<Jonathan44>
I am not figuring it out clever. Is there any other advice you can give me? I am a little confused because it looks like the cmake error is saying the error is on line 138 `CMake Error at CMakeLists.txt:138 (install):` but that line is `set(PKG_CONFIG_USE_CMAKE_PREFIX_PATH ON)` which appears to have nothing to do with `install`.
<lunik1>
ok I will have a read, thanks
<clever>
Jonathan44: can you put all of the nix files into a gist?
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
Jonathan44: i dont see how that ccd derivation would ever work, it has no source, so nix-build has nothing to build
<Jonathan44>
I am sorry, I copied an old version of the file.
<GrimSleepless>
bhipple: Thank you for your review. I made the changes you asked for. I have also added a commit to add the missing dependency of iptables
Ilya_G has quit [Remote host closed the connection]
<ajs124>
evils: that link seems broken
<evils>
s/thml/html/
Fare has joined #nixos
<ajs124>
Ok, I'll go ask in that channel, if someone can take a look at it, thanks!
<Jonathan44>
Why can the src not be "."?
<clever>
Jonathan44: it must contain a / and must not be quoted
<clever>
Jonathan44: src = ./.; is most common
<Jonathan44>
Oh.
<gchristensen>
clever: check this out:
<{^_^}>
[nixpkgs] @grahamc opened pull request #83327 → nixos-rebuild: support --upgrade-all and document --upgrade → https://git.io/JvS3q
<clever>
gchristensen: heh :D
<clever>
gchristensen: i wonder what spured that on!
<clever>
gchristensen: personal, i dont touch --upgrade, it conflicts with --rollback (which wont fully undo the upgrade)
kleisli has joined #nixos
<gchristensen>
hehe
<{^_^}>
[nixpkgs] @JeffLabonte opened pull request #83328 → [19.09] protonvpn ng 2.2.0 to 2.2.2 → https://git.io/JvS3s
<Jonathan44>
Oh, clever, fcl also depends on Eigen.
mbrgm_ has joined #nixos
<clever>
Jonathan44: already ahead of you
<clever>
Jonathan44: nearly done compiling
<Jonathan44>
Oh, wow
mbrgm has quit [Ping timeout: 240 seconds]
mbrgm_ is now known as mbrgm
<simpson>
Jonathan44: Once you get the hang of this, you'll be able to do this much faster. A lot of this is grepping nixpkgs, looking up release tarball URLs and versions, filling out license metadata.
<Jonathan44>
Yeah, I am sure. I am quite new to nix (and not familiar with C/C++ build systems).
exprez135 has joined #nixos
<Jonathan44>
Mine is building now too. Eigen is in nixpkgs so it is easier (:
<Jonathan44>
Um, clever. How can I use that in a shell. nix-shell is throwing errors.
<clever>
Jonathan44: you need to use -A to tell it which thing you want a shell on
<clever>
Jonathan44: either `-A libccd` or `-A fcl`
<Jonathan44>
Oh, thanks.
<Jonathan44>
I am going to have to look up the documentation on the scope thing you used.
<clever>
Jonathan44: i'm not sure its fully documented, but it basically lets you make a private set of packages, and you can still add overlays to it later
<Jonathan44>
Okay.
johnny5 has quit [Ping timeout: 240 seconds]
<Jonathan44>
I get the idea, but I am afraid I am going to need to become more familiar with other parts of nix before I can do it myself.
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
cosimone has quit [Quit: Quit.]
butterthebuddha has joined #nixos
<Jonathan44>
Can I that in a shell which includes more dependencies?
<clever>
Jonathan44: you could just add them to the buildInputs
<atemu12[m]>
Jonathan44: You can always add more `buildInputs`
<Jonathan44>
clever: `let fcl-thing = import(./default.nix);` and `fcl-thing.fcl` doesn't work.
<clever>
Jonathan44: depending on how you do it, that may give you an env with fcl pre-built, not fcl's deps
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvS3N
habbah has quit [Ping timeout: 240 seconds]
<atemu12[m]>
I'm trying to exclude a list of packages from systemPackages via an overlay but for some reason my solution doesn't work, how can I debug this?
<clever>
atemu12[m]: an overlay cant modify systemPackages, the most you can do is make that package null, like `firefox = null;` so installing firefox does nothing
<atemu12[m]>
Well, that'd explain why
<atemu12[m]>
clever: Do you know of a good way of doing that for a list of packages?
<clever>
atemu12[m]: for example, clevers_machines.nix does a bunch of config that i put on many of my machines, but i could just omit that from the imports list on a pi
<OmnipotentEntity>
So... who here knows the awful oroborous of fuckity that is CUDA? Specifically, I seem to need LD_LIBRARY_PATH to contain `/run/opengl-drivers/lib` to make a CUDA program go. If I have to put that in a nixpkg derivation using wrapProgram should I use that path or should I use `${linuxPackages.nvidia_x11}/lib` instead? grahamc's github for CUDA examples has the latter, but other packages in the repo seem to have the former.
<clever>
OmnipotentEntity: if xorg is enabled, then LD_LIBRARY_PATH will contain /run/opengl-drivers/lib system wide
<atemu12[m]>
<clever "atemu12: for example, clevers_ma"> That isn't going to work nicely with my config layout but you just gave me an idea
<OmnipotentEntity>
I thought I did, but I suppose a better question is: How does one go about doing that the proper way? Is there documentation I can read to ensure I did it correctly?
martin__ has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @jonringer opened pull request #83334 → python: make update script use installable path → https://git.io/JvSsY
<OmnipotentEntity>
I do have programs.zsh.enable set to true
jumper149 has quit [Quit: WeeChat 2.7.1]
lunik1 has joined #nixos
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mexisme_ has joined #nixos
<clever>
OmnipotentEntity: not sure whats going on then
<OmnipotentEntity>
I should note that nix-shell does still run bash by default, and running nix-shell with no arguments and running bash both do not set LD_LIBRARY_PATH either.
<clever>
OmnipotentEntity: it could either be /etc/bashrc or /etc/profile
<clever>
OmnipotentEntity: it may only run for "login shells", so a non-login bash wont do it
<OmnipotentEntity>
lookes like /etc/profile calls a set-environment script in the nix store
<bqv[m]>
i feel like i should try a different window manager to exwm
<OmnipotentEntity>
:q
<bqv[m]>
it's been a while now
<bqv[m]>
maybe there's something better
<OmnipotentEntity>
sorry wrong window
<srid>
Why is `nix-env -f ... -i` behaviour so strangely? The -f argument takes a tarball path (on Github), but it uses the previous fetch of the tarball. I just pushed a new revision to the git repo, but nix-env is not fetching (I did an uninstall before running nix-env)
<clever>
srid: nix show-config | grep tarball-ttl
<clever>
srid: it caches the tarball for 1 hour by default
<srid>
Can I temporarily disable this?
