<jackdk>
clever: thank you. Is that per-patch (I think I am doing the only patch, but I'm calling appendPatch on haskell packages)
<clever>
jackdk: its for all patches
<jackdk>
righto. is it better style to tweak patchFlags or fix line endings in the patch itself?
<clever>
patchFlags would be simpler
<samueldr>
maybe not edit the patch itself but makes the edit and re-generate the patch accordingly?
anton_ has joined #nixos
drakonis has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
aleph- has quit [Ping timeout: 268 seconds]
<jackdk>
bah, neither --binary nor --ignore-whitespace seem to do line ending conversion how I want
eadwu has joined #nixos
q6AA4FD has joined #nixos
vk3wtf has quit [Ping timeout: 268 seconds]
erictape1 has joined #nixos
janat08 has joined #nixos
erictapen has quit [Ping timeout: 250 seconds]
Ariakenom has quit [Quit: Leaving]
<janat08>
interesting thing, Ive installed the os (there was grub, and no live usb is inserted). Yet when running nixos-rebuild switch it considered the os unbootable for false resons sure enuf
<janat08>
filesystems specifies root and grub.devices configured
mkoenig has joined #nixos
iqubic has quit [Ping timeout: 250 seconds]
vk3wtf has joined #nixos
<janat08>
lol installed over my older install
janat08 has quit [Quit: Page closed]
<jackdk>
any haskell nixpkgs maintainers here? I'd really like to see jacereda's fix for aeson make it into master (#53748)
<q6AA4FD>
iqubic: if you check nixpkgs/pkgs/development/haskell-modules/configuration-hackage2nix.yaml
sinner has joined #nixos
<iqubic>
Yes?
<q6AA4FD>
it generates hackage-packages.nix, which is where almost every haskell package is specified
sinner is now known as Guest29060
<iqubic>
But for some reason it isn't pulling the latest version of megaparsec for me.
<q6AA4FD>
but it's a pain to read. anyways, the yaml file specifies version to scrape/parse fom hackage, which are turned into boilerplate expressions
cyounkins has joined #nixos
<q6AA4FD>
that's likely because the latest version of megaparsec isn't specified in the .yaml file
rprije has joined #nixos
d4g has quit [Ping timeout: 258 seconds]
<iqubic>
Why isn't it pulling the latest version of megaparsec? Seems like it ought to be able to do that.
<iqubic>
q6AA4FD: So does that mean someone needs to submit a pull request in order for it to pull the latest version?
<q6AA4FD>
that's my understanding, though i haven't talked to any of the maintainers just read through the public doc (which is the folder hackage2nix in github/NixOS/cabal2nix)
<jackdk>
That's what these are --> [nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update
<iqubic>
huh?
<jackdk>
the message in irc is a bot showing the PR where peti is updating the autogenerated hackage-packages.nix
<iqubic>
But you'd think that would pull the latest version of all the haskell packages.
<iqubic>
Cool.
<jackdk>
IME nixpkgs master is usually fairly close, but then there is often a lag between that hitting the channels, and there are tweaks that sometimes happen
<iqubic>
so how shall I go about getting version 7.0.4 of megaparsec installed?
<q6AA4FD>
if you need a later version of a hackage package for your own purposes i'd recommend this: 1: `nix-env -i cabal2nix` 2: `nix-env cabal://thing_you_need` 3: make a custom package with an overlay or package override
vk3wtf has quit [Ping timeout: 252 seconds]
<q6AA4FD>
oh, and 4: pin that specific version in your .nix file for your project
<iqubic>
that sounds really hard to do. Is there nothing simpler?
<q6AA4FD>
that's frankly the simplest way i could think of. i'm working on a situation like that right now for a package i plan to upstream to nixpkgs at some point, so if i remember i can link you that when i'm done
<q6AA4FD>
oh, and nix-env should be cabal2nix up there
<iqubic>
Well, according to the documetation I'm looking at, megaparsec 7.0.4 was released on Sep 4th, 2018. I wouldn't expect it to take this long to hit the main nixpkgs.
<jackdk>
you could consider doing a callCabal2Nix in your default.nix to override megaparsec
<iqubic>
Mainly what I'm puzzled by is why it's been like 3 months and we still don't have the latest version in nixpkgs.
oldandwise has joined #nixos
<jackdk>
I think it's built against stackage LTS resolvers, and might still be on LTS-12 which is megaparsec 6.5.0
drakonis has quit [Read error: Connection reset by peer]
<iqubic>
I see.
<jackdk>
also nixpkgs master has megaparsec 7.0.4, so your channel may be old
<iqubic>
How do I update my channel?
<jackdk>
(I still get megaparsec 6.5.0 from the channels)
justanotheruser has joined #nixos
aleph- has quit [Ping timeout: 244 seconds]
fusion809 has quit [Remote host closed the connection]
<iqubic>
So how do I go about updating my channel???
o1lo01ol1o has quit [Remote host closed the connection]
<disasm>
iqubic: it's only in unstable. We have a 6 month release cycle. If it doesn't get in for the release, you won't have it.
vk3wtf has joined #nixos
<iqubic>
disasm: What does that mean?
<iqubic>
Do that mean that you only update the main channel every six monts?
<iqubic>
So the thing jackd said about me having an outdated channel is just wrong?
<disasm>
iqubic: sudo nix-channel --list
<disasm>
that'll tell you what release nixos is using
<disasm>
It means every 6 months we cut a new release (19.03 is the new one being worked on in unstable/master). If you're on 18.09 then you'll never see the newer megaparsec you're looking for
<disasm>
Unless someone takes the time and the community approves a backport of the related commits.
<disasm>
Which a hackage bump is probably not going to be backported.
<iqubic>
I see.
<iqubic>
So what options do I have for getting the latest megaparsec?
doyougnu has joined #nixos
cyounkins has quit [Remote host closed the connection]
<ottidmes>
iqubic: you can get 7.0.0 at least in 18.09, so not the latest, but maybe good enough?
cyounkins has joined #nixos
<iqubic>
That's great.
esmerelda has joined #nixos
<iqubic>
Except that I have 18.09 and not 7.0.0
<iqubic>
which concerns me.
<ottidmes>
iqubic: I was not finished answer, you can get it via: megaparsec_7_0_0
<ottidmes>
* answering
<ottidmes>
iqubic: and I guess you could just copy that bit directly from hackage-packages.nix and modify it for the latest version, it will just rebuild from source, but that is doable in my experience with megaparsec
<ottidmes>
iqubic: got to go now, good luck!
cyounkins has quit [Ping timeout: 272 seconds]
justanotheruser has quit [Ping timeout: 252 seconds]
<iqubic>
why do I have to use megaparsec_7_0_0?
<jackdk>
you could do a callCabal2nix in your default.nix to pull in dependencies that you especially need. qfpl's waargonaut does this, but its nix use is a bit convoluted. I don't have a handy simple example
fusion809 has joined #nixos
<nh2>
iqubic: in general, you can override versions of packages to your liking. megaparsec_7_0_0 is a package already pre-made for you containing version 7. Which version would you like to have? Also, if you would like to use any Stackage snapshot of your choosing, instead of using the Stackage snapshot that's in nixpkgs currently, consider using the `stack2nix` tool
oida has joined #nixos
<iqubic>
I want version 7.0.0
kisonecat has joined #nixos
<iqubic>
I just wonder why that's not the default version.
<disasm>
my understanding is nix pulls in a pinned hackage. We bump versions by switching to a newer hackage.
<disasm>
esmerelda: you need to fork the repo, push your branch to your fork, and then PR it from there
<disasm>
that's because you don't have write access to nixos fork of nixpkgs
Taneb has quit [Quit: I seem to have stopped.]
<jackdk>
yeah so the usual github workflow is you make a fork within the github UI, set a remote to your fork, push to it, then GH will show you a button to make a PR from the ref you just pushed
<jackdk>
er, that^
<disasm>
lately, I've noticed just pushing a new branch gives me a URL to open a PR :) But I haven't tested that across forks yet.
