<colemickens>
Any pipewire adventurers have working volume control in Plasma? AFAICT everything I've tried just works with the pulse compatibility socket, other than Plasma itself.
o1lo01ol1o has joined #nixos
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos
<Yaniel>
IIRC 0.3.17 was supposed to make that work, but yea the volume icon isn't showing up in the system tray
<simpson>
If it were possible to use standard `src = ...;` and derivations, like the rest of nixpkgs, then I think that a lot of the apparently-necessary complexity will melt away.
raghavsood has joined #nixos
o1lo01ol1o has joined #nixos
<Fare>
simpson, it used to and it will again, but that was the "best" way I found to be able to conditionally override the src only if the sha256 has changed.
<Fare>
when, e.g. using a local checkout in an overlay
<infinisil>
Fare: Neat! I'll probably check that out sometime soon
<Fare>
I could use something that uses the passthru to achieve the same effect
shorberg has quit [Ping timeout: 240 seconds]
shorberg_ has joined #nixos
<Fare>
infinisil, once it's stable and featureful enough, I'd love to move it to lib/ and promote its use.
<infinisil>
Frankly, I don't think that's gonna happen
<Fare>
Neither do I.
Diagon has joined #nixos
LilleCarl has quit [Ping timeout: 272 seconds]
<infinisil>
Unless it went through a successful RFC process
<Fare>
And so, in the meantime I'll play with it and use it for Gerbil.
LilleCarl has joined #nixos
<Fare>
Not worth a RFC process until it's stable and has its own variant of callPackage.
<infinisil>
nixpkgs really does need some cleanup though, so any kind of ideas for how to improve it are great
chang has joined #nixos
<Fare>
What other great features are needed beside callPackage?
<infinisil>
What scope are you thinking of?
<Fare>
All the variants of fixed-points, scopes, makeExtensible, etc., seem pretty limited to me, anyway
<Fare>
makeScope.
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<Fare>
and then, each language builds its own ad-hoc version on top
<infinisil>
Yeah that's really awful
<gchristensen>
afaik almost all of them use newScope pretty consistently, no?
<infinisil>
Even with newScope, there's many inconsistencies
<Fare>
I don't grok the bootstrap process enough to understand whether POP can help or not, but it wouldn't surprise me if it could.
<gchristensen>
truth
<simpson>
At the same time, though, I was able to hack one up for Monte in a weekend. newScope and callPackage are very very confusing, but only once per actual new scope instantiated.
spudly1 has quit [Ping timeout: 256 seconds]
<gchristensen>
nixpkgs has a bit of a "thousand flowers" thing
<Fare>
That's one thing that bugged me: some of these functions take self: super: which I understand, and some magically extract the super for the self or other shenanigans, and I'm utterly confused.
<infinisil>
Fare: Another big thing is the function call layers and their overridability. See e.g. how a python package gets called by `callPackage`, which underneath calls `buildPythonPackage`, which underneath calls `mkDerivation`, which underneath calls `derivation`, and each of the arguments to these functions should be overridable
raghavsood has quit [Remote host closed the connection]
gustavderdrache has quit [Quit: Leaving.]
xcmw has joined #nixos
gustavderdrache has joined #nixos
Lord_of_Life has quit [Read error: Connection reset by peer]
<Fare>
infinisil, I am both amazed and appalled by all these layers. Coming from Jsonnet, there's a single override mechanism for everything and special syntax for the common case `a +: b` for adding overrides to a field.
<infinisil>
I'm not sure how jsonnet is used, but nixpkgs is a beast of customizability
<infinisil>
And all these features were added because they were needed at some point (albeit without much coordination)
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JIB1r
<gchristensen>
also, jsonnet has different goals than nix with I assume different semantics
<gchristensen>
though perhaps a subset of semantics
<Fare>
there's also a feature from BETA that I like: empty arguments that can be either specialized by inheritance or by applying the pattern as a function taking inputs.
Lord_of_Life has joined #nixos
<Fare>
gchristensen, there's a big semantic overlap.
<Fare>
(and jsonnet sucks in its own ways; just very different ways than nix)
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JIB17
<simpson>
But if one actually considers how Python interpreters load modules at runtime, then the *number* of layers is correct.
hlolli has quit [Ping timeout: 246 seconds]
<Fare>
infinisil, the first thing I did in jsonnet was build a library layer with fixed points, type descriptors, etc., not too different in some ways from the structure of nixpkgs, though of course much more primitive and smaller in scope.
<Fare>
that was before I had delved deep in nix.
<Fare>
simpson: the number of layers is incorrect???
<infinisil>
Fare: Have you seen the NixOS module system yet?
leungbk has joined #nixos
<Fare>
not sure what you call "the module system"
<Fare>
so I suppose, no.
cole-h has quit [Ping timeout: 256 seconds]
<simpson>
Fare: There's one layer for Nix's primitive derivations, one for nixpkgs' common derivation technology, one for Python being a PITA, and one for calling Nix expressions from files and integrating them into the tree.
<infinisil>
The thing that evaluates nixos systems, based on options you set
<Fare>
I've used it, as a nixos user, but have never written anything or it, much less looked at the code that glues it together.
<Fare>
(ok, I might have once or twice sent a trivial patch to a module)
<infinisil>
The module system imo is one of the best abstractions you can have. As with everything, there are some flaws with it, but in general it allowed NixOS to develop into what it is now
<simpson>
Fare: BTW, when it comes to this layering, note that the two different override mechanisms work at two different layers: One is for overriding any nixpkgs derivation, and the other is for overriding how Nix expressions are loaded from files.
