<clever>
you could also use nix-copy-closure to copy the current profile to another machine, and then run `nix-env --set -i /nix/store/thatpath` to overwrite your current profile with the copy
<fiatjaf>
mostly
<clever>
creating the derivation itself from nix is a bit more tricky
<oldandwise>
can we have the latest haskell 8.4.3 in nixos jellyfish? Or i'm stuck at version 8.2.2
jb55 has quit [Quit: WeeChat 2.2]
<fiatjaf>
I can't copy it because systems are different, one is x86_64 and the other i686
<fiatjaf>
I guess no one uses i686 anymore, right?
<clever>
fiatjaf: then youll want to use the above gist to create a mostly matching profile, that is 32bit (it will default to the same arch as the host nix-env was compiled for)
<fiatjaf>
yes, I understand the essence of that
<fiatjaf>
thanks
<fiatjaf>
oldandwise, I know nothing about nix, but maybe you could 'nix-env -i /nix/store/g5ggsbqkwqiiq3qkpw8ji42axrna117y-ghc-8.4.3/' ?
<fiatjaf>
(what would be wrong with that?)
<fiatjaf>
I'm using nix the package manager and I love the concept, but it really uses a lot of disk space and sometimes installing a simple tool takes a lot of time. is that normal? would things be better on nixos? what are the advantages of nixos besides the pureness?
<clever>
fiatjaf: you can regain some of the space it doesnt really need by running nix-collect-garbage
<clever>
without flags, it will only collect the unused things
<oldandwise>
fiatjaf: thanks, it works.
drakonis has joined #nixos
<{^_^}>
[nixpkgs] @samueldr opened pull request #48692 → Backports misc. fixes for the new ofborg eval time checks → https://git.io/fxwz5
<oldandwise>
so, the install was just a bunch of `copying ..`, so it's taking from binary cache, right?
<clever>
oldandwise: yep, downloading things from the cache
<oldandwise>
clever: but, if i do the same in my 32-bit netbook, it would recompile all from source, if no binary cache matches the checksum..!?
chreekat has joined #nixos
drakonis_ has quit [Ping timeout: 250 seconds]
<clever>
oldandwise: the hash in that path is a hash over the build instructions, which includes the arch
<clever>
oldandwise: so it will just blindly download the exact build you gave it (a 64bit build) which will then fail to start
<oldandwise>
clever: ok. So, each one has his own fork of nixpkgs github, point it as the channel. Do changes in personal repo, and then do pull request? is this the big picture?
mayhewluke has quit [Ping timeout: 250 seconds]
<clever>
oldandwise: no need to fork nixpkgs, if you properly use the nix expressions (rather then using the nix-env cmd fiatjaf gave), it will detect what arch its being ran on, and build for that arch (or download a prebuilt one)
<oldandwise>
clever: I meant, I want to be able to do my own nix packaging too..
<clever>
oldandwise: oh yeah, every contributor will have his own fork, where he makes some changes, and then files a PR
mayhewluke has joined #nixos
<oldandwise>
clever: i saw, how it `fixed` or patch small tools (gtest, imgcat), to prepare in the nix way.. so am understanding a little
Growpotkin has quit [Remote host closed the connection]
<sir_guy_carleton>
what's the difference with chroot and chrootenv?
jasongrossman has joined #nixos
fragamus has joined #nixos
<Arahael>
statusbot: I don't think chroot does anything special with the environment variables.
<Arahael>
statusbot: It's not even a security mechanism.
sb0 has quit [Quit: Leaving]
<Lisanna>
can I have the nix sandbox allow network access but still do all its usual filesystem namespace chroot stuff?
<clever>
Lisanna: declare what the output hash is and it will do that
<Lisanna>
clever I mean, apart from fixed-output derivations x_x
<Lisanna>
I mean it's mostly just for experimentation, I guess I could just let the drv fail
<clever>
nope
buckley310 has joined #nixos
iyzsong-x has joined #nixos
reinzelmann has joined #nixos
<sir_guy_carleton>
is it safe to use previous nixos generations for a difference release?
<kyren>
sir_guy_carleton: I'm not sure I understand what you mean, won't rolling back to a previous nixos generation mean that you're now on the previous release?
<sir_guy_carleton>
well, i have several release generations
<Unode>
hi all, I'm trying a 64bit app in wine64 and keep seeing: /nix/store/g240wk2bcrpdgin3zhd0s5ipwsxzizi3-wine-wow-3.0.2/bin/wine: not an ELF binary... don't know how to load it
<sir_guy_carleton>
i have serveral generation for 18.03 not yet gc'ed, and i want to check stuff there
<Unode>
I'm using: WINEPREFIX=~/.wine64 wine64 binary.exe - The error above is shown and nothing else happens
sir_guy_carleton has quit [Quit: WeeChat 2.2]
<Unode>
any suggestions?
nD5Xjz has quit [Ping timeout: 244 seconds]
<buckley310>
i would like to replace the default nvidia driver. i have coppied "pkgs/os-specific/linux/nvidia-x11" to a local path, modified it, and added the following configuration, but it seems to have no effect: nixpkgs.overlays = ([ ( self: super: { nvidia_x11 = (pkgs.callPackage /etc/nixos/nvidia-x11 {}).stable; } ) ]);
<kyren>
Unode: I don't actually know what's going on, but what happens if you run 'file' on the binary
<jasongrossman>
kyren: Thanks for your writing on Reddit.