<jackdk>
works for me when I push to my nixpkgs fork
o1lo01ol1o has joined #nixos
<esmerelda>
ah okay!
vk3wtf has quit [Ping timeout: 252 seconds]
o1lo01ol1o has quit [Ping timeout: 246 seconds]
LnL has joined #nixos
LnL is now known as Guest83973
<esmerelda>
can't get nox-review to work.. it keeps saying there's nothing to do
<jackdk>
I've been able to get by without nox-review
<mtesseract>
Hi, is it possible to reference a built Nix derivation by its Nix store path in a Nix expression? I want to write a system configuration that includes a derivation, which is not part of nixpkgs or another repository.
adamantium has joined #nixos
adamantium has joined #nixos
adamantium has quit [Changing host]
adamantium has joined #nixos
adamantium has quit [Changing host]
lokado has joined #nixos
shpx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 258 seconds]
ddellacosta has quit [Remote host closed the connection]
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
lokado has quit [Client Quit]
mtesseract has joined #nixos
shpx has joined #nixos
juhulian has joined #nixos
<moredread[m]>
I'm updating mediathekview. There currently also is zdfmediathk, which isn't working anymore and has been superseeded by mediathekview. What is the correct way to remove the package? As it hasn't been working for over 2 years now I guess there aren't any users left.
orivej has joined #nixos
lokado has joined #nixos
<moredread[m]>
I have asked the question in #nixos-dev again
<lokado>
Has anyone use kde upgraded to the latest nixos-unstable channel? I got "Power Management module can't be initialized" when I enter the desktop
mckeanalicej has joined #nixos
<mckeanalicej>
how do you compile nix with libsodium? is there a flag I need to pass to the configure script?
<mckeanalicej>
(im trying to bootstrap thats why im complining by hand)
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
shpx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
esmerelda has joined #nixos
saltsa has quit [Quit: leaving]
endformationage has quit [Quit: WeeChat 2.3]
shpx has joined #nixos
aleph- has joined #nixos
vk3wtf has quit [Ping timeout: 252 seconds]
MinceR has quit [Ping timeout: 240 seconds]
MinceR has joined #nixos
shpx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
simukis has joined #nixos
esmerelda has quit [Ping timeout: 250 seconds]
rauno has quit [Ping timeout: 268 seconds]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<iqubic>
How do I modify my X11 cursor on nixos?
simukis has quit [Quit: simukis]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jasongrossman>
iqubic: The best way is to install oneko, and then type at a terminal:
<jasongrossman>
oneko -dog -name one &
<jasongrossman>
Otherwise, you can do it in your window manager. For example, in stumpmw I do:
<jasongrossman>
(run-shell-command "xsetroot -cursor_name left_ptr -solid black -name root-window")
rprije has quit [Ping timeout: 245 seconds]
<iqubic>
What does oneko do?
<iqubic>
oneko isn't what I want it be.
<iqubic>
I think I'll look into xsetroot.
tzemanovic has joined #nixos
<jasongrossman>
Yes, if you don't want to do it with your window manager then xsetroot looks good.
<iqubic>
I can't do it with my window manager. I use EXWM and it doesn't give me that ability.
<jasongrossman>
Then you're in a similar boat to me, but you can use your window manager to run xsetroot, as I do.
<iqubic>
Most tiling window managers lack that feature.
<iqubic>
How do you pick which cursor to set it too?
<jasongrossman>
Yeah - I was being vague when I said "in your window manager".
jackdk has quit [Ping timeout: 268 seconds]
tzemanovic has quit [Ping timeout: 252 seconds]
Guanin has joined #nixos
reinhardt has joined #nixos
mtesseract has joined #nixos
kisonecat has quit [Ping timeout: 246 seconds]
amfl_ has joined #nixos
amfl has quit [Ping timeout: 240 seconds]
hamishmack has joined #nixos
<Guanin>
Hey, I'm noticing a strange bug and do not know how to debug it. Every now and then, my X server crashes and I get back to the display-manager. Here is the dmesg/journal snippet around such a crash: https://gist.github.com/Amarandus/f27e0f7c208aa9ac4d45fce4263400b2
amfl_ is now known as amfl
<Guanin>
I also noticed that the segfault is not always caused by libXUL, but also by e.g. libQt5Sql.so.5.11.1 or libX11.so.6.3.0
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jasongrossman has quit [Read error: Connection reset by peer]
<q6AA4FD>
i have a file placed at ~/.nixpkgs/overlays/foo.nix, and i'm unable to access any of the derivations i've set in it. is there anything i can do to debug this?
<Sophos[m]>
hello, attempting to write a nixpkg expression and something is failing. How can I drop into the environment at the point of failure to poke around and figure out what might be going on?
<q6AA4FD>
Sophos[m]: you can run nix-build with -K to keep the failed build in /tmp/, you can generate a shell in the same setup with nix-shell
mckeanalicej has quit [Quit: Page closed]
<Sophos[m]>
I have traversed to the tmpdir though running nix-shell complainss about no default.nix, was I not supposed to execute nix-shell there?
<Sophos[m]>
(apologies, I am quite new to this)
<q6AA4FD>
Sophos[m]: you want to give nix-shell the nix expression i think? i don't use nix-shell except with cabal2nix which is a whole other story
<Sophos[m]>
heh
MichaelRaskin has quit [Quit: MichaelRaskin]
<q6AA4FD>
but -K will give you the failed build, so you should be able to look at the logs and try your own hand at compiling it
<q6AA4FD>
Sophos[m]: reading the manual, you can build an environment for a build with nix-shell by invoking `nix-shell '<nixpkgs>' foo`
nek0 has quit [Ping timeout: 252 seconds]
Eram has quit [Remote host closed the connection]
mtesseract has joined #nixos
doyougnu has quit [Quit: WeeChat 2.2]
hyper_ch2 has joined #nixos
rauno has joined #nixos
mtesseract has quit [Client Quit]
<sphalerite>
q6AA4FD: how are you trying to access them? They should appear in nixpkgs like any other package
<Sophos[m]>
joy, running `nix-build $NIXPKGS -A my-package' fails, but attempting the manual steps via nix-shell is working... ugh
<Sophos[m]>
I am writing my own. $NIXPKGS points to my local fork
<sphalerite>
Sophos[m]: does it access the internet?
<Sophos[m]>
yes
<sphalerite>
Sophos[m]: then that's the problem
<sphalerite>
Builds aren't allowed network access
<sphalerite>
(except fixed-output derivations)
<Sophos[m]>
so, I need to used a fixed-output derivation?
<sphalerite>
That's why the source for a package is usually specified with fetchurl, which creates a fixed-output derivation that gets the source, which is then passed into the builder that has no network access
<sphalerite>
if you can guarantee the output will always have the same hash, yes.
<sphalerite>
You're in luck! We have fetchNuGet :)
<sphalerite>
pkgs/top-level/dotnet-packages.nix has some examples
<Sophos[m]>
nice. time to patch things to use that...
nek0 has joined #nixos
<Sophos[m]>
that's neat that packages can't access the net like that. Really useful, even if it causes some headaches...
<sphalerite>
yes, it's a huge help for purity/reproducibility
<q6AA4FD>
sphalerite: they weren't, until i added the overlay directory to NIX_PATH, even though the documentation says that every file in ~/.nixpkgs/overlays/ should be loaded automatically
<sphalerite>
^ the implementation of that behaviour
<q6AA4FD>
shpalerite: i'm not on nixos, so that's a possible explanation. would it be worthwhile for me to debug it or is it a relatively unused feature
<sphalerite>
q6AA4FD: I don't think that should affect it
<sphalerite>
q6AA4FD: what does your overlay file look like and where exactly did you place it?
<sphalerite>
q6AA4FD: overlays are a widely used and powerful feature :)
<q6AA4FD>
sphalerite: you have a preferred paste site? if it's an important feature i'll do my part lol
<sphalerite>
no, no real preference besides "not pastebin.com" :p
aleph- has quit [Ping timeout: 240 seconds]
<sphalerite>
oh yeah, and what command you're running would be good t oknow too
jackoe has joined #nixos
<q6AA4FD>
https://ix.io/1yim <-- this was at ~/.nixpkgs/overlays/pullover_lay.nix | i was running `nix-env -i klfc`
<q6AA4FD>
it wasn't that exactly file, but the derivation didn't change consequentially
<jackoe>
Is this an okay place to ask nix (but not nixos) related questions?
