<camsbury>
hey all - having issues with CuDNN in a nix-shell. I'm using `tensorflowWithCuda` and `Keras`, and I'm getting the error in the following paste
<{^_^}>
#63817 (by ambrop72, 3 weeks ago, merged): wine: Set WINELOADER in wrapper, wrap also wine64.
shiver has joined #nixos
ilikeheaps has joined #nixos
Soo_Slow has quit [Remote host closed the connection]
<Xyliton>
ivan: how long would it take for this fix to be available on 19.03?
ilikeheaps has quit [Ping timeout: 245 seconds]
<kalbasit>
how to access a member of a set from a variable without doing this `"pythonPackages.${pythonPackage}"`? Where `pythonPackages = python37Packages` and `pythonPackage = "aria"`.
<iqubic>
Also, that seems to require having things plugged in at the time that I start the X Server.
<clever>
iqubic: with xrandr and xfce, hotplug is fine
hio has quit [Quit: Connection closed for inactivity]
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
<iqubic>
I'm not using XFCE.
<Ralith>
if you're not comfortable writing elisp to configure your window manager, maybe you shouldn't use a window manager that requires you to write elisp to configure it
<iqubic>
I'm comfortable writing ELisp. It's just, I don't know what sort of ELisp I need.
<iqubic>
Specifically, I have no idea when the exwm-randr-screen-change-hook is called.
vmandela has joined #nixos
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
<iqubic>
Actually, reading through the documentation is the way to best understand this.
abathur has quit [Ping timeout: 246 seconds]
<Ralith>
it's almost like it's there for a reason :p
likivik has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
Havvy has joined #nixos
<iqubic>
Yeah. I love that I can just pull up documentation for anything in emacs.
<iqubic>
Also, I just found out that "M-x man" exists to pull up the man page of any command and put it into a buffer. Makes it real nice to search for a specific feature I want.
<likivik>
Greetings! New nixos user here. Can anybody help me with nvidia (prime, dynamic) drivers and steam games? I followed instructions on Nvidia and Steam pages from wiki, did rebuild and reboot. Trying to start steam with primerun.sh from terminal - results in 'Fatal Server error: (EE) No screens found. Would appreciate and all help!
<iqubic>
Can you start steam by just running it from the command line?
<iqubic>
like enter "steam" into the command line and see what happens?
<likivik>
yes, i can
mexisme_ has quit [Ping timeout: 246 seconds]
<iqubic>
So why do you need primerun.sh?
<likivik>
it starts fine, but i'm pretty sure it uses intel video to start games.
<day|flip>
make me wonder if your running both intel and nouveau
fragamus has joined #nixos
endformationage has quit [Quit: WeeChat 2.5]
<day|flip>
likivik: what kernel are you running?
<likivik>
tried with stable 4.19 and now with 5.1.18
palo1 has joined #nixos
<day|flip>
from services.xserver.videoDrivers = [ "nvidia" ]; to services.xserver.videoDrivers = [ "nvidiaBeta" ];?
b has quit [Ping timeout: 272 seconds]
palo has quit [Ping timeout: 272 seconds]
palo1 is now known as palo
<likivik>
Oh, maybe I didn't write it but I have muxless/non-MXM Optimus card, so I believe that doesn't apply? I'm using the Nvidia PRIME section of https://nixos.wiki/wiki/Nvidia
gerschtli has quit [Quit: WeeChat 2.4]
gerschtli has joined #nixos
zupo has joined #nixos
zupo has quit [Client Quit]
<likivik>
also $ lspci -v shows something weird: 0a:00.0 3D controller: NVIDIA Corporation GM108M [GeForce 840M] (rev ff) (prog-if ff) DeviceName: NVIDIA GeForce GT 840M !!! Unknown header type 7f Kernel modules: nvidiafb, nouveau
zupo has joined #nixos
<day|flip>
maybe you can try blacklisting nouveau at boot: boot.kernelParams = [ "nouveau.modeset=0" ];
iqubic has quit [Remote host closed the connection]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<day|flip>
ill brb
day|flip has quit [Quit: ERC (IRC client for Emacs 26.1)]
<likivik>
Thank you, was just trying to figure out how to do it with nix:)
fragamus has joined #nixos
fragamus has quit [Client Quit]
day|flip has joined #nixos
Church- has quit [Ping timeout: 248 seconds]
likivik has quit [Ping timeout: 260 seconds]
likivik56 has joined #nixos
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
Mateon1 has quit [Remote host closed the connection]
FireFly has quit [Quit: Goodbye]
FireFly has joined #nixos
Xal has quit [Ping timeout: 245 seconds]
Mateon1 has joined #nixos
Xal has joined #nixos
slack1256 has quit [Remote host closed the connection]
ar1a has quit [Quit: ZNC 1.6.6+deb1 - http://znc.in]
<{^_^}>
#64785 (by Taneb, 6 days ago, open): q: init at 1.7.1
<bahamas>
does nix automatically download updates if there have been updates to the corresponding channel?
