matklad has quit [Remote host closed the connection]
freeman42x]NixOS has quit [Ping timeout: 268 seconds]
freeman42]NixOS has quit [Ping timeout: 265 seconds]
<achambe>
Is there a way we can mirror tarballs.nixos.org ?
<achambe>
and is there any info about how it gets populated?
ndrei has quit [Ping timeout: 265 seconds]
<tnks>
the cudatoolkit package goes out of it's way to put one of the shared library in a separate output directory... but it's just 1MB out of 2GB of stuff.
day|flip has joined #nixos
nD5Xjz has joined #nixos
<clever>
tnks: that can help if the derivation is in the binary cache, and you only want to link to the library
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<tnks>
clever: wait... sorry it makes sense if someone wants to link to only it... I'm being silly.
<tnks>
one of these Makefiles is being really clever and finding a root path, which is getting broken, because the libraries are in two places... the larger 2GB set and the smaller one forked off.
<samueldr>
[time-appropriate greetings] I think I may need some guidance to integrate something with the nixos-homepage build
ruhatch has quit [Quit: Connection closed for inactivity]
<samueldr>
hmm, I'll come back to this later, sorry to disrupt the conversation ;)
sehqlr has quit [Ping timeout: 264 seconds]
<day|flip>
anyknow why adding xorg config leads to video driver not working at boot? did something change. cuz these two config don't work for me. https://hastebin.com/jekumokome.vbs
<tnks>
although I've done a fair share of building with Nix, I'm still not in the habit of doing builds in nix-shell... but I've actually got a super long build in front of me now, and thinking I should get with the program.
<tnks>
so I did a "nix-build -K" to get the tmp directory with the build in the right shape.
<tnks>
and I can nix-shell the derivation that failed to build.
<tnks>
but I get a touch lost at all the phases I need to run in the shell to reproduce a proper build.
<tnks>
does anyone here have a recommended set?
xcmw has joined #nixos
<vaibhavsagar>
tnks: I think you're supposed to be able to source the environment variables from a file
<tnks>
vaibhavsagar: oh yeah, I do see that environment variables file.
<vaibhavsagar>
but that's almost never worked for me in practice :)
<tnks>
wat? that makes me sad.
<vaibhavsagar>
I mean, it does have the environment variables
<tnks>
I suppose it's fine, though. I get the environment variables from a nix-shell too, right?
<tnks>
and I can go all --pure to be more careful.
<tnks>
mostly I avoid this because I'm unsure about phases.
hamishmack has quit [Quit: hamishmack]
<tnks>
I wish there was just one function I could call to run everything once in a nix-shell.
hamishmack has joined #nixos
<vaibhavsagar>
tnks: nix-shell is supposed to give you a build environment with dependencies
<tnks>
I suppose somethings are not idempotent, though.
<tnks>
vaibhavsagar: yeah, but I'd hope the environment variables in the -K dump have those dependencies too, right?
<vaibhavsagar>
yes
logzet has quit [Remote host closed the connection]
<{^_^}>
[nix-pills] @dudebout opened pull request #59 → fix rename mistake in 02de506 → https://git.io/vx7tf
ericsagnes has joined #nixos
Guest29 has joined #nixos
<{^_^}>
[nix] @andrew-d opened pull request #2059 → Fix missing $DESTDIR when installing programs → https://git.io/vx7tq
<Guest29>
Anyone use Wireguard? I've followed the Nixos Wiki, but cannot seem to get it working for the life of me.
sigmundv__ has quit [Ping timeout: 255 seconds]
<{^_^}>
[nix-pills] @dudebout opened pull request #60 → fix import issues in pill 20 → https://git.io/vx7t8
<{^_^}>
[nixpkgs] @martinetd opened pull request #38626 → linux kernel generic: use passAsFile for kernelConfig → https://git.io/vx7t2
<gchristensen>
cool PR
<gchristensen>
I think passAsFile is likely way underused
<tnks>
as nixpkgs gets larger and larger, what's the likelihood of it being split up?
<tnks>
something about this model feels untenable.
<gchristensen>
not unlikely
<tnks>
how active are discussions about it?
<gchristensen>
but it is critical the solution for that be easy to use, pleasant, and composable.
<gchristensen>
not very
<gchristensen>
its worked okay for ~15yrs. it'd be good to think of improvements, but it isn't the top of anyone's mind afaict right now
<tnks>
holy shit... that long... I didn't realize.
<Dezgeg>
wasn't nixos a separate repo at some point though, but got merged in
<gchristensen>
yeah, it was. I think merged because it was too hard to managed separately
<tnks>
I'm surprised people can follow the issue tracker at all.
<tnks>
so many concerns overloading one system.
<kuri0>
what prefix do i have to build packages for nix with ?
<kuri0>
is the default /usr ok ?
<gchristensen>
tnks: me too ...
<boomshroom>
kuri0: Each package gets its own prefix in /nix/store/
<kuri0>
boomshroom, yes but in most applications you can change the prefix when doing ./configure or cmake
<gchristensen>
set the prefix to $out
<gchristensen>
in the nix expression
<kuri0>
gchristensen, what about when doing ./config --prefix=???
iqubic has quit [Remote host closed the connection]
<gchristensen>
you should probably write a nix expression for it
<boomshroom>
kuri0: I take it you haven't used Nix for very long? It's very different from most other systems.
<kuri0>
uses it for around 2 days
<gchristensen>
kuri0: welcome aboard :)
<kuri0>
what would about if i have a proprietory application who's prefix is hardcoded /usr ?
<clever>
kuri0: it will need to be patched
<boomshroom>
kuri10: Welcome! I hope you'll come to appreciate Nix's strengths in time. It's grown to become not just my favourite package manager, but my favourite build tool.
<achambe>
kuri0: there is a wiki article of ways to do it
<achambe>
the build-fhs-userenv seems to use a chroot to support some things
<clever>
an FHS user env is almost never the answer
markus1189 has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<achambe>
sure, i mean if you have a real proprietry application you can test
<achambe>
otherwise why worry now?
markus1199 has quit [Ping timeout: 264 seconds]
<kuri0>
none of those seems to be applications which have stuff stored in /usr ?
<kuri0>
ill try that guide on the debian gimp and see what happens
<boomshroom>
kuri0: Is it a precompiled binary that specifically tries to call something in /usr, or does it try to open a dynamic library in /usr, or is it something else?
<kuri0>
boomshroom, they have stuff stored in /usr/share/programname/
<achambe>
is there some sort of way for github users to signal they are actively working on an issue?
<achambe>
it seems really frustrating that I can't see or coordinate with people working blocking issues.
<boomshroom>
kuri0: you would likely need to open up the binary and replace instances of /usr with the the store path. patchelf might be able to do that, but I don't know if it works on arbitrary strings.
<boomshroom>
achambe: Github has an assignment system, but I don't know how it works.
<kuri0>
boomshroom, yes but /usr and the nix prefix are not the same length
<kuri0>
i do this for flatpaks (/usr to /app) since its the same length
<kuri0>
i think you need LD_PRELOAD library which redirects it to the prefix
<boomshroom>
kuri0: I know. Which means you have to completely change the binary. It should be possible if you can update every pointer to things later in the file but difficult. There's a chance that patchelf could do so.
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vx7m2
<{^_^}>
→ b2f577d0 by @ryantm: mediainfo-gui: 17.12 -> 18.03.1
<{^_^}>
→ 89bbbf97 by @matthewbauer: Merge pull request #38074 from ryantm/auto-update/mediainfo-gui
monokrome has joined #nixos
<boomshroom>
kuri0: From what I can tell, pantheon is just a Gnome session. Gala seems to be the window manager it uses and a derivation shouldn't be too hard for it.
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vx7YY
<{^_^}>
→ b97550f4 by R. RyanTM: wireshark-gtk: 2.4.5 -> 2.4.6
<{^_^}>
→ 15ab41c0 by @matthewbauer: Merge pull request #38498 from r-ryantm/auto-update/wireshark-gtk
aristid has joined #nixos
Elorm[m] has joined #nixos
<boomshroom>
day|flip: You might be able to do overrideAttrs (old: { url = "${new}"; }), but I'm not sure it will work with fetchzip. You could try that.
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vx7YP
<{^_^}>
→ 7f957242 by @ryantm: oidentd: 2.0.8 -> 2.2.2
<Elorm[m]>
So I'm having this issue where ldm gets stuck in a login loop if I place a couple of startup scripts scripts in my .xsession. I'm able to login again once I delete it.
<{^_^}>
→ cd636254 by @matthewbauer: Merge pull request #38064 from ryantm/auto-update/oidentd
<Elorm[m]>
Is that the right place for them ? In .xsession ?
<boomshroom>
day|flip: By default, set elements aren't in scope when evaluating the set itself. If you want set elements to reference each other you need to add rec in front of the set: `oldAttrs: rec { ... }`
<fearlessKim[m]>
I want to forward a call like `y = {...} @args: (x args) // { additions }` but then nix says `x` hasn't received necessary arguments. How can I achieve that ?I tried removing the `{...} @` to no avail
<boomshroom>
fearlessKim[m]: What are x's arguments?
<fearlessKim[m]>
boomshroom: {config, pkgs, lib, ...} (`x` being the nixops machine generation function)
<boomshroom>
fearlessKim[m]: Then naturally those same attributes would need to be in y's argument. You can add them to y's formal parameters to catch it earlier, but what really needs to happen is that those attributes get passed in at y's call site.
<fearlessKim[m]>
boomshroom: nixops passes the arguments, I've copied the parameter names (which works) but surely there is way to bypass that
* samueldr
shouldn't put his 16:9 screen in a vertical orientation
<samueldr>
I'm writing "only one screenful" as a PR description
<boomshroom>
fearlessKim[m]: I'm not too familiar with nixops. You can try adding default parameters to those attributes in case nixops doesn't provide them.
