, likely due to the network environment/configuration.
<viric>
AH
<LnL>
and the builders, do they run not-os or a regular nix install
<viric>
triangle you mean the triangular danger sign :)
<clever>
the host connectivity was also able to inspect my ipv4 private's, and found my toxvpn
<viric>
clever: same here... I wonder if this test is reliable for anything
Nobabs27 has quit [(Quit: Leaving)]
<clever>
LnL: rasbian with nix set to multi-user mode
<viric>
clever: that's common in webrtc peer-to-peer
<viric>
clever: disable that in the browser if it annoys you :)
<clever>
i can see it actualy helping, if you friend me in toxvpn, it can bypass all nat trouble
<clever>
though toxvpn ignores the normal subnet rules, so it may not detect it
cpennington has quit [(Remote host closed the connection)]
<clever>
oh, and i should see whats ip with ipv6
<viric>
clever: I understand. I'm gathering proof to change ISP
sibi has joined #nixos
<clever>
hmmm, v6 working just fine
<clever>
must have been the server then
<viric>
clever: I think that STUN nat traversal does not work, although stun servers consider my NAT full-cone
Nobabs27 has joined #nixos
hotfuzz_ has quit [(Quit: WeeChat 1.7.1)]
<michalrus>
How can I have a function with default arguments, and call another such function in it, passing some arguments, including when not defined?
orivej has quit [(Read error: Connection reset by peer)]
<LnL>
clever: wut Current: user: Current not implemented on linux/arm
justelex has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #26099: lldb-4: Patch to fix libedit usage on Linux (master...fix/lldb-libedit) https://git.io/vHsD2
NixOS_GitHub has left #nixos []
petar has quit [(Remote host closed the connection)]
<clever>
LnL: where did you see that?
hotfuzz has joined #nixos
<LnL>
that's another test that fails for me
petar has joined #nixos
<michalrus>
Well, whatever, I went with args@{ ... }: (import ./default.nix args).env
<ekleog>
hmm, anyone remembers the latest name of nixos-{assimilate,cannibalize}? it seems to be none of these two, and I can't find documentation about it
<ekleog>
looks great, this kexec! I was starting to think of using qemu in a ramdisk, that's much better!
<ekleog>
thanks :)
<DIzFer[m]>
and nixos-infect ekleog
betaboon has quit [(Quit: This computer has gone to sleep)]
houqp has joined #nixos
<ekleog>
hmm, nixos-infect points to nixos-assimilate as a replacement... :/ (I'm pretty sure I remember some tool where you touched /etc/NIXOS_{CLEAN or something like this} and it'd move all your files from / to /old-root, allowing to slowly move over from one system to the next, but as that's for a clean install other systems also would work :))
<clever>
ekleog: thats nixos-in-place i believe
<clever>
ekleog: but the existing partition layout will persist
hiberno has joined #nixos
<ekleog>
hmm, I think I'll try and do with kexec, as it's closest to what I'm looking for, thanks to the two of you! :)
isidore has joined #nixos
justelex has joined #nixos
jensbin has joined #nixos
isidore has quit [(Client Quit)]
<clever>
ekleog: also beware, the kexec is configured to reboot every hour on the hour, as a safety
isidore has joined #nixos
<clever>
so you have ~4 minutes left, but you can always "shutdown -c" to cancel the ticking clock, and "systemctl stop autoreboot.timer" to prevent it entirely
Guest38005 has quit [(Quit: Lost terminal)]
jensbin has quit [(Client Quit)]
takle has joined #nixos
phreedom has joined #nixos
<ekleog>
oh wow, that's nicely thought!
<ekleog>
if you're doing things so safely, I wondered, how do you manage unattended systems running NixOS? (I guess these are the reason for this safety for kexec)
<clever>
if your server has dhcp, it should come back up after the kexec and let you in
<clever>
but if the network setup is weird, its easy to get wrong, and then you loose all control
<clever>
and without a control panel from the datacenter, you cant remotely reboot
goibhniu has quit [(Ping timeout: 240 seconds)]
<ekleog>
I was thinking of mostly nixos-rebuild test ; then at some point nixos-rebuild [boot-only-once?] ; then nixos-rebuild boot once having checked the new configuration correctly boots
<clever>
nixos-rebuild test, in combination with "shutdown -r +15" will activate it, and then reboot in 15 minutes
<clever>
if you can still access, cancel with shutdown -c
<ekleog>
issue being this boot-only-once (I guess there's no going without at least a remote "reboot" button for cases like kernel panics, but if you have an idea...)
