<{^_^}>
[nixpkgs] @jonringer pushed commit from @freezeboy to master « massif-visualizer: migrate to Qt5.15 »: https://git.io/Jk2iH
jbox has joined #nixos
<{^_^}>
[nixpkgs] @freezeboy opened pull request #104438 → kstars: migrate to Qt5.15 → https://git.io/Jk2Pf
<jared-w>
If I just run nixos-rebuild switch, it borks because it tries to build /nix/store/6ml2yqk6brk3fshsp3ydc3r09jbi62iv-gst-plugins-bad-1.18.0.drv which is broken. But if I go into the nix repl for the exact same nixos-unstable commit, the derivation corresponding to gst_all_1.gst-plugins-bad is /nix/store/q938zvyvn89y1bzdfqxgv9i847rgr0iy-gst-plugins-bad-1.18.0.drv which works
<{^_^}>
[nixpkgs] @lovesegfault opened pull request #104439 → pythonPackages.psycopg2cffi: init at 2.8.1 → https://git.io/Jk2Pt
<jared-w>
Am I going crazy? I'm not quite sure what could cause that type of situation.
zupo has joined #nixos
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm_ is now known as mbrgm
sss2 has joined #nixos
MasseR has quit [Ping timeout: 256 seconds]
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos
Soo_Slow has quit [Read error: Connection reset by peer]
MasseR has joined #nixos
Soo_Slow has joined #nixos
veleiro has quit [Read error: Connection reset by peer]
iH8c0ff33 has joined #nixos
veleiro has joined #nixos
<supersandro2000>
jared-w: maybe something is overwriting it somewhere?
werner291 has quit [Quit: werner291]
<supersandro2000>
are you using flakes or channel? Maybe you are not on the commit you thought you are
<jared-w>
I think I found the issue. If I did, I'm going to feel pretty silly about it.
<jared-w>
No channel, I had an update script that essentially wraps nixos-rebuild with -I to point directly to the pinned nixpkgs so that everything is guaranteed hermetic and I don't rely on the nix path
<jared-w>
Of course, that only works if the script dynamically gets the correct nixpkgs rather than statically doing it by substituting in with string interpolation...
iH8c0ff33 has quit [Ping timeout: 260 seconds]
bitmapper has quit [Quit: Connection closed for inactivity]
<jared-w>
Which is certainly an oversight considering I have fixed this exact bug before and somehow reintroduced it. Pffh
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.snowflake-connector-python: 2.3.5 -> 2.3.6 »: https://git.io/Jk2XV
mahogany has quit [Quit: Konversation terminated!]
hlisp has joined #nixos
<DigitalKiwi>
infinisil: i got matterhorn
<DigitalKiwi>
i don't even use it that's all evils ;_;
<DigitalKiwi>
still working on glirc...which i do use...i fixed it once already a week ago and now it broke again...
LnL has quit [Ping timeout: 260 seconds]
LnL has joined #nixos
hlisp has quit [Ping timeout: 264 seconds]
hlisp has joined #nixos
pushqrdx has quit [Ping timeout: 256 seconds]
hlisp has quit [Ping timeout: 256 seconds]
bbarker has joined #nixos
noudle has quit []
<jared-w>
ah yep, I am just dumb and was passing the wrong nixpkgs to nixos-rebuild. Yey
jonatanb has joined #nixos
<jared-w>
DigitalKiwi: have you ever thought about rewriting glirc in rust so that the haskell ecosystem churn in nixpkgs doesn't constantly break it? :p
justanotheruser has quit [Ping timeout: 240 seconds]
jonatanb has quit [Ping timeout: 240 seconds]
oxalica has joined #nixos
Soo_Slow has quit [Quit: Soo_Slow]
Maxdamantus has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @freezeboy opened pull request #104440 → mellowplayer: migrate to Qt5.15 → https://git.io/Jk2DB
Lord_of_Life_ has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @freezeboy opened pull request #104442 → heaptrack: migrate to Qt5.15 → https://git.io/Jk2D9
justanotheruser has joined #nixos
rajivr has joined #nixos
<DigitalKiwi>
jared-w: no i try not to think about rust
<DigitalKiwi>
haskell4lyfe
<DigitalKiwi>
i'm going to start sending peter an invoice every time he breaks me at this rate though ;_;
<DigitalKiwi>
for counseling
<rogerr>
if i don't want kbfsfuse and keybase to always be running, and only run when i run keybase-gui, i install pkgs.kbfs/keybase/keybase-gui then run kbfsfuse service & keybase service & keybase-gui. it kinda works but i can make it better right?
bbarker has quit [Remote host closed the connection]
bbarker has joined #nixos
ris has quit [Ping timeout: 256 seconds]
<DigitalKiwi>
copying path '/nix/store/ik69l2b5y2fkxh2fwdfmxi43gld5nhhc-config-value-0.8' from
<astro>
hi, I'm trying to make a font package, but I can't get it to work. currently I just copy the .otf file to $out/share/fonts/opentype. fc-cache doesn't seem to find it. is there more that I need to do?
<leo60228>
virt-manager is failing to build in my nixos config
<leo60228>
it's fine built using the cli
<leo60228>
this is the exact same nixpkgs
<leo60228>
i'm using flakes fwiw
<fresheyeball>
I have this 8.6.5.so: ELF section name out of range
<fresheyeball>
collect2: error: ld returned 1 exit status
<fresheyeball>
with code that I know builds
<leo60228>
(technically, not virt-manager, libosinfo and spice-gtk are both failing to build)
<fresheyeball>
leo60228: what is the error
proofofkeags has joined #nixos
<leo60228>
spice-gtk> meson.build:4:0: ERROR: Could not execute command "/build/spice-gtk-0.37/build-aux/git-version-gen /build/spice-gtk-0.37/.tarball-version".
