<radvendii>
right, and am i correct that there's no way to recover the attribute that was being passed into python before?
<radvendii>
it's not like home-assistant could add onto it
<AmandaC>
I'm not clear on why those need to be done with an overlay in the h-a expression though, vs just done as expressions in the `let ... in`block like some other expressions use to pin versions
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JOX5N
cfinch has quit [Read error: Connection reset by peer]
<hexa->
to prevent multiple version of the same package inside the same python environment
<hexa->
that's just not possible
<hexa->
and component-packages.nix already reference the attribute name within python3Packages
<AmandaC>
but the python environment is built per-package isn't it? instead of using the nixpkgs expression for the dep, use the expression from you `let ... in`
<AmandaC>
Or is python deps different from, eg, `buildInputs`
<hexa->
there is a home-assistant python environment, that contains all the dependencies for the enabled components
<hexa->
they need to be propagated
<hexa->
I'm sure there is a better solution than how we handle it right now, I am by now means an expert in overlays and overrides
<aleph->
Well this is fun then, sigh.
<AmandaC>
trying to figure out where I saw an instance of this pattern I'm thiking of recently
<radvendii>
aleph-: in this case, you may indeed be able to do what i tried originally with recursiveUpdate
<aleph->
Oh?
<radvendii>
i'm not sure what packageOverrides does besides that
<aleph->
Was that the second paste?
<radvendii>
idr
<radvendii>
i'll recreate it
<aleph->
Yep it is
<radvendii>
oh okay you got it then?
<aleph->
Yep, gonna try that
cole-h has quit [Ping timeout: 252 seconds]
Jd007 has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<radvendii>
okay, so the effect of packageOverrides is burried somewhere in makeScopeWithSplicing, and I don't think I'm prepared to go that deep tonight
<hexa->
AmandaC: is it a somepackage' = somepackage.overridePythonAttrs thing?
<aleph->
Ah right that is a thing. Most of my stuff isn't up streamed (yet...)
<hexa->
for me it is far easier to submit things upstream than to maintain them myself
<hexa->
overrides are tough
<AmandaC>
hexa-: it wasn't python specifically, but it was that yeah, then instead of using somepackage in the normal nixpkgs, you'd use your pinned version
YoDawg has joined #nixos
<aleph->
Yeah I need to start doing it more. The review process takes a while.
<aleph->
By the time I get to it usually my stuff is staled or someone grabs it and picks it up.
<AmandaC>
like, passing it to buildInputs or similar. Granted, I've only barely touched the pythin nix stuff
<YoDawg>
Hey, good evening, did anyone have any luck setting a custom GDM background image/theme?
<hexa->
that is what the python packageset maintainer has to say about the topic :)
<hexa->
I've been there.
<aleph->
Heh.
<AmandaC>
I see. Strange, not sure how-- oh. If another package depends on pyjwt it'll be duplicated
<hexa->
yep, and you can't have two
<AmandaC>
/now/ I see
<hexa->
because `import pyjwt` is … well …
<aleph->
Okay stupid question is there a way to pin nixpkgs to a commit and patch in derivs? Never had to so not sure off hand.
<hexa->
you can have a fork of nixpkgs and track your patches on top of it?
<aleph->
Yeah that's what'll I have to do I guess.
johnw has joined #nixos
<radvendii>
oh, have we finally actually banned the "name" attribute?
<radvendii>
oh never mind, i just wasn't paying careful attention to the error message
Jd007 has quit [Quit: Jd007]
xcmw has joined #nixos
<AmandaC>
Don't suppose anyone can tell me how I'd run the tests for samba. I've not had any issues with a functionally-identical config applied on my homelab, but better to check those boxes for my PR: https://github.com/NixOS/nixpkgs/pull/114393
<{^_^}>
#114393 (by AmandaCameron, 8 weeks ago, open): nixos/samba: Add a wants=network-online.target to the target.
<radvendii>
is it possible to *unpin* a flake input?
