<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to master: https://git.io/JUJnN
<superherointj>
Hello. I'm failing to get right the syntax for the option "systemd.network.networks.<name>.routes.*.routeConfig". My problem is at the '.*.' part.
<superherointj>
I tried to do something like this: "routes = [ { "Destination" = "192.168.60.0/24"; "Gateway" = "192.168.50.40"; } ];"
invokesus has joined #nixos
alexherbo2 has quit [Ping timeout: 260 seconds]
<superherointj>
It errors as: "error: The option `systemd.network.networks.kvm0-vswitch.routes.[definition 1-entry 1].Destination' defined in `/etc/nixos/configuration.nix' does not exist."
<zeta_0>
is there something wrong with the `search nixos packages` web page ? nothing is loading ?
<zeta_0>
the `options` section isn't loading either, how strange.
<samueldr>
zeta_0: tried inputing a search?
<samueldr>
it doesn't load the whole package set first
<samueldr>
(anymore)
<samueldr>
don't forget to send it off with enter
<zeta_0>
samueldr: things aren't auto-loading by default in the search nixos packages section, and the auto-completion is not working when i type things in, oh so it's a new feature that you guys added, that surprised me.
<samueldr>
the complete-as-you-type is, for the time being, gone, though work is happening to bring it back
<zeta_0>
samueldr: why this change though ?
<samueldr>
there are limitations in letting the browser download and manage search through the whole package set
<samueldr>
we already weren't indexing the whole thing
<samueldr>
and we couldn't index some of the package sets otherwise it basically made the load terrible for lower-specced computers
<samueldr>
now the search happens server-side, which allows us to index way more
<samueldr>
there are further plans to make the search more useful that couldn't be done any otherway
<samueldr>
I think all plans are out in the open in the nixos search repository (linked in the footer)
<zeta_0>
samueldr: oh i see, it was getting out of hand, and some of the pkgs did not show up in the search, so i wasn't sure if they existed, i'm happy you guys are at least attempting to fix this issue.
<zeta_0>
anyways i'm trying to build nixos on a raspberry pi 4, but it's taking an eternity, i guess i could try leaving it overnight.
<zeta_0>
the searching does seem so much faster, so that's good.
<zeta_0>
samueldr: thanks for the explanation.
<zeta_0>
samueldr++
<{^_^}>
samueldr's karma got increased to 256
gv has quit [Remote host closed the connection]
gv has joined #nixos
proofofkeags has quit [Remote host closed the connection]
<zeta_0>
does the nix unstable channel use emacs 27, because the nixos search does not show it ?
<manveru>
sorear: that looks awesome, would you mind posting it on the forum?
griff_ has joined #nixos
<sorear>
is the forum the discourse?
<manveru>
yes
cole-h has quit [Quit: Goodbye]
<manveru>
at the very least this should make it easier to discover for other newcomers :)
<garbas>
higherorder: hey! yes, I'm working from NixOS. I don't have access to any osx machine. Can you report your issues to nixos-search issue tracker?
CMCDragonkai has quit [Remote host closed the connection]
CMCDragonkai has joined #nixos
<yurb>
Do I understand correctly nix has to trust a binary cache server to build the packages correctly?
<yurb>
I.e., a malicious binary cache server can serve a modified binary for the same derivation hash and nix won't be able to tell without rebuilding itself?
<yurb>
(rebuilding the binary in question locally)
<sorear>
yes
<yurb>
okay, makes sense
<yurb>
is there some way to see if a given a nixpkgs commit produces the same bit-to-bit binary for a given package on two different machines?
<yurb>
or derivation, rather
sigmundv__ has joined #nixos
IRCsome has joined #nixos
otulp_ has joined #nixos
IRCsum has quit [Quit: leaving]
<yurb>
preferably down to every dependency
<alexarice[m]>
∿ und̷e̷l̷ě̷t̷e̷d̷: you could probably just hash the two store paths?
<yurb>
(I understand that in theory every build should be deterministic)
<bqv>
Build with --check
<yurb>
but is that checked on the binary caches for every package?
<yurb>
let's say there is an upstream source that does some weird random() stuff in their build scripts
<bqv>
I mean, no, but nixpkgs endeavours to make sure theres no crazy stuff like that
<bqv>
I don't think hydra does check builds
<angerman>
is anyone working on nix on *bsd in any form?