<leo60228>
libosinfo> [0/110] Generating osinfo_enum_types.h with a meson_exe.py custom commandninja: fatal: posix_spawn: No such file or directory
<leo60228>
(there's no elaboration on either of these, btw)
<growpotkin>
Can someone merge my PR before it gets stale? It's dead simple, fixes issues with google-fonts crashing Thunderbird and Emacs: #103779
<{^_^}>
https://github.com/NixOS/nixpkgs/pull/103779 (by alexameen, 6 days ago, open): Separate AdobeBlank-Regular.ttf from google-fonts package, moved to google-fonts.adobeBlank
<growpotkin>
It already went through review, but it's getting buried
waleee-cl has quit [Quit: Connection closed for inactivity]
veleiro has quit [Remote host closed the connection]
justanotheruser has quit [Ping timeout: 240 seconds]
SanchayanM has joined #nixos
palo1 has joined #nixos
SanchayanMaity has quit [Ping timeout: 256 seconds]
palo has quit [Ping timeout: 256 seconds]
palo1 is now known as palo
iH8c0ff33 has joined #nixos
iH8c0ff33 has quit [Ping timeout: 256 seconds]
lordcirth has quit [Ping timeout: 260 seconds]
lordcirth has joined #nixos
lordcirth has quit [Remote host closed the connection]
<danderson>
hm, is there an easy way to write a string into a file in $out, within an existing derivation?
<danderson>
like the various writeText derivations, but without generating an extra derivation
<danderson>
Context is I'm making a pkg-config file for a library that doesn't provide its own, but requires one to link to it (it's complicated). I can easily interpolate the contents into a string, and I'd like to dump that string into $out/pkgconfig in postInstall
kalbasit has quit [Ping timeout: 256 seconds]
<siraben>
danderson: you could just use shell scripting in the buildPhase
iH8c0ff33 has joined #nixos
<danderson>
siraben: sure, but the escaping is a bit ick
<danderson>
I guess I can make it work tho
<siraben>
danderson: I think you might be able to just put it in a while in the same folder as the derivation and copy that into $out
<siraben>
s/while/file
<danderson>
that's what I was thinking, but I can't find the levers to do that
<danderson>
ah, passAsFile apparently
<danderson>
passAsFile of variable "foo" creates a fooPath envvar pointing to the file containing that var
<Theuni>
does anyone know of a (simple) way to include a package in an environment but only use specific binaries or filter out unwanted ones?
<Theuni>
I keep fighting against inetutils on MacOS because I need the telnet client but I don't want stuff like 'ping' to be in there because the setuid stuff doesn't work.
<danderson>
hm. From within a derivation builder, can I find out the absolute store path I'm building for?
<danderson>
pkgconfig files want an absolute path to lib directories and such. I can put the pkgconfig file in a separate derivation, but ideally I'd avoid that
<danderson>
oh, $out is that
<danderson>
oh but I can't get at $out from a nix var, bah
<siraben>
Theuni: try using `--pure` when you enter nix-shell?
<Theuni>
sorry, i'm talking about stuff I install into my user environment
<cole-h>
You could make an overlay that takes the previous package and copies `bin/telnet` to $out and nothing else
<siraben>
What would I use `install -Dm755` vs `cp` vs `ln` for `installPhase`?
<siraben>
s/What/When
<cole-h>
Personal preference, probably.
<cole-h>
I don't think I'd use `ln` though
<cole-h>
No particular reason why.
<siraben>
Hm
<siraben>
How would the symlink still be valid after the build finishes?
<danderson>
yesss I made it work
<cole-h>
Probably wouldn't
<cole-h>
But the reason my thing would (probably) work is because that package would be pulled into the story by string interpolation
<cole-h>
s/story/store/
<danderson>
the nix manual _seems_ to say that if you reference another derivation anywhere in a derivation's attrset, that sets up a dependency
<danderson>
but that part of the manual is very hard to follow, so... shrug
iH8c0ff33 has joined #nixos
<cole-h>
Yep.
<cole-h>
Might not work if it was the installPhase of some derivation trying to symlink from the build environment (because the build environment is ephemeral)
FRidh has joined #nixos
cole-h has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @FRidh pushed 233 commits to staging-next: https://git.io/JkacH
<typetetris>
How comes `nixpkgs.fetchurl` is allowed to access the network within its builder.sh. How does it escape the sandbox? Is it enough, that is has an `outputHash*` attribute so the sandbox will not have disconnected network?
Morfio has joined #nixos
domogled has joined #nixos
<{^_^}>
[nixpkgs] @FRidh opened pull request #104469 → python27Packages: no longer build and prefer → https://git.io/Jka4C
lordcirth__ has joined #nixos
lordcirth has quit [Remote host closed the connection]
<deadpixels>
Hi! I'm pretty new to nixOS and I have an issue with my network connection. I'm using a wired connection and directly after the first reboot everything worked fine, but when I booted up the computer this morning it can't find the network interface anymore. I'm haven't changed any network settings in configuration.nix. Does anyone have any advice on how to troubleshoot the issue?
nikivi has joined #nixos
<typetetris>
A network interface disappearing? Sounds like a kernel/hardware issue. Do you use predictable interface names? (If you don't know, what it is, you are probably using it.)
