<{^_^}>
[nixpkgs] @Synthetica9 opened pull request #49769 → [WIP] Implement RFC0035 II: Electric Boogaloo → https://git.io/fxjcN
thomasjm has joined #nixos
<thomasjm>
question -- is it possible to regenerate the nix DB using only the store? (i'm hoping to improve how multi-layer images work with dockerTools)
phreedom_ has joined #nixos
phreedom has quit [Quit: No Ping reply in 180 seconds.]
jperras has quit [Read error: Connection reset by peer]
jperras has joined #nixos
oida has quit [Ping timeout: 256 seconds]
oida has joined #nixos
jtojnar has quit [Quit: jtojnar]
fresheyeball has quit [Quit: WeeChat 2.0]
jperras has quit [Quit: WeeChat 2.2]
romanofskiWork has quit [Ping timeout: 268 seconds]
dbmikus has quit [Quit: WeeChat 1.9.1]
<{^_^}>
[nixpkgs] @globin pushed 0 commits to docs-make-channel: https://git.io/fxjWf
<{^_^}>
Found in packages: which, busybox, debianutils
DeepwoodMotte has joined #nixos
Mr_Keyser_Soze has quit [Ping timeout: 268 seconds]
<DeepwoodMotte>
Hi! I'm looking for bit of help. I have Docker installed, and I only use it ocasionally. What I would normally do is `systemctl enable docker` whenever I wanted to use it, but that no longer works. The only way I can get it to work is by enabling it in configuration.nix, which seems to have the service always-on.
<DeepwoodMotte>
Is there a way to enable the docker service as needed?
fendor has quit [Read error: Connection reset by peer]
justanotheruser has quit [Ping timeout: 252 seconds]
sigmundv__ has quit [Read error: Connection reset by peer]
DeepwoodMotte has quit [Ping timeout: 246 seconds]
DeepwoodMotte34 is now known as DeepwoodMotte
peacememories has joined #nixos
<DeepwoodMotte>
colemickens Unfortunately I'm not getting that to work for me. I put that line in my configuration.nix, ran `nixos-rebuild switch`, and rebooted the machine for good measure. Trying to run any docker still warns me the daemon is not running.
erictapen has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fxjl8
rprije has joined #nixos
hlolli has quit [Remote host closed the connection]
romanofskiWork has quit [Ping timeout: 268 seconds]
<DeepwoodMotte>
i think I found out what I was doing wrong regarding docker. I was installing docker via `environment.systemPackages` instead of `virtualisation.docker.enable = true`.
init_6 has joined #nixos
<colemickens>
Oh yeah, it wouldn't have actually been running at all then
<colemickens>
Sorry, I hadn't realized you'd omitted that due to wanting lazy start :)
romanofskiWork has joined #nixos
wykurz has quit [Ping timeout: 240 seconds]
<DeepwoodMotte>
No worries! There was no way for you to kow!
<DeepwoodMotte>
know*
selfsymmetric-mu has joined #nixos
<selfsymmetric-mu>
Aether looks cool (https://getaether.net/). It just launched, so there's nothing in nixpkgs for it yet. It installs with a snap file. Can snaps be installed in a NixOS system yet? I don't see "snap" in nixpkgs.
<simpson>
selfsymmetric-mu: Given this README, I wouldn't touch this code at all, but maybe some build instructions are in there https://github.com/nehbit/aether
<colemickens>
jfc another chat app
lassulus has quit [Ping timeout: 268 seconds]
lassulus_ is now known as lassulus
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<__red__>
so I have a mono application. My default.nix happily builds it but I end up with an application.exe
<__red__>
so - I guess I need to write a simple "wrapper"... mono /nix/store/jhegdkjhewqgkdjhgqewkjhd-appname-ver/bin/application.exe
<__red__>
?
<__red__>
is that considered 'normal' ?
badmutex has quit [Ping timeout: 246 seconds]
badmutex has joined #nixos
Diagon has quit [Quit: Leaving]
jasongrossman has joined #nixos
orivej has joined #nixos
ottidmes has quit [Ping timeout: 272 seconds]
zack_moe_ has quit [Ping timeout: 272 seconds]
maximili_ has joined #nixos
<DeepwoodMotte>
I ran `nix-prefetch-url --unpack https//someurl.com/`in order to calculate the sha256 sum of that file. How do I now remove the files it downloaded/unpacked.
<DeepwoodMotte>
ls
maximili_ has quit [Remote host closed the connection]
maximiliantagher has joined #nixos
<DeepwoodMotte>
looks lke `nix-collect-garbage` did the trick for me.
maximiliantagher has quit [Ping timeout: 240 seconds]
maximiliantagher has joined #nixos
maximiliantagher has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @jtojnar pushed 159 commits to gnome-3.30: https://git.io/fxjEj
dbmikus has quit [Remote host closed the connection]
Rusty1 has quit [Quit: Konversation terminated!]
ddellacosta has joined #nixos
dramforever has joined #nixos
badmutex has quit [Quit: Quit]
mayhewluke has quit [Ping timeout: 244 seconds]
mayhewluke has joined #nixos
ddellacosta has quit [Ping timeout: 240 seconds]
marusich has joined #nixos
hyper_ch2 has quit [Ping timeout: 256 seconds]
rprije has quit [Ping timeout: 252 seconds]
katyucha has joined #nixos
b has quit [Ping timeout: 245 seconds]
katyucha has quit [Read error: No route to host]
justbeingglad has joined #nixos
justbeingglad has left #nixos [#nixos]
sir_guy_carleton has quit [Quit: WeeChat 2.2]
jasongrossman has joined #nixos
peterHK_ has joined #nixos
maximiliantagher has quit [Remote host closed the connection]
<peterHK_>
hi
reinzelmann has joined #nixos
smolboye has joined #nixos
smolboye_ has quit [Ping timeout: 246 seconds]
smolboye has quit [Ping timeout: 240 seconds]
<{^_^}>
[nix] @goodwillcoding opened pull request #2516 → Update multiuser install remove instruction with correct path and users + group removal → https://git.io/fxjz0
maximiliantagher has joined #nixos
smolboye has joined #nixos
maximiliantagher has quit [Ping timeout: 252 seconds]
smolboye has quit [Ping timeout: 240 seconds]
dvim has joined #nixos
vk3wtf has quit [Ping timeout: 252 seconds]
<run500>
i cant figure out how to package a python application to use a specific version of python
<run500>
im doing this at the beginning: let buildApp = pythonPackages.buildPythonApplication.override { python = pkgs.python37; };
Mateon3 has joined #nixos
<run500>
which seems to work for the interpreter, but some of the modules are being built for 3.6
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon3 is now known as Mateon1
<run500>
actually no.. that doesnt work at all. i cant find anything in the docs that talks about this
hyper_ch2 has joined #nixos
<run500>
perhaps this is supposed to work: let pythonPackages = pkgs.python37Packages; in pythonPackages.buildPythonApplication ...
<run500>
but then it seems gst-python has a module for 3.6 in its store path; ...-python-1.14.0/lib/python3.7/site-packages/gi/overrides/_gi_gst.cpython-36m-x86_64-linux-gnu.so
<run500>
maybe a bug in gst-python package?
romanofskiWork has joined #nixos
romanofskiWork has quit [Client Quit]
dramforever has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixos-homepage] @edolstra merged pull request #249 → nixos/packages: display license long name first → https://git.io/fxhpD
<{^_^}>
[nixos-homepage] @edolstra pushed 2 commits to master: https://git.io/fxjgH
periklis has joined #nixos
shabius has joined #nixos
periklis has quit [Read error: Connection reset by peer]
peterHK_ has quit [Ping timeout: 256 seconds]
marusich has quit [Quit: Leaving]
Ariakenom has joined #nixos
smolboye has joined #nixos
maximiliantagher has joined #nixos
vk3wtf has joined #nixos
maximiliantagher has quit [Ping timeout: 272 seconds]
jasongrossman has quit [Ping timeout: 252 seconds]
<lionello>
How can I avoid having a lot of `system = "x86_64-linux"` all over the place? Started passing "system" into each package by putting `{system}:` on top
<lionello>
I'm sure there's a better way to do this
periklis has joined #nixos
<gchristensen>
what do you use `system` for in the package?
ThatDocsLady has joined #nixos
<lionello>
To make sure any `import nixpkgs` has it
<lionello>
I guess I could move all of those to the top instead...
<lionello>
For example, ` importedNixpkgs = (import nixpkgs { inherit system; });
<lionello>
(where nixpkgs is a pinned revision of the nixpkgs tree)
<periklis>
curl: Couldn't find host cache.nixos.org in the .netrc file; using defaults
<periklis>
and curl-ing the urls, i get 404 from varnish
<periklis>
the only difference that i could suspect is that at work i get fastly
<gchristensen>
fastly uses varnish
<gchristensen>
the binary cache is checked for every derivation, so the cache issues a lot of 404s. it is not unexpected. it seems maybe something else is going on
maximiliantagher has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fxjot
<periklis>
although with fallback=true it should progress, or not?
<gchristensen>
a 404 indicates it must build, no need for fallback=true
<gchristensen>
fallback is for when it fails to get a reply from the cache
<periklis>
ok then it's unrelated to the last line: `error: unexpected end-of-file`
<periklis>
the interesting thing is that it happens after instantiation
goibhniu has joined #nixos
Tucky has joined #nixos
<periklis>
and nix-index could not retrieve 43871 paths :(
periklis has quit [Remote host closed the connection]
periklis has joined #nixos
Itkovian has joined #nixos
<{^_^}>
[nixpkgs] @srhb opened pull request #49777 → Revert "NixOS tests: Wait for shell for 10x longer (50m)" → https://git.io/fxjKR
tonic_ has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to master « haskell-cpython: revert "set to null" »: https://git.io/fxjKV
<srhb>
periklis: Have you been debugging dns etc. already?
<periklis>
srhb: not yet, since dig ok afaik for cache.nixos.org
peterHK has joined #nixos
thc202 has joined #nixos
<peterHK>
hey, we have a server which was deployed with nixops. is there a way to get the configuration.nix somehow? (the nixops deployment was never pushed)
<srhb>
peterHK: There is not configuration.nix in a traditional sense
<srhb>
peterHK: The expression is that which is used by nixops
<srhb>
peterHK: It builds the entire finished system closure and pushes that to the server, it does not build a configuration.nix and runs that on the server.
