<NixOS_GitHub>
nixpkgs/master ce8bd32 Tuomas Tynkkynen: ncurses: Add temporary mirror for sources...
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vAzUS
NixOS_GitHub has left #nixos [#nixos]
<pjan>
as a result, all the builds I try to do this way fail.
Niall has quit [Quit: leaving]
Niall_ is now known as Niall
<rardiol1>
does nixos set the "-pipe" flag on gcc by default when building a package?
kelleyNif has quit [Quit: Leaving.]
astrofog has quit [Quit: Quite]
Lisanna has quit [Remote host closed the connection]
d4g_ has joined #nixos
Synthetica has quit [Quit: Connection closed for inactivity]
Arcaelyx has joined #nixos
<samueldr>
when giving an URL to nix-build, it downloads said URL to the cache and... how do I evict that cache (if I e.g. used master.tar.gz)
coot has quit [Ping timeout: 260 seconds]
d4g has quit [Ping timeout: 260 seconds]
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master b332924 Peter Hoeg: hugo: 0.32.2 -> 0.36.1
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vAzko
<samueldr>
the nix manual, for NIX_PATH, common options, nix-build and nix-shell only says «it is interpreted as the URL of a tarball that will be downloaded and unpacked to a temporary location.» for http and https URLs
mbrgm has quit [Ping timeout: 268 seconds]
mbrgm has joined #nixos
coot has joined #nixos
nuncanada has quit [Quit: Leaving]
dbmikus has quit [Remote host closed the connection]
coconnor has quit [Ping timeout: 276 seconds]
coconnor has joined #nixos
elasticdog has quit [Ping timeout: 264 seconds]
elasticdog has joined #nixos
<gchristensen>
iirc the cache isn't evictable, and lasts for 5min
<samueldr>
isn't it downloaded to /nix/store?
<gchristensen>
I don't know ...
<samueldr>
it says temporary location, and it defaults to /tmp when no $TMPDIR, looked there and nothing "nixy", so I assumed it was unpacked in the store
<samueldr>
especially since I seem to have a matching unpacked tarball in /nix/store (just looked)
thc202 has quit [Ping timeout: 240 seconds]
Lisanna has joined #nixos
<clever>
samueldr: ive read the code involved in it
<clever>
samueldr: first, it will hash the url, and lookup a cache file in $HOME
<clever>
and based on the TTL, either reuse the old store path, or re-query the webserver
<samueldr>
gchristensen: not 5 minute :S
<clever>
then, based on the cache-control headers, it may discover it hasnt changed, and re-use the storepath again
<clever>
and if you choose to have it unpacked, it will generate a derivation that unpacks, compute the output path, and check the store for that
<clever>
and may reuse the output
<samueldr>
clever: I've got a bunch of files in my $HOME, is it in something .nix-*anything* ?
<clever>
one sec
<clever>
[clever@amd-nixos:~]$ ls -ltrh .cache/nix/tarballs/
<samueldr>
(or if I'm pointed towards the right bits of code that would also be good!)
<clever>
-rw-r--r-- 1 clever users 143 Feb 21 14:13 0d4h36hxmzd1dw4109idlxppb0ly013laj79b1bw1068mi0zgh2m.info
<clever>
samueldr: passthru adds attributes to the derivation, but does not actually exposed them as env vars within the derivation
<samueldr>
nixpkgs it seems :)
<clever>
samueldr: so they have no effect on the hash of that derivation
<samueldr>
ah!
<samueldr>
great!
<taohansen>
I want to define a command to run before the fetch phase of my written derivation scoped within the derivation. is there a pre-fetch phase I can define?
<clever>
so its the same as (mkDeriation ..) // { foo = bar; }
<taohansen>
I looked through the Nixpkgs manual, doesn't look like it
<clever>
but it will persist after a .overrideAttrs
d4g has joined #nixos
TheAppleMan has quit [Quit: No Ping reply in 120 seconds.]
yegortim1 has joined #nixos
pkill9 has joined #nixos
yegortimoshenko has quit [Ping timeout: 255 seconds]
owickstrom has quit [Ping timeout: 268 seconds]
d4g_ has quit [Ping timeout: 256 seconds]
owickstrom has joined #nixos
kelleyNif has joined #nixos
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 1cacfe3 Vladimír Čunát: nixos.tests.boot-stage1: fix missing libelf
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vAzqt
<cmcdragonkai1>
When running htop, because all my programs are within nix/store
<cmcdragonkai1>
Every command path stores with `/nix/store/......`
<cmcdragonkai1>
Is there a way to hide the prefix of this path?
<cmcdragonkai1>
So I can see what programs I am actually running.
<samueldr>
gchristensen: guessing it would be `identity (system)` for now?
<gchristensen>
samueldr: sure, I could even imagine just doing key: value, key: value, .... for whatever is present in the metadata
<gchristensen>
(if you wanted to do that, I would delete the attempt_id from that object on my end)
<samueldr>
hmm, no need to, I can easily list known in a well-formated manner, and still allowing the other props to be iterated over, and yet, "using" (but discarding) the attempt_id
<gchristensen>
cool
<samueldr>
I'm thinking the additional (or even full raw) data could be added as a title to that bar
<adisbladis[m]>
cmcdragonkai1: Press "P" ;)
<samueldr>
not discoverable, but if one day you go metadata crazy, it won't fill the secondary information line
<cmcdragonkai1>
adisbladis[m]: Oh yes :D!
<gchristensen>
samueldr: I wonder about a floating box that just shows formatted json?? I dunno :P
<gchristensen>
samueldr: what is super easy? ;)
<samueldr>
ah, or something like that
<samueldr>
it can all be incremental
<gchristensen>
yeah
<samueldr>
(as in, easiest solution first)
<gchristensen>
SGTM :D
<samueldr>
wow, thanks adisbladis[m], I was reducing fontsize and increasing windowsize to cope
<adisbladis[m]>
Haha np :)
fragamus has quit [Ping timeout: 265 seconds]
<gchristensen>
samueldr: btw this API is live
<gchristensen>
(commented on the issue)
__red__ has quit [Ping timeout: 248 seconds]
__red__ has joined #nixos
<samueldr>
nice, this'll make it easier to work on
<samueldr>
it probably will happen by friday
yegortim1 has quit [Remote host closed the connection]
Mateon1 has quit [Quit: Mateon1]
yegortim1 has joined #nixos
Mateon1 has joined #nixos
<gchristensen>
=) thank you!
heath has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #35310: gcc7: fix native and cross-musl build, sync w/gcc6 fixes (master...fix/gcc7-musl) https://git.io/vAzmu
NixOS_GitHub has left #nixos [#nixos]
<gchristensen>
samueldr: I hacked it in to the -ofborg- log pane for now, just to see what it looks like. it looks bad. haha, but ... it is there.......
heath has joined #nixos
<samueldr>
yes, will probably be much better to add it to the actual tab
<samueldr>
though, at least you can (until then) know which machine shows particular issues
<gchristensen>
yeah
<samueldr>
(that's to do with that issue, right)
<gchristensen>
exactly :)
kelleyNif has quit [Quit: Leaving.]
garbas has quit [Quit: WeeChat 1.9.1]
coconnor has quit [Ping timeout: 255 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 39f61d5 Will Dietz: gcc7: fix native and cross-musl build, sync w/gcc6 fixes
<NixOS_GitHub>
[nixpkgs] dtzWill pushed 2 new commits to master: https://git.io/vAzYw
<NixOS_GitHub>
nixpkgs/master ebec142 Will Dietz: Merge pull request #35310 from dtzWill/fix/gcc7-musl...
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vAzYK
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 89343b4 Peter Hoeg: dbus-broker: 9 -> 11
<Lisanna>
Can I access a nixops machine's logical configuration from its physical config file?
<Lisanna>
trying to setup key deployment which is a physical configuration setting, but I want to grab the name of a user from the logical configuration of the machine
spear2 has joined #nixos
<Lisanna>
wondering if I can use the "config" argument provided in the physical configuration function, but not sure if the logical config can be accessed through there, and the nixops manual doesn't say anything about it
tertle||eltret has quit [Quit: Connection closed for inactivity]
<Lisanna>
Oh, there's no such thing as "logical" and "physical" nixops configurations - that's just an invention of the nixops manual. I guess they both get merged at the end
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #35313: jetbrains-editors: minor updates (master...patch-87) https://git.io/vAzO6
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] gnidorah opened pull request #35314: autorandr: embed path to xrandr (master...autorandr) https://git.io/vAzO1
NixOS_GitHub has left #nixos [#nixos]
schoppenhauer has quit [Ping timeout: 256 seconds]
schoppenhauer has joined #nixos
snajpa has quit [Ping timeout: 248 seconds]
Adluc has quit [Ping timeout: 255 seconds]
srk has quit [Ping timeout: 256 seconds]
<fearlessKim[m]>
Lisanna: I hope they reword that part, I found that really confusing when starting nixops
yegortim1 has quit [Remote host closed the connection]
yegortim1 has joined #nixos
liori has quit [Remote host closed the connection]
Biappi has quit [Ping timeout: 256 seconds]
jtojnar has quit [Ping timeout: 240 seconds]
zzamboni has quit [Quit: Leaving.]
liori has joined #nixos
simukis has joined #nixos
pie_ has quit [Ping timeout: 240 seconds]
hamishmack has joined #nixos
iyzsong has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to staging: https://git.io/vAzct
<NixOS_GitHub>
nixpkgs/staging 102ed10 Tuomas Tynkkynen: e2fsprogs: 1.43.8 -> 1.43.9
NixOS_GitHub has left #nixos [#nixos]
<Lisanna>
hrmph. nixops deploy from macos to linux sucks
<Lisanna>
can't believe I have to setup remote building to make nixops work
davidak has joined #nixos
MercurialAlchemi has joined #nixos
<dhess>
Lisanna: Let's say it needs to build a package that you want to deploy. How else would it work?
<Lisanna>
build it on the target <_<
<dhess>
but... that's remote building
<dhess>
NixOps on Linux doesn't work that way
<Lisanna>
Right... I'm not incredulous that it needs to use remote building
<Lisanna>
it's that I need to manually configure it
<Lisanna>
i.e. nixops doesn't just do that automatically
<Lisanna>
sounds like there's some fixes for this, but they haven't all gone in yet
<dhess>
There's nothing like that coming down the pipe that I'm aware of.
<dhess>
The only things I know of that are going to improve things are maybe that you don't need to set nixpkgs.system in all of your machine/container definitions
<dhess>
yeah that's got nothing to do with what you're asking for.
<Lisanna>
what will it do?
<Lisanna>
and, I don't mind setting nixpkgs.system manually
<dhess>
All it will do is set nixpkgs.system properly in the container. Apparently right now it doesn't.
<dhess>
that's it.
<Lisanna>
...okay, so, what's the most practical solution to this problem then? Do I need to go and create a remote building configuration to use nixops?
<dhess>
From macOS? Yes.
<Lisanna>
or are there easier ways to tell nixops to just build it on the target
<dhess>
The easiest way is probably to use LnL's nix-docker
<Lisanna>
I've been trying something like that actually... a nix docker build slave
<Lisanna>
I've never done remote building before though it's not working well for me ):
<dhess>
NixOps can't build on the target. For no other reason, imagine that it needs to build a package *before it has even deployed the target*
<Lisanna>
NIX_REMOTE_SYSTEMS is set, but nixops seems to be ignoring it
<dhess>
LnL has already done the hard work for you with nix-darwin and nix-docker
<dhess>
I too had problems getting remote builds working on macOS until I installed nix-darwin
<dhess>
and now it all works beautifully.
<Lisanna>
nix-darwin...
<Lisanna>
this is nixos for darwin?
<dhess>
It's modules and configuration for macOS
alex`` has joined #nixos
<Lisanna>
oh wow
<dhess>
system-wide configuration, unlike the stuff you do in your own ~/.nixpkgs or wherever
<dhess>
it's very nice!
<Lisanna>
how much does it stomp around on my existing darwin system?
griff_ has joined #nixos
<dhess>
A bit. It will write files to /etc. But then so does the multi-user Nix installer, so if you're already running that, this isn't too much different.
<Lisanna>
ok
<Lisanna>
well, seems worth trying!
<dhess>
everything else it keeps in /run/system, which it also creates, but that is again no different to Nix creating /nix
<dhess>
Lisanna: I was skeptical at first, but now I love it.
<dhess>
I would prefer that it (and Nix) didn't touch /etc, but oh well. I think it's worth the hassle.
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging 54ac74c Tuomas Tynkkynen: gnutls: 3.6.1 -> 3.6.2
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to staging: https://git.io/vAzCQ
<Lisanna>
in my nixops configuration, when I try to deploy it, I'm getting "error: attribute ‘nixosVersion’ missing, at /nix/store/l2a5g79gsyzx6rji3z082hdqgj4393n2-nixops-1.5.2/share/nix/nixops/eval-machine-info.nix:284:107" - am I doing something silly?
<Ke>
btw. has anyone else been missing some sort of LRU caching for the garbage collection on nixos
<Ke>
ie. adaptively do a partial cleanup of packages that have not been recently referenced
knupfer has quit [Ping timeout: 256 seconds]
<Ke>
and by reference in this case I mean using the variable ${mypkg} in a build expression, but not necessarily for a package that is in use
leat has quit [Ping timeout: 276 seconds]
<simpson>
Ke: What's your use case, and how does the current GC not work for you?