<{^_^}>
[nixpkgs] @pmeiyu opened pull request #106308 → unbound: Add AF_NETLINK to allowed address families. → https://git.io/JIBM4
<infinisil>
Unfortunately, the module system is too expensive to use for the package set. But I'm sure it would work wonderfully if we ignored that
<infinisil>
Well, not entirely actually
<simpson>
(The Python abstraction layer also has a kind of overriding, s.t. we can reuse the entire subsystem for both CPython and PyPy.)
<Fare>
one thing I find amazing is the convergence of ideas—all these things constantly reinvent variants of the same pattern. I talked to some of the inventors of what is now known as prototype objects and is IMNSHO best examplified by those Nix override systems, and they have no idea all the amazing things we're doing, just like we mostly have no ideas what they were doing—mostly graphical interfaces and AI, from what I can tell.
<Fare>
simpson: couldn't / shouldn't they be unified somehow?
<Fare>
That's the one thing I like about jsonnet: every attrset is actually an object, and the same mechanisms are used for all the overrides.
<simpson>
Anything is possible. But it's clearly a terrible idea, by Chesterton's Fence; you *just* implemented your brand-new system, without a full grokking of what exists.
<Fare>
Add it can work deep: { a +: { b +: c } } adds c to field b of field a of the super pattern.
Dotz0cat has joined #nixos
<Fare>
simpson, I am well aware of that, and wish to understand what I'm missing in the existing systems.
ifthenelse has quit [Quit: Ping timeout (120 seconds)]
ifthenelse has joined #nixos
<Fare>
simpson, since the existing systems aren't well documented, my approach is to build an alternative that *IS* documented, and so the old systems can be explained by contrast, the new system improved and the improvements documented too.
<simpson>
I think we've covered all of it. The existing systems factor out into layers which separate concerns. This separation is strong enough to support multiple different language ecosystems.
Lord_of_Life has quit [Ping timeout: 256 seconds]
<Fare>
simpson: covered at too large a grain to understand where things break down, how they are used, how they work inside.
<Fare>
infinisil, what makes the module system expensive?
Lord_of_Life has joined #nixos
spudly1 has joined #nixos
<infinisil>
Eating now, bbl
sigmundv__ has joined #nixos
Fare has quit [Quit: Leaving]
<pumpy>
what are you eating
sigmundv_ has quit [Ping timeout: 260 seconds]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
nixos-search#251 (by praduca, 4 hours ago, open): Searching for options have inconsistent behaviour
marek has quit [Ping timeout: 256 seconds]
Fare has joined #nixos
<Fare>
Aha, something I forgot to ask: often in configurations, a field depends not only on its parent (the surrounding extensible attrs), but on its parent's parent, etc.
anders^ has joined #nixos
<Fare>
In Jsonnet, we had to manually store the context in a field, and make sure to update that field whenever we modify the object, so it doesn't get out of synch.
<Fare>
I'm not sure what if anything has this role in Nix, and how much is automated or manual, there.
ifthenelse has quit [Quit: Ping timeout (120 seconds)]
<infinisil>
Fare: Hmm not sure what you mean again
<infinisil>
An overlay can override any attributes that the fixed-point function already defined, or define new ones
<Fare>
infinisil: let's say {a={b=pkgs.c;};} well, if you override anything anyway, you'll want the next pkgs there, not the one from the original call. Or maybe that works because pkgs is already the outermost fixed-point and we're not extending "from without", but from within?
<infinisil>
In Nix you'd declare a fixed-point function for that:
<infinisil>
> fun = self: { a = { b = self.c; }; }
<{^_^}>
fun defined
<infinisil>
Then you can write an overlay to change/add attributes:
<infinisil>
> overlay = self: super: { a = super.a // { x = 10; }; c = 20; }
<{^_^}>
overlay defined
<infinisil>
And then you can apply that with lib.extends
<simpson>
infinisil++
<{^_^}>
infinisil's karma got increased to 386
<infinisil>
> lib.extends overlay fun
<{^_^}>
<LAMBDA>
<infinisil>
> newFun = lib.extends overlay fun
<{^_^}>
newFun defined
<infinisil>
And finally, you fix the fixed-point
<infinisil>
> :p lib.fix newFun
<{^_^}>
{ a = { b = 20; x = 10; }; c = 20; }
<infinisil>
Tada!
jybs- has quit [Ping timeout: 240 seconds]
jybs- has joined #nixos
<Fare>
infinisil: that works if you always extend the outer self. Now what if I don't want to run through 1000-long attribute paths, and want to overlay "just" the inner object... and still have the far parents be updated?
<lovesegfault>
Should wrapProgram be available in postInstall?
<lovesegfault>
I'm trying to use it but I keep getting command not found
<infinisil>
Fare: Yeah that's not well supported right now, but you could do `self: super: lib.recursiveUpdate super { some.deep.attribute = 10; }`
<infinisil>
lovesegfault: You need to add `pkgs.makeWrapper` to `nativeBuildInputs`
<Fare>
my main beef with the current system is that each step along that path may require a different kind of override vs extend vs whatever other mechanism.
