hio has quit [Quit: Connection closed for inactivity]
the-kenny has quit [Ping timeout: 255 seconds]
mexisme has quit [Ping timeout: 248 seconds]
lovesegfault has quit [Quit: WeeChat 2.4]
freeman42x has quit [Ping timeout: 252 seconds]
Rusty1 has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjcVD
the-kenny has joined #nixos
redj has quit [Read error: Connection reset by peer]
Qubasa has quit [Ping timeout: 252 seconds]
buffet has quit [Ping timeout: 246 seconds]
buffet has joined #nixos
pie___ has quit [Ping timeout: 252 seconds]
mexisme has joined #nixos
selfsymmetric-mu has joined #nixos
sestrella has joined #nixos
sestrella has quit [Client Quit]
<selfsymmetric-mu>
I am trying to build `emacs-jupyter`. It requires `emacs-zmq` which requires `libzmq`. Ideally I would like to be able to build an Emacs that has this available. If that's not possible I would like to simply provide `libzmq` as a library in my `nix-shell`. How do I get this? I added `pkgs.libzmq` to my `buildInputs` but apparently that isn't available.
<jackdk>
packages.zeromq?
<jackdk>
er, s/packages/pkgs/
<selfsymmetric-mu>
I tried that. Didn't work for me.
<selfsymmetric-mu>
And when it's not available Emacs tries to build it. But then it tries to dump it in the Nix filesystem, thence getting booted since it's readonly.
<___mat>
hi all! i was trying to install the haskell make replacement `shake` on macos using nix, but it seems like i cant find it available to my nix-env. https://nixos.org/nixos/packages.html#shake it shows up on this page, but i am unable to see it when i query for available packages in my env (`nix-env -qa 'shake.*'`)
<trh_>
hey folks -- a few minutes ago i attempted to boot up my usual NixOS running on a Macbook Pro, but was surprised to boot directly into MacOS. Usually I get a boot screen and can choose either a NixOS generation or to boot into MacOS.
<trh_>
I've had the system running for a couple months now and this hasn't happened before.
___mat has quit [Quit: Page closed]
<trh_>
All I did was boot into MacOS, run some audio software for a few hours, then shut down.
<trh_>
Anyone have an idea of what could cause this or how to fix it?
<trh_>
For context, if it's useful, I have NixOS installed to the same EFI partition as MacOS uses for booting; I was told that was safe to do, but this leads me to think perhaps MacOS overwrote something.
<trh_>
Would appreciate pointers to links or docs on this -- haven't foud anything useful so far
<sphalerite>
I'm interested in cross-compilation, so I may put some more effort into it at some point
<sphalerite>
I'll let you know if I get anywhere :)
<pbb>
thanks
hio has joined #nixos
mexisme has joined #nixos
dashkal has quit [Ping timeout: 268 seconds]
toppler` has quit [Ping timeout: 248 seconds]
toppler` has joined #nixos
<bahamas>
hello. is there a way to find the channel that corresponds to a particular commit? I have one Docker image that pins nixpkgs to a tar archive while another one uses nixpkgs-unstable. I need the 2 to match
<averell>
that is not really a function, unless you mean latest, many commits would start in master, become a release, still be the basis for the next one, etc.
<averell>
maybe i misunderstood, and you have a tar archive which is a certain channel at some point? anyway, you can do git branch --contains on the NixOS/nixpkgs-channels repo.
dashkal has joined #nixos
<bahamas>
averell: I have a base Docker image that uses nixpkgs-unstable. I have an image that inherits from that one which uses a pinned version of nixpkgs. this difference causes a version mismatch. currently, I removed the pinned version and am trying with latest in the second image. I'll see if that breaks the build
<etu>
bahamas: And the branches makes the channels
<bahamas>
etu: so a channel is a git branch
<etu>
bahamas: Kinda
goodwill has quit [Quit: ...]
<etu>
bahamas: A channel is a branch that have been approved by the tests for that channel
<bahamas>
etu: so all channels are git branches, but not all git branches are channels
wfranzini has joined #nixos
<etu>
bahamas: That commit is on master, and it's quite old so that commit exists in both 18.09 and 19.03. But it was also reverted by 1fa2503f9dba814eb23726a25642d2180ce791c3
<etu>
bahamas: Yeah, and not all channels are up to date with the branch because of tests :-)
<bahamas>
ok. hopefully the image will build with nixpkgs-latest instead of the old, pinned version. if not, I'm not sure what the best solution is
<bahamas>
is there a way to not copy the entire store when you add a channel?
nD5Xjz has quit [Ping timeout: 244 seconds]
<bahamas>
I mean when you run `update`
nD5Xjz has joined #nixos
<bahamas>
I guess it doesn't make sense, but I would like to not wait so long for the entire store to be copied locally
init_6 has joined #nixos
Makaveli7 has joined #nixos
Jetien_ has joined #nixos
nD5Xjz_ has joined #nixos
nD5Xjz has quit [Ping timeout: 268 seconds]
Jetien_ has quit [Ping timeout: 248 seconds]
nicknameos has quit [Ping timeout: 256 seconds]
<immae>
bahamas: can you be more precise about your question: You want to download all the nix expressions when you run update? If so, it’s already the case. You want to "upgrade something" (like your installed packages for instance) ? then you need to have a nix expression that builds that "something" and run it after the channel update. You want to download all the existing pre-built binaries corresponding to the
<immae>
channel ? Well I don’t think so :)
marusich has quit [Remote host closed the connection]
rtjure has quit [Ping timeout: 246 seconds]
anon1 has joined #nixos
<symphorien>
a question for nix gurus: is it possible to filter a list of derivations by whether the host platform is within meta.platforms ?
<bahamas>
immae: here's my motivation. I'm building a Docker image with nix and it's downloading the entire nix store. I was wondering if there was a way to download nix expressions only when they're needed
justanotheruser has quit [Ping timeout: 246 seconds]
sigmundv_ has joined #nixos
<juhe>
yep, was there couple of times, I plan to use RPi's and there's one kirkwood HW around, but didn't see the initial image, so that would be harder nut.
dansho has quit [Ping timeout: 252 seconds]
<juhe>
I'd like to have all those arms around on nixos, it saves so much maintenance work...
mexisme has quit [Ping timeout: 258 seconds]
<genesis>
if i've time i'd be interesting to try an old tomtom device support
mexisme has joined #nixos
<juhe>
if you can get at u-boot (or any boot loader used by embedded device) and have kernel with support for all (or most of the) devices, it's relatively easy
<pbb>
has anyone tried nixos on the nintendo switch?
<pbb>
I guess it would be a good platform to maintain an installation with so many patches necessary
<genesis>
juhe : the difficulty is that support for some hw on his is too old to use recent software this builders are made for
<genesis>
i've to port old linux driver to recent one, specially the screen
<genesis>
i'm specialised in embedded linux
<genesis>
but most of the time, i've worked with openembedded or buildroot
<genesis>
and since it's a gps with closed drivers and datasheet ;)
<{^_^}>
[nixpkgs] @vcunat pushed 15 commits to release-19.03: https://git.io/fjciF
<{^_^}>
[nixpkgs] @vcunat pushed 43 commits to staging-19.03: https://git.io/fjcib
<juhe>
genesis: oh, yes, legacy stuff... Yes, the maintenance is a burden and porting requires to understand hw, old state of sw, the new state of sw and the difference(s) in order to do the port... And it consumes time. But is *fun* :).