<bahamas>
I ran nix-shell and it downloaded some derivations, although I had done a full setup in the past
hyper_ch2 has quit [Remote host closed the connection]
hmpffff has joined #nixos
<teto>
bahamas: if your local channel advanced, then nix-shell will fetch the matching derivations yes
hexa- has quit [Ping timeout: 276 seconds]
hmpffff has quit [Client Quit]
<sb0>
the pypy3 package has been broken on nixos-unstable for quite some time now, and I don't understand why... anyone knows what is going on?
<bahamas>
teto: I have a question about how I should approach pinning. is it safe to just use a stable channel or should I pin that channel to a specific commit as well?
xelxebar has quit [Remote host closed the connection]
xelxebar has joined #nixos
<teto>
bahamas: depends what you want, if you don't pin, when you update the channel, you will have to redownload some packages.
<bahamas>
teto: how bad is it if I pin? missing security updates would be the biggest issue, right?
<dminuoso>
bahamas: You can pin single packages.
<teto>
bahamas: yes
<dminuoso>
Mmm, something akin to ⌜>= 0.1 && < 0.2⌝ is not trivially possible with how nix works I guess?
<rnhmjoj>
is it possible to add new arguments to a package? i'd like to extend a package but the expression is generate by a shell script, so i think i must use .override or something like that
<teto>
bahamas: dminuoso has a point. Depends on your motive but you can pin only some packages (I pin the big ones or some packages that are likely to break)
magnetophon has quit [Ping timeout: 244 seconds]
ericsagnes has quit [Ping timeout: 272 seconds]
<teto>
dminuoso: python and lua do some checks on package bounds (haskell too ofc)
ottidmes has quit [Quit: WeeChat 2.4]
hyper_ch2 has joined #nixos
<bahamas>
teto: so if I just pin the channel, there is still a risk that a security update will break a package
cfricke has quit [Ping timeout: 268 seconds]
<dminuoso>
teto: Well yeah but that's not part of nix really. I mean I cant trivially state `I want the latest openssl version in that range` unless there was some meta attribute set openssl_1_0
<dminuoso>
bahamas: You dont need to pin the entire channel. You'd rather have a sort of "additional pinned channel" that you can subscribe to for single packages.
___laika has quit [Ping timeout: 268 seconds]
hmpffff has joined #nixos
<dminuoso>
bahamas: Think having access to both `nixpkgs` and `nixpkgsAtSomeCommit` floating around. Then you can use nixpkgs.foo and nixpkgsAtSomeCommit.bar
<FRidh>
sb0: no maintainer perhaps? I know it's a pain to maintain that package becuse it takes forever tobuild
<dminuoso>
In the end a channel is just a big attribute set.
<dminuoso>
Or at least its an URL pointing to one.
magnetophon has joined #nixos
<bahamas>
dminuoso: well, what I mean is that I don't want the project to rely on the channel that is set as nixpkgs on the machine where I run nix. that's why I want to specify at least the channel in the project
<bahamas>
dminuoso: yes, I've already seen that page
<dminuoso>
bahamas: A channel is just a way to maintain a nixpkgs set. You can introduce your own nixpkgs unrelated to an outside channel in a pinned manner like this.
<dminuoso>
bahamas: If you are really nice, parametrize over it and just have it a default of your derivation
<dminuoso>
So that a user can override that pinned nixpkgs if they need to.
<bahamas>
my goal right now is to reduce the probability that something will break in the future when I or someone else run nix-build or nix-shell
<{^_^}>
[nixpkgs] @worldofpeace pushed 4 commits to master: https://git.io/fjMS6
<bahamas>
it sounds like pinning a specific commit in the channel I'm using is the most conservative approach
veske has joined #nixos
fendor has joined #nixos
hexa- has joined #nixos
<teto>
bahamas: if that's your goal, yes pinning the whole thing seems right
cransom has quit [Quit: WeeChat 2.4]
Jackneill has quit [Ping timeout: 246 seconds]
aswanson has quit [Quit: WeeChat 2.4]
ericsagnes has joined #nixos
shomodj has joined #nixos
bahamas has quit [Read error: Connection reset by peer]
<jackdk>
and he has one on using the free group to do an advent of code problem. in coding, it's like you add the minimal amount of structure to get the interface you want
<Taneb>
Poking around some old Nix code we have and there's stuff like 'import { outPath = ./.; revCount = 0; gitTag = "dirty"; } {}'. Is this form of import documented?
<srhb>
Taneb: It's usually related to things that are sometimes hydra inputs.
<gchristensen>
right, those are arguments hydra knowshow to pass -- and no, I don't think they are well documented
<Taneb>
Ah, that's a shame, I'd like to know more about this
<gchristensen>
what would you like to know?