<kreisys>
clever: I solved it :D
<{^_^}>
[nixos-homepage] @samueldr opened pull request #209 → Adds a completely re-made packages listing (explorer). → https://git.io/vx73f
srdqty has joined #nixos
<samueldr>
eek, now let's hope the wall of text doesn't scare people
<day|flip>
boomshroom: im almost there. but now im getting this: The option `myfsm' defined in `/etc/nixos/fsm.nix' does not exist.
<clever>
kreisys: now to figure out why this segfaults constantly...
<boomshroom>
day|flip: If the package is the kind that gets called with callPackage, then you should just return the derivation, not a set containing the derivation.
akkad has joined #nixos
<{^_^}>
[nixos-homepage] @samueldr closed pull request #189 → Splits the navigation in two navbars... → https://git.io/vx73q
<day|flip>
boomshroom: any links I can look at?
<day|flip>
now im not sure what to do
<kreisys>
clever: I basically just had to recreate the input derivations in the same context. it's a bit of a hack but I basically just used the same function I used to extract the tarballs from the npm registry to sort of do it again but since they're already in the store it doesn't actually end up doing anything other than the desirable side effect that it makes nix see them as derivations again
<kreisys>
now I gotta clean up all that awful code
<boomshroom>
day|flip: ` { stdenv, pkgs, ... }: pkgs.fantasque-sans-mono.overrideAttrs (old: { ... })` overrideAttrs is generally prefered to overrideDerivation, but it may be nessisary in this situation.
<kreisys>
I had to process what you said for a while (and smoke a couple bowls) to figure it out :|
<boomshroom>
kreisys: Sorry. overrideAttrs should pretty much be used whenever you can. That said, if you can't use it, overrideDerivation might work.
sehqlr has quit [Ping timeout: 260 seconds]
<Izorkin_>
help. how to use virtualisation/qemu-vm.nix ?
<kreisys>
boomshroom: oh sorry lol I was referring to something else. it would've been rude of me to say that in this context! :>
<boomshroom>
kreisys: I actually didn't realise that you weren't day|flip, so my bad!
<b5509cd>
iso sig would be nicer than https for a domain I think
<achambe>
grahamc: are there signed iso's ? (I see your gpg key on the security page)
xy2_ has joined #nixos
<achambe>
b5509cd: I think you are right, now we have to trust the developer + cloudflare + the website host. gpg would lower that to just the developers.
sehqlr has quit [Quit: Konversation terminated!]
sehqlr has joined #nixos
MercurialAlchemi has joined #nixos
ma27 has joined #nixos
ericsagnes has joined #nixos
johnw has quit [Ping timeout: 260 seconds]
hamishmack has joined #nixos
ma27 has quit [Ping timeout: 265 seconds]
<Myrl-saki>
clever: Ping
<Myrl-saki>
clever: If I build from source, I won't be able to copy my /nix/store to a different installation reliably?
fragamus has joined #nixos
<boomshroom>
Myrl-saki: If you build from source, the first thing you should do after is build nix with nix.
<Xal>
is there some way to get nix/hydra to build a statically-linked executable
johnw has joined #nixos
marusich has quit [Quit: Leaving]
rauno has joined #nixos
sehqlr has quit [Ping timeout: 240 seconds]
xAFFE has left #nixos [#nixos]
<elvishjerricco>
Xal: Yea. You can link in the `pkgs.glibc.static` derivation to statically link libc. You basically need to do this for all libraries you want to link in statically
tomberek has quit [Ping timeout: 260 seconds]
<boomshroom>
elvishjerricco: Would it be possible to create a package environment that reimports nixpkgs with the libc set to pkgs.glibc.static?
<{^_^}>
→ 6e4096d7 by @ivanbrennan: nixos/less: add configFile option
<{^_^}>
→ 53611a9b by @Mic92: Merge pull request #38629 from ivanbrennan/less-configFile
<elvishjerricco>
Xal: Hm. I guess you need to add `-L${glibc.static}/lib` to your linker flags
xy2_ has quit [Quit: WeeChat 1.9.1]
<Xal>
elvishjerricco: I need to substitute a variable in the source Makefile but I can't modify any of the sources
<Xal>
how cann I dot that?
<Xal>
for an output I would make a patch that replaces the relevant bit with @MYVAR@ and then I would substitute it in in installPhase, but I want to modify the makefile before building
<elvishjerricco>
Xal: Most variables in a make file can be set on the command line when you call `make`. You can set these from nix with e.g. `buildFlags = ["C_FLAGS=-L${glibc.static}/lib"];`. Does this help?
<elvishjerricco>
Otherwise you can always resort to patch files
<elvishjerricco>
`patches = [./mypatch.patch];`
<Xal>
that's a more reasonable way of doing it, thanks
<Xal>
I'll path the Makefile to use an environment variable
<Xal>
patch*
<{^_^}>
[nix] @IohannesArnold opened pull request #2060 → Add all dependencies to source prerequisites → https://git.io/vx7cT
<Myrl-saki>
Who here uses Nix as a build tool only?
<Myrl-saki>
User programs are on aptitude/etc, wbut builidng is through Nix.
<elvishjerricco>
Myrl-saki: I know some people at work like that
<Myrl-saki>
elvishjerricco: Reminds me of the reflex project.
<elvishjerricco>
how so?
<Myrl-saki>
elvishjerricco: Wait, no, reminds me of that blog post.
<Myrl-saki>
elvishjerricco: Also, it reminds me of the reflex project because to build it, you have to install Nix, which is kinda funny, IMO.
<Myrl-saki>
Not in a bad funny way.
<elvishjerricco>
eek. I hope no one is linking code against nix-env installed libraries. That's a worst practice for sure :P
<elvishjerricco>
Nix is definitely *primarily* a build tool for me. But it's secondly a deployment system and thirdly an operating system.
<Myrl-saki>
elvishjerricco: Do you deploy to FHS?
<elvishjerricco>
No I deploy to NixOS :P Which I guess transitively means it's in fact secondly an operating system.
<Myrl-saki>
Haha. :P
<Myrl-saki>
NixOS is pretty nice, really.
<elvishjerricco>
Agreed
<elvishjerricco>
as far as linuxes go, it's definitely my favorite.
<Myrl-saki>
I've been using Arch for like 2 years, then the remaining years were on NixOS after figuring about it.
<elvishjerricco>
But I actually kind of hate linux for any purpose other than software developement
<Myrl-saki>
elvishjerricco: What's your main OS?
<elvishjerricco>
NixOS :P My main computer use is software development
<elvishjerricco>
otherwise it'd for sure be macos
<Myrl-saki>
elvishjerricco: Ah. :P
<elvishjerricco>
MacOS is just so freakin bad for software development
<elvishjerricco>
better than windows I guess...
<Myrl-saki>
elvishjerricco: I use Linux for everything, and everything seems to work very well until audio.
<elvishjerricco>
or until nvidia :P
<Myrl-saki>
elvishjerricco: Guess who doesn't use NVidia! :D
<Myrl-saki>
Oh, yeah, OpenCL too.
<Myrl-saki>
But I think that's more of a NixOS thing.
Tucky has joined #nixos
* joepie91
has a complicated opinion of Linux
<Myrl-saki>
If I understand correctly, Nixpkgs has Mesa OpenCL disabled, so I don't have AMD OpenCL drivers.
<Myrl-saki>
joepie91: What are your opinions about it?
<joepie91>
it's like, it's good that it exists, and a lot of stuff is supported (for some value of 'supported'), and there's a lot effort going into it with the right intentions... but there is just _so much_ rotten stuff under the hood, technical debt, undocumented things that are nevertheless used widely, unnecessary duplication of work as a result of that, etc
<Myrl-saki>
Also, I'm glad that Wine works *so well*, in a non-sarcastic tone.
<joepie91>
organizationally, Linux as an ecosystem is not doing very well
jrolfs has quit [Ping timeout: 240 seconds]
jrolfs_ has joined #nixos
<Myrl-saki>
My primary usage of Wine is games, and all 32-bit games I've tried so far works.
<joepie91>
and the complexity is insane at points; not due to inherent complexity of what is being implemented, but due to poorly maintained code or duplication of effort
<joepie91>
these are fixable issues but right now it's... yeah
<joepie91>
:P
<elvishjerricco>
My main problem with the Linux ecosystem is: While I appreciate that I *can* customize everything, I really wish I didn't *have* to. Ubuntu is one of the only distros with sane defaults and it's still garbage
<Myrl-saki>
joepie91: I agree, I wonder if it's because it started out as a hobbyist project.
<joepie91>
Myrl-saki: I think it's far more complex than that
<Myrl-saki>
Like, what if it started out as an organizational project from the start.
<joepie91>
Myrl-saki: one constantly recurring theme I've seen, is that the set of 'developers who like tinkering and refining things' almost doesn't overlap at all with the set of 'developers who understand UX or the importance of documentation'
<joepie91>
you get a sort of self-selection for developers who think they're done once the code is written
<Myrl-saki>
joepie91: I think the KDE devs work pretty well on the "UX" part, even though I don't use most of their tools anymore, it may just be personal taste though.
<joepie91>
and like, no amount of organization is going to fix the widespread proliferation of that "all that matters is code" attitude
<joepie91>
Myrl-saki: eh. like, KDE does make some attempts, but there are a number of really obvious UX issues
<joepie91>
and there's nowhere near enough user testing taking place, for example
jperras has joined #nixos
<Myrl-saki>
joepie91: Luckily, I use command-line tools for the most part, but even then, there are some issues. :/
<joepie91>
one nice example of where KDE drops the ball: it seems like all UX features are considered in isolation without looking at the interactions
<fearlessKim[m]>
I guess nowadays all the components are here, just need someone to glue things (which Ubuntu did/does well). As every component moves fast, it's hard to keep up things which means the 10guys working on ubuntu desktop can't keep up
<joepie91>
you can see this in eg. the case where if you drag a link in Konversation to another window, it tries to download the file/page behind the URL and then initiate a DCC transfer
<joepie91>
which makes _no sense_ from a UX perspective, and can unintentionally leak information
<Myrl-saki>
Wow.