<{^_^}>
[nixpkgs] @FRidh pushed to master « Revert "Revert "cachix: drop the build from the release jobset"" »: https://git.io/JkaEG
<patagonicus>
deadpixels: That's not enabled by default, AFAIK. My computer also says it's not running, it's just using dhcpcd. And now I noticed that my configuration.nix doesn't even mention my interface.
sangoma has quit [Read error: Connection reset by peer]
<deadpixels>
ok, but it "should work" without it right?
<typetetris>
Using networkd is still considered experimental an 20.09 according to docs.
<deadpixels>
by it I mean networkd
<typetetris>
Yes, ofc.
<typetetris>
no
<typetetris>
No there is no networkd running by default on NixOS. Yes network should work. I mean my system is nixos 20.09 and I type here in chat :)
<typetetris>
I just use networkmanager and `nm-applet`.
<patagonicus>
If you didn't set anything in your configuration.nix, it should "just" get an address via DHCP, yes. Does `ip link` show that it's "state UP"?
<typetetris>
For that I have `networking.networkmanager.enable = true` in my config.
supersandro2000 has joined #nixos
mallox has joined #nixos
<deadpixels>
Ah I got it to work! In the default nixos configuration there are two outcommented lines reg networking: networking.useDHCP = false; and networking.interfaces... . I enabled those two lines for some obscure reason. WHen I commented them out and rebuilt it immediately found the network interface
<typetetris>
deadpixels: So root cause was that the interface name in `networking.interfaces.<whateverwashere>.useDHCP = true` didn't match your interface with the lines uncommented.
<typetetris>
deadpixels: You are now using `networking.useDHCP = true` which is the default.
<deadpixels>
exactly
<deadpixels>
the line is networking.interfaces.enp0s3 but my listed interface is actually called enp3s0. going to try to explicitly set that and see if it works
<patagonicus>
The option does say "Using this option is highly discouraged and also incompatible with networking.useNetworkd. Please use networking.interfaces.<name>.useDHCP instead and set this to false." So I think the "right" thing to do is setting it to false and setting the one for your interface to true.
<patagonicus>
My guess is that, for some reason, in the installation environment the interface name was different, so the generated configuration.nix was wrong. But that doesn't explain why it worked after your first reboot (maybe a warm reboot didn't fully reset it, but a cold reboot did?).
<deadpixels>
Yesterday I did quite a few full reinstalls, most likely it only worked during the installation process.
<deadpixels>
I'm quite new to this and had some issues getting EFI to work
<deadpixels>
setting networking.interfaces.epn3s0.useDHCP = true worked
<typetetris>
Maybe you copy and pasted the config from somewhere? The interface name should be stable on the same system, that was the whole point of the predictable interface names
<bpye>
Hmm, anyone got any thoughts on how I can improve my deployment experience for a couple physical machines I manage with nixos? currently I have my config in git and I use scp and ssh to update the config remotely. The plus of this is that it's simple and I can config autoupdate on the machines which is useful for my home server, the negative is
<bpye>
that it's kind of manual, just a bash script. I've tried nixops in the past but it never really worked for me, is there anything else I should give a shot?
<deadpixels>
Aaah yes that's true I did. I copied it from an earlier install!
<typetetris>
deadpixels: But from a different machine it has to be, maybe a VM?
seku has quit [Quit: Connection closed]
<deadpixels>
yep, I set it up on virtualbox first. And it wasn't until I was on the new machine that I removed the comment from those two lines... facepalm moment
<typetetris>
bpye: I know, you don't want to hear this, but try nixops again. Thats exacty its scenario.
<{^_^}>
[nixpkgs] @timokau opened pull request #104482 → maintainers: add rasking to the sage team → https://git.io/Jkaz1
nikivi has joined #nixos
joesventek has joined #nixos
Graypup_ has joined #nixos
vandenoever has joined #nixos
vandenoever has joined #nixos
<patagonicus>
deadpixels: Don't worry, I've done much stupider things. :D Once you get a NixOS system running, the great thing is that you can always figure out what change made something break (for example I'm currently looking for the change in nixpkgs that is breaking mpv for me, while having the system mpv pinned to an old version that works).
medvid has quit [Ping timeout: 246 seconds]
dstzd has joined #nixos
medvid has joined #nixos
<deadpixels>
yes I really like the explicit nature of it
<typetetris>
Never considered using something else, since I started using it. (Just sometimes an arch vm for stuff I am too lazy to package and which goes to /opt or /usr/local.)
<deadpixels>
now I just need to figure out how to setup a proper rust dev environment :D
orivej has joined #nixos
oyren[m] has joined #nixos
hnOsmium0001 has quit [Quit: Connection closed for inactivity]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<deadpixels>
by the way, any idea why the networking.interfaces configuration is not part of hardware-configuration.nix?
<deadpixels>
instead of in the configuration.nix file I mean
<patagonicus>
I was wondering the same earlier.
<patagonicus>
https://github.com/NixOS/nixpkgs/commit/5ee383ea8c31cd7c8489c2b076aac9c51f63b55c this added it and places it in configuration.nix. It doesn't say why there instead of hardware-configuration.nix. Maybe because it's something you'll likely change (if you're using static addresses) and thus you wouldn't want it to be overwritten by rerunning
<deadpixels>
I see, well I guess you can just extract that part of the configuration into a machine local module which your versioned configuration.nix depends onb
<thblt>
My very simple package update PR hasn't got any reaction in a week, anything I'm expected to fix so it can be processed, or should I just wait? https://github.com/NixOS/nixpkgs/pull/103797 Thanks!