<jasongrossman>
kyren++
<{^_^}>
kyren's karma got increased to 1
<Unode>
kyren: PE32 executable (console) Intel 80386, for MS Windows
<kyren>
jasongrossman: which writing haha?
<Unode>
hum...
<kyren>
Unode: huh, that's not what I expected
<Arahael>
kyren: Your reference to "the binary" is very ambigious. ;)
<Unode>
:)
<kyren>
I mean it is, but I meant 'binary' as in the 'binary' referenced in "not an ELF binary"
<Unode>
if you mean wine, it's not a binary. It's a shell scrippt
<Arahael>
Unode: In that case, the shebang line is relevant - that is, the #! line at the start.
<kyren>
right, I'm just trying to follow the trail to figure out what's printing "not an ELF binary", but I don't actually have wine installed atm and don't actually know what I'm doing :P
nD5Xjz has joined #nixos
<Unode>
ok guys, seems like there is something weird happening but in any case turns out the 'binary' I was running is actually not 64bits.
<kyren>
jasongrossman: oh, you're very welcome! I hope it's accurate
<Unode>
The installer was 64bits, the binary is 32... Lovely windows world...
<kyren>
well, I guess I'm glad I got you to run file on the OTHER binary in that case
<Unode>
I now moved the installed folder to ~/.wine32 and it worked
<Arahael>
Unode: That's odd, usually it's the other way around. :)
<Unode>
Arahael: the fact that these are mixed together is mind blowing enough :)
<Unode>
but in any case, there does seem to be some issue with wine64
<Unode>
need to look into a test-case to open an issue.
<Arahael>
Unode: "Lovely windows world" is a good description. :)
<kyren>
which build of wine runs both 64 bit and 32 bit binaries
<Arahael>
But... Consider that windows has maintained quite reasonable ABI stability over decades no w.
vk3wtf has quit [Quit: WeeChat 2.2]
<kyren>
I thought there was a build of wine that did that, but it's possible that was a hallucination
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<kyren>
sir_guy_carleton: sorry I didn't respond before you left before, you probably just figured out that you can just roll back across releases, there's nothing really special about release boundaries other than maybe some gotchas with system.stateVersion
<kyren>
if you live on nixos-unstable, every update can be like just as exciting as a whole new release
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #48695 → matrix-synapse: move from python2 to python3 [WIP] → https://git.io/fxwK5
<sir_guy_carleton>
kyren: that's what i figured. i just removed some programs i used nix-env to install after, just in case
mayhewluke has quit [Ping timeout: 245 seconds]
mayhewluke has joined #nixos
rprije has joined #nixos
rprije has quit [Ping timeout: 272 seconds]
hamishmack has quit [Ping timeout: 244 seconds]
jedahan has joined #nixos
robstr has joined #nixos
oldandwise has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Unode>
kyren: wine-wow runs both
<Unode>
but the 64bit one seems to have some issues
jedahan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rprije has joined #nixos
lawlesseel has quit [Remote host closed the connection]
hyper_ch2 has joined #nixos
lawlesseel has joined #nixos
rprije has quit [Ping timeout: 268 seconds]
rprije has joined #nixos
bohadi has quit [Quit: WeeChat 2.2]
bohadi has joined #nixos
bohadi has quit [Client Quit]
<Arahael>
"wow" is "windows on windows".
judson has joined #nixos
<judson>
I am currently trying to figure out if there's an existing NixOS module to add an extra loopback interface.
<judson>
I'm trying to set up a purely-local unbound config, but docker gets very confused.
<judson>
Because, if you don't know this joke yet, docker filters out localhost IPs from /etc/resolv.conf before copying it to containers - because obviously the containers don't have a resolver on 127.0.0.1
vbernat has quit [Quit: The future belongs to those who believe in the beauty of their dreams.]
<judson>
So what I'm planning is a dummy interface with a 172 address.
<Lisanna>
does remote nix building try to open multiple SSH connections to the target? ):
Itkovian has joined #nixos
Effilry has joined #nixos
alex`` has joined #nixos
Effilry is now known as FaerieFly
<elvishjerricco>
Lisanna: Yea, I think it does one SSH connection per derivation. But you can configure SSH to share a connection so that it transparently only connects once
<Lisanna>
elvishjerricco strange... in theory it's just building one derivation on the target, but I have a thing setup where after the derivation finishes building the host kills sshd, preventing further ssh connections. but hydra shows the build as always failing due to ssh connection problems
<elvishjerricco>
Lisanna: Huh. I would have thought it was only once per derivation. It's also likely that Hydra does it differently
<Lisanna>
elvishjerricco could also be that build vs. collect results are separate steps?