<clever>
srid: you can either use `--option tarball-ttl 1234` to change it, or you can just wipe ~/.cache/nix/tarballs
<srid>
Right, right, thanks!
ZaraChimera has quit [Ping timeout: 240 seconds]
<srid>
Really, all of this should be done automatically for `nix-env -f ... -u`
<OmnipotentEntity>
clever, what would you suggest then at this point? I'd reinstall but that shouldn't matter on NixOS because every upgrade is equivalent to a reinstall.
<OmnipotentEntity>
It looks like the default was set to false.
mexisme_ has quit [Ping timeout: 260 seconds]
<clever>
OmnipotentEntity: oh, i see that, line 130 in the file i linked
<clever>
OmnipotentEntity: try changing it to true?
mexisme_ has joined #nixos
<OmnipotentEntity>
well, if it's the case that the default is false, wouldn't I want to add that to the wrapProgram then?
<energizer>
when i do `src = ./.`, nix attempts to read my 50GB ./big_data.csv into memory, runs out of ram, and bails. i want it to succeed instead of failing, perhaps by transferring the file in chunks or using file-copy operations. what are the issues involved here?
<clever>
OmnipotentEntity adding it to the wrapProgram would also be a solution
<clever>
energizer: builtins.filterSource
<clever>
energizer: if you want to just exclude the csv file entirely
<clever>
energizer: pkgs.requireFile or pkgs.fetchurl if you actually want it in the store
<energizer>
clever: sure. what if i actually want the data to be stored?
<energizer>
ah
<OmnipotentEntity>
right, circling back around to the original question, should I used ${linux.nvidia_x11}/lib or /run/opengl-drivers/lib, do you think?
<OmnipotentEntity>
And it's great that something isn't mysteriously broken on my box :D
<clever>
OmnipotentEntity: you should use /run/opengl-drivers/lib, so it picks up the right drivers for the kernel
<OmnipotentEntity>
roger doger.
<OmnipotentEntity>
Thanks!
<OmnipotentEntity>
should I build against ${linux.nvidia_x11} though?
<OmnipotentEntity>
or should I build against the system drivers?
<OmnipotentEntity>
I suppose I should also build against the system drivers huh
<clever>
OmnipotentEntity: i think mesa is the right one, but you may need nvidia_x11 for the cuda headers
<OmnipotentEntity>
I'm just trying to work through a review and make sure I'm doing it the right way
<jared-w>
does anyone know a way to get `sh <(curl https://nixos.org/nix/install) --daemon` to run completely non-interactively?
justan0theruser has joined #nixos
justanotheruser has quit [Ping timeout: 260 seconds]
<cole-h>
Hook it up to `expect`, maybe?
<energizer>
clever: what is the difference between requireFile, fetchurl, copyPathToStore?
<clever>
energizer: requireFile is typically used for things behind a license agreement, like java
<clever>
energizer: it gives you an error saying to download the thing, and run `nix-store --add-fixed sha256 ./foo`
<clever>
energizer: then its in the store, and things work as normal
<clever>
energizer: pkgs.fetchurl is how most things work, but requires you to host it on an http server somewhere
<clever>
energizer: copyPathToStore is an internal function inside nix
<OmnipotentEntity>
clever, looks like it works just fine using the system drivers. So I've removed the linuxPackages reference. Thanks for all of your help
<OmnipotentEntity>
:D
<jared-w>
cole-h: printf 'n\n' | .... turned out to be the simpler solution
<jared-w>
cole-h: was gonna try expect at first but couldn't be assed to install it just to try a hunch out
<cole-h>
clever: As the resident Nix expert, would you mind explaining the difference between `import ./default.nix` and `import ./default.nix {}`?
<clever>
cole-h: after importing the file, it treats it as a function, and passes it an empty set
<cole-h>
In hindsight, that makes a lot of sense. Thanks! clever++
<{^_^}>
clever's karma got increased to 370
<OmnipotentEntity>
oh, karma system
<OmnipotentEntity>
clever++
<{^_^}>
clever's karma got increased to 371
<OmnipotentEntity>
I definitely owe you for earlier too, just wasn't really aware of it.
<jared-w>
maybe you should /nic Entity then :p
<OmnipotentEntity>
?
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bhipple has joined #nixos
<energizer>
is there anythign i can do to prevent big_data.csv from being world readable?
<gchristensen>
(OmnipotentEntity: not so omnipotent, was the joke :))
<OmnipotentEntity>
oh, that joke.
<OmnipotentEntity>
I'd ditch this nick if I hadn't had it forever.
ottidmes has quit [Ping timeout: 264 seconds]
xkapastel has quit [Quit: Connection closed for inactivity]
<OmnipotentEntity>
I thought he was saying I could workaround a restriction on increasing karma too rapidly by changing my nick, which I thought was weird.
<gchristensen>
ah hehe
<OmnipotentEntity>
Also it's Omniscient that knows everything ;D
<clever>
energizer: everything in the nix store must always be readable
butterthebuddha has joined #nixos
<OmnipotentEntity>
I'm very clearly of the almighty idiot archetype
butterthebuddha has quit [Client Quit]
mexisme has joined #nixos
<energizer>
clever: i can get around that with a user-level store?
mexisme_ has quit [Ping timeout: 246 seconds]
<clever>
energizer: you could maybe use `--store local?root=$HOME`, but that has many bugs
LysergicDreams has quit [Ping timeout: 265 seconds]
LysergicDreams has joined #nixos
GDiaX has quit [Read error: Connection reset by peer]
<jluttine>
in package derivation, is it possible to add a dependency in such a way that the resulting executable would find stuff from the dependency's share directory?
