<NixOS_GitHub>
[nixpkgs] Nadrieril opened pull request #34064: nixos/ckb: Add option to restrict access to the devices to a particu... (master...ckb-gid) https://git.io/vNzBN
NixOS_GitHub has left #nixos [#nixos]
shabius has joined #nixos
fragamus has joined #nixos
dan_b has quit [Ping timeout: 240 seconds]
la_putin has quit [Read error: Connection reset by peer]
<taohansen>
my Gentoo install of Nix is giving me a little more trouble than my Guix/Nix abomination. `sudo` doesn't know the path of my local Nix packages
pjan has joined #nixos
pkill9 has quit [Ping timeout: 268 seconds]
s33se_ has joined #nixos
markus1199 has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
s33se has quit [Ping timeout: 240 seconds]
markus1189 has quit [Ping timeout: 256 seconds]
<adisbladis>
taohansen: Because the command invoked by sudo does not inherit the environment variables of the user
<adisbladis>
taohansen: sudo -E should work
ryanartecona has quit [Quit: ryanartecona]
boxofrox has joined #nixos
xcmw has quit [Ping timeout: 256 seconds]
pie__ is now known as pie_
magnetop` has quit [Ping timeout: 256 seconds]
fragamus has joined #nixos
pie__ has joined #nixos
marusich has joined #nixos
pie_ has quit [Ping timeout: 268 seconds]
<pie__>
anyone else getting usb redirection permissions errors with virt-manager?
newhoggy has joined #nixos
marusich has quit [Ping timeout: 265 seconds]
newhoggy has quit [Ping timeout: 240 seconds]
Drakeson has joined #nixos
Lisanna has joined #nixos
Drakeson has quit [Ping timeout: 248 seconds]
<azdle>
Is there something I need to install to give java (openjdk-9) its set of CA certs?
<pie_>
for the record it does work if i run it as root :/
<pie_>
oh huh, adding a udev rule worked
<pie_>
\o/ yay
newhoggy has joined #nixos
newhoggy has quit [Remote host closed the connection]
nuncanada has quit [Quit: Leaving]
<taohansen>
adisbladis: ah perfect, thank you!
rogue_koder has quit [Quit: Konversation terminated!]
ison111 has joined #nixos
hakujin has joined #nixos
<ison111>
I'm confused about adding perl modules. I ran nix-generate-from-cpan as recommended here https://nixos.wiki/wiki/Perl and got the output. But it says to add it to pkgs/top-level/perl-packages.nix and I'm not sure how to do that. I suppose it's supposed to be somewhere in ~/.nix-defexpr/channels but all I have is channels_root which I can't edit.
hakujin has quit [Ping timeout: 255 seconds]
Capprentice has joined #nixos
Drakonis has quit [Read error: Connection reset by peer]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] adisbladis pushed 1 new commit to release-17.09: https://git.io/vNzKd
<NixOS_GitHub>
nixpkgs/release-17.09 4bb14e4 Jan Tojnar: librsvg: 2.40.17 → 2.40.18...
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lopsided98 opened pull request #34068: perf: apply patch to fix 4.9 build when inheriting kernel make flags (master...perf-4.9-fix) https://git.io/vNz6t
NixOS_GitHub has left #nixos [#nixos]
Drakeson has joined #nixos
newhoggy has joined #nixos
Drakeson has quit [Ping timeout: 256 seconds]
schoppenhauer has quit [Ping timeout: 256 seconds]
schoppenhauer has joined #nixos
la_putin has quit [Remote host closed the connection]
<NixOS_GitHub>
nixpkgs/master b81c65c Dan Peebles: libsnark: init at 9e6b19ff
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vNziI
NixOS_GitHub has left #nixos [#nixos]
Lisanna has quit [Ping timeout: 255 seconds]
Rusty1_ has quit [Quit: Konversation terminated!]
ssmike has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lukeadams opened pull request #34069: phantomjs2: set QT_QPA_PLATFORM to allow use in daemons (master...phantomjs-headless-fix) https://git.io/vNzii
<NixOS_GitHub>
[nixpkgs] lukeadams closed pull request #34069: phantomjs2: set QT_QPA_PLATFORM to allow use in daemons (master...phantomjs-headless-fix) https://git.io/vNzii
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lukeadams reopened pull request #34069: phantomjs2: set QT_QPA_PLATFORM to allow use in daemons (master...phantomjs-headless-fix) https://git.io/vNzii
<krey>
does anyone know why this doesn't work? { ("a"+"b") = "3"; }
tmaekawa has quit [Client Quit]
stackshadow[m] has joined #nixos
<simpson>
krey: You expected something like { ab = "3"; }?
<krey>
simpson: yup
<contrapumpkin>
{ ${"a" + "b"} = "3"; }
<contrapumpkin>
when in doubt, add moar money
<krey>
contrapumpkin: thank you for the money
joelpet has quit [Ping timeout: 240 seconds]
blahdodo has quit [Quit: Bye bye]
<tilpner>
Ralith - Yes, I linked nix-diff earlier. Did you find the issue?
krey has quit [Quit: Lost terminal]
blahdodo has joined #nixos
<Ralith>
tilpner: yep, there was something scribbling away in a dark corner in my source tree
<Ralith>
thanks!
<Ralith>
this'll also be useful for diagnosing why the hell I'm recompiling mono and vulkan all the time
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ison111 has quit [Quit: WeeChat 2.0]
<tilpner>
Ralith - Another approach would have been to manually diff the output of "nix show-derivation", then diffoscope the inputs you found that differ
Itkovian has joined #nixos
joelpet has joined #nixos
Ivanych has quit [Quit: Leaving.]
<Ralith>
tilpner: ooh, handy
iyzsong-x has joined #nixos
Ivanych has joined #nixos
iyzsong has quit [Ping timeout: 268 seconds]
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
klntsky has quit [Remote host closed the connection]
ylwghst has quit [Remote host closed the connection]
<stphrolland>
Hi. I don't understand why, since 1 or 2 days, when I try to do nixos-rebuild switch or nixos-rebuild switch --upgrade, it always fails compiling rustc. I don't understand why the large chanel 17.09 is not providing me Rust already compiled as a binary. 2/ it's strange because I don't have Rust as a package to install in my configuration.nix... Any idea how to troubleshoot ?
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<stphrolland>
THe compilation of rust has failed at two different places, so maybe it's simply my hardware that is too small to be able to compile rust. That's the likelyhood I think.
cinimod has joined #nixos
<cinimod>
`nix-shell -p "with import <nixpkgs> {}; python3.withPackages (ps: [ ps.numpy ps.jupyter ])" -I nixpkgs=/Users/dom/nixpkgs` fails for me on OSX
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vNzDU
bpa has joined #nixos
<cinimod>
Presumably `jupyter`
bpa has quit [Read error: Connection reset by peer]
<clever>
it clearly says its the package send2trash
bpa has joined #nixos
ylwghst has joined #nixos
ma27 has joined #nixos
<cinimod>
Indeed but I didn't explicitly ask for `send2trash`
<cinimod>
It must be a dependency
FRidh2 has joined #nixos
hakujin has joined #nixos
el_putin has joined #nixos
la_putin has quit [Remote host closed the connection]
<stphrolland>
Is there a place where I can check the status of a channel ? I really don't undestand why I cannot nixos-rebuild any more my nixos-17.09 system.
<el_putin>
what does that have to do with the status of a chennel
bpa has quit [Read error: Connection reset by peer]
<el_putin>
channel*
bpa has joined #nixos
hakujin has quit [Ping timeout: 240 seconds]
<stphrolland>
I dont get the binary cache for Rust, I find this suspicious
hiroshi has quit [Ping timeout: 276 seconds]
<stphrolland>
My thinking, I don't know where to start
<clever>
error: build of ‘/nix/store/z6592cjldc5clypy1jrhamc56fvkw55j-python3-3.6.4-env.drv’ failed
<clever>
cinimod: run nix-store -q --tree on this path
<stphrolland>
tilpner: what do you mean by 'here' ?
<cinimod>
`send2trash` seems a very modest package
<tilpner>
stphrolland - On my laptop. nixos-17.09 doesn't have substitutes for rustc. I'm currently failing to navigate Hydra to find out why
<stphrolland>
tilpner: so you mean you have the same issue as me ? I am reassured.
<stphrolland>
tiplner: and does rustc compiles fine on you laptop ? on mine it takes hours, then fails, either in the tests or not in the tests, at a different place.
<Aleksejs>
is there any option like gtk.config or something to modify .gtkrc file?
<stphrolland>
there's this option nixpkgs.config.allowBroken to be set to true, do you advise it so as to bypass the rust compiling ? I think i'm gonna try...
<tilpner>
I don't think it'll help
<stphrolland>
well it will have to compile so as to see it broken, of course...
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to staging: https://git.io/vNzD1
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging 5fafb7e Vladimír Čunát: clang-3.8: fixup build after #33953 917429233b
<stphrolland>
what is rustup ?
<tilpner>
FWIW, nixos-unstable has working substitutes for rustc, so you could avoid all this by just using that channel
<tilpner>
(Of course it should also work on nixos-17.09, but I have no idea what's wrong, and it might take a while for someone to fix it)
<stphrolland>
just to have an idea, in the past when there were issue like that, how long did it take to fix ?
<tilpner>
I don't know :/
bpa has quit [Read error: Connection reset by peer]
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vNzyO
<NixOS_GitHub>
nixpkgs/master 81a3863 Kai Harries: lbdb: 0.45.3 -> 0.46
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to staging: https://git.io/vNzy3
<NixOS_GitHub>
nixpkgs/staging 94f4857 Vladimír Čunát: Merge branch 'master' into staging...
NixOS_GitHub has left #nixos [#nixos]
hiroshi has quit [Ping timeout: 276 seconds]
hiroshi has joined #nixos
el_putin has quit [Quit: Konversation terminated!]
el_putin has joined #nixos
ThatDocsLady has joined #nixos
<steveeJ>
would someone mind to verify if their libvirt/virt-manager uses usb-auto-redir? this means, you connect to the console of a guest and plug a usb device, which should be automatically redirected. it's not working for me
<Moredread>
tilpner: is there a way to use a substitute for dependencies of another package? ripgrep needs rustc for compilation and pulls it in for me
<tilpner>
Moredread - I'm not sure I understand your question correctly. If rustc is pulled in, and everything is setup correctly, it will use binary substitutes automatically, regardless of how/why you use it.
<tilpner>
Moredread - If you meant replacing the rustc ripgrep needs with the rustc of nixos-unstable, that's possible
<tilpner>
(But there would be little point, as you could just use ripgrep from nixos-unstable instead)
<Moredread>
yeah I meant the later
<LnL>
cinimod: generally master, if the bot detects a large rebuild we can change it to staging
<Moredread>
thx that should work
<Moredread>
and it did :)
<LnL>
is there a problem with rustc?
bpa has quit [Read error: Connection reset by peer]
<Moredread>
LnL: it doesn't compile on 17.09 it seems
<vcunat>
(except for builds skipped due to some other failing)
<vcunat>
s/two retries/two attempts/
uwap has joined #nixos
<sphalerite>
vcunat: haha, seems hacky but I guess it should work. Thanks!