<AmandaC>
no. flakes are meant to be reproducible, so all inputs are pinned
<AmandaC>
( using the flake.lock )
<radvendii>
bleagh.
<radvendii>
just for testing...
ericsagnes has joined #nixos
stree has joined #nixos
<radvendii>
IMO that restriction should be eased for "path:" inputs, but if people tell me i'm wrong, i'll accept that
<AmandaC>
I mean, you can `nix flake update` and it'll pin to the new master
<lukegb>
AmandaC: `nix-build ./nixos/tests/samba.nix` is probably the easiest way
<radvendii>
AmandaC: yeah, but i have to remember to do it every time
<radvendii>
it's also not possible to have "co-dependent" flakes in the same repo
<radvendii>
well, it's possible but tedious, because you have to remember to update the inputs whenever you change one
Jd007 has joined #nixos
smitop2[m] has joined #nixos
YoDawg has quit [Ping timeout: 240 seconds]
<radvendii>
this is a weird error: "error: unsupported Git input attribute 'sha256'"
<radvendii>
it's complaining about sha256 in builtins.fetchGit
<AmandaC>
lukegb: well, it didn
<AmandaC>
lukegb: well, it didn't report any errors?
<lukegb>
yeah, if it outputs a store path at the end the test succeeded
<AmandaC>
ah
<AmandaC>
Then I can tick that off, then
<radvendii>
oh, i'm silly. fetchGit indeed does not accept sha256 as an argument. it uses rev instead
<srid>
Anybody here use `org-mode` to take daily notes in the form of `2021-04.22.org`, etc. and would like test something out (static site generated out of your .org files) quickly?
<hexa->
fetchgit does accept a sha256 argument
<hexa->
in addition to rev
<radvendii>
huh. it doesn't like it when i put it in
<radvendii>
srid: no, but that sounds really cool. i have a similar setup with markdown, and use hakyll to generate a website
gustavderdrache has quit [Quit: Leaving.]
<srid>
radvendii: I was going to give: `nix run github:srid/memoir -- -d ~/org`
<srid>
its gonna beat the likes of workflowy for the future-proof inclined
<srid>
its using https://github.com/srid/ema which is my pet project to provide beyond awesome static site dev env with super hot reloadsssssss
<cwnovusordoseclo>
I'm having some issues getting ACME working for one of my subdomains. It errors with `mkdir: cannot create directory ‘/var/www’: Permission denied` when trying to use `security.acme.certs."icecast.domain.tld"`. My webroot is `/var/www/challenges`. any ideas?
<radvendii>
does /var/www/ already exist?
<cwnovusordoseclo>
Probably, I've got a couple other subdomains
<mounty>
Hello; I'm a naive NixOS user and have been using nix-env -iA to get packages and commands that I need in a hurry, rather than extending /etc/nixos/configuration.nix -- is there any way to find what I've added with nix-env so that I can add it to configuraiton.nix and update properly?
<evils>
mounty: nix-env -q
<mounty>
Thanks so much evil: that's just what I need.
<radvendii>
is there a way to get ssh to cache failure? so it doesn't try the same host again within a certain amount of time?
<radvendii>
i have a remote builder that is sometimes up and sometimes down, but when it's down it's not like it's going to come up again when nix gets to the next package, so it checking every time wastes time
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cwnovusordoseclo>
radvendii: can you temporarily null route it? those should fail nearly immediately
<cwnovusordoseclo>
I realize that doesn't really answer your question
<radvendii>
cwnovusordoseclo: no, but i might be able to build a solution out of that. how do i null route it?