<q6AA4FD>
jackoe: that's my understanding
<q6AA4FD>
sphalerite: my old `NIX_PATH=/nix/var/nix/profiles/per-user/root/channels:nixpkgs=/home/quinn/.nix-defexpr/channels/nixpkgs` my new `NIX_PATH=$NIX_PATH:/home/quinn/.nixpkgs/overlays`
<q6AA4FD>
not a copy paste, so i may have typo'd, but you get the idea
<jackoe>
Fantastic, does anyone know why overriding haskell's lens package would cause infinite recursion?
<jackoe>
the specific line being: lens = haskellPackages.callHackage "lens" "4.16.1"
<sphalerite>
jackoe: most likely using self in some place that you need to use super
<sphalerite>
not sure though. Maybe the implementation of callHackage uses lens?
<jackoe>
sphalerite: thanks! I'll give that a shot. I'll look into callHackage too.
<sphalerite>
q6AA4FD: and it works when you manually set the NIX_PATH to point to it, and not when you don't have an overlays entry? And which command are you running?
<q6AA4FD>
yes, it works when i set it manually
<q6AA4FD>
but not otherwise; and i'm running `nix-env -i klfc`
rauno has quit [Ping timeout: 250 seconds]
rauno has joined #nixos
<sphalerite>
what about nix-env -iA nixpkgs.klfc ?
Ariakenom has joined #nixos
<q6AA4FD>
attribute 'klfc' in selection path 'nixpkgs.klfc' not found
<q6AA4FD>
i have a package that doesn't seem to provide releases at github.com/archive/${version}, is the next best thing fetchgit with $rev, or fetchzip with the url from the release page?
<q6AA4FD>
sphalerite: i'm not sure what you mean by both variants? if you mean where $NIX_PATH is fixed, the command works fine. and when my $NIX_PATH is as it was before, it fails with that error message
<sphalerite>
q6AA4FD: odd. I'm out of ideas, maybe open up an issue for nixpkgs and @ me in it (I'm lheckemann on github)
<sphalerite>
q6AA4FD: if the release page is github's releases thing, I'd say fetchFromGitHub with rev = "<tag>"; where you replace <tag> with the name of the git tag
<sphalerite>
if upstream releases their own tarballs, use those with fetchurl
dermetfan has joined #nixos
<esmerelda>
also, I notice that there's only one package using dotnet core... is there a plan to add dotnet alongside mono?
<arianvp>
Can anybody tell when JosW comes online he should mail me instead of ping on IRC?
<arianvp>
I keep missing him
<arianvp>
Because he doesn't have a bouncer
<Sophos[m]>
as a build dependency for omnisharp-roslyn I need to grab the nuget package cake. well, I have it: `cake = fetchNuGet { ... };` how can I then utilize cake in the buildPhase?
<Sophos[m]>
utilize => run
<Sophos[m]>
where is it installed to?
Eram has joined #nixos
Eram has quit [Client Quit]
jasongrossman has quit [Ping timeout: 258 seconds]
<esmerelda>
Sophos[m]: you can access it by $cake!
<q6AA4FD>
sphalerite: i'm drafting it now, i'll @ you on it too. and the tag thing worked, i see that rev= doesn't have to be an int, it's just whatever they name it (in this case v${version})
<esmerelda>
also happy to see another .net person here :) I thought I was the only one
Tucky has joined #nixos
tzemanovic has joined #nixos
<Sophos[m]>
esmerelda: hey, my gf is in C# classes, so I'm having a bit of fun with dotNet for the next few months :)
tzemanov_ has joined #nixos
<esmerelda>
:D nice! I work for them, and then linux from home to keep my sanity
<sphalerite>
arianvp: you can also use ,tell JosW please email me!
<sphalerite>
,tell arianvp ,tell is useful
<{^_^}>
sphalerite: I'll pass that on to arianvp
<Sophos[m]>
esmerelda: that sounds fun!
<Sophos[m]>
sweet, got $cake working now. 😁
tzemano__ has joined #nixos
Eram has joined #nixos
tzemanovic has quit [Ping timeout: 252 seconds]
tzemanov_ has quit [Ping timeout: 245 seconds]
<arianvp>
,tell JosW could you email me instead? You seem to keep missing me. Email is in my commit log on GitHub (arianvp)
<{^_^}>
arianvp: 5 minutes, 52 seconds ago <sphalerite> ,tell is useful
<{^_^}>
arianvp: I'll pass that on to JosW
vk3wtf has quit [Ping timeout: 252 seconds]
<arianvp>
Thanks!
<q6AA4FD>
sphalerite: alright, i made theissue
<esmerelda>
Sophos[m]: btw I'm working on fixing the Python plugin in VS Code, so I have a little context on how the extensions work if you need any help with your PR
<{^_^}>
[nixpkgs] @fpletz pushed commit from @elseym to release-18.09 « nixos/containers: explicitly set link up on host for extraVeths »: https://git.io/fhndH
<jasongrossman>
Hi hi. I have what I think is a very basic networking question. I've set up wireguard using the instructions on the NixOS wiki, and I can ping my local wireguard address successfully. But my internet traffic is not going over the wireguard tunnel - it's still using my wireless interface directly (wlp2s0) instead of my new wireguard interface (wg0). Should I be doing something to tell the OS to use the wireguard interface?
o1lo01ol1o has joined #nixos
<lassulus>
you need to set your default gateway to the wireguard server (mind, that you probably need a special route, to connect to your wireguard server still through the normal internet). You need to enable nat on the wireguard server
Tucky has quit [Quit: WeeChat 2.2]
<jasongrossman>
lassulus: Thanks. The server is a commercial VPN server, so I should just need to set my default gateway then.
Tucky has joined #nixos
* jasongrossman
searches NixOS options for "gateway".
<lassulus>
I would try it first manually with `ip route`
<jasongrossman>
lassulus: I'm not sure what the syntax for that is. Might be easier just to set the gateway in configuration.nix.
<jasongrossman>
networking.defaultGateway.interface, I guess.
<Mic92>
jasongrossman: this won't work for vpn as it is
<Mic92>
because you need still a route to your wireguard vpn
<Mic92>
otherwise it would be a loop.
<jasongrossman>
Mic92: Damn.
<lassulus>
hmm, I think you can also put 0.0.0.0/0 to the allowedIPs of the wireguard peer?
<jasongrossman>
lassulus: Yes. Done that.
<Mic92>
lassulus: if you do that in wg-quick it will work
o1lo01ol1o has quit [Ping timeout: 244 seconds]
thadthemad has joined #nixos
<Mic92>
because it uses a routing policies for that.
Guest29060 has quit [Ping timeout: 246 seconds]
jasongrossman has quit [Read error: Connection reset by peer]
jasongrossman has joined #nixos
<jasongrossman>
I'm going to be in and out as I test my network.
rprije has joined #nixos
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Mic92>
jasongrossman: you can just test a single ip or range forst
<Mic92>
i.e. ip route add 8.8.8.8 via $wireguard_ip
<tilpner>
^ useful to look up how other people configured something
jasongrossman has quit [Ping timeout: 252 seconds]
<siers>
Is androidsdk that's on 18.09 only for one of the versions? That is, if you want others then you're basically out of luck, except for channel switching?
<klntsky>
nix.useSandbox and nix-shell --pure - AFAIU they do the same sort of thing. Are there any differences besides what's described in `man nix-shell`? Also, does ``"real" Nix build'' (as written in the manual) mean ``build with useSandbox''?
<sphalerite>
klntsky: no, they don't do nearly the same thing
carlosdagos has quit [Quit: Connection closed for inactivity]
<sphalerite>
klntsky: nix-shell doesn't do any sandboxing, --pure just makes it discard existing environment variables to reduce the amount of impurity
<sphalerite>
klntsky: while the build sandbox uses namespacing to restrict the build's view of the filesystem and other things
<sphalerite>
klntsky: and yes, nix-build will use the sandbox if enabled in nix.conf.
jasongrossman has joined #nixos
jasongrossman has quit [Remote host closed the connection]
<drager>
I have installed terminus_font (via configuration.nix fonts.font) but when I try to use `xos4 Terminus 10` for AwesomeWM it doesn't look as it should. Any debugging tips?