<peterHK>
i thought so :( so no way to get the config without the nixops deployment
<srhb>
s/There is not/there is no
<srhb>
peterHK: If you have access to the server, you can probably at least partially reverse engineer it from examining the running system closure.
<srhb>
peterHK: (Well, fully if you _really_ dig deep)
<srhb>
I forget whether nixops copies the drv to the target system, but if it does nix-store --query --deriver /run/current-system will be helpful
<periklis>
that's interesting on mojave
<periklis>
dtrace: error on enabled probe ID 2172 (ID 161: syscall::write:return): invalid kernel access in action #12 at DIF offset 68
<periklis>
ok i think i found my error, thanks to dtruss: error: file 'nixpkgs' was not found in the Nix search path (add it using $NIX_PATH or -I)
<periklis>
which is not visible in a multi-user installation on darwin
maximiliantagher has quit [Ping timeout: 245 seconds]
patrl has joined #nixos
<patrl>
anyone using the dwarf-fortress-full package? I keep getting a dfhack error "Not a known DF version - XML version mismatch" on latest 18.09
Taneb has joined #nixos
Khetzal has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @gebner opened pull request #49778 → nixos/rngd: do not pass --version flag → https://git.io/fxj1i
justanotheruser has quit [Ping timeout: 268 seconds]
__monty__ has joined #nixos
ddellacosta has joined #nixos
leotaku has quit [Ping timeout: 240 seconds]
<Izorkin>
ho is used
<Izorkin>
Who is used opensmtpd?
ddellacosta has quit [Ping timeout: 252 seconds]
<__monty__>
We say "Yo ho!" but we don't say "Ho," 'cause ho is disrespectful, yo!
<Izorkin>
error running with pki, how to fix?
<srhb>
Izorkin: Check the git log for the nix file of the package for a list of probable suspects.
johanot has joined #nixos
vk3wtf has quit [Ping timeout: 264 seconds]
maximiliantagher has joined #nixos
<periklis>
srhb & gchristensen : i think i find a curious case. After dtracing for the error 'unexpected end of file', i found that NIX_PATH does not include nixpkgs. However, when i call nix build e.g. with `sudo -E nix build nix-index` it works.
<srhb>
periklis: That seems unlikely to be the cause of the fetch errors
<srhb>
periklis: If it fails to download <somehash> from the cache, clearly it had _some_ nixpkgs.
<sphalerite>
periklis: it also seems odd that `nix build nix-index` would work at all
<srhb>
patrl: The error is also there on current unstable. Maybe create an issue and ping @numinit in it? It seems likely to be fallout from the recent refactor.
<periklis>
sphalerite: why odd? `nix-build -f . nix-index` in my nixpkgs checkout should work like always
<srhb>
patrol: Or maybe lassulus knows
<sphalerite>
periklis: oh right, yes, with `-f .` it should work
maximiliantagher has quit [Ping timeout: 240 seconds]
Khetzal has joined #nixos
justanotheruser has joined #nixos
<lassulus>
oh yeah, someone should open an issue for that, I wanted to look into that but got sidetracked this weekend
aanderse has quit [Ping timeout: 250 seconds]
<patrl>
srhb: thanks for the confirmation on the error. Will do!
<srhb>
patrl: Great :) Feel free to ping the involved parties (git log of the file is very helpful to find the people likely to help out) :)
<periklis>
srhb: & sphalerite in addition i see the following error in system.log: nix-daemon[37650]: objc[37650]: +[__NSPlaceholderArray initialize] may have been in progress in another thread when fork() was called. We cannot safely call it or ignore it in the fork() child process. Crashing instead. Set a breakpoint on objc_initializeAfterForkError to debug.
<srhb>
periklis: Sounds... Fun. I have no idea about OS X, sorry.
<Arahael>
How do I fix this with 'stack build'? error: attribute 'ghc844' missing, at (string):1:43
<srhb>
Arahael: Presumably by using an lts release that points at a ghc that is in your nixpkgs version, or by overriding the version of ghc used.
peterHK has quit [Ping timeout: 256 seconds]
<periklis>
srhb: Oh boy, this fixed the issue
<srhb>
periklis: wow, I'm useful. :-)
<Arahael>
srhb: Doesn't stack download the required haskell, though?
<srhb>
Arahael: Not when using nix integration.
<srhb>
It delegates that to nix itself.
<Arahael>
Ah, ok. So I can either somehow update my nixos, or downgrade my stack?
<locallycompact>
how do I do nix-copy-closure to a machine I deployed with nixops
<srhb>
Arahael: I think you just want to specify a different lts release (that is, one that uses a ghc that is actually in your nixpkgs)
<Arahael>
srhb: I think you're right. Btw, I'm using the default one from 'stack new simple'.
<srhb>
locallycompact: Extract the key for root from your nixops install or create a user for yourself with the necessary ssh/store permissions
<srhb>
Arahael: Sorry, I don't use stack so I'm minimally useful on this topic. :)
<Arahael>
That's ok. :)
kisik21 has joined #nixos
<srhb>
Arahael: fwiw nixos-unstable since yesterday's bump has ghc844
<Arahael>
Setting it to LTS 12.14 worked well, and that's only 2 weeks ago.
<Arahael>
Ah, I think I'm probably best avoiding unstable for now. Thanks. :)
kisik21 is now known as vika
sigmundv has joined #nixos
juliendehos has joined #nixos
brodul has joined #nixos
rprije has joined #nixos
<hyper_ch2>
q: has anyone every used pandoc and is conversion good from md to like word?
<srhb>
pandoc is excellent.
<locallycompact>
I use pandoc a lot but I don't output to word
<srhb>
I've very rarely had to output to the varioud word processor formats. I find there's no better alternative, at least.
<hyper_ch2>
thx, will have a look
maximiliantagher has joined #nixos
orivej has joined #nixos
<locallycompact>
I made a custom user in the nixbld group on my server but when I try to nix-copy-closure I get error: cannot add path '/nix/store/1s753m9awnwgdlfrj5i0yb5ba3kcd45q-extra-udev-rules' because it lacks a valid signature
<Arahael>
Weird, I'm struggling to build my application that uses http-client, due to it's dependency on zlib.
<Arahael>
I tried using: nix-shell -p zlib; but it still fails to find z or zlib.h
<Arahael>
(That is, I modified simple.cabal to add 'http-client' to the build-depends section.)
fendor has joined #nixos
<srk>
Arahael: you can use cabal2nix . > simple.nix
angelsl has left #nixos ["Detaching"]
<Arahael>
srk: Let me try installing that.
erictapen has joined #nixos
<Arahael>
srk: How does that work with stack, though?
<srk>
for stack it is possible to add zlib to nix: section
<srk>
cabal2nix allows you to build with nix
<srk>
and also use nix-shell to enter the env with deps for your app
<srk>
and use cabal build / new-build there
<Arahael>
Right - shouldn't `nix-shell -p zlib` have been adequate?
<srk>
doubt so
<lionello>
Does it make sense to do a `{ system ? "x86_64-linux" }:
<Arahael>
Just looking for the documentation for cabal2nix.
<lionello>
...on top of file, and then `inherit system;` in any import nixpkgs?
betaboon has joined #nixos
<Arahael>
This seems complex. :(
<srk>
not really
<srk>
if you have a .cabal file it's quite easy
<srk>
you just need to call cabal2nix .
<Arahael>
srk: And then?
<Arahael>
srk: And I don't really have a .cabal file yet, except as the result of 'stack new simple'.
<srk>
on it, sec
<Arahael>
(I then added http-client, but that's it, for now, I plan to make other changes)
<Arahael>
I'm used to using stack on macos.
<srk>
you can use nix on macos as well ;)
<Arahael>
I'm sure, but right now I'm using nixos, and was hoping I could use stack regardless. :)
<srk>
you can, but it's more painful imho
<Arahael>
(And I don't really understand why nix-shell isn't sufficient, it's missing zlib, right? So start a shell with the zlib package added to it?)
<Arahael>
On other systems, I found stack _less_ painful.
<Arahael>
__monty__: I found stack *awesome*, because before I discovered stack, as a new haskell dev, who was barely figuring out this "show" business... Stack stopped wasting time for me.
orivej has joined #nixos
<Arahael>
I was otherwise trying to find haskell packages, somehow landing on an obscure one, then picking another haskell project to satisfy another requirement, and then having versioning hell.
sir_guy_carleton has joined #nixos
<Arahael>
It didn't help that brew, debian, and #haskell, were violently at odds with each other, and I didn't even realise until a while after. stack helpt keep things standard and consistent.
<Arahael>
To me, it made haskell possible to learn. :)
maximiliantagher has joined #nixos
<__monty__>
I rarely encountered that problem with cabal. Switched to stack, was happy for a little while. Then I noticed it was actually worse. Add to that all the damage stack has done to the community's image and I was just completely done with stack.
<srk>
they should just switch to nix as a backend at this point :D
pie_ has joined #nixos
<Arahael>
__monty__: I don't think stack has caused any damage to the community's image, putting any and all sorts of crap into hackage did that. And this cabal thing? That is super confusing.
<Arahael>
There's the cabal-the-library, cabal-the-app, cabal-the-build-system, or something?
<srk>
:D yeah
<__monty__>
Arahael: Nope, just the library and cabal-install the tool afaik.
<Arahael>
__monty__: I distinctly remember struggling with a third cabal item, but i can't remember what it is. Maybe it was cabal the configuration file format.
<__monty__>
I don't know why people even care about the difference? You use cabal-install, that there's a library of the same name is irrelevant.
<steveeJ>
any idea why a nmve device wouldn't show up in stage1?
<{^_^}>
[nixpkgs] @Mic92 merged pull request #49777 → Revert "NixOS tests: Wait for shell for 10x longer (50m)" → https://git.io/fxjKR
<pie_>
i recently saw that adding some packages to systemPackages seems to be able to set some other necessary things, do i need to enable config options to run virtualbox properly or just add it to systempackages?
<Arahael>
__monty__: I can't remember, but it was somehow super relevant when I was originally trying to figure it out.
<srk>
steveeJ: missing kernel module
<pie_>
hm maybe theres sometihng in the vbox nix script, i couldnt find any documentation
justbeingglad has joined #nixos
<__monty__>
Arahael: Since this is offtopic let's move to #nixos-chat if you want to discuss it further.
<steveeJ>
srk: I added nmve and nmve_core
maximiliantagher has quit [Ping timeout: 240 seconds]
<Arahael>
I work in the enterprise, too - stack also appealed to me because it gave a possibility of developing a company-blessed, curated package collection.