<cwnovusordoseclo>
I might be using the wrong terminology. Add a host file entry to a nonroutable IP
<hexa->
ip route add blackhole <cidr>
<hexa->
or just clear --builders ""
<AmandaC>
someone earlier talked about iptables -j REJECT-ing a binary cache
<AmandaC>
that should also work
<radvendii>
honestly, i think we need better UI for remote builders in general. I should be able to specify --builders "foo" if foo is defined in my configuration.nix, without having to specify all of the extra arguments every time
<radvendii>
oh those are all interesting solutions
<radvendii>
thank y'all
nrh^ has quit []
<cwnovusordoseclo>
@hexa- I didn't know about route blackhole, thanks for that
<deltab>
you could put a wrapper script around ssh and have that record and check for recent failure
<radvendii>
i should maybe just remove my personal host. nixbuild.net is easy, fast, and cheap enough that it might just not be worth the hassle
<chipb>
hexa-: radvendii: I think you might prefer 'prohibit' instead of 'blackhole' here, else it may hang trying to connect on account of the silent discard...
<radvendii>
chipb: ah, that's good to know. would that be "ip route add prohibit"?
<chipb>
or I guess 'unreachable' is a bit more appropriate for the circumstance.
superherointj has quit [Quit: Leaving]
<radvendii>
hexa-: i can't just clear builders, because i have another builder that *is* reliable (nixbuild.net) that i'd like to fallback on
<chipb>
yeah, just as their recipe before, subsituting blackhole.
<radvendii>
chipb: thanks
<chipb>
(though honestly, I'm not super familiar with iproute2's route subcommand myself.)
<twoolie>
hey all, i'm trying to get a list of all the services that are enabled in a nixos config
<twoolie>
i currently have a bit of code like this:
<chipb>
oh, nevermind. local senders shouldn't hang: "The local senders get an EINVAL error."
<twoolie>
in pkgs.lib.filterAttrs isEnabled config.services;
mounty has quit [Quit: Konversation terminated!]
<twoolie>
but this blows up on services like beegfs that are deprecated/removed
<twoolie>
anyone know if something like this is already builtin to nixpkgs somewhare?
<radvendii>
twoolie: i noticed you've asked that a couple of times here. i'm not bothered, just wondering if it might serve you better to post this on discourse.nixos.org
<radvendii>
unless i'm misremembering something, which is also possible
mounty has joined #nixos
YoDawg has quit [Ping timeout: 240 seconds]
<radvendii>
oh btw you can use attrs ? enable instead of pkgs.lib.hasAttr "enable" attrs
<{^_^}>
#30756 (by FlorentBecker, 3 years ago, closed): nixos: pptp in NetworkManager requires kernel module nf_conntrack_pptp, which is not present by default
<jesystani>
so i just rewrote my config, and ive booted and opened up a terminal, and i dont seem to have anything avaliable to me in path? the programs i installed are there, but i dont have ls, or mkdir, or date or bash or anything like that
<jesystani>
under x it only have my zsh plugins folder, and my /etc/profiles/per-user one set, but in tty it also has /run/wrappers/bin, /home/me/.nix-profile/bin, /nix/var/nix/profiles/default/bin and /run/current-system/sw/bin
<jesystani>
ie those are the ones missing under X, if that helps?
<{^_^}>
[nixpkgs] @SuperSandro2000 merged pull request #120201 → [staging-next] python3Packages.httplib2: allow local networking on darwin → https://github.com/NixOS/nixpkgs/pull/120201
<radvendii>
it somehow doesn't surprise me that those are the paths that are missing, as they define most of the system, but i'm not sure why they wouldn't be set under X
waleee-cl has quit [Quit: Connection closed for inactivity]
sangoma has joined #nixos
melchizedek6809[ has joined #nixos
stree has quit [Ping timeout: 265 seconds]
<SumnerEvans[m]>
I'm revamping the matrix-synapse module. What is the best way to test/inspect the config files that I am creating?
<__anb>
I see nix support NIX_{STORE,STATE,CONF}_DIR environment variables, does it mean I can point it to anywhere else without recompile the program? Asking because the docs I read said you have to reconfigure the program when you want to change it.
<SumnerEvans[m]>
srhb: thanks! I did see that, but it doesn't seem to test much (it looks like it just tests that the server runs, not that it's config is correct). In Home Manager, I would create an "expected" file, but I don't see that pattern in nixpkgs.