<joepie91>
are there any existing issue threads / RFCs / etc. on 1) making the import statement auto-add .nix to the end of an import path, or 2) restructuring NixOS service modules to better support multiple service instances, stateful service dependencies (eg. "service X depends on PostgreSQL with a particular database Y existing"), and such?
<sphalerite>
Sophos[m]: curl has openssl on by default
<Sophos[m]>
nice, that works. how are options specified thougH?
<sphalerite>
drager: right, so the font is there correctly. I'm not sure how font specifications for awesome work, but if it takes a fontconfig spec you can use Terminus:scalable=false :)
rprije has quit [Ping timeout: 250 seconds]
<sphalerite>
Sophos[m]: you can see the options that it takes in the build expression
<sphalerite>
Sophos[m]: and `curl.override {sslSupport = false;}` for instance will give you a curl that doesn't support SSL
<drager>
sphalerite: Alright, thanks!
<q6AA4FD>
i'm trying to debug a build with nix-shell, but i can't find any documentation about the phases for the haskell builder, and configurePhase and buildPhase don't seem to be doing anything
<q6AA4FD>
sphalerite: you need to pass that into a package, like curl = curl.override, right?
<q6AA4FD>
oh, and thank you for the link
<sphalerite>
q6AA4FD: yes
Lears has joined #nixos
<q6AA4FD>
sphalerite: well now i have another problem, if you're game. when i run nix-shell '<nixpkgs>' -A mypackagefromoverlay, i don't get the update configurePhase, setupCompilerEnvironmentPhase, or i think anything else
[Leary] has quit [Ping timeout: 258 seconds]
<q6AA4FD>
nvm, you need to use with haskellPackages. i don't think that's very idiomatic though
<q6AA4FD>
nevermind, that's not it...
Taneb has joined #nixos
jasongrossman has quit [Remote host closed the connection]
GlennS has quit [Quit: GlennS]
<siers>
Can android-studio work with the androidsdk from nix store?
<esmerelda>
I was going to respond to a PR comment on my Python extension thing. the reviewer pointed out that I might want to make the language server its own package
<esmerelda>
so I set to work on that, figured out how to make CoreCLR packages building, but then I needed to test it.
dermetfan has joined #nixos
<esmerelda>
and I decided if I'm building *that* rather than getting it as a binary, I might as well compile the extension in the first place
ThatDocsLady has joined #nixos
<esmerelda>
so then I figured out how to make them both build, but I need some extra node packages to make all that work, and... etc.
<esmerelda>
I pulled on a thread and unraveled my sweater :/
<Sophos[m]>
well, I've facepalmed and headdesked enough for one night with cake and nix. I have a feeling cake and nix just cannot mix currently. Too many assumptions made by cake
<Sophos[m]>
and not enough configuration options
<Taneb>
Sophos[m]: what is cake?
brodul has joined #nixos
<brodul>
Has someone build thunderbird with sound
<brodul>
I want to have sound for notifications
<wedens>
can nixops be used to deploy to online.net vps (scaleway)?
hydraz has joined #nixos
regulus_ has quit [Ping timeout: 246 seconds]
johanot has joined #nixos
jackoe has quit [Quit: leaving]
<joepie91>
wedens: yes, though not automagically
<joepie91>
afaik it requires manual setup from ISO and then you can use it from NixOps like any "existing NixOS system"
<ivan>
you can make an image on scaleway and restore it to another machine
<drager>
sphalerite: Didn't work out with scalable=false. However, I'm unsure what format it wants it in. By doing `Terminus 8` it looks better but not as I expect.
fusion809 has quit [Remote host closed the connection]
fusion809 has joined #nixos
<siers>
sphalerite, awesome
Makaveli7 has quit [Quit: Leaving]
<sphalerite>
__monty__: if you want /bin to contain stuff from different packages, you can't just bind mount multiple packages' /bin onto each other
rprije has quit [Ping timeout: 244 seconds]
<fusion809>
Nix used to run fine on my Arch system, but now whenever I install a new package, remove a package, update one, with Nix I get the error `error: cloning builder process: Operation not permitted
<fusion809>
error: unable to start build process`. Tried re-installing Nix too, with the https://nixos.org/nix/install script, yet I still get this error. In fact, the install with that script fails due to this error.
<sphalerite>
__monty__: so you need to make a package that contains a unified /bin and bind-mount that
<sphalerite>
fusion809: that sounds like it thinks it's multi-user
<sphalerite>
but there's no daemon running
<fusion809>
How do I fix that?
<symphorien>
fusion809: are user namespaces enabled ?
<sphalerite>
oh yeah, or that maybe
<symphorien>
or disable the sandbox
<fusion809>
Not sure, I've left things the default.
<symphorien>
now that sandbox is enabled by default, on systems without user namespaces, the installer fails with this unhelpful message
<__monty__>
sphalerite: Ah, makes sense. Does prevent it from working across nfs though. Maybe an overlayfs would be a more robust approach?
<sphalerite>
__monty__: does it?
<sphalerite>
I don't see why it wouldn't work with nfs
<Mic92>
fpletz: I suppose we need to rework the way rambox's node modules are updated.
<symphorien>
there is a limit on the number of layers for an overlayfs (like 16 or something)
<Mic92>
symphorien: I think this was increased significantly.
<Mic92>
fusion809: I think this was enabled by default.
<__monty__>
sphalerite: Symlinks are resolved on the client so if they point outside the nfs root, the nix store for example, they'll point to whatever's mounted on the client at /nix/store/...
<Mic92>
fusion809: I mean the sandbox
ng0 has joined #nixos
<symphorien>
ah maybe
<Mic92>
archlinux still don't enable this in their kernel
<fusion809>
Ah. So what do I do about this? My ~/.config/nixpkgs/config.nix is just allowUnfree = true, lol, so if an error is occurring for me, with my simple one-user set up there's seriously wrong.
<Mic92>
fusion809: sandbox = false in /etc/nix/nix.conf
<__monty__>
sphalerite: Actually, why not simply bindmount all the files that need to go into /bin? That'd solve the problem, no?
<__monty__>
And no need for an overlayfs, which wouldn't work well anyway because of that limit.
<fusion809>
Is this conf file, really as simple as sandbox = false, or do I have to format it as a nix file like { sandbox = false; }?
eadwu has joined #nixos
<fusion809>
Nvm, I learnt by trying
<Mic92>
fusion809: you also need to restart your nix-daemon if you have one running
<fusion809>
I don't
<Mic92>
fusion809: just sandbox = false
<fusion809>
Unless it's running in the background
<fusion809>
I just stuck to the defaults w.r.t. to daemon, so if it is enabled by default it is running
<Mic92>
fusion809: there is a nix-multiuser package in archlinux, In that case you would have a nix-daemon otherwise not.
<Mic92>
if your /nix directory is owned by your account, you don't have one.
<sphalerit>
monty: can files be bind mounted?
<__monty__>
sphalerit: Yep : ) and afaict it's recommended practice with NFSv4.
<symphorien>
are several thousands bind mounts reasonable ?
<symphorien>
nixos-rebuild prints how many symlinks you get for /run/current-system/sw and it is pretty high
ramses_ has joined #nixos
spooky_ has joined #nixos
<Mic92>
symphorien: bind mounts should scale well. Only some applications might be a bit slower when parsing /proc/mounts
spooky_ has quit [Read error: Connection reset by peer]
spooky_ has joined #nixos
thadthemad has joined #nixos
mtesseract has quit [Read error: Connection reset by peer]
janat08 has joined #nixos
spooky_ has quit [Ping timeout: 240 seconds]
__Sander__ has quit [Quit: Konversation terminated!]
janat08 has quit [Ping timeout: 256 seconds]
spooky_ has joined #nixos
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Ping timeout: 244 seconds]
thadthemad has quit [Ping timeout: 240 seconds]
IRC-Source_19571 has joined #nixos
symphorien has quit [Quit: WeeChat 2.2]
tmaekawa has joined #nixos
erictapen has quit [Ping timeout: 258 seconds]
<IRC-Source_19571>
After upgrading 18.03 -> 18.09 by adding the nix-channel `https://nixos.org/channels/nixos-18.09` and doing a `nixos-rebuild switch --upgrade` I still get nixos-version reporting 18.03. How do I perform a full system upgrade?
leo_ has joined #nixos
hyper_ch2 has joined #nixos
<goibhniu>
hi IRC-Source_19571 did you add it to your root profile?