<__monty__>
I'm happy as a clam with nix+cabal new-* fwiw.
<Arahael>
__monty__: Sure, I need to get to bed.
<srk>
steveeJ: to boot.initrd.kernelModules?
<__monty__>
Stackage is completely seperate from stack.
<steveeJ>
srk: yes
<Arahael>
__monty__: Different name, at least. :)
<Arahael>
__monty__: Anyway, sorry about that, but I'm still annoyed at how hard it was to get into haskell with just cabal.
<Arahael>
srk: Will keep your tabs open until I look into them again later in the week. :) Thanks for your help!
jbgi has quit [Ping timeout: 264 seconds]
<Arahael>
srk: (Also, LoRaWAN looks cool!)
<srk>
Arahael: :) yw!
<__monty__>
Arahael: Tbf, your problems sound like they were more due to your magpie-ing for libraries to do things than the tool.
<Arahael>
__monty__: How else does someoene who barely know haskell get started?
<__monty__>
By writing basic haskell.
jbgi has joined #nixos
lionello has quit [Ping timeout: 252 seconds]
<srk>
steveeJ: not sure actually, you can try poking udev again with udevadm trigger --add if you have shell, might also be a missing udev rule but that's improbable
<Arahael>
__monty__: Sure. brew install haskell;
<Arahael>
Things are much better now, though.
dvim has quit [Quit: WeeChat 2.3]
<{^_^}>
[nixpkgs] @gebner merged pull request #49778 → nixos/rngd: do not pass --version flag → https://git.io/fxj1i
<steveeJ>
srk: the live system I have on an MMC can see it when booted. I'll go over `lspci -k` to see what's loaded
justbeingglad has left #nixos [#nixos]
<srk>
mm, that's handy, was always digging this from lspci -vvv /o\
<__monty__>
Arahael: You asked why someone'd stop using stack, I answered. No need to defend stack, if it works for you then fine, just know there's shortcomings to it too and cabal's new-* commands have worked better for me than either the old ones or stack. I can see good reasons to use it in an industrial setting, I can also see good reasons to avoid using it in an industrial setting.
<Arahael>
__monty__: Ah, fair enough, you're right right.
<Arahael>
*quite right.
sir_guy_carleton has quit [Quit: WeeChat 2.2]
<Arahael>
__monty__: I certainly need to keep an open mind. G'night!
<steveeJ>
srk: typo :-D nmve vs nmve
<steveeJ>
ops, did the same typo again..
<srk>
:D
<ben>
has nix-env learned how to install manpages yet?
Thra11 has joined #nixos
<__red__>
Arahael, srk : Lora is kinda cool. Been building boards with it for a few years now (although I have avoided LoraWAN specifically)
<srk>
__red__: why? spec isn't that bad for application protocol
<srk>
about to write implementation for it for stm32l4 & rfm95w
garbas has joined #nixos
<__red__>
srk: Mainly because I don't want / need an infrastructure underneath it
<__red__>
and yeah- the hoperf boards are awesome. Ten Thumbs Up
maximiliantagher has quit [Ping timeout: 245 seconds]
<__red__>
OTP upgrades are the very antithesis of nix tho
<__monty__>
ben: What do you mean? I get manpages from nix-env.
<__monty__>
ben: Maybe your MANPATH isn't manly enough?
<jonge>
hey there, i have a c++ project with a working default.nix file. i would like to have some nice way to create a docker image that contains all the nix paths so that `nix-build` works in that docker image. i also know how to use dockerTools in nixpkgs, but i don't know how to express "put that nix-build environment there so it works on another source checkout"
<jonge>
the thing is that the devops admin told me "well you can use whatever build tools you want. my requirement is that you can give me a docker image that is selfcontained and that can run without internet". so i am trying to build just that.
jbgi has quit [Ping timeout: 250 seconds]
<ocharles>
Hi all. How do I configure Hydra to notify Slack? It seems I need a `<slack>` section with a `jobs` regex key and a `url` key - but what do I put in the url?
<srhb>
ocharles: It'll be a hooks.slack.com/... url
<simpson>
jonge: Can't stop laughing at the idea of a "devops admin". The basic idea would be to `FROM nixos/nix` with a standard Dockerfile, and call nix-shell during the Docker build with your default.nix in order to fetch everything into the container's Nix store.
<simpson>
And then running the container's nix-build later on will reuse that store. I can't guarantee that no Internets is needed, because if your default.nix changes what it wants, then you'll have to rebuild the container.
<__red__>
simpson: there was a time when all admins were devops
<the-kenny>
ekleog: Ping. I noticed you updated opensmtpd recently. After adapting my config, I get hard segfaults when connecting to Gmail for relaying a mail. Do you have any idea what this could be?
<__red__>
back when SunOS roamed the lands
<simpson>
__red__: Sure. But today jonge's problem is that somebody won't let the devs own their own CI process, I think.
<srhb>
jonge: Theoretically, you'll just have to traverse the buildInputs and provide those.
fpob has joined #nixos
maximiliantagher has joined #nixos
<srhb>
It strikes me as very odd that it's a _build_ process that's not allowed to have internet access.
<srhb>
(As opposed to the _built_ process)
<ben>
__monty__: I feel like `nix-env -iA nixpkgs.gcc && man gcc` should not fail
<simpson>
srhb: The Docker build server is probably living under this devops admin's desk~
<ben>
srhb: build process not having internet access seems reasonable to me. it's a goal here too to make builds not depend on external resources being up.
<srhb>
ben: Well, then I'd expect at least an internal server providing deps somewhere.. But maybe I'm insane. :)
<ben>
srhb: yeah, i was assuming there is an internal docker registry involved
jbgi has joined #nixos
<jonge>
simpson: you're right about the devops-admin thing. It's a crippled (mostly windows) development infrastructure and i am fighting hard to give my team non-demotivating development processes/infrastructure with nix.
<jonge>
they are using some teamcity build service they made their entire build processes dependent upon and it's even a very old version from which they can't upgrade for weird reasons. so the only chance to get nix in there is docker...
lionello has joined #nixos
maximiliantagher has quit [Ping timeout: 268 seconds]
<simpson>
This is a reasonable approach. At my $DAYJOB I accidentally did `FROM nixos/nix` without really thinking, but nobody complained, so now it's there.
<ekleog>
the-kenny: wow. no :/
<ekleog>
do you have any logs around the segfault?
<the-kenny>
ekleog: Yeah, journalctl omitted that line. Just a second
<the-kenny>
ekleog: Updated that gist
<__monty__>
ben: Hmm, manpage availability does seem a little hint and miss.
<the-kenny>
ugh no. Let me get you a clean one.
<ekleog>
the-kenny: ugh. that's not good :( like, ip and sp are *way* too close in this gist
<the-kenny>
It doesn't seem to be the same issue either - I'm not using pki via `listen` or similar.
<the-kenny>
This is a completely local relay
<ekleog>
well, both cases appear strongly related to tls
<the-kenny>
that's true
<jonge>
srhb: i also don't get why the build server is not allowed to access the internet for _building_ software. when you ask the admins it's "bla bla security can't change it, company policy bla"
<simpson>
jonge: Probably a policy from an era when builds *didn't* need network connectivity.
<jonge>
srhb: this results in someone building a working offline package for _weeks_ (because you can't automate things nicely in windows) and then they are not touching it for years because it works - lacking all the optimization and static analysis features as well as (security-!) bugfixes in toolchains and libraries.
<ben>
__monty__: My understanding is that everybody went and split manpages into a separate output for a lot of derivations, but nobody taught nix-env to install that output, so you have to write a bunch of custom nix per package to get at manpage now
<ekleog>
the-kenny: sanity-check: you're using nothing from opensmtpd-extras, right? I just now remembered this package should most likely be deleted
<the-kenny>
No, just the most basic `opensmtp.enable = true; opensmtpd.serverConfig = ...;`. Plus some debugging args in `extraServerArgs`
<ekleog>
ok :/
<__monty__>
ben: Maybe you need to use nix's man?
<ekleog>
the-kenny: can you try with an override from libressl to libressl_2_8? I don't think it's the issue, but just in case…
<ben>
__monty__: I am using nix's man :(
vk3wtf has joined #nixos
ghusbands has joined #nixos
<Izorkin>
ekleog: pasting journalctl, with libressl_2_8 not working
<betaboon>
can anyone tell me where i can find documentation on how something like this works: `{ pkgs, ... }@args : { .......}` i wonder about the @args stuff
<clever>
betaboon: the attribute set the function was called on, is stored in args
<ekleog>
Izorkin: ok so it looks like there's no segfault here…
<clever>
betaboon: yeah
<ghusbands>
Is there a way to get a list of installed package versions in a format that I can give to others, for them to get exactly the same setup as me, down to the version of every package?
<betaboon>
clever: thanks :)
* ekleog
doesn't have a 6.4 deployment with tls yet, so trying to reproduce in a nixos test
FRidh has joined #nixos
maximiliantagher has joined #nixos
sushi_ has quit [Ping timeout: 245 seconds]
orivej has quit [Ping timeout: 252 seconds]
orivej has joined #nixos
<ottidmes>
ghusbands: give them your configuration/customizations/nixpkgs commit
maximiliantagher has quit [Ping timeout: 244 seconds]
<disasm>
ghusbands: are you referring to just things installed in `nix-env`?
ghusbands has quit [Ping timeout: 256 seconds]
<eeva>
I'm trying to hunt down a “invalid character in Base64 string” any generic tips?
<eeva>
Nix is pretty much unusable for me :/
<ekleog>
the-kenny: is it OK with you if I share your gist on OpenSMTPD's issue tracker?
<eeva>
(both for rebuilding system config, or nix run, or nix build)
patrl has quit [Ping timeout: 264 seconds]
<ekleog>
I need to context-switch, and will summarize information from your bug report and Izorkin's there, as it appears more likely to be an upstream issue than a packaging issue to me at the mement
<symphorien>
ekleog: does nix show-config work ?
<symphorien>
eeva: ^ sorry
<betaboon>
is there something like `builtins.trace` but showing a complete attrset ?
<eeva>
it does symphorien
<symphorien>
did you try nix-build -vvvvvv something ?