<jluttine>
now i'm just copying files in installPhase from the dependency's share directory to $out/share
<peelz>
Why does this fail with "cannot coerce a function to a string, at /nix/store/......../nixos/pkgs/build-support/mkshell/default.nix:28:3": https://termbin.com/hurf
<srhb>
peelz: List elements are separated by whitespace
<srhb>
peelz: (So, wrap it in a paranthesis)
<peelz>
srhb: oh thanks lmao. I always trip on this and forget about it
<srhb>
peelz: :)
klntsky has quit [Remote host closed the connection]
klntsky has joined #nixos
wavirc22 has quit [Ping timeout: 240 seconds]
<peelz>
Is there a reason to have shell.nix wrapped in a function? It seems to work whether it's a function or the result of "mkShell"
<bqv[m]>
essentially have an emacs lisp dsl within nix
<MichaelRaskin>
Nix is slow
<bqv[m]>
so is emacs...
<bqv[m]>
i get your point though, but it's a one-time cost
<MichaelRaskin>
You mean one-time cost of improving Nix speed?
<MichaelRaskin>
It's per-evaluation otherwise
<energizer>
i'd rather have it lookLikeNix than look-like-elisp, i like being able to write in a consistent style
<bqv[m]>
i'm not anticipating having to evaluate often
<bqv[m]>
and yes that's my plan energizer
<bqv[m]>
i'd prefer it be entirely one or the other
<bqv[m]>
and since my elisp config is quite convoluted and nebulous, that'll be quite hard with just interpolation
<terlar>
Anyone using direnv with nix? I just ran into a quite unexpected situation due to how hooks and stuff works. If you have shellHook and buildInputs etc they are exported as environment variables, this means that if you from within the direnv switched directory run another nix-shell towards another target it will inherit the current environment from the nix-shell. I found it quite unexpected
<peelz>
terlar: AFAIK that's just how the default nix-shell behaves. I don't think direnv has a special solution for it. You might wanna try nix-shell --pure if env contamination is a problem
<energizer>
terlar: also check out lorri
felixfoertsch has joined #nixos
<terlar>
I am using lorri, still has the same issue
<terlar>
Yes, --pure is working as expected, but I do want to propagate some normal env variables. Now I am thinking I have to within the direnv unexport all the nix special env vars
<peelz>
There's also --keep
<peelz>
you could alias nix-shell in your direnv lmao
<peelz>
wait actually idk if you can do that
<terlar>
Ah, didn't know that one, does that work togather with --pure or is it an alternative, which will only propagate the --keep ones?
<layus[m]>
how would `exec sway` be any different from just `sway` ?
shyim has joined #nixos
<Jank[m]>
layus: Yes, sway can be loaded from a tty
<bqv[m]>
you shouldn't need the `exec`
<bqv[m]>
that just replaces the running shell with sway
<layus[m]>
Yes, so my issue remains... strange enough.
leonardp has quit [Ping timeout: 240 seconds]
<Jank[m]>
Are you able to run an older version of sway? I have mine pinned to 1909 release because I had a similar issue that I didn't feel like debugging
<layus[m]>
Oh yes, old versions run fine.
<Jank[m]>
Just 1.4, right?
Desetude has joined #nixos
<layus[m]>
I rolled back to my nixos build from February
mananamenos has joined #nixos
<Jank[m]>
Yeah, well I've noticed that 1.4 does have an issue similar to yours as well, so you're not alone in this
<Jank[m]>
Yes, this appears to be a similar result for me. Running nixos 19.09 with just sway as an unstable package.
Mateon3 has joined #nixos
Mateon2 has quit [Ping timeout: 246 seconds]
Mateon3 is now known as Mateon2
leonardp has joined #nixos
<Jank[m]>
Hey, anyone know how I can solve `fatal error: gdk/gdkx.h: No such file or directory` for a `#include <gdk/gdkx.h>`? I'm adding `gtk2-x11` to the buildInputs, but no change.
wavirc22 has joined #nixos
<manveru>
Jank[m]: try `gtk2-x11.dev` maybe?
vandenoever has quit [Read error: Connection reset by peer]
felixfoertsch23 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
felixfoertsch has quit [Ping timeout: 240 seconds]
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
<Jank[m]>
Hmm, I swear I tried that already and the build script gave an error, but now it seems fine with it. We'll see if it finishes
o1lo01ol1o has joined #nixos
<Jank[m]>
Nope. Same error.
ajp has quit [Quit: No Ping reply in 180 seconds.]
FRidh2 has joined #nixos
phreedom has quit [Ping timeout: 240 seconds]
FRidh has quit [Read error: Connection reset by peer]
phreedom has joined #nixos
<manveru>
does your configure script maybe use pkg-config?
ajp has joined #nixos
<Jank[m]>
Yes it does
orivej has quit [Ping timeout: 256 seconds]
<manveru>
and you added pkgconfig to your nativeBuildInputs?
<Jank[m]>
I am compiling a netplay version of dolphinEmu
<Jank[m]>
yes
CMCDragonkai_ has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has quit [Ping timeout: 256 seconds]
<Jank[m]>
I am doing a `dolphinEmu.overrideDerivation` and only modifying version, src, and buildInputs so far
<manveru>
sorry, i'm really not sure what's going on then...
<manveru>
not really a C expert :|
<Jank[m]>
Alright thanks anyway
butterthebuddha has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ottidmes has joined #nixos
<Jank[m]>
Fixed that issue! Using the gtk2-x11.dev, and modifying the include statement to be `#include <gtk-2.0/gdk/gdkx.h>` solved it. Is there a more elegant way of doing this than modifying the source code?
<Jank[m]>
Well actually that doesn't even work because inside `gdk/gdkx.h` there's an include statement for another `gdk/*.h` file. How do I add `${gtk2-x11.dev}/include/gtk-2.0` to the search path for header files?
pluplog has joined #nixos
nschoe has joined #nixos
xkapastel has joined #nixos
<srk>
something like NIX_CFLAGS_COMPILE = "-L${libmysqlclient}/lib/mysql -I${libmysqlclient}/include/mysql";
<srk>
sometimes this stuff works automagically with pkg-config
<Jank[m]>
Hmmm, I'm looking through other packages that use that library and they don't seem to do that, but I also don't see how they can get it to work at all.
<srk>
interestingly it outputs a whole bunch of flags!
<boogiewoogie[m]>
hey, did anyone manage to get routersploit working? I've got it installed, however testing is done instantly without any exploits actually being checked.. :x
<Jank[m]>
How would I go about using those flags? It should be just add it to build inputs, right? Is there something weird because I'm doing an override?