<Taneb>
What the intention is behind it, what the attributes mean, are there any other attributes that could be in there, etc
<gchristensen>
the idea is sometimes you want a build to know where it came from, for example so the manual can say what version of nixpkgs it is for
<{^_^}>
[nixpkgs] @abbradar merged pull request #65212 → nixos-test-driver: allow configuration of net frontend and backend → https://git.io/fjMg2
<{^_^}>
[nixpkgs] @abbradar pushed commit from @thefloweringash to master « nixos-test-driver: allow configuration of net frontend and backend »: https://git.io/fjMQC
<maralorn>
fendor_: I am not experienced enough to tell. But it looks to me that you have a shell-script there called "build.hs" Isn't that a bit unusual?
<maralorn>
Just from the file extension.
<fendor_>
that was a typo, it is actually build.sh
mkoenig has quit [Ping timeout: 244 seconds]
<maralorn>
Ah, okay.^^ I took it seriously because a lot of haskell projects have a build.hs.^^
<fendor_>
yeah, ironically, this project already has a haskell build script, Shake.hs which I also used for experimentation
<fendor_>
next question, I tried the same with a bit more complex application, I have: https://hastebin.com/fuyoxamolu.bash with the error shown at the bottom. Using `make` in the directory works, but shows the same warnings
srid6390803 has quit [Ping timeout: 272 seconds]
npmccallum has joined #nixos
<fendor_>
I have to do a hit and run, sorry, have to go, gonna just ask it again later, ciao
sigtrm has quit [Read error: Connection reset by peer]
mudri has joined #nixos
sigtrm has joined #nixos
Tucky has quit [Quit: WeeChat 2.5]
<alexarice[m]>
nilsirl: I think some packages have dev versions, it depends on the package/language really
<nilsirl[m]>
I am trying to contribute to a project but I don't know how to compile it :joyful:
<nilsirl[m]>
On other distros to compile it I just run scons
<nilsirl[m]>
But on nixos I am not sure how to do it
<infinisil>
nilsirl[m]: That's really what nix-shell is made for. Run `nix-shell '<nixpkgs>' -A hello` to get an environment with the deps for hello
<infinisil>
`unpackPhase` to unpack the source, the cd into the source directory, `configurePhase` to configure it, then `buildPhase` to build it (which is usually mostly just `make`)
bakakuna has quit [Ping timeout: 252 seconds]
<alexarice[m]>
infinisil: beat me to it, though I didn't realise you could pass an attribute into nix-shell, thanks for the info
<{^_^}>
[nixpkgs] @fpletz merged pull request #65088 → prometheus-nginx-exporter: switch to official version & general prometheus-exporters cleanup → https://git.io/fjMkF
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
primeos has quit [Quit: WeeChat 2.4]
<nilsirl[m]>
When I run buildPhase it says "no Makefile, doing nothing". This is fine since the project I am working on uses "scons". The thing is that when I run scons (within nix-shell) I get a compile error "In file included from platform/x11/os_x11.h:37:0,
<nilsirl[m]>
from platform/x11/godot_x11.cpp:37:
<nilsirl[m]>
./drivers/alsa/audio_driver_alsa.h:38:10: fatal error: alsa/asoundlib.h: No such file or directory"
<nilsirl[m]>
infinisil: build tool doesn't resolve the #include.
SovereignBleak has quit [Quit: Connection closed for inactivity]
hmpffff has quit [Quit: nchrrrr…]
<infinisil>
Ah, yeah so unfortunately there's no general way to run the *actual* phases. In your case you probably need to call `sconsConfigurePhase` or so
<infinisil>
And sconsBuildPhase
PLPD-Bot has quit [Remote host closed the connection]
<infinisil>
I think
fragamus has joined #nixos
dhs has joined #nixos
xkapastel has joined #nixos
<infinisil>
Actually there might be a general way, but I forgot how to do it
<nilsirl[m]>
I have sconsBuildPhase but not sconsConfigurePhase
<nilsirl[m]>
And when I run sconsBuildPhase I get the same error as when running scons "normally"
<infinisil>
Yeah the configure thing doesn't seme to exist, was just a guess
<nilsirl[m]>
Isn't there something I can do with nix-build?
dhs has quit [Remote host closed the connection]
hmpffff has joined #nixos
cia_49 has joined #nixos
dwdr has quit [Ping timeout: 245 seconds]
cia_49 has quit [Remote host closed the connection]
Thra11 has joined #nixos
veske has quit [Quit: This computer has gone to sleep]
nDuff has joined #nixos
nakkle has quit [Ping timeout: 276 seconds]
nakkle has joined #nixos
<vaibhavsagar>
is there some way I can prevent overlays in `.config` from being loaded?