<infinisil>
Yeah that's pretty bad, what I showed above won't handle that at all
<Fare>
ok. Just checking that I didn't miss anything obvious.
h0m1 has quit [Ping timeout: 264 seconds]
<infinisil>
However, the NixOS module system doesn't have any problems with that
<Fare>
what does it do / how does it solve or work around the issue?
<simpson>
lovesegfault: It might have to be done in preInstall, not sure TBH.
<simpson>
Oh, wow, completely failed to notice infinisil++ with a correct answer.
<{^_^}>
infinisil's karma got increased to 0o603
<infinisil>
:)
<samueldr>
simpson: how would you wrap something that isn't there yet ;)
<lovesegfault>
infinisil: bingo bango bongo
<lovesegfault>
infinisil++
<{^_^}>
infinisil's karma got increased to 388
<Fare>
Fare++
<{^_^}>
Fare's karma got decreased to -3
<Fare>
:-)
<Fare>
Fare--
<infinisil>
Fare: The module system is based on a set of options with a known type. Users can give option definitions which are then checked against the correct type and merged properly
<simpson>
samueldr: Oh, good point. Taking a Goldilocks approach, it should be done in installPhase, then? I don't know when $out becomes immutable.
h0m1 has joined #nixos
<lovesegfault>
I don't want to touch installPhase b/c I'm using naersk
<lovesegfault>
it's easier for me to just set postInstall
<simpson>
Yeah.
<Fare>
infinisil: eventually, I would like POP to support that kind of things, too, with meta-annotations for types, method combinations, etc.
<infinisil>
Fare: So the whole option namespace is an attribute set like { environment = { systemPackages = mkOption { ... }; ... }; ... } which gets checked against a matching user input
<samueldr>
simpson: after the derivation is "done", so at no point AFAIK during the mkDerivation it'll be immutable
<samueldr>
and it's postInstall generally, since you want to have the usual install hook running
<infinisil>
Fare: This is a very simplified explanation of the module system though. It has a ton of very powerful features
<Fare>
I'm also not sure anymore what overlapping or non-overlapping a.b.c = { ... } statements do. Are they merged at the syntax level?
<infinisil>
Yeah just syntactic sugar
shorberg_ is now known as shorberg
<Fare>
CLOS (the Common Lisp Object System) has multiple inheritance, defaults, types, method combinations, multiple-argument dispatch, and MOP that can control all that and much more.
<infinisil>
Fare: Can check `nix-instantiate --parse -E '<expr>'` to see how this sugar is already undone during parsing
<Fare>
The same thing could be done for Nix, and then the module system and extensibility system could be merged.
<infinisil>
Do you know what the module system is capable of?
<infinisil>
The NixOS module system that is
sigmundv__ has quit [Ping timeout: 256 seconds]
<Fare>
Only vaguely.
<infinisil>
I mean I can't speak for CLOS, but the module system is really featureful, and I imagine almost any attempt to replace it to lack in some way
<Fare>
I know you can declare options, then somehow check and use them.
<{^_^}>
[nixpkgs] @maralorn opened pull request #106315 → rust-synapse-compress-state: init at 0.1.0 → https://git.io/JIBpf
<pjt_tmp>
fancy-ass speedy nvme server and a full clone is still taking 5+ minutes >_<
<lovesegfault>
,locate libwayland-client.so
<{^_^}>
Found in packages: wayland
<lovesegfault>
So, I have a bin that is trying to dlopen a library. How can I get that to work?
<infinisil>
,runtimeDeps
<{^_^}>
In order of preference: Patch source OR ((if it uses PATH -> wrap with new $PATH) AND (if it uses dlopen, (patchelf --set-rpath in postFixup OR wrap with new LD_LIBRARY_PATH)))
SanchayanMaity has quit [Remote host closed the connection]
<energizer>
"Boot sector info: According to the info in the boot sector, sdb1 has 1048576 sectors.. But according to the info from the partition table, it has 1046528 sectors."
SanchayanMaity has joined #nixos
mlen has quit [Ping timeout: 260 seconds]
hyper_ch4 has joined #nixos
hyper_ch2 has joined #nixos
hyper_ch5 has quit [Ping timeout: 256 seconds]
hyper_ch4 has quit [Ping timeout: 272 seconds]
SanchayanMaity has quit [Remote host closed the connection]
<ryantm>
Repost: The Final Comment Period for RFC 0080 (https://github.com/NixOS/rfcs/pull/80) has started and, barring any blocking issues, will be merged after 2020-12-15. Your opinions, comments, and approvals are welcome!
<pjt_tmp>
wish I'd found that sooner--hopfully there's an answer here. Tlsh is both a library (diffoscope depends on it) and a cmdline tool. I haven't packaged for both at once before.
Guillaum has quit [Ping timeout: 260 seconds]
<srhb>
pjt_tmp: Yeah, it looks like the sort of thing where you'd want to first do the cmake build and then generate bindings to it with toPythonModule or something.
<srhb>
But yeah, I think you'll want to coerce the cmake build into doing the right thing with mkDerivation, and only afterwards make python bindings using that function.
ddellacosta has quit [Ping timeout: 272 seconds]
<pjt_tmp>
we also have a "buildPythonApplication"--is that prefered at all?