<joepie91>
in isolation, each of those features is justifiable - download a link when clicked and show it locally, drag a file means you want to send it, etc. - but clearly nobody ever thought through the interactions
<joepie91>
there are a lot of these kind of papercuts in KDE software
<joepie91>
and it results in way, way too many unexpected behaviours
<joepie91>
admittedly KDE has gotten better over time, but it's still nowhere near what I would consider acceptable from a 'shipping a product' perspective, really
<Myrl-saki>
joepie91: It seems to be a recurring pattern in Linux to have "easter eggs," w.r.t. features?
<Myrl-saki>
joepie91: I'm pretty sure that there's way more history involved, but you can observe the same thing with the clipboard system.
<joepie91>
oh man, clipboards are a whole different bucket of fun :P
jmeredith has quit [Quit: Connection closed for inactivity]
<joepie91>
Myrl-saki: but yeah, I'm not sure what you mean with 'easter eggs' in this context, but lack of clear documentation (for developers *and* users) on how stuff is implemented and how it behaves, ie. unexpected behaviour, can lead to these kind of issues real easy
Guest25108 has joined #nixos
<joepie91>
two particularly illustrative examples of lack of documentation around core stuff:
jperras has quit [Ping timeout: 268 seconds]
<joepie91>
* the official 'getting started with dbus' tutorial for dbus has said that it's a "work in progress" and "incomplete" for years now, and there's still no comprehensive explanation of it, despite literally every distro relying on it
<Myrl-saki>
joepie91: I guess by easter eggs, I mean "deviate from Windows" in a sense, considering that Windows is pretty much the norm w.r.t. UX. There are a lot of niche illogical interactions that people make, which others subsequently integrate to their workflow.
<Myrl-saki>
that people develop*
<joepie91>
* the freedesktop dbus notifications spec has lived in some random GNOME dev's userspace for years, and now lives on the GNOME developer site; despite being presumably a desktop-environment-independent specification that's actually a freedesktop thing, not a GNOME thing
<joepie91>
it's also been unclear for quite some time *which* of the versions of the spec was the authorative one
<joepie91>
no clear data on who implements what
<joepie91>
etc
<joepie91>
once you start digging into the constituent parts of a typical Linux system, you see this _absolutely everywhere_
<joepie91>
most of the parts have no authorative documentation at all, specs are ad-hoc and never formalized
<joepie91>
interoperability mostly consists of "you'll have to read the code for these other 5 implementations to see what they do, I guess"
<Myrl-saki>
joepie91: I guess modularity comes at this price?
<joepie91>
the documentation that _does_ exist is often poorly written, difficult to understand, uses jargon that is not defined anywhere but rather exists as institutional knowledge, outdated, poorly hyperlinked, etc. etc.
<Myrl-saki>
Not saying that it's inherent to modularity
<joepie91>
Myrl-saki: no, not at all
jluttine has quit [Ping timeout: 265 seconds]
<joepie91>
it's just that almost nobody in this ecosystem recognizes the importance of docs :)
<joepie91>
I mean, there's a reason that "just read the source code" is a trope
<joepie91>
this is genuinely the view held by many of the people working on this stuff, in my experience
<Myrl-saki>
joepie91: RTFM should be RTFS. :P
<joepie91>
the result is that you just don't have docs and everybody just tries to implement what everybody else implements
<joepie91>
with no attempt at standardization or reconciliation
<joepie91>
and you often get honed away when you try to address the lack of docs...
<Myrl-saki>
joepie91: This topic reminds me of the XCB project.
<Myrl-saki>
joepie91: Up until now, the standard for keyboard IO is still "use Xlib"
<Myrl-saki>
joepie91: I'm not sure if that's a good thing or a bad thing, maybe they're trying to discuss on a sane API for doing so.
<joepie91>
Myrl-saki: not something I've particularly closely tracked, but my experience is that a standardization process that isn't open in some way isn't really worth a damn, and that if you can't *find* the discussions on API design or standardization, then for all practical purposes they don't exist :P
jluttine has joined #nixos
<Myrl-saki>
joepie91: Do you think that "make it work first, generalize later" and derivatives contriube to this problem?
<joepie91>
Myrl-saki: not necessarily. but it's important that the "generalize and standardize" step comes before the point where people start writing alternative implementations
<Myrl-saki>
(Also, FWIW, my code is severely lacking in comments. :x)
<joepie91>
(or anything serious that ties into the existing implementation)
<joepie91>
Myrl-saki: code lacking comments can be totally fine, btw.
<joepie91>
in fact, code should generally have few comments; comments are for describing *why* the code does something, not *what* it does, and if you have to write a lot of explanations about why your code does something then that probably hints at a design issue :)
reinzelmann has quit [Ping timeout: 260 seconds]
<Myrl-saki>
joepie91: Yeah, I agree. The general rule I use for commenting is for (a) hacks to workaround a library bug (b) deviating from "good code" in order to increase speed, or (c) the algorithm that I used, with a link to the page that I took it from.
<joepie91>
yeah, those are good things to be commenting
<Myrl-saki>
joepie91: Oh, right. This reminds me, the best part of the Linux ecosystem is unmaintained code deep into the dependency tree.
<Myrl-saki>
Or *cough* left-pad *cough*
<joepie91>
Myrl-saki: btw, as an example of how a standardization process *can* work just fine in a modular ecosystem; see JS, where there are a lot of different implementations of Promises, but they are all interoperable with each other because some people wrote an 'interoperability spec' that defines the base API each implementation needs to provide to interact with others
<joepie91>
Myrl-saki: forget about leftpad, very likely everything you think you know about it is wrong :)
<joepie91>
specifically, leftpad had absolutely zero to do with dependencies, and everything with poor registry policies
<Myrl-saki>
joepie91: What do you mean by registry policies?
boomshroom has quit [Ping timeout: 265 seconds]
<joepie91>
unfortunately, the popular interpretation of that incident was, well, completely wrong, and mostly served as a sort of twisted "ha, I told you so!" attempt by people who had been ragging on JS' module system for some time without actually understanding how it works or why
<joepie91>
Myrl-saki: the problem with the leftpad incident was that npm allowed dependencies to be deleted
<joepie91>
at all
<Myrl-saki>
joepie91: Oh, right.
<joepie91>
it had nothing to do with small modules, or deep dependency trees, or whatever - it was purely and solely a policy fuckup
<joepie91>
that has since been rectified, although not entirely to my liking, but that's a separate discussion
<joepie91>
(and I have... Opinions, about npm as a company)
<joepie91>
anyway, most every other package registry got this right and just flat-out disallows deletion of packages
<joepie91>
under any circumstances, usually
<joepie91>
npm's policy now is to only allow deletion within 24 hours of publication if there's no evidence of the package being used by anybody, or something like that
<Myrl-saki>
joepie91: How about renaming packages?
<joepie91>
renaming packages is fine, provided the old name remains reserved and existing versions remain accessible at their original location
<joepie91>
honestly, package registry policy isn't that difficult; it's mostly 1) never break anybody's install process unless your *intention* is to break it (eg. for insecure packages), 2) never allow name reuse, 3) do whatever you can to prevent account or package takeovers
<joepie91>
there's a lot of bikeshedding beyond that about stuff like namespacing vs global namespaces and whatever, but that's all just that - bikeshedding
<joepie91>
the one single most important thing is "don't break stuff, don't compromise stuff"
<joepie91>
npm *seems* to follow those three points now, although it's a process
<joepie91>
Myrl-saki: as for unmaintained code deep in the dep tree... this is a problem in Linux especially with drivers, and I've been told that this has to do mostly with not having access to the original hardware anymore to test changes against
reinzelmann has joined #nixos
<Myrl-saki>
joepie91: Is the interoperability spec defined at the API level? Think, interfaces(Java)-esque?
<{^_^}>
→ a56b5d67 by @etu: php70: 7.0.28 -> 7.0.29
<{^_^}>
→ 9b59a912 by @etu: php71: 7.1.15 -> 7.1.16
<joepie91>
effectively every Promises implementation in JS-land conforms to it
<{^_^}>
→ 13000810 by @etu: php56: 5.6.34 -> 5.6.35
<Myrl-saki>
joepie91: Right. On the same note, I "wish" Haskell had an easy way of switching between String implementations, but that'd probably cause more confusion.
<Myrl-saki>
joepie91: Like, the kind of thing that would be nice to have, but will probably be problematic to implement.
<joepie91>
heh
<Myrl-saki>
Anyways, game time with friends. joepie91 Thanks for the chat.
<joepie91>
same :)
wangoe has joined #nixos
zzamboni has joined #nixos
rauno has quit [Remote host closed the connection]
jensens has joined #nixos
<telent[m]>
So who's at nixos London meeting tonight?
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Izorkin_>
How to create nix-config to build custom vm with qemu-kvm?
<telent[m]>
I was going to bring some hw for a nixwrt demo until I realised how much stuff I'd need to unplug (router, rasp pi, switch, cables, power supplies ...) & decided it'd be easier to show it remotely over ssh
<{^_^}>
→ d836d320 by Timo Kaufmann: flint: Always create proper so file, support blas
hotfuzz has quit [Ping timeout: 268 seconds]
<{^_^}>
→ 27bafa07 by @7c6f434c: Merge pull request #38120 from timokau/flint-improvements
MichaelRaskin has quit [Quit: MichaelRaskin]
<Aleksejs>
hi, is there a way to update location of .nix files for nixops deployment?
<vaibhavsagar>
Aleksejs: you can import .nix files from wherever you want, do you have an example of something you want changed?
<Aleksejs>
vaibhavsagar: I created a deployment with nixops create path/to/foo -d foo. Now I moved files to path/to/bar. Should I destroy foo deployment and re-create it or can I somehow edit path to files?
vaninwagen has joined #nixos
<vaibhavsagar>
oh, I see what you mean now, I misunderstood your question earlier
<Izorkin_>
How to create nix-config to build custom vm with qemu-kvm?
<achambe>
Izorkin_: I don't understand your question. What is a custom vm?