<goibhniu>
Cool, it's an easy mistake to make. The reason is that your user can have a different version of nixpkgs. It's worth mentioning (although, might already be obvious) that nix commands for your user will use the version defined in $HOME/.nix-channels. If you have added a channel there and run `nix-channel --update` as that user.
<tilpner>
Mic92: I modified my kexec file to build a self-extracting tarball-script, so that you only scp, chmod +x, ./kexec. Is that something you'd want in nixos-generators?
<tilpner>
(chmod +x is unnecessary, I forgot)
<lassulus>
doesn't nixos-generators already do that?
<lassulus>
ah, wait, that sounds cool
<lassulus>
just open a PR
<tilpner>
Last I checked, nix-tests didn't. But I didn't check for a while
Ariakenom has quit [Quit: Leaving]
strobelight has joined #nixos
Ariakenom has joined #nixos
<tilpner>
Why does config.nix have nginx enabled...
<{^_^}>
[nixpkgs] @orivej opened pull request #53941 → llvm_6, llvm_7: build all default targets → https://git.io/fhnxu
<tilpner>
That seems like a weird default
jomik has joined #nixos
<jomik>
Heya! I was pondering a solution to my direnv problems... I realize that my issue is caused by nix-shell. It ruins my shell environment somehow, so direnv can not use this. I.e. I cannot use emacsclient nor can I use `thefuck`..
<jomik>
So, I was wondering if there was a "simple" way to create a setup sort of like what home-manager does, for use with direnv.
<jomik>
There is a solution that would install packages in profiles, when I use `nix-env`. But I would prefer a config file, rather than the imperative installation.
janat08 has joined #nixos
cyounkins has joined #nixos
Guest83973 has joined #nixos
Guest83973 has quit [Changing host]
<janat08>
after nixos-install efibootmgr remains the same
<hyper_ch2>
if it's a test server, you could just use clever's kexec with dhcp (and auto-reboot after 5min or so) and see if it works
<duairc>
Is it just me or does my NixOS system use three or four times the RAM my Debian system did even when not much is happening?
<gchristensen>
probably just you
<gchristensen>
how are you measuring used ram?
danielrf has quit [Ping timeout: 268 seconds]
<duairc>
Like I just have a web browser with a couple of tabs, an email client, Pidgin and a few terminals and it's using 9G
<hyper_ch2>
unused ram == wasted ram
<duairc>
The "used" column in free -m, the green bar in htop
<duairc>
I get that, I do
<gchristensen>
"used" includes kernel caches
<gchristensen>
the important one is "available"
<duairc>
It says there's 6G (out of 16G) available
orivej has quit [Ping timeout: 240 seconds]
<duairc>
I am using /tmp on tmpfs (as I was on Debian), but df -h says there's only 1.2 MB used in /tmp, so it can't be that
<hyper_ch2>
my server has 32gb ram, used 4, free 1.4 cache 26.5
<gchristensen>
at any rate, no, nixos doesn't inherently use more ram when compared to another distro
cyounkins has joined #nixos
Makaveli7 has quit [Quit: Leaving]
jomik has joined #nixos
<Mic92>
jomik: does it override your TERM? (This message has been postponed on 2019-01-14 13:46:08.)
<duairc>
The reason I want to blame Nix is because this seems to happen only after I attempt a couple of nix builds... when I start my system and open the same apps it uses somewhere between 3 and 4 GB of RAM. But during the builds loads of RAM gets used, and then afterwards it stays like that. I mean maybe it all being cached.
<goibhniu1>
duairc: FWIW, I'm running KDE/plasma, firefox, pidgin, thunderbird, emacs, konsole and even an electron app, and it's using about 3G
<{^_^}>
[nixpkgs] @hedning opened pull request #53946 → attr: fix issues with 2.4.48 → https://git.io/fhnpd
<jomik>
Mic92: Sorry, I have been having some issues with KDE freezing, when using multiple monitors.. XD
<{^_^}>
[nixpkgs] @sbourdeauducq opened pull request #53947 → pyvcd: init at 0.1.4 → https://git.io/fhnpF
<Mic92>
jomik: I had the same bug.
<Mic92>
but no solution unfortunaly.
<duairc>
Like when I try to find what individual applications are using RAM, it just doesn't add up at all. Maybe it is just cache. But surely the available column should be lower then?
<jomik>
Mic92: I don't recall it overriding TERM. I am not sure why that'd cause emacsclient to be unable to find my emacs server though.
<Mic92>
jomik: it happend to me when using direnv-mode an my own nix function for direnv
<jomik>
$TERM is `alacritty` for in and out of a direnv dir
<Mic92>
Then this fine
<jomik>
Pretty sure it is a `nix-shell` issue tbh.
<jomik>
`thefuck` gets messed up if I install a python package in nix-shell. Because then `thefuck` can't find its python dependencies...
<jomik>
emacsclient can not find the server.
<jomik>
But if I don't have a nix-shell direnv loaded, it finds it just fine.
<Mic92>
jomik: this is because thefuck picks up PYTHONPATH.
<jomik>
Yeah.
reinhardt has quit [Quit: Leaving]
<jomik>
I thought it would be wrapped by nix, when installed :D
<jomik>
?
<Mic92>
I am surprised buildPythonApplication does not do exactly that.
<Mic92>
Thre might be some counter example but for the most part you want to ignore PYTHONPATH
npmccallum has joined #nixos
<jomik>
Hmm?
<jomik>
I guess it messes it up here, because the paths are not merged, they're just overwritten. I can fix that by manually saving my pythonpath before evalling the result of the nix expression.
<jomik>
Still have the issue of emacs not finding the server socket.
<jomik>
It is in `/tmp/emacs1000/server
<Mic92>
jomik: inside a nix-shell you have TMPDIR=/run/user/$UID.
<jomik>
nix-shell does set TMP, TEMP and TMPDIR to
<jomik>
yeah.
<Mic92>
I changed it to a fixed location
<jomik>
But I tried setting them to /tmp
<jomik>
which is where the emacs socket is
pointfourone has joined #nixos
<Mic92>
jomik: you have to specify that also when starting emacsclient.
<Mic92>
because emacsclient does not read your configuration to figure out where the socket is
<jomik>
OH
hydraz is now known as ^W
<jomik>
Unsetting them helps!
<jomik>
So I just have to unset them after loading the env.
<janat08>
can some1 tell hom i'm not booting into grub
<janat08>
after install
primeos has joined #nixos
jomik has joined #nixos
<jomik>
Mic92: Are you using GNOME then?
<Mic92>
jomik: awesome wm.... Configured it years ago, I will stay with it until wayland becomes a thing, one want to use.
<samueldr>
janat08: (among others) grub.efiSupport will need to be set to true, and it's systemd-boot.enable will need to be false; then you'll also need to nixos-rebuild with --install-bootloader to be sure it is used
endformationage has joined #nixos
<janat08>
ty
<samueldr>
janat08: do you know if your EFI firmware is fussy or buggy and doesn't like being configured by the system to boot non-default paths? (some are)
<jomik>
Ahh, yeah okay :P I have considered Awesome WM as well. It's a pretty difficult choice, I feel. Cuz you have to put in hours to configure something, then try it out D:
iqubic` has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<samueldr>
ah, with nixos-rebuild, `build` only builds, it will not do anything involving system modifications, including switching to the configuration or changing boot options
<samueldr>
so, janat08, you either want to nixos-rebuild boot or nixos-rebuild switch with --install-bootloader
tstat has joined #nixos
<janat08>
nixos-rebuild boot --install-bootloader building Nix... building the system configuration... Warning: do not know how to make this configuration bootable; please enable a boot loader.