<srhb>
It's preferred for python _applications_, that is, when it really is a python program (I'm not sure whether that's the case here)
<pjt_tmp>
the comment there implies that it's optional, but I still don't want to break stuff
hyper_ch5 has joined #nixos
<pjt_tmp>
do you know of an existing library/tool in python I could use as a reference?
hyper_ch2 has quit [Ping timeout: 240 seconds]
<srhb>
Hm, not really. I think the main problem here is that even with a mkDerivation the cmake/gnumake setup is so nonstandardthat you'll have to drive it by hand.
<srhb>
I would start there and get the binaries building.
<srhb>
In other words, I don't think a reference will help you (yet) because this package is probably like no others.
<pjt_tmp>
that's...intimidating
hlolli has joined #nixos
<srhb>
It's probably not so bad. Start by converting it to a mkDerivation and drive the build by hand according to the upstream build instructions, by filling in buildPhase, installPhase etc.
<srhb>
Once you have something that actually produces $out/bin, you can start worrying about making the packaging proper for both python and the rest of the components.
<srhb>
pjt_tmp: Sorry I can't help in more detail right now, workday has started, but if you do make an attempt, I may be able to take a look tomorrow or something. Or you could just open an issue for now, maybe someone more fluent in that tool or python can solve it quickly :)
<pjt_tmp>
srhb: That's alright. one last thing for now though:
o1lo01ol1o has joined #nixos
<pjt_tmp>
if I'm in my nixpkgs copy, is this a good way to test it builds?: nix-build -I nixpkgs=. -A python3Packages.tlsh
<srhb>
pjt_tmp: The include is redundant, nix-build defaults to default.nix in CWD.
<srhb>
pjt_tmp: But otherwise yes.
<pjt_tmp>
good
<pjt_tmp>
does `nix build` without the dash do anything useful that the dashed one doesn't?
<srhb>
I would avoid it until the experimental interface(s) stabilize.
<pjt_tmp>
makes sense
<srhb>
It'll behave very differently depending on the exact version of nix you're using, so.. Best avoid confusion :)
<pjt_tmp>
right
<pjt_tmp>
srhb: thanks for the initial help
<srhb>
You're welcome, good luck.
o1lo01ol1o has quit [Ping timeout: 260 seconds]
kalbasit has quit [Ping timeout: 256 seconds]
Guillaum has joined #nixos
hyper_ch4 has joined #nixos
mallox has joined #nixos
vidbina_ has joined #nixos
Darkmatter66_ has quit [Ping timeout: 264 seconds]
<Hash>
Tring to figure out what is up with this os
<Hash>
if you don't have python, you go nix-shell -p python
<Hash>
and you have python
<Hash>
without installing anything?
<Hash>
Am I too high?
<thibm>
Hash: it "installs" python but makes it available *only* in the ephemeral shell environment
<Hash>
So why does it even say it's not there? Why have to type this? Why not have python already for exmaple
<Hash>
Oh
<thibm>
So it's there, but you cannot access it from outside this shell. Except if you put it in your "default profile", for example
<Hash>
And how is this different from apt install python
<Hash>
Because of the hashing scheme, different versions of a package end up in different paths in the Nix store, so they don’t interfere with each other.
<Hash>
So now you're loading multiple versions of the same lib
cryptopsy has joined #nixos
<thibm>
Hash: apt will install python system-wise. All users, all shell can access *this* python once it's installed
<cryptopsy>
how do i get started with nixos?
<thibm>
Hash: with Nix, all users can choose if they want python or not, and *which* python they want
<cryptopsy>
i want to build nixos in a chroot to get the feel of it
<thibm>
Hash: yeah you can have multiple versions of the same libs/programs. That's why there are "profiles"
<thibm>
They take care of giving you access to the ones you wanted
<cryptopsy>
is nixos a compiled OS?
thc202 has joined #nixos
<thibm>
cryptopsy: compiler OS: you meant with prebuilt binaries?
<cryptopsy>
i mean like in gentoo
<cryptopsy>
i compile everything
<thibm>
ah, the opposite. Then by default, no (you get binaries), but it's very easy to compile packages (if you change flags or whatever) or even to compile everything
<thibm>
so yes and no
<cryptopsy>
how do i set up a nixos installation from a chroot on my current drive?
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<thibm>
cryptopsy: I think you can use the regular installation procedure, with `nixos-install --no-bootloader` (replace /mnt by a folder on your drive) and then `nixos-enter` (never tested that)
<pumpy>
when im installing nixos i do the basics then ssh in. weird thing though when i ssh in and edit configuration.nix, i can't right click to paste into the file using vi. but i *can* paste text into the command line so it's just a vi problem. how i can fix pls?
<thibm>
pumpy: try `:set mouse=` and try pasting again
cole-h has quit [Ping timeout: 240 seconds]
davidv7 has quit [Remote host closed the connection]
davidv7 has joined #nixos
<{^_^}>
[nixpkgs] @layus opened pull request #106333 → haskellPackages.cryptohash-sha512: fix by jailbreaking → https://git.io/JIRMR
<{^_^}>
[nixpkgs] @Yarny0 opened pull request #106340 → patchelf: add patch: false alarm for non overlapping sections → https://git.io/JIRQS
<pumpy>
i just tried something and the same password hash on a different OS i was able to use in my nix user password hash option and login with the same pw! is that expected?