Alling has joined #nixos
<Alling>
How can I install Dropbox on NixOS?
civodul has joined #nixos
<achambe>
Alling: what have you tried?
<Alling>
Adding dropbox to packages works, but when I run it, Firefox opens with a login page, and when I log in, I get an error "You're using an old version of Dropbox" in said page.
<Asmadeus>
Looks like you found the minus yourself
<Myrl-saki>
yorick: Thanks.
<Myrl-saki>
I was looking at the wrong term lol
<Myrl-saki>
I was searching "string substitution" instead.
<Guest25108>
When defining an option, is something like this allowed? types.listOf (types.either A B)
<yorick>
Guest25108: I don't see why not
Guest25108 has quit [Changing host]
Guest25108 has joined #nixos
Guest25108 is now known as adamt
<adamt>
yorick: Ok, i'll try figuring what the problem is then. :P
<fearlessKim[m]>
I have to fallback to nix1 as nix2 keeps spouting "out of memory" whatever I try. I thought nixpkgs wasn't supposed to use nix 2 features but I got on rebuild `undefined variable ‘placeholder’ at /home/teto/nixpkgs/pkgs/development/libraries/glib-networking/default.nix`
<yorick>
fearlessKim[m]: fixed in nix master
<fearlessKim[m]>
yorick: I have the same problem with nix master
MercurialAlchemi has quit [Ping timeout: 260 seconds]
<Myrl-saki>
I'll compile on x86 before doing anything lol
humanoyd has joined #nixos
<Asmadeus>
Is there a way to tell what pulls in a build listed in nixos-rebuild dry-build? I don't get why it insists on building linux twice
simukis has joined #nixos
<LnL>
nix-store -q --tree /nix/store/<name>.drv
<LnL>
that will show the dependency tree
Alling has joined #nixos
<Alling>
I get disconnected every time I close my laptop, but yeah ... Dropbox folder is still empty. ^
<Alling>
^^*
<kuri0>
How much RAM does Nix require since I got out of memory when installing a package on a 1GB QEMU virtual machine
<Asmadeus>
LnL: Great... except that both are the same, except for the linux-config*.drv... and when I do the same command on the linux-configs, they're exactly the same aside of the linux-config hash itself
<etu>
kuri0: more :D
<Asmadeus>
I don't see why I would have both configs. I guess it's a side effect of the config-in-kernelPatches in boot.crashDump.enable = true?
ericsagnes has quit [Ping timeout: 268 seconds]
<adamt>
reg. nix types: What does "types.listOf (types.either A B)" actually mean? I expected a list where each individual element could be of type A or B, but it seems like it can be a list of either A or a list of B
<kuri0>
how do people use it on 1GB RPi then ?
<LnL>
Asmadeus: possibly, I guess something else is depending on the kernel without those config changes
<achambe>
kuri0: there are patches coming soon to fix some ram usage issues
<Asmadeus>
Ugh. Any magic nix-store reverse-tree command?
<kuri0>
nix still seems beta quality :\
<kuri0>
its better than all package managers otherwise
<adamt>
kuri0: You can also build the entire machine/system on a different host, and then push it to the tiny VM. :P
<LnL>
nix 2.0 was a full rewrite so there are some rough edges
<kuri0>
maybe it uses so much ram since its written in C++ while others are C
<etu[m]>
kuri0: I was setting up my raspberry pi with nixos this weekend and created a swapfile :p
<achambe>
no, its because the hash code currently is loading the whole nar into ram
<etu[m]>
Nah
<Alling>
adamt: I would also expect that to mean something like Haskell's [Either a b].
ma27 has joined #nixos
<kuri0>
has anyone tried using selinux or apparmor to sandbox nix packages ?
<Asmadeus>
Sigh, I guess I'll give up trying to set DEBUG_INFO on this kernel and just try to reproduce my bug in a non-nix VM....
<kuri0>
so that they can't read each other's configuration files
<LnL>
Asmadeus: nix-store -q has --requisites and --references but tree is only in one direction AFAIK
<Asmadeus>
Hm will give references a try
<adamt>
Alling: Maybe it's because i'm using submodules for A and B, but it seems like nix expects every element to be of the same type as the first one specified in the either. Hmm.
<Asmadeus>
Both configs are referenced by the same items as well, so meh
MercurialAlchemi has joined #nixos
<Alling>
adamt: So you mean you get like [a], not even Either [a] [b]?
<adamt>
Alling: Yeah, that seems like it.
<Asmadeus>
Anyway sorry for the rant, I love nix and have had no problems with it whatsoever until I tried building a kernel, which is something stupidely simple to do manually - the nix framework is supposed to make builds simpler not harder :/
<{^_^}>
[nixpkgs] @shlevy opened pull request #38639 → Don't bypass unsupported platforms with allowBroken. → https://git.io/vx7iD
<adamt>
Maybe it's because my types are defined like...: a = types.submodule ({ ... }: { options = { foo = mkOption .... and nix just sees everything as a submodule or something, instead of a proper type. Not sure
MercurialAlchemi has quit [Ping timeout: 265 seconds]
tmaekawa has joined #nixos
alaricsp has quit [Ping timeout: 240 seconds]
jperras has joined #nixos
warbo has joined #nixos
pkill9 has quit [Quit: WeeChat 2.0.1]
tmaekawa has quit [Quit: tmaekawa]
<Myrl-saki>
clever: Ping.
<Myrl-saki>
clever: Were you the guy who taught me, around a year or 2 ago how to use my x86 machine as a build slave for ARM?
jperras has quit [Ping timeout: 260 seconds]
<warbo>
hello, I'm running emacs --daemon as a systemd service on NixOS
<warbo>
since I live in emacs(client), I don't want it to restart when I run "nixos-rebuild switch"
<warbo>
I have "reloadIfChanged=true" and an "ExecReload" script, which seemed to work for 16.09
<warbo>
but since upgrading to 17.09 the service seems to get stopped during nixos-rebuild switch
<avn>
Works at least for system servives, I have reloadIfChanged=true on my pulseaudio unit.
<warbo>
yeah, I'm using system services for everything (single user laptop)
iqubic` has joined #nixos
<warbo>
looking through journalctl I see "stopping emacs daemon..." "Unit cannot be reloaded because it is inactive" "State stop-sigterm timed out. Killing" "Main process exited, code=killed" "unit entered failed state" "Failed with result timeout"
iqubic has quit [Ping timeout: 245 seconds]
<warbo>
I have Type="forking", so I don't know why it would be "inactive"
Guanin has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
avn has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @expipiplus1 opened pull request #38640 → asciidoctor: 1.5.6.1 -> 1.5.6.2, also add mathematical and coderay → https://git.io/vx7XX
<warbo>
hmm, maybe it's due to a name conflict since NixOS added its own 'emacs' service
<warbo>
I'll try renaming mine to 'myEmacs'
reinzelmann has joined #nixos
alexteves_ has joined #nixos
Alling has quit [Ping timeout: 260 seconds]
Tobba has joined #nixos
ThatDocsLady has joined #nixos
MercurialAlchemi has joined #nixos
tmaekawa has joined #nixos
tmaekawa has quit [Client Quit]
avn has joined #nixos
vidbina has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @7c6f434c pushed to master « nixos/tests.chromium: actually notice the tab crash »: https://git.io/vx7Ds
jrolfs_ has quit [Ping timeout: 256 seconds]
dj_goku has quit [Ping timeout: 268 seconds]
freusque has joined #nixos
TweyII has joined #nixos
<freusque>
what is the standard way to fetchTarball with a hash?
<freusque>
I want to download a gzipped file and specify a hash.
<TweyII>
I'm trying to set up the proprietary NVIDIA drivers on my Optimus laptop. I've been trying this for a while, but with no luck — I get to initializing the X11 server in my boot process, but then everything hangs and won't respond (to ctrl-alt-del, &c.). I was looking back through my logs and saw: Apr 09 12:15:42 uruz systemd-modules-load[683]: Module 'nvidia' is blacklisted
<achambe>
freusque: what did you try?
<TweyII>
Why would nvidia be blacklisted? I don't have anything in my configuration that explicitly blacklists it
<freusque>
achambe: I tried fetchurl and fetchzip
<freusque>
I'm wondering if there is a trivial build helper for this somewhere
<freusque>
I didn't find anything by code searching gunzip in nixpkgs
<achambe>
lol
coot_ has quit [Quit: coot_]
<freusque>
:-)
<achambe>
I recently used builtins.fetchTarball and specified a sha256
<{^_^}>
→ 346c0ac3 by @shlevy: nix-daemon: Exit successfully when interrupted.
<{^_^}>
→ 3ec1b3da by @edolstra: Merge pull request #2061 from shlevy/nix-daemon-interrupt
fendor has joined #nixos
<Myrl-saki>
TimePath: It doesn't. I'm going the Gentoo route.
<dbe>
What do I do in a mkDerivation to install a file with a file ending? Setting installPhase to 'cp myfile.txt $out' works fine, but I'd like to store the result in /nix/store/hash-file.txt, not /nix/store/hash-file.
<shlevy>
Myrl-saki: would love to see some benchmarks
<Myrl-saki>
shlevy: "Barely any change," probably.
ndrei has joined #nixos
<Myrl-saki>
shlevy: Might be more useful in the Raspi though.
<TimePath>
Myrl-saki: nixos is simultaneously a better gentoo and arch :)
<Myrl-saki>
TimePath: A better Arch, sure, but Gentoo, IMO not.
<Myrl-saki>
TimePath: IMO, Gentoo has better "USE_FLAGS," than NixOS.
<TimePath>
Myrl-saki: system configuration still makes it better imo
<Myrl-saki>
these paths will be fetched (186.27 MiB download, 382.85 MiB unpacked):
<Myrl-saki>
That's amazing.
<TimePath>
I thought 32GB would be heaps
<Myrl-saki>
I didn't know nixos-rebuild's dependencies are that... large.