<vcunat>
it is hacky
<sphalerite>
vcunat: oh but "some other failing" includes failed dependencies
<sphalerite>
So only the ones lowest in the dependency tree will get retried
<vcunat>
not really
MP2E has quit [Remote host closed the connection]
<sphalerite>
No?
<vcunat>
the first instance will skip them
<vcunat>
but if the second instance manages to build the package, it will continue to reverse deps
<sphalerite>
Right but say we have libuv -> cmake -> foo. Then if libuv fails the first time, it won't try to build cmake the first time, and only try it once the second tike
<sphalerite>
Time*
<vcunat>
correct
<sphalerite>
So I guess I want two attempts at realisation per derivation really
<vcunat>
BTW, there's a nix option to repeat *successful* builds (to check determinism).
<sphalerite>
Yeah I know that one
<sphalerite>
In this case I just want to get libuv building in spite of its nondeterministic test suite
<vcunat>
If a test in there is problematic, I'd just disable it in nixpkgs.
<vcunat>
We have done so for a few libuv tests already.
joelpet has quit [Remote host closed the connection]
<vcunat>
(But I haven't noticed libuv failures on Hydra recently.)
<steveeJ>
Mic92: have you recently tried to use the rls-preview component? like any other rustup component it lives in the toolchain, and by default we don't have the toolchain's bin directory in PATH. do you have any better ideas than just fixing the latter?
<Mic92>
steveeJ: I have not used rls-preview so far, only the compiler. How is the toolchain supposed to be used then? rustup only register itself as binary for rustc, cargo, rust-gdb, rust-lldb and so for.
<steveeJ>
Mic92: I think that rls is added to that list
<Mic92>
steveeJ: have you just tried to update rustup?
<steveeJ>
Mic92: no, I've used 1.3 from nixpkgs. however, we can't statically produce wrappers/symlinks for components as they can be added at runtime
<stphrolland>
Moredread: thank you. Thanks to you I found th epackage that was bringing rustc in the dependencies.
<stphrolland>
Moredread: and now I can nixos-rebuild nicely without rustc comming in the way
<NixOS_GitHub>
[nixpkgs] idontgetoutmuch opened pull request #34076: Support the us package (master...masterNG) https://git.io/vNzH1
NixOS_GitHub has left #nixos [#nixos]
alunduil has joined #nixos
<Mic92>
steveeJ: I can add an rls/rustfmt symlink to rustup though. rustup is supposed to dispatch it to the correct target.
<stphrolland>
So I am subscribed to channels nixos-17.09 and unstable (for one problem package). From my readings, I think I have understand there is a master branch. Is this master branch different from unstable ? If not, what's the address for subscribing to master channel ?
<stphrolland>
I mean "If yes", instead of "If not" in my last sentence :-)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vNzQe
<steveeJ>
Mic92: I wonder if we should just rely on the user modifying their PATH. apparently rustup should install binaries into cargo's bin directory, but that hasn't happened on here
andymandias has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 26f3f06 to 861f6a7: https://git.io/v5aIf
Capprentice has quit [Remote host closed the connection]
<sphalerite>
stphrolland: unstable is master, just delayed until some tests have passed
Capprentice has joined #nixos
<stphrolland>
sphalerite: thx, much clearer for me
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh opened pull request #34077: Python: package updates and fixes (master...python-unstable) https://git.io/vNzQW
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
This applies to both nixos-unstable and nixpkgs-unstable, but nixpkgs-unstable has a smaller test set which makes it dangerous (like master) to run nixos from
<sphalerite>
!channels
<sphalerite>
!channel
Capprentice has quit [Max SendQ exceeded]
ssmike has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 861f6a7 to 1d55378: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 6c5149f Frederik Rietdijk: python.pkgs.crayons: init at 0.1.2
<NixOS_GitHub>
nixpkgs/python-unstable 92e535d Frederik Rietdijk: python.pkgs.flask-limiter: init at 1.0.1
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable efb915e Frederik Rietdijk: python.pkgs.flask-common: init at 0.2.0
<sphalerite>
there we go. stphrolland ^ goes into more detail
alex`` has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 1d55378 to a6ae893: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 3045f66 Frederik Rietdijk: python: SQLAlchemy: 1.2.0 -> 1.2.1
<NixOS_GitHub>
nixpkgs/python-unstable 5f47916 Frederik Rietdijk: python: stripe: 1.77.0 -> 1.77.1
<NixOS_GitHub>
nixpkgs/python-unstable 78ff151 Frederik Rietdijk: python: Sphinx: 1.6.5 -> 1.6.6
NixOS_GitHub has left #nixos [#nixos]
ixxie has joined #nixos
dbe_ has quit [Ping timeout: 240 seconds]
<Mic92>
steveeJ: does rustup extends this directory dynamically, or is it only touched on version upgrades?
<Mic92>
I mean ~/.cargo/bin
<Yaniel>
cargo install puts binaries there
ylwghst has joined #nixos
<Yaniel>
`cargo install`*
<pmeunier>
Hi! I've read some of the issues about GCE disk size on nixops, and I don't understand: has anyone succeeded in deploying a machine with a disk smaller than 100Gb?
<pmeunier>
if so, how? also, how are the images built? I'm willing to write it up in the manual if I can do it.
<sphalerite>
pmeunier: (disclaimer: I've never touched the GCE stuff, including using it) I believe the images are built by hydra so you should, I think, be able to find how the images are built in nixpkgs/nixos/release.nix
<pmeunier>
sphalerite: thanks (and sorry for being stupid)
<sphalerite>
pmeunier: I don't see any stupid in any of your messages :p
<psychon>
hi, can someone give me a good idea for https://github.com/NixOS/nixpkgs/issues/14164 ? AwesomeWM has $LD_LIBRARY_PATH set via a wrapper script; this variable is inherited by processes spawned by awesome; replacing this with an rpath does not work; are there any other options available?
<sphalerite>
psychon: It's not a beautiful solution by any means, but the easiest I can think of is patching awesome to unset LD_LIBRARY_PATH before execing child processes
<psychon>
changing env vars is not thread safe :-(
<psychon>
so that requires copying the whole env to change it, but yeah, that could actually work
<sphalerite>
that shouldn't be a problem when using exec since exec replaces the whole process, no?
andymandias has joined #nixos
<sphalerite>
so either you fork beforehand, which gives you a new process, or you're stopping all the threads since you're replacing the process image. I think?
<psychon>
...or awesome uses some wrapper from glib to spawn processes and so things become slightly more complicated, but yeah ;-)
thc202 has joined #nixos
<psychon>
sphalerite: do you want to be mentioned somewhere as the author of this idea? or should I just propose a patch?
ixxie has quit [Ping timeout: 248 seconds]
yarny has joined #nixos
vcunat has quit [Ping timeout: 256 seconds]
<sphalerite>
Nah don't worry about it, it was just the first thing that came to mind
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/python-unstable b47ff59 Frederik Rietdijk: python.pkgs.sqlalchemy: use current pytest
<NixOS_GitHub>
[nixpkgs] FRidh pushed 3 new commits to python-unstable: https://git.io/vNz7L
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable f72a465 Frederik Rietdijk: CPython and PyPy: update meta.maintainers
<NixOS_GitHub>
nixpkgs/python-unstable c0c616b Frederik Rietdijk: python.pkgs.pytest_30: remove unused version
<sphalerite>
it's not a deeply thought-through concept or anything :p
<psychon>
well, my first thought was something like "perhaps gobject-introspection can be patched to embed absolute paths in its .typelib files"; your idea is a lot better
<psychon>
yeah, but it's not really related since in the case of awesome it (well, some library it links to) uses dlopen() to load libraries
<psychon>
so that dlopen() needs to get an absolute path as argument, which seems to be exactly what PR #12558 did
<sphalerite>
yeah
<sphalerite>
maybe check the typelib files involved to see if they contain absolute paths?
<psychon>
now I just have to remember where I put my nixos VM...
<yarny>
Hi, my machine is observing something strange going on in the release-17.09 channel. Two hours ago, my cronjob did "nixos-rebuild --upgrade boot", which fetched "nixos-17.09.2826.5141f28405e", then tried to locally build libreoffice (this usually comes from the binary cache for me).
<avn>
sphalerite: patcher for gobject-introspection can fix other apps/libraries as well
<yarny>
20 min ago, I stopped the rebuild and did a "nix-channel --upgrade" by hand, which surprisingly fetched "nixos-17.09.2799.d9a2891c32e" (apparently an *older* verision). A "nixos-rebuild" then fetched libreoffice from the cache, but tried to (and failed) build wireshark-cli, which is also unusual.
<sphalerite>
yarny: release-17.09 is not a channel
<Aleksejs>
is there any way to see what graphics driver is currently used?
<tilpner>
Something is weird about the channel currently, I can't explain it though
<yarny>
I remember having read somewhere that the number after "17.09" (here 2799) is the number of commits from the initial 17.09 release and therefore like a increasing version number. So I'm surprised that this number actually decreased, and I suspect it has something to do with my update-failures.
<tilpner>
And if you follow that URL, you indeed get d9a289 as nixpkgs input
<tilpner>
Hold on, this may have been fixed between my first line and now
<tilpner>
Try again, yarny?
<yarny>
One moment ...
<yarny>
OK, I just "nix-channel --update"ed to 17.09.2826.5141.." again.
<yarny>
Oh no: "nixos-rebuld dry-build" claims to rebuild libreoffice again.
<tilpner>
:/
<yarny>
.. so I'm apparently at the "newest" version again, but the binary cache still seems to lack some binaries (at least libreoffice-5.2.6.2)
<tilpner>
There was a similar issue earlier with rustc
* psychon
currently builds the X.org X11 server
<samae>
Anyone can point me on a way to use overlays to define new configuration modules?
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<tilpner>
samae - You can technically define new modules that way, but you can't use them, and there is no point in defining them in overlays
<samae>
:s
<tilpner>
samae - You're supposed to put the new modules into imports
<samae>
any ressources on how to do that?
<tilpner>
samae - You write a module, put it somewhere, and then put "imports = [ ./path/to/new/module.nix ];" into an existing module
<tilpner>
samae - That can include <path-items> and paths into the store (e.g. from fetchFromGitHub), so you don't *have to* manage the modules locally
<fearlessKim[m]>
don't we have bash functions "nix_debug"/nix_info/nix_warn, it is kinda boring to keep checking for NIX_DEBUG before echo-ing a message?
hakujin has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 12 new commits to python-unstable: https://git.io/vNzdC
<NixOS_GitHub>
nixpkgs/python-unstable 29e9608 Frederik Rietdijk: python.pkgs.PyLTI: move expression
<NixOS_GitHub>
nixpkgs/python-unstable 3fa9702 Frederik Rietdijk: python.pkgs.pytest_27: remove old version
<NixOS_GitHub>
nixpkgs/python-unstable e2f041a Frederik Rietdijk: python.pkgs.PyLTI: 0.4.1 -> 0.5.1
NixOS_GitHub has left #nixos [#nixos]
el_putin has quit [Remote host closed the connection]
el_putin has joined #nixos
fuyuuri has quit [Quit: WeeChat 2.0]
leat has quit [Quit: WeeChat 1.9.1]
hakujin has quit [Ping timeout: 255 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vNzdw
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable ee8f15e Robert Schütz: pythonPackages.pytestrunner: 2.6.2 -> 3.0...