<teto>
in the output schema at https://nixos.wiki/wiki/Flakes, it doesn't say which attribute is run by `nix develop`. Also is it posible to reference a flake in a subfolder of a git repository ?
ardumont_ is now known as ardumont
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #106350 → ocamlPackages: add ppxlib 0.20.0 and split ppxlib versions into different attrs → https://git.io/JI0ui
werner291 has joined #nixos
raghavsood has joined #nixos
orivej has joined #nixos
lukego has quit [Ping timeout: 240 seconds]
adamse has quit [Read error: Connection reset by peer]
adamse has joined #nixos
lukego has joined #nixos
<niksnut>
teto: devShell.<system> or defaultPackage.<system>
<niksnut>
teto: yes, e.g. 'nix run github:edolstra/nix-warez?dir=blender'
gentauro has quit [Read error: Connection reset by peer]
gentauro has joined #nixos
gvolpe[m] has joined #nixos
gvolpe[m] is now known as GabrielVolpe[m]
chang has joined #nixos
skrzyp has quit [Ping timeout: 264 seconds]
o1lo01ol1o has joined #nixos
<ptotter[m]>
how can I tell a single invocation of nix-shell to allow broken packages? (i.e. not by modifying a config file)
<teto>
I've done a rm -rf ~/.cache/nix and now nix run complains about "cannot open sqlite database ~/.cache/nix/eval-cache-v2". How can I rebuild it ?
<DavHau[m]>
<Mic92 "DavHau: I found a new use case f"> How can it mess up programs outside fhsuserenv?
jiribenes has joined #nixos
dvicente has quit [Remote host closed the connection]
<niksnut>
teto: strange, it should be created automatically
NeoCron has joined #nixos
<{^_^}>
[nix] @regnat pushed 2 commits to ca/properly-store-outputs: https://git.io/JI0MZ
<niksnut>
~/.cache/nix/eval-cache-v2 should be a directory btw (containing sqlite files)
<Mic92>
DavHau[m]: if you have different version of glibc in LD_LIBRARY_PATH than you outside program uses
<teto>
the command I run `nix run 'github:teto/neovim/flake?dir=contrib#nvim' ` with nix 2.4pre20201201_5a6ddb3
chang has joined #nixos
<teto>
there is an sqlite database in the folder eval-cache-v2 but with a different name
<teto>
maybe because the daemon differ ? I should restart ?
jonatanb has joined #nixos
pumpy has joined #nixos
<pumpy>
i just tried something and the same password hash on a different OS i was able to use in my nix user password hash option and login with the same pw! is that expected?
<teto>
arf I though --refresh could help but now I get nix --refresh run 'github:teto/neovim/flake?dir=contrib#nvim' ~/hm
<teto>
'/nix/store/nsvdk4qbjiam19vapkaasjq3i4rfkszc-source', 0, 1607438551)': attempt to write a readonly database (in '/home/teto/.cache/nix/fetcher-cache-v1.sqlite')
tcunha has quit [Ping timeout: 260 seconds]
jonatanb_ has joined #nixos
Doraemon has joined #nixos
jonatanb has quit [Ping timeout: 258 seconds]
boxofrox has joined #nixos
NeoCron has quit [Ping timeout: 258 seconds]
<teto>
because ofthe previous issue, I tried a sudo -E nixos-rebuild that recreated ~/.cache/nix and assigned it to root. now it works fine
medvid has quit [Ping timeout: 240 seconds]
chang has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<halfbit>
if I have a derivation where the actual source I want to build after configure is in a subdirectory, can I simply add like a preBuild step that cd's in?
<simpson>
I have a machine which thinks that it is currently April 2020. It thinks that most SSL certificates are invalid. The time cannot be manually set; an "Automatic time synchronization is enabled" message appears.
<simpson>
Further, asking it to be disabled with `timedatectl set-ntp false` gives "Changing system settings via systemd is not supported on NixOS". This is fine, but does this mean that I have to wait ~30min for the next NTP sync before SSL will work?
o1lo01ol1o has joined #nixos
<steveeJ>
simpson: what's the status of `systemd-timesyncd.service`?
<simpson>
steveeJ: Active, running, idle. The log starts in September, but then it thinks "system clock time unset or jumped backwards" and continues in February.
<steveeJ>
simpson: what happens if you restart it?
<gchristensen>
🎉 thanks to raghavsood we have a Darwin builder in ofborg 🎉
<raghavsood>
Happy to help give back to a community that's saved me countless hours!
<simpson>
steveeJ: Somehow it's all working again; it now knows that we're in December. Go figure, thanks. Guess it just needed a second opinion.
<{^_^}>
[nixpkgs] @mweinelt pushed 2 commits to staging-20.09: https://git.io/JIEYE
<chpatrick>
hey, do you have any advice for packaging a python module that's built by cmake?
<chpatrick>
the output dir has "__init__.py lib pmeshlab.cpython-37m-x86_64-linux-gnu.so tests2 which doesn't look like other nix python packages
<chpatrick>
and I can't import it if I put it in an environment
<chpatrick>
I'm creating it with `toPythonModule (pkgs.libsForQt5.callPackage ...`
Dotz0cat has quit [Ping timeout: 256 seconds]
<hexa->
chpatrick: have you tried adding cmake to nativeBuildInputs?
<chpatrick>
hexa-: yes, it builds fine
magnetophon has quit [Ping timeout: 240 seconds]
<chpatrick>
the problem is that the resulting directory isn't picked up by python if I want to use it in an env
<hexa->
are you using buildPythonPackage?