<gchristensen>
Dezgeg: that should probably not be done by root anyway
<dbe>
I'm stupid. The name should contain the file ending. Got it. (for logs)
<makefu>
musicmatze[m]: i was referring to the question of lewo regarding fetch from signed git branch to rebuild nixos -> module
<Dezgeg>
that is true (and maybe it isn't)
zzamboni has quit [Ping timeout: 260 seconds]
<lewo>
musicmatze[m]: do you already implement a such module?
ashgillman has joined #nixos
knupfer has quit [Remote host closed the connection]
lord| has quit [Quit: WeeChat 2.1]
knupfer has joined #nixos
<musicmatze[m]>
Ummm... No, I don't. I don't know who proposed this for the timeline.
zzamboni has joined #nixos
<gchristensen>
Dezgeg: I think it is :/ the old perl script was definintely running as root
<lewo>
musicmatze[m]: hmm, it could be me... but it's not the case:/
jperras has joined #nixos
<makefu>
:D
xcmw has joined #nixos
winem_ has joined #nixos
ashgillman has quit [Ping timeout: 245 seconds]
Mercuria1Alchemi has joined #nixos
<lewo>
musicmatze[m]: I had added a "preStart" option to the autoUpdate nixos module to pull signed commits before rebuilding the system (but never submitted upstream)
Mercuria1Alchemi has quit [Client Quit]
knupfer has quit [Ping timeout: 260 seconds]
Mercuria1Alchemi has joined #nixos
Mercuria1Alchemi has quit [Client Quit]
knupfer has joined #nixos
jperras has quit [Ping timeout: 240 seconds]
Mercuria1Alchemi has joined #nixos
MercurialAlchemi has quit [Ping timeout: 255 seconds]
tmplt_ has quit [Ping timeout: 264 seconds]
rauno has joined #nixos
seafood has joined #nixos
<{^_^}>
[nixpkgs] @magnetophon opened pull request #38644 → faust: add more faust2appl scripts → https://git.io/vx75U
zzamboni has quit [Quit: Leaving.]
zzamboni has joined #nixos
<musicmatze[m]>
lewo: I'm not sure why you're telling me this? I have nothing to do with the autoUpdate functionality!
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<makefu>
musicmatze[m]: i think because of the agenda of the nixos meetup where i guessed that you had something to do with that :D
<warbo>
yay, I solved my issue! Turns out my emacs service had local-fs.target in its "requires", and nixos-rebuild switch was stopping and starting the filesystem targets (local and remote), hence why emacs was getting stopped :)
zzamboni has quit [Quit: Leaving.]
<warbo>
(in case it's useful, since I know the channel gets logged) I tracked it down by digging through the nixos-rebuild script, to see that it does 'nix-build "<nixpkgs/nixos>" -A system' to build the switch-to-configuration script; built that and executed it with 'sudo perl -d /path/to/switch-to-configuration test', which let me step through the Perl code
zzamboni has joined #nixos
<avn>
btw folks, `modern` `nix` command use nixpkgs. prefix for packages, `nix-build` uses `pkgs.` (I haven't use channels, only git checkout)
<avn>
how I can correct it, and have similiar prefix to both commands?
<avn>
(alsi I curious, how I can disable fancy output in `nix build`)
rosa has joined #nixos
<warbo>
avn: as far as I'm aware, 'pkgs' is just the name of a set, e.g. nix-build -E 'with import <nixpkgs> {}; pkgs.bash'
<warbo>
avn: so if you want the word 'nixpkgs' to work in the same way as 'pkgs', just define a set called 'nixpkgs'
<warbo>
you could do this in ~/.nixpkgs/config.nix or /etc/nixos/configuration.nix
<warbo>
just put e.g. 'nixpkgs = pkgs;' somewhere in the top-level overrides
ilyaigpetrov has joined #nixos
zzamboni has quit [Ping timeout: 256 seconds]
<avn>
well, toplevel overrides looks interesting idea. I just want to have similiar attrpath when use both tools
rauno has quit [Ping timeout: 260 seconds]
<mg->
To upgrade is still just nix-channel --add https://nixos.org/channels/nixos-18.03 nixos and nixos-rebuild switch --upgrade right? And I still want to keep system.stateVersion = "16.09"; in /etc/nixos/configuration.nix ?
<tazjin>
Is there something that can be set in a derivation to tell Nix that the build output is uncacheable?
<achambe>
mg-: good question, the default comment tells us to wait updating the state version until the release notes say
<achambe>
but the release notes never say
<warbo>
tazjin: you could put something like { cacheBuster = with builtins; toString currentTime; }
<{^_^}>
[nixpkgs] @Profpatsch opened pull request #38647 → modules/profiles/minimal: sound is disabled by default → https://git.io/vx7pT
<achambe>
mg-: actually, the release notes do say, you can have a state version of 18.03 now
<achambe>
but you should check the notes
leat has quit [Ping timeout: 260 seconds]
<warbo>
tazjin: the nice thing about currentTime is that it's set once when Nix starts evaluating, so even if it takes ages to evaluate all occurrences will be the same
<{^_^}>
[nixpkgs] @Mic92 opened pull request #38648 → CODEOWNERS: add Mic92/LnL7 for rustc → https://git.io/vx7pC
<mg->
achambe: hm. When I upgraded to 17.09 I was told to keep it at 16.09. Do you have a link?
<zybell_>
mg-: system.stateVersion is about the directory layout you installed nix with, meaning:where the installer put important files. Unless you want to move these files around to their new locations I would think it better to keep the number which unfortunately looks like the version which choosed the directories.
coot_ has joined #nixos
hihello has joined #nixos
marcinkuzminski has quit [Quit: ZNC - http://znc.in]
deba5e12 has quit [Quit: WeeChat 1.9.1]
<hihello>
Hi, can anyone point me to something that can help me to accomplish this -> I want to be able have the latest gnome3(everything) while being on 18.03?
jperras has quit [Ping timeout: 265 seconds]
<WhittlesJr>
Hey guys. My /etc/os-release is empty on some RPi3s that were deployed with NixOps, and I can't use them as remote builders as a result. Any thoughts?
<{^_^}>
→ 2d01bb7f by @endgame: haskellPackages.blank-canvas: disable tests
<{^_^}>
→ c2ce7070 by @endgame: Whitespace lint
<{^_^}>
→ 62e06d98 by @peti: Merge pull request #38613 from endgame/blank-canvas-dontcheck
<WhittlesJr>
gchristensen: I get "machine-name> generating new SSH keypair... done machine-name> warning: cannot determine NixOS version". Then I later get "error: a 'aarch64-linux' is required to build '/nix/store/....drv', but I am a 'x86_64-linux'".
<WhittlesJr>
gchristanses: But my nix.buildMachines and nix.distributedBuilds is configured correctly
<kreisys>
Is there any documented best practice on how to deal with build-time secrets so that they don't end up in the store?
<gchristensen>
... weird, WhittlesJr ...
<hihello>
goibhniu: environment.systemPackages = [ unstable.PACKAGE_NAME ]; -> Is there like an alias for all the gnome package pulled by setting gnome3 to default?
<goibhniu>
sorry hihello, I don't know
rosa has quit [Quit: rosa]
<Dezgeg>
hm, who is responsible for setting up /etc/os-release? the installer or nixos-rebuild switch?
<zybell_>
kreisys:there shouldn't be any build time secrets.
<hihello>
goibhniu: Thanks anyway. Guess I'll get to searching. :)
<gchristensen>
kreisys: you can use impure environment variables ... impureEnvVars (https://nixos.org/nix/manual/) but ideally don't have them
<WhittlesJr>
gchristensen: So I'm trying to deploy to these RPi3s from an x86_64 using the RPi3s themselves as builders but it's not working. And my ssh connectivity is fine.
rosa has joined #nixos
<gchristensen>
Dezgeg: activation time (switch)
<LnL>
WhittlesJr: what's the build that's failing?
<samueldr>
I personally hardcoded a path to something like /etc/secrets/some_file in the compiled package, and manage it without nix, kreisys, though I too would like to see what others are doing
<Dezgeg>
ok, not the rpi image build scripts to blame then
<kreisys>
zybell_: in real life there have to be sometimes... just like we can fetch a tarball using a fixed output derivation we should be able to fetch a tarball that's hosted behind an auth wall
<zybell_>
kreisys:Thats a fetch time secret;-) How about prefetching?
<gchristensen>
yeah, best bet IMO is to fetch them before you run nix-build. it isn't pretty, but alas ...
<hihello>
goibhniu: I'm slow... Two words -> Package overides
<gchristensen>
think though about the implications of the artifacts you're fetching ending up in the nix store, and the value of the credentials preventing you from fetching them without creds in the first place
<kreisys>
zybell_: I was looking into using builtins.fetchurl (which is supposed to take place during eval time I believe?) but that had 2 problems... 1) can't specify http headers and 2) only supports sha256 while I need to use several types of hashes (because npm -_-)
silver has joined #nixos
<kreisys>
I'd rather avoid having to prefetch manually prior to building because I want my devs to actually adopt this
<zybell_>
A ssh-key can be avoided by manually loging in with ssh as master session and having ssh config set up to slave to that master session.
joe3 has quit [Ping timeout: 256 seconds]
<WhittlesJr>
LnL: Sorry, that's not evaluating for me... I assume you want system to be "aarch64-linux", but I'm not sure what else I would have to change for that to evaluate
rosa has quit [Quit: rosa]
<kreisys>
zybell_: do you have a link detailing that method by any chance?
<LnL>
WhittlesJr: yeah system = "aarch64-linux"; in your case :)
hihello has quit [Ping timeout: 260 seconds]
<woffs>
Hi. I'm trying to fix 18.03 on i686 for me. How to override gcc = gcc6 for qt and haskellPackages.hslua?
jperras has joined #nixos
adamt has quit [Ping timeout: 260 seconds]
<zybell_>
I did it to login to web space that didn't allow authorized-keys and found out about it from 'man ssh_config'/ControlMaster . I worked it out by myself so I cant give you a link atm. If you really need one, holler and I will google around, but it will take a while.
jrolfs_ has joined #nixos
fendor has quit [Remote host closed the connection]
<shlevy>
puffnfresh: About to review nix-script-store-plugin and possibly integrate it into nix-plugins. Can you add a license if you don't have an objection? Also, would you object to renaming it command:// or stdio:// or some such?
fendor has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to master « multi-ghc-travis: update to latest git version »: https://git.io/vx5TK
szicari has quit [Quit: szicari]
xcmw has joined #nixos
szicari has joined #nixos
xAFFE has left #nixos ["Rechner geht in den Ruhezustand"]
selaux has quit [Remote host closed the connection]
<{^_^}>
→ 399f43c3 by @samueldr: nix-shell: Fixes use with ruby shebangs.