<SumnerEvans[m]>
srhb: true, I could. But I don't see any existing patterns for that.
cr4y1_ has joined #nixos
beertoagunfight has quit [Ping timeout: 246 seconds]
<srhb>
SumnerEvans[m]: No, I don't think it's a pattern that's used often in nixpkgs (which is perhaps a shame) -- but worst case you just need to expose the generated config through some option and eventually `diff ${config.matrix.whatever.renderedConfig} ${expected}`
<srhb>
SumnerEvans[m]: Though I suppose it's a bit expensive to do it in a NixOS test, but at least you have the module at hand.
<__anb>
Can anyone point me to a document about how the hash is generated? I'm moving things around, and want to avoid rebuilding.
<srhb>
__anb: The hash of store paths?
cfinch has quit [Read error: Connection reset by peer]
sss has joined #nixos
<__anb>
yes.
<srhb>
__anb: Every input (dependency) is recursively part of the hash of every package. I think you'll need to provide a bit more detail to answer your question :) What are you trying to move?
cole-h has joined #nixos
LilleCarl has joined #nixos
<__anb>
srhb: sure. I'm new to nix, and on macos-aarch64, I guess it's a hard start. There's a nixpkgs branch I used to bootstrap nix, but I guess later I'll move to an official one once it gets merged. I don't know if the path of the nixpkgs will be used to calculate the hash or not. Also the nix binary changed during the bootstrap, is it another factor?
phalrax has joined #nixos
<srhb>
__anb: I'm not very familiar with nix on macos, however; the path of _nixpkgs_ -- that is, the source tree that constitutes the main package set -- should not matter, since it, as well as most other sources, is a fixed-output derivation. Meaning, in essence, that the above statement about hashes was "lies to children" -- the hash of sources are usually determined from their _output_ (essentially a
<srhb>
__anb: There's a lot of detail and nuance here, but essentially, every change to anything used in the build (recursively) will bubble up through the dependencies and change all hashes.
<cole-h>
dotlambda: Nice, we did similar things at similar times :D
<__anb>
srhb: I switched from one nix binary which was used to bootstrap, to another binary, which was installed by the previous one. When I try to use the new binary, it wants to rebuild the whole system again...
redmp has quit [Ping timeout: 252 seconds]
<srhb>
__anb: We'll have to see the difference in the derivations to know why that's the case.
<twoolie>
radvendii: thanks for the script. unfortunatly i'm trying to get the result as a nix attrset to do some extra processing on it. I'll take your sugestion and post on discorse though.
cfinch has quit [Remote host closed the connection]
beertoagunfight has quit [Ping timeout: 260 seconds]
<__anb>
srhb: I think I know what's wrong, the new bootstrapped binary is built with default store path, which is not the one I'm using. Even with NIX_STORE_DIR set, nix-env is able to query package, but not install it. I guess the variable is not honored everywhere.
<JordiPlayGames[m>
thibm: What package count will have in NixOS 21.05???
<tejing>
there are a lot of options of how to count, so there isn't really a single answer
* thibm
sighs
<thibm>
Can someone kick this spammer?
<JordiPlayGames[m>
tejing: So, 90.000 packages will be available in NixOS 21.05???
<Kinnison>
presumably that 90,000 is multi-counting some things such as the vast number of different boost versions, or how vscode is repeated a few times.