<chpatrick>
hexa-: no, I don't know how to combine that with cmake
<chpatrick>
because it wants to use setup.py, not building with cmake
<hexa->
format = "other";
<hexa->
(i guess)
jonatanb has joined #nixos
<hexa->
but I'm not sure, maybe the setup.py will trigger the build process
<chpatrick>
hexa-: thanks, I'll try that
<hexa->
you definitely need buildPythonPackage though
<FRidh>
that's not true
<FRidh>
buildPythonPackage is a convenience function. Most things can be done with toPythonModule and the right hooks.
<hexa->
uh ok :)
<FRidh>
chpatrick: do you have an expression you could share?
<FRidh>
the Python modules need to be in ${python.sitePackages}
jonatanb has quit [Ping timeout: 240 seconds]
<FRidh>
what's most convenient to use really depends on the build tool used
<halfbit>
how do I statically link something, do I need to wrap things in pkgsStatic or something
<halfbit>
like buildInputs = [ pkgsStatic.openssl ]; for example?
<FRidh>
$ nix build -A pkgsStatic.my_nixpkgs_pkg
hnOsmium0001 has joined #nixos
iH8c0ff33 has joined #nixos
magnetophon has joined #nixos
<halfbit>
what if its not in nixpkgs
<halfbit>
I have my own default.nix
magnetophon has quit [Read error: Connection reset by peer]
liminal18 has quit [Remote host closed the connection]
<FRidh>
you can use pkgsStatic.callPackage to call your expression
thibm has quit [Quit: WeeChat 2.6]
liminal18 has joined #nixos
<slimeee>
would someone mind explaining to me what the nix.package option entails? I read what's on the wiki, but I'm still confused. What should i set this option to? If I install nix flake, I apparently have to set it to pkgs.nixFlakes, but I don't understand what that does.
magnetophon has joined #nixos
<FRidh>
slimeee: it sets the version of nix the system (daemon) will use. By default its `pkgs.nix`. For flakes its `pkgs.nixUnstable` nowadays
cole-h has joined #nixos
cosimone has joined #nixos
<FRidh>
to enable flake support you need to set also `experimental-features = nix-command flakes ca-references` in `nix.extraOptions`
FRidh has quit [Quit: Konversation terminated!]
<slimeee>
gotcha, thanks.
Darkmatter66 has joined #nixos
<{^_^}>
[nix] @regnat pushed to ca/properly-store-outputs « Use a better prefix for the storage of realisations in binary caches »: https://git.io/JIEs2
liminal18 has quit [Quit: Leaving]
gxt__ has joined #nixos
cr4y1 has joined #nixos
gxt_ has quit [Ping timeout: 240 seconds]
malook has joined #nixos
<halfbit>
when I do pkgsStatic.callPackage like that, I get a warnings about openssl 1.0 that I have never seen
slimeee has quit [Quit: WeeChat 2.9]
<halfbit>
I guess openssl 1.1 doesn't have a static build option?
<{^_^}>
[nixpkgs] @jonringer pushed commit from @seppeljordan to master « pythonPackages.nix-prefetch-github: 4.0.2 -> 4.0.3 »: https://git.io/JIEGE
<halfbit>
unclear where that's coming from, --show-trace didn't really help me, is there a nice way of seeing what is requesting openssl 1.0 to be built?
<pumpy>
i just tried something and the same password hash on a different OS i was able to use in my nix user password hash option and login with the same pw! is that expected?
<{^_^}>
[nixpkgs] @jonringer pushed commit from @sevenfourk to master « vimPlugins.lens-vim: init at 2020-04-24 »: https://git.io/JIEcR
<simpson>
pumpy: Somewhat, yes. And it's great when it works. It can't be relied upon, sadly, but yes, those crypt/passwd hashes are designed to be forward-compatible.
<halfbit>
pkgsStatic doesn't seem to obey my overlays
<halfbit>
:(
<halfbit>
that's a bummer, is that a known thing, with a known workaround or am I just doing it wrong?
<rpgwaiter>
How should I go about troubleshooting network issues on nixos? nmcli isn't seeing my network card (Intel 9650) and I added iwlwifi to the available kernel modules.
secondry has quit [Remote host closed the connection]
gueorgui_ is now known as gueorgui
secondry has joined #nixos
<secondry>
Can anyone tell me why this derivation fails?
<secondry>
this is the important bit:
<hexa->
rpgwaiter: probably update to a newer kernel using boot.kernelPackages
<{^_^}>
[nixpkgs] @SCOTT-HAMILTON opened pull request #106369 → flask-appbuilder: fix flask-babel dep → https://git.io/JIECY
<rpgwaiter>
@hexa-: I'm on 5.9
cosimone has quit [Quit: cosimone]
<secondry>
let \n n = import <nixpkgs> {}; h = import <nixpkgs.haskellPackages> {}; ... blah blah ... buildInputs = [ "h.xmonad" ] ...
<secondry>
the \n is a literal newline in this instance, there are more, but I didn't include them
<secondry>
it says that h.xmonad does not exist... I've tried entering the identifier manually, (nixpkgs.haskellPackages.xmonad)... Still no avail. I've installed that exact path via nix-env though.