<{^_^}>
→ 9478f886 by @samueldr: Adds `nix-shell` test for special-cased ruby interpreter.
<{^_^}>
→ a4c9b259 by @samueldr: Hardcodes `nix-shell` instead of `/usr/bin/env nix-shell`...
<{^_^}>
[nix] @shlevy merged pull request #2056 → nix-shell: Fixes use with ruby shebangs. → https://git.io/vxQlp
ryanartecona has joined #nixos
xcmw has quit [Client Quit]
joe3 has joined #nixos
adamt has joined #nixos
adamt is now known as Guest17535
<WhittlesJr>
LnL, gchristensen: Sorry, I'm a little stupid. The distributed build problem was my fault (had the config in the wrong place). But the empty os-release is real. Not sure if it matters. Sorry for the noise.
<WhittlesJr>
Well I got further with my RPi3 deployment, but... https://pastebin.com/17PmUCw6. It's expecting hashes of 0 for everything built on the RPi3s.
Guest17535 has quit [Ping timeout: 256 seconds]
knupfer has quit [Ping timeout: 264 seconds]
troydm has joined #nixos
<telent_[m]>
```
<telent_[m]>
install: cannot stat '/nix/store/pbfbn27lwkrr2305i0cnxhbs6hpg7rlc-Libsystem-osx-10.11.6/bin/getconf': No such file or directory
<telent_[m]>
```
<telent_[m]>
trying to install rabbitmq_server on darwin, any ideas?
<telent_[m]>
macos 10.13.4 apparently
<telent_[m]>
(sudden thought: I hope the matrix->irc gateway does something sensle with markdown formatting)
<gchristensen>
(it does not)
Guest17535 has joined #nixos
<telent_[m]>
sorry. ok, well, the bit inside teh backticks is an error message from running nix-shell -I nixpkgs=$HOME/nixpkgs -p rabbitmq_server
warbo has left #nixos [#nixos]
<telent_[m]>
where $HOME/nixpkgs is current HEAD checkout of nixpkgs from github
<kreisys>
zybell_: whoa thanks that's actually awesome! it really cuts down on those git fetches (got quite a few of those)
winem_ has quit [Ping timeout: 240 seconds]
fendor has quit [Ping timeout: 265 seconds]
mmercier has joined #nixos
<Dezgeg>
WhittlesJr: try running nix-store --verify --check-contents on the rpi
<Dezgeg>
I probably need to fix the rpi3 image building scripts to have hashes in the initial database
<WhittlesJr>
Dezgeg: Thanks, I'll try that. And I assume I'd have to run that on every pi on the network (since they're all being used as builders?)
<ocharles>
How do I generate a sha256 to give to builtins.fetchTarball?
<Dezgeg>
yes
<ocharles>
Note that when obtaining the hash with nix-prefetch-url the option --unpack is required. aha
mizu_no__ has joined #nixos
<WhittlesJr>
Dezgeg: OK. And is adding the --repair flag optional or necessary? (Or even recommended?)
<telent_[m]>
I still do date | sha256sum then run the derivation to get an error message then paste the actual sha256 in :-)
<Dezgeg>
hmm
<Dezgeg>
I can't remember right now
spear2 has quit [Ping timeout: 260 seconds]
<{^_^}>
[nix] @shlevy pushed to master « Export required C++ version in pkgconfig. »: https://git.io/vx5mM
<WhittlesJr>
Dezgeg: ok. my assumption is that --repair is for redownloading wrong things. But the problem here is that all of the hashes are just missing, I think? (Pretty much everything is saying "fixing missing hash" and "updating size field")
<Dezgeg>
I think it should just be recomputing the hash
mizu_no__ has quit [Client Quit]
<Dezgeg>
not downloading anything
<Dezgeg>
if --repair makes it download things then don't pass --repair
<WhittlesJr>
Dezgeg: Got it.
fendor has joined #nixos
jtojnar has quit [Remote host closed the connection]
<WhittlesJr>
Dezgeg: It seems to only redownload if it computes a hash that is not what it should be. For missing hashes it just computes the hash.
<Dezgeg>
I don't know why hash mismatches are happening though
knupfer1 has joined #nixos
<WhittlesJr>
Dezgeg: I do get a handful, and they're all related to python2.7-azure-mgmt-*****
kmicklas has joined #nixos
<Dezgeg>
hmmh
knupfer has quit [Read error: Connection reset by peer]
knupfer1 is now known as knupfer
pacmeister has joined #nixos
<pacmeister>
Hello
<pacmeister>
Can I ask general noobie questions here or is this IRC for development?
<shlevy>
pacmeister: we have #nixos-dev :) Ask away!
<WhittlesJr>
Dezgeg: That did the trick! Thank you.
sehqlr has joined #nixos
<pacmeister>
cool
<fendor>
morning, got problems with stack and gloss, when executing a gloss applications, i get the error: https://hastebin.com/buqigovoke.pl
<Dezgeg>
ok, great. it should get fixed for new images soonish
Xyoff is now known as unacceptable
<pacmeister>
does nixos care if i create my own configs in my home directory? e.g. installing oh-my-zsh without nix-env, or creating my own .desktop files in ~/.local/share/applications
<pacmeister>
like will it break something?
<telent_[m]>
hmm, looks liek redis has the same issue on darwin. I am speculating but
<Myrl-saki>
clever: I guess I'll figure out soon? Do the nixpkgs of both machines have to be the same, or will the .drv of the RasPi get passed to my desktop?
<clever>
it will push the .drv over
<Myrl-saki>
clever: Thanks for the help.
<clever>
yep
<Myrl-saki>
clever: Do I have to add /etc/nix/machines to NIX_PATH?
<clever>
no
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #38653 → home-assistant: make hass cloud integration work → https://git.io/vx5ZM
jensens has quit [Ping timeout: 240 seconds]
<zybell_>
fendor:Probably not. What is stack for? Is it a script or binary? How is the *full* command you give? What is the meaning of flags you give? Why is there a second DISPLAY in the picture? Obviously you are doing something not right, please take care to give helpers as much and *accurate* info as you can. Your helper may be behind a small bw connection(as I am), that prevents him to download your sw to play with it. Maybe he doesn't want to
<zybell_>
install any sw he himself doesn't need. You ask now for half a day and nobody touches that question. In your place I would think about,why not.
telent has quit [Ping timeout: 268 seconds]
dan_b has quit [Ping timeout: 265 seconds]
vidbina has joined #nixos
<fendor>
stack is a build tool for haskell projects. i can provide an example via git. I dont know why any of that stuff is happening, i am pretty sure it is some config issue involving nixos, stack and glut/opengl, and i also found some resources about it.
<fendor>
i dont know what information i should provide thats why i am pretty vague
coot_ has quit [Quit: coot_]
jrolfs has joined #nixos
<Myrl-saki>
clever: How to verify that the remote build is working?
jrolfs has quit [Client Quit]
<clever>
Myrl-saki: check to see if gcc is running on the remote machine
<Myrl-saki>
clever: Doesn't seem so.
<Myrl-saki>
clever: Oh wait, your module is still compiling.
<Myrl-saki>
clever: I'll report back once it's done
<zybell_>
ok, as a build tool stack itself sholdn't touch a XServer, probably some build script. Since opengl is involved I would suspect some build rule for opengl. Normal build rules sholdn't touch X either, but tests can. That would also explain the second DISPLAY. Do you need the tests?
<fendor>
zybell_, i am not executing any tests, when executing the program, it dies as soon as it tries to open a window
knupfer has quit [Remote host closed the connection]
<zybell_>
But *why* does it try to open a window? I do *not* see any reason it should. Btw I would like to see the full command line.
<clever>
fendor: i suspect that whatever your building has fully built, and the exec in `stack exec` is now running a gui program
xcmw has joined #nixos
pacmeister has joined #nixos
<clever>
fendor: does xterm work from the same shell?
ambro718 has joined #nixos
<fendor>
clever, yes, it does fully compile and xterm works from the same shell
<pacmeister>
how would one go about getting kdialog/qtdialog to work with chromium? it should support it without a patch (unlike firefox which opensuse had to make a patch for)
<hakujin>
clever: do you know of a way to use the nix build environment for side effects? e.g. I want to use the host nix store to `nix-copy-closure`. I understand this isn't the, uh, recommended fashion.
goibhniu has quit [Ping timeout: 268 seconds]
<clever>
hakujin: its simpler to just manually run nix copy after nix-store has finished
Turion has joined #nixos
Ariakenom has joined #nixos
hakujin has quit [Quit: WeeChat 2.1]
logzet has joined #nixos
xcmw has quit [Ping timeout: 240 seconds]
<avn>
Folks, does anyone remember, what package provides x11-xcb for pkg-config?
<clever>
[clever@amd-nixos:~]$ nix-locate x11-xcb
<clever>
xlibs.libX11.dev 371 r /nix/store/2802m84rk70vac9n6wz6d8cgzzzk2600-libX11-1.6.5-dev/lib/pkgconfig/x11-xcb.pc
<clever>
avn: just add xlibs.libX11 and pkgconfig to buildInputs
freeman42x]NixOS has quit [Ping timeout: 256 seconds]
blankhart has quit [Ping timeout: 268 seconds]
vidbina has quit [Ping timeout: 260 seconds]
<cbarrett>
i'm on 18.03 but I'm not seeing nix 2.0
<cbarrett>
(i'm on a mac using the darwin channel)
<cbarrett>
I'm sure it's something obvious that I'm missing
<clever>
cbarrett: what does `type nix-store` return?