<Lisanna>
s/results/outputs
<Lisanna>
ugh this is so impure and gross
Czen has quit [Remote host closed the connection]
<elvishjerricco>
Lisanna: Yea, I wouldn't be surprised if Hydra did that. I thought that wasn't the case for normal distributed building though
<elvishjerricco>
Regardless, this would all be fixed transparently by just enabling ssh multiplexing.
Ariakenom has joined #nixos
bohadi has joined #nixos
<sir_guy_carleton>
wee i think i solved one breakage with 18.09
<Arahael>
If you configure ssh to reuse connections, you don't even have the tcp connection delay. :)
<{^_^}>
[nixpkgs] @Mic92 pushed 3 commits to release-18.09: https://git.io/fxwSQ
Thra11 has quit [Ping timeout: 272 seconds]
<elvishjerricco>
Is there any way to convince `nix eval` to treat derivations as just their outPath? I'd like to be able to e.g. eval a list of derivations without it recursing into all the attrs of the derivation, since that fails to eval.
<elvishjerricco>
--raw does this, but you can only use it if you're eval-ing one derivation.
<elvishjerricco>
Can't do it for data structures with derivations inside them
shabius_ has quit [Remote host closed the connection]
shabius_ has joined #nixos
shabius_ has quit [Remote host closed the connection]
<manveru>
i found it easier to simply filter the files i want to include instead of exclude :)
<srhb>
siers: Ah, okay.
<siers>
you can do that with "!patterns" in gitignore :P
* siers
will go on a hike now
siers has quit [Quit: WeeChat 2.0]
phreedom_ has joined #nixos
phreedom has quit [Ping timeout: 256 seconds]
rprije has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fxwAH
<thibm>
Is it possible to do a nixos-rebuild --target-host X switch but using X's nixos channel instead of localhost's? i.e evaluate configuration.nix on X
sigmundv__ is now known as sigmundv
<thibm>
If I understood well, nixos-rebuild is using root's 'nixos' channel. I think the easiest way to do what I want to do (but requires a manual step) is to override this. But I don't want to override localhost's root's nixos channel
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fxwpb
<TweyIV>
How does nix-store decide whether a path is alive? I have a package I built in `nix repl`, which now lives in my /nix/store. I'd like to use nix-store --delete /nix/store/*-my-package, but nix-store complains it's still alive, even though nix-store -q --referrers /nix/store/*-my-package gives no referrers
Acou_Bass has joined #nixos
<srhb>
TweyIV: referrers, in memory, garbage roots
<srhb>
Try -q --roots
<srhb>
Actually, roots will include memory roots too :)
<{^_^}>
[nixpkgs] @alyssais opened pull request #48700 → qt59.qtwebkit: fix build on Darwin → https://git.io/fxwhZ
<srhb>
Taneb: ^
<srhb>
Might be interesting.
<Lisanna>
thibm I am sleepig now, tbh that's all the potential insight I have on what you're trying to do :p if it doesn't work, ,it probably should. file a github issue against nix if it doesn't work and you're sure it's not you don't have SSH setup right
<Taneb>
srhb: thanks, I'll pass this on
<Lisanna>
'cause you can put https:// URIs in NIX_PATH stuff, so ssh:// or ssh-ng:// should also work in theory
<srhb>
Lisanna: It doesn't work, but I definitely agree that it should.
<thibm>
Lisanna: ok, many thanks.
thibm has quit [Quit: WeeChat 2.0]
<srhb>
(It just shells out to a fetcher, I think, and that fails parsing ssh-ng:// altogether, and fails with unsupported protocol on ssh://)
<srhb>
hrm. Can pls has pkgs.override
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fxwhM
<TweyIV>
srhb: Just says {memory:122}
<TweyIV>
But the nix repl is closed :(
<srhb>
TweyIV: lsof?
<TweyIV>
srhb: I ran lsof on everything on there and it didn't return anything
pie___ has joined #nixos
<srhb>
TweyIV: grepping for the hash?
<TweyIV>
srhb: What's the ‘122’?
<TweyIV>
srhb: In what? :þ It's literally something I ran in `nix repl` and then I closed the repl
<srhb>
TweyIV: AS in, sudo lsof to get all open files, and grep for your package hash?
<TweyIV>
srhb: No results :(
<srhb>
I suspect the 122 is something in /proc, but still trying to find out _what_
<srhb>
Like, a file descriptor or something
<TweyIV>
Yeah, that was my suspicion
<srhb>
Guess we need to check the source code...
pie__ has quit [Ping timeout: 244 seconds]
<srhb>
TweyIV: Specifically, the logic is in findRunetimeRoots in gc.cc
<TweyIV>
srhb: Looks like it means it's the 122th root
<TweyIV>
122nd, rather
<srhb>
Uh, where'd you get that from?