<Ke>
we use nixtest for transient builds that are not always used, but the dependencies may or may not be rebuilt or cleaned
<Ke>
if the dependency is referred to often, it could not be cleaned up
asuryawanshi has quit [Remote host closed the connection]
<Ke>
like testing
<Ke>
I think the current gc cleans everything that has no references at that time
<Lisanna>
Ke there used to be something like that, but it depended on the "last accessed" filesystem attribute and it was removed
<Ke>
yeah atime is bad, and not probably suitable for our purposes
rauno has quit [Ping timeout: 256 seconds]
deepfire has quit [Ping timeout: 260 seconds]
<Ke>
well I guess relatime might be at some level tolerable
marusich has quit [Ping timeout: 256 seconds]
asuryawanshi has joined #nixos
<Lisanna>
Ke yeah, I agree, relatime should have been fine
<Lisanna>
don't think it was worth getting rid of, should've just switched it to relatime and kept it as an option
<Ke>
well I am not saying it's completely fine, some snapshotting systems might not like it
<Ke>
in general I think atime is the worst invention ever
<Lisanna>
haha
<Ke>
ie. turning reads into writes is the thing I hate the most
marusich has joined #nixos
<Ke>
it has actually even caused moderate data corruption on btrfs
<Ke>
IIRC
<Ke>
well mostly corrupt atimes on send
<LnL>
I only run a gc every 3 months or so
<Ke>
LnL: let's assume we are doing a lot of test builds
<LnL>
I'm not? :)
<Ke>
=o)
<Ke>
anyway, in theory there is some database of the packages which is referenced on builds and access times could be bumped there on new builds, right?
asuryawanshi has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos
<LnL>
store paths and the nix db don
<LnL>
don't have timestamps AFAIK
akaWolf has quit [Ping timeout: 256 seconds]
<Ke>
you mean, I could make higher leve solution that keeps the references and kills them with LRU or something
<LnL>
sure, just saying that you don't have a lot of info like that to go with
<LnL>
making sure whatever you care about has a gc-root is probably a lot easier
<Ke>
but the thing is not care about, but caching
<LnL>
nix-collect-garbage() { : }
<LnL>
best caching ever :D
<sphalerite_>
Ke: yeah that would be really useful. I think in general nix could collect and store more useful data, one thing I'd really like is being able to access build durations more easily
rauno has joined #nixos
<sphalerite_>
with enough logging you could implement that sort of thing as a separate tool which wires up that data to nix-store --delete. Might be useful to mention on https://github.com/NixOS/nix/issues/1829
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<LnL>
more serious, it should be possible to determine paths that are used frequently by intercepting or hooking into the daemon somehow
zzamboni has quit [Quit: Leaving.]
zzamboni has joined #nixos
<dhess>
build logs should do it
<Ke>
yeah, I think genlop parses build logs on gentoo
<dhess>
then you can take it out of the critical path, too
<dhess>
You could also put a Nginx reverse proxy in front of your binary cache and collect logs there
<dhess>
or S3 logging or whatever.
<dhess>
I guess that only works if you are not using cache.nixos.org though, which is unlikely
<joko>
Hello, could anyone help me with restrict-eval? Hydra evaluator returns now errors like "error: access to path /nix/store/xxxx is forbidden in restricted mode"
<sphalerite_>
and build time stats would be useful for deciding on which builder to assign a given derivation as well
tommyangelo has quit [Ping timeout: 256 seconds]
thc202 has joined #nixos
<LnL>
why did apt upgrade my libstdc++ time -_-
Itkovian has joined #nixos
knupfer has joined #nixos
ixxie has joined #nixos
leat has joined #nixos
davidak has joined #nixos
<FRidh>
Mic92: I personally find the jupyter PR a bit messy, but I have neither time nor incentive to look at it any further, so feel free to take that on.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/staging d322a5b Robert Schütz: pythonPackages.bibtexparser: enable tests
<NixOS_GitHub>
nixpkgs/staging 4a9a7d3 Robert Schütz: pythonPackages.binaryornot: 0.4.0 -> 0.4.4
<NixOS_GitHub>
[nixpkgs] FRidh pushed 27 new commits to staging: https://git.io/vAzgd
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging e117650 Robert Schütz: pythonPackages.bibtexparser: add propagatedBuildInputs
jbarthelmes[m] has joined #nixos
rauno has quit [Ping timeout: 240 seconds]
jensens has joined #nixos
__Sander__ has joined #nixos
rardiol1 has quit [Remote host closed the connection]
rauno has joined #nixos
rardiol1 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] adisbladis opened pull request #35327: go: Remove old versions 1.6 and 1.7 (master...old_gophers) https://git.io/vAzVm
NixOS_GitHub has left #nixos [#nixos]
rardiol1 has quit [Remote host closed the connection]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 2 new commits to master: https://git.io/vAzVr
<NixOS_GitHub>
nixpkgs/master eb5ad9b Peter Hoeg: dbus-broker: 9 -> 11
<NixOS_GitHub>
nixpkgs/master 1c67ee4 Peter Hoeg: Merge pull request #35325 from peterhoeg/u/dbusb...
NixOS_GitHub has left #nixos [#nixos]
ThatDocsLady_ has joined #nixos
<Lisanna>
I'm having trouble building nixops from a custom git checkout, it the builder fails with "... is not a valid wheel filename"
jensens has quit [Remote host closed the connection]
jensens has joined #nixos
gfvgfvgfv has joined #nixos
<srhb>
Hmm. Any idea why gitlab 10.4.x hasn't been backported? I'm unsure whether to follow the (continuous) 10.3.x patches, or 10.4.x (which is still a minor version afaik?)
<srhb>
Maybe just an oversight, I don't know
<gfvgfvgfv>
Hi. How does one embed a kernel patch into configuration.nix? (Is this actually a good way to add a custom kernel patch?)
<srhb>
gfvgfvgfv: boot.kernelPatches
<gfvgfvgfv>
Would this work with a "name = <...>; patch = ''diff --git <...>''"?
<gfvgfvgfv>
With no extra .patch files, just a diff embedded within the configuration.nix file.
<srhb>
gfvgfvgfv: No, as far as I can see patch is a path to a patch file. However, you can just use pkgs.writeText or something to write it to the store.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] clefru opened pull request #35328: trezor: Add pyblake2, click, libusb1 and ecdsa as propagatedBuildInput (master...pytrezor-fix) https://git.io/vAzKW
NixOS_GitHub has left #nixos [#nixos]
Biappi has joined #nixos
zzamboni has quit [Quit: Leaving.]
zzamboni has joined #nixos
zzamboni has quit [Client Quit]
zzamboni has joined #nixos
coot has joined #nixos
griff_ has joined #nixos
griff_ has quit [Client Quit]
zzamboni has quit [Client Quit]
Mateon3 has joined #nixos
vidbina has quit [Ping timeout: 276 seconds]
griff_ has joined #nixos
vidbina has joined #nixos
Mateon1 has quit [Ping timeout: 255 seconds]
Mateon3 is now known as Mateon1
zzamboni has joined #nixos
sigmundv has joined #nixos
<gfvgfvgfv>
srhb: is there any example of calling writeText from within the building environment? I have about five hours of experience with Nix and that's totally not enough :(
roberth has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] gebner pushed 1 new commit to master: https://git.io/vAz6x
<NixOS_GitHub>
nixpkgs/master a14ab86 Gabriel Ebner: fcitx-engines.table-other: 0.2.3 -> 0.2.4
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] adisbladis pushed 2 new commits to master: https://git.io/vAz6h
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master b846847 Clemens Fruhwirth: trezor: Add pyblake2, click, libusb1 and ecdsa as propagatedBuildInput
<NixOS_GitHub>
nixpkgs/master 52e419b adisbladis: Merge pull request #35328 from clefru/pytrezor-fix...
f0i has joined #nixos
iyzsong has joined #nixos
<manveru>
gfvgfvgfv: what do you want to do?
<gfvgfvgfv>
manveru: I have a tiny two-line kernel patch that is no use for anyone else. I want to place it somewhere sensible and build a patched kernel.
pie_ has joined #nixos
TweyII has joined #nixos
<manveru>
the easiest is to just make a file yourself and reference it from your configuration.nix :)
TweyII is now known as Guest54223
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 831ef47 adisbladis: electrum: Patch compat with python trezor>=0.9.0
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] adisbladis pushed 1 new commit to master: https://git.io/vAzi1
<manveru>
otherwise, you can use `patches = let myKernelPatch = writeText "my-kernel.patch" ''write patch here''; in [ myKernelPatch ];` i guess
kim0 has quit [Quit: Connection closed for inactivity]
ravloony has quit [Remote host closed the connection]
griff_ has quit [Ping timeout: 256 seconds]
tommyangelo_ has joined #nixos
monotux has quit [Read error: Connection reset by peer]
monotux has joined #nixos
ilyaigpetrov has joined #nixos
pie_ has quit [Ping timeout: 276 seconds]
MercurialAlchemi has quit [Ping timeout: 264 seconds]
d4g has quit [Ping timeout: 264 seconds]
slyfox has quit [Ping timeout: 264 seconds]
jeschli has quit [Ping timeout: 264 seconds]
jaeckel has quit [Ping timeout: 264 seconds]
anderslundstedt has quit [Ping timeout: 264 seconds]
epta has quit [Ping timeout: 264 seconds]
tk_ has quit [Ping timeout: 264 seconds]
globin has quit [Ping timeout: 264 seconds]
laumann has quit [Ping timeout: 264 seconds]
raxius has quit [Ping timeout: 264 seconds]
Gohla has quit [Ping timeout: 264 seconds]
jeschli has joined #nixos
tk_ has joined #nixos
globin has joined #nixos
<gfvgfvgfv>
manveru: brilliant! boot.kernelPatches = let myPatch = pkgs.writeText "mypatch.patch" ''<...>'' in [ { name = "mypatch"; patch = myPatch; } ]; did the trick. Should I document it somewhere? It looks like a good example for ad-hoc package patching =)
<lexdoingsomethin>
Is it okay & fine to use recursive set in /etc/nixos/configuration.nix ? like { config, pkgs, ... }: rec { #and so on
tommyangelo has joined #nixos
_rvl has quit [Ping timeout: 240 seconds]
hyper_ch has quit [Ping timeout: 240 seconds]
jaeckel has joined #nixos
<manveru>
lexdoingsomethin: while i prefer doing let/in instead, i don't see why it wouldn't work
robstr has quit [Remote host closed the connection]
<lexdoingsomethin>
it works fine for me now. I used it to satisfy conditionals like IF ELSE inside option set. just clarifying if that is good practice.
Gohla has joined #nixos
<TweyII>
I've noticed that binaries compiled with a Nix toolchain always seem to look in /run/opengl-driver/lib/drm now for drivers even when LD_LIBRARY_PATH is unset and the binary's RPATH points to /nix/store/… for libGL
_rvl has joined #nixos
<TweyII>
Why is that? Is the /run/opengl-driver/lib/drm path in mesa_noglu's libGL?
hyper_ch has joined #nixos
<manveru>
pretty sure that's the case, that's on all platforms if i remember right
<manveru>
(well, linux anyway)
<avn>
TweyII: export LIBGL_DRIVERS_PATH="${mesa.drivers}/lib/dri" to set dri path explicitly
<TweyII>
avn: Ahh, so the /run/opengl-driver/lib/dri path is just a default value for that?
<lexdoingsomethin>
How I can get end value of any particular option, let's say powerManagement.enable , in the current state of my NixOs?
<Lisanna>
lexdoingsomethin from a nixos module or from the command line?
<manveru>
`nixos-option powerManagement.enable`
<lexdoingsomethin>
from command line prefferably
<avn>
TweyII: yes, and up again -- it a bit doubtful default. I believe it from old times, when nvidia? blob has some_dri.so loadable via default libGL
<manveru>
it'll take the value from your config, not from the current "state"
<lexdoingsomethin>
oh, manveru, that's great, i like it
<TweyII>
alp, guibou: Ooh, I did stumble across this, thank you
<TweyII>
alp, guibou: Any plans to get it into nixpkgs?