<wirew0rm>
the nixpkgs manual recommends to look into lib/debug.nix for documentation of debug stuff
drewr has quit [Ping timeout: 250 seconds]
<betaboon>
wirew0rm: must have overread that section :/
shabius has quit [Ping timeout: 252 seconds]
adamt has joined #nixos
<disasm>
fpletz: how's the networkd stuff coming along? :)
mayhewluke has joined #nixos
<adamt>
Hiya. I'm having issues with setting additional sudo rules. I can add to security.sudo.extraRules, no problem, but the sudo module is hardcoded to add a SETENV rule for all of the wheel group, which ends up at the bottom of the list. Since sudo uses the last match, and not the most specific match, I need to something append my own rules to the end of the list.
<adamt>
I tried using mkOrder (0,1, 100, 900..), and no matter what my own rules are added before the predefined one.
<adamt>
Aaaand I would rather like to avoid mkForce, since I add rules in multiple places.
<adamt>
Any suggestions for a pretty solution? :-)
<clever>
adamt: what about mkAfter ?
<wirew0rm>
betaboon: no problem, happens to me all the time. it was meant more as a source/further-reading :)
<infinisil>
Yeah mkAfter should work (which is just mkOrder 1500)
<betaboon>
wirew0rm: had debug.nix opened as soon as you mentioned traceValSeqN XD
<adamt>
clever: Ah. That's clever. How does multiple invocations of mkAfter interact though?
<clever>
adamt: i believe they all just set a prio or 1500, and then it will be the order of the modules being imported
<adamt>
oh, so then mkOrder 1500 one place, and mkOrder 1600 another place should work as well? I thought the default order was 100
<srhb>
adamt: It's 1000 iirc.
<srhb>
(500 being before, 1500 after)
pie_ has quit [Ping timeout: 256 seconds]
<disasm>
`mkAfter = mkOrder 1500;`
<adamt>
Thanks guys, sorry for wasting your time instead of just trying higher numbers. :P
<disasm>
adamt: it's never a waste of time, I learned about mkAfter and mkOrder because of your question!
<adamt>
Where would one find the default ordering value though? Not that I don't believe you guys, I'm just curious. :-)
<adamt>
infinisil: Thanks. It looks so innocent and hidden :P
<srhb>
adamt: A nice property of nix functions looking like "foo = x: ..." is that you can almost always grep -r "foo =" through the sources, and anywhere that has inline documentation will be immediately discoverable.
<adamt>
srhb: Good point. Then I just need to know what to grep for next time. :P But I see mkAfter and mkBefore are defined just after mkOrder, so it would have saved me this time. :-)
<ekleog>
the-kenny: sorry to poke again, but is it OK with you if I share the link to your gist? as this is a publicly logged chan I'd assume that yes, but always prefer to ask
<srhb>
adamt: It *often* works at least. :)
jasongrossman has joined #nixos
<srhb>
adamt: Works is being done making inline documentation be generated into real documents as well, which will make this easier yet. :)
<adamt>
srhb: Yeah, I think that was mention during NixCon as well. It's going to be great.
periklis has quit [Ping timeout: 252 seconds]
<eeva>
betaboon: hmm, so it's not a problem with my substituters :/
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<eeva>
symphorien: ^^ sorry
<betaboon>
XD
<{^_^}>
[nixpkgs] @peti pushed to master « haskell-cpython: don't bind the obsolete name "python34" »: https://git.io/fxjhc
steshaw has quit [Quit: Connection closed for inactivity]
<eeva>
I'm at the point where I'm poking randomly trying to understand *what* base64 string is failing to decode
random_yanek has quit [Ping timeout: 245 seconds]
Itkovian has joined #nixos
<symphorien>
for example trusted-public-keys in nix show-config contains a base64 string, is it valid ?
<{^_^}>
[nixpkgs] @peti pushed 10 commits to haskell-updates: https://git.io/fxjhg
<eeva>
It's only the cache.nixos.org, so I should check its key
<srhb>
eeva: Did you post your -vvvv log somewhere yet?
<srhb>
More eyes...
worldofpeace has joined #nixos
<azazel>
uhumm if I have the same set of options that i need in two places (two different and loosely related config.a.b.c attributes) should I make a submodule or could I assign them to a variable and then reuse that variable on the places where I need the definition... or maybe again is better to make a function that returns a set with the options?
<srhb>
azazel: I think it's really hard to give generic advice on that without further context. Either might be the right approach.
<{^_^}>
[nixpkgs] @Zimmi48 opened pull request #49780 → coqPackages.ssreflect: refactor choice of source version → https://git.io/fxjjl
<srhb>
eeva: fwiw I can reproduce it by setting eg. the hash of hello to be a base64 hash but replace one of the characters with a '. So it might just be in a local hash expression?
<eeva>
Not sure what you mean exactly
patrl has joined #nixos
<eeva>
What do you run to reproduce this error?
<srhb>
I edited the hash for nixpkgs.hello in my local nixpkgs checkout and set the hash to "MeBmE3qWJnbon2nRtlOC3pWn732RS4y5VvQepy4PU's=", then nix-build . -A hello in the repository.
<exarkun2>
Okay so last week I ran into this collision (collision between `/nix/store/9ril7mcqrvd17gg0mdfqnqw5ad6ymfpn-python2.7-hyperlink-18.0.0/.pytest_cache/v/cache/nodeids' and `/nix/store/rz5lfh4w1hgx37ashi0xlx149lgfn8w5-python2.7-spake2-0.8/.pytest_cache/v/cache/nodeids') and now I'm trying to fix it. I added `rm -rf .pytest_cache` to the `preFixup` of the Python spake2 package in my nixpkgs-overlays.nix but it makes no
<exarkun2>
difference.
<exarkun2>
What am I doing wrong?
<clever>
eeva: try the -vvvvv again, but also add -j 1, then pastebin that log
<srhb>
eeva: My only point was that it could easily be a package that you're depending on that has a broken hash, and not just a configuration option.
<eeva>
I see
<azazel>
srhb: i'm updating a patch for adding configuration for prometheus 2 in parallel to prometheus 1 where for example both have an accompanying service named "alertmanager" which has been updated but the config is mostly the same ( for context https://github.com/NixOS/nixpkgs/pull/46069 )
<{^_^}>
#46069 (by azazel75, 8 weeks ago, open): Prometheus: upgrade configuration module to use the 2.x package
<simpson>
exarkun2: I think that that cache is generated during the check phase, so I'd try using the postCheck hook to remove it.
<simpson>
preFixup might just be too late.
<eeva>
srhb: yeah, this started 3 days ago
<exarkun2>
dang. And here I was worrying I might be running the cleanup too early.
* simpson
pondering py.test adding its own hook
<clever>
building of '/nix/store/dwiqyq7z55pml2zb20shs88xbiimx6q7-source.drv': woken up
<eeva>
According to nixos-upgrade.service
<clever>
eeva: can you pastebin that .drv file?
<eeva>
yes
<exarkun2>
simpson (IRC): When I started looking into the issue, I saw that pytest has a hook that tries to do this cleanup. I didn't manage to understand why it's not working. Maybe something about doing it at the wrong time like me?
* exarkun2
checks to make sure he spelled `postCheck` correctly
<azazel>
simpson: I remember something on this... wasn't pytest cache already added to the list of stuff to cleanup? I remember a similar discussion with someone and a lookup on the issues
<clever>
eeva: try `nix-store -r` on each path listed in your last pastebin, what ones work and what dont?
<exarkun2>
hmm when I added `doCheck = false;` the package didn't get rebuilt.
<clever>
eeva: and what is the contents of ~/.config/nix/nix.conf ?
sushi_ has quit [Quit: Konversation terminated!]
ilikeheaps has joined #nixos
<exarkun2>
yea my `postCheck` isn't being used at all afaict. Added a big echo to it, appears nowhere in build output - though changes to the value of postCheck _do_ provoke a rebuild...
orivej has quit [Ping timeout: 245 seconds]
<eeva>
clever: no such file :>[
<clever>
eeva: and what about running nix-store on each of those paths?
<clever>
-r
<eeva>
yeah, doing that
<exarkun2>
Bleh `postCheck` isn't making it into the derivation.
<exarkun2>
(so why does it rebuild?)
<eeva>
clever: all fail with the same error
<clever>
eeva: can you pastebin `nix show-config` ?
<exarkun2>
Apparently spake2 package runs its tests during installCheckPhase.
<exarkun2>
(and checkPhase)
<azazel>
sh, weird
<exarkun2>
I wonder why preFixup would be too late. Can I not just make any changes I want to anything in $out up until the very end? And then whatever is left is what goes into the package?
<simpson>
exarkun2: Given how Nix works, I think it'd be acceptable to only run the tests once during checkPhase. It's nice that they did that, though, I guess?
<clever>
exarkun2: yeah, thats how things should work
sb0 has quit [Ping timeout: 272 seconds]
civodul has joined #nixos
<exarkun2>
trying the cleanup in both postCheck and postInstallCheck now
<simpson>
exarkun2: Oh, installCheck is later than fixup. So that's why; you indeed weren't late enough and I was wrong on my first guess. You can hook with `postInstallCheck`.
<exarkun2>
simpson (IRC): Ah yea, I just noticed that too.
<exarkun2>
(after just randomly trying a bunch of stuff)
ddellacosta has joined #nixos
<exarkun2>
and I guess I can drop the cleanup from postCheck because it's useless
<eeva>
clever: am I missing this?
<exarkun2>
aaaaaaaaaaaa (still collision)
iyzsong has quit [Read error: Connection reset by peer]
<clever>
exarkun2: it should just download that from the binary cache
<clever>
eeva: oops, ^^
<eeva>
yeah… Not in my store :/
<clever>
eeva: and the above curl command output what?
<{^_^}>
OpenSMTPD/OpenSMTPD#886 (by Ekleog, 17 seconds ago, open): Errors related to TLS with OpenSMTPD 6.4.0p1 (incl. segfault)
<eeva>
thanks *a lot* for your patience
<the-kenny>
awesome, thanks!
<Izorkin>
ekleog: thanks
<ekleog>
that's not really awesome, though :p
Thra11 has quit [Ping timeout: 245 seconds]
<exarkun2>
I can't get my `postInstallCheck` script to run at all. It's just ignored. If I invent a custom phase and put it in postPhases then it runs and fixes the problem. But isn't `postInstallCheck` a real thing?
* ekleog
would have preferred things to just work :( if anyone sees a way to fix the nixos test to include some TLS operations I'd be happy to take it, haven't managed to get `nixos/tests/common/letsencrypt` to actually properly generate a certificate in the allocated time :/
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<das_j>
Hey, why does `nix run` not set the PYTHONPATH when using pyton2.withPackages?