<srhb>
That sounds wrong
<TweyIV>
size_t n = 0; for (auto & root : findRuntimeRoots()) roots[fmt("{memory:%d}", n++)] = root;
<TweyIV>
findRootsNoTemp
<srhb>
Huh. That seems not very useful.
<TweyIV>
That seems to be the only place that constructs the "{memory:%d}" string
<TweyIV>
Yeah :þ
<srhb>
It is. But I assumed the %d was coming from findRuntimeRoots somehow
<srhb>
But yes, I agree with your interpretation.
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell-fgl: ignore the new cabal revision published on Hackage »: https://git.io/fxwj5
<srhb>
ah, so that's the NAME of the root.. Wonder if we can query that somehow
simukis has joined #nixos
<srhb>
TweyIV: Can't find any good way short of modifying nix itself, sorry. I bet it's something silly like a CWD for some shell though.
<TweyIV>
:(
<TweyIV>
Alright, thanks!
<TweyIV>
It's not too much of a problem, I assume it'll go away eventually
<srhb>
TweyIV: Reboot should do it dead at least :-P
<Mic92>
should contain something like: /run/booted-system/kernel-modules/lib/modules
<domenkozar>
hmm can't deploy busybox
<domenkozar>
sec
<Mic92>
You rather want binutils.
alana has joined #nixos
<Mic92>
If this is related to strings
hyper_ch2 has quit [Quit: Page closed]
<domenkozar>
ah
<domenkozar>
this might be because it's a nixos container
vaibhavsagar has joined #nixos
<Mic92>
yes
<Mic92>
you can if you really want to also loading kernel module there.
<alana>
So first time installing nixos, when I run nixos-install it's saying "attribute 'filesystems."."' .../hardware-configuration.nix:25:3 already defined at .../hardware-configuration.nix:15:3" Could someone tell me what that means?
<Mic92>
But I would not recommend it
<domenkozar>
Mic92: there you mean on the guest?
<domenkozar>
I did
<domenkozar>
modprobe ip6_tables on host
<srhb>
alana: Sounds like you got some mounts wrong possibly
<domenkozar>
seems to work now
<srhb>
alana: "." should not normally be generated.
<Mic92>
domenkozar, yes, just put it in boot.kernelModules
<srhb>
alana: Where did you mount your new filesystems?
<domenkozar>
Mic92: shouldn't that be the default anyway?
mayhewluke has quit [Ping timeout: 268 seconds]
<Mic92>
domenkozar this should be the case if you have the firewall enabled there?
<domenkozar>
at least with firewall enabled
<domenkozar>
hmm
<Mic92>
ip6tables -S
<Mic92>
should contain rules then
alana_ has joined #nixos
<alana_>
Ouch that sucks
<alana_>
:(
<domenkozar>
Mic92: right :)
<domenkozar>
thanks for the help
<Mic92>
alana_ can you provide the whole example file? you have defined a mountpoint twice
<srhb>
Mic92: Notice that the mount point is "."
alana has quit [Ping timeout: 256 seconds]
<Mic92>
yes. this also looks very wrong
<alana_>
Mic92: Example file as in console output?
<joko>
Does anyone know if it is possible to define a network in systemd-networkd to be manually enabled? I would like to setup wireguard as a client and I would like to be able to manual enable it
<Mic92>
joko: I think you have more luck with wg-quick for your particular use-case.
<Mic92>
networkd does not support on demand networks
<joko>
Mic92: ok, thanks, I will have a look
<joko>
Mic92: speaking of your configuration, I got reminded of a setup of yours with LXC containers in nixos. Do you have any tutorial / guide I could look into?
rprije has quit [Ping timeout: 244 seconds]
rprije has joined #nixos
jabranham has joined #nixos
<joko>
I would start from lxd, even tried to update it to a more recent version, but it seems to require a distributed flavour of sqlite and might be harder to use it at this point
<Mic92>
joko: I used to have legacy archlinux containers that got now replaced by nixos services. What is your use case for LXC. In some older revisions the code is still there.
<alana_>
I realy appreciate yalls help but its rly late and I have a stats test coming up so I'll work on this later if yall dont mind
<alana_>
sorry!
<alana_>
byeeee I love yall <3
<srhb>
alana_: Good luck.
<FaerieFly>
good luck on the test!
<alana_>
thanks <3
alana_ has quit [Quit: Page closed]
<Mic92>
joko, I probably would also go for lxd these days
<joko>
Mic92: well, I would like to fire up some Ubuntu containers for mainstream stuff until I port them to NixOS
<joko>
Mic92: roger, I would try my luck again with lxd
jabranham has quit [Quit: ERC (IRC client for Emacs 27.0.50)]
jedahan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<aminechikhaoui>
What's the situation of Nix installer on MacOS high sierra ? just got a weird issue while installing nix for a friend, the installation is barfing about missing libunc.dylib path :/
<acowley>
Is there a straightforward way to have a src or srcs attribute that includes multiple fetch* expressions and dumps all the source files into one directory?