<NixOS_GitHub>
nixpkgs/python-unstable c44d515 Florian Klink: curtsies: don't set name attribute anymore
<NixOS_GitHub>
nixpkgs/python-unstable 79745a0 Florian Klink: pyte: don't set name attribute anymore
<NixOS_GitHub>
[nixpkgs] FRidh pushed 6 new commits to python-unstable: https://git.io/vNzdd
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable cddfa94 Florian Klink: guzzle_sphinx_theme: don't set name attribute anymore
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh closed pull request #33758: python: remove unneeded name attributes, remove rec, make afew an application (master...py-fixes) https://git.io/vNs2F
NixOS_GitHub has left #nixos [#nixos]
nuncanada has joined #nixos
el_putin has quit [Ping timeout: 256 seconds]
<ixxie>
hmmm... is nix-build always pure?
<hyper_ch>
hi ixxie
<LnL>
what tp you mean by that?
jensens has quit [Ping timeout: 265 seconds]
<LnL>
nix-build also evaluates so it’s not pure in that regard
vcunat has joined #nixos
<ixxie>
heya hyper_ch
<ixxie>
well LnL - I am wondering whether it uses my system environment when building
acarrico has quit [Ping timeout: 265 seconds]
<ixxie>
my question is really how to reliably test that my packages correctly build
hellrazor has joined #nixos
<hyper_ch>
never used nix-build :)
ndrei has quit [Remote host closed the connection]
<typetetris>
hmm, you can unset PATH in the builder and construct a PATH only from the inputs. As far as I understand it, thats how it is usually done. But if the buildsystem of the package looks in hard coded paths, you can't prevent that. (Just by making sure, the path isn't there.)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb pushed 1 new commit to master: https://git.io/vNzF7
<NixOS_GitHub>
nixpkgs/python-unstable 75c2a76 Frederik Rietdijk: python.pkgs.xvfbwrapper: add missing test dependency
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vNzFA
NixOS_GitHub has left #nixos [#nixos]
Itkovian has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/python-unstable 7f77cce Frederik Rietdijk: python.pkgs.pytest_29: remove old version
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to python-unstable: https://git.io/vNzbY
<NixOS_GitHub>
nixpkgs/python-unstable ee91c45 Frederik Rietdijk: python.pkgs.pytest_32: remove old version
NixOS_GitHub has left #nixos [#nixos]
<ixxie>
typetetris: maybe I could run nix-build inside a pure nix-shell xD
<gchristensen>
ixxie: nix-build is pure
<gchristensen>
ixxie, typetetris: nix-build runs inside a build sandbox which cannot access anything outside of the provided inputs
<typetetris>
gchristensen: Maybe I misunderstood something, but how can https://github.com/NixOS/nix-pills/issues/40 be possible then? Clearly that build used my systems libexpat, even if expat wasn't mentioned in buildInputs or anything like that?
<gchristensen>
aahhhhh sorry, this is true for Linux. MacOS sandboxing is almost here!
<ixxie>
thanks gchristiansen :) also: does buildPythonPackage come with some sort of basic environment?
<ixxie>
I notice I don't need pytest
<gchristensen>
I don't know
<typetetris>
gchristensen: Will there be sandboxing for linux too? For example with a chroot environment or something like that? That would be really grea.
<typetetris>
s/grea./great./
deltasquared has joined #nixos
<gchristensen>
typetetris: Linux already has sandboxing
simukis has quit [Quit: simukis]
<FRidh2>
ixxie: buildPythonPackage comes with python, setuptools, pip and wheel
simukis has joined #nixos
<typetetris>
Then why did the behaviour of the above issue happen? The build should have failed, telling me libexpat wasn't in the sandbox?
<typetetris>
On is sandboxing on linux only provided with Nixos and not with nixpkgs on Arch (for example)?
<FRidh2>
typetetris: by default sandboxing is off
<ixxie>
FRidh2: I seem to be able to build certain python packages without fullfilling test-requirements.txt
<FRidh2>
you need to turn it on in either `/etc/nix/nix.conf` or pass it as an `--option`. With NixOS you can set `nix.useSandbox = true`
<ixxie>
FRidh2: but I am starting to get the feeling *requirements.txt is not reliable usually
<FRidh2>
ixxie: tests aren't always automatically run
<ixxie>
FRidh2: btw I think I finally correctly squashed those commit xD
<typetetris>
FRidh2: Thanks.
<typetetris>
I will read up about it.
<ixxie>
FRidh2: I see. Well I will just trust that if it builds, its okay
<lejonet>
How would I do if I want to rename some attributes in an attrset? I have an attrset that I'm going to turn into a ini file with the toINI generator, but the ini file option has options that have spaces in them, and I've named the options with camelCase as is the norm for nixos
simukis has quit [Client Quit]
<lejonet>
should I rename the options instead?
simukis has joined #nixos
<lejonet>
or should I use removeAttrs after moving the options to the right name instead?
<joko>
lejonet: I would vouch for removeAttrs to keep conformity
<FRidh2>
lejonet: have another attrset that acts as translator?
<ixxie>
FRidh2: I have another PR in the works that depends on the Jupyterhub one; can I just submit it immediatly or should I wait?
<lejonet>
(e.g. just before calling toINI, doing a replacement in the attrset and then do removeAttrs)
<lejonet>
joko: that is what I'm leaning towards
<typetetris>
Is it correct, that nix-pill 17 is outdated, as nowadays one would use overlays instead of package overrides?
<FRidh2>
ixxie: let me have a look at those PR's
<lejonet>
FRidh2: I haven't found how I would do that, I don't see anything in the built-in functions that would allow me to do that
<joko>
typetetris: packageOverrides works for the time being, but I would recommend using overlays. I am also transitioning slowly to them
<ixxie>
FRidh2: okay, I will submit the second one now
<lejonet>
FRidh2: because that would indeed be the smoothest and cleanest solution
<deltasquared>
hmm, fetchurl... where are you hiding in the docs
<typetetris>
joko: Thanks. Are they deprecated or something like that or will they be support for the time being?
<FRidh2>
deltasquared: no docs for that one
<joko>
lejonet: A translator is not that bad
ambro718 has joined #nixos
teh[m] has joined #nixos
<deltasquared>
FRidh2: ಠ_ಠ
<lejonet>
joko: yeah, that is what I was initally looking for, but haven't found anything about it
<tilpner>
lejonet - You should give examples of the transformation. Maybe it can be done automatically
<lejonet>
joko: could you point me to the docs where that is explained?
<lejonet>
tilpner: please explain what you mean with examples of the transformation
<lejonet>
(as in how I would "code" it in nix)
<tilpner>
lejonet - E.g. "fooBar" -> "foo bar" would be easy
<lejonet>
tilpner: ah, so -> can be used for that? Would I apply that to the attrset with like a map or so?
aborsu has joined #nixos
<FRidh2>
typetetris: the pill explains how overrides work, that has not changed. overlays are an additional method for overriding not just packages but package sets. Do note that packageOverrides is mostly replaced by overlays
<joko>
typetetris: I am not sure when packageOverrides is going to be removed, all I can say is that it works for now
<tilpner>
lejonet - Yes, you can map keys. If you have a function that goes from camelCaseKey to finalKey, you're set
<lejonet>
tilpner: That is what I had hoped to find, awesome :)
<joko>
lejonet: you can't use "->" directly, you need mapAttrs
<deltasquared>
FRidh2: was only asking about fetchurl due to wanting to find documentation about this "fetchgit" thing I heard yesterday in the same place
<lejonet>
tilpner: do you have any example code for such a function? or is it just to do a function that has a list of the transformation, that then is applied to the input value unconditionally, but only those that meet the criterias of the "fooBar" -> "foo bar" element will be transformed?
<deltasquared>
seems interesting these expressions wouldn't be in the store or under /nix somewhere
<ixxie>
FRidh2: so cstrahan will lose credit for his work
<ixxie>
FRidh2: is there no way around this?
cinimod has quit [Quit: Connection closed for inactivity]
peacememories has joined #nixos
<FRidh2>
ixxie: yes and no. You could add and remove changes from a commit, in such a way that the committer info is kept. But considering the struggling to get this right I suggest you don't bother with it.
<azazel>
hyper_ch: well, it's a mixed situation of learning and tinkering ;-) I've found a nice project on github that I plan to use as a trace to develop my own "setup": https://github.com/r-raymond/nixos-mailserver
<hyper_ch>
had a look at that as well
<hyper_ch>
azazel: but kexec is really nice :)
<azazel>
hyper_ch: coming from python, I've still to understand some few things of Nix...
psychon has left #nixos [#nixos]
<hyper_ch>
azazel: only a few? then you know it much better than me...
<azazel>
probably because i'm too dumb to see the other things ;-)
<deltasquared>
lol, whose idea was it to git archive | tar to copy a git repository to the target store path? :P
<deltasquared>
I mean sure it *works* but that took me a moment to figure out what was going on
<azazel>
my confidence has yet to reach a critical point
<hyper_ch>
azazel: I'm just a basic nixos user and reveiced a lot of help from all the friendly people in here
<hyper_ch>
azazel: but the thing about nixos is, once it's installed, it's not easy to whack it into an unworking state
<hyper_ch>
it's not impossbile but the atomic upgrades make it very hard
<hyper_ch>
(furthermore you have zfs where you can autosnapshot)
<deltasquared>
rm -rf /nix/store # nah just kidding
<hyper_ch>
deltasquared: try it :)
<azazel>
:-)
<hyper_ch>
doesn't work even as root
<deltasquared>
hyper_ch: I'm pretty sure it would be explosions. I'll let you go first :P
<hyper_ch>
deltasquared: it's mounted as ro by default
<deltasquared>
(should point out there won't be any magical protections on a mere hosted nix setup)
<hyper_ch>
as root you'd have to remount it as rw
<hyper_ch>
not really sure how the nix daemon handles that on the file
<deltasquared>
hyper_ch: I'm not running nixos right this moment... ;)
<deltasquared>
though point taken for the regular case
<hyper_ch>
deltasquared: if you don't run nixos, then there's no /nix/store to even rm -Rf :)
<deltasquared>
hyper_ch: sure there is, I have a hosted nix installation on arch linux here
<hyper_ch>
so return signal should be 1 :)
<hyper_ch>
deltasquared: try to touch /nix/123
<hyper_ch>
curious if it's also ro for you
<deltasquared>
hyper_ch: no, it is not. I can do that just fine
<deltasquared>
/nix is just a regular directory, no fancy protections going on here
<deltasquared>
hyper_ch: thatsthejoke.jpg
<hyper_ch>
right
<gchristensen>
on NixOS it is mounted read-only, deltasquared
<gchristensen>
well, /nix/store is
<hyper_ch>
yeah, nix isn't
<hyper_ch>
root@subi:/# touch nix/store/123
<hyper_ch>
touch: cannot touch 'nix/store/123': Read-only file system
<deltasquared>
gchristensen: I hadn't actually paid close enough attention to the mounts table when running the portable setup I made last week to spot that. TIL
vidbina has quit [Ping timeout: 248 seconds]
<deltasquared>
I mean it figures, immutability should be the default etc.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vNzN7
<NixOS_GitHub>
nixpkgs/python-unstable 25d1d49 Frederik Rietdijk: python.pkgs.bleach: relax test requirement
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
joko, tilpner: I'm getting it whining on "value is a string while a Boolean was expected" when trying to do "fooBar" -> "foo bar" in nix-repl (just fiddling around to try and get my head around it works before) the code I tried was: map ("fooBar" -> "foo bar") "fooBar"
<tilpner>
lejonet - While "->" is part of Nix syntax, it does something completely different
<tilpner>
lejonet - You need to use mapAttrs', like I said
<hyper_ch>
my new smart headphones still need to charge before they can make me smart also :(
<lejonet>
tilpner: then I'm not really getting how I should apply the transformations, should I do mapAttrs (name: value: nameValuePair (something to apply transformation to name) (value)) attrSet?