<{^_^}>
hydra#588 (by ledettwy, 38 weeks ago, open): Consumable system features
<andi->
ahh just saw you commented on that..
ubert has joined #nixos
<Taneb>
The issue is at work we've got some huge builds that need >50GB of RAM, and if someone happens to be running one locally...
<andi->
add a required feature and only allow N numbers of jobs on those machines?
<andi->
thats probably the only reasonable way right now
<yorick>
cross compiling arm doesn't work *well* exactly
<yorick>
but it may work
knupfer has quit [Ping timeout: 252 seconds]
Anton-Latukha has joined #nixos
fishi0x01 has quit [Ping timeout: 250 seconds]
n_db has quit [Remote host closed the connection]
<Taneb>
andi-: I think to do that we'd also need support in Nix for consumable system features, which I don't think exists even in a PR
<Taneb>
Unrelated idea: function which reads a .gitignore and gives you a source filter
lovesegfault has joined #nixos
<das-g[m]1>
When writing a shell.nix with pkgs.mkShell, is there a way to specify a default command to run (e.g. my IDE) unless overridden by `nix-shell --command` or `nix-shell --run`?
<yorick>
das-g[m]1: there's shellHook but it always gets run
<Taneb>
andi-: neat!
ricekrispie2 has joined #nixos
<das-g[m]1>
Is shellHook documented anywhere? I see it being used in examples on https://nixos.org/nixpkgs/manual/, but can't see any specification there on what it does.
<BuGless>
I'm trying to get node 11.15 running on nixos (and have a minimal installation). So far I have nixos running, but the question is, how do I get node 11.15 in there?
TaoHansen[m] has joined #nixos
<TaoHansen[m]>
I'm having trouble running a binary from the command-line. Is there something I should be prepending that is NixOS specific?
nD5Xjz_ has quit [Ping timeout: 244 seconds]
<jasongrossman>
TaoHansen[m]: If it has any dependencies then it needs to be made into a Nix package.
fishi0x01 has quit [Ping timeout: 268 seconds]
johanot has quit [Ping timeout: 250 seconds]
nD5Xjz has joined #nixos
<jasongrossman>
(That's not quite strictly true - e.g. you can run a shell script that depends on bash without packaging it - but it is generally true and it's what I'm guessing your problem is.)
<TaoHansen[m]>
jasongrossman: It's a game binary usually run under Steam. Steam itself can run these games just fine but I need to do it manually because of a gamebreaking bug that hasn't been patched yet.
johanot has joined #nixos
<{^_^}>
[nixpkgs] @flokli opened pull request #61133 → shairport-sync: set sysconfdir to /etc/ → https://git.io/fjcPb
<jasongrossman>
TaoHansen[m]: Does steam-run help?
<xorAxAx>
hmm, #nixops only has 9 people currently, is that one the right channel and how does everybody else here deploy their nixos cloud instance? :-)
<TaoHansen[m]>
Let me check.
<jasongrossman>
xorAxAx: I deploy my cloud instances using NixOS (without NixOps), using ssh. I believe that's common.
<TaoHansen[m]>
jasongrossman: Ah you mean this `steam steam://rungameid/859580`?
<makefu>
a lot of people are doing it similar to krops (scp sources to the host, perform a rebuild switch)
<jasongrossman>
TaoHansen[m]: Sorry - I don't know what that is. I meant this (from the NixOS wiki):
<TaoHansen[m]>
I need it to execute whatever dependency logic it has for a binary outside of itself
<xorAxAx>
makefu, ah, interesting
<jasongrossman>
steam-run is a helper that can be used to run external programs in the steam FHS environment. This means that some third-party software, mostly games, can be run easily on NixOS using steam-run. Before trying other solutions, try starting your game this way, where start-game.sh is the script used to start the game on Linux.
<jasongrossman>
$ steam-run ./start-game.sh
<jasongrossman>
<TaoHansen[m]>
jasongrossman: Ah gotcha. I don't have it. I'll check the repos
<xorAxAx>
makefu, which backends does krops have?
<jasongrossman>
TaoHansen[m]: :-) Hope it works.
<makefu>
xorAxAx: krops uses ssh
<makefu>
essentially it creates scripts which use ssh and rsync to copy sources and run the rebuild
nD5Xjz has quit [Ping timeout: 250 seconds]
<xorAxAx>
ah
nD5Xjz has joined #nixos
<xorAxAx>
what does the page mean by "Precise versioning for every machine.", makefu?
<jasongrossman>
Yes, that's exactly what I do, only manually.
<TaoHansen[m]>
jasongrossman: Me too. 🙂 Thanks for the help.
nschoe has quit [Ping timeout: 258 seconds]
<makefu>
xorAxAx: you normally pin the exact nixpkgs version you want to deploy instead of following a channel (even though this is of course still possible)
<xorAxAx>
ah
nD5Xjz_ has joined #nixos
nD5Xjz has quit [Ping timeout: 246 seconds]
<xorAxAx>
i think you can get the same result with nixops if you use a defexpr that pins, right, makefu?
<genesis>
TaoHansen[m] : i fear nix-locate only list library you already have, donno, i dont play much with it
nD5Xjz_ has quit [Ping timeout: 258 seconds]
<genesis>
"* this tool can only find packages which are built by hydra, because only those packages
<genesis>
will have file listings that are indexed by nix-index
<genesis>
"
<BuGless>
I now have a nixos installation of following nixos-19.03-small which has a disk-footprint of about 1.6GB. How would I go about remove packages from this installation to reduce the footprint to somewhere around 200MB?
<infinisil>
(Also using nix-index, but no need to install it and create the index)
ottidmes has joined #nixos
<BuGless>
xorAxAx: systemctl disable udisks2
<BuGless>
Failed to disable unit: File /etc/systemd/system/udisks2.service: Read-only file system
<BuGless>
Any clues?
numerobis has left #nixos ["WeeChat 2.4"]
<xorAxAx>
BuGless, man configuration.nix, search for udisks2
Elon_Satoshi has quit [Read error: Connection reset by peer]
<genesis>
infinisil : i keep that for nix-locate failed to, else it would pollulate the channel.
domogled has quit [Ping timeout: 250 seconds]
<genesis>
he would have discover that with times ;)
fishi0x01 has joined #nixos
hamishmack has joined #nixos
Elon_Satoshi has joined #nixos
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
<BuGless>
xorAxAx: Thanks, that got me down from 1.6GB to 1.4GB. But I still have a long way to go from there to 200MB. In Debian I can do "dpkg -l" and then apt-get purge ... packages to get them removed. Is there something equivalent in nixos which allows me to trim all the excess fat?