<axelf4>
Hi, I am trying to create a Nix flake for developing a Rust crate and eventually putting it in an ARM container. Is edolstra/import-cargo the way to go?
<{^_^}>
[nixpkgs] @FRidh pushed to staging « python27Packages: no longer build and prefer »: https://git.io/JkaFs
<{^_^}>
[nixpkgs] @FRidh closed pull request #104469 → python27Packages: no longer build and prefer → https://git.io/Jka4C
thblt has joined #nixos
Lord_of_Life_ has joined #nixos
<thblt>
Emacs users: Emacs can technically show the source code of C functions that are exposed to elisp. NixOS emacs derivation already packages the Elisp source code along with Emacs itself, do you think it would make sense to bundle the C source as well?
lordcirth__ has quit [Remote host closed the connection]
_mecairS` has quit [Quit: Hi. If i ever be an Angel of Light to guide you, to lead you in every way, will you also think of me at night and day.. https://bit.ly/36blphN]
<Cadey>
fuzzypixelz: what do you mean? like using something such as DWM instead of a full desktop environment like KDE?
<pinpox>
Hi, I'd like to create some custom packages and store them in a git-repo or directory. E.g. some personal golang projects I'd like to be able to "just install". How do I set that up, so I can put those packages in my configuration.nix?
<pinpox>
Also stuff like personal scripts and such, that is not something that would be useful to submit to nixpkgs
FRidh has quit [Remote host closed the connection]
<typetetris>
nasirhm: Just add `networkmanager` to your systempackages or install them as a user `nix-env -iA nixos.networkmanager` assuming your on nixos.
<infinisil>
lejonet: How about changing the type of `protocols` to `listOf format.type`
<infinisil>
Then users are forced to use a list
<infinisil>
lejonet: But also, how it is currently done, if the user specifies any protocol, all the default ones will be removed, is that behavior intended?
<lejonet>
infinisil: no, that is what I'm trying to figure out to ensure doesn't happen
<lejonet>
and yeah, lets change the type to listOf
<infinisil>
lejonet: I see, I think all you need for that is to remove the mkDefault in the assignment of protocols
<lejonet>
infinisil: then I get a "settings.packetbeat.protocols is defined in several places" type of warning
<infinisil>
Full error?
<nasirhm>
typetetris: Where are systempackages localted ?
<lejonet>
One moment, I'll get you a pastebin of it
<typetetris>
nasirhm: Usually you have an `environment.systemPackages = [ ... ]; ` attribute in your `/etc/nixos/configuration.nix`. It would go in that list, thats already there.
<nasirhm>
__red__: I added NetworkManager in Environment packages and nmcli is installed, but when trying to connect to network, It works for a moment and then the network seems connected but ping returns : Network is Unreachable.
Psykobunny has joined #nixos
Psykobunny has quit [Excess Flood]
lordcirth__ has joined #nixos
flal_ has joined #nixos
neiluj has joined #nixos
neiluj has joined #nixos
neiluj has quit [Changing host]
<neiluj>
hello
<neiluj>
would it be possible to automate the update of packages from nixpkgs?
lordcirth has quit [Ping timeout: 260 seconds]
<neiluj>
like a bot that crawls the git sources and check for latest releases, updates hashes and checks the build with hydra
flal_ has quit [Remote host closed the connection]
<dminuoso>
neiluj: Absolutely, haskellPackages is built that way
<simpson>
So you're going to have to explain exactly *how* you want to further improve automation. nixpkgs-update produces PRs like https://github.com/NixOS/nixpkgs/pull/104505
<neiluj>
except you must specify which package to update and provide version numbers
<neiluj>
i was thinking of a bot working in the background
<dminuoso>
neiluj: Do you see my link above?
<neiluj>
and we could even automate the merging once tests pass
<neiluj>
dminuoso: yes, i was talking about the nixpkgs-update program
gueorgui has joined #nixos
<ryantm>
Most packages don't have tests and testing all reverse dependencies is computationally intensive
<simpson>
neiluj: We *do*, just in a different way from what you might expect; we automate bumping the channel after tests pass.
<gchristensen>
automatic merging is pretty scary
<Jezhehim[m]>
I'm trying to use `services.xserver.displayManager.session` to let me log in to a graphical session that is _just_ fullscreen emacs
<simpson>
I've seen systems where merging could be as simple as a single maintainer clicking a single button in order to approve PRs, and further, I think that we *have* that for some subsystems already?
<Jezhehim[m]>
Can anyone suggest what I'm doing wrong?
<dminuoso>
Jezhehim[m]: Mmm, you need a window manager at least?
<dminuoso>
Oh no yo udont
<neiluj>
gchristensen: i agree, we could maybe let a bot merge once some conditions are met in a pr (approvals from maintainers and pushers)
<neiluj>
that's what coqbot does
<dminuoso>
Jezhehim[m]: What is DISPLAY set at?