<tilpner>
lejonet - Look at lib.mapAttrs' (k: v: lib.nameValuePair (lib.toUpper k) v) { foo = 42; } in nix-repl
<lejonet>
Because if I need to hard code the transformation in the function given to mapAttrs, I'll have to have one call for each transformation, and that is all okay and so, but then I kindof need to know how to actually apply it
* deltasquared
reads the default.nix for fetchgitlocal
<deltasquared>
o.o what's it doing to my poor git index
b has joined #nixos
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] peti pushed 1 new commit to master: https://git.io/vNzAk
<NixOS_GitHub>
nixpkgs/master c6678c5 Peter Simons: pycurl: don't pollute $out/lib with system libraries...
<tilpner>
lejonet - Run nix-repl '<nixpkgs>' to bring lib into scope
kai_w has joined #nixos
<lejonet>
tilpner: well I get that, but how do I apply the "fooBar" -> "foo bar" function on the name? doing lib.mapAttrs' (k: v: lib.nameValuePair ("fooBar" -> "foo bar" k) v) { fooBar = 42; } gives me same error
<tilpner>
lejonet - Here, toUpper is your transformation function
<lejonet>
Yeah, that is what I'm using
<lejonet>
(nix-repl '<nixpkgs>'
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vNzAt
<NixOS_GitHub>
nixpkgs/python-unstable 21a736b Frederik Rietdijk: python.pkgs.secp256k1: 0.12.1 -> 0.13.2
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
)
<tilpner>
lejonet - I'll PM the rest
<lejonet>
tilpner: thank you, I'm very lost xD
simendsjo has joined #nixos
<azazel>
hyper_ch: smart headphones? :-)
<hyper_ch>
azazel: nuheara iqbuds :)
<hyper_ch>
but I'm sceptical... my smartphones didn't make me smart either
<azazel>
i bet that
<samae>
praise the person(s) who wrote the nat.nix module \o/
<hyper_ch>
what does that do?
<samae>
it sets up nat for you :D
<deltasquared>
I would guess set up port forwards and things of that nature to make NAT work
<hyper_ch>
it does?
<deltasquared>
as well as poke the sysfs knobs
<deltasquared>
(logical guesses based on my experiences on other distros)
stphrolland has quit [Remote host closed the connection]
<hyper_ch>
you use nixos on a router?
<deltasquared>
FRidh2: actually it turns out those expressions *are* present in the nix store, though a little bit out of the way. /nix/var/nix/profiles/per-user/$user/channels appears to be an alias to the relevant store entry
* gchristensen
also uses nixos on his router
mario_ has quit [Quit: mario_]
<lassulus>
what router does support nixos?
<gchristensen>
an old, regular system
<deltasquared>
depends if you mean actual router form factor hardware or not
<FRidh2>
deltasquared: eventually copies can exist in the store
<lassulus>
and does it provice an access point?
<azazel>
gchristensen: and what do you use for the firewall rules? the standard "firewall" config seems pretty basic
<FRidh2>
aborsu: nice work
<gchristensen>
lassulus: it did for about a year, before I got a Ubiquiti AP
<hyper_ch>
I use turris omnia for router currently :)
<deltasquared>
gchristensen: don't look at my firewall scripts, they're a bit of a hack. :P
<gchristensen>
:)
<deltasquared>
(incidentally the whole reason I'm on here today, to try and grok how to throw them into something I can get on an nixos system)
* azazel
feels a bit nostalgic about shorewall, and wonders if it make sense to use it in nixos
<gchristensen>
deltasquared: depends on what they are
<deltasquared>
gchristensen: a bunch of shell scripts and systemd units.
<gchristensen>
ack
<joko>
azazel: I am porting it for my employer, hopefully I will be able to release it as open source once it is ready
<deltasquared>
gchristensen: I'm getting my head around some of the basic stuff first so I can have a go and blow stuff up to figure out the best approach (which may include some modifications to the scripts themselves)
<deltasquared>
most noticeably where some of the "primitives" like runCommand come from
<deltasquared>
(because curiosity, and why not!?)
<azazel>
joko: wonderful! any ET?
<joko>
But imho one should write nftables
<joko>
azazel: I would say end of next month, I still have some other deadlines to catch first
Isorkin_ has joined #nixos
<kai_w>
virt-manager says 'could not detect a default hypervisor', but adding connections works fine. Is there any way to set the default hypervisor to connect to?
<azazel>
joko: I've yet to tackle on nftables, but with shorewall it's possible to have very complex configurations in just few lines, it's awesome
<azazel>
... (and intepolating strings is the right job for perl) :-)
Isorkin has quit [Ping timeout: 240 seconds]
vcunat has quit [Ping timeout: 256 seconds]
<deltasquared>
joko: my firewall scripts use nftables, they're essentially generators for per-interface skeleton stuff and then I just write the rules for individual interfaces. https://github.com/thetaepsilon/nftfirewall
<azazel>
oh, made in bash...
<deltasquared>
I think actually there is only one bashism I end up using
<deltasquared>
but for the most part they are posix sh
<deltasquared>
https://ptpb.pw/QbD5 would be an example fragment - these specify the rules for my current profile's outbound packets on my wireless interface.
<deltasquared>
(the jumped-to things are some inbuilt chains I have written)
michaelpj_ has joined #nixos
iyzsong-x has quit [Ping timeout: 268 seconds]
<joko>
deltasquared: this is awesome! Thanks for mentioning ti
<deltasquared>
joko: ... hey now, be careful, I haven't written any concrete documentation for it yet
<deltasquared>
(this is mine. thetaepsilon is my github handle)
<deltasquared>
joko: just let me know if you end up using it. although it's on github I tend to assume nobody does.
<deltasquared>
though then again I guess I would assume so anyway until someone started whining that I broke something for them ;)
<joko>
deltasquared: will do, thanks :D
<deltasquared>
that moment when you realise you may have just dug yourself a hole
<joko>
Hahah
<joko>
Somewhat related, I have never found an application-level firewall for Linux, something like Little Snitch for macOS. Is anyone aware of something relevant?
Rusty1 has joined #nixos
<deltasquared>
joko: not really. if it was system services I generally resorted to ensuring distinct services had their own UID, but for user-scale application filtering... no idea
<deltasquared>
last time I pondered this the conclusion I drew was that one would need e.g. a SELinux-enabled setup to be able to securely distinguish between apps running as the same user from say nftables
<deltasquared>
android got around this problem by giving each app it's own uid and just using that for rules in iptables >_>
<gchristensen>
joko: I've heard of something ... let me see if I can find it
<joko>
gchristensen: So far I have seen Douane and OpenSnitch, has anyone tried them?
<gchristensen>
yeah those are thee ones I've seen, and no I haven't tried them
<joko>
What's the third one?
<gchristensen>
those are the ones* my keyboard repeats e's a lot :(
<joko>
lol, no worries
yarny has quit [Quit: Page closed]
<azazel>
application-level firewall?
<deltasquared>
azazel: "this app can send to this port, this one can't" etc. firewall rules with application level granularity
<aborsu>
If I want to add modules to pkgs/development/node-packages/node-packages-v6.json can I add multiple modules in one commit?
<aborsu>
Or should I add them one by one and regenerate pkgs/development/node-packages/node-packages-v6.nix each time
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vNzpD
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging 176a035 Frederik Rietdijk: Merge pull request #34077 from NixOS/python-unstable...
<ixxie>
FRidh2: I am in the process of fixing the jupyterhub PR; aborsu is helping me maintain the history but cstrahan added a couple dozen node libraries and commiting them seperately would be a pain, so I figure I would group them.
coot_____ has joined #nixos
<ixxie>
FRidh2: would that be acceptable? I could call the commit something like `node-packages: added various jupyterhub dependencies.` this would also help clarify why they were added...
coot______ has quit [Ping timeout: 260 seconds]
<FRidh2>
each node expression / package set should be in a separate commit
<NixOS_GitHub>
[nixpkgs] FRidh deleted python-unstable at 21a736b: https://git.io/vNzhK
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh created python-unstable (+9 new commits): https://git.io/vNzhi
<NixOS_GitHub>
nixpkgs/python-unstable a6710a5 Frederik Rietdijk: python: alembic: 0.9.6 -> 0.9.7
<NixOS_GitHub>
nixpkgs/python-unstable d73f1a9 Frederik Rietdijk: python: ECPy: 0.8.2 -> 0.8.3
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/python-unstable e1abb6d Frederik Rietdijk: python: asana: 0.6.5 -> 0.6.7
mmlb has joined #nixos
<deltasquared>
rargh, functional languages are at once awesome and mind bending...
<gchristensen>
what're you having a tough time with?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rycee pushed 3 new commits to master: https://git.io/vNzhS
<NixOS_GitHub>
nixpkgs/master 663d827 Robert Helgesson: eclipse-plugins-jsonedit: init at 1.0.1...
<NixOS_GitHub>
nixpkgs/master 3a70f21 Robert Helgesson: eclipse/plugins: support multiple plugins JARs
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master b57981e Robert Helgesson: eclipse-plugin-cup: use `buildEclipsePlugin`...
<deltasquared>
gchristensen: nothing in particular thus far, I am becoming acquainted with the build support functions nix offers
<deltasquared>
source code is the best documentation, after all. it can't lie (assuming it's in sync)
<hyper_ch>
source code can't lie?
<deltasquared>
hyper_ch: ... well, not unless constructed deliberately to mislead
<gchristensen>
deltasquared: sounds good, let us know if you need help with something. Nix was my first functional language
<deltasquared>
hyper_ch: in this context seeing as I couldn't find any documentation on I think it was fetchgit, I elected to understand myself what it was doing
<MichaelRaskin>
hyper_ch: like a contract with devil — exact words
<deltasquared>
MichaelRaskin: or a genie.