<ottidmes>
I plan to buy a new laptop today, I have reduced my choices to two laptops, they have almost identical specs except for the display and the size of the laptop. The biggest choice factor is whether I am better of with a 3200x1800 monitor (13.3" with big bezels) or a 1920x1080 monitor (15.6" with almost no bezels). I hope people here have experience with these sizes on laptops running NixOS and could help me
<ottidmes>
decide. I like looks / thin bezels / bigger screen of the 15.6" but the like the higher resolution of the other, on paper, but I have no experience with such a hiDPI display on a laptop and wonder if that will problematic in practice with Linux software and its far from prefect hiDPI support
<BuGless>
ottidmes: I would decide based on weight/mass of the laptop. Pick the lighter one.
<Synthetica>
ottidmes: Wayland has decent hidpi support, even for mixed scales
<Synthetica>
What enviornment were you looking to run?
<gchristensen>
X11 handles hidpi quite well too as long as you don't have an external display which isn't hidpi, afaik
<Synthetica>
In my case, I went for a 14 inch 1080p display because of scaling concerns, but this was ~2 years ago
Elon_Satoshi has quit [Read error: Connection reset by peer]
<ottidmes>
BuGless: the 13.3" is 1.4kg and the 15.6 is 1.7kg my current is an old 13.3 of 1.7kg which is perfectly fine for me, so I decided that that will not be the determining factor
<avn>
Folks, does anyone have working example of docker+zfs on nixos?
<ottidmes>
Synthetica: software development environment mostly, to work in remote locations with at most one display connected to it if available
jasongrossman has joined #nixos
<Synthetica>
(I was thinking DE/WM)
lovesegfault has quit [Quit: WeeChat 2.4]
<ottidmes>
Synthetica: ah like so, right now I am happy with bspwm
<ottidmes>
Synthetica: I am afraid of using wayland because I don't like the prospect of the potential problems it could bring, and I would have to change to the i3 variant that works on wayland, while I would prefer to say with bspwm
<BuGless>
error: The unique option `fileSystems./.device' is defined multiple times, in `/etc/nixos/configuration.nix' and `/root/.nix-defexpr/channels/nixpkgs/nixos/modules/virtualisation/google-compute-config.nix'
<ottidmes>
Yaniel: do they simply not scale or are they scaled like some applications on Win10 are scaled just by zooming in (which makes it less sharp, but bigger)?
<BuGless>
Is there a way to override a config definition that already exists in an imported config file?
<{^_^}>
[nix] @grahamc merged pull request #2594 → installer: update macOS version check to 10.12.2 → https://git.io/fhvGk
<yorick>
Synthetica: way-cooler is way less complete
<yorick>
it probably has a superset of sway bugs
<ottidmes>
BuGless: mkForce probably?
<yorick>
ottidmes: I encourage you to run wayland and report all the bugs you find
silver has joined #nixos
<yorick>
but there will be a lot
<Synthetica>
yorick: I'm running sway as my daily driver and not encountering too many bugs, am I doing something wrong? 😅
<yorick>
Synthetica: recently I'm down to a few firefox crashes a week instead of a few a day, and my terminal does not get blurry if I move it to the hidpi screen before disconnecting the external ones, and it mostly doesn't crash if you never use floating windows or the scratchpad
<xorAxAx>
BuGless, good question, i am not sure
woffs has quit [Quit: Gateway shutdown]
<xorAxAx>
BuGless, you can try nix-du
<yorick>
Synthetica: firefox is still glitching all over the place but I'm told that might be fixed with mesa 19
<yorick>
(which is a mass rebuild)
pie___ has joined #nixos
<Synthetica>
yorick: Firefox on Xwayland or native?
<yorick>
native
<yorick>
xwayland is not really usable with scaling
<ottidmes>
if only the 15.6" laptop had the 3200x1800 resolution as well, then I would just have chosen that one, its only 4cm larger than my current 13.3" laptop, my biggest fear is finding it too small (definitely the case right now with 1366x768) and regretting not going with the 3200x1800 13.3" instead, but if in practice I only get scaling issues without much gain, than that fear is unwarrented, I simply have no
<ottidmes>
usage experience with such displays on laptops
woffs has joined #nixos
knupfer has quit [Remote host closed the connection]
orivej has joined #nixos
<deadloko>
When i'm trying to launch qtcreator through nix-shell --pure i'm getting error Could not initialize GLX, any help?
knupfer has joined #nixos
<yorick>
deadloko: you need to keep the LD_LIBRARY_PATH entries that point it to /run/opengl-driver/lib
<Baughn>
ottidmes: HiDPI used to be badly supported. It works much better now.
<Baughn>
You're going to find that the VTY has ridiculously small text, but that's actually configurable.
<yorick>
I think linux 5 might have fixed that
<Baughn>
(Also, I just acquired a 4k 13.3" laptop as well. This should be interesting.)
<deadloko>
yorick: thanks!
<Baughn>
yorick: Can't use Linux 5 yet, so meh. It doesn't work with ZFS.
<gchristensen>
that is odd, I'm on 5.0.9 and ZFS
<yorick>
Baughn: it should
<Baughn>
Unstable ZFS? I don't think there was a release since that kerfloofle.
<yorick>
Baughn: there was
<gchristensen>
no, 0.7.13
<yorick>
0.7.13 should work just fine, 0.8.0-rc4 works on linux 5.0
<ottidmes>
I guess I will just go with the 13.3" although it being less nice looks wise, its better spec wise, and hiDPI is the new normal, so there is enough attention to it to work any issues out in due time
fishi0x01 has joined #nixos
<yorick>
I'm running 0.7.13 on 5.0.1, apparently
<gchristensen>
ottidmes: my one caution to you is once you go to hidpi, it is hard to use non-hidpi again :P
<{^_^}>
[nixpkgs] @rnhmjoj opened pull request #61138 → Avoid collisions between essential packages → https://git.io/fjc1N
<xorAxAx>
whats the easiest way to add a command to configuration.nix that is run after/while -rebuild switch runs?
<gchristensen>
a systemd service is the usual right answer
<andi->
xorAxAx: system.activationScripts
<gchristensen>
what do you want to do?
<xorAxAx>
andi-, thanks
<gchristensen>
:/
<xorAxAx>
run "nix-channel remove"
<gchristensen>
what for?
<andi->
xorAxAx: cosnider an systemd unit.. it is the saner way in the long term.. probably
<andi->
if it is something that can't be done otherwise my alternative
<xorAxAx>
to remove the nixos-infect-authored nixos channel from a nixops machine
<gchristensen>
activation scripts should not contain anything which could be a systemd service
<xorAxAx>
i have created systemd services already on my other nixos machine
<xorAxAx>
so yes, it wouldnt fit my usecase
abcrawf has quit [Remote host closed the connection]
rauno has joined #nixos
<gchristensen>
andi-: do you know if we have any handy documentation on how to write activation scripts in such a way that your system still boots?
srid has joined #nixos
cfricke has quit [Ping timeout: 258 seconds]
<andi->
gchristensen: Not really, I usually reduce mine to a single line that is calling another scripts that does the entire thing.. Means I can do things like exit, worry less about syntax errors, …
<gchristensen>
https://nixos.org/nixos/options.html#system.activationscripts has some info but doesn't include (maybe that is the place to fix it) a warning that it must not assume networking
<{^_^}>
[nixpkgs] @adisbladis opened pull request #61139 → python-packages: Set pythonPackages attr in pythonPackages scope → https://git.io/fjc1j
<ottidmes>
Yaniel: I am going for the 15.6" full HD after all, because I will end using it on its own most of the time, so than a bigger screen (size not resolution) is preferred and I will need to take it with me to customers, so it looking much nicer is also preferrable, so I will not be using a hiDPI display after all
gagbo has joined #nixos
<gchristensen>
do you travel a lot?