<samueldr>
good, that's some progress
* samueldr
checks your config again
pointfourone has joined #nixos
<samueldr>
progress in the sense that nixos is attempting to install the bootloader, and we have output explaining that it can't
<{^_^}>
[nixpkgs] @Ma27 opened pull request #53952 → nixos/gitea: minor fixes → https://git.io/fhnhj
<samueldr>
add `boot.loader.efi.canTouchEfiVariables = true;` otherwise nixos will not change them
<janat08>
lol i meant to change it
thadthemad has joined #nixos
socksy has joined #nixos
<samueldr>
oh, I might have missed something relatively obvious: if you're still in the installation image, you can nixos-install, nixos-rebuild (unless you used nixos-enter) will rebuild the installer config in-memory
<socksy>
hey, I'm setting up my new work's environment using an fhs environment with buildFHSUserEnv. One of the dependencies in the electron app is node keytar, which only works if gnome-keyring is installed ( https://github.com/atom/node-keytar/issues/74 ). I have it in my list of packages but it still doesn't find it, reporting "org.freedesktop.DBus.Error.ServiceUnknown: The name org.freedesktop.secrets was not provided by any .service files
<{^_^}>
atom/node-keytar#74 (by blaatunga, 1 year ago, open): It still requires gnome-keyring even on KDE
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fhnjt
<socksy>
also, there are multiple ways of installing gnome-keyring, and I think I've tried them all now but a bit confused what the difference is between e.g. gnome3.gnome-keyring and gnome3.gnome_keyring
<ottidmes>
,declarative fusion809
<{^_^}>
fusion809: There are multiple ways of managing declarative profiles. 1) Attrset, compatible with imperative use of nix-env https://git.io/fAQHW ; 2) buildEnv, providing more control over the paths that are linked into the profile https://git.io/fp0aU ; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fhnj3
pointfourone has quit [Quit: Leaving]
tzemanovic has joined #nixos
<sphalerite>
socksy: those are identical
<sphalerite>
socksy: the problem, and I'm not sure this can be solved within an FHS user env, is that gnome keyring is talked to via dbus, and your dbus session bus is unaware of the dbus service files provided by gnome-keyring
<sphalerite>
socksy: so AFAIK the only way is to make sure that whoever runs the software has gnome-keyring in their user or system profile :|
<socksy>
could I get it working in the "host" environment?
<socksy>
I only care atm for development, not packaging
<{^_^}>
[nixpkgs] @dywedir pushed commit from @P-E-Meunier to master « Adding webpack-cli to the node packages »: https://git.io/fhnj4
kisonecat has quit [Ping timeout: 268 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
thadthemad has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
Makaveli7 has joined #nixos
<duairc>
About my RAM issues, ps_mem shows I'm using 3.3G, which is line with what I'd expect, but everything else is showing more than twice that. It seems like the rest is being used by ZFS somehow?
thadthemad has joined #nixos
<tilpner>
Did you set the ARC cache size? Are you using deduplication?
<gchristensen>
oh, heh, hyeah
<duairc>
I never explicitly enabled nor disabled any of those things
<gchristensen>
duairc: ZFS's caches don't appear in the available section either, because it uses them differently -- but the cache is still evictable for sure
<gchristensen>
AND, yes, ZFS's caches will dutifully use all your RAM as it can
<duairc>
So if I open stuff that needs that RAM, I shouldn't worry, the kernel will figure it out?
<gchristensen>
yep
<duairc>
(without swapping)
<duairc>
Okay, good! And I wasn't using ZFS on Debian so that would explain why it's different
o1lo01ol1o has quit [Ping timeout: 258 seconds]
kisonecat has joined #nixos
das_j has quit [Remote host closed the connection]
das_j has joined #nixos
<{^_^}>
[nixpkgs] @rvolosatovs opened pull request #53954 → mbed-cli: Init at 1.8.3 → https://git.io/fhceG
<gchristensen>
yeah, ZFS has very different memory use :)
<duairc>
If I'm on nixos-unstable, which has GHC 8.6, is there any way I can install GHC 8.4?
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @rvolosatovs opened pull request #53955 → pythonPackages.fuzzywuzzy: Init at 0.17.0 → https://git.io/fhcea
<{^_^}>
[nixpkgs] @rvolosatovs opened pull request #53956 → pythonPackages.mbed-ls: Init at 1.6.2 → https://git.io/fhcew
o1lo01ol1o has quit [Remote host closed the connection]
ddellacosta has joined #nixos
<elvishjerricco>
My memory meter in my status bar became completely useless after I switched to ZFS :P
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<elvishjerricco>
I should figure out how to set it to ignore the kernel's memory usage
<infinisil>
freedan42x: systemd services can do that
<freedan42x>
infinisil, systemd-services not found
tcosmo_ has joined #nixos
<infinisil>
That's not a command
<infinisil>
freedan42x: Do you not know what systemd services are?
<freedan42x>
infinisil, tbh - no. I'm noob at Nix
<infinisil>
Has nothing inherently to do with Nix. systemd is the thing that manages a lot of things on a lot of Linux distros
<tcosmo_>
Hello everyone :) I'm struggling trying to make a python systemd service because of the propagated build inputs. Indeed, when I try to add them to the path, nixos says "The value of the environment variable ‘PATH’ in systemd service ‘timelist.service’ is too long."
<infinisil>
And its services can run arbitrary stuff when things happen
<tcosmo_>
Would you know if there is a better way to create the python env of my package within my systemd service?
<sphalerite>
tcosmo_: as it happens, there's a convenience buildEnv wrapper for python in the form of python.withPackages
<freedan42x>
infinisil, thank you; `Actually, how to write script that runs several commands?`
<tcosmo_>
sphalerite: Thank you!! Would you be aware of any service that uses it so that I can copy?
<infinisil>
freedan42x: systemd.services.<foo>.script will take a bash script to run
<infinisil>
(that's a NixOS option)
<tcosmo_>
sphalerite: Just found web-servers/uwsgi.nix thank you!!!
<freedan42x>
infinisil, oh, thx
dermetfan has quit [Ping timeout: 250 seconds]
<Church->
samueldr: So I found something worse then packaging Ruby code.
<gchristensen>
is it unpackaged ruby code?
<Church->
gchristensen: Writing xlib code
<Church->
Trying to figure out how to split a root window into having several child windows. Which I sorta know how.
<gchristensen>
hehe
<Church->
But then figuring out how to assign an application to each window
<Church->
That last one is driving me barmy. There's nothing that says how to do so.
JosW has joined #nixos
<gchristensen>
sounds like a nightmare
<Church->
If I could just map the applications I'd be fine!
<Church->
I'm literally just hardcoding a root window and four children, with a hardcode app per window.
<Church->
Solely for my own personal use as I learn more xlib
thadthemad has joined #nixos
janat08 has quit [Ping timeout: 256 seconds]
<jasongrossman>
Church-: Have you looked at window manager code like TinyWM?
thadthemad has quit [Client Quit]
<Church->
jasongrossman: So I have a port of TinyWM I made as a base. But it doesn't do anything for child windows if I recall
<Church->
Or wait... did I miss something stupid.
Makaveli7 has joined #nixos
shibboleth has quit [Quit: shibboleth]
kisonecat has quit [Ping timeout: 240 seconds]
janat08 has joined #nixos
<shazow>
has anyone managed to get flatpak apps with tray icons working on NixOS? I'm getting some dbus-related warnings on everything I try (e.g. discord, steam) -- I tried adding xdg-desktop-portal-gtk but no change (running on i3wm fwiw, other native app tray icons like nm-applet show up)
<janat08>
i followed wiki about unfree, chrome still has no matches
<gchristensen>
are you trying to run nix-env -iA? or put it in environment.systemPackages?
Guest56801 has quit [Ping timeout: 258 seconds]
freeman42x[nix] has joined #nixos
<janat08>
trying to run nix-env -qaP
Guanin has joined #nixos
<janat08>
i have allow unfree everywhere: nixpkgs.config = { allowUnfree = true; # more stuff };
freedan42x has quit [Quit: Leaving]
abathur has joined #nixos
<tcosmo_>
Hello, do you know why there is "ps" in the following expression:
<tcosmo_>
python35.withPackages (ps: with ps; [ numpy toolz ])
<gchristensen>
janat08: nixpkgs.config ONLY impacts the configuration.nix. if you want to run nix-env as a user, you have to set the allowUnfree option _per user_. so, if you want to run it as root , you have to put it at ~root/.config/nixpkgs/config.nix
<__monty__>
tcosmo_: withPackages takes a function and gives a package set to it.