<MichaelRaskin>
And then a compiler or a CPU gets too smart and just breaks the contract outright.
<deltasquared>
side channels or "optimisations"?
<deltasquared>
(don't get me started on memory orderings)
<gchristensen>
why not both!
<gchristensen>
get paid twice!
<deltasquared>
gchristensen: that must be intel's plan recently >_>
<gchristensen>
three times in intel's case, when everyone buys new CPUs :)
<MichaelRaskin>
Meltdown played out relatively well for Intel: in server market it damages Intel _and_ ARM in favour of AMD, but how to push out AMD is a well studied art at Intel.
<hyper_ch>
I wonder if Intel could get sued
<deltasquared>
"this is the 9001th time we have beaten AMD, and we have become exceedingly efficient at it."
<MichaelRaskin>
If someone even claims that 1995 papers were enough for the actions by Intel to count of negligence, we could see some useful change…
<deltasquared>
MichaelRaskin: as in, papers in 1995 that suggested this could happen?
<lejonet>
hyper_ch: iirc they are being sued for meltdown and spectre
<MichaelRaskin>
Yes.
<lejonet>
same as AMD is being sued for being "misleading" with their info too
<lejonet>
deltasquared: its hinted in Intels own documentation that the hole that now is meltdown and spectre exists
<hyper_ch>
from what I've gathered intel didn't just think it could happen but they kinda knew it can
coot_____ has quit [Ping timeout: 255 seconds]
<MichaelRaskin>
The actual high-assurance security researchers published in 1995 that having an architecture full of side-channels (cache is not the only one) _and_ speculative execution _and_ no clear documentation on unrolling side-channel-relevant changes is bad.
<deltasquared>
gchristensen: just reading nix-build's man page, it states that it's a wrapper around nix-instantiate and nix-store --realise. where it talks about "high level to low level" derivation... how are these two related? I had not previously seen this step mentioned
hakujin has joined #nixos
<lejonet>
This was also raised by Theo of openbsd already in 2006, but he was ofc brushed off as a "paranoid security person", but this is why openbsd isn't affected by them, because they already implemented their version of the KAISER/KPTI patch since 2006
<MichaelRaskin>
Of course nobody reads the actual research papers, so in 2005 it had to be reinvented.
<lejonet>
Atleast no one with the actual power to make the changes :P
<gchristensen>
deltasquared: it is high level expression to low level derivation
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar opened pull request #34081: gobjectIntrospection: use absolute path for cairo GIR (master...gir-abs-cairo) https://git.io/vNzj4
NixOS_GitHub has left #nixos [#nixos]
<MichaelRaskin>
Yes, it took ten years for _OpenBSD_ people to get paranoid.
<gchristensen>
deltasquared: nix expressions get turned in to a "derivation" which is the build instruction
<deltasquared>
gchristensen: those .drv entries in the store I take it?
<gchristensen>
deltasquared: that is the '.drv' yep
<MichaelRaskin>
I wonder what Skyfall is — speculative MMIO?
<MichaelRaskin>
Just the first idea I have when I hear a claim that something worse than Meltdown is yet unfixed.
<lejonet>
MichaelRaskin: yeah, that is going to be interesting to see, both skyfall and solace
dbe_ has quit [Ping timeout: 240 seconds]
<deltasquared>
I have my doubts about whether those are legit or not, but we shall see
srid_ has joined #nixos
statusfailed has joined #nixos
<MichaelRaskin>
Unclear what one can win out of creating these two websites without revealing something that is considered interesting.
<deltasquared>
MichaelRaskin: having a domain name ready, I guess
aborsu has left #nixos ["Au revoir"]
<MichaelRaskin>
And? I mean, if someone else finds something worth the name Skyfall, they can just call it ArmageddonAttack.
<MichaelRaskin>
(not yet registered)
<MichaelRaskin>
Or ZombieFlaw!
<lejonet>
I'm just waiting for some tired security researcher naming their attack "IntelIsStupidAgain" or "WhyDoWeContinueWithX86"
<MichaelRaskin>
JustSwitchToRiscV
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] zimbatm opened pull request #34082: cargo-vendor-bin: init at 0.1.13 (master...cargo-vendor-bin-0.1.13) https://git.io/vNzjK
NixOS_GitHub has left #nixos [#nixos]
srid[m] has joined #nixos
<deltasquared>
BestHardwareIsStupidHardware
<lejonet>
RiscVForPresident?
michaelpj_ has quit [Ping timeout: 248 seconds]
<lejonet>
ARMDidNothingWrong :P
periklis has joined #nixos
<ixxie>
I need to commit some node packages, but due to the autogenerated nature I cannot be sure of the version
<ixxie>
I added the package to node-packages-v6.json and generated the nix file, but grepping inside for the package yields two different versions
<ixxie>
and I am not sure which is the one I initialized it in xD
<MichaelRaskin>
Actually, speculative MMIO might be even more likely on aarch64
fragamus has joined #nixos
<gchristensen>
MichaelRaskin: why so?
<MichaelRaskin>
Age of architecture. Speculative MMIO on x86 would have already broken something by accident.
aarvar has quit [Ping timeout: 256 seconds]
aanderse has quit [Ping timeout: 276 seconds]
dywedir has quit [Ping timeout: 256 seconds]
srid_ has quit [Quit: leaving]
srid has quit [Quit: WeeChat 1.9.1]
srid_irccloud is now known as srid
srid has quit [Changing host]
srid has joined #nixos
srid has joined #nixos
ToxicFrog has left #nixos ["WeeChat 1.9.1"]
<deltasquared>
gchristensen: I may be being thick here but my brain is slowly beginning to fry learning new concepts... where's the documentation for mkDerivation hiding?
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
i-am-the-slime has joined #nixos
knupfer1 has joined #nixos
knupfer has quit [Ping timeout: 265 seconds]
knupfer1 is now known as knupfer
peacememories has joined #nixos
vidbina has joined #nixos
Isorkin_ has quit [Ping timeout: 248 seconds]
<juhe>
Hello, I've a problem setting (or better said keeping) cursor scheme in xmonad, problem is the default cursor theme set up by lightdm is reset when I run xmonad (none + xmonad specifically). Any idea how to resolve it?
fyuuri has joined #nixos
civodul has joined #nixos
<lassulus>
hmm, unity3d is broken for me. It fails with the error: RPATH of binary /nix/store/ba1m7m8kg43jlp9n6zr52khalfwclm7b-unity-editor-5.6.1xf1/opt/Unity/Editor/Data/Tools/libRL.so contains a forbidden reference to /build
tomberek has joined #nixos
fyuuri has quit [Read error: Connection reset by peer]
Isorkin has joined #nixos
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<andjscott>
samueldr: I didn't know that the matrix bridge was discouraged. Sorry about that
<samueldr>
not discouraged, but there are limitations that should be known :(
<samueldr>
the bridge itself is rude, not using the bridge
nuncanada has quit [Read error: Connection reset by peer]
tomberek has quit [Ping timeout: 260 seconds]
<MichaelRaskin>
I would say there are cases where collapsing is actually the correct behaviour by bridge — it replaces the need for pastebin
<MichaelRaskin>
But long messages in general are treated just like a pastebin linnk would be — opened if it is expected to be details of a conversation one follows (or participates in), ignored otherwise.
<LnL>
yeah, just start off with a short message
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<deltasquared>
argh, XSLT stylesheets? all I wanted was a command that went from docbook to (x)html.
<LnL>
also highlights won’t work for irc in long messages
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] adisbladis pushed 1 new commit to master: https://git.io/vNgfz
<deltasquared>
anyone here who can help me get something useful out of the docbook sources that comes with nixpkgs?
aanderse has joined #nixos
peacememories has joined #nixos
<deltasquared>
there are some xslt stylesheets available that can supposedly spit out xhtml (i.e. something I could shove into a browser), though it's giving me warnings I'm not sure how to interpret.
<deltasquared>
“Element include in namespace 'http://www.w3.org/2001/XInclude' encountered in book, but no template matches.”
Itkovian has joined #nixos
<LnL>
what are you trying to do, build the manual?
<deltasquared>
java? just to build some docbook stuff? welp
<MichaelRaskin>
Nix has to show off how it can magically install all the dependencies without breaking the rest of the system, right?
<joepie91>
deltasquared: pandoc may be what you're looking for btw, if you want a just-works docbook-to-something-else converter
<deltasquared>
joepie91: I recall not being happy with the amount of dependencies pandoc wanted to drag in.
<joepie91>
it's a little annoying to have to twiddle with its mountain of settings sometimes, and it has a few rough edges, but overall it works very well as a quick X-to-Y converter
<joepie91>
ah, no idea
* joepie91
does not really care about amount of dependencies :P
acarrico has quit [Ping timeout: 264 seconds]
<deltasquared>
you've never been a grumpy anti-bloat arch neckbeard evidently
<deltasquared>
which I can become at times -_-
<MichaelRaskin>
I use A/B LFS at some time.
<MichaelRaskin>
You Arch people don't even know what powers your boxes.
<MichaelRaskin>
used
<MichaelRaskin>
But come on, Nix handles dependencies just fine.
<deltasquared>
re arch, them fightin' words! :P
<MichaelRaskin>
NixOS people refuse to know, either.
<deltasquared>
right, how would one manually register a gc root for nix? I was just building the nixpkgs manual and that worked fine, however as it was within the store itself at the time it's refused to make that point a symlink.
<MichaelRaskin>
Both Arch and NixOS-mainline agreed to use systemd!
<MichaelRaskin>
nix-build -o
<MichaelRaskin>
You can just rebuild, the output path is already valid, so nothing expensive will happen.
<deltasquared>
MichaelRaskin: cheers.
Arcaelyx has joined #nixos
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
<joko>
Any thoughts on system.autoUpgrade and how to deal with a) kernel upgrades which require reboots and b) combining it with a Nixops deploy?
<joepie91>
samueldr: very similar packaging indeed, but never ran across that in my childhood so that wouldn't be it
<kolb>
so I noticed fontconfig is actually in many locations, suffixed with -dev/-bin/-lib, I have a nix expressions that gets library paths via ${package}/lib but that doesn’t work for fontconfg subsequently. How do I address "fontconfig-$version-lib"?
<joepie91>
possibly something else that cloned that design though, good catch
<MichaelRaskin>
kolb: fontconfig.lib
<MichaelRaskin>
This is called multiple-output derivation
<kolb>
MichaelRaskin: \o lisper high five and thanks!
jb55 has quit [Ping timeout: 264 seconds]
<MichaelRaskin>
Oh, I am being tracked. I mean, I haven't mentioned Lisp today.
<MichaelRaskin>
Only yesterday.
<kolb>
I just recognized your name from ELS
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<MichaelRaskin>
Ah
<kolb>
we had a technocratic vs my-idiot-worldview debate ;-)
<kolb>
another thing: any way I can "GC" only specific packages?
<MichaelRaskin>
Ah, now I remember who you are. Not really idiotic, just… I cannot believe that can be made to work fast enough this time.
<kolb>
but that is only for files on my system, right?