<cbarrett>
nix-store is /nix/var/nix/profiles/default/bin/nix-store
vidbina has joined #nixos
<clever>
cbarrett: `sudo -i` then `nix-env -iA nixpkgs.nix`
<avn>
clever: ty, looks it works ;) Just some "if we have mass rebuild here, why not fix also...."
<{^_^}>
[nixpkgs] @thoughtpolice pushed 21 commits to pgsql-fixes: https://git.io/vx5CV
<{^_^}>
→ 9fdec8ec by @thoughtpolice: nixpkgs: deprecate the badly named 'postgresql100' attribute
<{^_^}>
→ 495794c9 by @thoughtpolice: nixpkgs: move postgresql patches into a common directory
<{^_^}>
→ 02a51bc5 by @thoughtpolice: nixpkgs: reorganize the postgresql extensions
sanscoeur has joined #nixos
<cbarrett>
clever: i'll try it
<{^_^}>
[nixpkgs] @thoughtpolice pushed 0 commits to pgsql-fixes: https://git.io/vx5Cr
<mkaito>
I'm building a remote system with `nixos-rebuild --target-host`, and there are a few nix files on the remote that I want to put in `includes`. Do I need to download them first, or can I tell Nix that these files are on the target host?
<Myrl-saki>
LnL: I get an untrusted substituter.
<LnL>
run that as root or a trusted user
<Myrl-saki>
LnL: i have.
<Myrl-saki>
Errr.
<LnL>
euh
<Myrl-saki>
LnL: Is root a trusted user by default?
<LnL>
yes
<Myrl-saki>
LnL: I mean, implicitly*
<Myrl-saki>
LnL: If, say, I set a trusted user explicitly, is root no longer trusted?
<LnL>
no, I'm pretty sure it's always considered trusted
<LnL>
and otherwise you can bypass the daemon with --store local
<LnL>
cbarrett: that's weird, with nix-darwin you should have it
<cbarrett>
I think i may be stuck on an old version somehow
Myrl-saki has quit [Quit: WeeChat 2.0]
<cbarrett>
darwin-rebuild changelog gives the newest entry as 2017-07-28
Myrl-saki has joined #nixos
<Myrl-saki>
LnL: Thanks.
<LnL>
what about nix-shell -p nix-info --run nix-info
<clever>
Myrl-saki: also, you can set a default for that in nix-daemon's env
<clever>
not sure about that config you pasted
<Myrl-saki>
clever: So, in my case, trusted-users=pi?
<clever>
yeah
<Myrl-saki>
Errr, I think I have to place it in /usr/local/etc
Drakonis[m] has quit [Changing host]
Drakonis[m] has joined #nixos
Drakonis[m] has joined #nixos
<Myrl-saki>
clever: LnL: Should I consider rebuilding nix with a different prefix?
<clever>
id say rebuild nix with nix
<LnL>
:D
<Myrl-saki>
clever: LnL: The /usr/local throws me off a bit, but I'm scared it might conflict with apt.
<cbarrett>
back
<Myrl-saki>
clever: True, makes sense.
<LnL>
eww apt
johnw has joined #nixos
<Myrl-saki>
unable to open SSH connection to 'ssh://disnix@192.168.1.101': cannot connect to 'disnix@192.168.1.101'; trying other available machines...
<Myrl-saki>
It's something!
<Myrl-saki>
LnL: Shh
<Myrl-saki>
I presonally hate apt, but it's probably the only sane RasPi OS.
<LnL>
that looks better, let me find a gist...
<cbarrett>
oh boy
<cbarrett>
i figured it out
<cbarrett>
NIXPATH is darwin-config=/Users/cbarrett/.nixpkgs/darwin-configuration.nix:/nix/var/nix/profiles/per-user/root/channels
<gchristensen>
I can never remember how to call borgbackup
<LnL>
gchristensen: hmm, what was the awesome thing you did recently then?
rauno has joined #nixos
<gchristensen>
hehehe I sent a patch to explain what features were required when it couldn't use any existing remote builders
<gchristensen>
(but it isn't merged)
<Myrl-saki>
I have machine foo. On foo, I want to build c, with the dep graph a->b->c. I have `a` already built on foo(the master). I have a build slave, bar, which wants to build b. Will foo copy a to bar or will bar rebuild a?
<LnL>
oh! yeah that's what I was thinking of
<gchristensen>
hmm I should wait for this backup to finish prior to deleting some partitions on another disk ...
humanoyd has quit [Quit: WeeChat 2.1]
<cbarrett>
hm
<cbarrett>
something seems to have put this in bashrc
<{^_^}>
[nixpkgs] @matthewbauer pushed 5 commits to master: https://git.io/vx5Kg
<{^_^}>
→ 24246f2d by @Synthetica9: i3-wk-switch: init at 2017-08-21
<{^_^}>
→ f3c5939a by @Synthetica9: fix license
<{^_^}>
→ eba94285 by @Synthetica9: Following @teto's suggestion with regards to pname
<Myrl-saki>
Wow, qemu is still not done with ./configure
<{^_^}>
[nixpkgs] @dezgeg pushed to master « rustc: Disable failing test on aarch64 »: https://git.io/vx5KF
MP2E has joined #nixos
<clever>
Myrl-saki: whats the last thing it output?
<Myrl-saki>
clever: I mean ./configure under qemu.
<clever>
ah
<clever>
yeah, that can take forever
<Myrl-saki>
clever: It has been running for around 30 minutes already.
<clever>
ive had perl configure run for several hours
<Myrl-saki>
clever: Lmao. Yep. It's perl!
<clever>
just go to bed :P
<Myrl-saki>
It's still 3 AM, too early. x
<Myrl-saki>
:x
<clever>
with any luck, it might be done by the time you wake up
<clever>
you still need luck if you want it done by the time you wake up :P
<Myrl-saki>
clever: So true.
<clever>
also, its 4pm here!
<Myrl-saki>
clever: How do I prioritize the local machine w.r.t. building ?
<clever>
not sure how you can do that, other then turning remote building off entirely
<Myrl-saki>
clever: Oh. :(
<Myrl-saki>
clever: Also. I have machine foo. On foo, I want to build c, with the dep graph a->b->c. I have `a` already built on foo(the master). I have a build slave, bar, which wants to build b. Will foo copy a to bar or will bar rebuild a?
<{^_^}>
→ f3045a5a by Marko Poikonen: photoflow: init at unstable-2018-03-06
<{^_^}>
→ f1c8417c by @jtojnar: Merge pull request #35545 from MtP76/photoflow
dweller has joined #nixos
<kreisys>
LnL: do you have any idea whether it's possible to prevent macos starting mdworker (and a bunch of other daemons.. including /usr/libexec lsd *eyeroll*) for each and every nixbld user being used? I'm pretty sure that's what's making my laptop melt whenever I try to build something...
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/vx56y
<{^_^}>
→ a842f0e9 by @ciderale: graph-tool: 2.16 -> 2.26
<{^_^}>
→ db8d8ac3 by @matthewbauer: Merge pull request #34864 from ciderale/py3k-graph-tool-update
<{^_^}>
→ 83dc28ca by @ciderale: graph-tool: use buildPythonPackage with format = "other";
jrolfs_ has quit [Ping timeout: 264 seconds]
jtojnar has joined #nixos
<kreisys>
Asking LnL because he wrote nix-darwin but if anyone else knows.. please! I don't need another stove top :P
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #34318 → mono{48,50,54}: make sure configureFlags are used in configurePhase → https://git.io/vNMZb
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vx56h
<{^_^}>
→ f2622e6a by @cumber: mono{48,50,54}: make sure configureFlags are used in configurePhase
<{^_^}>
→ 0dd1bb16 by @matthewbauer: Merge pull request #34318 from cumber/fix/mono-use-configure-flags
<{^_^}>
[nixpkgs] @kazcw opened pull request #38659 → iaca: init at 3.0, 2.1 → https://git.io/vx5iO
asuryawanshi has quit [Remote host closed the connection]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
asuryawanshi has joined #nixos
telent has quit [Ping timeout: 264 seconds]
dan_b has quit [Ping timeout: 260 seconds]
<avn>
Folks, do we need /build/mozilla-release-239e434d6d2b8e1e2b697c3416d1e96d48fe98e5/obj-x86_64-pc-linux-gnu/dist/firefox-59.0.2.en-US.linux-x86_64.stylo-bindings.zip with stylo stuff?
<cbarrett>
kreisys: maybe mdutil has something
asuryawanshi has quit [Remote host closed the connection]
<LnL>
kreisys: well, you probably don't want it for the store
asuryawanshi has joined #nixos
fragamus has joined #nixos
fendor has quit [Quit: Connection closed for inactivity]
<LnL>
kreisys: try mdutil -i off /nix and mdutil -E /nix
coot has quit [Ping timeout: 276 seconds]
fragamus has quit [Client Quit]
<kreisys>
I think I'm looking for something more general because for each of my working nixbldN users it spawns mdworker, lsd, distnoted and trustd. I'm pretty sure none of which are needed
<kreisys>
LnL: let me try your sleep builder test. I wonder if it's happening because of something my builder is doing
<LnL>
gchristensen: do the hydra builds still have those random processes?
simendsjo has joined #nixos
johnw has joined #nixos
<gchristensen>
Don't know, I could check. How should I test for that? I'm AFK but can test in somewhere between ten minutes and like 3 hours depending on how this line at a government office goes.
Turion has joined #nixos
fragamus has quit [Client Quit]
zybell_ has joined #nixos
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
<zybell_>
Myrl-saki:prioritizing local: first make localhost a remote builder,second set an unique build attribute on buildhost 'localhost', third require that buildattribute in your recipe.
<clever>
zybell_: fourth, deal with nix deadlocking because its building something on localhost but has to wait for itself to finish before it can start
<zybell_>
clever:-j2?