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JOMJd
<typetetris>
Hi there, someone who could help a bit with `buildRebar3` and an erlang build? I have a project using rebar3_proper rebar plugin and I thought I provided rebar3_proper, but rebar3 still insists on touching hex pm to check its version.
m4ts has quit [Read error: Connection reset by peer]
jumper149 has joined #nixos
<jumper149>
I want to use lm_sensors? Is there anything I need to look out for on NixOS? Or can I just run `sudo sensors-detect`?
cfinch has quit [Remote host closed the connection]
cfinch has joined #nixos
m4ts has joined #nixos
exondation has joined #nixos
malook has quit [Quit: malook]
<sphalerite>
jumper149: I don't recall exactly what files it tries to write and whether they all make sense on nixos, but I have run it before and it's helped me identify the drivers to add to boot.kernelModules
phalrax has joined #nixos
avaq has quit [Remote host closed the connection]
avaq has joined #nixos
vidbina_ has quit [Ping timeout: 240 seconds]
Qwerky has joined #nixos
cfinch has quit [Read error: Connection reset by peer]
<lukegb>
I have a [client.<instance-name>] entry in /var/lib/ceph/radosgw/<cluster name>-<instance-name>/keyring which has mon/osd allow rwx caps
<devhell>
lukegb: okay, I'll check that that's also the case with mine, thank you!
<rmcgibbo[m]>
Inside nixpkgs, nixos modules are organized separately from packages usually. Does anyone have a good example I could follow of an out-of-tree nixos module+package combination (I don't use flakes)? I was thinking of making a normal package and then defining the nixos module under meta.passthru, but I'm not sure if that would be convenient to `import` from configuration.nix.
pen has quit [Ping timeout: 240 seconds]
lordcirth has joined #nixos
<sterni>
I think flakes has a custom way of distributing modules
<sterni>
but usually you can do ${sourcetarball}/module.nix just fine in imports
<sterni>
although with flakes you don't have the src right?
<sterni>
nevermind me :)
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<yorick>
[20.09] so I'm trying to configure network.interfaces.x.ipv4.routes, but it seems that they don't get restored if the link goes down and back up again
lordcirth has quit [Ping timeout: 250 seconds]
endformationage has joined #nixos
<radvendii>
why don't /usr/bin/env shebangs work during a derivation build? is that expected? i.e. should i have to patchShebangs my bulid scripts in postPatch / preBuild?
<sterni>
yep
<puck>
radvendii: /usr/bin/env doesn't exist in the derivation sandbox for some reason
<lukegb>
devhell: -> #nixos-dev
<lukegb>
it sounds like your keyring file is in the wrong place
<radvendii>
puck: huh. i would guess it's for some sandboxing reason, but i can't actually think what that reason would be
<puck>
radvendii: something about impurity (there's /bin/sh, but it's a busybox static binary that has been provided at Nix build time)
<KarlJoad>
I am trying to build a RISC-V toolchain inside a nix-shell. But, I am always getting a 2 exit code. I think I'm missing dependencies, but have no idea what I should be looking for. I have logs from the script run in both the nix-shell and an Ubuntu VM to diagnose.
lordcirth__ has quit [Ping timeout: 276 seconds]
codygman` has quit [Remote host closed the connection]
<shofius>
aleph-: do you run your whole config in unstable?
ahmedelgabri has quit [Ping timeout: 245 seconds]
<cgbrewster[m]>
When configuring substituters in nix.conf, if one substituter has a network error (maybe the cache is down), will nix gracefully go to the next one in the list?
<rmcgibbo[m]>
cgbrewster: my experience last week was that it did not.
<cgbrewster[m]>
rmcgibbo: cool, thanks! Do you happen to know if this is intentional, would a patch be accepted to allow the fallback behavior?
dhess has joined #nixos
<rmcgibbo[m]>
I do not know. It was not the behavior that I wanted, so personally I would really like such a patch :)
<radvendii>
cgbrewster[m]: i configure substituters in configuration.nix, and my experience is that it does this semi-gracefully
<radvendii>
oh
<radvendii>
there was a trick. i forgot i changed something
<radvendii>
i set ConnectTimeout 5 on the host that is sometimes down
<radvendii>
otherwise it will hang for a while before moving on
<radvendii>
i will warn you that even with ConnectTimeout 5, it will hang for 5 seconds for every package it builds
<radvendii>
i would love to find a way for it to remember that the host is down for the entire `nix build` command
<rmcgibbo[m]>
Looking at the commented out section of my cfg, I had done it with `nix.binaryCaches = [ "my local build machine that's not always online"]`;
<cgbrewster[m]>
Yeah, I'm not sure the 5 second per package will work unfortunately
<cgbrewster[m]>
Well thanks for the info anyways! Might toy around with a patch or change up what I'm doing 😃
<radvendii>
cgbrewster[m]: yeah, it's kind of a pain. there was someone last night who was helping me write a nix module that would wrap ssh in a script that remembers if the ssh has failed recently and fails immediately if so
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<aleph->
shofius: Using nixops which defaults to unstable technically so yeah.