fishi0x01 has quit [Ping timeout: 246 seconds]
fishi0x01 has joined #nixos
iyzsong has quit [Ping timeout: 252 seconds]
<ottidmes>
gchristensen: not really, only by car, but I assume you mean by train/plain, i.e. long distances, then no
<gchristensen>
ah ok
Twey has joined #nixos
<gchristensen>
I used to be a die-hard 15" laptop fan, but when I started taking planes and trains a lot it became a real drag
<ottidmes>
its also only 4cm wider than my current laptop which was comfortable in the train when I still went with the train a lot, so I doubt that increase in size will be a problem (due to the small bezel the increase in size is less so)
<Yaniel>
I used to be a die-hard 13" laptop fan, but then my bearings got rusty
<yorick>
gchristensen: does it deploy instantly?
<gchristensen>
no
<gchristensen>
I wish I could make the deploy log public, but it sometimes accidentally includes secrets :(
<Yaniel>
more seriously though, I discovered that the actual size of 15" laptops was roughly the same as of my old 13" one
<Yaniel>
because bezels had shrunk
<Synthetica>
gchristensen: It feels like bad practice to keep the config and the code in the same repo tbh
<Synthetica>
I did that once and I still feel dirty for it
<ottidmes>
anybody hosting wiki.js on NixOS? I am thinking of using it work the documentation of a small team project
<attente[m]1>
hi, i did an upgrade this morning and now apparmor is failing to load
toppler` has joined #nixos
<{^_^}>
[nixpkgs] @lheckemann pushed commit from @thoughtpolice to release-19.03 « linuxPackages.perf: add libopcodes as a buildInput »: https://git.io/fjcDd
<attente[m]1>
was the new 5.1.0 kernel not built with apparmor compatibility?
blackriversoftwa has joined #nixos
<{^_^}>
[nixpkgs] @volth opened pull request #61141 → perl-packages: Set perlPackages attr in perlPackages scope → https://git.io/fjcDb
<attente[m]1>
tbh i don't even really know why apparmor is enabled by default
WhatisRT has joined #nixos
cfricke has joined #nixos
<WhatisRT>
Hi! Is there any good way to install all dependencies of a package? Specifically, say I have a local package pkg, and I launch a nix shell with `nix-shell -A pkg`. How do I get access to things in that shell from the outside?
<bahamas>
WhatisRT: what do you mean by outside and why do you want that?
<WhatisRT>
(my issue is that my editor needs to run a command that only works inside the nix shell, but I don't want to start it from the nix shell)
<bahamas>
I see
sigmundv_ has joined #nixos
<Synthetica>
WhatisRT: Why not? That honestly seems like the easiest approach to this problem
<WhatisRT>
Synthetica because I have multiple different such projects
nD5Xjz has quit [Ping timeout: 255 seconds]
orivej has joined #nixos
<qyliss^work>
You could configure your editor to run shell commands as nix-shell --run "$command"
orivej_ has joined #nixos
<{^_^}>
[cabal2nix] @peti pushed to master « git-annex: improve the post-process hook to install the completion files »: https://git.io/fjcyf
orivej has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjcyJ
nD5Xjz has joined #nixos
orivej_ has quit [Read error: Connection reset by peer]
<WhatisRT>
qyliss^work that doesn't work for some reason. It seems to want to make a streaming connection to that process which fails
<{^_^}>
[nixpkgs] @adisbladis opened pull request #61142 → treewide: Make all usage of python2Packages explicit → https://git.io/fjcyI
orivej has joined #nixos
<WhatisRT>
Maybe for a more specific issue, the only thing that I seem to be missing are a bunch of libraries for idris. The documentation tells me that `nix-env -iE 'pkgs: pkgs.idrisPackages.with-packages (with pkgs.idrisPackages; [ contrib ])'` should work, but I get the error `error: attribute 'idrisPackages' missing, at (string):1:7`
<WhatisRT>
I'm not really proficient in the nix language, so I don't know what I should do to fix this. I only know that in the .nix files, idrisPackages works just fine
<yorick>
what does nix-build --dry-run do?
<WhatisRT>
it takes a second do finish and produces no output
<yorick>
oh, that was an unrelated question :P
<yorick>
I just don't know what that does
<WhatisRT>
ah :D
<infinisil>
WhatisRT: What does nix-end -qAp idrisPackages return?
<astal>
I downloaded the package i needed, which was installed in /nix/store/hash-package-name. Should i copy it from there to another directory now or should i work with it in the same place?
<infinisil>
Ugh nix-env..
<WhatisRT>
This gives the same error, actually
<infinisil>
Okay I'm a bit out of ideas
orivej_ has joined #nixos
<astal>
can anyone answer me?..
<infinisil>
WhatisRT:Oh no.. I think it might be this ugliness: nix-env -iE 'pkgs: (pkgs.nixos {}).idrisPackages.with-packages (with (pkgs.nixos {}).idrisPackages; [ contrib ])'
<goibhniu>
astal: can you give some more details? If you installed it, it should be on your path
<infinisil>
(I wrote the original nix-env idris install instructions, and because of how I set up nix-env, they worked for me)
<infinisil>
Guess it's time to rewrite as a nix file
orivej has quit [Ping timeout: 255 seconds]
o1lo01ol1o has joined #nixos
<WhatisRT>
infinisil: with nixos it just tells me that attribute nixos is missing
<WhatisRT>
and with nixpkgs it reinstalls a package and then prints `Abort trap: 6`
<deadloko>
xD
<infinisil>
The heeeellll
<infinisil>
(same for me, but still what the hell)
<infinisil>
Yeah just use a nix file
<aswanson>
pie_: you online? I finally got that issue with PYTHONPATH/reticulate resolved, but I'm wondering if there's enough friction here to open an issue with Nix
<astal>
goibhniu: i install package through nix-env (nix-env -i jira). after it, all binary files was installed into /nix/store/hash-jira-version
<deadloko>
another 30mins of qtcreator compilation, i think, i should finally use ccache >_<
init_6 has quit []
<WhatisRT>
With that file, I get error: value is a function while a list was expected
<astal>
goibhniu: but jira has a configuration files. this files located in /nix/store/hash-jira-version. should i move this files to other dir or i need work with it in store?
orivej_ has quit [Remote host closed the connection]
<gchristensen>
angerman: do you know Git's pickaxe?
rprije has joined #nixos
<angerman>
gchristensen: apparently not. Teach me (or do you even have a gold standard documnetation?)
<gchristensen>
«git log -S libffi pkgs/development/compilers/ghc» will search every commit and every patch for libffi, in that directory
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fjcyh
<deadloko>
angerman: like `git blame -L10,+1 ./pkgs/development/compilers/ghc/8.6.4.nix` this would return hash
detran has quit [Ping timeout: 252 seconds]
selfsymmetric-mu has quit [Remote host closed the connection]
<angerman>
deadloko: sure, but git blame is pretty useless on that very file :-/
<angerman>
I think it might have been 398ac54593d1a1300f63d3a6d4869f66126b9e86
<attente[m]1>
does anyone know why the new kernel 5.1.0 breaks apparmor?