<avn>
TweyII: I usually make scripts for game loading -- to use libGL/dri that explicitly built with
<lexdoingsomethin>
Lisanna I will add record about nix-option to my personal NixFAQ
<lexdoingsomethin>
thnx
<alp>
TweyII, well, if there's demand for it I'm sure guibou won't mind :) but he only recently got it into a state is rather happy with, which would explain why it's not there yet (AFAIK)
[0__0] has quit [Remote host closed the connection]
sdhandsucks has joined #nixos
prooftechnique_ has joined #nixos
owickstr- has joined #nixos
Fuuzetsu_ has joined #nixos
jophish has joined #nixos
jeschli1 has joined #nixos
Faster-Fanboi_ has joined #nixos
aszlig_ has joined #nixos
Fuuzetsu_ is now known as Fuuzetsu`
bgamari- has joined #nixos
avn_ has joined #nixos
^il has joined #nixos
sdhand has quit [Ping timeout: 264 seconds]
Guest88515 has quit [Ping timeout: 264 seconds]
dan_b has quit [Ping timeout: 264 seconds]
owickstrom has quit [Ping timeout: 264 seconds]
Niall has quit [Ping timeout: 264 seconds]
DavidEGrayson has quit [Ping timeout: 264 seconds]
avn has quit [Ping timeout: 264 seconds]
abrar has quit [Ping timeout: 264 seconds]
duncan^ has quit [Ping timeout: 264 seconds]
owickstr- is now known as owickstrom
prooftechnique has quit [Ping timeout: 264 seconds]
hotfuzz has quit [Ping timeout: 264 seconds]
LnL has quit [Ping timeout: 264 seconds]
sweater has quit [Ping timeout: 264 seconds]
joachifm has quit [Ping timeout: 264 seconds]
aszlig has quit [Ping timeout: 264 seconds]
{`-`} has quit [Ping timeout: 264 seconds]
Fuuzetsu has quit [Ping timeout: 264 seconds]
etu has quit [Ping timeout: 264 seconds]
globin has quit [Ping timeout: 264 seconds]
srhb has quit [Ping timeout: 264 seconds]
Fuuzetsu` is now known as Fuuzetsu
Unode has quit [Ping timeout: 264 seconds]
vil has quit [Ping timeout: 264 seconds]
woboats has quit [Ping timeout: 264 seconds]
bgamari has quit [Ping timeout: 264 seconds]
hiberno has quit [Ping timeout: 264 seconds]
Faster-Fanboi has quit [Ping timeout: 264 seconds]
jeschli has quit [Ping timeout: 264 seconds]
srhb has joined #nixos
woboats has joined #nixos
abrar has joined #nixos
Unode has joined #nixos
hotfuzz has joined #nixos
[0__0] has joined #nixos
hiberno has joined #nixos
<TweyII>
guibou: As a nix-shell user, it'd be nice if I could get it from my usual nixpkgs and provide the driver e.g. as an argument to my shell function when invoking nix-shell
etu has joined #nixos
<manveru>
doesn't the driver have to match what's in the kernel?
<TweyII>
guibou: I'm not saying you have to, just that if you're worried about the awkwardness of providing the argument, I'd like to point out that I have valid use cases where that's acceptable :)
<guibou>
manveru: yes, unfortunately
hyper_ch has quit [Ping timeout: 260 seconds]
<TweyII>
It seems like we could write a script to probe for that information, too
<manveru>
so you can't make fully pure opengl packages
<TweyII>
manveru: Yeah :(
<guibou>
manveru: OpenGL packages are not pure by definition ;(
<manveru>
alright, thought i missed something :)
<guibou>
TweyII: Actually I don't see why you may want to provide this in a nix-shell. I'm usually installing globally (for my user) the useful nixGLXXX and let impure nix-shell do its work.
<guibou>
TweyII: I can see the usage if you want to work in a pure nix-shell, but I think even on NixOS, pure nix-shell and OpenGL does not work
<manveru>
pure nix-shell just means it cleans out env, not a whole chroot
robstr has quit [Remote host closed the connection]
hyper_ch has joined #nixos
spear2 has quit [Read error: Connection reset by peer]
robstr has joined #nixos
kgz has joined #nixos
<TweyII>
guibou: We're trying to have it so that nix-shell provides everything that's required to build and run the application, and indeed a pure nix-shell can still depend on a driver package (choosing the right driver package is a bit tricky, but not something I have to worry about yet)
sarah_ has joined #nixos
<guibou>
TweyII: the OpenGL status is weird (even on nixos). Most of the time, people are building OpenGL application using the "mesa" dependency.
<guibou>
and when they are running it, they pick what's is configured in the impure environment
<guibou>
TweyII: but hell yeah, if you think it can be a good idea, I'll put it in nixpkgs
<lexdoingsomethin>
How do I configure screen timeout, powersaving via nixOs options?
robstr has quit [Ping timeout: 240 seconds]
lambdamu has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nix/master 98031b6 Will Dietz: nix build: Don't create output links with --dry-run....
<NixOS_GitHub>
[nix] edolstra pushed 8 new commits to master: https://git.io/vAzy4
<NixOS_GitHub>
nix/master 0ffa615 Will Dietz: busybox shell: enable various useful/expected features...
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nix/master 3780435 Will Dietz: tests: Add (failing) tests for reported --dry-run issues.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra closed pull request #1842: busybox shell: enable various useful/expected features (master...fix/sandbox-shell-features) https://git.io/vNhXp
<etu>
I've seen references to https://sta.li/, but it just return 403 for me
pkill9_ has joined #nixos
* lexdoingsomethin
likes idea of statically linked musl applications
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar pushed 2 new commits to master: https://git.io/vAz9Z
<NixOS_GitHub>
nixpkgs/master 022cbdd David Kleuker: xdg-utils: fix homepage URL
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 6b707cf Jan Tojnar: Merge pull request #35320 from davidak/patch-1...
zzamboni has quit [Quit: Leaving.]
zzamboni has joined #nixos
<etu>
hodapp: wow
<manveru>
lol
<manveru>
"This destroyed 3 production server after a single deploy!"
<hodapp>
dtz: oh, cool! thanks
<lexdoingsomethin>
sta.li at this current state AFAIK is dead and not usable project, there is however some progress in bitfrost and alpine statical link toolchains
<hodapp>
Li[m]: what do you mean, writing a nix tree from it?
<Li[m]>
writing a nix tree to build oasis
<etu>
dtz: Today! ?!?
<etu>
woo
<hodapp>
oh, I'm just not used to hearing it called a tree
<Li[m]>
hodapp: like nixpkgs, the tree that builds nixos, but only to build this small legacy-free system
<hodapp>
Li[m]: oh, I see what you mean
<Li[m]>
its build system should make it easier
<hodapp>
Li[m]: are there other examples of this online?
<hodapp>
hmm. I've seen NixOS-specific stuff in nixpkgs but haven't ever actually looked into the fact that I'd guess that one would use Nix to, well, bootstrap NixOS
<Li[m]>
hodapp: there are some forks, im aware of triton, but there are more I forgot, dont know how active. nixos is biased towards bigger builds to get more features, but there is a movement, which was once called suckless nixos, to have another tree which is biased to minimal build options/libs
<Mic92>
is there a irc channel dedicated to darwin?
<lexdoingsomethin>
Li[m] looks promising, yes I had it in my list of static projects, along with https://github.com/minos-org/minos-static . It would be nice to have two things 1) repo of all linux application in static binary tarballs 2) and nix based toolchain around build scripts to build those applications with custom MAKE\C\FLAGS configs
<Li[m]>
I think just dropping the entire xorg world is the right thing to do now
<hodapp>
Li[m]: yeah, that's the SLNOS I just asked about and that some folks were talking about
<hodapp>
are folks that close to ditching X?
<Li[m]>
hodapp: I am
<hodapp>
for wayland?
<Li[m]>
mobile open OS is coming, and its just gonna be wayland
<lexdoingsomethin>
actually i was wondering to implement both too, by myself
<Li[m]>
fuchsiaos is wayland
Synthetica has joined #nixos
<Li[m]>
and so is chromeos
<Li[m]>
android doesnt really matter at this point, but theres libhybris to run wl on android stack, so X is pretty monolithic and useless for future-looking dev
<Li[m]>
maybe someone will leak qualcomm graphic drivers and well all run nixos on our samsung
<manveru>
anyone here using scuttlebutt?
<hodapp>
I tried running X on Android once just to try it and it was horribly nonfunctional
<hodapp>
while this late 2012 I doubt it's gotten much better
<hodapp>
mostly what I remember is that it was crashing before I even got as far as xterm
<Li[m]>
X wasnt designed properly, wayland is a better way to use the hardware better and secure the windows/input better, which is why the first WM have so few features
<hodapp>
I've read plenty about X and Wayland but have yet to actually use Wayland regularly
<TimePath>
X's network trabsparency is a neat concept, but all the major toolkits use pixmaps now anyway
<hodapp>
from what I recall, doesn't Wayland still allow a decent level of network transparency too?
<TimePath>
transparency*
<Li[m]>
sway is about to take a big leap forward when they stabilise the new compositor library
<Li[m]>
which is also being ported to rust bindings for a rust tiling wm
<Li[m]>
I guess stable nixos doesnt have the best wayland support now
<hodapp>
sway?
<Mic92>
I still wait for wlroots update, before I consider wayland again
<Li[m]>
gnome is on wayland by default now on a lot of distros
<Li[m]>
so a lot of people are on wayland now
<Mic92>
does firefox/chrome runs nativly now?
coot has joined #nixos
Myrl-saki has joined #nixos
spear2 has joined #nixos
freeman42x]NixOS has joined #nixos
ravloony has joined #nixos
tommyangelo has quit [Read error: Connection reset by peer]
<Profpatsch>
ofborg had a few problems because of broken master, but the tests seem to work now.
<Profpatsch>
At least on x86_64
<lexdoingsomethin>
when I boot up laptop, I see strange artefacts , already in DM, after I log in sometimes screen goes darker by itself, and window become transparent in a bad way (I did not enforced transparency in my system, and they mistakenly show prevous screens, sessions, frames) . anyone knows what the cause might be?
aarvar has quit [Ping timeout: 256 seconds]
knupfer has quit [Ping timeout: 240 seconds]
<lexdoingsomethin>
I tried to screen shot that, but screen shot did not captured graphics artefacts(
mrkgnao has joined #nixos
<goibhniu1>
what GPU have you got lexdoingsomethin?
goibhniu1 is now known as goibhniu
<lexdoingsomethin>
nvidia
<lexdoingsomethin>
10XX
<lexdoingsomethin>
and nvidia driver
Wizek has quit [Ping timeout: 255 seconds]
sjanssen has quit [Ping timeout: 255 seconds]
griff_ has joined #nixos
sjanssen has joined #nixos
f0i has quit [Ping timeout: 276 seconds]
pie_ has joined #nixos
indika has quit [Ping timeout: 255 seconds]
<guibou>
What can be the reason for which derivation does not have an "override" attribute ?
indika has joined #nixos
<lexdoingsomethin>
this issue is occasional, it happens from time to time, and I'm not awared of clear steps to reproduce
<etu>
lexdoingsomethin: What channel are you on? Kernel version?
habnabit has quit [Ping timeout: 255 seconds]
<lexdoingsomethin>
etu 17.09 , 4.9.82
fendor has quit [Ping timeout: 255 seconds]
<etu>
Hmm, I got a 10XX card (not laptop) and it works fine. But I'm on nixos-unstable and 4.14
tk_ has quit [Ping timeout: 240 seconds]
<lexdoingsomethin>
i did nix-build --upgrade yesterday, but the issue still existed after reboot. another strange this - issue do disappear by itself after 3-5-10 minutes, from time to time. so it fixes itself occasionally.
fendor has joined #nixos
<joko>
Is it possible to import a nix file from a derivation (= sth I get via fetchgit)?
<lexdoingsomethin>
but unpleasant thing about it - that it occurs for me unpredictably, e.g. when I code , screen could go darker so letters became barely readable
<joko>
So far I was using its outPath, but it no longer works
<etu>
lexdoingsomethin: I don't know if 4.14 is in 17.09... But I'm out of other ideas than to upgrade since you have quite modern hardware.
<manveru>
joko: try with .out instead?
gfvgfvgfv has quit [Remote host closed the connection]
<manveru>
might need to be `in builtins.map (drv: "${drv.out}") [ systemd-zfs-generator gummibootr ];`
<manveru>
depends on what fetchcode does... that's not something i know
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] mdorman opened pull request #35333: connman: fix build with upstream patch (master...connman-fix) https://git.io/vAzFr
NixOS_GitHub has left #nixos [#nixos]
<joko>
manveru: Right, it's a wrapper for fetchgit
<lexdoingsomethin>
I did upgrade recently. issue is still there. I'm wondering what could automatically trigger it : powersaving modes, thermal settings, cooler settings, not sure where to dig
pkill9 has joined #nixos
<goibhniu>
what desktop/wm are you using lexdoingsomethin?
<goibhniu>
lexdoingsomethin: are there compositing options you can play with?
pkill9_ has quit [Ping timeout: 240 seconds]
<etu>
Lisanna: Indeed it is.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] manveru opened pull request #35334: Patchwork (master...patchwork) https://git.io/vAzFd
NixOS_GitHub has left #nixos [#nixos]
the-kenny has quit [Ping timeout: 256 seconds]
<lexdoingsomethin>
goibhniu I don't know) I'm not familiar with it. probably I don't have compositor.
<joko>
manveru: with "${drv.out}", Hydra still complains: access to path '/nix/store/rs285514z306wy14y5bnkqi36c64j9b6-systemd-zfs-generator-1e33bd2' is forbidden in restricted mode
<manveru>
ah... right
<manveru>
that only works in nix-shell i guess
<manveru>
but there's a way for sure
<lexdoingsomethin>
maybe I could dump any variables, and values when system affected by screen lag bug, and when it is not, and compare it.
alex`` has quit [Ping timeout: 276 seconds]
<goibhniu>
lexdoingsomethin: did you check if this is a known issue on other distros? ... it could be an i3wm/nvidia bug?
<lexdoingsomethin>
ok, I'll check, np.
the-kenny has joined #nixos
rauno has quit [Remote host closed the connection]
<joko>
manveru: using plain drv.out shows the following: "The option `GIT_SSL_CAINFO' defined in `/nix/store/nh639rljcbb2wli5rfd5sqs219326nfn-git-export/machines/munch-hardware.nix' does not exist." Most likely default.nix from that repository is not imported
<manveru>
and if you add "/default.nix" to the path?