<aleph->
Decided to finally make the push to using it now that some other bugs were fixed.
<aleph->
Err push to using unstable rather then a release.*
cfinch has quit [Read error: Connection reset by peer]
snk__ has left #nixos [#nixos]
cfinch has joined #nixos
<cole-h>
rmcgibbo[m]++ Thanks for the additional aarch64 runs. I really only wanted 1 more to ensure I fixed the aarch64 build issues :D
<{^_^}>
rmcgibbo[m]'s karma got increased to 7
<cole-h>
(I'm inclined to believe that the scancode-toolkit failure is misleading; at least, it works in pkgsCross)
<davidak[m]>
NobbZ: pypi and you are right that it is probably not included there
<lom>
tmpfs is /tmp
<lom>
it's in your memory
<tpw_rules>
KarlJoad: did you run out of tmpfs space? does df say anything is at 100%?
<simpson>
NobbZ[m]++ PyPI source packages can be incomplete, since they're packaged by a script which can fail to include everything from git/svn.
<{^_^}>
NobbZ[m]'s karma now has 2 digits!
<simpson>
KarlJoad: $ df -h
<davidak[m]>
NobbZ[m]++
<{^_^}>
NobbZ[m]'s karma got increased to 11
<lom>
does free -h also show tmpfs space?
<lom>
just wondering
<tpw_rules>
i think it's accounted for in buf/cache
<KarlJoad>
My user ran out of tmpfs space. /run/user/1000 has a full tmpfs.
<davidak[m]>
then no tests for nix users
<tpw_rules>
davidak[m]: or package from github. please
<KarlJoad>
How can I clear my user's tmpfs out?
<tpw_rules>
reboot would be best
<davidak[m]>
tpw_rules: github has no release. why can't this be easy :/
<davidak[m]>
yeah i do it
<KarlJoad>
That's what I figured. Then, second part, is there a way to increase the size of this user's tmpfs? A ulimit or something?
<NobbZ[m]>
<simpson "NobbZ[m]++ PyPI source packages "> This is on purpose. When I install some library from pypi (as a regular python user) I assume it has already been tested and do not want to run the tests again, instead I prefer the download and install sizes as small as possible
M0x0815[m] has left #nixos ["User left"]
lordcirth__ has joined #nixos
<matthewcroughan>
I need to make a change to obs-studio and or mpv to add ffmpeg-full instead of ffmpeg to its inputs.
<tpw_rules>
KarlJoad: you can remount it with a bigger size. but remember tmpfs lives in RAM, although it is swappable. you could look and see if anything looks deletable, it's kind of weird that it filled up
<matthewcroughan>
What would you rather do? Change mpv and obs-studio to ffmpeg-full? Or would you compile srt support into the ffmpeg library by default?
<matthewcroughan>
The reason is because mpv and obs-studio should both be able to read srt:// streams, but they can't right now because srt isnot compiledi into `ffmpeg` only `ffmpeg-full`
<matthewcroughan>
so what should the change be to? ffmpeg? Or to the obs-studio and mpv derivations.