<vaibhavsagar>
seems very impure to have my expressions be affected by arbitrary file content
noogie has quit [Quit: Lost terminal]
<sauyon>
wouldn't configuration.nix be impure by that logic
__Sander__ has quit [Quit: Konversation terminated!]
<nilsirl[m]>
infinisil: genericBuild seems to work but I am not sure yet
<nilsirl[m]>
infinisil: genericBuild seems to work but I am not sure yet
srid6390803 has quit [Ping timeout: 245 seconds]
<infinisil>
nilsirl[m]: Yeah that does work, but it'll do a full build every time
<infinisil>
There's no granularity like a configure phase, build phase, etc.
hmpffff has quit [Quit: nchrrrr…]
MmeQuignon has joined #nixos
srid6390803 has joined #nixos
<alexarice[m]>
Do emacs packages seem to be broken for anyone else? getting a lot of Error's connecting to github and I get 500 errors when I manually navigate to the links
primeos has joined #nixos
<mkaito>
sounds more like a github than an emacs issue to me
<infinisil>
alexarice[m]: It's on github's side, I'm having trouble accessing the webpage
srid6390803 has quit [Ping timeout: 268 seconds]
<alexarice[m]>
there's nothing wrong with the nix derivation then? I guess I'll wait a bit
<infinisil>
emacs packages are kind of the exception because they get fetched from github directly, instead of using cache.nixos.org
<alexarice[m]>
I seem to have trouble pushing to github as well, maybe they are having problems
Neo--- has quit [Ping timeout: 264 seconds]
<tilpner>
They've been giving me server errors for a few hours too
<tilpner>
And now I can't deploy anymore :c
jmeredith has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @globin pushed commit from @johanot to master « nixos/gitlab: gitlab-workhorse requires exiftool on path to process uploaded images »: https://git.io/fjMbj
<{^_^}>
[nixpkgs] @globin pushed commit from @johanot to release-19.03 « nixos/gitlab: gitlab-workhorse requires exiftool on path to process uploaded images »: https://git.io/fjMNe
<DariusTheMede>
clever: vaibhavsagar has sorted me out thanks
<vaibhavsagar>
wait what
<vaibhavsagar>
oh hi :)
<DariusTheMede>
:)
<vaibhavsagar>
I'm still upset that overlays in `.config/nixpkgs` are automatically picked up
<pbb>
The application I am trying to package has dependencies from it's language's packaging system, that is not tracked in nixpkgs (npm, rubygems, maven, rust crates, ...). There is tooling to adapt the dependency definitions from the language's package management system to nix (yarn2nix, bundix, ..), but since the original dependency definitions don't contain usable hashes for all dependencies, or are missing the hashes for git dependencies, these dependency
<pbb>
I am looking for the best solution for the following problem that I run into quite a lot:
<pbb>
2) downloading the dependencies during build time with a fixed-output derivation (jd-gui package, rust crates)
<pbb>
1) generating a .nix file describing the dependencies including all needed hashes and including it in the packaging (gemset.nix, yarn.nix)
<pbb>
definitions need to be combined with information from the internet. There are multiple solutions that I see being used:
<pbb>
Which one is actually more favorable? The first one results in difficult-to-maintain packages and spams nixpkgs with large files, the second one is not strictly pure, and it is kind of working around nix using the fixed-output derivation, I think I read edolstra discouraging it.
shomodj has quit [Ping timeout: 268 seconds]
<clever>
vaibhavsagar: i would expect that to work, try putting a syntax error in an overlay from ~/.config/nixpkgs and then paste me the error it gives
mexisme_ has quit [Ping timeout: 276 seconds]
<clever>
vaibhavsagar: is it an overlay, or config.nix ?
<vaibhavsagar>
I think it's in `pkgs.nix`, so I'll fix that
<pbb>
infinisil: I did not say the nix description is generated and then imported. For example the rust tooling just downloads the dependencies using cargo in a fixed-output derivation.
<clever>
vaibhavsagar: ah yeah, your `inherit (import <nixpkgs>{}) lib;`
<infinisil>
pbb: Ahh like that, I see
<slabity>
What is the status of this 'flakes' proposal? I really like the concept and want to move my modules over to using it.
* vaibhavsagar
shakes fist at implicit overrides
<pbb>
infinisil: The same could be done for npm (or ruby gems with a little hassle)
<clever>
vaibhavsagar: that one is broken by the syntax error, but the overlays shouldnt impact other things, so it may be a false alarm, purify it and then see what happens next
<infinisil>
pbb: Well I prefer the second approach if you can make it as reproducible as possible. With the rust dependency sha256's i remember there being some problems because they weren't entirely deterministic and could therefore change over time
<vaibhavsagar>
clever: seems fine after I fixed that
<clever>
vaibhavsagar: not sure what the original issue was then, enless your overlays are messing with lib
<vaibhavsagar>
clever: DariusTheMede was having an issue getting my notebooks repo working because of his overlays
MmeQuignon has joined #nixos
<pbb>
infinisil: but I think there is no consensus on what solution should be preferred? Or is there existing discussion about this topic?
<infinisil>
pbb: I don't know of any discussion about this, maybe you can open an issue on this
<pbb>
In nixpkgs or in nix? I think it is a general nix thing, but there should probably be a decision on how it will be handled in nixpkgs.