<bqv>
#freebsd-nix
<yurb>
(I'm playing devil's advocate here - just trying to figure out the threat model of NixOS)
civodul has joined #nixos
<alexarice[m]>
∿ und̷e̷l̷ě̷t̷e̷d̷: I don't think all packages are completely deterministic
<dminuoso>
Mmm, what's the protocol to cause *all* haskell packages to be built with -split-sections ?
<alexarice[m]>
some log the date they were built or weird stuff like that
fendor__ is now known as fendor
<kraem>
can i somehow/go 13
<kraem>
oops
<regnat>
setthemfree: No, it's not enforced. There's tooling to check it (nix-store --check) and semi-independant initiatives to improve it (https://r13y.com/), but it's not fundamentally required by the Nix model
<maralorn>
∿ und̷e̷l̷ě̷t̷e̷d̷: No, nixos does not yet have fully reproducible builds. I think the mental model is, that for a given input (hash) every occurring output is valid. In the sense that they don't differ in and relevant way.
<init_6>
balsoft: Hello Friend! How are you?
<maralorn>
But yeah you either build yourself or you trust out build server hydra. No way around that.
<dminuoso>
The channel has both haskellPackages.ip (1.5.1) and haskellPackages.ip_1_7_1 (1.7.1) - how do I make it use ip_1_7_1 instead of ip?
<dminuoso>
(Im using callCabal2Nix here)
<dminuoso>
Should I just do a hackery override in the sense of { ip = super.ip_1_7_1; } ?
<srk>
someone ask exactly this yesterday - global overlay, overrideCabal or developPackage with overrides
<srk>
asked
<dminuoso>
Mmm, so overrides it is
<dminuoso>
Cheers
<dminuoso>
Not sure what developPackage is for, exactly. So far, Im just doing something like: let overrides = { foo = lib.dontCheck super.foo; myPackage = self.callCabal2nix "myPackage" ./. {}; } in haskellPackages.override { inherit overrides }
<dminuoso>
Dont even recall the reason for mixing my package in as an override, but I had good reasons for it. :p
hyper_ch2 has quit [Ping timeout: 260 seconds]
<srk>
developPackage is just another abstraction combining few different things together (callCabal2nix, overrides, modifiers..)
hyper_ch2 has joined #nixos
<srk>
I prefer to use "lower" level things to achieve the same, similar to what you do
<srk>
it's like higher-order function to do the same
<dminuoso>
ah I see
arjen-jo1 has joined #nixos
<dminuoso>
Yeah, I dont interact with nix too often, so a bit more low level is preferrable so I dont need to remember what these higher level things do.
<{^_^}>
#96127 (by hmenke, 1 day ago, open): shadowsocks service: support plugins
<henri>
The maintainers of both don't seem to be too active.
<{^_^}>
[nixpkgs] @golddranks opened pull request #96271 → Disable trifecta tests on AArch64, as they tend to fail and prevent the build → https://git.io/JUUEZ
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/JUUEg
<{^_^}>
[nixpkgs] @peti opened pull request #96272 → Update Haskell package set to LTS 16.11 (plus other fixes) → https://git.io/JUUEV
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell-language-server: fix overrides for new version of hie-bios »: https://git.io/JUUEi
<munksgaard>
Hi! I'm having trouble with cabal and zlib on NixOS. I have a haskell project that depends on libz.so.1, so I've added zlib (and zlib.out and zlib.dev) to my buildInputs in shell.nix, but I still get this error when I try to run `cabal repl`: http://paste.debian.net/1161131/ Does anyone know what could be going wrong?
<dminuoso>
munksgaard: I just use `shellHook = ''export LD_LIBRARY_PATH=${pkgs.lib.makeLibraryPath buildInputs}:$LD_LIBRARY_PATH'';
Acilim has joined #nixos
<dminuoso>
No idea whether that's "the right solution", but all my shell.nix contain that :P
<dminuoso>
And it works at least
andreas303 has joined #nixos
<munksgaard>
dminuoso: That seems to work. But clearly, there must be a nicer solution than manually hacking the library path? I thought these kinds of things were exactly what NixOS was trying to avoid?