<KarlJoad>
tpw_rules: I'm trying to compile a fairly large framework and I've been doing it all morning. I don't know how I filled the 3.2GB tmpfs, but I did.
lordcirth_ has quit [Ping timeout: 250 seconds]
<tpw_rules>
KarlJoad: then there's probably something in there that looks like compilation output or whatever you can safely delete. there's no problem per se wandering in and deleting stuff, but you might confuse some application. rebooting ensures everybody is done with it
<KarlJoad>
Fair enough. I'll just reboot.
exondation has joined #nixos
KarlJoad has quit [Quit: ERC (IRC client for Emacs 27.1)]
attila_lendvai has joined #nixos
hexchen has joined #nixos
<davidak[m]>
<NobbZ[m] "This is on purpose. When I insta"> NobbZ: do you know if there is a rule or recommendation for this practice? could we ask pypi to change it? i think pypi is the official source for most python package and we should use it, but also test packages when possible
<tpw_rules>
davidak[m]: you can file an issue on the git repo
<davidak[m]>
for every python package on pypi?
<tpw_rules>
pypi doesn't control what you upload
<davidak[m]>
they might have best practices
<KarlJoad>
tpw_rules: I was cleaning up some lorri roots and might have done something. I'm not sure. Rebooting cleared it out.
<KarlJoad>
Still my question stands, can I increase the tmpfs size allocated to my user through configuration.nix?
<tpw_rules>
i think it's systemd's responsibility. presumably you could write an appropriate unit
evils has joined #nixos
lom has quit [Read error: Connection reset by peer]
<KarlJoad>
It's definitely systemd's because it lives in /run/user/uid. I'll have to check.
<davidak[m]>
tpw_rules: i know SQLAlchemy has tests, because they always take long and we fetch it from pypi, so some packages have tests included
chenin has joined #nixos
<KarlJoad>
tpw_rules: It is systemd. RuntimeDirectorySize is the appropriate systemd variable. By defualt it's 10% of RAM.
<KarlJoad>
Doesn't seem like there's a configuration.nix variable for it. I'll just leave it. I have a RAM upgrade coming soon, so it's not a huge deal. I still have no ideal how i filled it hough. Zoom seems to have some effect though...
<AmandaC>
KarlJoad: are you sure it's not just something that is handled by a blanket value? Like how units have a service/unit/etc Config
<tpw_rules>
KarlJoad: if this is a temporary problem you could say e.g. mount -o remount,size=5G /run/user/1000 or whatever.
<KarlJoad>
AmandaC: What do you mean by blanket value? systemd's RuntimeDirectorySize defaults to 10% of total RAM capacity. I could write a systemd unit, but this was the first time I encountered this issue, so I think it's a bit of a weird time.
<KarlJoad>
tpw_rules: This is the very first time I encountered this, so I think it was a weird one-off thing.
<{^_^}>
[nixpkgs] @lukegb opened pull request #120452 → [20.09] nixos/test-driver: use a variety of different Tesseract settings for OCR → https://github.com/NixOS/nixpkgs/pull/120452
<lom>
It seems like my nvidia graphic drivers always stop working after i upgrade my channels with nix-env --upgrade and/or rebuild --upgrade. Does anyone have experience with that?
<lom>
last time this happened i played around with my config for a hour and a half
<lom>
and it magically started working again
zupo has quit [Client Quit]
mallox has joined #nixos
mallox has quit [Client Quit]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
cfinch has joined #nixos
cfinch has quit [Read error: Connection reset by peer]
attila_lendvai has quit [Ping timeout: 260 seconds]
beertoagunfight has joined #nixos
cfinch has joined #nixos
<{^_^}>
[nixpkgs] @lukegb merged pull request #120452 → [20.09] nixos/test-driver: use a variety of different Tesseract settings for OCR → https://github.com/NixOS/nixpkgs/pull/120452
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Raito_Bezarius>
if I have a nix shell with a dependency using src = ./some/local/path; does all derivations that depends on it will be rebuilt once I re-run the nix-shell?
<Raito_Bezarius>
it seems like I'm not getting this behavior though I expect it
<Raito_Bezarius>
(though if I change the version, it performs a rebuild)
<Raito_Bezarius>
(but I'd like to do some editable package source thing)