Drakonis has quit [Ping timeout: 250 seconds]
<tilpner>
pbb: Right, this is an open problem still
Soo_Slow has quit [Remote host closed the connection]
<lambda-11235>
I don't seem to be getting any haskellPackages, is there an obvious reason for this?
<pbb>
tilpner: I was aware of the recursive nix discussion, although I don't fully understand all implications. It seems to me like recursive nix can solve the issue in those cases where all hashes are in the lockfile, like yarn.lock without git dependencies, but where hashes are missing the problem is not solved by recursive nix
<pbb>
I would also be interested to discuss this at nixcon if anyone else is interested
<tilpner>
pbb: Yes, but just gathering all hashes can be done with less space overhead than putting the full expressions into nixpkgs
<clever>
pbb: another thing recursive nix can give you is a ccache replacement
<pbb>
clever: sounds really cool, I am looking forward to playing with that ^^
<clever>
pbb: what if gcc was replaced with a shell script, that would pre-process the c code, and remove all comments (like ccache), then it just did `nix-build gcc.nix --arg source ./foo.c`
<clever>
pbb: then if foo.c matches a previously built thing, it builds instantly
o1lo01ol1o has quit [Remote host closed the connection]
<clever>
pbb: nix-shell and nix-build could even share intermediate products!
<pbb>
tilpner: I don't fully understand the last thing you wrote
Mrmaxmeier has joined #nixos
<pbb>
you mean solution 2) I described earlier or another alternative involving recursive nix?
<tilpner>
pbb: It would be a combination of both. You would still run a tool to update the hashes, but the rest of the expression generation can be done later during build. Smaller files, and not as difficult to maintain anymore
<pbb>
so where would those hashes be stored and in what format?
<tilpner>
pbb: But admittedly that's not an amazing improvement over the current state, it "only" reduces file sizes. I might be missing something too
<pbb>
just a smaller format for the same hashes we already have with yarn.nix and gemset.nix?
<tilpner>
pbb: Still in nixpkgs, probably per-tool formats
<tilpner>
Yes
<pbb>
ok I get it now, thanks for explaining :-)
<pbb>
of course it would be nice to not need those at all, or for yarn2nix maybe to only store the hashes for git dependencies and generate the rest on the fly from the upstream yarn.lock
<{^_^}>
[nixpkgs] @aanderse opened pull request #65273 → linux/kernel: enable automatic memory hotplug support → https://git.io/fjMNb
srid6390803 has quit [Ping timeout: 258 seconds]
<pbb>
tilpner, infinisil: do you have a suggestion how I could name the issue we discussed with hashes being missing in the upstream dependency description? And what repository should I open this issue in?
<tilpner>
pbb: Maybe I misunderstood what you meant. Wasn't your point that putting large expressions into nixpkgs is ugly, but shoving the fetching of hashes into a fixed-output build is too impure?
<{^_^}>
#65274 (by petabyteboy, 13 minutes ago, open): the future of npm tooling in nixpkgs (yarn2nix)
<pbb>
tilpner: yes, there are two approaches to language package management with nix, but they both have problems, and I don't know which one to use
<pbb>
and if one is better than the other I think this one should be used in all tooling
<tilpner>
But the first one works fine outside of nixpkgs
<pbb>
gchristensen: but an internet connection is not available during the build
<pbb>
(unless you do the dirty "one big hash" thing)
<gchristensen>
I evidently don't have enough context
<tilpner>
(I'm still arguing for the "condensed many hashes" method I mentioned earlier)
<tilpner>
(For that you need internet anyway, but then it works offline-ish)
<pbb>
tilpner: I also think that is the most comfortable option, but maybe gchristensen can explain why the "one big hash" option is not nice
<pbb>
gchristensen: the context is actually the following: this came up when I was packaging mastodon. using the approach of prefetching all hashes and generating a nix expression and putting it into nixpkgs was seen as not so nice because it bloats nixpkgs, and because it makes it a lot of efford to override the source with a fork or new version
<gchristensen>
well because it is a lie
<pbb>
someone complained in the PR that it's so hard to use a forked version when you need to get bundix and yarn2nix and generate yarn.nix and gemset.nix for every update. so it made me think if there is a better way
<gchristensen>
yeah
<gchristensen>
it is complicated for sure
<{^_^}>
[nixpkgs] @globin merged pull request #65179 → nixos/bind: allow manual additions to zone config fragments → https://git.io/fjM8d
<eraserhd>
Is there a better way to escape '' inside a '' string, other than ${"''"}?
<pbb>
maybe it is actually easier to have a talk about this irl at nixcon if anyone cares. I think those issues are the ones holding me back the most when using nix.
lordcirth has quit [Remote host closed the connection]
lordcirth has joined #nixos
o1lo01ol1o has quit [Ping timeout: 245 seconds]
shomodj has joined #nixos
nDuff has quit [Ping timeout: 268 seconds]
day|flip has joined #nixos
<pbb>
adisbladis: what do you mean by "cache busting" in your last comment in that PR?