<dminuoso>
munksgaard: I dont think its solvable.
<dminuoso>
munksgaard: The reason is: cabal needs to load libz as a dynamic library, but that dependency is not because of cabal, but because of what you're running against.
<munksgaard>
dminuoso: I don't know enough about nix to understand why that's a problem. When libz is added as a build dependency, couldn't it just be automatically added to the LD_LIBRARY_PATH?
<dminuoso>
munksgaard: at what point would that happen?
<dminuoso>
The derivation of cabal cant know what dynamic libraries cabal will need for repl
<dminuoso>
If we had some cabal/shell builde, that could take care of it I guess
orivej has joined #nixos
<dminuoso>
munksgaard: The stack builder for example pretty much does the above
<{^_^}>
Found in packages: mesa, driversi686Linux.mesa
<infinisil>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<infinisil>
graf_blutwurst: ^
<graf_blutwurst>
much obliged!
<symphorien[m]>
,locate libgbm.so
<symphorien[m]>
Locally you can use nix-index/nix-locate
<{^_^}>
Found in packages: mesa, driversi686Linux.mesa
alexherbo2 has joined #nixos
mallox has quit [Quit: WeeChat 2.9]
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #96285 → viw: init at 20171029 → https://git.io/JUUwL
mallox has joined #nixos
ncl3 has joined #nixos
<{^_^}>
[nixpkgs] @golddranks closed pull request #96271 → Disable trifecta tests on AArch64, as they tend to fail and prevent the build → https://git.io/JUUEZ
<{^_^}>
[nixpkgs] @golddranks opened pull request #96287 → Disable trifecta tests on AArch64, as they tend to fail and prevent the build → https://git.io/JUUwg
ddellacosta has joined #nixos
<{^_^}>
[nixpkgs] @danieldk opened pull request #96288 → python37Packages.tensorflow: mark broken → https://git.io/JUUwD
<{^_^}>
[nixpkgs] @zowoq pushed commit from @saschagrunert to master « ginkgo: init at 1.14.0 »: https://git.io/JUUoZ
<graf_blutwurst>
got it to run. turns out the package was just "mesa" "driversi686Linux" seems to refer to the 32 bit version which lead to ELF errors
<{^_^}>
[nix] @Ericson2314 opened pull request #3959 → WIP: Derivations can output "text-hashed" data --- contains many other PRs → https://git.io/JUUo4
invokesus has joined #nixos
<{^_^}>
[nixpkgs] @vcunat opened pull request #96290 → nixos boot extra-utils: ensure we use working lvm2 → https://git.io/JUUoR
<krkrkr>
that works for ethernet cards but not for bridges
krkrkr has quit [Quit: Ping timeout (120 seconds)]
griff_ has quit [Quit: griff_]
growpotkin has joined #nixos
work_ has joined #nixos
leotaku has joined #nixos
is_null has quit [Ping timeout: 256 seconds]
dsg has quit [Ping timeout: 246 seconds]
philr_ has quit [Ping timeout: 240 seconds]
erasmas has joined #nixos
kenran has quit [Ping timeout: 240 seconds]
acarrico has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
is_null has joined #nixos
zupo has joined #nixos
dsg has joined #nixos
erasmas has quit [Read error: Connection reset by peer]
abathur has quit [Ping timeout: 240 seconds]
erasmas has joined #nixos
sgrunert has quit [Remote host closed the connection]
user_0x58 has joined #nixos
wnklmnn has joined #nixos
<asymmetric>
it seems the aspell data-dir config value is not set correctly. it's left at the default ($out/lib/aspell), whereas the dat files are actually saved under the dictionary's output
user_0x58 has quit [Ping timeout: 240 seconds]
user_0x58 has joined #nixos
<{^_^}>
[nixpkgs] @cdepillabout merged pull request #96287 → Disable trifecta tests on AArch64, as they tend to fail and prevent the build → https://git.io/JUUwg
<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to haskell-updates: https://git.io/JUUPR
user_0x58 has quit [Ping timeout: 256 seconds]
ManiacOfMadness has quit [Ping timeout: 272 seconds]
graf_blutwurst has quit [Remote host closed the connection]
meh` has joined #nixos
mallox has quit [Quit: WeeChat 2.9]
acarrico has quit [Ping timeout: 265 seconds]
griff_ has joined #nixos
<jasom>
I have fbdev working, but modesetting does *not* work despite the kms driver loading and /dev/dri/card* existing. https://gist.github.com/jasom/574fbd18b8d30ec6f4811e5ca7fe1516 is output when starting the display-manager. If it matters this is on aarch64
<LouisDK>
I'm trying to add an appmenu module for GTK but it seems like it trying to load the gtk2 variant in gtk3 apps. Could it have something to do with how "GTK_PATH" get set? https://github.com/NixOS/nixpkgs/pull/95412
sigmundv__ has quit [Read error: Connection reset by peer]
ManiacOfMadness has quit [Ping timeout: 244 seconds]
sigmundv__ has joined #nixos
<joesventek>
I'm trying to package an unfree software that is only distributed in binary form (as a .deb). What would be the easiest way to figure out the required buildInputs? Look at the output of `ldd binary` and search for corresponding pkgs?