<aminechikhaoui>
ah wait it's not missing but the error is file system sandbox blocked stat()
<acowley>
I'd have thought the installer was pretty solid on high sierra unless some recent mojave changes broke it on HS
dbmikus has quit [Quit: WeeChat 2.2]
<acowley>
I think I've run it on two HS machines without issue, but not in the mast two months
<kai_w>
Why does `(python3.override { packageOverrides = (self: super: {});}).pkgs` not have a `dotmap` attribute if `python3.pkgs` does?
<samueldr>
ah good, I was worried since I didn't see any breakage here :)
<samueldr>
though this is relevant to a need where I pin 18.03 and have issues with "special" characters (which 99% sure is due to that whole thing)
kiloreux__ has quit [Quit: Leaving]
revtintin has joined #nixos
kiloreux__ has joined #nixos
kiloreux__ is now known as kiloreux
<kiloreux>
What's the best nix language parser library ? I couldn't find a decent one in python/c++/c ? Could someone please recommend one to me ?
<bpye>
Uh, how should I be enabling ssh in the installer?
<bpye>
ssh server that is
<samueldr>
> If you would like to continue the installation from a different machine you need to activate the SSH daemon via systemctl start sshd. In order to be able to login you also need to set a password for root using passwd.
<{^_^}>
error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):203:1
<tilpner>
kiloreux - Check if you can use libnix*. There's also hnix
<samueldr>
kiloreux: using `nix-instantiate --parse` you can get an `Aterm format` representation of the ASTs
<samueldr>
(if a full blown library doesn't exist in the ecosystems you want)
<samueldr>
that format maybe has bindings already in your ecosystem, hopefully
bohadi has joined #nixos
<bpye>
Ack thanks samueldr
Ariakenom has quit [Ping timeout: 260 seconds]
<samueldr>
(btw, don't take the link to the manual as "read the fine manual", but more as a citation :) )
v0|d has quit [Remote host closed the connection]
<buckley310>
how can I override the nvidia driver in NixOS? It is apparently possible since someone this github user got it working, but I cannot, maybe this solution is outdated? https://github.com/NixOS/nixpkgs/issues/34938
<{^_^}>
#34938 (by ck3d, 35 weeks ago, closed): nvidia_x11 override via overlay not possible
<arianvp>
with nixops, is it possible to get the IP of a VM inside the nix derivation?
<arianvp>
such that my nixos config can for example depend on that IP for like configuring a load balancer?
<clever>
arianvp: yeah, i think it was networking.publicipv4 or something like that
Tucky has quit [Quit: WeeChat 2.2]
<arianvp>
how does it do that though? how does it figure that out during instantiation?
effsquared has joined #nixos
<arianvp>
I thought nix modules couldn't have side-effects etc
o1lo01ol1o has joined #nixos
<clever>
arianvp: nixops will get the IP from the vm system, before it builds the modules
<arianvp>
but what if the vm doesn't exist yet?
<clever>
when deploying, it will create the vm, query its state, then build the nixos with that ip as a param
<clever>
if the vm hasnt been created, the param will be null
<o1lo01ol1o>
how is scaling with respect to buildCores and maxJobs? If I have a 72 core system will setting buildCores = 72 scale sublinearly?
<clever>
o1lo01ol1o: it will run `make -j72` inside each build, and it will depend on the deptree within a given package, and if the package even has enableParallelBuilding = true;
<clever>
maxJobs is at the `nix-build -j5` layer, and will run up to 5 packages in parallel
<bpye>
My SSH connection seems to be hanging before I get a shell, I'm not sure what's going on...
<bpye>
Before the password prompt even
<bpye>
Wait no, that's just a broken ssh-agent
<bpye>
My bad
<o1lo01ol1o>
clever: thanks. So I guess I'll just have to wait and see if having 72 cores is significantly better than 32.
<arianvp>
meh the nixops docs are very ... vague. They do not distiniguis between options and config
<arianvp>
so I thought I had to _set_ networking.publicipv4
<arianvp>
:/
<arianvp>
but it's calculated
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
arianvp: this will generate nixos config, based on python variables
<clever>
behind the scenes, nixops generates an extra deployment file with that kind of state, and adds it to the nix eval
<arianvp>
aah
<arianvp>
so does the python like 'parse' the nixops modules, then generate more modules, and only thenafter is the entire set of modules eval'd?