<srk>
really depends on the buildsystem of the package
<Orbstheorem>
Hello, anybody using nixpkgs-wayland? I get a missing `wayland-server` when building wlroots (though I added the cachix cache, so I don't get why I building either)
<Jank[m]>
I think it's just cmake
<srk>
Jank[m]: reading previous comments it should *just* work if you have cmake and pkg-config in propagatedBuildInputs
<srk>
Jank[m]: if not it might require some patching
<ottidmes>
boogiewoogie[m]: Interesting piece of software, guess I will try it on my new router to see if there is anything to worry about
orivej has joined #nixos
<boogiewoogie[m]>
ottidmes: yep! please let me know if you can get it to work properly. :) my current expression [ https://hastebin.com/atuhavibuh ] builds fine, but when running the scan from inside the interactive shell, it instantly finishes without actually having checked any vulnerability.
wavirc22 has quit [Read error: Connection reset by peer]
<ottidmes>
boogiewoogie[m]: Thanks! when I give it a shot, I will let you know
aveltras has joined #nixos
linarcx has joined #nixos
opthomasprime has left #nixos [#nixos]
MoreTea has quit [Quit: WeeChat 1.9]
knupfer has joined #nixos
m1sosoup has joined #nixos
m0rph has joined #nixos
m1sosoup has quit [Ping timeout: 240 seconds]
zupo has quit [Ping timeout: 250 seconds]
phreedom has quit [Remote host closed the connection]
hax404 has quit [Remote host closed the connection]
jbgi has joined #nixos
<ottidmes>
is there a way to get the date of a version from a git repo, without having to clone the repo, cause then you could do: git log -1 --format=%as <commit>? It is on github so I know I could use the API to get the info, but that won't really work for a package meant to be put in nixpkgs
hax404 has joined #nixos
Cale has quit [Ping timeout: 260 seconds]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
<ottidmes>
Or rather my question would be, how do most packages get their version date in their manpages, do they hardcode them somewhere and commit that date, or can this done dynamically?
<gchristensen>
probably using SOURCE_DATE_EPOCH
Heirlung has quit [Ping timeout: 256 seconds]
kai_w has joined #nixos
wavirc22 has joined #nixos
<{^_^}>
[nixpkgs] @thoughtpolice pushed 5 commits to master: https://git.io/JvS0M
<ottidmes>
gchristensen: seems to be SOURCE_DATE_EPOCH=1, so I would still have to find a way to determine its value, kind of hoped this would be set based on fetchFromGitHub
<Miyu-saki>
I've already replicated optimus-manager using clones.
<klys>
are you using gdm3 or another dm to startx?
realrokka has quit [Ping timeout: 272 seconds]
<Miyu-saki>
I'm using lightdm.
<Miyu-saki>
But I don't have to.
<Miyu-saki>
I don't mind going to the command line, login, then manually starting X.
<klys>
thing is, it may be a matter of configuring lightdm to start two sessions
<klys>
yes, if you manually startx a second time, it will appear as :1 the second time.
<Miyu-saki>
I think that'd require a lot more hacking to the NixOS module system tho. Hm. I need to load with nvidia drivers in the other one. I'll just do the manual startx?
<klys>
reading man startx...
<klys>
there does not appear to be a way to set the .conf file from here, though startx and xinit are scripts that can be modified.
mekeor has joined #nixos
<ocharles>
mpickering: huh, never really though about that. Just not convinced it'll tell me anything useful
<mpickering>
I assume people must profile c++ applications occasionally..
<Miyu-saki>
Will give it a try after dinner. thamks
<tilpner>
Miyu-saki: I've had luck with hardware.nvidia.prime.offload, no need for xrun
<tilpner>
(Unless you want power savings, I guess)
o1lo01ol1o has joined #nixos
<ocharles>
mpickering: sure, but you want to know things like "evaluating import performance", or more abstract measurements, or something like that
<ocharles>
But maybe it'll work. I need to learn those tools though
<mpickering>
I think if you compile `nix` with debug symbols then that should get you most the way
<tilpner>
I have successfully used that in past, but I found this by going backwards in wiki history to find the repo, and the going backwards in git history to find the script
zebrag has quit [Ping timeout: 260 seconds]
<tilpner>
It's a scary script, but it worked eventually
<ottidmes>
I just now noticed that on nixFlakes I get the following hashes: sha256-K0udUU06wVlixXKIipF+Ob2322bBVeQ+mwzfdZ2xapE=, why the change?
<niksnut>
those are SRI hashes
<niksnut>
they're more standard than what we were using before
<ottidmes>
niksnut: seems like a good reason, but how will that impact nixpkgs, where everything AFAIK still uses the old format? Will we slowly move towards this format?
<Miyu-saki>
tilpner: omg thanks
<Miyu-saki>
tilpner: thank youu
zupo has joined #nixos
<niksnut>
eventually nixpkgs should switch to SRI hashes
<niksnut>
in fact it should be safe to do so now, since they've been supported for a while
<tilpner>
niksnut: Will that mean renaming sha256 attributes to "hash"?
<Miyu-saki>
Holy crap, this is amazing.
<Miyu-saki>
tilpner: Are you building a NixOS system here? lol
<Miyu-saki>
And making it work on any OS
ryantm has quit [Ping timeout: 256 seconds]
<Miyu-saki>
By setting the environment variables
<Miyu-saki>
Because, wow.
<ottidmes>
niksnut: I see 4 packages indeed already using them in nixpkgs, thanks for clearing that up :)
<niksnut>
tilpner: yes (well, replacing, not renaming)
<tilpner>
Miyu-saki: Sort of. The readme implies it's meant to also work on non-NixOS systems
ryantm has joined #nixos
_e has quit [Quit: WeeChat 2.4]
<tilpner>
Miyu-saki: It's not my script, and the repo name is "crap", so, uhh...
<tilpner>
niksnut: Of course, the values have to be changed too
Desetude has quit [Ping timeout: 272 seconds]
gustavderdrache has joined #nixos
phreedom_ has quit [Remote host closed the connection]
pluplog has quit [Read error: Connection reset by peer]
phreedom has joined #nixos
asbachb has joined #nixos
mekeor has quit [Ping timeout: 246 seconds]
<bqv[m]>
hey
<bqv[m]>
is there anything stopping me from doing `mkOption { default = mkIf (cfg.???) ???; }` ?
<Miyu-saki>
tilpner: Now, what would be nice if I can somehow wire that up with prime render offload, because eGPU is bandwidth starved.
<gchristensen>
bqv[m]: I'm ... not sure ... but at least in nixos, it is not acceptable to have the default change
Desetude has joined #nixos
hax404 has quit [Ping timeout: 260 seconds]
<tilpner>
Miyu-saki: Sorry, I can't help you with that
<bqv[m]>
hm, ok
hax404 has joined #nixos
<Miyu-saki>
tilpner: No worries!