<samueldr>
no
<kolb>
cool
shabius has quit [Quit: Leaving]
jtojnar has quit [Remote host closed the connection]
thaega has quit [Remote host closed the connection]
<kolb>
is there a way to tell if a package has -lib/-dev splittings in advance? Or will ${pkg.lib} work for all packages that have DSOs?
xcmw has joined #nixos
<vcunat>
kolb: you can inspect pkg.outputs
<MichaelRaskin>
Well, there is lib.getLib
<vcunat>
yes, you were faster :-)
<MichaelRaskin>
I just didn't waste time on complicated approaches!
<vcunat>
right, that's the typical use case
<vcunat>
I had in plan to put .lib attribute even into packages without such output, but that would need some restructuring of the framework, and "more important" nix ideas keep coming all the time.
<cinimod>
This works `nix-shell -p "with import <nixpkgs> {}; python.withPackages (ps: [ ps.matplotlib ])"`
<cinimod>
`Your branch is up-to-date with 'origin/master`
pkill9 has quit [Ping timeout: 268 seconds]
MP2E has joined #nixos
hakujin has quit [Ping timeout: 256 seconds]
alex`` has quit [Quit: WeeChat 1.9]
pkill9 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 394e989 Robert Helgesson: perl-Log-Contextual: 0.008000 -> 0.008001
<NixOS_GitHub>
[nixpkgs] rycee pushed 1 new commit to master: https://git.io/vNgOl
NixOS_GitHub has left #nixos [#nixos]
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<FRidh2>
ison111: don't use `nix-shell p python36Packages.setuptools`. It's an old method that is no longer supported. For most packages it works (typically those that use buildPythonPackage) but for some it does not.
<ison111>
FRidh2: I see, thanks. Also, do you know if it's possible to make an environment that installs something from pip?
hakujin has joined #nixos
<FRidh2>
ison111: check pypi2nix
i-am-the-slime has quit [Ping timeout: 256 seconds]
<cinimod>
<FRidh2> ison111: don't use `nix-shell p python36Packages.setuptools`. It's an old method that is no longer supported. For most packages it works (typically those that use buildPythonPackage) but for some it does not.
<cinimod>
What should I use?
<cinimod>
> <FRidh2> ison111: don't use `nix-shell p python36Packages.setuptools`. It's an old method that is no longer supported. For most packages it works (typically those that use buildPythonPackage) but for some it does not.
<FRidh2>
instead of python3Packages you can use python3.pkgs
<cinimod>
And then `nix-shell -p "callPackage ./foo.nix {}" -I nixpkgs=/Users/dom/nixpkgs`
<cinimod>
:thumbs-up:
<cinimod>
👍
i-am-the-slime has joined #nixos
toogley has joined #nixos
<toogley>
how can i find out which package wants rustc pkg?
<toogley>
(defined in my configuration.nix )
<typetetris>
I noticed that nixos creates a link farm at / for me. So I have paths like /bin/sh. Is this strictly necessary on nixos, wouldn't an environment per user (under ~/.nix-<something>/) suffice? Can I deactivate that global environment somehow ?
fragamus has joined #nixos
<MichaelRaskin>
We-ell, glibc's system() gets unhappy without /bin/sh
<joepie91>
typetetris: generally NixOS will only link paths globally that are somehow necessary for correct functioning, eg. because some piece of software demands a global path (think windowing toolkit themes, and what MichaelRaskin mentioned, etc.)
<MichaelRaskin>
And the limit of portability of scripts in various languages is /usr/bin/env shebang…
jb55 has quit [Ping timeout: 256 seconds]
i-am-the-slime has quit [Ping timeout: 256 seconds]
<joepie91>
typetetris: you can probably disable it by modifying pathsToLink or whatever that option was called, but chances are that you won't end up with a functioning system :)
<MichaelRaskin>
And not having /etc/passwd might mean changing more than desired…
<typetetris>
Ok .. maybe I was to ambitious :)
<MichaelRaskin>
No, it is not pathsToLink, it is activation script
<typetetris>
Sounds like pain, trying this.
<joepie91>
typetetris: I agree that it's a bit of a ugly spot on an otherwise isolated distro, but there are some things that are just not easy to make other software un-assume :P
<MichaelRaskin>
I am trying to get purely read-only /etc/ (with symlinks to writeable targets where needed) on something nix-based but not NixOS; this is already a pain
kp__ has joined #nixos
<typetetris>
hmm, which programs want to write configuration files under `/etc` ?
<MichaelRaskin>
On the other hand, with read-only /etc I will finally have what NixOS promised but only approximated: the switch is just atomic replacement of the current-system symlink, all the rest goes through it.
<MichaelRaskin>
typetetris: samba, for example
<typetetris>
What a horrible design decision.
<MichaelRaskin>
Also, I do want writeable /etc/passwd, just not writeable via /etc
<MichaelRaskin>
Well, samba config can be just generated.
dbe_ has joined #nixos
<MichaelRaskin>
But there is /etc/resolv.conf in some cases
ixxie has joined #nixos
<typetetris>
True.
<typetetris>
And one doesn't want to switch generations for each change of wlans.
<MichaelRaskin>
I think NixOS CUPS service wants /etc/cups
<MichaelRaskin>
And it even assumes that it is a symlink to /var/lib/cups
nuncanada has joined #nixos
<MichaelRaskin>
(And this has to be writeable, because printer list and there are interesting situations)
<samueldr>
hyper_ch: contents blocker are the most probable cause of connection issue, AFAIK
<typetetris>
If I configure a grub device in /etc/nixos/configuration.nix how often will grub be installed? I thought one time should suffice, every other change only should need to change the grub.cfg? (I ask since I have a strange bios, where I have to add a trusted boot loader in the bio every time grub64.efi gets written.)
<gchristensen>
azazel: stomp.js which is STOMP over websockets
zarel has joined #nixos
i-am-the-slime has quit [Ping timeout: 276 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 8 new commits to master: https://git.io/vNgG4
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 55f966b Charles Strahan: nodePackages.clean-css: init at 4.1.9
<NixOS_GitHub>
nixpkgs/master c8300a0 Matan Shenhav: pythonPackage.python-oauth2: init at 1.0.1
<NixOS_GitHub>
nixpkgs/master 850799e Charles Strahan: pythonPackages.pamela: fix resolution of libpam.so
<ixxie>
w00t!
i-am-the-slime has joined #nixos
joko has quit [Quit: leaving]
jtojnar has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 opened pull request #34088: xcbuild: platform version (master...xcbuild-platform-version) https://git.io/vNgG9
NixOS_GitHub has left #nixos [#nixos]
juhe has quit [Quit: Leaving.]
Slabity has joined #nixos
i-am-the-slime has quit [Ping timeout: 256 seconds]
<Slabity>
Would it be difficult to have a tmpfs as the root filesystem in a nixos system? Like only having /nix and /home on actual physical devices?
dywedir has quit [Read error: Connection reset by peer]
<Slabity>
Then have the init system set everything up with all the symlinks necessary, with the tmpfs system being readonly.
i-am-the-slime has joined #nixos
vcunat has quit [Quit: Leaving.]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vNgZU
<NixOS_GitHub>
nixpkgs/staging 75d8096 Frederik Rietdijk: Merge remote-tracking branch 'upstream/master' into HEAD
NixOS_GitHub has left #nixos [#nixos]
kai_w has quit [Quit: Konversation terminated!]
ma271 has quit [Ping timeout: 276 seconds]
hakujin has joined #nixos
jensens has quit [Ping timeout: 256 seconds]
i-am-the-slime has quit [Ping timeout: 256 seconds]
<NixOS_GitHub>
[nixpkgs] peti pushed 5 new commits to master: https://git.io/vNgno
<NixOS_GitHub>
nixpkgs/master 8bf8581 Peter Simons: hackage2nix: disable broken Hydra builds...
<NixOS_GitHub>
nixpkgs/master a8d9a94 Peter Simons: haskell.lib.markBroken: explicitly disable Hydra builds...
<NixOS_GitHub>
nixpkgs/master 9688c0e Peter Simons: haskell-hlint: use latest version by default
NixOS_GitHub has left #nixos [#nixos]
magnetophon has quit [Remote host closed the connection]
<ixxie>
FRidh2: how do I reliably tell if tests are actually run?
<ixxie>
Also, relatedly: how can I tell if something needs to be a propogatedBuildInput or a mere buildInput?
mk-fg has quit [Ping timeout: 256 seconds]
shiver has quit [Remote host closed the connection]
coot______ has quit [Quit: coot______]
dan_b has joined #nixos
coot______ has joined #nixos
ma27 has joined #nixos
leapingfrogs has joined #nixos
mk-fg has joined #nixos
mk-fg has quit [Changing host]
mk-fg has joined #nixos
FRidh2 has quit [Quit: Konversation terminated!]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] mnacamura opened pull request #34089: SDL2_{gfx,mixer,net,ttf}: fix Darwin build (master...sdl2-darwin-fix) https://git.io/vNgcZ
NixOS_GitHub has left #nixos [#nixos]
leapingfrogs has quit [Remote host closed the connection]
shiver32 has joined #nixos
shiver32 has joined #nixos
shiver32 has quit [Read error: Connection reset by peer]
<Kim>
So browsing around I saw something mention `lib.getBin`, but I can't seem to find it in the various manuals on the site. Is this a feature that was pulled out?
<MichaelRaskin>
It is still in lib, and I do not think it is going away soon.
<MichaelRaskin>
Not yet documented — well, there are many useful things in Nixpkgs that are not documented…
<Kim>
So to the source code it is xD
<etu>
Kim: Sadly true for many projects...
<etu>
Kim: If you haven't discovered ripgrep for searching code, I suggest trying it :)
<MP2E>
ripgrep is a lifesaver when it comes to combing through nixpkgs fast
<Kim>
I tend to use ack; haven't tried ripgrep though
shiver has joined #nixos
shiver has quit [Changing host]
shiver has joined #nixos
<etu>
the_silver_searcher (ag) was developed to be faster than ack, then ripgrep came around and is faster than ag :)
<etu>
But the idea is the same
<MichaelRaskin>
Dunno, once nixpkgs is in read cache, GNU grep is fast enough for me
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vNgCY
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 74e2b60 Tuomas Tynkkynen: libsigsegv: 2.11 -> 2.12
<etu>
ripgrep respects things like .gitignore properly :)
<etu>
and don't search ignored areas
<samueldr>
(ag does too)
<etu>
samueldr: nope
<Kim>
Huh. Nifty
<etu>
ag has it's own agignore
<samueldr>
maybe the keyword is properly
<MichaelRaskin>
To be honest, I just search pkgs/ or nixos/
<Kim>
I'll have to play with it
<etu>
samueldr: yeah, it does partly, but not properly :)
<MP2E>
the real point of all this is that searching nixpkgs with a tool similar to grep (or grep itself) is pretty handy :p
<Kim>
Hm. Neat. Seems someone else has already started on a package for Shashlik
bpa has quit [Read error: Connection reset by peer]
jensens has joined #nixos
bpa has joined #nixos
ryanartecona has quit [Quit: ryanartecona]
<va1entin>
I have a question regarding the binary cache: I am on the 17.09 channel, but when I try to install or start a shell with libreoffice, NixOS builds it from scratch. This started today, when I did a nixos-rebuild. Does anyone have an idea what might be causing this?