is_null has quit [Ping timeout: 240 seconds]
<joesventek>
Or are there any tools that assist in finding the dependencies?
<{^_^}>
[nixpkgs] @primeos pushed commit from @raboof to master « helm: compile with gcc9 (#95862) »: https://git.io/JUUyQ
thc202 has joined #nixos
LouisDK has quit [Ping timeout: 258 seconds]
<pie_>
hm...do i need to do something for a container to get n internal network device and a ve-containername device?
<pie_>
i have privateNetwork enabled
adam_ has joined #nixos
<adam_>
Do any of you all use neovim with nixos?
<adam_>
I would like to use the build in lsp client in nvim nightly
<adam_>
Is there a way to specify which nvim version nixos uses? I would like to either build from source, or at least use a build which is fairly recent
<siraben>
Can someone review #96051? It's ready for merge.
<lordcirth>
I have Ubuntu 18.04 on a 3-SSD raidz root. How crazy is it to try switch to NixOS in-place by creating another ZFS filesystem? Am I likely to bork the existing setup?
<pie_>
basically im trying to make it more convenient and appliance-y
ris has joined #nixos
ATuin has joined #nixos
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #96303 → tty-solitaire: init at 1.3.0 → https://git.io/JUU9E
<dminuoso>
How can I have nix-build be verbose about how it invokes things like ghc?
dbmikus has quit [Ping timeout: 240 seconds]
<infinisil>
nix-build itself doesn't control the output, the derivation builder does
<{^_^}>
[nixpkgs] @B4dM4n opened pull request #96309 → qt5.qtwebkit: fix build with bison-3.7 → https://git.io/JUU7c
mmohammadi9812 has quit [Remote host closed the connection]
Darkmatter66_ has quit [Read error: Connection reset by peer]
mananamenos has quit [Ping timeout: 265 seconds]
proofofk_ has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
fendor_ has quit [Remote host closed the connection]
proofofkeags has quit [Read error: Connection reset by peer]
proofofk_ has quit [Ping timeout: 265 seconds]
<lordcirth>
If I set zfs mountpoint=legacy to install NixOS, will that immediately break my current Ubuntu boot from that pool?
fendor has joined #nixos
<lordcirth>
Ah, I can do it per filesystem, nvrm
<infinisil>
bqv: I am awake, but not really interested right now in messing with flakes or haskell.nix, sorry!
__Sander__ has quit [Quit: Konversation terminated!]
<bqv>
sure!
drakonis_ has quit [Ping timeout: 246 seconds]
lsix has quit [Ping timeout: 272 seconds]
bvdw4 has joined #nixos
bvdw4 has quit [Client Quit]
Extends has joined #nixos
bvdw8 has joined #nixos
<{^_^}>
[nixpkgs] @yvesf opened pull request #96310 → update qcad to 3.25.0.0 → https://git.io/JUU5V
bukkitgerman8608 has quit [Quit: Ping timeout (120 seconds)]
stephank has quit [Quit: Ping timeout (120 seconds)]
patagonicus3 has quit [Quit: Ping timeout (120 seconds)]
bvdw has quit [Read error: Connection reset by peer]
bvdw8 is now known as bvdw
<Extends>
Hi, does anyone know how can I get past the "restricted mode" in the flakes ? I'm trying to read a file from a subdirectory of my flake configuration, but I don't know why, nixos tells me that it isn't allowed in restricted mode, whereas it absolutely doesn't break reproductibility
fendor has quit [Read error: Connection reset by peer]
fendor has joined #nixos
<bqv>
you can't get past it, you should work with it. is the subdirectory staged/committed to your repo?