<Miyu-saki>
It was more of just me thinking out loud
<Miyu-saki>
Also, doesn't seem to survive a vt change sadly. :/
<Miyu-saki>
But hey, it mostly works.
<tilpner>
Miyu-saki: I remember having to stop the current display-manager, switching to tty1 or 2 (not sure), running the script with a special invocation, and then switching to the tty it used for X
<tilpner>
Miyu-saki: And then something was up with the DPI, everything scaled weirdly, but that was good enough
wavirc22 has quit [Read error: Connection reset by peer]
wavirc22 has quit [Read error: Connection reset by peer]
work_ has joined #nixos
<jakobrs>
The wiki mentions it's experimental
<jakobrs>
And I've tried to get it working, but I've always failed
<ottidmes>
niksnut: I looked at flakes yesterday, but I did not use them yet because it does not seem to support dirty working directories as of yet. I get why this should not be used for published flakes, but for local development this to me would be a must have. Is support for this eventually planned?
waleee-cl has joined #nixos
wavirc22 has joined #nixos
dtz1 is now known as dtz
hax404 has quit [Ping timeout: 246 seconds]
<mjrosenb>
possibly silly question: I've installed idea-ultimate on MacOS, and it *mostly* works (because java), but it looks like the thing that got installed was the linux version.
<mjrosenb>
any ideas what I did wrong, since when I looked at the nix expr for it, I do remember seeing it attempting to install from the a dmg.
hax404 has joined #nixos
hax404 has quit [Remote host closed the connection]
<{^_^}>
nix#3231 (by Ma27, 18 weeks ago, open): Using dirty git checkouts as flakes
<ma27[m]>
those were my findings from november, I'm not sure about the current state tbh
<ma27[m]>
but I'd be fairly interested in knowing how to use dirty checkouts as well :)
<niksnut>
note that files do have to be tracked by git
<niksnut>
untracked files are invisible
<niksnut>
but tracked files can be dirty
<niksnut>
in fact you can even have dirty inputs (e.g. --override-input nixpkgs /path/to/dirty/nixpkgs)
<ottidmes>
ah, so only new files would require action, but if I just change some file, I would be able to rebuild without having to go through git first (which would be painful)
<niksnut>
right
<ottidmes>
and that --override-input would allow for untracked files, are you implying that? if so, is it also possible to define those not through the command line? (would be awesome if I can just set those up for my dev machine in a file, rather than having to wrap all calls)
phreedom has quit [Remote host closed the connection]
<niksnut>
no, that doesn't allow untracked files either
phreedom has joined #nixos
orivej has joined #nixos
<niksnut>
and yes you don't have to use the command line, you can also define a (possibly dirty) input in flake.nix, e.g. inputs.nixpkgs = { type = "git"; url = "/path/to/nixpkgs"; }
<ottidmes>
and does it copy over the git repo in some way, or does it just use the repo on the path as-is
hax404 has quit [Quit: WeeChat 2.7.1]
chagra has joined #nixos
<ottidmes>
it seemed from the description of path inputs for example, that they do copy over the whole path to the store, like "${/some/path}" does for Nix
<ottidmes>
(those paths might contain secrets in my development directory I would not like copied)
<gchristensen>
I think `hash = ` support is too new, we usually wait 1 year before using them in nixpkgs
jakobrs has quit [Quit: WeeChat 2.7.1]
<ottidmes>
Valodim[m]: AFAIK you should either use the change proposed, i.e. sha256 = "..."; or hash = "sha256-K0udUU06wVlixXKIipF+Ob2322bBVeQ+mwzfdZ2xapE="; (SRI format)
<Valodim[m]>
aight, that's a clear answer. no point having prs like that stuck due to bikeshedding
<FRidh2>
no it's ok already, we actually use it in some places
<gchristensen>
FRidh2: when did it get introduced?
<ottidmes>
gchristensen: not long ago "niksnut | in fact it should be safe to do so now, since they've been supported for a while"
<ottidmes>
jakobrs: could be just the first time, should go away if you set nix.package = pkgs.nixos-unstable.nixFlakes; (or wherever you get nixFlakes from)
<jakobrs>
the warning is just because I was using nix-shell -p nixFlakes
<jakobrs>
where nix-shell is from nixos-19.09
erictapen has joined #nixos
<{^_^}>
[nixpkgs] @karolchmist opened pull request #83361 → bloop: add blp linking to bloop → https://git.io/JvS2u
kolbycrouch has joined #nixos
<ottidmes>
I will have to start updating some scripts to support the new `nix eval` API in a backwards compatible way, I mostly use its --raw flag that is missing from nix-instantiate AFAIK, ah well, it was flagged as experimental for a reason
asbachb has quit [Quit: Ping timeout (120 seconds)]
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
cosimone has quit [Client Quit]
cosimone has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
ottidmes: You can do `nix-instantiate --eval --strict --json` and then process it with jq
<infinisil>
E.g. if it outputs a string, use `jq -r .` to un-json it
<infinisil>
(or anything really, not just strings)
<ottidmes>
infinisil: true, I have used that before, guess I will use that trick again and not rely on the nix command for now in any of my scripts
plato[m] has joined #nixos
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
<plato[m]>
"nix search --json" outputs lots of packages in JSON, but then returns an exit code 1 and thus fails. This is a bug, isn't it? It also outputs "no results found".
cjpbirkbeck has quit [Quit: cjpbirkbeck]
<plato[m]>
I'm looking at the source code for this and it's...confusing to say the least.
<{^_^}>
[nixpkgs] @bachp opened pull request #83362 → nixos/boinc: simplify setup of boinc service → https://git.io/JvS2h
<mindtree[m]>
Hi folks! Do I need to do anything in particular to enable broadcast packets?
glittershark has joined #nixos
<mindtree[m]>
I'm interfacing with an old networked LASER that informs the network of its presence by sending broadcast packets once a second, but I'm not seeing it at the moment. Just wondering if I should be checking my nix firewall config (it's just default atm) or if my issue is likely elsewhere
jumper149 has joined #nixos
LysergicDreams has quit [Ping timeout: 240 seconds]
LysergicDreams has joined #nixos
<bahamas>
can anyone point me to the code that allows nixops to deal with virtualbox? I'm creating a machine with nixops and it times out while trying to connect
justanotheruser has quit [Ping timeout: 246 seconds]
orivej has quit [Ping timeout: 250 seconds]
rsa has quit [Ping timeout: 260 seconds]
rardiol_ has joined #nixos
<jakobrs>
I'm not sure if this is NixOS-specific or not but
<jakobrs>
If I do `touch ~/.config/pulse/default.pa`,
<jakobrs>
systemctl restart --user pulseaudio.service stops working
<mindtree[m]>
Ahh, turns out I just had to enable the UDP port that the LASER was broadcasting on and the TCP port is uses for comms. Case closed!