<etu>
samueldr: Then it might have changed, but ripgrep is still faster :p
<etu>
"Like The Silver Searcher, ripgrep defaults to recursive directory search and won’t search files ignored by your .gitignore files. It also ignores hidden and binary files by default. ripgrep also implements full support for .gitignore, where as there are many bugs related to that functionality in The Silver Searcher. Of the things that don’t work in The Silver Searcher, ripgrep supports .gitignore priority
<etu>
(including in parent directories and sub-directories), whitelisting and recursive globs."
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vNgCD
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 45e47c1 Tuomas Tynkkynen: devicemapper: 2.02.176 -> 2.02.177
coot_______ has joined #nixos
fleaz has quit [Quit: fleaz]
megfault has quit [Quit: WeeChat 2.0.1]
coot______ has quit [Ping timeout: 264 seconds]
<joepie91>
dash: you'll have a problem with that; the `npm install` can never succeed because the builder isn't allowed to use the network (and this is what node2nix takes care of anyway)
<joepie91>
dash: besides that, npm scripts are executed with ./node_modules/.bin at the start of the PATH, so if you replicate that, then you should be able to run `node build.js`
<ixxie>
anybody familiar with python packaging?
<ixxie>
I cannot figure out how to check if a test is running or not
<ixxie>
or if a package should be propagated or not
<joepie91>
dash: (node2nix basically generates the expression that produces the node_modules, in lieu of npm doing that during `npm install`, so after that's done you can just run the build script yourself if the PATH env var is correct)
<dash>
joepie91: eh, I don't mind patching stuff here to make it work, I just don't see where to hook into node2nix to run stuff during the build
<joepie91>
hmmm...
coot_______ has quit [Quit: coot_______]
<joepie91>
dash: a hack: specify it as the postinstall script in the package.json
ma27 has quit [Ping timeout: 276 seconds]
coot_______ has joined #nixos
<joepie91>
dash: it seems that the node2nix expression does actually run `npm install` at the end as a check, so presumably it should also trigger the postinstall hook then
<dash>
hm. yeah i guess it'd work as postinstall
<joepie91>
not sure if there's a nicer, less hacky way to do this
<dash>
thanks, i'll fool with it some more
<joepie91>
I feel like there should be :P
ma27 has joined #nixos
verite has joined #nixos
bpa has quit [Read error: Connection reset by peer]
<joepie91>
ah, the `src` is the tarball of that actually
swayf has quit [Read error: Connection reset by peer]
coot_______ has quit [Quit: coot_______]
coot_______ has joined #nixos
marusich has quit [Quit: Leaving]
swayf has joined #nixos
bpa has quit [Read error: Connection reset by peer]
bpa has joined #nixos
hiratara has quit [Ping timeout: 276 seconds]
raynold has quit [Quit: Connection closed for inactivity]
hiratara has joined #nixos
periklis has quit [Ping timeout: 276 seconds]
jb55 has quit [Quit: WeeChat 1.9]
jb55 has joined #nixos
shiver has quit [Ping timeout: 264 seconds]
mk-fg has quit [Ping timeout: 256 seconds]
threepointone has joined #nixos
fgaz[m] has quit [*.net *.split]
bennofs[m] has quit [*.net *.split]
AllanDaemon has quit [*.net *.split]
Diagonal[m] has quit [*.net *.split]
tobim[m] has quit [*.net *.split]
paperdigits[m] has quit [*.net *.split]
aniketd[m] has quit [*.net *.split]
mith[m] has quit [*.net *.split]
itorres has quit [*.net *.split]
danielrf has quit [*.net *.split]
georgew has quit [*.net *.split]
themistocle has quit [*.net *.split]
houli has quit [*.net *.split]
indika has quit [*.net *.split]
dhess has quit [*.net *.split]
siel has quit [*.net *.split]
admin[m]3 has quit [*.net *.split]
scott1 has quit [*.net *.split]
apurvapavaskar[m has quit [*.net *.split]
cornu[m]1 has quit [*.net *.split]
fogbugzz[m] has quit [*.net *.split]
equalunique[m] has quit [*.net *.split]
bachp has quit [*.net *.split]
NickHu1 has quit [*.net *.split]
aaronc[m] has quit [*.net *.split]
fearlessKim[m] has quit [*.net *.split]
wmertens[m] has quit [*.net *.split]
revoltmedia[m] has quit [*.net *.split]
DIzFer[m] has quit [*.net *.split]
nwspk has quit [*.net *.split]
hedning[m] has quit [*.net *.split]
midchildan[m] has quit [*.net *.split]
Rick[29SMPittsbu has quit [*.net *.split]
AlanPearce[m] has quit [*.net *.split]
Elephant454[m] has quit [*.net *.split]
reactormonk[m] has quit [*.net *.split]
regnat[m] has quit [*.net *.split]
ConorCurry[m] has quit [*.net *.split]
demyan[m] has quit [*.net *.split]
Drakonis[m] has quit [*.net *.split]
primeos[m] has quit [*.net *.split]
mankyKitty[m] has quit [*.net *.split]
necronian has quit [*.net *.split]
herzmeister[m] has quit [*.net *.split]
bendlas has quit [*.net *.split]
davidar has quit [*.net *.split]
copumpkin has quit [*.net *.split]
sphalerite has quit [*.net *.split]
TimePath has quit [*.net *.split]
paraseba has quit [*.net *.split]
lluchs has quit [*.net *.split]
scode has quit [*.net *.split]
dfranke_ has quit [*.net *.split]
pingveno has quit [*.net *.split]
philipcristiano has quit [*.net *.split]
dvim has quit [*.net *.split]
mgttlinger has quit [*.net *.split]
johs has quit [*.net *.split]
hexagoxel has quit [*.net *.split]
armijn has quit [*.net *.split]
bara has quit [*.net *.split]
alanz has quit [*.net *.split]
fadenb has quit [*.net *.split]
itscaleb has quit [*.net *.split]
bara has joined #nixos
siel has joined #nixos
dhess has joined #nixos
fadenb has joined #nixos
armijn has joined #nixos
nwspk has joined #nixos
lluchs has joined #nixos
philipcristiano has joined #nixos
dfranke has joined #nixos
hexagoxel has joined #nixos
johs has joined #nixos
paraseba has joined #nixos
mgttlinger has joined #nixos
houli has joined #nixos
pingveno has joined #nixos
scode has joined #nixos
georgew has joined #nixos
indika has joined #nixos
coot_______ has quit [Quit: coot_______]
dvim has joined #nixos
<neonfuz>
sauerbraten isn't building
themistocle has joined #nixos
<neonfuz>
for me
coot_______ has joined #nixos
<neonfuz>
if someone else here can try building it and tell me if it builds I'd appreciate it
<neonfuz>
(sauerbraten-r5000 is the package name)
aaronc[m] has joined #nixos
fogbugzz[m] has joined #nixos
dywedir has joined #nixos
apurvapavaskar[m has joined #nixos
mith[m] has joined #nixos
midchildan[m] has joined #nixos
hedning[m] has joined #nixos
regnat[m] has joined #nixos
Elephant454[m] has joined #nixos
itscaleb has joined #nixos
fearlessKim[m] has joined #nixos
ConorCurry[m] has joined #nixos
paperdigits[m] has joined #nixos
primeos[m] has joined #nixos
reactormonk[m] has joined #nixos
Drakonis[m] has joined #nixos
Rick[29SMPittsbu has joined #nixos
<neonfuz>
I think it's related to gcc-6.4.0/include/c++/6.4.0/math.h, and things not being in scope, so I think it's a gcc / stdlib version thing?
jtojnar has quit [Remote host closed the connection]
demyan[m] has joined #nixos
mankyKitty[m] has joined #nixos
danielrf has joined #nixos
aniketd[m] has joined #nixos
revoltmedia[m] has joined #nixos
tobim[m] has joined #nixos
wmertens[m] has joined #nixos
admin[m]3 has joined #nixos
Diagonal[m] has joined #nixos
DIzFer[m] has joined #nixos
bpa has quit [Read error: Connection reset by peer]
AllanDaemon has joined #nixos
NickHu1 has joined #nixos
scott1 has joined #nixos
<MichaelRaskin>
That might be… Sauerbraten package is quite old, and it might have suffered from some gcc update.
bennofs[m] has joined #nixos
necronian has joined #nixos
fgaz[m] has joined #nixos
TimePath has joined #nixos
bendlas has joined #nixos
bachp has joined #nixos
cornu[m]1 has joined #nixos
itorres has joined #nixos
<neonfuz>
yeah, :/
herzmeister[m] has joined #nixos
AlanPearce[m] has joined #nixos
davidar has joined #nixos
copumpkin has joined #nixos
equalunique[m] has joined #nixos
bpa has joined #nixos
sphalerite has joined #nixos
jtojnar has joined #nixos
shiver has joined #nixos
shiver has quit [Changing host]
shiver has joined #nixos
<neonfuz>
it fails right away, like in the first few files it compiles, I think in cube.h
avn has quit [Ping timeout: 260 seconds]
<samueldr>
neonfuz: using nixpkgs 17.09 or unstable?
<neonfuz>
unstable
<samueldr>
ah
<neonfuz>
is there a way to compile with a different gcc version or something to fix this?
avn has joined #nixos
<samueldr>
yeah, you probably had "code" already built, but as soon as it built the actual sauerbraten it failed for me too
mk-fg has joined #nixos
mk-fg has quit [Changing host]
mk-fg has joined #nixos
<MichaelRaskin>
Well, just checking out the source is not «right away»
va1entin has quit [Quit: WeeChat 1.9.1]
<neonfuz>
samueldr: is there some sort of code caching?
<neonfuz>
or build continuing (in nixos)
<MichaelRaskin>
You could pass stdenv = overrideCC stdenv gcc5;
* samueldr
checks how sauerbraten is made before answering
<samueldr>
17.09 did build correctly
<neonfuz>
alright, hmm
jtojnar_ has joined #nixos
<threepointone>
Can anyone point me to resource(s) where I can get a high-level overview of NixOS works? There are many things that are completely different than what I'm used to with other traditional Linux distributions and there are lots of symlinks which I'm not sure how to interpret. If there is a comparison somewhere about how NixOS achieves stuff vs how that stuff was achieved before, that'd really help.
<neonfuz>
can I build a 17.09 build and install it into my nix-env?
<samueldr>
you could checkout nixpkgs 17.09 in a folder, then `nix-build -A sauerbraten` at the root
<samueldr>
this will build it, leaving result as a symlink
<samueldr>
nix-env can be used too this way to install, but I'm not used yet to its use
<azazel>
threepointone: do you actually use any linux distribution?
shiver has quit [Remote host closed the connection]
<threepointone>
azazel: I've mostly used Debian and Arch before. I also used Gentoo for a short time. Right now, I've got no OS installed and I'm trying to learn about Nix and Guix. :)
<neonfuz>
MichaelRaskin: just wondering, can that then be installed into the nix-environment?