<__monty__>
The with ps, opens up that package set record so you don't have to type ps.nump ps.tools ps.etc...
ixxie has quit [Remote host closed the connection]
o1lo01ol1o has quit [Remote host closed the connection]
<tcosmo_>
__monty__: thank you for your answer but I dont get why ps replace "python35Packages"
<sphalerite>
tcosmo_: if you add (python36.withPackages …) to the path for the systemd service, you can use the python interpreter in the script or whatever and it will have all those modules availbale
<janat08>
is there a way in which i have to set up config.nix or does it get read automatically
<janat08>
cuz i had to create it
<tcosmo_>
sphalerite: Gonna try! But I dont understand why what it will produce wont be again too long
<sphalerite>
tcosmo_: because it creates a forest of symlinks to the individual packages
<sphalerite>
tcosmo_: sort of like a virtualenv with all the packages installed
<sphalerite>
and only adds that to the path
<tcosmo_>
sphalerite: ahhhh makes sense gonna try rn!
<tcosmo_>
sphalerite: cheers!!!
<{^_^}>
[nixpkgs] @nbp merged pull request #53397 → lib/modules: Add function to create option alias that respects priority → https://git.io/fhOT9
<ivan>
why does my machine complain about 'warning: substituter 'http://ra.wg:5000' does not have a valid signature for path ...' when on ra.wg I did `nix sign-paths -k /var/secrets/nix/secret_key --all` and `NIX_SECRET_KEY_FILE=/var/secrets/nix/secret_key nix-serve -l 10.10.0.1:5000`
<tcosmo_>
sphalerite: path = [ pkgs.python36.withPackages ( pkg.propagatedBuildInputs) ]; results in a error: cannot coerce a function to a string would you have any idea why?
<plakband>
I am trying to install NixOS on an old mac mini. The installation from the graphical live CD went flawlessly (with modeset), but when booting from disk I get a black screen after GRUB. Anyone have any tips? I'm now looking through the live cd config, since they work, but no luck yet.
<gchristensen>
was just reminded of this Pro Tip (tm): systemd.services.nix-gc.unitConfig.ConditionACPower = true;
<ivan>
plakband: at grub press e to edit boot options and add nomodeset?
mtesseract has joined #nixos
freeman42x[nix] has quit [Ping timeout: 258 seconds]
<ivan>
er, modeset, whatever you need
<ivan>
never mind I guess I misread
<mtesseract>
Hi, I have a question regarding types. When I evaluate ":t (import <nixpkgs> { }).fetchgit" in nix repl, it says "a set". But it is my understanding that fetchgit is actually a function which expects an attribute set as input ({ url = "..."; sha256 = "..."; ...}). Could somebody explain this discrepancy?
Guanin has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @fpletz pushed 2 commits to release-18.09: https://git.io/fhcme
gagbo has joined #nixos
<freeman42y>
is there a way to tell NixOS to connect only on ipv4 on ethernet? I connected my Windows laptop with the NixOS laptop and they have connected using IPV6 and SSH does not see the NixOS ssh server
slyfox_ has joined #nixos
slyfox has quit [Ping timeout: 240 seconds]
slyfox_ is now known as slyfox
o1lo01ol1o has quit [Remote host closed the connection]
<ottidmes>
freeman42y: I have used this in the past in NixOS to disable IPv6 for my ethernet: boot.kernel.sysctl."net.ipv6.conf.eth0.disable_ipv6" = 1; (note that you probably are not using eth0)
<{^_^}>
[nixpkgs] @lheckemann opened pull request #53965 → nixos/zfs: autoscrub only after boot is complete → https://git.io/fhcma
o1lo01ol1o has joined #nixos
<ottidmes>
freeman42y: it seems you can also just say: ssh -4 ... since that "Forces ssh to use IPv4 addresses only."
babic has joined #nixos
erictapen has joined #nixos
<dmj`>
I have a derivation in nix
<dmj`>
how can I make the set (the argument) overrideable
<ottidmes>
freeman42y: so you have like Win -IPv6-> NixOS laptop, but you then try to connect to a NixOS server from your NixOS laptop and it fails because it tries to connect over IPv6 as well, while your server does not have support for this?
<ottidmes>
freeman42y: if that is the case, could you not configure the laptop to connect to the server via ssh -4 at that point?
<freeman42y>
ottidmes, I am trying to connect from Windows laptop to NixOS laptop
freeman42y[nix] has quit [Remote host closed the connection]
cyounkins has quit [Remote host closed the connection]
Dagger has quit [Excess Flood]
<freeman42y>
I'm trying the configuration you mentioned. Hopefully it gets NixOS to use ipv4 and hopefully then Windows will see it
cyounkins has joined #nixos
civodul has joined #nixos
reinhardt has quit [Quit: Leaving]
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Dagger has joined #nixos
orivej has quit [Ping timeout: 245 seconds]
cyounkins has quit [Ping timeout: 258 seconds]
<freeman42y>
nope, it still does not work. It gets some IPv6 temporarily while it tries to connect and then after spinning the connection item a while it disconnects
<ottidmes>
freeman42y: there is also boot.kernelParams = [ "ipv6.disable=1" ]; but it caused some programs to complain for me
<freeman42y>
ottidmes, I wonder if this might not be fixable by disabling IPv6 on Windows side
<Dagger>
rather than disable v6, wouldn't it be better to just get the v6 working?
freeman42z has joined #nixos
<ottidmes>
freeman42y: AddressFamily in sshd_config might be used to limit it to IPv4
<freeman42z>
Dagger, <<rather than disable v6, wouldn't it be better to just get the v6 working?>> sure, but how?
<Dagger>
what do you mean by "SSH does not see the NixOS ssh server"?
<freeman42z>
Dagger, it doesn't see it because there is no IP allocated to NixOS. As I said, it connects for a bit, gives it an IPv6 IP but then it stops icon spinning and disconnects
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<Dagger>
hm... I don't have a specific answer, but it sounds like the sort of thing that e.g. network-manager would do if you configured the interface to "require v4, don't care about v6"
_kwstas has quit [Quit: _kwstas]
<Dagger>
it would set the interface up, wait for v4 to fail, and then tear it down again
freeman42y has quit [Ping timeout: 252 seconds]
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
<freeman42z>
Dagger, I changed from the GUI interface settings to require IPv4, let's see if it works
<Dagger>
FWIW you're connected to IRC over v6, so that part seems to work (assuming you're connecting from the same machine)
<Dagger>
are you connecting these two machines directly? i.e. with a single ethernet cable between the two?
plakband has quit [Quit: WeeChat 2.3]
<freeman42z>
Dagger, yup, direct connection with single cable between them
<freeman42z>
I've done this before with all sorts of machines and it worked
Makaveli7 has quit [Quit: Leaving]
<Dagger>
so there's probably no DHCP server on the link (unless you've configured one?) to give them IPs
<Dagger>
Windows will automatically generate an IP in 169.254.0.0/16. on Linux it depends
<freeman42z>
Dagger, afaik at least windows will act as DHCP for the link
<Dagger>
see if you can find a setting for "link-local v4", "auto config" or something similar
<Dagger>
otherwise... set it to ignore v4 and then either manually configure an address or just use the v6
<Dagger>
or set it to static and give it an IP in 169.254.0.0/16, that would make sense (I just prefer to push v6 whenever I can, and it works here because all v6 interfaces automatically get link-local addresses)
<freeman42z>
Dagger, I found the Link Local setting in the GUI. Thank you very much. Problem solved.
mtesseract has joined #nixos
ng0 has quit [Quit: Alexa, when is the end of world?]
<{^_^}>
[nixpkgs] @dywedir pushed commit from @jfrankenau to master « darktable: 2.4.4 -> 2.6.0 »: https://git.io/fhcsY
* gchristensen
is being punished by nix-shell using bashInteractive, despite it being an undeclared input
janat08 has quit [Ping timeout: 256 seconds]
<gchristensen>
even the smallest impurities are punishing!