<mindtree[m]>
* Ahh, turns out I just had to enable the UDP port that the LASER was broadcasting on and the TCP port it uses for comms. Case closed!
ottidmes has joined #nixos
erictapen has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @domenkozar pushed 4 commits to release-20.03: https://git.io/JvSVv
<cole-h>
Anybody here ever used androidenv to build an Android app? Having troubles with "accepting the license", even though I have `android_sdk.accept_license = true;` in my config.nix
lunik1 has quit [Ping timeout: 250 seconds]
<bqv[m]>
how are you consuming androidenv
erasmas has joined #nixos
<bqv[m]>
i imagine it's just not in the right file
gentauro has quit [Quit: leaving]
noudle has joined #nixos
<cole-h>
Oh, I was trying to make a shell out of said androidenv duh
jakobrs has quit [Quit: WeeChat 2.7.1]
<cole-h>
Building with `nix-build` seems to fail at the build step: "Buildfile: build.xml does not exist!"
gentauro has joined #nixos
<cole-h>
I guess androidenv doesn't support Kotlin builds yet?
cosimone has quit [Quit: Terminated!]
ddellacosta has joined #nixos
wildtrees has joined #nixos
bvdw has quit [Read error: Connection reset by peer]
zupo has joined #nixos
bvdw has joined #nixos
* cole-h
tries in straight Android Studio
gentauro has quit [Quit: leaving]
gentauro has joined #nixos
Rusty1 has joined #nixos
lunik1 has joined #nixos
opthomasprime has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @Ericson2314 pushed 5 commits to release-20.03: https://git.io/JvSw2
<lunik1>
ottidmes: will I run into the same kernel module issue linked above if I run the commands that early?
CptCaptain has joined #nixos
<ottidmes>
lunik1: not sure, I had to use it for something myself, to write to /sys/, but whether the issue also affects this place, that is for you to try
<jasom>
I'm getting some errors trying to run a program under wine, and some googling tells me that I may need 32-bit graphics drivers installed. Is there a way to do this under nixos? I have a nvidia GTX 1050 if it's driver specific.
dottedmag has left #nixos [#nixos]
<CptCaptain>
When using postgres with a copy-on-write filesystem, it is generally advised to disable COW on /var/lib/postgresql. Is there a good NixOS-style way of doing that?
<bqv[m]>
bind mount?
<bqv[m]>
add a filesystem
<CptCaptain>
Can you create btrfs subvolumes using autoFormat = true?
<watzon>
Correct me if I'm wrong, but Nix would help with the typical "installing multiple versions of Y on my PC" problem, right?
<lordcirth_>
watzon, yes, Nix is quite good at that
<watzon>
Because I seem to run into that a lot
<CptCaptain>
Valodim[m]: I mean, they are quite close to each other and NixOS is somewhere else
<MichaelRaskin>
Well, you probably end up installing them from different checkouts of Noxpkgs
<CptCaptain>
yum and apt are conceptually quite similar
<MichaelRaskin>
Nixpkgs is not too keen on having a ton of versions of something at once
<MichaelRaskin>
(in a single checkout)
<Valodim[m]>
CptCaptain: ah. yeah :)
<MichaelRaskin>
But there are no problems with using stuff from assorted checkouts
<Valodim[m]>
or having different versions of the same package per user, per directory, or just around
srid has quit [Quit: Connection closed for inactivity]
<watzon>
I need to do some reading still, but from what I've seen watching Andrew's streams you can have different local configs for different environments, right? Or am I thinking about this wrong.
<MichaelRaskin>
Yes, this is easy with Nix
<Valodim[m]>
if it looks seamless in his setup, he's probably using lorri or a similar mechanism to define dev environments with shell.nix'es
<bqv[m]>
my favourite thing is derivations that take a version as an input
<bqv[m]>
so that you can choose what version you want, and use an older one if you're happy with compiling it yourself
<noonien>
hello folks
Lysergic1reams has quit [Ping timeout: 240 seconds]
lunik1 has joined #nixos
<noonien>
does anyone have issues with nm-applet not reading vpn passwords from the keyring?
<watzon>
I'll have to look into lorri, because I want that lol
<noonien>
it seems the passwords are stored/updated, but never read
<scoates>
I'm the first to admit that my brain isn't working all that well these last couple weeks, but that definitely reads like that will only work on a reboot to me.
<LnL>
synthethic.conf only applies on reboot, but you can trigger it yourself
<scoates>
got it. thank you for spelling that out. If I manage to catch my breath, I'll submit a disambiguation PR (-:
<scoates>
(erm… unintentionally bad metaphor for the times; just meant I'm busy)
<LnL>
it's not merged yet
<watzon>
Ok I'm up and running 😁
<scoates>
ah. well it's this part that seems ambiguous to me: "This only applies on a reboot, but apfs.util can be used to trigger the creation (not deletion) of new entries."
melling has quit [Quit: WeeChat 2.7.1]
<scoates>
because the "This" appears to apply to the sentence it's in.
<watzon>
Question. How do you handle a user local configuration? I know you can do things globally in /etc/nixos/configuarion.nix, but what about on a user by user basis?
erictapen has quit [Ping timeout: 246 seconds]
<scoates>
"synthetic.conf changes are only applied on [re]boot, but to trigger an immediate, temporary (until synthetic is applied, on reboot) change, you can `/System/Library/Filesystems/apfs.fs/Contents/Resources/apfs.util -B`" something like that. I don't have my writer hat fully on at the moment. (-:
<colemickens>
How do I make sure this never happens? I don't want to ever copy store contents to my remote builder, only in reverse.
<colemickens>
Preferably, I'd just build on the remote and then ssh c@remote cachix push /nix/store/last-output, but I've been repeatedly butting heads with the CLI and not all of these things to be easy.