<__red__>
I've always been curious about coq and the idea of formal proofs - but never gone down that road
bbarker has quit [Remote host closed the connection]
<neiluj>
simpson: oh good, then maybe a thin layer around this tool with a bot crawling the sources and calling the nixpkgs-update on each program with the latest fetched version
<__red__>
nasirhm: that particular package area is a black-hole in my knowledge - sorry :-/
<Jezhehim[m]>
dminuoso: definitely have `DISPLAY=:0` (and `XDG_SESSION_TYPE=x11` fwiw)
<simpson>
neiluj: If it makes sense. We do this for Node.js; there's a script in nixpkgs which can be manually invoked in order to regenerate metadata.
<nasirhm>
__red__: No worries, I fixed it. Actually had to flush my wpa_supplicant configs.
<Jezhehim[m]>
I added `env > /tmp/emacs-env.sh` that start script
<Miyu-saki>
Any tips for nix copy closure to a nixops machine?
<Jezhehim[m]>
* I added `env > /tmp/emacs-env.sh` that start script to help with debugging
<Miyu-saki>
Right now, I decided for making a closure
<dminuoso>
Jezhehim[m]: Out of curiosity, if you use xterm as in the example documentation, does it work?
<Miyu-saki>
Closure size is too big for me to do that now though.
<nasirhm>
any ideas on controlling brightlight / laptop light in NixOS
<simpson>
neiluj: Note that we can't do this automatically because the Node.js ecosystem doesn't have a centralized metadata source which is robust enough to event-source or otherwise constantly scrape. Most language ecosystems don't. You may want to adjust your expectations for how fast a ports tree of this size should move.
<nasirhm>
<nasirhm "any ideas on controlling brightl"> (I used to use brightlight package in other distros)
<__red__>
what was the executable called in other distros?
<Jezhehim[m]>
dminuoso: good question. brb...
<nasirhm>
<__red__ "what was the executable called i"> `brightlight`
<neiluj>
simpson: sure, thanks for the links to these scripts :)
<nasirhm>
__red__: yep.
feezeboy has joined #nixos
feezeboy is now known as freezeboy
<dminuoso>
Jezhehim[m]: Oh. Are you running emacs as daemon?
sajin4 has joined #nixos
<freezeboy>
I have some troubles to enable a kernel module (I put boot.kernelModules = [ "v4l2loopback" ]) but it is not loaded and modprobe doesn't see it
<neiluj>
but maybe this would help to let a bot take care of executing those scripts and interacting with prs in the background
<__red__>
nasirhm: It's a *baby* program - shouldn't take a moment to make a package foir
<sajin4>
is there a clean/nice or even any way to add my own channel into /etc/nixos/configuration.nix. So that I don't have to manually add any channels or stuff but just drop my configuration.nix, rebuild and its done?
<nasirhm>
__red__: I would love to make a package out of it.
<__red__>
nasirhm: nasirhm: Welcome to NixOS - wanna contribute a package ;-)
<sajin4>
sure let me take a look
<__red__>
hheh, you should probably get vaugly comfortable with nix before you contribute
<__red__>
but I can show you how it's done
caef^ has joined #nixos
<nasirhm>
That'll be awesome __red__.
<__red__>
so I don't have a laptop here to test functionality - but I can talk you through how to test it quickly to make sure it works on hardware before we look at packaging it
<freezeboy>
nasirhm I missed the start of the discussion, what new package are you talking about ?
<__red__>
this may be verbose - wanna move to #nixos-brightlight ?
<adisbladis>
Miyu-saki: Did you by any chance copy this from me? :P
<__red__>
basically
<__red__>
nix-shell -p stdenv
<nasirhm>
__red__: Sure, let me join.
<__red__>
git clone rgr
<nasirhm>
freezeboy: It's `brightlight`
<freezeboy>
nasirhm a url ?
<Miyu-saki>
adisbladis: Most likely. :D
<adisbladis>
Miyu-saki: That's probably it then :P It's looks pretty much identical to my older exwm startup expressions :)
<adisbladis>
I like it. I think I should probably upstream the xsession stuff.
<Duponin>
Hi, I'm trying to write a simple systemd unit to start an Elixir application but I have no idea how to get path of `mix` binary (from package nixos.elixir). Someone have any hint please?
<{^_^}>
[nixpkgs] @xwvvvvwx opened pull request #104506 → pythonPackages.ledgerwallet: init at 0.1.2 → https://git.io/JkVEC
<freezeboy>
Duponin you have two choices: the path attribute where you can put your package or just ${pkgs.elixir}/bin/mix in the start comd
<freezeboy>
Duponin you have two choices: the path attribute where you can put your package or just ${pkgs.elixir}/bin/mix in the start command
<__red__>
now I just need to work out how to convince github to stop telling me to create a pull request for it since I have no intention of ever merging it
alp has joined #nixos
Morfio has quit [Quit: This computer has gone to sleep]
<cm2>
jtojnar: thanks again mate. i will RTFM now :)
<evils>
cm2: <3 jtojnar
<{^_^}>
jtojnar's karma got increased to 63
<cm2>
one last question - since i installed mercurial from unstable-small how do i get it back to stable eventually?
<cm2>
evils: he saved my ass BIG time
gueorgui has quit [Ping timeout: 256 seconds]
gueorgui has joined #nixos
<evils>
cm2: for installing from stable, either change the channel in the nix-env command to the channel you want, or maybe if you had a stable one installed, maybe just nix-env -e mercurial to remove the env one?