<clever>
and if you loose control, it will reboot itself, and come back with the old config
<clever>
you can configure the kernel to reboot upon panic
<ekleog>
oh that's great
<ekleog>
so only thing missing is the boot-only-once helper, I guess it's possible with grub but don't know exactly how to do it
<MichaelRaskin>
viric: re: WebRTC: with Talky.io I found out that Firefox can route WebSockets via SOCKS5 but not via https proxy with CONNECT
<clever>
ekleog: the grubenv and next_entry stuff is in grub by default
<MichaelRaskin>
A thing that you just happen to learn when running a browser in a separate network space…
<clever>
ekleog: if you add boot.kernelParams = [ "panic=30" "boot.panic_on_fail" ]; then that should make 2 of the fatal situations reboot
<ekleog>
clever: oh yeah just checked it out on my machine :)
<clever>
ekleog: and the grub-reboot program is designed to modify the grubenv for you
<ekleog>
this looks absolutely great! I'll make sure to try all this out, write it up and ping you when done :) (that said, not promising anything soon, the unattended computer I'm planning on running isn't up yet, so it's not top-level item on my todo-list, but pretty high :))
<ekleog>
(that said I personnally prefer /boot ext4, / and /nix btrfs, and maybe as that's for a server I'll split out /var and /var/log ; but it shouldn't be hard to do that using your kexec :))
<ekleog>
(did you consider packaging your kexec in nixpkgs?)
<viric>
MichaelRaskin: well, it's normal for websockets. But the media doesn't go through websockets, only signaling
<viric>
MichaelRaskin: weird for the https proxy problem
<clever>
ekleog: you can either do the install manualy over ssh (thats how the kexec was previously used), or edit the justdoit script to do btrfs instead
<clever>
ekleog: i have also been considering putting this kernel+initrd into /boot, and giving it its own grub entry, via a nixos option
<clever>
ekleog: then you can just pick the installer from grub, and do recovery of any scale
<MichaelRaskin>
viric: actually, I think that the media will work fine with SOCKS5 by SSH to a server with public IP
<ekleog>
hmm, I was thinking maybe that'd be a bit too much as it's in my mind mostly useful for installing, but you're right there's also a use case for an option!
<MichaelRaskin>
SOCKS5 vs CONNECT difference fo XMPP-over-WebSockets did surprise me
<viric>
MichaelRaskin: socks5 over ssh won't do udp
<clever>
ekleog: that would also be required for cases where the existing os doesnt have kexec enabled
<viric>
MichaelRaskin: I wish it did
<viric>
MichaelRaskin: but it would be udp-over-tcp
<clever>
ekleog: manualy copy the kernel+initrd to /boot and make them the default in grub.cfg
<clever>
viric: ive also been told that tcp over tcp will result in massive latency
<MichaelRaskin>
Let me try…
<viric>
clever: of course :)
<clever>
the inner most tcp uses packet loss to adjust its throttling
<ekleog>
clever: oh indeed, good point :) (but isn't that possible by just taking the .xz, unpacking it and recovering the kernel and initrd from it? -- not tested yet)
<clever>
and the outer most tcp refuses to drop any packets
<clever>
so the inner tcp just goes ham, and fills up buffers everywhere
<viric>
clever: all pretty simple
<clever>
ekleog: which .xz?
<ekleog>
clever viric: I've been told that story about tcp over tcp too, then read someone actually benchmarking it and it seems it worked better in at least some cases
isidore has quit [(Quit: WeeChat 1.8)]
<clever>
ekleog: i think it will vary, depending on if your trying to push more data then the network can allow
<ekleog>
clever: the one generated by config.system.build.kexec_tarball from your script? (didn't try it out yet so I may be wrong)
<viric>
ekleog: yes, the cases of not bulk transfer
<clever>
ekleog: yeah, that tarball will contain a kernel, initrd, and the full closure of bash + kexec
<clever>
ekleog: you can just unpack the tar and run the kernel+initrd however you like, as long as the kernel cmdline is intact
justelex has quit [(Quit: Konversation terminated!)]
<clever>
ekleog: the testing i recently added does just that with qemu
<shlevy>
niksnut: Is hydra's build queue documented anywhere?
goibhniu has quit [(Ping timeout: 255 seconds)]
<clever>
ekleog: and if you want to skip unpacking the tar, you can nix-build -A config.system.build.image instead, that will give you the files as a directory, rather then a tar
<clever>
the tar is only there to get the full closure of kexec into a machine that lacks nix
<simpson>
heath: In particular, k8s is supposed to be able to scale quite well, so you probably wouldn't be able to match it with nixops alone. However, it's possible to blend nixops and k8s in several ways, like clever has linked.