<gchristensen>
attente[m]1: I've not heard that, but it definitely happens -- maybe you could try and look in the log for related changes, or even bisect the kernel?
<gchristensen>
attente[m]1: or, if you can author a NixOS test which shows the bug, I could do the bisection for you
<{^_^}>
[nixpkgs] @NeQuissimus pushed 4 commits to release-19.03: https://git.io/fjcSE
bbarker has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @NeQuissimus pushed 3 commits to release-18.09: https://git.io/fjcSz
<attente[m]1>
gchristensen: i think it could be reproduced directly with `boot.kernelPackages = pkgs.linuxPackages_latest;` and `security.apparmor.enable = true;`
aanderse-work has joined #nixos
<attente[m]1>
gchristensen: although i don't have that second line, i think apparmor is being enabled via lxd in my config
<deadloko>
how do i find what pkg produced nix/store path?
<gchristensen>
attente[m]1: and then what happens?
rauno has quit [Ping timeout: 252 seconds]
tschai[m] has quit [Ping timeout: 264 seconds]
tschai[m] has joined #nixos
sigmundv_ has quit [Ping timeout: 252 seconds]
<bricewge>
How can I generate a `cargoSha256` manually?
knupfer has joined #nixos
<bricewge>
I don't understand the docs when it says "Currently it [ cargoSha256] is obtained by inserting a fake checksum into the expression and building the package once. The correct checksum can be then take from the failed build. "
<catern>
bricewge: when you build a package with the wrong checksum, it still builds, and in the error message it tells you the real checksum
<bricewge>
sha256sum on `Cargo.toml` or `Cargo.lock` don't seems to cut it...
attero[m] has quit [Ping timeout: 264 seconds]
<Taneb>
`tofu bricewge
nD5Xjz_ has quit [Ping timeout: 246 seconds]
<bricewge>
catern:
attero[m] has joined #nixos
nD5Xjz has joined #nixos
<Taneb>
bricewge: set cargoSha256 to 52 0's
<bricewge>
That isn't waht I'm experiencing
<Taneb>
And try to build, it will go "That wasn't the right hash, this was the right hash", and tell you the right hash
<catern>
bricewge: what are you experiencing? I don't know the rust infrastructure so there might be some quirk about it
<bricewge>
I only get: "error: failed to select a version for the requirement `clap = "= 2.33.0"`"
<bricewge>
No sign of any hash in the output.
<catern>
hmm I'm not exactly sure how the Rust infrastructure works, but that sounds like you need to regenerate the version pinning or whatever; that sounds like an actual build failure due to version mismatches, not a hash problem
linarcx has quit [Quit: WeeChat 2.4]
<catern>
(more specificaly, I'm confident that's a Rust/cargo-level error message, not a Nix error message about hashes or anything)
linarcx has joined #nixos
MmeQuignon has quit [Quit: WeeChat 2.4]
<bricewge>
Ok, thanks catern, I'll dig deeper with cargo then.
<pbb>
bricewge: I ran into this with a project where the Cargo.toml file was changed without regenerating Cargo.lock
<pbb>
try running cargo on the project outside nix once
<arianvp>
worldofpeace: the lockscreen issue vanished. I think it was a transient error because is 'rebuild-switch'd from lightdm to gdm
<arianvp>
instead of rebooting
<bricewge>
pbb: I'm trying it. It seems quite involved just to bump a version.
<attente[m]1>
anyways, i think maybe the main problem here is "Cache read/write disabled: interface file missing. (Kernel needs AppArmor 2.4 compatibility patch.)"
<gchristensen>
but it is nice that it auto-pastebinsz
<attente[m]1>
ha, ok, good to know for future reference
sigmundv_ has joined #nixos
<attente[m]1>
gchristensen: can i report this issue somewhere?
<gchristensen>
yep! open up a github issue on nixos/nixpkgs
<gchristensen>
can you try it on the other kernel Nixpkgs provides and provide a report of which it works / which is broken?
<attente[m]1>
i can, sure
<attente[m]1>
thanks!
<gchristensen>
thank you!
<xorAxAx>
,locate aclocal
linarcx has quit [Quit: WeeChat 2.4]
rauno has joined #nixos
linarcx has joined #nixos
<symphorien>
xorAxAx: I guess you want autoreconfHook
<xorAxAx>
maybe
<xorAxAx>
i am currently looking at a project that pulls in 20-30 github repositories via bazel
<xorAxAx>
and i am inclined to do a non-sandboxed build
WhatisRT has quit [Remote host closed the connection]
mwilsoncoding has joined #nixos
<mwilsoncoding>
anyone in the Elixir/Erlang space able to tell why I'm getting an "Unchecked dependencies for environment prod: * telemetry ... * mime ... * plug_crypto ..." _in spite_ of the fact that I have added these packages into nixpkgs, listed them in the inputs to the package I'm building, `plug_1_8_0`, and included them in the `beamDeps`?
Shouou has joined #nixos
cfricke has quit [Ping timeout: 245 seconds]
Ariakenom has joined #nixos
Ariakenom has quit [Remote host closed the connection]
<mwilsoncoding>
I can even see in the output for `nix-env -if ./nixpkgs -A beamPackages.plug` (I rewrote the entry for `plug` to point to my updated v1.8.0) that it copies the packages into the `_build/prod/lib` directory prior to building, so I'm thinking I missed something, I just don't know where
bahamas has quit [Quit: Lost terminal]
<gchristensen>
thanks Synthetica!
* Synthetica
immediately adds "Rust developer" to their resume ;)
<{^_^}>
[nixpkgs] @dawidsowa opened pull request #61146 → kcc: init at 5.5.1 → https://git.io/fjc9V
Ariakenom has joined #nixos
<gchristensen>
you've earned it!
amir has quit [Quit: Have to disappear. Read you later guys!]
jabranham has joined #nixos
<gchristensen>
there is a really interesting way of using NixOps which I think could use bolstering
Ariakenom has quit [Read error: Connection reset by peer]
<gchristensen>
NixOps' deploy command has a `dry-activate` option where it is guaranteed to not actually execute the deploy -- just copy all the pieces to every server it will deploy to, and then tell you what would be restarted.
Ariakenom has joined #nixos
<gchristensen>
this means you can have an automatically triggered build-and-dry-activate step where you stage all the deployed artifacts on all the servers, and you can walk away and make coffee while this happens.
<aanderse-work>
i love dry-activate :)
<mwilsoncoding>
gchristensen: similar to `terraform plan`?