<aasg>
I've got a module option of type "attrsOf submodule ...". Is there a way for a default value inside the submodule to refer to another option value in the submodule?
iH8c0ff33 has quit [Ping timeout: 256 seconds]
vandenoever has quit [Read error: Connection reset by peer]
<taylskid_>
I am trying to use both `services.postgresql.ensure{Users,Databases}` and `servives.postgresql.initialScript` but only my changes to the ensure options seem to be getting picked up anywhere. Are these options mutually exclusive. The documentation doesn't indicate that or anything
alp has joined #nixos
<aasg>
infinisil (IRC): That worked nicely, thanks. Now to remember where else I needed something like that…
arinov_ has quit [Ping timeout: 256 seconds]
arinov has joined #nixos
Fare has joined #nixos
<bpye>
taylskid_ If you go look at the source for the PostgreSQL service initialScript is only used on first run whilst ensure* is always applied
<bpye>
No worries, I think for most uses initialScript probably isn't too helpful. It looks more suited for a container use case for example
<wouter_>
Hey, I have a question about Nix, specifically the attrsOf function: this function should merge all definitions into a joined attribute set. However: the set that is in default is not added to this list if we write to it somewhere else. Is there any way to always add the values in default to the list?
<wouter_>
Now we're using apply = x: {attributesHere} // x; but maybe there's a nicer way
<andrewrk>
howdy. I am hacking up proof-of-concept and trying to make it work on nixos - so this is not your typical support request. My question is: how could I determine the RPATH of Xlib.so on nixos programmatically?
<andrewrk>
for example I am typing this question into HexChat which is a gui application. so if I were to ldd hexchat ELF file it would reveal the RPATH of Xlib.so
<andrewrk>
what would be a robust way to programmatically determine this value, that is likely to work on everyone's nixos installations?
<simpson>
Why is Xlib being loaded dynamically? You could provide it as a runtime input, right?
<andrewrk>
you mean to prompt the user for it? the goal of the PoC is to make a static linux executable that works on all linux distributions (not only nixos) regardless of the dynamic linker path, system libc, or window manager, and open a window and show an opengl triangle
<andrewrk>
so far I have solved the dynamic linker problem and now I am solving the "figure out the .so paths of the window manager that is running" part
<simpson>
Hm. Because of how X11 works, you should be able to bring your own Xlib/XCB without issues; it's libGL that you'll need to always look up at runtime. The `libglvnd` library can help with that.
Morfio has quit [Quit: This computer has gone to sleep]
proofofkeags has joined #nixos
<andrewrk>
then the question becomes how to find libglvnd at runtime
<andrewrk>
it may not even be installed
<andrewrk>
also s/opengl/vulkan/
seku has joined #nixos
<simpson>
Oh! GL is *not* Vulkan; GL enjoys GLX, an X11 extension which allows for GL to be sent over the X11 protocol ("indirect rendering") but IIUC Vulkan refuses to do that, and you'll need something like EGL.
<simpson>
andrewrk: FWIW it seems like libglvnd can't be built statically, but XCB can. I don't actually *know* how libglvnd works, and you could probably crib its source to grok how to find libGL.so dynamically.
<zeorin>
Hey all. I'm trying to add a udev rule in my nixos configuration
<zeorin>
It needs to run a command. Do I have to provided the absolute path to it or can I just use the command name?
magma has quit [Client Quit]
<Jezhehim[m]>
dminuoso adisbladis Miyu-saki I ended up using `services.xserver.displayManager.sessionPackages` with a `.desktop` file I'd put together when using Arch and it works fine: https://gitlab.com/-/snippets/2041904
<dminuoso>
Jezhehim[m]: Can you file a bug report at the very least?
<dminuoso>
Jezhehim[m]: If the documented example doesn't work, then at the very least its a documentation bug.
<zeorin>
From looking at udev rules they generally provide an absolute path to the executable. How do I get the absolute path to the executable in my nix configuration, since it can change with any update?
<{^_^}>
[nixpkgs] @mweinelt pushed 3 commits to release-20.09: https://git.io/JkVbN
<rogerr>
if i don't want kbfsfuse and keybase to always be running, and only run when i run keybase-gui, i install pkgs.kbfs/keybase/keybase-gui then run kbfsfuse service & keybase service & keybase-gui. it kinda works but i can make it better right?
<leo60228>
rogerr: i don't use keybase, but from a bit of googling + prior experience with similar things, i believe you want to have keybase set up to use socket activation
alp has joined #nixos
<leo60228>
from what i can tell, keybase doesn't officially support it, but it should *work* if you can configure it in systemd
<cmcaine>
I'd like to compile a C++ project called Julia from slightly customised source. I've found this derivation that will build Julia 1.5.3 from the github sources, but I don't know how to use that to compile my local source tree.
<cmcaine>
I'd really appreciate someone giving me a hand getting this to work, or setting up an environment where I can just type `make` as I would on other distros.
FRidh has quit [Quit: Konversation terminated!]
deadpixels has quit [Ping timeout: 240 seconds]
vandenoever has quit [Read error: Connection reset by peer]
vandenoever has joined #nixos
<cizra>
cmcaine: I think the usual way of developing stuff is by using nix-shell.
<jdnixx[m]>
<cmcaine "I'd really appreciate someone gi"> couldn't you just do src = import ./yourlocalsource.tar.gz instead of fetchzip {}
<jdnixx[m]>
I'm not exactly sure how that works sorry lul wish I could help moar. I suck at packaging too
<jdnixx[m]>
but I imagine if all the dependecies & stuff are the same then that's the only part of hte derivation you'd really have to change
<jdnixx[m]>
my q: why can't you use `let` straight in the body of configuration.nix?