<{^_^}>
[nix-pills] @jonringer merged pull request #161 → Pill 13: Reorder words so sentence makes more sense → https://git.io/JkfLw
<{^_^}>
[nix-pills] @jonringer pushed commit from @pingiun to master « Reorder words so sentence makes more sense »: https://git.io/JIECQ
<{^_^}>
[nixpkgs] @NeQuissimus pushed to master « scala: Make overridable »: https://git.io/JIECN
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
ixxie has joined #nixos
<ixxie>
heyo folks
<ixxie>
I've got a weirdness going on which, I don't even know where to begin to lookup
sangoma has joined #nixos
<ixxie>
when I open VSCodium, and try to open a folder or file, it launches the file browser (nautilus in my case); however, it launches the file explorer with all characters of all text in the UI replaced with a rectangles
<ixxie>
I've just updated my channels and rebuilt nixos to make sure everything is uptodate
malook has quit [Quit: malook]
<ixxie>
otherwise, both VSCodium and Nautilus exhibit no issues, so I don't know how to get to the bottom of this
<rpgwaiter>
I keep getting an error when enabing drivers about needing hardware.enableRedistributableFirmware set to true, but it definitely is already set to true. Any idea what would cause this?
magnetophon has quit [Ping timeout: 246 seconds]
<pumpy>
what's the nixos command to tail firewall pass/block events pls?
o1lo01ol1o has joined #nixos
<lordcirth>
pumpy, dmesg or journalctl -k for kernel logging. Not sure if there is a way to extract just firewall events, other than grep
peelz_ has joined #nixos
peelz has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JIErc
<{^_^}>
[nixpkgs] @otavio opened pull request #106379 → [20.09] Backport Skype 8.66.0.74 from 21.03 release → https://git.io/JIErR
<gchristensen>
LilleCarl: some lawyers say there is a problem, some lawyers say there isn't. Ubuntu has really bet the farm on it not being a problem, so I feel fairly confidentthat there is no problem ... otherwise there would have been a "stop doing that" action
<halfbit>
override is the thing I need, thanks!
<{^_^}>
[nix] @matthewbauer opened pull request #4334 → WIP: aarch64-darwin build of Nix → https://git.io/JIE6t
magnetophon has quit [Ping timeout: 258 seconds]
<LilleCarl>
gchristensen: Didn't know ubuntu shipped ZoL now. Then i can go along doing the easy googling of how people interpret it in the case of Ubuntu
<{^_^}>
[nixpkgs] @jonringer pushed commit from @nyanloutre to master « pythonPackages.pyfuse3: init at 3.1.1 »: https://git.io/JIE62
<gchristensen>
LilleCarl: my (not a legal opinion as I'm not a lawyer) thinking is that if there was actually an issue with a license conflict, at this point, somebody would have sent a legal cease and desist letter to ubuntu. since it isn't, the only thing Linux's devs can do is be loud and grumpy. small dogs / big bark, etc :)
<LilleCarl>
I'm with you, was just curious. Not trying to start a GNU war :d
RickHull has joined #nixos
<gchristensen>
no worries, cool :)
<Yaniel>
isn't it the oracle lawyers that are more concerning anyway
<hexa->
pretty sure those were the ones being referred to
rtjure_ has quit [Ping timeout: 272 seconds]
<lordcirth>
There are linux devs who are concerned about it as well, but partly because of Oracle
<LilleCarl>
And Oracle isn't afraid to lawyer up considering the Java API stuff with Google
<lordcirth>
Also, "Oracle lawyers" is redundant, as Oracle is a law firm with a legacy software division
<LilleCarl>
lordcirth++
<{^_^}>
lordcirth's karma got increased to 0x9
<jmercouris>
lordcirth++
<{^_^}>
lordcirth's karma got increased to 10
<jmercouris>
lordcirth++
<{^_^}>
lordcirth's karma got increased to 12, it's a crit!
<dminuoso>
lordcirth: Oracle also has a special forces division that comes with blackhawk helicopters, rappelling onto your roof top, guns blazing, if you're missing a license.
<gchristensen>
(further oracle jokes could be had in #nixos-chat :P)
<seku>
hah
o1lo01ol1o has joined #nixos
magnetophon has quit [Read error: Connection reset by peer]
leotaku_ has joined #nixos
leotaku has quit [Ping timeout: 240 seconds]
magnetophon has joined #nixos
magnetophon has quit [Read error: Connection reset by peer]
<teto>
I would like to update the configuration/linux-kernel.xml : should I convert it to commonmark ?
<halfbit>
when setting crossSystem I'm seeing some of the nativeBuildInputs also being built which I'm a bit surprised by
mananamenos has quit [Read error: Connection reset by peer]
<eacameron>
I'm running gnome on a 4k screen and things are a bit...tiny. Is there a way to scale things up a bit. The gnome settings only let me select 200%, 300% etc. I think I need like 130%
miscalculated has quit [Ping timeout: 260 seconds]
<RickHull>
Hi, I created https://github.com/rickhull/hello-nix, which uses direnv, lorri, and niv to customize a shell environment for building projects. As noted, it's derived from the work of others, and it was mostly intended for my own reference
miscalculated has joined #nixos
davidv7 has quit [Remote host closed the connection]
<RickHull>
lately though, I've been using nix-shell with shell.nix for a similar effect, helping manage the environment for ruby projects
davidv7 has joined #nixos
mkaito has joined #nixos
mkaito has joined #nixos
mkaito has quit [Changing host]
<RickHull>
note that my hello-nix is just documentation -- effectively a blog post, not any real code other than examples. is it still up to date / basically correct?
<colemickens>
I don't recommend bothering trying under X11, but that's just me.
<kaliumxyz>
where is goBuildPackage defined?