<clever>
zybell_: happens even with 1 thread
<clever>
nix gets a local lock on the storepath, because its being built
<clever>
then ssh's into the slave (itself) and says "build this!"
<clever>
the slave see's the lock there, assumes another thread is building it already, and waits
<kreisys>
LnL: seems like those processes also stick around for a while after the build is finished -_-
nh2 has joined #nixos
<kreisys>
I love apple. but I hate apple *sigh*
nh2 has quit [Client Quit]
nh2 has joined #nixos
nh2 has quit [Client Quit]
<clever>
kreisys: check pstree, what is the parent?
<LnL>
kreisys: do you also see them with nix-build -E 'with import <nixpkgs> {}; runCommand "foo" {} "sleep 1000"' like I did?
<norfumpit>
how do you all do your gnome configuration? like things youd normaly use gnome-tweak-tool to set, stuff like themes and so on. gsettings stuff
robstr has quit [Quit: WeeChat 1.9.1]
<clever>
norfumpit: the xfce session settings include an option for ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt
<clever>
norfumpit: running the gnome settings daemon
<LnL>
lol
<zybell_>
set COMMAND on ssh key to deal with this
<clever>
LnL: and that would be my desktop locking up once more, lol
<kreisys>
clever: launchd of course :P
<clever>
kreisys: pid 1 or other?
<kreisys>
clever: pid 1.
<kreisys>
LnL: I haven't tried yet. still waiting for all those processes to go away
<clever>
it was probably started by something that has since quit
<zybell_>
clever:login
<kreisys>
oh also @clever I tried enabling sandbox but apparently the nodejs package isn't pure? it complains that it's trying to open some dylibs
<kreisys>
that *might* be the trigger
ndrei has quit [Remote host closed the connection]
<zybell_>
kreisys:I believe mdworker is started for every user by login to be able to mount diskimages as user.
<kreisys>
And it's not disable-able?
<kreisys>
I don't see why nix builder would need to mount dmgs
<zybell_>
And the other 'helpful' procs for similar reasons
<clever>
kreisys: id recomend doing the initial work on linux, darwin is just weird
Unode has joined #nixos
<LnL>
heh
<kreisys>
I thought the darwin impurity issues have been already dealt with
fede_run has joined #nixos
<LnL>
kreisys: sandboxing doesn't work with frameworks yet
<fede_run>
Hello!
<clever>
kreisys: there are non-free libraries that cant be put into the store
<kreisys>
I c but can't I build a pure version of nodejs somehow?
<zybell_>
I dont see it either, but my view isnt important, OSXs is. And for OSX is login the last chance to start sth as root for the user. Follows:Dont use login. Try su.
raving has joined #nixos
<kreisys>
zybell_: ... try su for starting nix-daemon?
Turion has quit [Remote host closed the connection]
<fede_run>
I want to greet to everyone, it's my first week with nixos and I find it great, I love the ability to do reproducible build. There is only one flaw, I can't find anyware a way to add localization to plasma (it isn't a big flaw I can stick with English).
<kreisys>
like I'm not really sure why the nixbldN users go through a full-on login
<zybell_>
try su in nix-daemon to switch to builduser.
<zybell_>
I think they are run by sshd, which ofc does a full login.
<zybell_>
Sorry mixed that up with another issue.
Guanin has quit [Remote host closed the connection]
Guanin has joined #nixos
jrolfs_ has joined #nixos
<kreisys>
k they aren't going anywhere. I even tried to murder all of them but they keep coming back. I'm gonna give up and reboot -_-
<kreisys>
with node I get that, and also this: "/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation"
<nick_l>
Is there anyone who upgraded to 18.03 on a desktop system without issues?
<telent[m]>
So, dunno how to match irc usernames to real names but if anyone here was at (or was organising) the London meetup - I had some great conversations, thanks!
<telent[m]>
<- nixwrt guy
ryanartecona has quit [Quit: ryanartecona]
<tilpner>
nick_l - Maybe. I had a weird keyboard freeze once, but I'm not sure it's related
<kreisys>
LnL: otool shows both the @rpath one and the absolute one
<{^_^}>
[nixpkgs] @tadfisher opened pull request #38660 → btrbk: fix SSH filter script; clean up build inputs → https://git.io/vx55q
johnw has quit [Ping timeout: 263 seconds]
<nick_l>
How can I use multiple NixOS machines to compile work generated by a configuration.nix configuration?
<nick_l>
That is, I have a fairly fast network between two locations and I would like to use the cores from the other location in order to rebuild my system faster.
<mpickering>
I'm trying to debug why a executable fails to load a dynamic library. I have tried setting (DY)LD_LIBRARY_PATH but that doesn't work. How can I diagnose why it is failing?
<mpickering>
good idea. I've never used it before though. Another problem is that the executable is being invoked by the main executable I am packaging.
<nick_l>
LnL: that could use improvement. I just looked at the Perl script (I thought we got rid of Perl as a dependency), but it seems fairly limited.
<kreisys>
try this. this triggers the aux processes
<nick_l>
Does "beep" work on your systems?
knupfer has quit [Ping timeout: 260 seconds]
<LnL>
nick_l: sorry to many conversations at the same time, with nix 2.0 it was rewritten so that's no longer a perl script
<{^_^}>
[nixpkgs] @7c6f434c pushed to release-18.03 « libreoffice: wrap to set JAVA_HOME, SAL_USE_VCLPLUGIN and DBus session if not set »: https://git.io/vx5Fj
<LnL>
kreisys: interesting, I wonder if frameworks trigger these somehow
<kreisys>
I sure hope not! that would be hard to get rid of!
asuryawanshi has joined #nixos
detran` has quit [Ping timeout: 240 seconds]
<zybell_>
kreisys:other buildInputs trigger too? other npm commands?
detran` has joined #nixos
<{^_^}>
[nix] @shlevy merged pull request #2060 → Add all dependencies to source prerequisites → https://git.io/vx7cT
<{^_^}>
[nix] @shlevy pushed commit from @IohannesArnold to master « manual: Add all dependencies to source prerequisites (#2060) »: https://git.io/vx5bP
<{^_^}>
[nixpkgs] @dezgeg pushed to master « nixos/make-ext4-fs: Use closureInfo »: https://git.io/vx5b1
<kreisys>
so I'm guessing apple might actually be doing the right thing and I'm being a troll
<kreisys>
I looked up trustd and looks like it's simply in charge of providing processes with CA certificates
<kreisys>
it makes sense that npm would want to use that
detran` has quit [Ping timeout: 256 seconds]
ottidmes has joined #nixos
<katona>
Hi, any idea how could I get KDE git unstable on nixos? I mean I have seen the fetch.sh file in the plasma-5 dir, and tried to change one specific package to use fetchgit instead of fetchurl, but I didn't get it to work.
detran` has joined #nixos
lord| has joined #nixos
Mercuria1Alchemi has quit [Ping timeout: 264 seconds]
nick_l has quit [Ping timeout: 260 seconds]
<zybell_>
that makes sense, but not in all npm commands. There must be some where it doesn't. does loading npm by itself trigger trustdb? Or does it need a dl command?
Ariakenom has quit [Quit: Leaving]
detran` has quit [Ping timeout: 240 seconds]
kerrhau has joined #nixos
kerrhau has quit [Changing host]
kerrhau has joined #nixos
szicari has quit [Quit: szicari]
<wilornel>
Hi #nixos! I'm trying to install cmake. I need version 3.10 or above. I found https://nixos.org/nixos/packages.html#cmake and when I try nix-env -iA nixos.cmake, I get installing 'cmake-3.8.2'
<wilornel>
How can I install cmake 3.10?
<mpickering>
dtruss immediately crashes when I try to run the application through it, it was never meant to be easy was it
<puffnfresh>
shlevy: sure thing, I'll license it same as nix-plugins
<puffnfresh>
actually it was 99% copied from Nix itself, should have that license
<shlevy>
puffnfresh: Cool, thanks!
hiratara has quit [Ping timeout: 255 seconds]
<shlevy>
Hmm
<boomshroom>
Wow. GHC takes a really long time to compile!
pkill9 has joined #nixos
<boomshroom>
Any chance that we could get an unregistered build of GHC in the repos? It would definitely save a few hours when you want to inspect the C output.
hiratara has joined #nixos
<zybell_>
boomshroom:What are its build inputs?
<puffnfresh>
boomshroom: you mean like ghc821Binary?
kerrhau has quit [Ping timeout: 260 seconds]
<boomshroom>
zybell_: I mean built with the --enable-unregistered configure flag.
<{^_^}>
[nixpkgs] @dezgeg pushed 656 commits to unstable-aarch64: https://git.io/vx5pb
<{^_^}>
→ ded1281f by @ryantrinkle: Merge branch 'master' into docker-dirlinks
<{^_^}>
→ dce2c258 by @ryantrinkle: dockerTools: optionally preserve directory symlinks
<{^_^}>
→ cbe48055 by @xrvdg: hl1110 : init at 3.0.1-1
<tmplt>
How can I use packages from a specific channel when starting a nix-shell?
<zybell_>
mpickering:ofc it is, but how does nix select it, where does it get from, and how will that ver be compiled for *reproducibility*? Does ghc (the new one) care about the hash of the old one?
hiratara has joined #nixos
Rusty1_ is now known as Rusty1
<boomshroom>
I should mention that nix was set to build only 1 package. So it's not building anything but my GHC.
<mpickering>
look in "top-level/haskell-packages.nix"
<mpickering>
The "bootPkgs" attribute is set to the version of the compiler which is needed to do the bootstrapping
<cufisz_>
zybell: thanks for the suggestion. Any idea on what all the relevant configs may be or where to find them?
<cufisz_>
ottidmes: that worked!
<ottidmes>
cufisz_: Great to hear :)
<cufisz_>
that's a one time thing right? no changes to /etc/nixos/configuration.nix required.
<ottidmes>
cufisz_: One time thing, it has worked fine for me since, across reboots, without any changes other than sound.enable = true; (I already had enabled pulseaudio and added my main user to the audio group)