<manveru>
tbh i've never tried making this work with imports, only import itself
LnL7 is now known as LnL
pop has quit [Ping timeout: 255 seconds]
pop has joined #nixos
kelleyNif has joined #nixos
<joko>
Maybe I should create a proper (:P) impurity by using a NIX_PATH value / Hydra input
pop is now known as Guest78130
<manveru>
well, git submodules would work :)
<manveru>
but that's a PITA
kakuzade has joined #nixos
pie_ has quit [Ping timeout: 240 seconds]
alex`` has joined #nixos
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra pushed 3 new commits to master: https://git.io/vAzNo
<NixOS_GitHub>
nix/master e8d53bf Eelco Dolstra: Revert "Enable sandbox builds on Linux by default"...
<NixOS_GitHub>
nix/master ddbcd01 Shea Levy: Fix restricted mode when installing in non-canonical data dir
<NixOS_GitHub>
[nix] edolstra closed pull request #1815: Fix restricted mode when installing in non-canonical data dir (master...data-dir-non-canon) https://git.io/vNiEH
NixOS_GitHub has left #nixos [#nixos]
ghuntley has quit [Ping timeout: 255 seconds]
ghuntley has joined #nixos
<gchristensen>
hi, manveru
pie_ has joined #nixos
<manveru>
oh damn
<manveru>
always the wrong nick :P
<gchristensen>
I've long-ago setup an alert on `grahamc` ;)
<gchristensen>
that is called Import From Derivation
<manveru>
:D
<manveru>
alright... so we need a package for yarn2nix finally
<manveru>
though i'm not sure how that'd work exactly without copying the .nix files
<gchristensen>
They smarts of yarn to nix might need to be copied in, indeed
<manveru>
MoreTea:
<manveru>
zimbatm:
robstr has joined #nixos
<kakuzade>
Hi. I tried to uninstall the clementine music player with nix-env -e but it still appears at the kde boot. Can it be uninstalled completely?
<manveru>
it's pretty stable these days... so i think we could finally do that
griff_ has joined #nixos
asuryawanshi has quit [Ping timeout: 260 seconds]
lord| has quit [Quit: WeeChat 2.0.1]
<manveru>
Kakuzade: remove it from your configuration.nix
<manveru>
gchristensen: thanks for the help, gonna get on it after lunch
zzamboni has quit [Quit: Leaving.]
humanoyd has joined #nixos
zzamboni has joined #nixos
m0rphism has quit [Quit: WeeChat 2.0]
<kakuzade>
manveru: I installed it in my user profile with nix-env -iA
<joko>
Is there any workaround to import from a derivation when nix is in restricted mode? As if the child derivation is calculated in the parent one (the one calling it), so that the latter does not have to access the nix store
Thra11 has joined #nixos
FRidh has quit [Remote host closed the connection]
<Lisanna>
joko curious, what is "restricted mode"?
<guibou>
what is "callPackageS"? I was using it instead of "callPackage", it is mostly similary, but not totally.
deepfire has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jfrankenau opened pull request #35335: libgudev: add gobjectIntrospection to nativeBuildInputs (master...libgudev-gobjectIntrospection) https://git.io/vAzx9
<guibou>
joko: I was using "callPackageS" since a will because of a typo error, and it was working... except that the resulting derivation wasn't provided with "override"
<joko>
guibou: callPackage is defined in lib/customisation.nix of nixpkgs, I couldn't find callPackageS
<guibou>
joko: pkgs/top-level/splice.nix
ma27 has joined #nixos
kakuzade has quit [Quit: Page closed]
mrkgnao has joined #nixos
Rusty1_ has joined #nixos
<joko>
guibou: this one seems to be from cross-compilation and I am not sure how to use it
pie_ has quit [Ping timeout: 255 seconds]
<guibou>
joko: do you think I should open a bug to inform that callPackages behaves a bit diferently than callpackage and that it may be surprising?
pie_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rbvermaa pushed 1 new commit to release-17.09: https://git.io/vAzhu
<NixOS_GitHub>
nixpkgs/release-17.09 883a4eb Rob Vermaas: imbalanced-learn: init at 0.3.2...
NixOS_GitHub has left #nixos [#nixos]
<dtz>
callPackages I thought was just callPackage but for sets of packages instead of a single one
<NixOS_GitHub>
[nixpkgs] edolstra pushed 2 new commits to staging: https://git.io/vAgvv
<NixOS_GitHub>
nixpkgs/staging 8d490ca Eelco Dolstra: libseccomp: Use multiple outputs
<NixOS_GitHub>
nixpkgs/staging f67b251 Eelco Dolstra: libatomic_ops: Use multiple outputs
NixOS_GitHub has left #nixos [#nixos]
yann-kaelig has joined #nixos
sdhandsucks is now known as sdhand
sdhand has quit [Changing host]
sdhand has joined #nixos
wrl has quit [Read error: Connection reset by peer]
wrl has joined #nixos
JosW has joined #nixos
ryantrinkle has joined #nixos
<guibou>
Apparently "ldd" package is not wrapped (well, it is able to find the library in the nix-shell, but it does not modify the rpath accordingly). Is there a wrapped version available?
<guibou>
"lld" sorry.
dnovosel has joined #nixos
rauno has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #35337: release-cross: minor cleanup, add aarch64-musl (master...feature/release-cross-minor-cleanup) https://git.io/vAgJV
NixOS_GitHub has left #nixos [#nixos]
nuncanada has joined #nixos
<clever>
guibou: it sounds like you need to add the library to the rpath of the program in question, with patchelf
<32NABLE54>
[nixpkgs] dtzWill pushed 2 new commits to master: https://git.io/vAgTI
<32NABLE54>
nixpkgs/master 76879b8 Will Dietz: release-cross: minor cleanup, add aarch64-musl
32NABLE54 has left #nixos [#nixos]
<32NABLE54>
nixpkgs/master 12f20b3 Will Dietz: Merge pull request #35337 from dtzWill/feature/release-cross-minor-cleanup...
reinzelmann has quit [Quit: Leaving]
ryanartecona has joined #nixos
pie_ has joined #nixos
tommyangelo_ has joined #nixos
rauno has joined #nixos
ravloony has quit [Read error: Connection reset by peer]
Arcaelyx has joined #nixos
<guibou>
clever: yep, indeed. I wanted to know if there was another solution.
zzamboni has quit [Quit: Leaving.]
pie__ has joined #nixos
chocopuff has joined #nixos
dnovosel has quit [Ping timeout: 276 seconds]
pie_ has quit [Ping timeout: 256 seconds]
<chocopuff>
hi, I'm interested in the .nix configuration of a system. I'm currently a Gentoo Linux user, and hope to apply for the Google Summer of Code program with them. An idea I had was to implement some of the ideas from nixOS's functional configuraion approach. If I understand correctly, everything is configured within the one file?
<chocopuff>
For me, so far, the appeal is that everything is centralized. The configuration defines a system, hence it would be easy to share configurations?
<ij>
Are there some attributes that add dependencies only when creating a nix-shell of the derivation?
<chocopuff>
Sorry if this is not the right place to ask though!
<manveru>
ij: maybe shellHook
<chocopuff>
I'm hoping to help make sharing a Gentoo users system configuration easier, and I really liked the nix approach
jtojnar has quit [Quit: jtojnar]
zzamboni has joined #nixos
<manveru>
chocopuff: well, in theory it's the right place, but it's gonna be ... challenging to replicate the NixOS approach for Gentoo
robstr has quit [Remote host closed the connection]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #35338: release-cross: add nix and nixUnstable (master...feature/release-cross-nixen) https://git.io/vAgIG
raynold has quit [Quit: Connection closed for inactivity]
robstr has quit [Remote host closed the connection]
robstr has joined #nixos
<manveru>
chocopuff: what might work is a partial approach though, kinda like nix-darwin works
ma27 has quit [Ping timeout: 252 seconds]
<chocopuff>
manveru: is *everything* configured through that?
<manveru>
the biggest issues i see are: gentoo packages won't be able to be built via nix, and the init system needs to be built via nix as well
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<chocopuff>
manveru: ahhh, I was thinking more along the lines of create a configuration file system for Gentoo, similar to nix
<manveru>
no, of course not, since you can't build a darwin kernel
<chocopuff>
not necessarily using nix
<chocopuff>
manveru: I mean, is everything configured through the .nix?
<gchristensen>
chocopuff: the root of the problem is you're playing the game Chef / Puppet / Ansible plays because of the fundamental differences between Nix and Gentoo's package manager
<chocopuff>
I'm assuming Chef/ Puppet / Ansible are irc handles?
<gchristensen>
no
<manveru>
:D
<chocopuff>
gchristensen: ah, can you clarify
<gchristensen>
in that: nix and NixOS is fundamentally declarative from top to bottom, and Gentoo is not. Chef / Puppet take a "declarative" language but then convert it in to commands that need to run to change the computer to match the declared system
<gchristensen>
Chef / Puppet are Configuration Management tools
<chocopuff>
I guess there is a reason why nixOS had to be created from the ground up
<chocopuff>
it definitely wasn't w/o reason then
<gchristensen>
yeah
<gchristensen>
every single time the NixOS configuration is changed, an entirely brand new system is built from scratch. there is some smarts to remove duplication, but the idea is it starts from scratch every single time. whereas, other systems have to change what they have to meet their desired goal
<manveru>
what NixOS does differently is that it declares the system, builds it, and then replaces the existing one with the new build by basically changing one symlink (and restarting systemd services)
<chocopuff>
gchristensen: Isn't that somewhat system intensive?
fendor has quit [Quit: Connection closed for inactivity]
<chocopuff>
ah, this is starting to make more sense to me
<maurer>
Not really - the declarative nature means that it can fairly aggressively cache things
<gchristensen>
chocopuff: Nix's ability to not duplicate work is very good :)
<maurer>
Unlike with gentoo where you may occasionally have to emerge -e to arrive at a stable state, nixos is always in a stable state, and knows whether a rebuild is required with precision
<gchristensen>
(and well defined, it isn't based on heuristics... but this is getting a bit in to the weeds)
<chocopuff>
oh, is this the "functional -> yay aggressive optimizations safely" thing
<gchristensen>
yeah, exactly
<LnL>
yeah, the only thing that happens _every_ time you rebuild is evaluation
<LnL>
everything else gets cached if nothing changes
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] dtzWill pushed 2 new commits to master: https://git.io/vAgLF
<NixOS_GitHub>
nixpkgs/master 6daf813 Will Dietz: release-cross: add nix and nixUnstable...
<NixOS_GitHub>
nixpkgs/master f1dfc22 Will Dietz: Merge pull request #35338 from dtzWill/feature/release-cross-nixen...
<chocopuff>
darn, I didn't expect to be some tempted :D
<chocopuff>
*so
<gchristensen>
:)
<chocopuff>
gchristensen: does this mean in fact, you actually can't go around poking in the physical system configs?
<maurer>
Not directly no
<gchristensen>
you can't
<chocopuff>
*it's not a good idea
<maurer>
they are read-only/immutable
<gchristensen>
/nix/store is read only, even to root, you really cannot go poking around
<chocopuff>
yeah, that makes sense
<LnL>
yeah, except if nix doesn't know about it
<LnL>
it won't fix things like database state magically for you
<gchristensen>
LnL++ if you manually put some config in to /etc/my-cool-config it can be changed, because Nix didn't create it and put it in the store
<chocopuff>
gchristensen: yes, I was thinking of /etc/my-config specifically
<gchristensen>
but it is trivial to put your own config in to /etc/my-config via nix, and get the protections
<chocopuff>
I mean, say you edit a config directly in /etc, doesn't that get wiped though when you rebuild
<manveru>
you can't modify it, it's linked to the read-only store
<maurer>
chocopuff: Sure, but any configs that are built by nix you won't be able to edit
<manveru>
so to modify, you have to go through your nix config
Myrl-saki has quit [Quit: WeeChat 1.9.1]
<chocopuff>
gchristensen, maurer: so say I wanted to edit /etc/X11/xorg.conf.d I edit it through my nix config right?
<gchristensen>
yep
<maurer>
chocopuff: Yes, though these days x autodetection is good enough that you'll rarely have to
<chocopuff>
ah good to see my impression get confirmed
<maurer>
chocopuff: A side benefit of doing things this way is that config reversion can be done through nix as well
<chocopuff>
maurer: funnily enough, that *is* the appealing part to me
<gchristensen>
yeah, perfect reverts
<chocopuff>
being able to do it all through the config
<chocopuff>
it kinda feels weird though
<chocopuff>
as if the standard hierarchy is somewhat useless with this kind of powerful abstraction
m0rphism has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] manveru opened pull request #35340: yarn2nix: Init at 0.1.0 (master...yarn2nix) https://git.io/vAgtg
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
chocopuff: you have that with e.g. opensuse as well, through yast iirc
<chocopuff>
well you still need to know your way around the standard hierarchy if stuff goes haywire
<maurer>
So, while I'd reccomend using nixos for what you're describing, if you really wanted to bring this stuff to gentoo, what I think would make sense would be a config language that generates a set of per-package use flags, plus a package set, plus some subset of /etc/ files, then triggers a double emerge -e
<maurer>
chocopuff: Atomic here means that you don't end up partway through configurations - you end up in one or the other
<maurer>
In gentoo, if I describe what my system should look like with use flags, and trigger an emerge
spear2 has quit [Remote host closed the connection]
<maurer>
I can end up in a state where some of my new packages are installed but not all, and the system does not function
<maurer>
And then I'm stuck dancing with revdep-rebuild
<chocopuff>
maurer: how oddly specific :P
erasmas has joined #nixos
<gchristensen>
chocopuff: I suspect many of us nixos users have at one point experimented with Gentoo :)
<manveru>
i used it for a few years :)
<maurer>
imo NixOS is basically Gentoo Done Right in the sense that Subversion is CVS Done Right
<maurer>
sadly whatever Git is in this analogy has not yet been invented
<sphalerite>
chocopuff: centralised config. But having just centralised config without the other stuff nixos provides isn't such a big win
<chocopuff>
gchristensen: I haven't really had that much experience with Linux distrors, have only been a user for slightly over a year
<manveru>
maurer: maybe once we get recursive nix...