<heath>
disnix, nixops, ...
<heath>
trying to think of something equivalent
<heath>
clever: not really sure what i'm looking at having not used kubernates myself
<simpson>
Well, most of the k8s features have equivalences, but the entire composition of features doesn't. This is mostly because Nix is so different under the hood.
<clever>
heath: ah, i think its just a way to enable the kubernetes daemon within nixos systems
<heath>
neat!
<clever>
then you can spawn normal (non-nix based) kubernetes containers on it
orivej has joined #nixos
earldouglas has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] armijnhemel opened pull request #26105: WIP: hackish script to compare JSON from Guix and Nix (master...nix-guix-correlate) https://git.io/vHspl
NixOS_GitHub has left #nixos []
<UrsidaeJohn>
I'm trying to get some python code running, and I think that nix is using the wrong version of a package. I'm telling it to use pkgs.python35Packages.pygtksourceview, which is python3.5-pygtksourceview-2.10.1, but when I run nix-shell it fails with the error "pygtk-2.24.0 not supported for interpreter python3.5m" which means that it's using one of the python27Packages packages
<cornfeedhobo>
does nix only maintain a single build description? or do you all maintain the old versions?
<cornfeedhobo>
i'm just getting started so ... i could be asking poor questions, feel free to correct me.
<UrsidaeJohn>
cornfeedhobo I'm new at this too, but I think that nixpkgs generally only keeps one version
<cornfeedhobo>
hmm, maybe i have a poor understanding of nix then
<cornfeedhobo>
i was hoping nix, given the hash namespacing, would allow me to have arbitrary versions of software on the same system
<MichaelRaskin>
Well, NixPkgs contains only the latest version definition unless there is a good reason
<cornfeedhobo>
like, some systems i need python 2.7.8, some i need 2.7.13, specifically
<UrsidaeJohn>
yeah, so you _could_ have multiple versions of the same software, you just have to write the derivation yourself. Is that correct?
<MichaelRaskin>
You can have different versions installed at the same time, including the possibility of installing packages from the different NixPkgs versions in the same store (and even in the same profile, if you are careful~)
<cornfeedhobo>
okay. that is what i was hoping. hmm, how could one create a repo that supports storing each version persistently?
<UrsidaeJohn>
in my case, it looks like nix is using a python 2 package when I'm explicitly telling it to use a python 3 package. How could that happen?
<MichaelRaskin>
2.7.13 being incompatible with 2.7.8 code is, to be honest, worse than I expect from Python
<MichaelRaskin>
I am not sure what exactly happens; it is also a good questin which python you pass
<MichaelRaskin>
cornfeedhobo: given that NixPkgs is hard to manage at the current size, storing all old versions persistently may lead to unsustainable growth
<cornfeedhobo>
is this because the calculation and planning time, or raw size?
<MichaelRaskin>
We do have enough packages that ship multiple versions, you could use them as examples to add some more
<cornfeedhobo>
oh really? do you have some examples i could look at?
<MichaelRaskin>
First of all, whether all-packages.nix is too big to edit is already a realistic topic…
<MichaelRaskin>
boost has many versions
<MichaelRaskin>
autoconf does
<MichaelRaskin>
Linux kernel
<MichaelRaskin>
Wine
<MichaelRaskin>
LLVM
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 4 new commits to master: https://git.io/vHsjJ
<NixOS_GitHub>
nixpkgs/master f618a6c Tim Steinbach: linux: 4.11.2 -> 4.11.3
<NixOS_GitHub>
nixpkgs/master 446c57f Tim Steinbach: linux: 4.9.29 -> 4.9.30
<NixOS_GitHub>
nixpkgs/master 8f0ca4f Tim Steinbach: linux: 4.4.69 -> 4.4.70
NixOS_GitHub has left #nixos []
<heath>
15:02 < clever> then you can spawn normal (non-nix based) kubernetes containers on it
<dash>
cornfeedhobo: you can definitely have multiple versions on a single system if you need them
<heath>
that last bit doesn't sound so great
<dash>
cornfeedhobo: Even if they're not in nixpkgs you can provide them yourself
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 4 new commits to release-17.03: https://git.io/vHsjq
<NixOS_GitHub>
nixpkgs/release-17.03 3643979 Tim Steinbach: linux: 4.11.2 -> 4.11.3...