<arianvp>
I thought the calling to python happend during lib.evalModules so this is why I was confused how it got values back into the nix world
<clever>
it does a first nix-instantiate pass over things, with --to-xml
<clever>
python then parses that xml, to find the names and cloud-hosters of every vm, and what resources must exist
jabranham has joined #nixos
<arianvp>
aha, so it is indeed a two-phase system
<clever>
then it creates all of those resources, and generates a nix file with the state
<clever>
and then does a nix-build to build the machines
<arianvp>
aha thanks. that helps
<clever>
if you do `nixops deploy --build-only` and have never created, then it will just not set things like publicIPv4, and they will default to null
zduch4c has joined #nixos
<zduch4c>
any Emacs EXWM users here? how can I inhibit NixOS from loading the default EXWM config?
<jabranham>
I'm trying to write a systemd user service and having problems. The executable file it starts has the shebang: #!/usr/bin/env Rscript which works fine when I execute it but when I do systemctl --user start <my-service-file> it can't find Rscript
<arianvp>
jabranham: did you write the unit file manually or is it in nixos config?
lonokhov has quit [Quit: WeeChat 2.2]
<jabranham>
arianvp: manually
<arianvp>
if manual, that is because services are executed with a different environment than your shell
fragamus has joined #nixos
<arianvp>
env looks at PATH , which might be different inside the unit file
<zduch4c>
so… does NixOS by default execute default config of EXWM on start if set by services.xserver.windowmanager?
<arianvp>
you'll have to add this:
<arianvp>
sec
<arianvp>
are you on nixos or on just a nix machine?
<infinisil>
Judson: Crazy idea: Have a debug NixOS where every binary gets wrapped in a binary that outputs debugging information
<infinisil>
Via an overlay of course, and provide a binary cache for it
nkaretnikov has joined #nixos
<cransom>
i'd settle for (and it's not hard to do) just to write a wrapper that drops you into a systemd.service environment so i can see what the service sees.
<maurer>
infinisil: Would something like gentoo's splitdebug not be more appropriate?
<infinisil>
maurer: I have no idea what that is
<maurer>
infinisil: Gentoo can install debug information "separately" so that it is not part of the main binary, and gdb can be told where to pick it up
<maurer>
infinisil: now that we have multiple output derivations, this could be feasible for us as well
<infinisil>
Ah I see
<avn>
Folks, anyone use pylint with python27 within release-18.09 branch?
<maurer>
(the idea is that you can e.g. put your debug information on spinning rust, while keeping your ELFs on an SSD)
<nkaretnikov>
hi, i'm trying to package a haskell program and library. until now, cabal2nix has mostly worked just fine for me. with this one, however, i'm getting `sh: configure: No such file or directory` when building the library: (shell.nix of the executable makes use of shell.nix of the library). has anyone seen anything like that? it has a bunch of c files too, so not an easy thing to package, i suppose.
<nkaretnikov>
i can upload my nix files if anyone wants to take a look
<nkaretnikov>
ah, and that's on macos with nix, not nixos. just thought i should mention.
bohadi has quit [Quit: WeeChat 2.2]
drakonis has quit [Ping timeout: 268 seconds]
drakonis_ has joined #nixos
asymmetric has joined #nixos
Thra11 has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
<nkaretnikov>
ah, apparently it has to do with the fact that the package uses `Distribution.Make` rather than `Simple`.
worldofpeace has quit [Remote host closed the connection]
copumpkin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<hyper_ch>
ahmmm....... error: The option `services.znc.confOptions.networks.freenode.userName' defined in `/etc/nixos/configuration.nix' does not exist
<samueldr>
(the options.html listing lists "current-stable" only)
<hyper_ch>
this looks now so complicated... thx :)
mayhewluke has quit [Ping timeout: 245 seconds]
<electrocat>
ah, nix doesn't automatically do cross compilation when building systems not of the current architecture
<infinisil>
hyper_ch: Whoops, I'll look into it
<hyper_ch>
infinisil: why you need to look into it?
<samueldr>
infinisil: though you already have taken care of it!
<samueldr>
even wrote the doc in the release notes :)
<lunik1>
Hi I'm trying to bootstrap nix on a RHEL 6 derived system. I've managed to build nix itself, but when I then use that to try and build nix from nixpkgs many libuv tests fail (but libuv builds fine independeltly)
mayhewluke has joined #nixos
<infinisil>
Ahhh it's one of the removed ones, I see
<samueldr>
luck of the draw :)
<infinisil>
Phew
mekeor has joined #nixos
goibhniu has quit [Remote host closed the connection]
<Ralith>
what's the proper way to define a derivation source that's composed of multiple git checkouts?
<kyren>
Thra11: you can copy a closure to a NAR file, but afaik there's not a nice simple command for it, you can read about it in the man page of nix-store
<kyren>
specifically in the section about `nix-store --export`
<kyren>
Oh, wait.. reading comprehension, you said on a system without nix. Never mind, I don't think that exists
<Thra11>
Looks like hydra's aarch64 build of rustc 1.29 got oom killed or something (https://hydra.nixos.org/build/83020833/nixlog/1). Is somebody able to poke hydra to try again for transient failures? (My system succeeded in building it locally, so it doesn't seem to be actually broken)
<bpye>
makefu Is there any disadvantage doing it that way?