<Extends>
it is indeed
<bqv>
how are you referring to it?
<bqv>
(censor as needed)
<Extends>
I tried using ./<machine-hostname>.nix and /etc/nixos/hosts/<machine-hostname>.nix
<Extends>
what do you mean by censor ?
<bqv>
is ./hostname.nix the right path??
<bqv>
or should you have used ./hosts/hostname.nix
<Extends>
./hostname is the right path as I import the file in a default.nix in the same directory
<bqv>
and are you passing that to import, or putting it in "imports"
<Extends>
I'm putting it in imports
kenran has joined #nixos
<bqv>
...you're absolutely sure that when you tried ./hostname.nix, it was added to git's index?
<bqv>
cause if so, i'm not actually sure what's wrong there
rardiol has quit [Read error: Connection reset by peer]
<bqv>
you might try mitigating this by running with "--impure", which does what it says on the tin, allows impure everything. if even that fails, at least you know that restricted mode is not your issue
<Extends>
I'm sure sure sure
<Extends>
is --impure a flag of nixos-rebuild ?
<Extends>
or do I have to build my configuration with another way
<bqv>
i doubt nixos-rebuild will support it. you'll have to try nix build directly
<Extends>
that's what I thought; I'll try this right away
<adam_>
just do programs.neovim.enable = true and configure it similar to how the vim wiki for nixos works
<aterius>
I mean, or list neovim-nightly in home-packages
<aterius>
your call
<adam_>
okay I'll try that
<adam_>
so just have a plugins attribute in your programs.neovim?
<aterius>
If you want to manage plugins through nix
<adam_>
I am trying to go full nix haha.
<aterius>
My overlay doesn't affect any of this
<bqv>
Extends: i have no useful ideas, tbh. i do the exact thing you're doing, and it works, so at best i'd try frobnicating some things. maybe replace ./file.nix with (import ./file.nix), see if that helps
<dominikh>
I have a question about flakes, dependencies and the lock file: nixpkgs is a dependency like any other, pinned to a specific revision, right? Won't that potentially cause problems when using different flakes that depend on different versions of nixpkgs, thus pulling in different versions of dependencies? For example for inter-process communication that expects to be using the same version of a
<dominikh>
library
rardiol has joined #nixos
<bqv>
dominikh: nix supports multiple versions of a library installed side-by-side because it's definition of "installed" is different. in the same way, flakes support multiple simultaneous nixpkgs trees during evaluation. IPC is an edge case, and if it doesn't work, you can always override broken packages. That said, since nixpkgs painfully strives for package coherency, it's usually unlikely that you'll
<bqv>
hit even that
lsix has joined #nixos
<dominikh>
bqv: When you say "since nixpkgs painfully strives for package coherency", how does that relate to third party flakes?
<pikajude>
hi, i have a qemu guest on a nixos host and the host is connected to a vpn
<pikajude>
is there a way to enable the guest to use it as well?
rardiol has joined #nixos
<pikajude>
i know absolutely nothing about networking
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<bqv>
dominikh: because firstly, if you use mostly one version of nixpkgs and just cherry pick packages from others as needed, you will hit essentially no incoherency bugs, and secondly, in the case that you do, you can exchange those incoherency bugs for potential build failures by overriding the flake input, and relying on your "main nixpkgs" being coherent
<bqv>
e.g. in my flake i usually make the tradeoff of unifying nixpkgs from every flake, at the expense of getting a hell of a lot of cache misses
<bqv>
(actually, i think there's currently one that i'm not overriding)
<{^_^}>
[nixpkgs] @jonringer closed pull request #96246 → discord: use buildFHSUserEnv to avoid crashing → https://git.io/JUJ2J
<cole-h>
For your local machine, if you were to `time nix run` right now, what do you get?