<sphalerite>
Or even if it doesn't break, you're still in such states in between starting and completing the emerge
<gchristensen>
maurer: if Nix is CVS's SVN, I don't think I could handle the orders-of-magnitute-more-terrible interface that the Git of NixOS would be ... :P
<leo60228>
i have an optimus laptop, and i was only able to get it to run with bumblebee. i want to use vulkan, so i'm trying to figure out the proprietary nvidia drivers support for optimus (which uses nvidia for everything). how do i use it?
spear2 has joined #nixos
<leo60228>
my system didn't turn on when i set my driver to nvidia, i had to rollback
<sphalerite_>
cement: nix-env -f vim.nix -i
zzamboni has quit [Quit: Leaving.]
<sphalerite_>
cement: probably. Depends on what exactly your vim.nix looks like
<cement>
cool, thanks
<sphalerite_>
leo60228: there's an nvidia-prime module which may or may not help. I don't think it's in 17.09 though, only in unstable
<sphalerite_>
oh wait no I think it's not even been merged yet
Thra11 has quit [Ping timeout: 260 seconds]
<coconnor>
I think there is room for a nix desktop module distributed on top nixos that would best satisfy the "out of box desktop experience" customers.
<sphalerite_>
coconnor: I think nixos isn't really ready for that
<sphalerite_>
coconnor: one major missing piece for that is a graphical package (or rather configuration) manager UI
zzamboni has joined #nixos
<coconnor>
sphalerite_: a GUI package manager is unnecessary for progressing on this goal
<coconnor>
sphalerite_: and, based off how I've managed my own nix configurations, modules can be developed that simplify desktop configuration
<coconnor>
and, importantly, distributed separate from nixpkgs
ma27 has joined #nixos
rardiol1 has quit [Remote host closed the connection]
<sphalerite_>
at that poitn I imagine it looking like {pkgs, ...}: {services.xserver = { enable = true; desktopManager.plasma5.enable = true; }; hardware.pulseaudio.enable = /* just to annoy oxij */ true; environment.systemPackages = with pkgs; [firefox dolphin …];
<sphalerite_>
oh… author wrote "Big sorry guys, but that was quite a big hack and maintaince burden with a bunch of rough edges. Better to wait for proper solution from Nvidia."
<Profpatsch>
manveru: You mean copy the whole source into nixpkgs?
<Profpatsch>
I don’t think that’s a good idea.
jtojnar has joined #nixos
<manveru>
it might be possible to avoid copying the .js files... but can't avoid copying the .nix files :|
<manveru>
anw, gotta head out for some business now, maybe we can talk a bit later?
<Profpatsch>
sure
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] svsdep opened pull request #35344: keepass: fix auto-type and clipboard features (master...keepass-xdotool) https://git.io/vAgOH
<NixOS_GitHub>
[nixpkgs] YorikSar opened pull request #35346: linuxPackages.broadcom-sta: fix build with Linux 4.15 (release-17.09...release-17.09) https://git.io/vAgGF
<maurer>
Not fully on topic, but if you want to feel vindicated for bothering to do package management Right (e.g. sandboxes, repeatable builds etc) check out https://github.com/npm/npm/issues/19883
Itkovian has joined #nixos
<maurer>
viric: If you use nix on cygwin, you might know - has there been any progress on nix-on-windows (e.g. without cygwin or the linux subsystem)?
<maurer>
Last I checked the project had been scuttles
<maurer>
*scuttled
jtojnar has joined #nixos
<gchristensen>
maurer: was there ever a project to do that?
acarrico has quit [Ping timeout: 255 seconds]
tk_ has joined #nixos
<maurer>
gchristensen: There was a thing that could use mingw to do some stuff, but it did not do nearly enough to be viable for users
<fearlessKim[m]>
maurer: that's a fantastic link, I feel so bad for both the npm developers and their users xD
mmlb has joined #nixos
mmlb has quit [Client Quit]
<maurer>
gchristensen: Like, I think it was a patchset that did just enough to get nix able to build nix, but that was about it?
<maurer>
(the core issue being that mingw doesn't actually provide you with a posix IO subsystem)
<gchristensen>
oh I see
<fearlessKim[m]>
like the comment in the thread "Just mopping up my tears whilst reinstalling Arch for the 3rd time this month."
jtojnar has quit [Ping timeout: 264 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edolstra pushed 1 new commit to nix-2.0: https://git.io/vAgnV
<gchristensen>
Lisanna: everybody makes mistakes =)
<Lisanna>
yup
<Lisanna>
hrm, is there some tricky way to configure remote building so that two system features cancel eachother out? i.e., if you do a build on a machine with "x", then don't let any builds with "y" happen on that machine at the same time
knupfer has quit [Ping timeout: 255 seconds]
<Lisanna>
I'm thinking I'll probably have to write a custom build-remote.pl for that tho
ryantrinkle has quit [Ping timeout: 240 seconds]
magnetophon has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edolstra pushed 1 new commit to nix-2.0: https://git.io/vAgcY
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to master: https://git.io/vAglO
<NixOS_GitHub>
nixpkgs/master d328ba7 Frederik Rietdijk: Merge pull request #35243 from FRidh/postgis...
<NixOS_GitHub>
nixpkgs/master b2f53c7 Frederik Rietdijk: postgis: get rid of composableDerivation...
Itkovian has quit [Read error: Connection reset by peer]
<fearlessKim[m]>
nix repl ditches readline :'(
<manveru>
fearlessKim[m]: yeah :( my vi mode is gone
<fearlessKim[m]>
manveru: so lineoise has no vi mode, that was my only concern :'( the rest looks dope though
<manveru>
yeah, i probably will survive it... guess eelco just wanted to get rid of all dependencies
<manveru>
maybe there's a way to make it work via rlwrap later
knupfer has joined #nixos
<manveru>
can someone confirm that aarch-64 is arm64?
<manveru>
i have no clue about arm...
<sphalerite_>
manveru: main reason not to continue using readline is that nix repl is integrated into nix and that would mean that binary distributions of nix would have to be GPL-licenced
<sphalerite_>
manveru: can confrim :)
<manveru>
thx
acarrico has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
<manveru>
poor soul who has to compile electron on ARM
<sphalerite_>
fearlessKim[m]: probably also fiddly in terms of completion
<Ke>
I have 4 A72-cores on my router, it could very well be self hosting anyway
ryanartecona has quit [Quit: ryanartecona]
griff_ has quit [Quit: griff_]
<Ke>
mostly though it's about memory
<sphalerite_>
Ke: yeah especially for a router. I'm running nixos on my ARM chromebook and it's self-hosting with X11 and stuff, although it does take a day or two when there's a lot to rebuild
<sphalerite_>
That's on 2GB of RAM
<Ke>
sphalerite_: C201?
<sphalerite_>
yep
<Ke>
rtyping on mine =o)
<Ke>
though only compiling the kernel myself
<sphalerite_>
with nixos? ;)
jensens has quit [Ping timeout: 240 seconds]
<Ke>
no debian
<manveru>
rlwrap has completion, but one would need to write some perl to reimplement the nix-repl completion... or maybe just pass through the tab
<sphalerite_>
oh yeah, I haven't got a nixpkgs-built kernel running on it yet
magnetophon has quit [Remote host closed the connection]
<sphalerite_>
manveru: nix-zsh-completions could be helpful as a reference for that
sigmundv has quit [Ping timeout: 256 seconds]
<Ke>
sphalerite_: kind of sad that stock kernels never are in a for that is loadable by the vboot anyway
<Ke>
even if you had the image, you still have to package it yourself
<sphalerite_>
I run a mainline kernel
<Ke>
sure
digitus has joined #nixos
<Ke>
I run 4.9.82
<sphalerite_>
4.14.15 here
<sphalerite_>
for now
<Ke>
I bet you don't use crypto
<sphalerite_>
crypto what? the vboot stuff? Yeah permanently on developer mode mostly because I'm scared of flashing the firmware again :/
<manveru>
i might tinker with that when i got some time
<sphalerite_>
I plan to, eventually, but for a start I want to install to the eMMC.
<Ke>
the driver was broken on somewhere around 4.12
<manveru>
seems like i write a surprisingly large amount of perl recently ... :|
<sphalerite_>
did you have difficulties with the eMMC in yours too?
<Ke>
or not broken, but suffered bitrot due to other changes
<Ke>
I had difficulties some kernel versions ago, but haven't seen them in ages
robstr has joined #nixos
ma27 has quit [Ping timeout: 252 seconds]
<Ke>
I am planning to replace this unit with the new Samsung chromebook plus anyway soon
ryanartecona has joined #nixos
<Ke>
will be running 4.9 until EOL
<sphalerite_>
hm. Because for me it's very dodgy, depending on the alignment of the stars it will eventually just hang completely on any access to the eMMC. Which is not ideal when I'm trying to use it :)
<sphalerite_>
Or if I apply the tuning patch it gives I/O errors right from the start so the kernel gives up on it, which is how I'm using it now because it doesn't just spontaneously start hanging
<Ke>
oh right, I don't actually use eMMC for anything execpt for loading the kernel
nh2 has quit [Quit: Leaving.]
<Ke>
and yeah it's about the tuning thing
<Ke>
still it works well enough to write the new kernel
<sphalerite_>
the tuning patch being one that modifies the device tree to completely remove the tuning
<sphalerite_>
I just keep the kernel and the rootfs on the sd card
<Ke>
what's on the eMMC then?
<sphalerite_>
oooh but then you get to boot without pressing ctrl-U I'm guessing?
<sphalerite_>
Nothing atm
roberth has quit [Ping timeout: 256 seconds]
<samueldr>
sphalerite_: using gbb *something* (in chromeos) it's possible to change the default boot to usb IIRC
<Ke>
there is a setting for the boot order, you can have the external boot as default
pie___ is now known as pie_
<sphalerite_>
I was hoping to put chrome os back on it for firmware settings and debugging purposes, but it turns out the chromeos recovery won't work with libreboot
<Ke>
crossystem is also packaged for debian
<samueldr>
gbb_flags iirc
<sphalerite_>
yeah I've also been trying to package crossystem in nixpkgs, a bit fiddly though
<sphalerite_>
I guess I could make a debian chroot and use it from there
<sphalerite_>
good to know! I'll look into that
<Ke>
well I get screen full of IO errors with it
<Ke>
not sure if it's the debian or the libreboot
<samueldr>
I also reduced the timeout of the developer screen using it, and it doesn't beep anymore
<Ke>
but the old setting by chromeOS seemed to work anyway
<samueldr>
(though a CS10 chromebit for myself)
<samueldr>
(without alternative firmware, and no safe way to recover)
<Ke>
I think the alternative FW is the big reason to get these chromebooks
<sphalerite_>
yeah that's the main reason I went for this one
<Ke>
well you also get some security by obscurity
<Ke>
and low cost and good battery life
<sphalerite_>
I mean, I still ruin the liberty of the system by using the proprietary wifi firmware
<sphalerite_>
yes!
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #35348: data/mime-types: make derivation fixed-output (master...patch-88) https://git.io/vAgBO
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
and it's so small and light as well
<Ke>
yup
<samueldr>
being Free doesn't mean we *have* to use only free stuff, we have the *freedom* though to change that (imho)
<Ke>
even the cheap build seems pretty ok, when it's so light and small
<sphalerite_>
yes! and the keyboard isn't even as terrible as I'd expect from something this cheap
[0x4A6F] has joined #nixos
ma27 has joined #nixos
TonyTheLion has joined #nixos
<TonyTheLion>
hey! Is there a way to nixify a project that already uses cmake and has install targets, such that the build doesn't autoinvoke the install targets?
<Yaniel>
you can disable the install phase
<Yaniel>
or specify a custom one
knupfer has quit [Ping timeout: 265 seconds]
<sphalerite_>
TonyTheLion: should be pretty easy, modulo any nasty things that the CMake scripts might do
<TonyTheLion>
I set the 'installPhase = ""' in my script
<sphalerite_>
TonyTheLion: what do you mean though by "such that the build doesn't autoinvoke the install targets"?
<sphalerite_>
set it to ":"
<TonyTheLion>
but the cmake still attempts installation to dirs it has no access to
<markus1199>
Can someone help me debug a new package built? I linked all libraries, but I get '[20506:0222/182008.034488:FATAL:udev_loader.cc(38)] Check failed: false.' when running the executable.