<adisbladis>
pbb: That for any change in the dependency graph you invalidate the entire cache
shomodj has quit [Client Quit]
<pbb>
thanks :-)
<adisbladis>
Let's say you have a `buildGoPackage` modSha256 and you update one of your deps, now you need to download them all again :/
<pbb>
yeah I know that problem
<gchristensen>
and build them all again
<pbb>
oh yes it's so annoying with buildRustPackage :<
<gchristensen>
and cache them all again
<adisbladis>
gchristensen: Though for the go case you'll build them all again anyway
<gchristensen>
and good luck trying to get it to work properly on macos and linux
fusion809 has quit [Remote host closed the connection]
<day|flip>
I notice that with my little custom kernel config
<adisbladis>
day|flip: Yeah... NixOS is the first distro to convince me of genkernel as a concept ^_^
<{^_^}>
[nixpkgs] @nbardiuk opened pull request #65276 → vimPlugins.ghcid: Init at 2019-07-04 → https://git.io/fjMxJ
<gchristensen>
oh?
noudle has joined #nixos
<day|flip>
after deleting my gc. i have to rerun nixos-rebuild switch --update will i think it mostly udev rules get deleted some how. and have to re build everthing. get it all back after rebooting again
Jackneill has quit [Remote host closed the connection]
<gchristensen>
day|flip: I can't parse your question
<adisbladis>
I wish we could reach a concencus about fixed-output derivations and when they're supposed to be used
<adisbladis>
gchristensen: That sounds sensible to me. Obviously there are other people out there that think buildRustPackage/buildGoModule are reasonable approaches
<day|flip>
i notice after deleting my gc. my udev rules dont run. so i have rebuild once more after doing nix-collect-garbage -d
<slabity>
aminechikhaoui: Looks like adding an EBS volume killed it with a kernel panic. I'll see if I can figure out why
<gchristensen>
I've never heard of gvac-source
<aminechikhaoui>
slabity oh I see, that's a bit odd, what ami/nixpkgs version are you using
<tilpner>
,feeds = https://feed.nix.tx0.co provides Atom feeds for nixpkgs issue labels to notify you about new issues tagged with that label from within e.g. Thunderbird
<{^_^}>
feeds defined
<aminechikhaoui>
tilpner++
<{^_^}>
tilpner's karma got increased to 29
<gchristensen>
hot DOG tilpner!
<slabity>
aminechikhaoui: Not sure. The latest one that NixOps looks for I'd presume?
<kraem>
i ran `find /nix/store -type d | grep emacs` to find something and i found gvac-source which seems to be a clone of the nixpkgs repo - was just curios if that whats get updated when calling `nixos-rebuild --upgrade`
<slabity>
It looks like the volume was not formatted, despite having `autoFormat = true;` in the filesystem config
<gchristensen>
tilpner: do you scrape these?
ris has joined #nixos
<tilpner>
gchristensen: If by scraping you mean "extracting information from e.g. HTML", then no. Their API gives access to issues (and sometimes even works)
<samueldr>
I was seeing that the channel updated, but no announcement here
<gchristensen>
oh!
<samueldr>
guess it must have been during the interval it does whatever it does
<gchristensen>
it is a pull-based system :)
mexisme_ has joined #nixos
jmeredith has quit [Quit: Connection closed for inactivity]
<slabity>
aminechikhaoui: Did not know that was possible. I'll try it, but I still want to know how to attach volumes properly. It looks like it never gets formatted and initd dies when trying to mount it
<slabity>
Jeez, I have like, 15 terminated EC2 instances in my console now from all my attempts :P
gratto has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<slabity>
Looks like it worked. Thanks for the help
fragamus has joined #nixos
<slabity>
animechikhaoui++
<slabity>
aminechikhaoui++
<{^_^}>
aminechikhaoui's karma got increased to 2
<aminechikhaoui>
attaching volumes is done later after the boot in regular deployments which takes care of formatting the volume, then rebooting won't try to format again
day|flip has quit [Quit: ERC (IRC client for Emacs 26.1)]
<pbb>
gchristensen, adisbladis: about fixed-output derivations, what would be the bad thing about generating yarn.nix / gemset.nix in a fixed-output derivation? it would make mantainance a lot easier
<pbb>
and it doesn't have the caching issue
<adisbladis>
pbb: How are you going to consume these files? IFD?
<adisbladis>
Tbh once you've done that you might as well do that step in an update script
<pbb>
that's a seperate problem in my opinion. yes for hydra it would still be necessary to add those files to nixpkgs, but for overriding the source with a fork or when updating to a new version
<pbb>
and could this be solved with recursive nix in the future?