lordcirth has quit [Remote host closed the connection]
<bqv>
something extremely unpleasant, because i'm running a painful haskell build
<cole-h>
Heh
<bqv>
at a guess, on a pristine tree, it'd be under 5 seconds for the first two hosts (this one and another on LAN), and a minute or two for the last one (server)
<cole-h>
Next time you have a mostly no-op deploy, please run that for me and report back. I'm interested in your results, since for a no-op `nixops deploy`, it's 23 seconds for me.
werner291 has quit [Quit: werner291]
cryptomonad has joined #nixos
<bqv>
sure
<cole-h>
bqv++
<{^_^}>
bqv's karma got increased to 21
drakonis_ has joined #nixos
<cole-h>
(I'm frequently away from IRC these days, so if I'm not around, feel free to ,tell me)
andreas303 has quit [Remote host closed the connection]
<cryptomonad>
I'm packaging a script pkgs.chez (scheme) that includes shared libraries (using load-shared-object). What's the best way to tell scheme about the location of the shared libraries? I understand in C that's the job of ld-wrapper but there's nothing like that for scheme in nix. I can get it to work with LD_LIBRARY_PATH but I assume that's not the proper way to do it.
andreas303 has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<Extends>
bq: I think I might have found the problem: as I'm using builtins.readDir to iterate over my files, I import my files using imports = [ "/etc/nixos/hosts/sigma.nix" ] (because ./"sigma.nix" doesn't work), is there a way to convert builtins.attrNames (builtins.readDir dir)) to files instead of strings ? so that nix would understand that it comes from the same directory
fendor has quit [Read error: Connection reset by peer]
<clever>
Extends: ./. doesnt end in a / so `./. + "foo.txt" in a bar dir, winds up pointing to /path/to/barfoo.txt
<Extends>
yep
<bqv>
hey, there aren't by any chance, any caches that build ghc865's packages?
<bqv>
actually, i remember someone saying cache.nixos.org never deletes anything, maybe i could go back to an old nixpkgs and build using that..
shibboleth has quit [Quit: shibboleth]
<cole-h>
infinisil: Ping
LouisDK has joined #nixos
<{^_^}>
[nixpkgs] @aanderse opened pull request #96316 → nixos/redmine: replace extraConfig option with settings option → https://git.io/JUUFP
pamplemousse has joined #nixos
Dr8128 has quit [Ping timeout: 258 seconds]
hexa- has quit [Read error: Connection reset by peer]
<Church->
Nice.
hexa- has joined #nixos
<pamplemousse>
Hey :)
<pamplemousse>
Could someone explain the difference between `nix-shell -p python3Packages.some_package` and `nix-shell -p 'python.withPackages(ps: with ps; [ some_package ])'` to me?
<cole-h>
python3Packages.some_package -> this package is available in the environment. python3.withPackages... -> python3 is available in the environment, with access to this package
<cole-h>
I believe
<pamplemousse>
What do you call the "environment" in this context?
<cole-h>
The shell environment.
<pamplemousse>
So, "python3 is available in the environment" means... the binary for python3 is in the PATH of the shell?
<cole-h>
Yes
alp has quit [Ping timeout: 272 seconds]
<pamplemousse>
I get that a package could be available to python (via a well set PYTHONPATH). But I am not too sure of what that means for a package to be available in the environment...
<bqv>
if the package has binaries, the former would bring those binaries into scope
<pamplemousse>
Maybe: the scripts exposed by a package are available in the PATH.
aswanson has quit [Quit: WeeChat 2.7.1]
<bqv>
yes
<bqv>
scripts rather than binaries, fair point
<pamplemousse>
I see
meh` has quit [Ping timeout: 258 seconds]
<pamplemousse>
Thanks cole-h and bqv :)
<bqv>
S6
<bqv>
*:)
<cole-h>
bqv: Does nixus support dry runs, do you know?
<bqv>
i don't believe so
<bqv>
you might be able to engineer one
<bqv>
the deploy process is almost fully exposed
<bqv>
you can inject code
<cole-h>
Does it support an analogue of `nixos-rebuild test`?
<bqv>
again, not directly, but you could probably engineer one
<cole-h>
OK, thanks.
<bqv>
e.g. i've hacked it a bit to replace the use of nix-copy-closure with nix copy, without having to touch the source
<cole-h>
bqv: Do you remember the flag for nix(1) (e.g. `nix build`) that shows the full output log as it happens?