<jdnixx[m]>
`let something = stuff in environment.systemPackages =` doesn't work
<jdnixx[m]>
but `environment.systemPackages = let something = stuff in ` works
<leo60228>
for the latter, you can get into a development shell for julia with `nix-shell '<nixpkgs>' -A julia_15`
<leo60228>
that command will drop you into a shell with the environment necessary to build julia
Fare has joined #nixos
<leo60228>
the most important things will be the commands unpackPhase (extracts the julia source archive), patchPhase (applies nixos's julia patches), and buildPhase (builds julia with the correct flags)
beertoagunfight has joined #nixos
<cmcaine>
leo60228, julia_15 isn't in nixos 20 where I'm at yet, but I can just download the derivation file and use that, right?
<leo60228>
yes, though you'll need to do something like `nix-shell -E '(import <nixpkgs> {}).callPackage ./julia.nix {}'` so it can find its dependencies
<leo60228>
flakes atm probably aren't suitable for a beginner, but with them you can just `nix develop nixpkgs/master#julia_15`
<stites[m]>
haha, okay -- another option is I can reuse their deb?
<stites[m]>
is that better?
jjakob has quit [Read error: Connection reset by peer]
<rogerr>
leo60228 ty ill look
hl has quit [Ping timeout: 260 seconds]
sss2 has quit [Ping timeout: 240 seconds]
<rogerr>
home manager have any way to control systemd units? i guess i need to make kbfs and keybase socket activation systemd units for keybase-gui to activate on demand?
<leo60228>
update: i think i figured out my weird nix problems
<bqv>
I made a fresh one since the url's dead
<axelf4>
How would I use ociTools.buildContainer from a flake-based project?
<leo60228>
the nixFlakes package in 20.09 is missing a fix for a bug breaking extra-sandbox-paths, and my poorly-thought-out workaround (use sandbox-paths instead) breaks a lot of stucck
<leo60228>
rogerr: yes, iirc it's systemd.services and systemd.sockets though i could be wrong
<rogerr>
leo60228 do the kbfs and keybase sockets already exist like stites[m] says? what's that mean too? if i understand right i need kbfs and keybase systemd units that require sockets, but i dont need to do anything for keybase-gui because it already expects kbfsfuse and keybase services to be running?
<stites[m]>
turning these on will enable them at login. keybase-gui should just work without any additional configuration
<rogerr>
all i know is i want keybase-gui to start kbfsfuse and keybase services on demand, and quit them when keybase-gui isnt running anymore
<stites[m]>
oh!
<rogerr>
that's not what i want
<rogerr>
i only want them running when keybase-gui is
<andrewrk>
simpson, I think I have an idea for how to do it. It would look like having support for the various popular window managers, and detecting one of them running. For example on my system the program could detect that xfdesktop is running, and from that pid look up the ELF file which would show the .so paths for the various X11 libs as well as libGL.so.1 => /nix/store/q4pa2k9vhvzix7psvpvpph0hs4d3a70h-libglvnd-1.3.2/lib/libGL.so.1
<andrewrk>
so then having detected this, the program could dlopen that libGL.so.1 and do graphics
<andrewrk>
not sure how to do it for vulkan yet
<andrewrk>
another problem is that the xfdesktop that is running is actually a wrapper script, so I would need some trick to get the path to the actual ELF file
<{^_^}>
[nixpkgs] @dwoffinden opened pull request #104530 → dns-over-https: init at 2.2.2 → https://git.io/JkwvM
<stites[m]>
rogerr: I think you'll just want to write a shell script that triggers the same start scripts in the linked systemd units
<stites[m]>
overwrite the desktop of keybase-gui
axelf4 has quit [Remote host closed the connection]
<stites[m]>
actually, I see one of two options:
<stites[m]>
- just copy those services, don't enable them but have available. *.desktop can start these services and have a systemd timer just turn them off (I don't know how you'd gracefully shutdown on killing the process), or
<stites[m]>
- write a wrapper of keybase-gui to do approximately the same thing
beertoagunfight has quit [Remote host closed the connection]
<evils>
shapr: are you sure you're using byobu right, and are there any errors?
turbo_MaCk has quit [Ping timeout: 260 seconds]
<shapr>
evils: pretty sure I'm using it correctly. By default byobu should start and work like a nicely configured tmux. I can copy my byobu config from by ubuntu install and see if that helps.
hyper_ch5 has joined #nixos
beertoagunfight has joined #nixos
<beertoagunfight>
To answer my own question from before with regards to the systemd service, I changed the target to basic.target and it works.
<beertoagunfight>
So, I was attempting to write my first derivation and I was wondering if I have to re-write the installPhase for this Makefile? Is there some magic I'm missing? https://github.com/jeremija/unipicker/blob/master/Makefile
<evils>
shapr: `byobu-tmux` may do what you want, seems like `byobu` is supposed to default to `tmux`, which also needs to be installed, but it isn't using it? (i don't use byobu though)
<shapr>
oh, good idea!
<shapr>
evils: that works!
hyper_ch4 has quit [Ping timeout: 272 seconds]
<shapr>
hurrah!
<shapr>
evils: how did you think up that solution?
<shapr>
I have ~20 years experience with Linux, the first five or so as a sysadmin. Even so, switching from Debian based distros is challenging for me.