<gchristensen>
then when you're back you've saved the interesting part, authorize/confirm the deploy, and it takes only a few seconds: no agonizing minutes of waiting for artifacts to upload or download -- _only_ the actual activation phase
<aanderse-work>
sometimes i don't care what will be activated and i just do a --copy-only
<gchristensen>
mwilsoncoding: similar to `plan`, but everything already exists
<mwilsoncoding>
:+1:
<gchristensen>
mwilsoncoding: after dry-activate every system already has exactly what it needs to effectfully deploy
<Guest61>
Speaking of noob questions. Does anyone have any experience with PAM in nixos? I'm trying to get a pam module with text = "auth optional pam_exec.so expose_authtok /home/name/bin/script" to work, but I've had no luck... Any ideas on how I could debug it? Nothing seems to show up in journalctl or journalctl --user
<arguser>
Synthetica: that's good and maybe points to some missing nix-channels on my side
amir has joined #nixos
tdammers has joined #nixos
<arguser>
Synthetica: care to show relevant channels?
<lunik1>
I'm trying to use clamav for on-demand scanning, I've added clamav to my systemPackages but when I run freshclam I get "ERROR: Can't open/parse the config file /etc/clamav/freshclam.conf", and the configuration file does not exist
mwilsoncoding has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @andir pushed 2 commits to release-19.03: https://git.io/fjc7H
domogled1 is now known as domogled
pie__ has quit [Ping timeout: 252 seconds]
Ariakenom has joined #nixos
vidbina has quit [Ping timeout: 246 seconds]
MmeQuignon has joined #nixos
mexisme has joined #nixos
orivej has joined #nixos
dbmikus_ has joined #nixos
mexisme has quit [Ping timeout: 246 seconds]
judson has joined #nixos
Jetien_ has quit [Ping timeout: 250 seconds]
Ariakenom has quit [Remote host closed the connection]
sigmundv_ has joined #nixos
sondr3 has joined #nixos
srid has joined #nixos
erasmas has joined #nixos
linarcx has quit [Ping timeout: 268 seconds]
linarcx has joined #nixos
detran has quit [Ping timeout: 248 seconds]
ddellacosta has joined #nixos
ubert has quit [Remote host closed the connection]
<samrose>
trying to use a node2nix override for the sodium-native package. Although the package is listed in node-packages.nix, I still see error error: attribute 'sodium-native' missing, at override.nix:11:18
<sondr3>
are there any plans of deprecating Python2 in a similar vein to Arch/Fedora?
Shouou has quit [Ping timeout: 252 seconds]
orivej has quit [Ping timeout: 248 seconds]
sigmundv_ is now known as sigmundv
<Guest61>
Hi all, just checking in if anyone has experience with PAM in nixos. Still working on getting pam_exec.so to work
<lordcirth_>
Guest61, not in nixos specifically, but I've tinkered with PAM before. What problem are you having?
detran has quit [Ping timeout: 245 seconds]
<Guest61>
lordcirth_: I'm trying to get a script to activate every time I login. Maybe what would be most useful is helping me understand how to debug a pam module
<FRidh>
sondr3: it's a continuous process
<Guest61>
I've tried looking into journalctl, and journalctl --user and that doesn't seem to show anything
<FRidh>
as far as I know nothing in the base system depends on it
<Guest61>
I've also tried adding log=/var/log/logfile to the text, and it doesn't seem to get created, or updated
knupfer has quit [Ping timeout: 248 seconds]
<Guest61>
Basically I've just written something like: text = "auth optional pam_exec.so expose_authtok /home/name/bin/script"
<Guest61>
But it doesn't seem to work
detran has joined #nixos
ddellacosta has quit [Ping timeout: 252 seconds]
ddellacosta has joined #nixos
<lordcirth_>
Guest61, /var/log/auth.log?
FRidh has quit [Remote host closed the connection]
<Guest61>
Maybe I'm also misunderstanding pam. So long as there is a module in /etc/pam.d/ with the line above, it should be called at every login attempt, right?
<Guest61>
Like if I use xlock or i3lock, that custom module should be run
<infinisil>
Aw man, my xmonad is not starting anymore after resuming from suspend :/
halfbit has quit [Quit: WeeChat 2.4]
<matthewbauer>
kalbasit: github only shows that when you hit the "fork" button, not when you create a new repo
<infinisil>
Even restarting display-manager doesn't fix it
<infinisil>
Rebooting will probably fix it though
<kalbasit>
matthewbauer: yep, that's why I'm confused as it gives the impression that Nix's development is moving there
<kalbasit>
matthewbauer: which might be true giveng the PRs and issues there
<kalbasit>
not sure I'm comfortable with that however
<gchristensen>
kalbasit: it definitely isn't moving
<gchristensen>
though I can understand the concern around how it looks
srid has joined #nixos
<gchristensen>
I'll see about getting it moved
<kalbasit>
gchristensen: thank you for confirming. Not sure if it's too late to go the fork way now, but a README (or a repo description) might be warranted
<kalbasit>
thank you!
<sondr3>
I've stumbled across that repo before, what is the Nix flakes thing anyways?
<gchristensen>
kalbasit: by way of explanation, Flakes development is largely funded by Target, and Target has hired Tweag to do the work. it initially started as a private prototype, and then public, but while it probably should have, development didn't move to NixOS/nix -- we're meeting with Target tomorrow and we'll bring it up there
<kalbasit>
Oh I see how the fork did not get created now, it makes sense
<kalbasit>
and makes me feel a whole lot better, thank you for the explanation!
<gchristensen>
+1 I'm glad to hear it :)
<sphalerite>
Aren't private forks a thing too?
<kalbasit>
You can't fork a public repo as a private fork
<sphalerite>
oh ok
knupfer has joined #nixos
rprije has quit [Remote host closed the connection]
<infinisil>
Hmm, so if flakes is what is coming for Nix in the future, can the community give any feedback to that? Because it sure doesn't look like it
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
<infinisil>
Well I guess you can file issues, but eh
<simpson>
infinisil: IIUC part of the framing is that there *isn't* yet such a community, because everybody has to start off by forking nixpkgs.
<infinisil>
simpson: Ah you think the flakes thing requires major changes to nixpkgs?
<gchristensen>
I wish I could participate more in this chat, because I think I have answers, but I'm in a video call for the next 55min :(
hio has joined #nixos
sigmundv has quit [Remote host closed the connection]
psyanticy has quit [Quit: Connection closed for inactivity]
asa_ has joined #nixos
asa_ is now known as Guest43735
<lordcirth_>
infinisil, more like, flakes would allow people to make and share packages without needing to get them accepted into 'nixpkgs'
<lordcirth_>
Easier than now, that is
Guest43735 has quit [Client Quit]
fusion809 has quit [Remote host closed the connection]
<sondr3>
so it would replace NUR in a sense?
<__monty__>
It sounds a bit more like PPAs.
<andi->
NUR is more like a registry with a central listing. I think flakes are more like: here are some sources, the list of dependencies and how ot build it (a derivation)
<__monty__>
NUR is like AUR, flakes is like ppa?
<andi->
and by the looks of it you'd the be able to do something like `{ mypkg = builtins.flake https://my-repo/; }` but I only looked briefly a month ago
fusion809 has joined #nixos
tryingtoregister has joined #nixos
<sondr3>
I quickly skimmed the linked gist, but think I'm missing the larger picture here, what does this give me as a user that we don't have today?
<samrose>
ls
<samrose>
:-D
fishi0x01 has quit [Ping timeout: 246 seconds]
<__monty__>
sondr3: Convenience.