LilleCarl has joined #nixos
* colemickens
can't even get gnome to launch right now though, so...
<kaliumxyz>
buildGoPackage*
<eacameron>
colemickens: What do you mean about not bothering under X11? Thanks BTW!
<colemickens>
kaliumxyz: a trick, you can do `rg "buildGoPackage ="` in a nixpkgs dir and find out. top-level.nix, line 16676: points to pkgs/development/go-packages/generic/
<colemickens>
eacameron: I have never, ever once, had fractional scaling work as documented or expected under X11 and usually just find myself jammed up
lordcirth_ has joined #nixos
<lovesegfault>
What's the handy way of doing "0.1.2" -> "0_1_2"?
<lovesegfault>
I remember there's something in lib but can't quite recall the name
<kaliumxyz>
thanks colemickens
<eacameron>
colemickens: What do you use instead?
sss2 has joined #nixos
lordcirth has quit [Ping timeout: 258 seconds]
<lovesegfault>
builtins.replaceStrings
dsg has quit [Quit: Lost terminal]
<colemickens>
eacameron: I'm bouncing around right now. I was in Plasma (X11, our Wayland session is apparently not working) all day yesterday. Today I'm back in Sway where I'm most comfortable. I want to get Gnome-Shell (Wayland) working, but right now ... gnome-shell-wrapped is actually segfaulting so...
<colemickens>
I'd like to spend a weekend trying to get Arcan/Durden going again because hey, it's that time of year.
<{^_^}>
[nixpkgs] @fruit-in opened pull request #106385 → vimPlugins.vim-nong-theme: init at 2020-12-03 → https://git.io/JIEyk
seku has quit [Quit: Connection closed]
<eacameron>
I just installed slack desktop client but I can't login because clicking "Sign In to Slack" opens up the resulting link page in a text editor....
<boxofrox>
Jonathan43: if I `nix-shell -p gcc -p glibc.dev`, I can see in the output of `env | grep -i flags` a nix variable that holds the path to the glibc include folder, inside which a stdlib.h resides. Hope this helps.
o1lo01ol1o has quit [Remote host closed the connection]
<Jonathan43>
One of the most annoying things about NixOS is that I can have nixpkgs build script and be just as far from building it from my own sources as before.
<Jonathan43>
I need to patch a package which is in nixpkgs for a project of my, and I need to reinvent the wheel to compile it.
<Jonathan43>
Or is there a way to do that easier?
<abathur>
to just add a patch?
<Jonathan43>
Something like that.
<Jonathan43>
I need to modify the library to use Thread Local variables...so I don't know exactly what needs changes, I need to test it.
<Jonathan43>
I was going to just clone the repo, figure out how to build it, and then test modifications.
<abathur>
there are usually multiple approaches to anything...
Acou_Bass has joined #nixos
<Jonathan43>
Yeah.
<Jonathan43>
Ah, maybe I just got something.
<abathur>
as far as literal patches go, you should just be able to override the derivation and do something like patches = old.patches ++ [ new patches ];
<abathur>
but, if you need to play around with the source for a while to know how to patch it, maybe you want to just override it to replace its source with your local source
<{^_^}>
[nixpkgs] @teto opened pull request #106395 → linux: mkDefault over the common config → https://git.io/JIENj
ransom has joined #nixos
<Jonathan43>
Yeah, that is what I was just trying to do.
<Jonathan43>
At first I tried to just use the normal build steps with a nix-shell with gcc, etc. But I ran into link errors.
Dotz0cat has joined #nixos
<Jonathan43>
Um, so I replaced the function header of the package with `{ pkgs ? import <nixpkgs> {} }: with pkgs;` and it built!
<Jonathan43>
Is that all the function parameters of nixpkgs expression are is a more explicit usage of the channel packages?
<Jonathan43>
Thanks, I have a hard time navigating the long nix manuals.
<Jonathan43>
Okay, I think I have figured out what I need.
<Jonathan43>
Thanks for you help abathur.
<simpson>
cirno-999: I don't know of a good place that talks about the maths without sounding like it's talking about JSON or other boring data structures. Instead, you might want to learn another language which manipulates similar objects in a different way, like jq (https://stedolan.github.io/jq/manual/)
<{^_^}>
[nixpkgs] @figsoda opened pull request #106397 → vscode-extensions.serayuzgur.crates: init at 0.5.3 → https://git.io/JIEpE
<tokudan>
,locate rlwrap
<{^_^}>
Found in packages: pakcs, rlwrap, monero.source
iH8c0ff33 has quit [Ping timeout: 256 seconds]
manjaroi3 has joined #nixos
manjaroi3 has quit [Client Quit]
civodul has quit [Quit: ERC (IRC client for Emacs 27.1)]
hplar has quit [Ping timeout: 260 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/JIEhj
<{^_^}>
[nixpkgs] @figsoda opened pull request #106398 → vscode-extensions.mskelton.one-dark-theme: init at 1.7.2 → https://git.io/JIEjq
<stevenroose>
So I'm going to try improve/revive an old PR to nixpkgs. How do I test-build a single pkg from nixpkgs? It'd be my first time contributing more than just a version bump.
Raito_Bezarius has quit [Ping timeout: 272 seconds]
xantoz has joined #nixos
<colemickens>
`nix-build . -A attrname` from a git checkout of nixpkgs at the pr commit, maybe rebased if it's really old