DariusTheMede has joined #nixos
day|flip has joined #nixos
<tilpner>
Avoidance of IFD isn't just relevant to Hydra. Single-stage evaluation is an incredible feature, and import/readFile-from-derivation prevent that
<pbb>
where is it used other than hydra?
<pbb>
but yeah I see it is not nice having to build a derivation to see that part of the dependency graph
<adisbladis>
If you have figured out how to generate a yarn.nix/gemset.nix you've already done the hard parts. Defering this to the build would just be a matter of convenience(?) for the package author?
<pbb>
and for users messing with the package
<pbb>
they could just override the source and dependencies FOD hash (first with garbage to get the correct one)
DariusTheMede has quit [Ping timeout: 246 seconds]
<adisbladis>
Tbh I find FOD makes it more difficult to do that. YMMV.
<pbb>
maybe I haven't seen good update scripts
<pbb>
I guess a good update script should make it just as easy to do that
<pbb>
but I don't know how I could make it that easy with an update script
<pbb>
the user overriding the source will always have to get the update script and put the generated files somewhere
bakakuna has joined #nixos
<pbb>
and changes in the mechanism used to generate those files (yarn2nix or bundix) are not reflected, and have to be accounted for manually
<clever>
pbb: i like having update.nix files with a shellHook, where you just run `nix-shell update.nix`
<clever>
pbb: you can then use buildInputs to get whatever you need (purely) and shellHook to run things impurely, and `exit 0` at the end to stop the shell from actually opening
<clever>
pbb: ${toString ./.} gets you the dir the nix file was in
noudle has quit []
<pbb>
clever: thanks
hedgie_lappy has left #nixos ["Konversation terminated!"]
<pbb>
clever: would you use callPackage or where does the update script source its' tools
day|flip has quit [Quit: ERC (IRC client for Emacs 26.1)]
o1lo01ol1o has joined #nixos
bakakuna has quit [Ping timeout: 248 seconds]
DariusTheMede has joined #nixos
tilpner has quit [Ping timeout: 245 seconds]
pamplemo1sse has quit [Ping timeout: 245 seconds]
tilpner has joined #nixos
<tilpner>
pbb: Sorry, phone call
DariusTheMede has quit [Ping timeout: 268 seconds]
<tilpner>
pbb: Deployment of server systems gets a lot easier if you can just send over a description of how the system should look like (list of derivations), and then have the server realise it
phreedom_ has joined #nixos
day|flip has joined #nixos
<tilpner>
pbb: But if you need to build something to get that list of derivations, you lose a lot of the benefit
<pbb>
ah yes
phreedom has quit [Ping timeout: 260 seconds]
srl295 has joined #nixos
domogled has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
dftxbs3e has joined #nixos
Neo--- has joined #nixos
zupo has joined #nixos
<alienpirate5>
I'm having trouble cross-compiling alsa-firmware to armv7l
<alienpirate5>
It's trying to compile a C binary `tobin` and then run it on the hos
<alienpirate5>
* It's trying to compile a C binary `tobin` and then run it on the host
<alienpirate5>
The problem is that the C binary is being cross-compiled and so cannot run on the host
<tilpner>
Why's that a problem?
<alienpirate5>
you can see for yourself by running `nix build nixpkgs.pkgsCross.armv7l-hf-multiplatform.alsa-firmware`
<alienpirate5>
it tries to run `tobin` as part of the compilation
<alienpirate5>
but since `tobin` was compiled to run on armv7l, not x86_64, it cannot run
<tilpner>
Oh, misread, sorry
<infinisil>
alienpirate5: What is it running that binary for?
<alienpirate5>
it includes the firmware files which are stored as C byte arrays
<alienpirate5>
then outputs them to binary files
<tilpner>
alienpirate5: If you're in a hurry, you can avoid fixing cross compilation with qemu-user
<nilsirl[m]>
I am trying to build a software using "SCons" (a build system). It seems to me that when SCons executes commands (e.g. g++), it doesn't use the g++ wrapper which allows access to libraries. However this software (godot) is in the nix packages collection which means it is possible to build it it just somehow doesn't work on my machine. Also when I run the commands SCons is suppose to execute it works which I what makes me
<nilsirl[m]>
believe SCons skips the wrapper somehow.
<nilsirl[m]>
* I am trying to build a software using "SCons" (a build system). It seems to me that when SCons executes commands (e.g. g++), it doesn't use the g++ wrapper which allows access to libraries. However this software (godot) is in the nix packages collection which means it is possible to build it but somehow doesn't work on my machine. Also when I run the commands SCons is suppose to execute it works which I what makes me
<nilsirl[m]>
believe SCons skips the wrapper somehow.
<clever>
nilsirl[m]: scons has some "handy" utils to make the build env pure, and unsets all env vars it doesnt understand
<clever>
nilsirl[m]: that breaks cc-wrapper
<nilsirl[m]>
How do I do?