<clever>
das_j: that handled by setup hooks usually, which only get ran by nix-shell and nix-build
<das_j>
clever: Is there a way to call them manually in nix run?
Itkovian has joined #nixos
<clever>
das_j: probably not
<das_j>
heck. Okay, I'll jsut stay with nix-shell
ghusbands has joined #nixos
maximiliantagher has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « ghc: add new version 8.6.2 »: https://git.io/fpeTT
<das_j>
clever++
<{^_^}>
clever's karma got increased to 42
<clever>
heh, hit the magic number!
<ghusbands>
disasm: I don't think so, but I'm not sure. (more to follow)
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ghusbands>
I want to publish a mini-whitepaper (or blog post) with exactly what's required to reproduce my results, including an easy way of getting exactly the same version of everything I was using.
Itkovian has joined #nixos
maximiliantagher has quit [Ping timeout: 252 seconds]
<ghusbands>
I thought Nix(OS) would be ideal for this, but I can't find a facility for publishing and then separately reproducing an exact/versioned set of packages.
<wirew0rm>
does anyone here have a nice workflow for looking at dependency graphs for nix expressions? gephi looks nice, but i think it would need some extra layouting algorithms for acyclic graphs to be really useful
<clever>
ghusbands: if you import a specific version of nixpkgs, an store all the expressions in git, you can reproduce the whole thing, as long as the src URL's arent deleted
<sphalerite>
ghusbands: if you publish the entirety of the nix expressions you used for it, along with the nixpkgs revision, it should be easy to reproduce
peterHK has joined #nixos
ilikeheaps has joined #nixos
<sphalerite>
ghusbands: typically that can be done once the expressions are obtained in a single command using -I nixpkgs=https://github.com/nixos/nixpkgs/archive/<commit-hash>.tar.gz
<sphalerite>
wirew0rm: "looking at" to what end?
<sphalerite>
I often find myself using inx-store -q --graph and just reading/searching the output directly in my text editor
<ghusbands>
clever, sphalerite: Thanks, folks, that gives me plenty to work with.
<clever>
ghusbands: you can also use things like `nix copy` to create a binary cache with all of the build products, as either an s3 bucket, static http server, or just a tarball
<wirew0rm>
sphalerite: i'm still trying to build my esp32 toolchain "the nix way" and I want to find out what stages are built in which order.
<wirew0rm>
sphalerite: raw output is very cluttered with all the things stdenv/etc pull in, so I would like to maybe filter out .patch, the tars, some trivial dependencies and then render in a useful way
<sphalerite>
wirew0rm: sounds like something that could probably be achieved with grep :D
<infinisil>
Convert to svg with `dot -Tsvg <(nix-store -q --graph <store path>) > graph.svg`
<infinisil>
(from the graphviz package)
<das_j>
Does anyone know how to apply overlays always? I'm currently defining them in a common file included on all hosts, but they would be also useful for non-system packages like for nix-shell -p
<wirew0rm>
infinisil: yes, that's what I tried first, but a cross compiled gcc does not yield a useful graph with that.
<infinisil>
Why not?
<wirew0rm>
it can probably be done with graphviz but I would need a lot of filters and layout options
<sphalerite>
das_j: put them in the nixpkgs-overlays NIX_PATH entry
<wirew0rm>
infinisil: ~200 nodes and ~1000 edges
<infinisil>
Hmm.. I see
<das_j>
sphalerite: What do I put there? An attrset to the overlays?
<infinisil>
wirew0rm: Have you checkout out --tree?
<sphalerite>
das_j: a nix expression file containing a list of overlays iirc. Not 100% sure though
endformationage has joined #nixos
<exarkun2>
does overrideAttrs work on Haskell packages? :/
<clever>
exarkun2: yes and no
<clever>
exarkun2: it works, but it applies after generic-builder.nix and they cease to be "haskell packages"
<wirew0rm>
infinisil: yes, same problems basically... each package depends on basic_builder and the like + all their dependencies
<clever>
exarkun2: overrideCabal applies before generic-builder.nix
<clever>
exarkun2: which one you want, depends on what your trying to modify
<exarkun2>
overrideCabal sounds plausible. I'm trying to insert a test dependency.
<clever>
exarkun2: a basic override that changes nothing
<exarkun2>
thanks
slabity has joined #nixos
<slabity>
Is there an option to enable KDE services without using the Plasma DE?
<slabity>
Sorry if that message showed up twice. Room kept thinking I was unregistered
<sphalerite>
slabity: only showed up once :) What sort of services?
<azazel>
I have to set an attribute at config.bla.bla"${dynamic}".foo ... anyone know how to do it?
<slabity>
Things like KWallet, KDE Connect, etc.
<srhb>
azazel: Exactly like thus.
<srhb>
azazel: (Well, probably without config)
<sphalerite>
wirew0rm: seems to me like using nix-store -q --graph togterht with some tool for computing graph differences on graphviz-formateed graphs would do this. I'm quite surprised that I haven't been able to find such a tool yet...
<sphalerite>
sorry about garbled writing, blame mosh :p
<sphalerite>
slabity: looks to me like it's a matter of putting the relevant package in your user or system profile (e.g. kdeFrameworks.kwallet) and ensuring kwalletd5 gets started on login
<azazel>
srhb: the config entries are services."prometheus${promVersion}".exporters.minio.minioAddress and services."prometheus${promVersion}".exporters.minio.minioAccessKey, but nix complains with: "dynamic attribute 'prometheus' at <the second line> already defined at <the firs line>"
<azazel>
but i'm in an empasse... in the repl I cannot even try a.b.c = "foo", what am I missing?
<sphalerite>
azazel: so the problem is that you have prometheus${promVersion} already defined as a non-dynamic attribute
<sphalerite>
azazel: { and } maybe? :)
<sphalerite>
going afk now
grp has joined #nixos
jperras has joined #nixos
<mog>
anyone else having dns issues on 18.09?
<mog>
suddenly firefox and chromium keep getting failed dns queries
<mog>
that work over curl or wget
<ekleog>
ocharles: hey! do you know the location of the postgresql socket on nixos? The options mention it but I can't find anything in /var/run :/
ilikeheaps has joined #nixos
<ocharles>
ekleog: I'm not sure it makes one unless you specify it in the config
pyrtsa has joined #nixos
<ocharles>
do you have enableTCPIP = true?
<ekleog>
no
<ekleog>
ocharles: ^
<ekleog>
I'm literally running a vanilla postgresql
<ekleog>
just .enable = true; that's created by services.matrix-synapse
jasongrossman has quit [Ping timeout: 244 seconds]
<ekleog>
somehow matrix-synapse manages to use postgresql's auth-by-socket, but I can't manage to make nextcloud do the same, it apparently needs an explicit path to the socket :/
<exarkun2>
Why isn't callPackage passing mkDerivation? nix-build --expr 'with import <nixpkgs> { }; callPackage ./hspec-jenkins-lae.nix { }' -> error: anonymous function at /.../hspec-jenkins-lae.nix:1:1 called without required argument 'mkDerivation'
<ekleog>
(well, at least I didn't set a password and it works, for matrix-synapse)
<ocharles>
ekleog: /tmp/.s.PGSQL.5432
<ocharles>
I believe
<ocharles>
that's what `strace psql` tried to open, anyway
<ekleog>
ocharles: oh good catch, thanks!
<ekleog>
oh I'm stupid, I had actually checked `lsof` but forgot I was doing so in mosh and thus had no scrollback ._.
<ekleog>
and it turns out it *was* in the scrollback
<azazel>
this doesn't work even in the repl: "let b = "2"; in {a."b${b}".c = 3; a."b${b}".d = 5;}"
<azazel>
sphalerite: for when you come back ^^^^
<infinisil>
azazel: Why would that work?
<infinisil>
It's not valid nix syntax
__monty__ has quit [Ping timeout: 268 seconds]
<sphalerite>
azazel: right, the problem is that you're defining a.b2 twice
<azazel>
infinisil: but why? the same without the dynamic part works: "{a.b2.c = 3; a.b2.d = 5;}"
__monty__ has joined #nixos
<exarkun2>
How do I reference fetchFromGithub given nixpkgs?
<sphalerite>
azazel: because nix can detect that you're setting attrs of the same set early enough there, and determine that there aren't any conflicts
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
azazel: what you can do is…
<sphalerite>
> let b = "2"; in { a."b${b}" = { c = 3; d = 5;}; }
<{^_^}>
{ a = <CODE>; }
<infinisil>
Ohh sorry, it was valid syntax, I was a bit confused by the {} (i thought it was a ${})
<azazel>
exarkun2: nixpkgs.lib.fetchFromGithub
<infinisil>
(for some definition of syntax)
Dedalo has joined #nixos
<sphalerite>
exarkun2: the path to nixpkgs or an already imported nixpkgs?
<sphalerite>
exarkun2: in the former case, (import <nixpkgs> {}).fetchFromGitHub (assuming you're getting it via NIX_PATH)
<sphalerite>
exarkun2: in the latter case, pkgs.fetchFromGitHub (assuming it's bound to "pkgs")
<azazel>
oh, i thought it was imported in pkgs.lib., i checked that the other day...
<__red__>
I just got my first report from a user on a Mac using the package I got added to Nix on MacOS
* __red__
does a little dance.
<__red__>
and it *WORKED*
maximiliantagher has quit [Ping timeout: 245 seconds]
jD91mZM2 has joined #nixos
ilikeheaps has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
<azazel>
sphalerite: thanks
<azazel>
sphalerite++
<{^_^}>
sphalerite's karma got increased to 36
<sphalerite>
__red__: \o/
<sphalerite>
azazel: no problem :)
<__red__>
sphalerite: I was pretty excited
<__red__>
sphalerite: I just need to get the rest of the packages ported over, but they're such a mess I put them in an overlay
hyper_ch2 has quit [Quit: Page closed]
<__red__>
but since I don't own a mac, I have no way of testing if the new packages build or not
<__red__>
and honestly, trying to get the additional packages into Nix is going to be a nightmare
<__red__>
not because of Nix per se, but because the code quality is... um, not great in many places
aminechikhaoui has joined #nixos
drakonis_ has joined #nixos
<catern>
feature of Nix that I really like lately: the ability to just do nix-build -A somepackage.src to get the source of a package and look at it
drakonis has quit [Ping timeout: 268 seconds]
<clever>
catern: ive also done: nix-shell -A somepackage, then ran unpackPhase
drakonis has joined #nixos
<azazel>
ah, nice...someone should write these down
maximiliantagher has joined #nixos
<grp>
clever: Hi. I want to make a config-checker pass to a generated config. I'd like it to work like an assert, so if it fails, there won't be a new generation unless it works
<grp>
I thought to make a derivation for the confs, but was wondering if there's a cleaner way
<grp>
(mkDerivation for the confs then use checkPhase to do whatever check I need)
<joko>
Anyone familiar with apparmor? One service (lxd) tries to run the following: apparmor_parser -I /nix/store/1hxayamd0s923f49c290ykqmawrvgahf-apparmor-profiles-2.12.0/etc/apparmor.d/ -b /nix/store/1hxayamd0s923f49c290ykqmawrvgahf-apparmor-profiles-2.12.0/etc/apparmor.d/ -rWL /var/lib/lxd/security/apparmor/cache /var/lib/lxd/security/apparmor/profiles/lxd-ipa and fails
maximiliantagher has joined #nixos
<joko>
While running the command in a nix-shell, it works
<joko>
The command fails while in systemd service as follows: Error: basedir /nix/store/1hxayamd0s923f49c290ykqmawrvgahf-apparmor-profiles-2.12.0/etc/apparmor.d is not a directory, skipping.