<bqv>
so i don't see why it wouldn't be doable to do similar for switch -> test
<bqv>
cole-h: i think it would be problematic if it could be set via --log-format and --option log-format. it's probably wiser to just have it check an environment variable
<cole-h>
bqv: Just like `--option experimental-features` and `--experimental features`? ;)
<adam_>
aterius: I've made great progress with the neovim nightly but I have one more question for you if you are still there?
<dminuoso>
but I suppose it doesnt have to be
<adam_>
I am trying to use the built in nvim-lsp. I added it to the plugins and it seems to have installed
<infinisil>
cole-h: Pong!
<adam_>
I am trying to add the vim-language-server for example. I got the node module installed, and I can check, it is on the path
<cole-h>
infinisil: I don't even remember my question :D But know this: you're gonna have a PR to nixus very soon!
<infinisil>
Oh neat, what about?
<cole-h>
infinisil: doas :)
<infinisil>
Ahh :D
<cole-h>
infinisil: Rather, generic privilege escalation, but "sponsored" by my love for doas
<infinisil>
Appreciated!
<dminuoso>
cole-h: also, it seems incorrect to do `{ haskell.packages = .. }` because that gets rid of all other attributes of haskell, no
<cole-h>
infinisil: A question does come to mind, though: how can I make it so nixus shows all logging? e.g. `cat /var/lib/system-switcher/system-#/log` displays additional logging not present on the terminal.
<cole-h>
(like the "stopping the following units" info)
<dminuoso>
bqv: That does not address my issue at all.
<infinisil>
There are a bunch of nixus problems, and I think all of them are due to bash, aka they'be be solved if I switch the implementation to haskell
<adam_>
Is it better to use vim.cmd('packadd nvim-lsp') in the lua <<EOF section?
<dminuoso>
Whether I use .extend or .override is not the problem
<bqv>
dminuoso: did you read the last part? :p
<infinisil>
cole-h: Well not *due* to bash, but using bash makes it much more difficult to fix them
<dminuoso>
Im doing a pkgs wide overlay, so if this on haskellPackages all is good, because its just 1 attribute in. Buf if you want this on a specific haskell package set, then that's 3 attributes in
<dminuoso>
Right
<dminuoso>
Im asking whether there's some handy bits to do this for me
<bqv>
no, because generally this is a pretty extraneous requirement
<cole-h>
infinisil: That was just ~30m of hacking, so there might be things to be changed. Especially unsure of the naming (`privilegeEscalationCommand` is a bit long), but I just followed nixops there.
<dminuoso>
be that as it may, but static-haskell-nix assumes an entire nixpkgs set in the correct shape
<dminuoso>
so I dont have much choice, now do I? :)
<bqv>
wasn't judging%
<dminuoso>
fair
<infinisil>
cole-h: Neat! I'll take a closer look later today or tomorrow :)
<cole-h>
:D
<cole-h>
bqv: OK, it's time to wade through your monstrosity of a flakes.nix and Frankenstein my own :P
<bqv>
lmao
<bqv>
i'm sorry in advance
<cole-h>
You should be
<cole-h>
My eyes are already melting
<cole-h>
Hehe
<cole-h>
I've said it before, but I kinda wish there was a way to use git repos instead of tarballs... tarballs can be a large download every time, while a git repo can be updated incrementally.
<dominikh>
wrt flakes and nixpkgs, have there been any thoughts on flakes affecting how software gets packaged for nixpkgs? e.g. if some software already has its own flake.nix, will nixpkgs just refer to that, or define its own derivation like it currently does?
<bqv>
that's a flamewar that hasn't quite been resolved yet
<adam_>
How do you all deal with situations where you want some python modules available globally, but others just for a particular project
<adam_>
however when you are in that project, you want to have BOTH the globally available packages, and the project specific ones
<cole-h>
Yikes
<cole-h>
I'd argue that's... generally not what you should do.
<adam_>
I am currently using nix-shell and specifying the python version + packages, then activate the shell
<adam_>
but there are certain things like linters/language servers which I want everywhere
<adam_>
and are not "project dependencies" but instead vim dependencies
<adam_>
cole-h: so what would you do instead?
<energizer>
adam_: install your project tools in your project