<sphalerite_>
stdenv interprets the empty installPhase as "use the default installPhase"
<sphalerite_>
alternatively, you could set phases = [ "unpackPhase" "configurePhase" "buildPhase" ]
__Sander__ has quit [Quit: Konversation terminated!]
<TonyTheLion>
sphalerite_: basically that cmake runs the install targets and therefore tries to install in dirs that it has no access to
<sphalerite_>
TonyTheLion: usually cmake's hook should pass -DCMAKE_INSTALL_PREFIX so that it installs to the right place
<sphalerite_>
TonyTheLion: did you write a configurePhase manually?
<TonyTheLion>
ye
<sphalerite_>
Just adding cmake to nativeBuildInputs should be enough
magnetophon has joined #nixos
<sphalerite_>
if you need to specify any additional parameters to cmake, use cmakeFlags
<sphalerite_>
so just removing your configurePhase should also fix it
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] matthewbauer opened pull request #35349: qt5: install pkg_config files on darwin (master...qt5-darwin-pkgconfig) https://git.io/vAgRD
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<flox>
I am having trouble combining <pkg>.overrideAttrs with <pkg>.override, can someone provide an example of how to do that?
<LnL>
502 probably means it's updating the repos
<aminechikhaoui>
Anyone knows at which point is /sys/kernel/debug/tracing mounted in NixOS ?
<sphalerite_>
flox: (pkg.override {foo}).overrideAttrs (o: {bar}) should do it
<aminechikhaoui>
didn't find anything in nixpkgs so maybe it's auto mounted based on a kernel setting ?
Neo-- has joined #nixos
<aminechikhaoui>
oh there is a systemd unit sys-kernel-debug-tracing.mount
<sphalerite_>
yeah was going to say, that sounds like something systemd would do
<flox>
arg, missing the brackets on the override... so close....
<sphalerite_>
Oh how I love being able to have some complex nix command like nix build -f /nixpkgs --argstr system armv7l-linux --keep-going i3 and just be able to replace "build" with "log"
<gchristensen>
ack it seems to be because ofborg doesn't have a `master` branch
<sphalerite_>
hound bug!
<sphalerite_>
Would that be, like, a flea?
<gchristensen>
lol
<gchristensen>
given the severity (ie: it just crashes on startup) it seems like a much bigger problem than a flea
<sphalerite_>
killer fleas from outer space
<Yaniel>
something like ENOHEAD maybe?
<gchristensen>
ok I had to drop ofborg searching from hound, but it works now
<sphalerite_>
it should probably fetch HEAD and not master
<gchristensen>
sphalerite_: or support multiple branches
<sphalerite_>
or that. but that's cooomplicated
<gchristensen>
thanks for the ping, aminechikhaoui
<datakurre>
Would you know about "/nix/store/.links/... has maximum number of links" with btrfs? Is it harmful or just annoyance? I could not find definitive answer by Googling.
<gchristensen>
datakurre: appears to just be annoying
<datakurre>
gchristensen: Thanks. That one I found. Do you know what those .links -directories are?
<aminechikhaoui>
gchristensen: sure, thanks for fixing it :)
<gchristensen>
datakurre: if I had to guess, I'd say you probably have optimizeStore enabled, which creates a zillion hard-links in /nix/store/.links to do the optimisation
<aminechikhaoui>
meanwhile I still don't find which part is enabling sys-kernel-debug-tracing.mount
<datakurre>
gchristensen: That it is. Thank you.
<sphalerite_>
Ke: also, do you use an initrd for your debian boot on your c201?
<Ke>
yes, but only because of the crypto
<sphalerite_>
hm ok so I must be doing stuff wrong for initrds not to work for me :D
<Ke>
sphalerite_: there seems to be strict 32M size limit or something
<Ke>
I used to have huge kernels and initrds
<sphalerite_>
oh, even though the partition can be bigger?
<Ke>
yes
<sphalerite_>
pff what a pain. Helpful to know though, thanks!
<ramses_>
Hi guys, I'm having an issue running ansible on NixOS, presumably because ansible is directly pointing to /usr/bin/python instead of using env. Anyone ran into this before?
<gchristensen>
what are you trying to use ansible for?
<gchristensen>
and, how did you get ansible?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] adisbladis opened pull request #35350: go: Remove old version 1.8 (master...drop/go-1_8) https://git.io/vAguR
NixOS_GitHub has left #nixos [#nixos]
vidbina has quit [Ping timeout: 255 seconds]
<ramses_>
I'm on 17.09, I just included ansible in system packages
<ramses_>
I'm using it to run an ansible script against remote servers
kitemikaze has joined #nixos
<ramses_>
I'm aware that this is probably more of an ansible bug, but maybe someone has a workaround in the meantime...
<gchristensen>
sphalerite_: maybe it put the system in to a special mode where it can secretly run fun instructions
<TweyII>
sphalerite_: Memcheck?
kelleyNif has quit [Quit: Leaving.]
knupfer has quit [Ping timeout: 260 seconds]
raynold has quit [Remote host closed the connection]
Wizek has quit [Remote host closed the connection]
<sphalerite>
gchristensen: idk. It's really weird, and it consistently fails the first few times after a reboot then works as if nothing was ever wrong
<gchristensen>
sphalerite: is this on an armv8/7 box?
Wizek has joined #nixos
raynold has joined #nixos
<sphalerite>
TweyII: ^ leads me to believe it's not a memoey issue f that kind
<troydm>
could somebody help me, I'm trying to write defautl.nix for rtl8192eu, so I've checked out nixpkgs repo added folder to pkgs/os-specific/linux/rtl8192eu
<troydm>
inside I've created default.nix
<troydm>
but now I'm trying to build it and nix-shell can't find it
<troydm>
am I missing something?
<clever>
troydm: you need to edit a file somewhere to reference the new rtl8192eu directory
<clever>
troydm: grep the whole nixpkgs for the name of another directory in that region
<troydm>
clever: okey got it
rattboi has joined #nixos
<ramses_>
Can I change a file in the ansible package from within configuration.nix to change a hardcoded python path into a call to env? If it works I can make a pull request for nixpkgs.
<ramses_>
I'm quite new to all this so I'm getting a bit lost with all these override functions
blob__ has joined #nixos
yegortim1 has quit [Remote host closed the connection]
<blob__>
Hello, I'm having trouble playing Hollow Knight as it just freezes after a key press in the menu (audio continues). It seems to be kind of popular so I'd like to ask whether anyone here on NixOS 17.09 has the same issue? (default steam config, no useful logs)
yegortim1 has joined #nixos
<troydm>
clever: how do I get sha256 for git commit?
rattboi has left #nixos [#nixos]
<ocharles>
Is there anyway to add a derivation to the Nix store from my local filesystem from the command line? nix-store --add and nix-store --add-fixed sha256 don't work :(
<ocharles>
cp /nix/store/y1s7vcx57clxfs1ssazslsq42y43q880-lua-5.1.5.drv lua-5.1.5.drv ; nix-store --add-fixed sha256 lua-5.1.5.drv => /nix/store/7vivmrip1bhjg105mj9aav3as1pyh9cd-lua-5.1.5.drv -- notice that hash has changed
kelleyNif has joined #nixos
<ocharles>
I have something that locally generates derivations, but I can't actually add them to the store. Really don't want to have to link to libnix just for this
<troydm>
clever: nvm got it
robstr has quit [Remote host closed the connection]
<clever>
boomshroom: you can also put that line into a shell-32bit.nix and then just point nix-shell at the file
<boomshroom>
I only needed i686-unknown-linux-gnu-ld, but i686-unknown-linux-gnu-gcc should work as well.
zzamboni has joined #nixos
yegortim1 has quit [Remote host closed the connection]
sud0x3 has joined #nixos
contrapumpkin has joined #nixos
Synthetica has quit [Quit: Connection closed for inactivity]
<nhill>
So, I have 'services.fish.enable = true;' in configuration.nix so that I can use fish as a default shell. I'm trying to do a fresh install with this config and nixos-install
<taohansen>
Thought I'd check back if anyone knew of a preFetchPhase I can declare for a fetchgit derivation?
<nhill>
I get 'services.fish' does not exist
<boomshroom>
clever: thank you. That works for now.
<maurer>
nhill: Why would fish be a service?
nh2 has joined #nixos
<nhill>
Is there a way I should be wrapping this sp that is works
f0i has joined #nixos
<maurer>
nhill: Have you tried adding it to systemPackages instead? What goes wrong there?
<nhill>
maurer: It's also in system packages
<clever>
nhill: i think its programs.fish
<nhill>
I had to add that line to let me user it as the default shell
<clever>
fish isnt a service that runs in the background
<nhill>
Because otherwise it would not correctly load PATH
<nhill>
gchristensen: so, you're saying it chroots to /mnt and then can't find it?
<gchristensen>
that is my guess
<nhill>
I wondered about that but saw nixos-intall has a --chroot option, so thought it wasn't default behavior
<clever>
gchristensen: there is also a bug with the recent nixos-install, where it just does the nix-build in the host env, then copies the result over
<gchristensen>
I don't know, I'm just guessing
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb opened pull request #35355: nixos: disable sound by default, if stateVersion > 18.03 (master...sound-disabled-by-default) https://git.io/vAgi9
NixOS_GitHub has left #nixos [#nixos]
<clever>
so all paths refering to the rootfs, are the host root
<nhill>
clever: wouldn't that actually solve my problem though?
<clever>
nhill: what command did you run to cause the rrror?
<nhill>
clever: nixos-install
<nhill>
clever: well, first nix-channel to add nixos-unstable
<aristid>
nhill: funnily i think i made a similar mistake
<aristid>
i tried nixos-rebuild build -Inixpkgs=/home/aristid/othernixpkgs
<nhill>
I can easily finish this install, by commenting out the lines and then adding the unstable packages after first boot
f0i has quit [Ping timeout: 264 seconds]
<gchristensen>
probably the best bet
<aristid>
if i override NIX_PATH instead, it seems to work better
<nhill>
But, I'd like to be able to just install a system using my main config so it's easier to automate
<pie_>
sometimes when I rebuild krita I get "/tmp/nix-build-krita-4.0.0-pre2c.drv-0/krita-1b1695a/plugins/extensions/pykrita/plugin/pyqtpluginsettings.cpp:19:24: fatal error: ui_manager.h: No such file or directory #include "ui_manager.h"" and I cant figure out why
<pie_>
if it builds it builds but sometimes when i try again a few days later this happend
<nhill>
is -I nixpkgs=/nix/var/nix/profiles/per-user/root/channels/nixos-unstable
<nhill>
the right location?
scribbler has joined #nixos
<pie_>
*happens
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #35358: nixos/label: no prepend '-' if there are no tags (master...patch-100) https://git.io/vAgMw
NixOS_GitHub has left #nixos [#nixos]
<pie_>
ah what the heck, ill open an issue
mrkgnao has quit [Ping timeout: 260 seconds]
scribbler has quit [Quit: scribbler]
ottidmes has joined #nixos
mkoenig has quit [Ping timeout: 260 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb closed pull request #35355: nixos: disable sound by default, if stateVersion > 18.03 (master...sound-disabled-by-default) https://git.io/vAgi9
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb reopened pull request #35355: nixos: disable sound by default, if stateVersion > 18.03 (master...sound-disabled-by-default) https://git.io/vAgi9
NixOS_GitHub has left #nixos [#nixos]
<aristid>
oh man
<aristid>
i feel so embarrassed by my fumbling misclicking :D
<gchristensen>
aristid: you should see the mistakes I've made!
mkoenig has joined #nixos
zzamboni has quit [Quit: Leaving.]
zzamboni has joined #nixos
knupfer has joined #nixos
<aristid>
gchristensen: also i feel like there is somebody who would be better off with forking nixpkgs and removing all traces of anything Poettering :>
goibhniu has quit [Ping timeout: 248 seconds]
<gchristensen>
they did, they called in SLNOS: Suckless NixOS, maintained by the Church of Suckless Nix, and oxij is the pope
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c closed pull request #35358: nixos/label: no prepend '-' if there are no tags (master...patch-100) https://git.io/vAgMw
<simpson>
Oh, I don't know anything about that. But there's several forks of nixpkgs, and their contributors still care about nixpkgs, AFAICT.
<gchristensen>
manveru: it ... does? but I don't think it is on github, because it doesn't look like they've done much based on GH: https://github.com/SLNOS/nixpkgs
marusich has joined #nixos
<aristid>
gchristensen: so uh, another stupid question: your evaluation succeeds, but if i try to evaluate my own system config with it, it says: "The option `sound.enable' is used but not defined."
<gchristensen>
neat ...
<aristid>
not sure what i did wrong :D
<manveru>
oh well then... on to the yarn2nix issue
<aristid>
so if i // anything in there, it will add an ignored variable to the mkIf node
<aristid>
mkMerge should fix it *hope*
marusich has quit [Ping timeout: 248 seconds]
rauno has quit [Ping timeout: 260 seconds]
<manveru>
Profpatsch: so, what do you think a better solution to my PR would be? because i'd rather have it all in one place instead of fetching two little js files...
rardiol1 has left #nixos [#nixos]
marusich has joined #nixos
<manveru>
ideally there'd be a way to use external .nix files in nixpkgs... but until that's implemented i'm not sure there's much i can do to avoid duplication
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb pushed 1 new commit to sound-disabled-by-default: https://git.io/vAg9T
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/sound-disabled-by-default 473cfed Aristid Breitkreuz: nixos/alsa: fix wrong use of mkIf
<maurer>
Avoiding systemd is unfortunately a super heavy overhaul
<maurer>
I too would be interested in this, but it'd require a rebuild of how services work at _minimum_
<manveru>
do we have a RFC for switchable pid1 at least?