<NixOS_GitHub>
nixpkgs/release-17.03 0cca564 Tim Steinbach: linux: 4.9.29 -> 4.9.30...
<NixOS_GitHub>
nixpkgs/release-17.03 97a3b9b Tim Steinbach: linux: 4.4.69 -> 4.4.70...
NixOS_GitHub has left #nixos []
<cornfeedhobo>
that's slick that you all use NOTICE to tell the channel
<cornfeedhobo>
very correct ;-)
<dash>
cornfeedhobo: i think it's github doing that :)
<cornfeedhobo>
really?! not a webhook to a custom service?!
<MichaelRaskin>
nixos-wiki-bot may stilll not have migrated form msg to notice
<cornfeedhobo>
okay. this was informative. i will dive deeper soon. thank you all.
johnramsden has joined #nixos
<pingveno>
Wouldn't it be better to put it through an IRC bouncer?
<johnramsden>
Does anyone know if there is a way to force a rebuild of all system packages?
hiratara has joined #nixos
<clever>
dash: have the bot connect to a bouncer, which itself stays connected 24/7
<pingveno>
clever: Took the words right off my fingertips.
<dash>
oh. i think github just doesn't want to maintain that many connections for whatever reason
<dash>
johnramsden: probably, but why do you want to?
<dash>
johnramsden: why you want to do it will affect how you do it
<clever>
dash: also, the project name is in the ident field, which can only be changed by reconnecting
<clever>
dash: so it cant just /nick its way between projects
<johnramsden>
dash: I'm having a problem with my plasmashell. I already ran a check of my store, which didn't seem to help. All of my plasma widgets are gone and I'm getting the error: Error loading QML file: file:///nix/store/smlpa9dvd0n9qc1dn3vni47f2fs47hai-plasma-desktop-5.8.7/share/plasma/plasmoids/org.kde.plasma.taskmanager/contents/ui/main.qml:168:9: Cannot assign to non-existent property "enabled".
sibi has quit [(Quit: Connection closed for inactivity)]
<clever>
johnramsden: if nix-store --verify --check-contents doesnt find anything, then rebuildings things wont make any difference
<johnramsden>
clever: It did, but that didn't fix anything. Is there an easy way to diff configurations. My old generations still work, so it's obviously something related to the Nix store.
<Profpatsch>
In the second file I added comments to the gobjectIntrospection hook to see if it fires.
<Profpatsch>
Apparently it doesn’t once I add gtk3 to the dependencies.
darlan has joined #nixos
goibhniu has quit [(Ping timeout: 240 seconds)]
python476 has quit [(Read error: Connection reset by peer)]
darlan has quit [(Remote host closed the connection)]
greymalkin has joined #nixos
erasmas has quit [(Quit: leaving)]
dbmikus has quit [(Quit: WeeChat 1.7)]
<johnramsden>
clever: It doesn't make much sense to me. I haven't really changed anything between the two configurations other than removing some kernel parameter's. If I did want to force a rebuild, how would I do that? I figure it's worth a try
<clever>
johnramsden: did you nix-channel --update between them?
<johnramsden>
Well, my system is set to system.autoUpgrade.enable = true;
<clever>
ah
<clever>
so it could be an upstream problem, and nothing to do with your config
<johnramsden>
Should I post a bug report?
<johnramsden>
Issue
<clever>
johnramsden: if you look at the last ~6 digits of a nixos build, you can see nix channel versions, try checking those out in a clone of nixpkgs
<clever>
johnramsden: and then nixos-rebuild -I nixpkgs=/home/clever/nixpkgs/ test/switch
<clever>
to find out which revision breaks it, and which revision fixes it
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vHGkm
<NixOS_GitHub>
nixpkgs/master a8d3cb8 Matthew Maurer: rustfmt: 0.8.1 -> 0.8.3...
NixOS_GitHub has left #nixos []
earldouglas has quit [(Quit: leaving)]
ertes has joined #nixos
erictapen has joined #nixos
<erictapen>
hello everyone, I just tried to install a haskellPackage (pkgs.haskellPackages.tidal in environment.systemPackages) but it doesn't show up in ghc-pkg list | grep tidal. Can anyone give me hint, what I am doing wrong?
<clever>
erictapen: you need to use ghcWithPackages to get it into ghc-pkg