<colemickens>
(Especially since the nix-copy-closure speed is shockingly slow and I suspect pushing to and then pulling from cachix will be faster. and helps prevent rebuilds between communtiy box wipes)
<clever>
colemickens: you can use nix-copy-closure to copy a drv file, then run `nix-store -r` on the remote machine to build it
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.xdis: 4.2.2 -> 4.2.3 »: https://git.io/JvS67
<b42>
teteros: you can also use systemd.tmpfiles.rules for this, C to copy or L to symlink
<jakobrs>
Apparently I need to `inherit config;`?
<jakobrs>
Is there a list of all the arguments that need to be `inherit`ed?
<aswanson>
I'm attempting to hack on a python application and trying to generate a dev environment with pypi2nix. when I run `pypi2nix -r requirements.txt` I get an error about being unable to access https://github.com/pypa/pip.git due to a self signed certificate
<aswanson>
is that something wrong with my setup, or do I just need to temporarily disable git's ssl verification temporarily?
CRTified has quit [Ping timeout: 240 seconds]
ddellacosta has quit [Ping timeout: 264 seconds]
ddellacosta has joined #nixos
jakobrs has quit [Ping timeout: 240 seconds]
chpatrick has quit [Quit: Connection closed for inactivity]
<watzon>
I installed the module, but it doesn't seem to be working
<watzon>
Tried with home-manager first, then with the global config
<cole-h>
What does "doesn't seem to be working" entail for you?
kloenk has quit [Read error: Connection reset by peer]
<watzon>
Well I don't have a .zshrc, that's what I was trying to copy into my home directory
kloenk has joined #nixos
<OmnipotentEntity>
oh my zsh is fun.
pbb has joined #nixos
<jakobrs>
I'm kind of concerned that having two different versions of bluez in use at once is going to be a problem
<drakonis>
is bluez a daemon?
<watzon>
Apparently lol. I'm trying to do things the Nix way and avoid just copying my .zshrc from my dotfiles
<jakobrs>
It's a "bluetooth stack"
shibboleth has quit [Quit: shibboleth]
<jakobrs>
Not sure what that means, just that bluez 5.53 works better than 5.50 in my case and that I'm not entirely sure how to upgrade it
<watzon>
I also don't have a .oh-my-zsh directory though, so I don't know what I'd source even if I did copy my dotfiles over
<Valodim[m]>
watzon: don't go overboard though ;)
<{^_^}>
[nixpkgs] @bbigras opened pull request #83373 → pixiecore: init at 2020-03-25 → https://git.io/JvSXR
FRidh2 has quit [Quit: Konversation terminated!]
<jakobrs>
The problem is that I don't see how nixpkgs overlays would be used to replace a package with one from a different channel
<psiperator[m]>
I just got oh-my-zsh working system-wide, just added zsh and oh-my-zsh to pkgs, and programs.zsh.ohMyZsh expression with enabled = true then whatever you want to configure like themes etc after
jakobrs has quit [Quit: WeeChat 2.7.1]
<psiperator[m]>
I think its nicer system wide so you get theming and autocomplete as root etc. you can still use a custom .zshrc for user specific stuff but i guess if you want to that the nix way thats where the home manager shenanigans start
<energizer>
is there an equivalent of fetchurl but for ssh?
<johnjay>
drakonis: i'm downloading the iso right now
<johnjay>
going to test in vm first
<drakonis>
that's fair
zeta_0 has joined #nixos
<cole-h>
drakonis: There's also `man configuration.nix` as a manpage alternative to /nixos/options.html
growpotkin has joined #nixos
vandenoever has quit [Read error: Connection reset by peer]
aveltras has quit [Quit: Connection closed for inactivity]
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
shafox_ has joined #nixos
cosimone has joined #nixos
captjakk has joined #nixos
shafox has quit [Ping timeout: 264 seconds]
<drakonis>
oh yes
<drakonis>
and you can query the manpage with /
<drakonis>
it'll allow you to look up text through regex
pluplog has quit [Read error: Connection reset by peer]
shafox__ has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 pushed 6 commits to release-20.03: https://git.io/JvS1Q
shafox_ has quit [Ping timeout: 258 seconds]
shafox__ has quit [Quit: Leaving]
<energizer>
if i ever feel cool because i'm using cutting-edge tools like nix i get a helpful reminder that no, i'm a barbarian grepping manpages like it's the 70s
<gchristensen>
lol
<jared-w>
energizer: You could use a fancy manpage thing written in node if you want. 10x the bugs, 500x the disk space, and 102.5% coolness factor or something
<drakonis>
a barbarian?
<drakonis>
what.
<evils>
on that topic, i'm trying to fix apropos, infinisil suggested building the mandb cache at build-time (rather than in an activationScripts entry), would this involve making a derivation that produces it for the generated system?
<jared-w>
gchristensen: have you ever run into issues with hydra where a channel update (on nixos.org) happens, and then a personal hydra starts rebuilding a crap ton of things from scratch for no reason even though it should be in the upstream binary cache?
<jared-w>
evils: the way most package managers do apropos is by re-generating the entire mandb cache every package install, iirc
<evils>
jared-w: we can do better
<jared-w>
I should hope so :)
<nilsirl[m]>
How can I install a program for which I have a "default.nix"?
<evils>
jared-w: i can't though :(
<tilpner>
evils: Uh oh, that sounds like a bad tradeoff for most people. It takes 17s here, and I almost never use apropos
<tilpner>
Please it make it configurable if you do add it
<evils>
tilpner: you're a good nixos user, you never use apropos because it doesn't work :P
<nilsirl[m]>
clever: thank you
<tilpner>
evils: But it does, after you manually build the cache
ZoomZoomZoom has joined #nixos
<evils>
tilpner: so the idea is to build it once if your system is being built with documentation enabled
<tilpner>
I just don't want to pay the cost for an always-up-to-date cache, if it's going to add 17s to every rebuild where I add a manpnage
<jared-w>
evils: what if you used https://nixos.org/nix/manual/#chap-post-build-hook a post-build-hook that ran a script which checked if a package had man pages and, if it did, regenerated the cache?
<evils>
tilpner: on the open PR #76841 infinisil suggested building it once for each system, and on the `cfg.man.enable` switch, making `cfg.man.apropos.enable` is trivial though
<jared-w>
For more speed you could conceivably make it async and just fire a zombie process off into the aether
<infinisil>
jared-w: Not the right approach for this. post-build-hook would run for everything that's built on the system
<infinisil>
Not just installed things
<jared-w>
infinisil: ah yeh, good point
<infinisil>
jared-w: Also, it wouldn't run for things downloaded from a cache
mrpi has quit [Ping timeout: 240 seconds]
<evils>
FWIW, if the db is up to date it's nearly instant to update, so if it's an activation script and you made no changes that change the mandb it's fairly low overhead?