<evils>
shapr: `byobu --help`
<shapr>
So I'm trying to understand the heuristics
<shapr>
oh, so easy
<evils>
:D
leo60228_ has joined #nixos
__monty__ has quit [Quit: leaving]
hyper_ch5 has quit [Ping timeout: 272 seconds]
jonatanb has quit [Remote host closed the connection]
<beertoagunfight>
I'm currently stuck at `install: cannot create regular file '/usr/local/bin/unipicker': No such file or directory`. Is there a way to emulate a unix FS in the installPhase?
inkbottle has joined #nixos
leo60228 has quit [Ping timeout: 264 seconds]
bbarker has joined #nixos
leo60228 has joined #nixos
leo60228_ has quit [Ping timeout: 265 seconds]
<simpson>
beertoagunfight: Pastebin your derivation? But usually you'll want to get the build script's prefix (e.g. --prefix) set to $out.
inkbottle has quit [Ping timeout: 240 seconds]
turlando has quit [Read error: Connection reset by peer]
turlando_ has joined #nixos
sajin4 has quit [Ping timeout: 245 seconds]
<beertoagunfight>
simpson: https://pastebin.com/Hj526GiL - It does build for now, but I'm not sure if this is the right™ way to do it. For instance, the make phase succeeded even though I didn't specify python as a dependency.
<beertoagunfight>
(It's also broken, I guess my `cp`s are incorrect)
turlando_ has quit [Client Quit]
bqv has quit [Quit: WeeChat 2.9]
alp has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @zowoq opened pull request #104535 → Revert "haskellPackages.trifecta: dontCheck on darwin" → https://git.io/JkwTM
bqv has joined #nixos
<simpson>
beertoagunfight: You're certainly close. You'll want something like `mkdir -p $out/bin/`, followed by `cp unipicker $out/bin/`. You can think of $out as holding an entire (slice of) the FHS inside of it; if you need /bin or /share, just mkdir them.
bqv has quit [Client Quit]
bqv has joined #nixos
cosimone has joined #nixos
Boomerang has joined #nixos
<beertoagunfight>
Ah I see what you mean, looking at the Makefile, it looks like I simply have to set $DESTDIR to $out. I'm trying to figure out how I can do that.
<evils>
shapr: thanks, i think this means i figured out why my attempt at packaging `hollywood` didn't work
turlando has joined #nixos
aasg has quit [Quit: Bridge terminating on SIGTERM]
<supersandro2000>
or you use install which can do mkdir, mv and chmod in one
redmp has joined #nixos
<supersandro2000>
and version should not start with v in nixpkgs. Add the v to rev to fetch it correct.
<beertoagunfight>
Oh thanks simpson, I just figured that out a moment ago. Now it does assume the correct path for the build output, but the build fails with: install: cannot create regular file '/nix/store/zgpvvabz4rljqi1khycf787r4k5zdnp1-unipicker-v2.0.0/bin/unipicker': No such file or directory
<beertoagunfight>
supersandro2000 - Gotcha! I'll correct it.
<supersandro2000>
and the file wget wants to fetch should be fetched before and hashed
<supersandro2000>
fzf and xclip are probably something for a wrapper.
<beertoagunfight>
Okay, thanks for the pointers. I had a feeling that something doesn't feel right :-D
<beertoagunfight>
It's all coming together slowly, there's a lot to process lol.
aasg has quit [Quit: Bridge terminating on SIGTERM]
<beertoagunfight>
"use install which can do mkdir, mv and chmod in one" - Does that mean installPhase = "true"?
cosimone has quit [Quit: cosimone]
zebrag has quit [Quit: Konversation terminated!]
<evils>
beertoagunfight: they mean the `install` command
blackriversoftwa has quit [Ping timeout: 240 seconds]
sss2 has joined #nixos
<lejonet>
Okay, this is driving me nuts, I've just updated from 20.03 to 20.09 and I cant for the life of me get firefox to stop segfaulting, I've tried with nuking .mozilla, using the official binary, building from source, everything just segfaults and the only thing I can find with strace is that it gets a null pointer deref somewhere after reading prefs.js and user.js (it crashes regardless if those 2 exist or not)
<lejonet>
Anyone else seeing this behaviour with firefox 82.0.3?
mananamenos has quit [Read error: Connection reset by peer]
<xenophile>
Hello. So I have nixOS on my laptop and I really want to swtich to it on my main machine but the only thing preventing me is windscribe vpn cli isn't availible as a nix package. Is it fairly difficult to package something from a .deb or .rpm?
<ivan>
lejonet: did you re-test 20.03 to see if it's not just your computer coincidentally failing?
meh` has quit [Ping timeout: 246 seconds]
<lejonet>
ivan: I'm about to do that, this is driving me up the wall
<ivan>
xenophile: it looks like windscribe has wireguard support which you can use on nixos via wg-quick
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ivan>
lejonet: check journalctl / dmesg for the crash info from the kernel
<xenophile>
:ivan Oh interesting. I'll go look into that. Thanks!
<lejonet>
ivan: there is none... that is what vexes me
meh` has joined #nixos
<ivan>
lejonet: try --disable-ion
hyper_ch4 has joined #nixos
<shapr>
back to my byobu questions, when I run byobu-config I get "ERROR: Could not import the python snack module"
<lejonet>
But that pops the question "But why is chromium working then?", because even tho they don't use the same jit, chromium is guaranteed to be using some type of JIT that should hit similar snags
<lejonet>
oh well, firefox compiles aren't +4h anymore, so it isn'
<lejonet>
isn't TOO much of a pain to build it from source :P