<makefu>
bpye: it does not spawn another instance of the task, but trigger a systemd service to start. if the job is still running nothing will happen
<bpye>
makefu Ah excellent, and if the job finishes successfully I assume it'll be run again?
nikivi has joined #nixos
<makefu>
bpye: you are losing some of the flexibility which are provided by timers
<makefu>
bpye: i am not really sure, my initial guess would be 'probably not' but maybe systemd timers provide some logic to achieve this
fragamus has joined #nixos
<bpye>
Oh I don't mean immediately, I mean when the timer is next hit - not a one shot but recurring
<Ralith>
samueldr (IRC): that is spectacularly poor documentation :/
emily has joined #nixos
<elgoosy>
hi, I have this in my hardware-configuration https://pastebin.com/TasaArpb How do i make it accesible for a normal user (not root)?
<makefu>
bpye: ah yes, according to the man page "Note that in case the unit to activate is already active at the time the timer elapses it is not restarted, but simply left running. There is no concept of spawning new service instances in this case."
<infinisil>
Actually, startAt just generates timer units :)
<bpye>
Ah, just a neat shortcut, thanks
<bpye>
I think #nixos has to be the most helpful Linux distro channel :)
run500 has quit [Remote host closed the connection]
<bpye>
I could pass it all on the command line but access tokens and such make it quite silly
<makefu>
setting config in /etc is really no problem, you can use environment.etc."rclone.conf".text
aleph- has joined #nixos
<elgoosy>
makefu: better adding allow_other in hardware-configuration, right? Mount command should be somehow hooked so that it executes every time and turn system on
<Ericson2314>
what's the easiest way to load the fixed nixos config in nix repl?
<grp>
fixed config?
<Ericson2314>
(`nix repl '<nixos-config>'` or whatever that NIX_PATH entry it is what I beleive just give the rool module file)
<grp>
nix repl '<nixpkgs/nixos>'
<Ericson2314>
grp (IRC): alright! thanks!
<nkaretnikov>
how do i affect c_include_path when in nix-shell? i added a c library to a haskell package with librarySystemDepends, but #include <libname> still fails to find it when inside the shell.
Lears has joined #nixos
[Leary] has quit [Ping timeout: 252 seconds]
<vaibhavsagar__>
nkaretnikov: your shellHook needs to set `C_INCLUDE_PATH`
<vaibhavsagar__>
including your library in `librarySystemDepends` only affects nix-build, not nix-shell IIRC
trevorriles has joined #nixos
hyper_ch has quit [Remote host closed the connection]
<nkaretnikov>
vaibhavsagar__: okay, any examples of that? i think i did it a couple of times, but that was a long time ago. ideally, i would like to provide the same experience via nix-shell and nix-build. without fiddling with nix-env etc.
<nkaretnikov>
here's how i'm testing: created /tmp/test.c with #include <libname.h> and main, then do gcc -E /tmp/test.c in the shell
<nkaretnikov>
i'm doing this^ however, nix-shell complains about librarySystemDepends being undefined, is it not the right way to do it?
<{^_^}>
[nixpkgs] @matthewbauer pushed commit from @Tmplt to release-18.09 « nixos/compton: fix corrupt colours with Mesa 18 on AMD »: https://git.io/fxoTm
<nkaretnikov>
... because i seem to see the same in the generic-stack-builder.nix
sigmundv has joined #nixos
<clever>
nkaretnikov: you want rec there
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
> rec { a = 5; b = a; }
<{^_^}>
{ a = 5; b = 5; }
<clever>
> { a = 5; b = a; }
<{^_^}>
{ a = 5; b = 10; }
<clever>
> a
<{^_^}>
10
<clever>
an old a somebody else made
elgoosy has quit [Remote host closed the connection]
Mateon2 has joined #nixos
Mateon1 has quit [Ping timeout: 246 seconds]
Mateon2 is now known as Mateon1
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
fragamus has joined #nixos
<nkaretnikov>
clever: yeah, it fixed that, but i still can't figure out how to export env vars: i do export Foo=foo in shellHook and env | grep Foo returns nothing in the shell
<clever>
nkaretnikov: every attribute on the derivation is automatically made into an env var
<clever>
nkaretnikov: what nix-shell command are you running?
<nkaretnikov>
nix-shell --pure -j1
<nkaretnikov>
with NIX_PATH=nixpkgs=$HOME/nixpkgs in front
<nkaretnikov>
so, if i set directly as in Foo=foo, then the generic c builder complains
<clever>
nkaretnikov: the .env at the very end accesses the env for building haskell packages, which ignores most things you pass to mkDerivation
<nkaretnikov>
okay, so how do i make it work?