<joko>
Maybe systemd is denying access to the /nix/store directory?
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<samueldr>
could it be that this is running inside the LX[cD] container, where the store isn't available? (wild guess)
ddellacosta has quit []
ddellaco1 has joined #nixos
Dedalo has quit [Ping timeout: 252 seconds]
<joko>
samueldr: you mean the lxd executable is in a container itself?
johanot has quit [Quit: WeeChat 2.2]
<ddellaco1>
hello, I think I've gotten myself into a weird state with home-manager anyone know what this means? error: file 'home-manager/home-manager/home-manager.nix' was not found in the Nix search path (add it using $NIX_PATH or -I)
Itkovian has joined #nixos
locallycompact has joined #nixos
<locallycompact>
Hi, so I just installed nixos on my thinkpad x230 but the audio hardware buttons don't work
<joko>
ddellaco1: are you setting programs.home-manager.path (and enable) in your ~/.config/nixpkgs/home.nix
<joko>
?
<ddellaco1>
joko: hrm let me confirm--I installed home-manager via the /etc/nixos/configuration.nix file, which is not how the instructions say to do it, so maybe that's my problem
<{^_^}>
[nixpkgs] @griff opened pull request #49792 → nixos/rspamd: Fix enable for locals and overrides → https://git.io/fpeEL
<ddellaco1>
joko: so I have ...enable = true, and ...path pointing to HM_PATH, which is pointing to the master tarball per https://github.com/rycee/home-manager
peti has joined #nixos
peacememories has joined #nixos
<peti>
Hi guys, I would like to start several syncthing daemons on my NixOS system -- one for every user. It feels like a parameterized systemd unit would be ideal to accomplish that. Now I wonder: can I define such a thing in configuration.nix?
peacememories has quit [Client Quit]
<joko>
ddellaco1: I'm not sure how one can install home-manager system-wide, why don't you try with the default instructions?
ddellaco1 is now known as ddellacosta
jbgi has quit [Ping timeout: 264 seconds]
drakonis_ has joined #nixos
<ddellacosta>
joko: fair enough. Thanks!
locallycompact has quit [Remote host closed the connection]
drakonis has quit [Ping timeout: 250 seconds]
<tazjin>
peti: I *think* there shouldn't be more to parameterised units than just adding @ to the name and using the placeholders
locallycompact has joined #nixos
<tazjin>
but for your use-case you may want to use a user unit instead
<catern>
hey #nixos, how do I make a string which has a lot of ${} in it which I don't want to replace?
<catern>
(it's a snippet of makefile)
<tazjin>
and then template the username/uid where you need it
<locallycompact>
Can anyone help me fix my audio hardware buttons in nixos?
<peti>
tazjin: Will systemd start user units at boot time?
<peti>
tazjin: I was under the impression that those get started only after the user logs in.
<tazjin>
ah, hmm, I don't think it will unless there's a user session
<joko>
peti: syncthing already has a user unit, the tough one is as you said, how to start it on boot
<tazjin>
you could try a templated unit, but I'm not sure how enabling that with parameters will look in NixOS config 🤔
<peti>
joko: syncthing has all kinds of systemd units, but those aren't integrated well into NixOS. The unit file NixOS generates has all kinds of settings that those files don't have.
adamt has quit [Ping timeout: 250 seconds]
<joko>
peti: nope, it works reliably here as a user service, but I start it manually
<Taneb>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<joko>
I suspect, you just need to enable the user service for the respective users and enable lingering
<peti>
joko: Well, still, the NixOS code sets all kinds of extra settings (like "STNOUPGRADE") and I assume those exist for a reason, so I'd rather not use the units shipped by upstream.
<emily>
I like how ,tofu works even when the correct hash is 0000000000000000000000000000000000000000000000000000
maximiliantagher has quit [Remote host closed the connection]
<Taneb>
emily: I think it would fail if that's the correct hash for something else
<Taneb>
Which is already in the store
ilikeheaps has quit [Ping timeout: 264 seconds]
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<tnks>
god help us all... what in the world has a hash of all zeros?
<Taneb>
tnks: nothing I've found so far
<tnks>
Yeah, I rely on that.
<tnks>
when I'm developing... not for anything serious, I mean.
<Taneb>
tnks: the tofu principle relies on it being almost certainly incorrect
<Taneb>
And I think if it is the hash for something, interested parties would like to know
<emily>
the good news is that the chances of anything having that hash, as with any other specific sha256 hash, are nigh-infinitesimal.
<emily>
if you get something that hashes to all zeroes you should be, like, thankful that you didn't get hit by lightning, but disappointed that you didn't win the lottery a thousand times in a row
coconnor has joined #nixos
<Myrl-saki>
Is there a way to nix copy very parallel-y again?
<Taneb>
Is it generally safe to just kill a nix build you suspect is infinite looping (on multi-user nix on other Linux)
lsyoyom has joined #nixos
aminechikhaoui has quit [Ping timeout: 250 seconds]
<Myrl-saki>
I've done `--export | gzip` for very small files(drvs), but this doesn't work for "dumb" substituters.
<Myrl-saki>
"dumb" stores*
ddellacosta has quit [Quit: WeeChat 2.2]
ckauhaus has quit [Quit: WeeChat 2.2]
locallycompact has quit [Ping timeout: 268 seconds]
<Myrl-saki>
I'm thinking of using LD_PRELOAD, but aahhhhhhh, that's ugly.
lsyoyom has quit [Ping timeout: 246 seconds]
Tucky has quit [Quit: WeeChat 2.2]
maximiliantagher has joined #nixos
aminechikhaoui has joined #nixos
schopp0r has left #nixos ["Error from remote client"]
jbgi has joined #nixos
jonreeve has joined #nixos
maximiliantagher has quit [Ping timeout: 252 seconds]
jpyen has joined #nixos
<jpyen>
Hey Folks, anyone ever used the hadoop services on nixos?
<jpyen>
I am wondering if it is possible to access the created hdfs, hadoop user? How does one execute an mapreduce with the services.hadoop setup?
<jonreeve>
Hi everyone. I seem to be out of disk space. When I look into it, it looks like I have lots of different copies of the same library, and often the same version.
<jpyen>
@jonreeve you can clean up with the nixos garbage collecter
<jonreeve>
I'm trying `nix-store --gc --print-roots`, followed by `sudo nix-collect-garbage -d`, but I'm still using 50GB of disk space
<jonreeve>
@jpyen, is there another way I can more aggressively collect garbage?
<tilpner>
Have you tried nix-store --optimise?
<jpyen>
maybe your different versions are needed by other software?
<sphalerite>
jonreeve: what does nix-store --gc --print-roots actually print?
<jonreeve>
A list of a ton of packages
vidbina_ has quit [Ping timeout: 245 seconds]
<Myrl-saki>
...
<jonreeve>
Cool, I'll try nix-du and nix-store --optimize
<nh2>
jonreeve: well `nix-store --gc --print-roots` doesn't really delete anything, only displays; likely you have GC roots (symlinks) in there that keep stuff alive; if you delete them, `nix-collect-garbage` should be able to collect that garbage
<jonreeve>
Hm I seem to have four copies of pandoc 2.2.1 in my /nix/store. Any way I can find out which derivations own these? And how to get rid of all the unnecessary ones?
sigmundv has quit [Ping timeout: 272 seconds]
<clever>
jonreeve: run `nix-store --query --roots` on each one
<clever>
jonreeve: if any have no roots, you can `nix-store --delete /nix/store/foo` them
<clever>
to find out how a root connects to a given pandoc, use `nix why-depends`
<elvishjerricco>
jonreeve: You can do `nix why-depends /nix/var/nix/profiles/...-233 /nix/store/...-pandoc-2.2.1` to figure out why the profile has that installed.
<elvishjerricco>
FWIW I don't think the readlink is necessary. I think Nix follows symlinks, doesn't it?
<symphorien>
only if the path is absolute
<symphorien>
or starts with ./
fendor has quit [Ping timeout: 260 seconds]
Ariakenom has quit [Ping timeout: 272 seconds]
<symphorien>
otherwise it is interpreted as an attribute name
bsima has quit [Quit: Bye!]
maximiliantagher has joined #nixos
jpyen has quit [Ping timeout: 256 seconds]
IRCsum has quit [Remote host closed the connection]
Judson1 has joined #nixos
maximiliantagher has quit [Ping timeout: 246 seconds]
IRCsum has joined #nixos
<Judson1>
Seems like version 18.09 added a check that udev rules run commands installed by udev itself. I used to have a rule that locked my screen when I unplugged my Yubikey, but since xautolock isn't installed by udev...
<Judson1>
So I was wondering about the reasoning there and about workarounds
<clever>
Judson1: you could just use the right ${pkgs.xorg.something}/bin/xautolock
<clever>
when setting the udev rule from configuration.nix
<Judson1>
Oh, that makes sense.
<Judson1>
Because the rule would bring in the program.
<clever>
Judson1: you can also use services.udev.path = [ pkgs.xorg.something ]; to make it available to all rules
<gchristensen>
that is a cool rule, Judson1
<Judson1>
I like it, and am glad to improve it :)
<gchristensen>
can you share?