<__monty__>
And maybe it could be used to organize nixpkgs into pieces with their own maintainers? Like, I could imagine haskellPackages to be a flake?
dansho has joined #nixos
tryingtoregister has quit [Ping timeout: 256 seconds]
o1lo01ol1o has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
ris has joined #nixos
<sondr3>
Cool, that's kinda what I figured
o1lo01ol1o has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
<samueldr>
splitting chunks of nixpkgs is unlikely, because of the n-m dependencies
<samueldr>
like, nixpkgs to haskell, fine, but if you somehow have nixpkgs, haskell and python and ruby and lua and and and it's going to be an issue to deal with finding dependencies working
asar has joined #nixos
<samueldr>
well, splitting major chunks at least :)
<Baughn>
I could imagine making the git repository for an application into a flake, for instance.
<Baughn>
But I don't know about that name. It has a lot of bad connotations.
ddellacosta has quit [Ping timeout: 246 seconds]
capnp has quit [Remote host closed the connection]
Jetien_ has joined #nixos
ambro718 has joined #nixos
juhe has quit [Remote host closed the connection]
Ariakenom has joined #nixos
Ariakenom_ has joined #nixos
<Baughn>
Well, that's bike-shedding. Flakes look useful. How far along is the implementation?
<kuznero[m]>
Hi All! Is that a very bad idea to create a nix package that uses virtualenv under the hood (python), same applies for npm and yarn packages (I am talking about pgAdmin4).
<Yaniel>
there are docs for building python-based packages
<Yaniel>
using virtualenv manually might not be an outright terrible idea, but it certainly is unnecessary (not sure how bad it would actually be)
<LnL>
you can if you really want for some reason, but there's no advantage compared to the standard python infrastructure
<Ericson2314>
nix question, are fixed output derivations still in a user namepsace?
<kuznero[m]>
Problem is that pgAdmin4 depends on some packages that are not yet ported. And since I know very little about python infra in NixOS, I wanted to shortcut basically.
<Ericson2314>
I did a `nix log` on a fetch git private and I get my own user name
<kuznero[m]>
I am first trying to build it :laughing: I was always trying to create packages for something that has already all the dependencies in place. In the upstream stable channel. I just never tried building a package that depends on another package in my our source tree that is not in the upstream. Basically, starting from how to organize it and so on. If I remember different technology stacks (programming languages) are done
<kuznero[m]>
in ~different ways in nixpkgs. But any recommendations on that front would be great!
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
<ambro718>
Where do I file issues about the build farm?
<yorick>
ambro718: hydra? or ofborg?
<ambro718>
yorick: official nixos hydra
<gchristensen>
ambro718: what is the nature of th ebug?
<LnL>
the #nixos-dev channel is a good place for more org related things
<ambro718>
gchristensen: VirtualBox tests, VirtualBox 6.0 has a regression with SW-based virtualization causing nixos tests to fail (where VirtualBox runs in KVM) .
<ambro718>
with nested virt enabled, virtualbox successfully used HW-based accel and stuff works
rprije has quit [Remote host closed the connection]
<gchristensen>
interesting, I wonder if this means we should roll back the 6.0 update until they fix the regression
rprije has joined #nixos
<ambro718>
gchristensen: I would like to get it in in spite of this problem. It's a hard bug to pinpoint, and it's about SW-based virtualization, which I suppose doesn't get much investment.
<gchristensen>
a problem, though, is I would be surprised if we don't have users who depend on SW virt
<ambro718>
gchristensen: nobody would probably notice it if we didn't have these tests which must use SW-based virt, and a 32-bit guest because of that.
<ambro718>
gchristensen: but it's not a NixOS problem. Why should we hold it back in NixOS if it was released with this problem?
Anton-Latukha has quit [Quit: Leaving.]
<ambro718>
gchristensen: it's also a hard problem to track down. I don't know if I will spend much more effort on trying to reproduce it in a way that can be sent to the devs.
<gchristensen>
in a sense, it is a problem NixOS should be careful about, though -- just because a project ships doesn't mean we should release it to the users, as it needs to be fit for release -- part of what Hydra and for
<gchristensen>
(I'm not saying we can't do it, or we won't do it, but I think enabling nested virt is not the first thing I want to turn to)
<gchristensen>
for example, it isn't uncommon for us to bisect projects to identify where they introduced bugs -- I'm happy to help with this
Plotinus_ has joined #nixos
<gchristensen>
can you link me to the failure on Hydra?
<ambro718>
gchristensen: there's no failure in Hydra because it was never merged
<gchristensen>
ah, link to the PR?
<{^_^}>
[nixpkgs] @catern opened pull request #61158 → perlPackages.public-inbox: init at 1.1.0-pre1 → https://git.io/fjcbY
pie__ has quit [Remote host closed the connection]
<ambro718>
thanks for looking into it. Anyway I'm afraid it won't be easy to apply cherry-picking because of the patching.
<gchristensen>
"Did you open an issue in upstreams issue tracker about the broken software virtualization support?" - have you?
Plotinus has quit [Ping timeout: 248 seconds]
<makefu>
,locate libpopt.so.0
<{^_^}>
Found in packages: popt
<ambro718>
Not yet. I do not have enough information. I will spend some more time on it and try to open something useful.
<gchristensen>
okay
<gchristensen>
ok, cool, then if I'm understanding properly somewhere between 5.2.28 and 6.0.6 virtualbox broke software virtualisation, and a potential fix is enabling nested virt,
<camsbury>
hey guys - trying to set up a basic binary cache - and I can curl it at the /nix-cache-info path successfully
<camsbury>
but when I go to build, it says the following
xkapastel has joined #nixos
<camsbury>
warning: don't know how to open Nix store ...
<camsbury>
where the cache server is just running `nix-serve`
asar has joined #nixos
MmeQuignon has joined #nixos
stites has joined #nixos
mmercier has joined #nixos
<camsbury>
nvm - just needed to prefix it with http://
<worldofpeace>
arianvp: yeah, I'd expect some of the configuration for gdm to require a full reboot to work properly. though that's not really fun for nixos (i don't want to reboot almost ever :P )
mmercier has quit [Quit: mmercier]
o1lo01ol_ has joined #nixos
o1lo01ol1o has quit [Read error: Connection reset by peer]
orivej has quit [Ping timeout: 246 seconds]
Glider_IRC_ has joined #nixos
ilikeheaps has joined #nixos
dbmikus__ has joined #nixos
<mog>
if i want to run a module from a different channel than one im on what is prefered way to do this ? i want to do networking.networkmanager.enable = true; but from the unstable branch not the 18.09 im running
<catern>
things are supposed to be in nativeBuildInputs if I run binaries from them, right?
<catern>
what if I run binaries from them and also link against them?
<catern>
with strictDeps, pkg-config setuphook doesn't pick up things in nativeBuildInputs
<clever>
i'm still not sure how cross-compiling is meant to work for things like that
<catern>
(run binaries from them at build time, I mean)
<clever>
QT for example, has both a qmake binary (acts like cmake), and the libs, and you need to mix arches, and teach qmake where to find the target libs
<camsbury>
anybody ever dealt with hosting a binary-cache, and pushing binaries to it? trying to do `nix copy` but running into issues once it tries to POST
<gchristensen>
what kinds of issues?