<jared-w>
oh, duh
mrpi has joined #nixos
<tilpner>
evils: It is not instant here. I just ran it again, 12s
<jared-w>
Hmm... A systemd timer to run the db update periodically if necessary?
<tilpner>
It's fine if it's possible to disable it
<evils>
jared-w: would that ever be necessary?
<jared-w>
evils: it means you don't need to touch the entire nix build or nixos-rebuild mechanisms at all
<evils>
tilpner: i'm going by the fact that rebuilding my system when the database already exists takes the the same amount of time as rebuilding it without apropos added
<evils>
infinisil: did you have something specific in mind when you suggested moving it to build-time?
<gchristensen>
indeed it seems the right time to index is at build-time of the system closure, since the philosophy of Nix is to move as much in to system build as possible, leaving state gunk for the very little we can't
<infinisil>
evils: Well, just that using activationScript should be avoided whenever possible. It's just impure and essentially what all other deployment tools do (massage the system into some state, as good as they can)
<evils>
ideally, one would attempt to reuse the most recent mandb output?
<gchristensen>
why should I build my apropos db on my 20 target servers instead of my 1 build host
ZaraChimera has joined #nixos
<evils>
gchristensen: because i can't think of a better way to do it :D
<infinisil>
evils: What's the problem with generating it at build time?
<evils>
infinisil: ^
<gchristensen>
does that mean you don't know how?
<evils>
jup
garbas has quit [Ping timeout: 250 seconds]
<gchristensen>
ah
<gchristensen>
sounds like a fun quarantine puzzel!
<evils>
as far as i can tell i'd want a derivation that depends on something like /nix/store/*-system-path/sw/share and outputs its db?
fling has quit [Ping timeout: 240 seconds]
<evils>
but then you have the system depending on something that depends on the system?
garbas has joined #nixos
<infinisil>
evils: You'd want to use `environment.systemPackages` probably
<infinisil>
I think using `environment.systemPackages` should work better for this
<tilpner>
There's probably a buildEnv created from systemPackages somewhre
<tilpner>
Try and reuse that
<infinisil>
Oh yeah
<tilpner>
Ahh, config.system.path
<tilpner>
Though that doesn't account for users.users.foo.packages
<infinisil>
Hm, a separate DB per user would make sense
<tilpner>
evils: See config/system-path.nix
gentauro has joined #nixos
gentauro has quit [Changing host]
ixxie has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 256 seconds]
<evils>
i think i'm missing some background, i'd want to depend on a part of the system config, but want the db be in the nix store? i've never seen anything that does that
<tilpner>
Yeah, this can collect a weird circular dependency if you're not careful
<evils>
it's not even that, sounds like i want to make a derivation that depends on a config instead of a derivation
<tilpner>
You can probably generate mandb from environment.extraSetup
<tilpner>
Though no caching then
<evils>
sounds like that'd be equivalent to what i have now
<tilpner>
No, it's very different
orivej has joined #nixos
<tilpner>
extraSetup runs after system.path is built
<tilpner>
Yours run every time I boot
<tilpner>
(And switch)
zupo has joined #nixos
<tilpner>
extraSetup runs as nixbld*, yours runs as root
xkapastel has quit [Quit: Connection closed for inactivity]
<tilpner>
If the one in extraSetup fails, the user gets an error
<tilpner>
If yours fails, the user can't boot
<infinisil>
environment.variables.MANPATH = pkgs.runCommandNoCC "mandb" {} "mandb -C $out <some way to pass in ${config.system.path} as the input>"
<infinisil>
Something like this ^
<infinisil>
Should work
<tilpner>
Huh, there's a --no-straycats
<evils>
tilpner: yea, they become an issue if there's too many /s
<evils>
mandb requires /var/cache/man to exist
zupo has quit [Client Quit]
captjakk has quit [Remote host closed the connection]
erictapen has joined #nixos
wiml has joined #nixos
linarcx has quit [Ping timeout: 250 seconds]
mehlon has joined #nixos
lsix has quit [Ping timeout: 272 seconds]
bvdw has quit [Read error: Connection reset by peer]
<betaboon>
can anyone suggest a way to add an assertion to an expression in an overlay to make sure i notice when nixpkgs has a version of a package greater-or-equal than in my overlay ?
gustavderdrache has quit [Quit: Leaving.]
captjakk has joined #nixos
<tilpner>
betaboon: foo = let newFoo = ...; in with lib; assert versionAtLeast (getVersion newFoo) (getVersion super.foo); newFoo;
gudea[m] has joined #nixos
<tilpner>
Though this won't catch equal versions
<{^_^}>
[nixpkgs] @afrepues opened pull request #83378 → Build MELPA package lists without null packages → https://git.io/JvSDH
<tilpner>
Use versionOlder and a negation for that
<betaboon>
tilpner: would it be possible to compare the rev of fetchsvn ?
<tilpner>
Uhh
<{^_^}>
[nixpkgs] @symphorien opened pull request #83379 → buildRustCrate: don't sort link flags → https://git.io/JvSD7
<tilpner>
Probably, but I have never used fetchsvn
<tilpner>
What I sent was the generic version that will work for all-ish of nixpkgs
chagra has quit [Quit: WeeChat 2.7.1]
mehlon has quit [Quit: Leaving]
<tilpner>
You can of course use foo.src.rev instead of getVersion name
<betaboon>
thank you :)
shibboleth has joined #nixos
zupo has joined #nixos
mananamenos has quit [Read error: Connection reset by peer]
hax404 has quit [Remote host closed the connection]
hax404 has joined #nixos
maddo has quit [Quit: See ya]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<benley>
anyone know how to use overrideCabal (or whatever thing) to override just the "revision" field in a hackage package? When I try I end up with sha256sum mismatches that I can't resolve because it matches either the .cabal file or the .tar.gz source but not both
<ZoomZoomZoom>
cole-h, huh, so I've added 'pkgs.' before the package name and now I get "undefined variable 'python37'" and "undefined variable 'callPackage'" errors
fendor has quit [Read error: Connection reset by peer]
<cole-h>
Can't help without seeing the full overlay ;^)
chagra has joined #nixos
<energizer>
ZoomZoomZoom: perhaps you mean pkgs.callPackage