<Judson1>
It is loginctl though, which means bringing in systemd
<clever>
Judson1: udev is part of systemd now, you cant win :P
<catern>
hey #nixos, these build scripts have hardcoded /usr/bin/sed, /usr/bin/grep, etc. should I patch them upstream to use just "sed" and "grep" from the PATH? or should I patch them in Nixpkgs to reference store paths?
<Judson1>
Two shakes
<gchristensen>
catern: upstream is best
sg2002 has quit [Read error: Connection reset by peer]
<catern>
but then they are using the PATH instead of maximum hardcodingness :)
<catern>
what if these are not build scripts, but actually outputs of the build? (well, they are still build scripts, but shared with other packages)
<catern>
then should I patch them upstream, or patch them in Nixpkgs?
Judson1 has quit [Ping timeout: 245 seconds]
peacememories has joined #nixos
<gchristensen>
quitting before he could paste it :')
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-18.09: https://git.io/fpeXh
<simpson>
theseriousadult: Offhand, I don't know. I'd probably try removing the constraint and seeing if it works anyway; Haskell packages tend to have wrong/bad dependency versions listed.
maximiliantagher has joined #nixos
<{^_^}>
[nixpkgs] @andyscott opened pull request #49799 → bazel: support tools/bazel workspace bootstrap scripts → https://git.io/fpe1m
bobnormal has joined #nixos
Rajsun has quit [Ping timeout: 240 seconds]
<bobnormal>
hi there. i noticed the default download URL is banned from china. i wanted to ask a major open source mirror in china to add nixos support but couldn't find a reference URL for how to set up mirroring.
<catern>
,locate bin ar
<{^_^}>
Found in packages: swift, binutils-unwrapped
<ottidmes>
betaboon: not sure what you mean exactly, but probably, if you could give more details, maybe I can give you an example
lsyoyom has quit [Ping timeout: 272 seconds]
<catern>
ugh. why does binutils-unwrapped build from source when I install it?
<catern>
normal binutils doesn't build...
<ottidmes>
catern: probably different prios
lsyoyom has joined #nixos
waleee has quit [Quit: WeeChat 2.2]
<catern>
yes, but surely normal binutils depends on binutils-unwrapped!
judson has quit [Killed (verne.freenode.net (Nickname regained by services))]
judson has joined #nixos
<judson>
Good news! I was wrong.
<rawtaz>
\o/
<simpson>
catern:
Synthetica has quit [Quit: Connection closed for inactivity]
<simpson>
catern: Yes, but normally you'd get your `ar` from the same place as your `cc`.
<judson>
I had a lingering change to filesystems that I hadn't built, and that was the problem.
<judson>
Saving up a glare for Mitchell if I see him again.
lsyoyom has quit [Ping timeout: 260 seconds]
patrl has quit [Ping timeout: 264 seconds]
<judson>
So, thinking it would be reasonable to have the fstab module make the substitution, no?
<{^_^}>
[nixpkgs] @luigy opened pull request #49800 → Update all-cabal-hashes to match master → https://git.io/fpeMn
<clever>
judson: yeah
lsyoyom has joined #nixos
<catern>
How do I format a string with '' that contains literal instances of ${}?
<catern>
oh, I do ''
<catern>
nvm
<clever>
,escape''
<{^_^}>
'' two single quotes: ''' bash curly bois: ''${} newline: ''\n tab: ''\t any character x: ''\x ''
KennosiS has joined #nixos
<judson>
Philosophically, would it be better to silently to the substitution, or define a type of "fstab mount point name" that cannot have tabs or spaces?
<ottidmes>
does anyone have experience with node packages? All I need is, run some patch over the source, `npm install`, and `npm run dist`. Outside of Nix this is as easy as those 3 steps, but I am not sure how to package it in Nix
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/fpeMp
<clever>
judson: i would prefer it to just silently do the right thing when you have spaces in the field
<azazel>
Mic92: thanks for stopping the review request :-)
<magnetop`>
I'm trying to package faust2jackrust, a script that takes faust code, turns it into rust, and builds it with cargo. When I use it, it breaks with: "ld: cannot find -ljack". I have libjack2 in the propagatedBuildInputs
sir_guy_carleton has joined #nixos
sir_guy_carleton has quit [Client Quit]
magnetop` is now known as magnetophon`
sir_guy_carleton has joined #nixos
<hyper_ch>
I have a weird issue on my server... after about 24h it loses capability for dns resolving
foldingcookie has quit [Ping timeout: 264 seconds]
<magnetophon`>
IOW: how do I make cargo find libs during the linking phase?
<azazel>
hyper_ch: it loses capability? any insight?
<hyper_ch>
weird thing is, in the nixos config I did specify dns servers but resolv.conf doesn't have any
<bobnormal>
hyper_ch: could be DHCP lease not renewed by local DHCP server (eg. cheap wireless router). does other network traffic work? try tcpdump while resolving something or check logs.
<colemickens>
When I have a module that provides a wrapped version of a command, how do I ensure that it is used, evne if the user places that package into the environment.systemPackages ?
<azazel>
hyper_ch: as bobnormal said, when ip gets renewed the dns may not be on the dhcp response
<hyper_ch>
azazel: I did set specific nameservers in the nixos config
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #49800 → all-cabal-hashes: Update all-cabal-hashes in 18.09 to match Nixpkgs master → https://git.io/fpeMn
<{^_^}>
[nixpkgs] @Ericson2314 pushed 6 commits to release-18.09: https://git.io/fpeyM
<hyper_ch>
azazel: but resolv.conf doesn't show any
<hyper_ch>
also lease time is more than 7000 days left
reinzelmann has quit [Quit: Leaving]
mayhewluke has quit [Ping timeout: 252 seconds]
<azazel>
hyper_ch: if you don't use dhcp, I usually add useDHCP = false; to both the networking set and each of the interfaces. I found that it interferes when I have servers with fixed addresses
maximiliantagher has quit [Remote host closed the connection]
fede_run has joined #nixos
<fede_run>
Hi!
magnetophon` has quit [Read error: Connection reset by peer]
maximiliantagher has joined #nixos
pie_ has joined #nixos
<pie_>
enabling virtualbox and running a rebuild switch killed my kde network applet thing, what gives
<pie_>
alternatively im not sure if and kernel modules got unloaded but my network interfaces were still up
<pie_>
it was fixed by a reboot but id rather not have rebooted if i didnt need to
<fede_run>
Has anyone installed MATLAB on nixos? I'm having some problems
maximiliantagher has quit [Remote host closed the connection]
<pie_>
fede_run: i heard theres something called fhsuserenv ...that might be easier...
ddellacosta has joined #nixos
<fede_run>
Thanks I'm going to look on the wiki maybe I can find something usefull
<catern>
lolwut
<catern>
ran fetchgit and it failed with "fatal: not a tree object"
<catern>
an error in git??
sanscoeur has joined #nixos
theseriousadult has quit [Remote host closed the connection]
<hyper_ch>
something is really weird, I did tell it to open ports and I do run syncthing on that port but when I nmap it from another system, I get port closed
jbgi has joined #nixos
crmlt_ has joined #nixos
civodul has joined #nixos
rfold has joined #nixos
<ottidmes>
hyper_ch: I too have syncthing on one of my machines, I rarely use it, and just tested it, and it no longer works, could be a coincidence though, I will look into it
<hyper_ch>
ottidmes: I found the issue... for some reason it started reading wrong config
<hyper_ch>
I set datadir to /home/user/Syncthing where also config.xml is located
<hyper_ch>
but now it seems to use /home/user/.config/syncthing
hyperfekt has joined #nixos
<catern>
would substituting/binary caching work when using builtins.fetchGit { url="whatever"; rev="commit hash"; }?
<catern>
or is that actually impure enough under the covers to break substitution?
<{^_^}>
[nixpkgs] @azazel75 opened pull request #49802 → Add Prometheus 2 service in parallel with 1.x version → https://git.io/fpe5p
<hyper_ch>
ottidmes: also port 21027 is now being required
<ottidmes>
hyper_ch: that is probably the issue then, since I do not see 21027 listed as open
<hyper_ch>
ottidmes: ok, that doesn't solve all my issues
<hyper_ch>
still wrong config being used
<exarkun2>
I'm trying to build a stack-based project with Nix and failing to get one of its dependencies resolved. I have a github entry in `extra-deps` of the project's `stack.yaml`. And I have an overlay that adds that dependency to `testHaskellDepends` (test dependency). The build fails with "Setup: Encountered missing dependencies:" and then mentions that dependency. What else do I need so the package can be found?
<hyper_ch>
ottidmes: well, port 21027 helped me on one installation... on the other still wrong config is being used
<fede_run>
I can't find how to use that, I set an fhs env and then i run ./install from matlab install folder but it doesn't work.
drakonis_ has joined #nixos
<{^_^}>
[nixpkgs] @Lassulus opened pull request #49803 → dwarf-fortress: fix most of customization options → https://git.io/fpedz
<ottidmes>
hyper_ch: I had to accept something from the interface to make it find changes faster, and now it works again for me
<hyper_ch>
ottidmes: I can't access the wui on the server
<ottidmes>
hyper_ch: I just do: ssh -L 8384:localhost:8384 server
<hyper_ch>
ottidmes: because I'm already using 8384 on my notebook as well
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
hyper_ch: ssh -L (port forward from remote to local) 8384(remote port):localhost(where to forward to):8384(local port) server(normal ssh name of remote machine)
<hyper_ch>
so how do I set local a different port?
<ottidmes>
hyper_ch: ssh -L 8385:localhost:8384 server, would then work
<ottidmes>
hyper_ch: I am quite sure it is: ssh -L 8384:localhost:8385 server, but both ways worked just now, maybe I still had a connection open
jasongrossman has joined #nixos
crmlt_ has quit [Remote host closed the connection]
<hyper_ch>
ottidmes: and how to open it in the browser now?
<hyper_ch>
nah, it was 8385:localhost:8384
<ottidmes>
hyper_ch: I was wrong, its the other way around, so: ssh -L 8385:localhost:8384 server
crmlt has joined #nixos
<ottidmes>
hyper_ch: I am going to make a alias for it, I always get confused, and rightfully so
halfbit has quit [Quit: WeeChat 2.3]
<hyper_ch>
ottidmes: thx
<fede_run>
Solved, I was missing some libraries in my fhs env
<ottidmes>
hyper_ch: so it was: ssh -L [8385](local port):[localhost:8384](what is being forwarded on the remote machine) [server](the rest of your SSH command)
pie_ has quit [Ping timeout: 256 seconds]
<hyper_ch>
yeah :)
hyperfekt has joined #nixos
<hyperfekt>
oh, i see the logging service was migrated. good, i wasx worried about that.