<clever>
camsbury: nix-serve doesnt support upload, so all uploads would bypass it and use the ssh:// protocol
<gchristensen>
ah right
<camsbury>
ah
<camsbury>
ty clever
<catern>
clever: yeah, Xapian has this kind of thing too - a xapian-config binary you're supposed to run to find the location of the libraries
mckeanalicej has joined #nixos
<mckeanalicej>
does anyone know the recommended way to install agda on nixos?
<clever>
catern: and then you have the mess of getting the x86 version of the binary to know where to even find the arm libs
<catern>
clever: right, no idea how that's supposed to work
<clever>
catern: i guess since the cross-compiled libs are always x86->arm cross-compile (changing that changes the $out hash), you could just put native x86 binaries into its bin/ folder
<catern>
well, maybe it's a shell script for portability, but that's actually still a problem on Nix with patched shebangs, hm...
<manveru>
i want the `gems` (an attrset of derivations) to build
o1lo01ol_ has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
<manveru>
so i might call those `ruby_2_5-gems` or something
<manveru>
oh wait, there was an attribute that made hydra go deeper, right?
* manveru
digs
<clever>
manveru: you may also want to experiment with hydra-eval-jobs
<clever>
hydra-eval-jobs release.nix -I .
<manveru>
recurseIntoAttrs i think
<clever>
you dont need to configure hydra, just nix-build '<nixpkgs>' -A hydra
<manveru>
cool, thanks
<clever>
that lets you manually run the evaluator and see what it will discover
<colemickens>
Let me know if someone wants to make a push and get the mesa/libdrm meson changes in;
<colemickens>
We're coming up on the 19.1 release soon which will require the current open PRs to be merged first.
<colemickens>
I can donate CPU and cached built packages if there are changes needed and further testing....
<colemickens>
I really don't understand how efforts like this are supposed to play out in NixOS development, given the insane number of packages getting rebuilt and "staging".
o1lo01ol1o has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
__monty__ has quit [Quit: leaving]
o1lo01ol1o has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @lheckemann opened pull request #61162 → Kernel config: use PREEMPT_VOLUNTARY → https://git.io/fjcNS
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #44086 → paperless: add package and service → https://git.io/fNRw2
<{^_^}>
[nixpkgs] @worldofpeace pushed 10 commits to master: https://git.io/fjcN9
o1lo01ol1o has joined #nixos
<qyliss>
colemickens: I'm interested in mesa -- I have a bunch of in progress stuff for it and 19.1 should have a patch that came out of one I sent to the mesa-dev.
toppler` has quit [Ping timeout: 258 seconds]
<colemickens>
qyliss: what'd ya patch in mesa, out of curiosity?
<ambro718>
basically disabling a bunch of suspected-unnecessary code in glibc to improve the performance of finding libraries
<ambro718>
I've been building my system with a patch for the past 2 days and soon I will be able to test, but it's just the stuff I have, only a very small part of nixpkgs.
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
srid has joined #nixos
<ambro718>
though the fact it's already quite far (KDE, Chromium) probably means there is no breakage, at least for x86_64
o1lo01ol1o has quit [Ping timeout: 245 seconds]
esmerelda has quit [Ping timeout: 245 seconds]
johanot has quit [Quit: WeeChat 2.4]
ilikeheaps has quit [Read error: Connection reset by peer]
ilikeheaps has joined #nixos
o1lo01ol1o has joined #nixos
shibboleth has joined #nixos
<NemesisD>
is there any way to protect projects that start a nix shell from getting garbage collected?
<NemesisD>
meaning if that shell derivation changes, anything not used anymore should get gced but what's currently used
esmerelda has joined #nixos
ixxie has quit [Ping timeout: 246 seconds]
Plotinus_ has quit [Quit: night]
<yuken>
Huh. Should I setup my VMs for Nix with QEMU, or is nix-vm better?
<yuken>
I'm not sure how nix-vm stuff works.
nD5Xjz has quit [Ping timeout: 258 seconds]
<{^_^}>
[nixpkgs] @adisbladis merged pull request #61139 → python-packages: Set pythonPackages attr in pythonPackages scope → https://git.io/fjc1j
<tilpner>
yuken: 1. What do you need VMs for? 2. What is nix-vm?
<yuken>
tilpner, not 100% sure yet, nixos-build-vms
<yuken>
Planning on using a VM to test my mail server configuration & Matrix stuff.
nD5Xjz has joined #nixos
teoma has joined #nixos
<tilpner>
I can't help much, but maybe check <nixpkgs/nixos/tests> for inspiration on VM tests
<{^_^}>
[nixos-hardware] @JorisE opened pull request #107 → Added basic support for the t470s → https://git.io/fjcAW
zupo has joined #nixos
<ambro718>
tilpner: one possibility is nixops, which can work with local virtualbox or libvirtd/qemu
<ambro718>
I mean yuken^
Ariakenom_ has quit [Quit: Leaving]
<ambro718>
if you are anyway planning to deploy with nixops it's particularly simple (you can have an argument in the nix expression that you set different values for on nixops create command line)
nD5Xjz_ has joined #nixos
<ambro718>
based on which you set different options, like how to deploy
<NemesisD>
infinisil: that's weird. i added pkgs.glibc to the contents of an image with dockerTools.buildImage but it doesn't show up in /bin
<qyliss>
You want pkgs.glibc.bin
<qyliss>
It's not in the default output
nD5Xjz_ has joined #nixos
nD5Xjz has quit [Ping timeout: 248 seconds]
galaxie has quit [Ping timeout: 256 seconds]
rprije has quit [Remote host closed the connection]
rprije has joined #nixos
<gchristensen>
ambro718: I just about finished importing virtualbox's svn repo in to git, so I can do a properly automatic bisect
<NemesisD>
qyliss: thanks!
dbmikus has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @veprbl pushed to master « gcc48, gfortran48: add myself as maintainer »: https://git.io/fjcxv
<NemesisD>
i'm really confused why this derivation i wrote for conda isn't working. i've loaded it into a docker image, i'm looking at a file in /conda-env/bin/tic, when i run `strace tic` i get: execve("/conda-env/bin/tic", ["/conda-env/bin/tic"], 0x7ffe7c3cace0 /* 10 vars */) = -1 ENOENT (No such file or directory) i can't figure out what it is trying to open and can't find
<{^_^}>
[nixpkgs] @veprbl pushed to release-19.03 « gcc48, gfortran48: add myself as maintainer »: https://git.io/fjcxf
hamishmack has joined #nixos
thc202 has quit [Ping timeout: 248 seconds]
<NemesisD>
could it be an ELF thing?
mbrgm_ has joined #nixos
<OmnipotentEntity>
worldofpeace, sorry, I've been AFK for a bit. You suggested that I change some code, and I was just trying to confirm my understanding.
<NemesisD>
i noticed that nix seems to run patchELF on things. i'm wondering if somehow these executables were not subject to that patching. is there something i can to do my derivation to make sure those executables get patched?
<{^_^}>
#61107 (by OmnipotentEntity, 1 day ago, open): mp3cat init at 0.5