<nhill>
maurer: Well, if someone want to knock themselves out I'm all for it
<KABA>
UnicodeDecodeError: 'ascii' codec can't decode byte 0xf0 in position 1820: ordinal not in range(128) builder for '/nix/store/...-python3.6-pytest-mock-1.7.0.drv Anyone came across this error on current trunk?
<gchristensen>
manveru: I don't think so
<maurer>
Yeah, I'm pretty sure that's been the position of nixos for several years at this point
<maurer>
"If you want to make a configurable choice of init system, branch, send us a PR when you got it, this shit looks hard"
<nhill>
Would just be nice if their intention was to do it so that it could be merged back into the master branch
<NixOS_GitHub>
[nixpkgs] sconybeare opened pull request #35361: wrap the monkeysphere executable with the necessary crypto libraries … (master...master) https://git.io/vAg9P
NixOS_GitHub has left #nixos [#nixos]
zzamboni has quit [Client Quit]
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master ba58b42 Gabriel Ebner: neovim: fix bundlerEnv
<NixOS_GitHub>
[nixpkgs] gebner pushed 1 new commit to master: https://git.io/vAg9D
coot has quit [Client Quit]
<KABA>
gchristensen: Thanks, the problem is that I get this error while nixos-rebuild-ing, isn't the environment variables set automatically for python packages if used with the setup script?
<gchristensen>
weird :/
<KABA>
gchristensen: :q
<KABA>
:q
<clever>
KABA: /exit
<KABA>
sorry :/ Shame I couldn't find the focus
<manveru>
i think getting an alternative pid1 in wouldn't be too hard as a start, i just haven't worked on it yet because it seems like a waste of time if i'm the only one writing services for it :P
<clever>
manveru: have you seen not-os?
<manveru>
yes, several times
davidak has joined #nixos
<manveru>
unfortunately not-os still seems to have no docs :(
<lejonet>
gaaah, what am I missing now... I just want to create a frikin patch lol ontop of master, I have this interesting problem that nixops doesn't want to deploy because config.system.nixosVersion is missing (I'm guessing this is a new, upcoming change that adds it somewhere) but I can't seem to set it correctly
rauno has joined #nixos
<gchristensen>
lejonet: nixosVersion is set by nixops automatically when instances are created, if the variable is missing it is because something went wrong with the state
<aristid>
is there a way to ban a github user from commenting on my PRs? :D
reinzelmann has quit [Quit: Leaving]
mkoenig has quit [Ping timeout: 260 seconds]
<gchristensen>
aristid: please PM me
<lejonet>
gchristensen: ah I see, then I know where the error is, because I've changed a path in the deployment, thus I need to create a new deployment then too
<aristid>
gchristensen: ok
<gchristensen>
yeah
<fpletz>
aristid: I feel your pain :)
<lejonet>
gchristensen: Thanks :)
<gchristensen>
lejonet: happy to help!
alexteves_ has quit [Ping timeout: 252 seconds]
mkoenig has joined #nixos
<lejonet>
gchristensen: I had a hunch that something might get borked if I change the path :P
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
<lejonet>
gchristensen: is there any way to modify an existing deployment with a new deployment file or is the easiest way to just remove old and add new as a new?
<gchristensen>
hmm maybe check nixops modify? I don't know! :)
sud0x3 has quit [Quit: WeeChat 2.0.1]
<clever>
lejonet: i believe you can change the deployment file with nixops modify
<clever>
lejonet: and if the machine names match up, it will just upgrade them to match the new expressions
johnw has joined #nixos
<lejonet>
clever: that is what I hoped, I'm not entirely sure how to use nixops modify, is it like nixops create that you just do nixops modify <deployment spec> <deployment> -d <deployment name> ?
<clever>
lejonet: i think so
<lejonet>
Hmm, it still seems to whine about the nixosVersion thingy
<clever>
lejonet: ?
<lejonet>
so I think I'm not using nixops modify correctly then
<gchristensen>
from nixops?
<lejonet>
Yeah
<lejonet>
"error: attribute ‘nixosVersion’ missing, at /nix/store/82gjbw920v9an39ik2jp1kjqb77l4adk-nixops-1.5.2/share/nix/nixops/eval-machine-info.nix:284:107" is what I get when I try to deploy
mkoenig has quit [Ping timeout: 264 seconds]
<clever>
lejonet: what does `nixops info` say?
<lejonet>
clever: I get the same error there, but it continues to show the deployment and all info related to it
<lejonet>
So I think I didn't manage to update the deployment spec
<clever>
lejonet: can you add the 2 files on line 9 to the gist?
<lejonet>
clever: sure
<ottidmes>
I am still getting the following warning (warning because it seems to work fine) when trying to run 2 of the 4 deb binaries I repackaged: libssl.so.1.0.0: no version information available (required by synergy-tests), The only information I can find is that it could be because of having multiple versions of libssl.so.1.0.0 present on your system, but I just checked on a system with just one version of openssl
<ottidmes>
present, the warnings are still there
aminechi1haoui has joined #nixos
<lejonet>
clever: updated
pkill9 has quit [Read error: Connection reset by peer]
<clever>
lejonet: the ( on line 3 is missing the matching )
mkoenig has joined #nixos
<clever>
lejonet: and line 2 is missing a ;
aminechikhaoui has quit [Ping timeout: 248 seconds]
<lejonet>
clever: probably me that mistyped, let me double check file (pasting it into the gist apparently didn't work...)
<clever>
and it will update the gist (you will also want to gist --login)
<lejonet>
clever: fancy :O
<lejonet>
But yeah, was me that fatfingered stuff, I have { } and a ; at the appropriate place :P
<clever>
lejonet: can you also add configuration.nix?
<lejonet>
clever: sure
<lejonet>
clever: added
<lejonet>
the cli tool was a lot easier xD
<clever>
lejonet: everything looks just fine, try `nixops deploy --build-only --show-trace` and add it to the gist
pkill9 has joined #nixos
leo60228 has quit [Quit: Page closed]
<yorick>
how do I check if an option is set? hasAttr doesn't seem to be doing it
robstr has joined #nixos
<lejonet>
clever: its clearly not liking that I changed the spec file, updated the gist with it
<clever>
lejonet: where does <nixpkgs> point?, nix-instantiate --find-file nixpkgs
<ottidmes>
yorick: An option will have to be set unless it has a default, so you would check its default value (often an empty value "", [], {}, null) or you could check dynamic attributes with <some_attrset> ? <some_attr_name>
<yorick>
ah, I'll use types.nullOr types.string
<lejonet>
clever: /etc/nixos/nixpkgs, I'm running with a clone of my fork of nixpkgs
kelleyNif has quit [Quit: Leaving.]
<lejonet>
Which I just updated to latest from upstream/master and am trying to create a PR ontop of that, but want to deploy and test it first :P
<clever>
lejonet: try pointing that at a normal channel temporarily, and see if `deploy --build-only` works
<lejonet>
I'm guessing just commenting out the nixPath and doing a rebuild is enough to do that, or is there an easier way?
<clever>
lejonet: the nixops state doesnt come that early into the nix eval
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] florianjacob opened pull request #35364: nixos/version: Copy /etc/os-release from nix store (master...os-release_copy) https://git.io/vAgbe
<lejonet>
clever: I have no clue what nixops problem is :(
ryanartecona has quit [Quit: ryanartecona]
<lejonet>
clever: I think this could be related: "nix-instantiate --eval-only '<nixpkgs/nixos>' -A config.system.nixosVersion" -> "error: attribute ‘nixosVersion’ in selection path ‘config.system.nixosVersion’ not found"
<clever>
lejonet: sounds related
<lejonet>
and from what I've been able to find, nixosVersion seems to be built from something called filecontent releaseFile
<lejonet>
I'm guessing filecontent being some helper function that fetches something from the variable releaseFile, and maybe that has gone missing or something?
<clever>
lejonet: and do you have a line like this?
asuryawanshi has quit [Ping timeout: 264 seconds]
<lejonet>
clever: yep
<BlessJah>
Assuming I want to build zfs-only system (i.e. no systemd/fstab mounts), can I have zfs manage mounts?
<BlessJah>
or mountpoint=legacy is still preffered way?
<clever>
BlessJah: the nixos boot scripts expect your rootfs to be mounted with the standard mount command
<clever>
BlessJah: so at least the rootfs and /nix have to be via legacy
<BlessJah>
but those are mounted so early that it should hardly matter, am I right?
<clever>
BlessJah: yeah
<lejonet>
clever: is there any easy way to create a nix-repl that can access system vars and such?
<clever>
lejonet: put the configuration.nix stuff into a file, then nix-repl /path/to/nixpkgs/nixos -I nixos-config=/path/to/configuration.nix
Lisanna has joined #nixos
<BlessJah>
are zfs-managed mounts mounted automatically, or do I need to tweak something? I had zpool imported (boot.zfs.extraPool) that didn't seem to get mounted by zfs
<clever>
BlessJah: i'm not sure, ive found them to be a bit unpredictable
<BlessJah>
I'll use legacy on root and try extraPool on data then
griff_ has quit [Ping timeout: 255 seconds]
<lejonet>
clever: nixosVersion does indeed not exist in that context, but the appropriate config.system.nixos.<attr> attributes seems to exist, so it would seem like the "alias" or w/e its called isn't created
<BlessJah>
I have EFI partition on sda1, sdb1, and zfs mirror on sda2, sdb2
<BlessJah>
should I mdstat raid1 sda1/sdb1 or is telling grub to use sda,sdb as boot devices enough?
<clever>
BlessJah: if you are using EFI, then you dont need to set the device in grub
<clever>
BlessJah: that device field is only for legacy booting
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dywedir opened pull request #35365: tectonic: 0.1.6 -> 0.1.7 (master...tectonic) https://git.io/vAgAN
NixOS_GitHub has left #nixos [#nixos]
<clever>
BlessJah: and i dont think the bios supports mdadm based raid, so if you did raid sda1&sdb1, it would have to be done without any headers prefixed onto them
kgz has quit [Ping timeout: 260 seconds]
griff_ has joined #nixos
<BlessJah>
how will grub know to use sda,sdb (or sda1,sdb1)?
<clever>
BlessJah: EFI booting with grub just auto-detects whatever you mounted to /boot/
boomshroom has quit [Ping timeout: 260 seconds]
<BlessJah>
it scans all disks for ESP and collects all menu entries?
<lejonet>
clever: I'm just wondering why nixops is referencing config.system.nixosVersion, when it is clear that its defined as config.system.nixos.version, is it some legacy thing maybe?
<clever>
BlessJah: i believe the menu entries are programmed into the bios, with the efibootmgr program
<clever>
lejonet: yeah
<BlessJah>
what's the purpose of ESP then?
<lejonet>
BlessJah: to hold the loaders :)
<lejonet>
BlessJah: aka efi_stub enabled kernels, grub, etc etc, the actual efi programs is stored in the ESP
JosW has quit [Quit: Konversation terminated!]
nh2 has joined #nixos
kgz has joined #nixos
<BlessJah>
ok, to avoid XY: I want to install nixos on zfs on raid1, I think I need to create ESPs sda1,sdb1 and ensure both are identical (e.g. mirror or raid them), then create ZFS sda2,sdb2 mirror (raid1)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] oxij opened pull request #35366: nixos: properly disable sound by default, if stateVersion > 18.03 (master...pull/35355-no-sound-v2) https://git.io/vAgpa
NixOS_GitHub has left #nixos [#nixos]
<BlessJah>
in case one drive fails system will boot from the other drive ESP
<clever>
BlessJah: zfs supports its own raid'ing, so you dont need to put it ontop of mdadm or anything
<BlessJah>
yep, I know
<gchristensen>
it quite likes not being on top of raid
<clever>
BlessJah: i'm not sure whats the best way to deal with your ESP though
hiratara has quit [Ping timeout: 276 seconds]
<BlessJah>
yep, furthermore hardware raid might be SPOF (i.e. dead raid card has to be replaced with exact same model)
hiratara has joined #nixos
AluisioASG has quit [Remote host closed the connection]
AluisioASG has joined #nixos
<jonge>
hey there. i am trying to get my bluetooth to work reliably. unfortunately, `sudo ps aux | grep pulse` tells me that 1.) i and 2.) user "gdm" have an instance of pulseaudio running.
<jonge>
is there any way to stop the gdm user to start pulseaudio, too?
<clever>
jonge: try killing it with the kill command?
<jonge>
clever: yeah did that. restarts immediately.
<clever>
jonge: then something else running as the gdm user might be trying to use audio?
<jonge>
clever: it's only me on that system.
<clever>
jonge: also, have you changed pulseaudio.package?
<jonge>
clever: yes i set hardware.pulseaudio.package = pkgs.pulseaudioFull; but makes no difference
<clever>
hardware.pulseaudio.package = pkgs.pulseaudioFull; has to be set to get bluetooth support
<clever>
jonge: and you then need to connect with something like blueman
<jonge>
clever: connecting works. the problem is that i cannot switch between headset mode and a2dp mode.