<aminechikhaoui>
I want to override sbt package with a different jre, and seems that sbt has jre as an argument to callPackage not to the derivation, is there a simple way to override jre as I guess overrideAttrs won't work in this case
<ottidmes>
aminechikhaoui: override should work for the arguments to a callPackage'd package
<clever>
aminechikhaoui: any params passed at the callPackage level are changed with .override
<ottidmes>
aminechikhaoui: override "It is used to override the arguments passed to a function.", i.e. the arguments passed to callPackage for sbt, i.e. you can just do sbt.override { jre = myJRE; }
<aminechikhaoui>
oh ok nice
<aminechikhaoui>
thanks clever and ottidmes !
<ivan>
does nix-serve actually work? I tried a lot of things but I'm seeing behavior like the stapled Sig:s are incorrect or something
<ivan>
does nix have a cache somewhere that would cause it to complain about signatures without hitting the cache again
<clever>
ivan: there was a bug in nix-serve at one point, where it generated invalid sigs
<clever>
ivan: nix does also cache the narinfo files locally
<clever>
[root@amd-nixos:~]$ ls -lh ~/.cache/nix/binary-cache-v5.sqlite*
<clever>
ivan: in there
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 250 seconds]
Mateon3 is now known as Mateon1
sanscoeu_ has joined #nixos
Dedalo has joined #nixos
sinner has joined #nixos
sinner is now known as Guest29717
sanscoeur has quit [Ping timeout: 240 seconds]
verite- has joined #nixos
erictapen has quit [Ping timeout: 268 seconds]
das_j_ has joined #nixos
teozkr_ has joined #nixos
pingveno_ has joined #nixos
Haskellfant has joined #nixos
systemfault_ has joined #nixos
zopsi_ has joined #nixos
chris|_ has joined #nixos
cbarrett_ has joined #nixos
erictapen has joined #nixos
pointfree_ has joined #nixos
teej_ has joined #nixos
lopsided98_ has joined #nixos
Ariakenom has quit [Quit: Leaving]
Glenn_S has joined #nixos
beaky1 has joined #nixos
nokomprendo has joined #nixos
Taneb has quit [Quit: I seem to have stopped.]
<Dedalo>
I'm trying to setup a samba share in NixOS. I'm using the istructions from the Wiki. I use as server my router with Tomato firmaware and a hdd connected via USB 2.0 (because 3.0 doesn't work). The hdd is formatted with ext3 and it works, I'm running a torrent client on it (on the router). I want to share this hdd using Samba, so I have enabled Samba in Tomato, with a username and a password. Then I have
<Dedalo>
modified my configuration.nix but I can't connect. I have seen that NixOS creates a dir /mnt/nas (I called the share nas), but I can't access. Any idea?
<infinisil>
But that hardly qualifies as a usecase
<Dedalo>
ivan: how do I specify vers=1 in the configuration.nix? fileSystems automount_ops?
<ivan>
if it's like for other filesystems then it's in options = [ ] but I haven't tested
abaiste has quit []
<ottidmes>
Dedalo: ivan: I can confirm that that is the case, I use it to specify vers=3.02 myself
hamishmack has joined #nixos
<ivan>
note something after Linux 4.14 regressed vers=1.0, it's kind of busted in 4.19, or it was for me when I tested a few weeks ago
<ivan>
writes executable files on the server, breaks pretty often
sanscoeur has quit [Remote host closed the connection]
<ivan>
had to switch to nfs4 which is working much better
_kwstas has quit [Quit: _kwstas]
<ivan>
maybe because it was pointed at a debian 9.6 samba instead of the very latest samba, dunno
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
<ottidmes>
infinisil: does every package have a stdenv attribute? if I just thought of a sure way to get the file location of the package via unsafeGetAttrPos for any callPackage'd package (I need override)
ThatDocsLady has quit [Remote host closed the connection]
<Dedalo>
I have also a Media DLNA Server on this router. I have no idea how to connect to it. Do I need to configure avahi client?
<duairc>
If I wanted to install ancient versions of GHC, what can I do? When I developed on my Debian system, for certain libraries I develop I'd test that they worked with very old versions of GHC. I see that old nixpkgs have derivations for newer versions of Haskell. They should theoretically still work, right?
<clever>
duairc: yeah, just give them unique names in nix-channel
<clever>
duairc: or just check it out in a git clone and ignore nix-channel entirely
<duairc>
And do I have to do anything special to install from that channel? If I'm adding a package to environment.systemPackages, pkgs will refer to the main nixpkgs, I guess I'd have to import the old channel differently?
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
maskerad has quit [Quit: leaving]
<benley>
someone remind me, is there a way to do the equivalent of python.withPackages or python.buildEnv but in a way that only puts the executables from the installed python deps in the output derivation, and _not_ the whole python distribution?
<benley>
I've got a case where package A provides the entrypoint to an app, but for it to be useful you have to install packages B, C, and D, but B C D depend on package A. And package A doesn't depend on B, C, D.
doyougnu has joined #nixos
<benley>
so I can package those all as libraries just fine, but I haven't figured out how to make a final "user" derivation that just includes the entrypoint to the whole app
<clever>
duairc: if you name the channel 'oldchannel' then you need to do `let oldpkgs = import <oldchannel> {}; in ...` in your configuration.nix
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
clever: I am trying to override derivation via scopedImport, but it seems scopedImport with __derivation does not seem to work, any idea?
<duairc>
clever: Okay, great, thanks.
cyounkins has quit [Remote host closed the connection]
<gchristensen>
many moons ago I had this problem wehre the firefox download file window was 2x larger than my screen, and someone here had a way to fix it... anyone remember? :D
cyounkins has joined #nixos
<clever>
ottidmes: a: are you ensuring that scopedImport is recursive?
<samueldr>
gchristensen: some gtk ini setting file
<gchristensen>
samueldr: ah, yes!
<gchristensen>
samueldr: your memory :o
sigmundv_ has joined #nixos
<duairc>
I actually use nix.nixPath to set nixpkgs to a local clone of the git repository, it'd be cool if I could somehow do nix.nixPath = ["oldchannel=/path/to/nixpkgs@specificcommit"] instead of needing to have multiple clones... is there anything like that?
<ottidmes>
clever: I want to experiment with unsafeGetAttrPos on the attrset given to a derivation, I then abort with a toJSON stringified version of this information so I can actually extract it :P
<clever>
or builtins.fetchgit
<clever>
ottidmes: :D, but i need to leave for groceries, i'll be back in an hour
<gchristensen>
ottidmes: dear god
<{^_^}>
[nixpkgs] @ElvishJerricco opened pull request #53978 → GHCJS: Fix 8.4 and add 8.6 → https://git.io/fhcEV
<ottidmes>
muwhahah muwhahah!
<ottidmes>
clever: ok
<ottidmes>
gchristensen: always see what the limits of your tools are ;)
<duairc>
clever: Yeah I suppose I could do that. Thanks!
strobelight has quit [Ping timeout: 252 seconds]
<gchristensen>
ottidmes: well, did you make scopedImport recursive?
<ottidmes>
gchristensen: they probably are rewritten like that, but scopedImport does not expect them named as such
<gchristensen>
eh?
<ottidmes>
nevermind
<ottidmes>
<nixpkgs> is rewritten as __findFile __nixPath, and + to __add apparently, but that does not seem to be the case for things like import and abort
<gchristensen>
ah
abathur has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed 7 commits to release-18.09: https://git.io/fhcul
<ottidmes>
my idea was not going to work though, mkDerivation manipulates too much stuff before it gets to derivation, so if I use the same trickery with a more complex import to override mkDerivation somehow, I could probably make it work for all packages defined with mkDerivation, but I have put it down as a nice puzzle when I feel like solving such a thing ;)
beaky1 has quit [Quit: WeebChat 2.0.1]
<ottidmes>
(the idea/goal was to get a more solid answer to the question, in what file is this derivation defined)
<gchristensen>
ottidmes: what methods have you attempted?
<clever>
ottidmes: mkDerivation is easy to override
<ottidmes>
clever: it is? like everywhere without changing the source code in nixpkgs?
beaky has joined #nixos
<clever>
ottidmes: do an override on stdenv i think, and set it to be stdenv // { mkDerivation = args: ...; }
<clever>
then everybody calls the "wrong" function
<ottidmes>
clever: but not everybody uses stdenv right?
o1lo01ol1o has quit [Remote host closed the connection]
<ottidmes>
gccStdenv for example
o1lo01ol1o has joined #nixos
<ottidmes>
but I guess that is defined in terms of