<hyper_ch>
hyperfekt: what do you mean'
<hyper_ch>
?
<hyperfekt>
The IRC logs were previously hosted on botbot.me, iirc, but that's shutting down in a few days
<hyper_ch>
the config.xml has <address>0.0.0.0:8384</address>
<ottidmes>
hyper_ch: hah, you definitely want that on 127.0.0.1
<ottidmes>
hyper_ch: otherwise other people can potentially modify your filesystems
<hyper_ch>
ottidmes: I definitively don't want 127.0.0.1
<ottidmes>
hyper_ch: this is for the admin UI only
sigmundv has joined #nixos
<hyper_ch>
ottidmes: I still definitively don't want 127.0.0.1 there
<ottidmes>
hyper_ch: it is configurable
<hyper_ch>
ottidmes: what do you mean?
<hyper_ch>
I have it to 0.0.0.0:8384
<hyper_ch>
yet it only listens to localhost
<ottidmes>
hyper_ch: port 22000 at least works properly both tested from my desktop and server with your nmap command
hamishmack has joined #nixos
<hyper_ch>
good idea going to unprivileged port
<hyper_ch>
wtf.... it just keeps ignoring the gui setting in the config
worldofpeace has joined #nixos
crmlt has quit [Ping timeout: 245 seconds]
<ottidmes>
hyper_ch: yeah I tried it myself, I get a big danger message and local nmap gives me open, but desktop nmap gives: 8384/tcp filtered marathontp and I cannot connect
worldofpeace_ has joined #nixos
sir_guy_carleton has quit [Quit: WeeChat 2.2]
<ottidmes>
hyper_ch: do you have a nginx or similar running? that is an easy fix
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
hyper_ch: I am just going to write some nginx config to allow me to access it via sync.server, only allowing local access of course
<hyper_ch>
:)
<hyper_ch>
you know you can set username and password for authentication
c0bw3b_ has joined #nixos
<colemickens>
wouldn't that bug defeat the entire point of running syncthing?
<ottidmes>
colemickens: why? its just the way to access the admin panel, which you rarely have to do
<hyper_ch>
I don't like it because (a) config files were moved out from the data dir... everyone who used it is suddenly f***** because it's not working anymore and (b) I configure hte config file and nothing happens - wtf?
<ottidmes>
colemickens: or did you mean the config dir bug?
<ottidmes>
hyper_ch: yeah, those things should not happen, its what makes updating your system a pain sometimes. This was especially bad with Arch Linux years ago, there was always something that broke, making me have to invest time to search the internet for potential fixes, but that was just my experience
<clever>
maximiliantagher: 5 pre-allocates a dummy space for a gitrev at compile time, 3/4 then modify the binary (after compiling) to inject a rev, which allows nix to cache the compile haskell code (for when no haskell changes occur), but still update the gitrev within it, so --version remains right
<maximiliantagher>
Got it
<maximiliantagher>
Tough it seems kind of overly complicated tbh
<clever>
maximiliantagher: there are simpler ways to do it, but they result in nix rebuilding the binary every time any commit happens, so you loose caching
patrl has joined #nixos
<maximiliantagher>
do you know of a simpler way off-hand? 99% of our commits are modifying Haskell code anyway so seems like we wouldn't get much caching benefit
<clever>
that older one will read $GITREV, and if it fails, it runs `git rev-parse HEAD`
zack_moe has joined #nixos
<clever>
(nix also deletes .git, so the fallback wont work)
<maximiliantagher>
Yeah that's what I've been running into
<clever>
maximiliantagher: in cardano's case, that template haskell was too deep into the dep stack, so it was recompiling 30+ minutes worth of code, when only the top layer has changed
<maximiliantagher>
Ah yeah, they have a ton of packages right?
<ddellacosta>
so I'm running into a problem using home-manager that I just don't know how to fix. I've tried installing it via the main nix configuration.nix file, uninstalling it there and reinstalling it using the README instructions, passing in arguments per the error, but nothing works. I'm stumped: error: file 'home-manager/home-manager/home-manager.nix' was not found in the Nix search path (add it using $NIX_PATH or
<ddellacosta>
-I)
<ddellacosta>
that happens whenever I run home-manager build or switch ^
<ddellacosta>
I'm a newbie, and probably shouldn't have tried using it in the first place (I found a solution to my x11 mouse size woes in a home-manager issue, and I was desparate), but I don't even know how to uninstall it at this point
<ddellacosta>
any suggestions?
<fede_run>
I've got a new problem, when I try to do nix-env -I idea-ultimate, curl gives me 403 forbidden error
<clever>
fede_run: you want -i, not -I
<bsima>
ddellacosta: home-manager shouldn't be installed via configuration.nix
<ddellacosta>
bsima: yeah, so if I uninstall it there I'm still stuck
<clever>
as well
jabranham has joined #nixos
<bsima>
ddellacosta: did you follow the install instructions in the home-manager readme?
<fede_run>
The problem is that it tries to download curl: (22) The requested URL returned error: 403 Forbidden
<fede_run>
error: cannot download ideaIU-2018.2.2-no-jdk.tar.gz from any mirror
<fede_run>
ideaIU-2018.2.2-no-jdk.tar.gz but now the release is 2018.2.5
<ddellacosta>
bsima: I did after the fact, but those fail for me too
<clever>
fede_run: somebody deleted it from the server, so the nix code will need to be fixed
<fede_run>
May I fix it?
<clever>
fede_run: youll need to update the url in the nixpkgs source, and the sha256, then file a PR to nixpkgs
<bsima>
ddellacosta: can you do a 'nixos-rebuild --rollback'?
<ddellacosta>
bsima: I guess so? I've installed and uninstall so many times though that I'm not sure what that'll do (not to mention all the other stuff I've installed since then)
<ddellacosta>
I'd love to know how I can simply uninstall it
<bsima>
ddellacosta: you could try nix-collect-garbage
jperras has quit [Quit: WeeChat 2.2]
phreedom has quit [Quit: No Ping reply in 180 seconds.]
vk3wtf has joined #nixos
phreedom has joined #nixos
<ddellacosta>
bsima: so I did that, and deleted the .config/nixpkgs/home.nix file as well, but I'm not sure what to do now so I can delete it entirely and try to start over. It appears to have scattered stuff all over my home directory...
<ddellacosta>
and home-manager is still in the path, although it appears it's coming from ~/.bin
<clever>
ddellacosta: ls -lh ~/.bin
<ddellacosta>
clever: sorry I take that back, it's from .nix-profile/bin
<clever>
ddellacosta: then it will be listed somewhere in `nix-env -q`
<ddellacosta>
anyways, I feel stupid because I'm just cargo culting all of this. I need to go and figure out what home-manager actually did to my system I guess
<ddellacosta>
clever: yeah that gives me 'home-manager-path' but what that is actually telling me is a mystery
<clever>
ddellacosta: `nix-env -e home-manager-path` to uninstall it from .nix-profile
<ddellacosta>
clever: okay wow, thanks, seems to maybe have done it? I'll now go try and read some docs so I'm not asking such clueless questions, thanks to you and bsima for the help
zolk3ri has quit [Quit: Lost terminal]
carlosdagos has quit [Quit: Connection closed for inactivity]
<jabranham>
Does cache.nixos.org not serve a binary for the kernel? I'm trying to do my first "nixos-rebuild switch --upgrade" and nix keeps insisting it needs to compile the kernel.
<clever>
jabranham: it should, what is the storepath of the kernel its trying to build?
jasongrossman has quit [Ping timeout: 272 seconds]
__monty__ has quit [Quit: leaving]
<jabranham>
clever: ugh, now it's trying to build strace. I dunno why.
maximiliantagher has quit [Remote host closed the connection]
thomasjm has joined #nixos
orivej has joined #nixos
dbmikus_ has quit [Ping timeout: 252 seconds]
dbmikus_ has joined #nixos
Ariakenom has quit [Quit: Leaving]
maximiliantagher has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 pushed 118 commits to staging: https://git.io/fpvUY
jasongrossman has quit [Ping timeout: 244 seconds]
<thomasjm>
question: when you build Nix from source and "make install", how do you actually initialize the Nix store/profiles/channels and start using it? I noticed that there exists [install prefix]/etc/profile.d/nix.sh so I ran that, but it produced a broken ~/.nix-profile symlink and a broken symlink in ~/.nix-defexpr. It seems there's still some steps I'm missing to initialize things, but I can't find documentation on it anywhere. I can't run
<thomasjm>
"nix-channel --update" either because I get the error "path '/nix/store/...-nixpkgs-unstable' is not in the nix store." Thanks in advance for any help
drakonis_ has quit [Ping timeout: 272 seconds]
<clever>
thomasjm: ~/.nix-profile will remain broken until you first run `nix-env -i something`
<clever>
thomasjm: and ~/.nix-defexpr should be fixed the first time you run nix-channel
<{^_^}>
[nixpkgs] @costrouc opened pull request #49807 → [WIP] Integration of kernels and extensions with jupyter → https://git.io/fpvUo
<thomasjm>
thanks clever. that's what i thought, but when I try something like "./bin/nix-env -i awscli" I get "error: selector 'awscli' matches no derivations"
<clever>
thomasjm: it looks in ~/.nix-defexpr to find nixpkgs by default
<clever>
thomasjm: you can also clone nixpkgs, and then nix-env -f ~/nixpkgs -i awscli
jasongrossman has joined #nixos
<thomasjm>
aha, that seems to work. thanks!
Thra11 has quit [Ping timeout: 250 seconds]
rfold has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @pbogdan opened pull request #49808 → steamPackages.steam-runtime-wrapped: don't include all of gcc → https://git.io/fpvTG
c0bw3b_ has left #nixos [#nixos]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
avn has quit [Ping timeout: 264 seconds]
patrl has quit [Ping timeout: 252 seconds]
drewr has joined #nixos
freeman42x has quit [Ping timeout: 252 seconds]
jasongrossman has quit [Ping timeout: 244 seconds]
MinceR has quit [Ping timeout: 246 seconds]
MinceR has joined #nixos
ddellacosta has quit [Quit: WeeChat 2.2]
fede_run has quit [Quit: leaving]
mayhewluke has quit [Ping timeout: 252 seconds]
mayhewluke has joined #nixos
avn has joined #nixos
erasmas has quit [Quit: leaving]
orivej has quit [Remote host closed the connection]