<clever>
jonge: in pavucontrol, on the configuration tab, what options do you have?
<jonge>
the options are handsfree low quality and a2dp. but selecting them does nothing. i still have the poor sound all the time if at all.
<clever>
not sure then
<jonge>
so i can basically reconnect N times and then it is finally recognized as sound device at all. in which mode it then operates is again random. cannot switch.
<clever>
ive found bluetooth to be a pain to work with
<jonge>
yeah this is the last thing i really want to have working because then i am all set to use nixos as my 100% work system
<gchristensen>
nice!
kitemikaze has quit [Remote host closed the connection]
<jonge>
nixos is totally awesome, but i really need bluetooth working for meetings
<ottidmes>
jonge: And you had it working under a different Linux distro?
mkoenig has quit [Ping timeout: 240 seconds]
<jonge>
ottidmes: yes. it just worked under ubuntu
griff_ has quit [Ping timeout: 240 seconds]
kitemikaze has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vAgjp
<NixOS_GitHub>
nixpkgs/master a43e33d Aristid Breitkreuz: nixos: disable sound by default, if stateVersion >= 18.03 (#35355)
NixOS_GitHub has left #nixos [#nixos]
mkoenig has joined #nixos
<ottidmes>
jonge: That is good to know, that means it should be able to work under Linux, did you try the option I suggested? hardware.pulseaudio.systemWide I no longer need to use it, but with other hardware (not bluetooth related though) it caused it to work properly again
<jonge>
ottidmes: yes i am currently permutating different settings, also the systemwide one. unfortunately i have the same flaky behaviour all the time
<jonge>
can it be a problem that `pactl list` lists bluez5 and bluez4 modules?
dan_b has joined #nixos
<ottidmes>
jonge: Even though it connects, couldn't it still be a driver problem? Did you check the output on your NixOS compared to your Ubuntu (maybe try via LiveCD if no longer installed)
<jonge>
yeah i'd need to get a livecd again. haven't it handy now
<ottidmes>
jonge: On all my machines (except my servers) I needed hardware.firmware = with pkgs; [ firmwareLinuxNonfree ]; for at least some hardware driver, maybe that helps?
ericsagnes has quit [Ping timeout: 268 seconds]
<jonge>
ottidmes: oh that is something i have not tried yet. one moment, thx
<sphalerite_>
ottidmes jonge: hardware.enableAllFirmware = true; does the same AFAIK
<jonge>
this is an all-intel board. but i would try it, sure
simukis has quit [Ping timeout: 265 seconds]
<taohansen>
clever: I’m trying to implement your fetchGitPrivate strategy clever but I’m being foiled at every turn. I’d like if there was a way to just run nixbld with the same access and permissions as root user but I understand there are security concerns
<NixOS_GitHub>
[nixpkgs] fpletz closed pull request #35292: Pulseaudio by default, if sound is enabled (master...pulseaudio-by-default) https://git.io/vAur1
NixOS_GitHub has left #nixos [#nixos]
<jonge>
in bluetoothctl i can see that my bluetooth devices (i am currently trying 2 different ones) connecting, disconnecting, connecting, disconnecting...
nh2 has quit [Ping timeout: 264 seconds]
<sphalerite_>
!m dtz
<[0__0]>
You're doing good work, dtz!
jtojnar has joined #nixos
szicari has quit [Quit: szicari]
<shlevy>
clever: Just had an absurd idea that made me think of you: A FUSE filesystem overlayed on top of a normal filesystem at /nix/store, that lazily fetches things from some cache and populates the underlying real fs when some nix store path is accessed
<clever>
shlevy: though both of them require uncompressed .nar files in a directory to operate
dnovosel has quit [Ping timeout: 260 seconds]
hiratara has joined #nixos
kelleyNif has quit [Quit: Leaving.]
kelleyNif has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] grahamc closed pull request #35041: tree-wide: no pulseaudio by default (master...pkgs/no-pulseaudio-by-default) https://git.io/vACmS
NixOS_GitHub has left #nixos [#nixos]
rogue_koder has joined #nixos
shoogz has quit [Quit: shoogz out]
gil2a4 has joined #nixos
<ottidmes>
I have a systemd program that needs the DISPLAY env variable, what is the best approach for that? I now have Environment = "DISPLAY=:0.0"; which works, but I saw other packages in nixpkgs use PassEnvironment = "DISPLAY"; but that does not seem to work. I even have set environment.variables."DISPLAY" = ":0.0";
<sphalerite_>
yeah the deb file is Doing It Wrong™
<sphalerite_>
see above
alfie has quit [Ping timeout: 255 seconds]
davidak has quit [Ping timeout: 255 seconds]
aarvar has joined #nixos
<ottidmes>
sphalerite_: Thanks!
alfie has joined #nixos
<clever>
sphalerite_: when i first got into nixos, the synergy stuff was running as root, with $DISPLAY, and worked, due to an un-discovered security problem
<clever>
sphalerite_: the xorg was just allowing connections from any user
<clever>
when that was fixed, synergy broken, because it wasnt set to read /home/clever/.Xauthority
<skrzyp>
Should I expect the 18.03 release to be released at March 2018? I'm about to test and install it soon, but I'm not sure if I'll be prompted to upgrade right after installing
<sphalerite_>
skrzyp: yes it should be released during or slightly after March
<sphalerite_>
skrzyp: you won't get prompted to upgrade though. You'll just stop getting updates, if you use 17.09 >.>
<skrzyp>
is the upgrade planned to break anything beyond the regular upgrade procedure?
<mog>
skrzyp also upgrading nixos is not a very big deal
<mog>
compared to other distros
<ottidmes>
So... should I run it as a systemd user service, and somehow run it as root (that does not feel right, but does seem to require root access)
<mog>
the release notes explain what changes skrzyp but you cant upgrade into a failed state which is nice
<mog>
and if you upgrade and arent happy you can just roll back
<skrzyp>
okay, I often get into "breaking changes" case with many other software
<skrzyp>
nice, we'll see, I'm going to install it trough this weekend
<mog>
good luck
<gillmanash>
Hi. I am running Nix on Ubuntu. I am setting up a python env, which is user env. Issue is user env includes bin link to dbus, which screws with Ubuntu's dbus version
<mog>
come here if you get stuck
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 closed pull request #35330: go_1_10: init at 1.10 and set as default (master...go) https://git.io/vAzDS
NixOS_GitHub has left #nixos [#nixos]
<skrzyp>
I've used NixOS for a while about 2 years ago, but discarded it as I've seen it needed some "polishing" in terms of Nix package manager
<sphalerite_>
not sure if I'm understanding correctly though
<sphalerite_>
__monty__: what problem are you getting? Is it just failing to resolve the hostname?
<sphalerite_>
I mean, I don't really know much about how this stuff work
mkoenig has quit [Ping timeout: 256 seconds]
vidbina has quit [Ping timeout: 264 seconds]
<gillmanash>
spalerite_: Hmm, I would have thought it would be normal to want Python with some packages installed by default without dropping into a shell
<__monty__>
sphalerite_: I think so, tried it a couple days ago, had to switch back to the ip address.
<gillmanash>
sphalerite_: Although, true, Nipype is strictly a dev dependency so I should really have it separately
<__monty__>
sphalerite_: Ping has no problems resolving the names. Neither does opera.
mkoenig has joined #nixos
taktoa has quit [Ping timeout: 260 seconds]
<sphalerite_>
__monty__: I realise this is not really helpful but I'd probably just try and get a decent DHCP+DNS setup that lets your router resolve names through normal DNS
<simpson>
gillmanash: You may eventually come to not want anything in your default environment, and instead to have nix-shell incantations. It makes many things much easier.
tertle||eltret has joined #nixos
<clever>
sphalerite_: i run nixos on my router, with static IP's via dhcp, and dns pointing to those IP's
<TonyTheLion>
can I somehow pipe the output of 'nix-build -I ~' to a file?
<TonyTheLion>
the normal redir doesn't seem to work
<sphalerite_>
TonyTheLion: it's stderr so 2>
<sphalerite_>
rather than >
<TonyTheLion>
thanks :)
dan_b has quit [Ping timeout: 256 seconds]
<sphalerite_>
__monty__: I don't really know,maybe there's a difference in the env vars
<gillmanash>
simpson: one issue I have, for example, is making sure emacs has python with elpy and other editing tools. Should I have these as a dependency to emacs? I do already customise my emacs derivation
<sphalerite_>
__monty__: try runnning it in an environment as similar as possible to nix-daemon's maybe
<__monty__>
I think I'm not using nix-daemon. I have a single user setup.
<sphalerite_>
__monty__: oh yeah, it's probably also worth checking if you don't already have LAN DNS provided by your router, some consumer routers do that for you, often under a TLD like .lan (check /etc/resol.conf for a "search" line)
<gillmanash>
sphalerite_: simpson: I mainly ask because I am python.pkgs.xvfbwrapper maintainer, and want to check I haven't committed a design flaw by listing xorgserver (provides xvfb) as a propagatedBuiltInput
<__monty__>
The reason I use avahi is because it allows me to create containers with a really simple overlay-fs, this changes the container's host and gets me automatic "dyndns" for all my containers. I special-cased the nix-cache container to have a static ip for now but I'd love to get that ugly/unportable detail out of my config.
<skrzyp>
Is the high Emacs users' share in NixOS due to these nice integrations with Nix?
Wizzard has joined #nixos
<__monty__>
skrzyp: I assumed it was the other way around.
<simpson>
gillmanash: That sounds somewhat reasonable. I'm not sure why that wouldn't be wanted.
<gillmanash>
skrzyp: yeah one of the initial attractions for me was ability to have consistent complex environments, emacs can easily be one of those..
<gillmanash>
simpson: Yeah it makes sense, but I wasn't aware initially that doing so would cause non-propagted inputs of xorgserver to be put on PATH.
<simpson>
Xorg wants lots of stuff on its $PATH. It's a very ambient package.
<gillmanash>
yes..
<gillmanash>
simpson: how do you manage development envs. Do you have common ones e.g., in overlays that you load in, or per-packages as {default,shell}.nix?
<gillmanash>
I already have many per project ones, but not sure of best way to have common ones.
<clever>
gillmanash: for this project, the release.nix provides linux and darwin derivations for the nearby default.nix
<ashkitten>
Hey, how can I fix this issue? I'm trying to switch to nixos-unstable but this is keeping me from doing it - https://ptpb.pw/Acn8
<clever>
gillmanash: and line 4 blocks the config.nix, but overlays didnt exist when i made that, so its not covered
<gillmanash>
clever: Very interesting, thanks for sharing. I have been doing other hacks to try make sure my overlays aren't affecting projects
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 3 new commits to master: https://git.io/vA2tz
<NixOS_GitHub>
nixpkgs/master 4e4ae18 Franz Pletz: browserpass module: add firefox support
<NixOS_GitHub>
nixpkgs/master 7ea15ef Franz Pletz: nixpkgs module: update config example...
<NixOS_GitHub>
nixpkgs/master a04f624 Franz Pletz: nixos/manual: fix build...
NixOS_GitHub has left #nixos [#nixos]
<clever>
gillmanash: of you import <nixpkgs> { config = {}; overlays = []; } then all config.nix and overlays are ignored
<gillmanash>
Still, I find it useful to have a non-prject specific python I can drop into with basic reasearch software (numpy, scipy). I'm not a developer.
<clever>
gillmanash: you could maybe have a git repo of helper files for tasks like that
<TonyTheLion>
can't I reference the '${out}' variable in a build flag? Such as '-DFOOPATH=${out}/lib'
<TonyTheLion>
it complains not knowing what '${out}' is
<gillmanash>
For now, I'll remove nipype, and hence xorgserver from the env. But I am concerned that the design of python3.pkgs.xvfbwrapper is flawed
<gillmanash>
clever: I have a monstrosity of a dotfiles repo which serves as that
dan_b has joined #nixos
<sphalerite_>
TonyTheLion: with nix 2.0 (released today) you can use ${builtins.placeholder "out"}
<gchristensen>
I'm not sure 2.0 has been _released_ released...
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to master: https://git.io/vA2qU
<NixOS_GitHub>
nixpkgs/master 53dc96c Franz Pletz: Merge pull request #32323 from caugner/nextcloud-2.3.3...
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
gchristensen: :p
mkoenig has joined #nixos
<sphalerite_>
Ke: for me crossystem from debian mostly returns errors about mosys, would you happen to know how to fix those?
<sphalerite_>
Heh, I was going to ask if anyone knows a nice wrapper script around chroot that mounts sys proc dev etc for you… then I realised systemd-nspawn is basically what I want
<clever>
sphalerite_: oh, ive also got some systemd-nspawn scripts in my narfuse repo
<TonyTheLion>
hmmm, I thought doing '-DMYPATH=$mynixpackage.outPath' should give me the path of that package
<clever>
sphalerite_: also, i managed to hard-lock pid 1 when the fuse process crashed
<sphalerite_>
TonyTheLion: oh you mean if you want to refer to some other package? Just use ${package}
<TonyTheLion>
ye
<TonyTheLion>
that is what I meant
<sphalerite_>
TonyTheLion: so if you want to call hello from some package's builder (without putting it in the PATH using nativeBuildInputs, for some reason), use ${hello}/bin/hello