<clever>
nilsirl[m]: cant remember what i was doing to fix it
knupfer has joined #nixos
<nilsirl[m]>
The thing is that this workaround must be known because otherwise it wouldn't be possible to do `nix-env -i godot` as godot is in the nix packages collection.
<srhb>
aveltras: wrapProgram and friends seem more fitting for that purpose
<srhb>
aveltras: (Because you get to keep the original derivation intact, meaning more cache)
acowley has quit [Ping timeout: 258 seconds]
<aveltras>
srhb: is it possible to add environment to an existing service definition ?
DariusTheMede has joined #nixos
DariusTheMede has quit [Ping timeout: 272 seconds]
mexisme has joined #nixos
Diagon has joined #nixos
erasmas has quit [Quit: leaving]
<alexarice[m]>
aveltras: you might be able to use an overlay to modify a package that a service uses
mniip has quit [Remote host closed the connection]
<alexarice[m]>
If a package is trying to build on my system and I think it should have been cached does that mean it will fail, in other words is every non broken package built by hydra before unstable moves forwards
m0rphism has quit [Ping timeout: 245 seconds]
drainful has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
ambro718 has quit [Quit: Konversation terminated!]
<samueldr>
it even works right through the network!
<andi->
samueldr: thankt for that. The wrapper line is a bit different from the one I found in another issue / the spice_gtk package
<andi->
We should make that a proper option - if this works :-)
<samueldr>
I guess so
<megfault>
samueldr: thanks for the help
<samueldr>
you're welcome :)
abcxyzchris has joined #nixos
<selfsymmetric-mu>
,locate bin yq
<{^_^}>
Found in packages: yq
shibboleth has quit [Quit: shibboleth]
<samueldr>
I was wholly surprised when I saw that usb redirection was working through the network, well enough that I was able to use proprietary tools to unlock a phone
<selfsymmetric-mu>
Hm...I think there's more than one yq out there. I'll have to figure this one out...
<samueldr>
while virtualbox's usb redirection wouldn't work for that purpose
<selfsymmetric-mu>
Anybody work with a lot of YAML? I'm looking for some lightweight query tools.
<andi->
samueldr: the wonders of spice ;-)
<abcxyzchris>
Is there any way I can 'locate' a file / path that is provided by a nixpkg package?
<samueldr>
abcxyzchris: I assume before installation?
<samueldr>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<samueldr>
nix-index (link provided in there) does wonders for that
<samueldr>
though it only can index what hydra builds, so it can't find unfree software
<abcxyzchris>
,locate /usr/share/lxc/templates
<{^_^}>
Couldn't find in any packages
<samueldr>
since nixos doesn't have /usr (except for /usr/bin/env) it won't find that
<abcxyzchris>
samueldr: Thanks that looks super usefull!
abf_ has joined #nixos
<samueldr>
,locate share/lxc/templates
<{^_^}>
Found in packages: lxc
lambda-11235 has joined #nixos
wildtrees has quit [Quit: Leaving]
pamplemousse has quit [Ping timeout: 272 seconds]
fresheyeball has quit [Ping timeout: 268 seconds]
stepcut has joined #nixos
<samueldr>
hm? didn't unused kernel configurations in out kernel builder fail the build previously?
<{^_^}>
[nixpkgs] @lpetre opened pull request #65285 → squashfuse - fix sha for 0.1.103 → https://git.io/fjDv8
<kalbasit>
`collision between `/nix/store/3pr3f51vf71slz3jvmhfq746aqsaaa0f-python3.7-jsonpatch-1.23/bin/jsondiff' and `/nix/store/xxhdwh0g54rkkkm0105c7wr90h5y073y-python3.7-jsondiff-1.1.2/bin/jsondiff'`
<kalbasit>
moto depends on jsondiff and on cfn-lint. cfn-lint depends on jsonpatch
stepcut has quit [Remote host closed the connection]
<selfsymmetric-mu>
,locate dhall-repl
<{^_^}>
Couldn't find in any packages
<selfsymmetric-mu>
:(
Diagon has quit [Quit: Leaving]
fresheyeball has joined #nixos
<fresheyeball>
clever: how can I run memtest when I am using systemd-boot?
Rusty1 has joined #nixos
atlas_ has quit [Ping timeout: 245 seconds]
liberiga has joined #nixos
atlas_ has joined #nixos
fresheyeball has quit [Ping timeout: 248 seconds]
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 250 seconds]
mbrgm_ is now known as mbrgm
magnetophon has quit [Ping timeout: 272 seconds]
stepcut has joined #nixos
bdesham[m] has left #nixos ["User left"]
abcxyzchris has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixops] @grahamc closed pull request #981 → Delete all the providers (and use plugins instead) → https://git.io/fNcqQ
stepcut has quit [Remote host closed the connection]
thc202 has quit [Ping timeout: 250 seconds]
stepcut has joined #nixos
lambda-11235 has quit [Ping timeout: 245 seconds]
fresheyeball has joined #nixos
<fresheyeball>
clever: I ran memtester and it passed