<clever>
you will want to do `lib.overrideDerivation drv.env (old: { buildInputs = old.buildInputs ++ [ portaudio ]; })` in the area between `then` and `else`
<catern>
hey #nixos, packages from Nix notably don't dynamically load NSS modules, which means you have to use nscd, right? What should I do when I'm in a container, and don't run anything but a single process? Do I still have to run nscd or what?
<clever>
catern: basic dns and /etc/hosts still work without nscd, you just loose advanced things like avahi
<nkaretnikov>
clever: wait, isn't there a way to keep everything inside f?
<Ericson2314>
config in submodule is for the current submodule, which isn't very useful
<clever>
nkaretnikov: the .env attribute is rather basic
<Ericson2314>
It's annoying to manually map over the attrsOf or listOF
<Ericson2314>
is there any other way I can do my per-enty config and then let the module system's combination do the reducing work?
[Leary] has joined #nixos
<clever>
Ericson2314: i think what you do is mapAttr or attrsOf over the entire config, outside of the submodule, when assigning other config (like systemd services)
<clever>
Ericson2314: this for examples defines some submodule options within fileSystems, then it just maps over a the result of running filter on all fileSystems
<Ericson2314>
I don't know what one can do in config beside assign other module's stuff, ya know?
<clever>
at start,up, it will run /etc/runit/1, 3 is for shutdown, i think 2 was for when it finishes booting?, and then the run's under /etc/service/name/run are services it keeps running
jedahan has joined #nixos
worldofpeace has joined #nixos
worldofpeace has quit [Remote host closed the connection]
jperras has quit [Ping timeout: 252 seconds]
<bpye>
Hm, looks like there is go2nix and dep2nix but nothing for go 1.11 modules?
<clever>
fetchGit will clone the repo, including the full history, and the cache you have is breaking things
<clever>
fetchTarball fetches the state at that rev, so it has less to download
<dmj`>
clever: sure, but that’s seconday, a message like “no common commits” seems to indicate it cannot find that hash from the remote
<dmj`>
secondary*
<clever>
the problem is that the branches you have in your local cache dont share any common commits with the remote git repo
<clever>
so it cant compute the differences for a fetch
<clever>
that typically happens if you try to fetch from 2 entirely different repos
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<clever>
dmj`: the cache is held in ~/.cache/nix/git/
<dmj`>
clever: I got that hash from inspecting the git rev from the channel I was using
<clever>
the problem is more likely that something else you fetched in the past, with fetchGit, is incompatible
<clever>
you may need to wipe ~/.cache/nix/git/
MP2E has joined #nixos
<dmj`>
clever: I don’t see why I would need to have that hash in my cache before fetching it
<clever>
dmj`: the problem is with the git internals, you must have common commits to fetch
<dmj`>
clever: I removed the cache, now I see
<dmj`>
error: Server does not allow request for unadvertised object 89b618771ad4b0cfdb874dee3d51eb267c4257dd
<dmj`>
clever: when I call nix-channel —update will it update the git cache
<dmj`>
clever: apparently not
mayhewluke has quit [Ping timeout: 252 seconds]
<clever>
this has nothing to do with nix-channel
thc202 has quit [Ping timeout: 245 seconds]
<clever>
dmj`: that is a limitation of the git protocol, you must know which branch the commit is in, try adding ref = "master";
mayhewluke has joined #nixos
<dmj`>
clever: when I call nix-channel —update won’t it fetch a newer hash of nixpkgs from nixpkgs-channels (nixpkgs-unstable in this case), and then update the git cache
<clever>
no, it will download a tarball of the channel, and ignore git entirely
<dmj`>
clever: ah, yea, that seems like a purer way to do things...
chreekat has joined #nixos
rprije has quit [Ping timeout: 246 seconds]
rprije has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
judson has quit [Remote host closed the connection]
asymmetric_ has quit [Ping timeout: 272 seconds]
fragamus has joined #nixos
drakonis1 has joined #nixos
nbardiuk has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @volth opened pull request #48725 → perlPackages: fix evaluate with allowAliases=false → https://git.io/fxonB
copumpkin has joined #nixos
Rusty1 has joined #nixos
copumpkin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<avn>
samueldr: Can I ask you backport 4f09e0584906a6792128330a9b44bfe6e729b94c and 67ea56717a54096f499b195a4f44d11fdbbcda2b to 18.09? (I hope it not pull other commits, otherwise pylint not useable with python 2.7)
<samueldr>
(while being checked out on the 18.09 branch)
<avn>
samueldr: Yep, I will do myself tomorrow, no problems ;)
<samueldr>
if you open such a PR, on which you have verified this all to work right, possibly with a test case, it would be wonderful
<samueldr>
oh neat! thank you!
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
judson has joined #nixos
<drakonis1>
so.
<drakonis1>
how's the backports proposal?
rprije has quit [Ping timeout: 245 seconds]
<judson>
This is kind of off topic, but dhess implied they'd solved this: I've got a 172 address for my local DNS resolver. It seems like the container can't reach that address, still.