<neonfuz>
I'm just trying to understand exactly how nix-env works
<ixxie>
threepointone: since all of this stuff is relatively new, we appreciate feedback :)
<MichaelRaskin>
neonfuz: well, you could probably pass -E 'expression'
<MichaelRaskin>
Or just write a Nix expression in a file (importing <nixpkgs>) and use -f
mk-fg has quit [Ping timeout: 256 seconds]
<threepointone>
ixxie: Sure. I'm gonna go through these three articles and then give you and idea about what I'm expecting to see as a beginner in a guide that tries to introduce NixOS.
shiver has joined #nixos
shiver has joined #nixos
<ixxie>
threepointone: there is lots we are aware is missing but we haven't had time to write yet, but we are happy with the start :) This wiki didn't exist in this form in August
shiver has quit [Remote host closed the connection]
ma27 has quit [Ping timeout: 256 seconds]
shiver has joined #nixos
alanz has joined #nixos
ma27 has joined #nixos
<threepointone>
ixxie: I'm a bit surprised to see that NixOS doesn't have an official wiki like Arch does. There is just a long manual which writes about things like VirtualBox and DNSCrypt but doesn't go into detail about the numerous symlinks spread throughout the system and how they all "fit together" to make things work.
<MichaelRaskin>
neonfuz: But maybe it is about glibc. And apparently we don't have an older one…
<MichaelRaskin>
Or maybe it is time to update Sauerbraten
<ixxie>
threepointone: the Wiki had a complicated history; it used to be plagued by spam, so it lost official support. We are now working on it again to get it up to standard, and will hopefully file an RFC in the coming months to ask for official support once again.
coot_______ has quit [Remote host closed the connection]
<ennui>
i'm running nixos and i want to install nixos to a specific device (/dev/sdb) which i'll use as the boot device on another system, where it'll show up as /dev/sda.
<ennui>
i've run nixos-install --closure my-nixos-system --no-bootloader
<ennui>
what's the correct grub-install command to finish the installation?
<ennui>
i'm not sure if the --root-directory argument is relevant
<ennui>
sphalerite, yes, i considered this, too, but GRUB doesn't understand UUIDs
<ennui>
i'm forced to use grub.device="/dev/sda"
<ennui>
forgot to mention: i'm using MBR
<sphalerite>
ennui: /dev/disk/by-id/stuff
<sphalerite>
ls -l /dev/disk/by-id to find out which one is the right one
<ennui>
okay, i'll try that. and then let nixos-install handle the GRUB installation?
<sphalerite>
yeah that should work
<ennui>
great, i'll try that
<ixxie>
sphalerite: any clue about how I can know if a test is being run when I build a package?
<ixxie>
(a python package to be exact?)
<sphalerite>
ixxie: look at the build log I guess. Why me specifically?
zarel has quit [Quit: Leaving]
<ixxie>
sphalerite: because you were just active and are experienced xD
bpa has quit [Remote host closed the connection]
<ixxie>
how do I inspect the build log?
<ennui>
ixxie, you can also check the 'doCheck' attribute of the derivation
<sphalerite>
ixxie: nix-store --read-log <path to drv>
<sphalerite>
Or read-logs, I always forget
<ixxie>
ennui: well, I am writing these derivations and the question is whether or not tests are redundent and a doCheck = false is necessary, so...
<ixxie>
cheers sphalerite
<ennui>
ixxie, it's true by default for python packages
ssmike has quit [Remote host closed the connection]
<samueldr>
ennui: while the configuration option can't use uuid (there are none for the drive) you should be able to target /dev/disk/by-id/_somthing_
savanni has quit [Read error: Connection reset by peer]
<samueldr>
I'm using this on a bios/mbr setup /dev/disk/by-id/ata-WDC_WD5001AALS-00L3B2_WD-WCASY6986843
gaqzi has quit [Read error: Connection reset by peer]
<ennui>
samueldr, thanks, that's what sphalerite suggested. i'm trying this out right now.
savanni has joined #nixos
<ixxie>
ennui: FRidh asked me to set it to false if tests are not necessary, so I am trying to figure that out
r0bby has quit [Read error: Connection reset by peer]
gaqzi has joined #nixos
<samueldr>
ahk, foiled again by not reading the backlog!
jmeredith has quit [Read error: Connection reset by peer]
<ennui>
:)
linuxdaemon has quit [Ping timeout: 246 seconds]
jmeredith has joined #nixos
grenade has quit [Ping timeout: 246 seconds]
feepo has quit [Ping timeout: 246 seconds]
madknight has quit [Ping timeout: 246 seconds]
boegel has quit [Remote host closed the connection]
r0bby has joined #nixos
<samueldr>
fun note: bios boot doesn't necessarily mean that gpt can't be used, but if you ever use gpt, grub will want a specific partition to use in addition to the normal partitions
justanotheruser has quit [Ping timeout: 248 seconds]
<samueldr>
(though, obviously, there won't be an ESP)
leapingfrogs has joined #nixos
wmertens[m] has quit [Ping timeout: 246 seconds]
mith[m] has quit [Ping timeout: 246 seconds]
iohidtbhfam[m] has quit [Ping timeout: 246 seconds]
adisbladis[m] has quit [Ping timeout: 246 seconds]
offlinehacker[m] has quit [Ping timeout: 246 seconds]
voobscout[m] has quit [Ping timeout: 246 seconds]
Wysteriary[m] has quit [Ping timeout: 246 seconds]
gridaphobe[m] has quit [Ping timeout: 246 seconds]
AntonLatukha[m] has quit [Ping timeout: 246 seconds]
benkolera has quit [Ping timeout: 246 seconds]
wak-work has quit [Ping timeout: 246 seconds]
sudoreboot[m] has quit [Ping timeout: 246 seconds]
mujx[m] has quit [Ping timeout: 246 seconds]
hl has quit [Ping timeout: 246 seconds]
scott1 has quit [Ping timeout: 246 seconds]
TimePath has quit [Ping timeout: 246 seconds]
bachp has quit [Ping timeout: 246 seconds]
Diagonal[m] has quit [Ping timeout: 246 seconds]
leapingf1 has joined #nixos
<MichaelRaskin>
You only now discovered?
herzmeister[m] has quit [Ping timeout: 246 seconds]
davidar has quit [Ping timeout: 246 seconds]
equalunique[m] has quit [Ping timeout: 246 seconds]
moredhel1 has quit [Ping timeout: 246 seconds]
wervenyt[m] has quit [Ping timeout: 246 seconds]
icetan has quit [Ping timeout: 246 seconds]
rihardsk[m] has quit [Ping timeout: 246 seconds]
stites[m] has quit [Ping timeout: 246 seconds]
andyjscott[m] has quit [Ping timeout: 246 seconds]
sdemos has quit [Ping timeout: 246 seconds]
r3mfKm[m] has quit [Ping timeout: 246 seconds]
bhipple[m] has quit [Ping timeout: 246 seconds]
seif_madc[m] has quit [Ping timeout: 246 seconds]
jack[m] has quit [Ping timeout: 246 seconds]
octalsrc[m] has quit [Ping timeout: 246 seconds]
peterhoeg has quit [Ping timeout: 246 seconds]
randomenduser[m] has quit [Ping timeout: 246 seconds]
wchresta[m] has quit [Ping timeout: 246 seconds]
sargon[m] has quit [Ping timeout: 246 seconds]
ison[m] has quit [Ping timeout: 246 seconds]
efx[m] has quit [Ping timeout: 246 seconds]
corngood has quit [Ping timeout: 246 seconds]
atopuzov[m] has quit [Ping timeout: 246 seconds]
Nobabs27[m] has quit [Ping timeout: 246 seconds]
olejorgenb[m] has quit [Ping timeout: 246 seconds]
ssmike has joined #nixos
<MichaelRaskin>
Makes sense, actually: MBR had the gap more often than not, GPT doesn't, so just Give Me A Contiguous Block Of Space.
<samueldr>
nah, that's knowledge I often forget, but used a couple of times already
AntonLatukha[m] has joined #nixos
seif_madc[m] has joined #nixos
<MichaelRaskin>
Because non-trivial FS code in the first sector… that just doesn't fit.
madknight has joined #nixos
<samueldr>
and it's exactly the reason
efx[m] has joined #nixos
sargon[m] has joined #nixos
sdemos has joined #nixos
<MichaelRaskin>
I know
vidbina has quit [Ping timeout: 268 seconds]
leapingfrogs has quit [Client Quit]
grenade has joined #nixos
<MichaelRaskin>
It is even called BIOS Boot partition, if I remember correctly
icetan has joined #nixos
corngood has joined #nixos
andyjscott[m] has joined #nixos
peterhoeg has joined #nixos
randomenduser[m] has joined #nixos
jack[m] has joined #nixos
atopuzov[m] has joined #nixos
rihardsk[m] has joined #nixos
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
wchresta[m] has joined #nixos
r3mfKm[m] has joined #nixos
moredhel1 has joined #nixos
ison[m] has joined #nixos
Diagonal[m] has joined #nixos
<ennui>
sphalerite, samueldr: disk/by-id solved it. thanks a lot! i'll add a note to the wiki.
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
iohidtbhfam[m] has joined #nixos
wmertens[m] has joined #nixos
gridaphobe[m] has joined #nixos
<Dezgeg>
don't celebrate just yet, only after you discover that some bioses refuse to do legacy boot from GPT partition tables and that your machine is not one of those
<ennui>
i'm using MBR :)
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
octalsrc[m] has joined #nixos
mith[m] has joined #nixos
adisbladis[m] has joined #nixos
scott1 has joined #nixos
boegel has joined #nixos
offlinehacker[m] has joined #nixos
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
Nobabs27[m] has joined #nixos
Wysteriary[m] has joined #nixos
olejorgenb[m] has joined #nixos
TimePath has joined #nixos
<samueldr>
Dezgeg: while it wouldn't surprise me if it happens, you know, bioses being largely buggy messes, it always feels strange to me
benkolera has joined #nixos
sudoreboot[m] has joined #nixos
<Dezgeg>
yes
wak-work has joined #nixos
bhipple[m] has joined #nixos
voobscout[m] has joined #nixos
hamishmack has quit [Quit: hamishmack]
<samueldr>
don't they generally load the boot sector part, which is still available in GPT disks?
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
stites[m] has joined #nixos
<samueldr>
why would they bother knowing anything about the partition scheme?
mujx[m] has joined #nixos
<MichaelRaskin>
Are you assuming BIOS makes sense?
herzmeister[m] has joined #nixos
davidar has joined #nixos
bachp has joined #nixos
hl has joined #nixos
linuxdaemon has joined #nixos
<samueldr>
I'd like it if it did :)
linuxdaemon has quit [Max SendQ exceeded]
<MichaelRaskin>
What's next, hardware RAID with sensible data layout?
wervenyt[m] has joined #nixos
<samueldr>
(I'm just lucky enough that where I installed witgh GPT+BIOS it never failed)
<Dezgeg>
yup
equalunique[m] has joined #nixos
linuxdaemon has joined #nixos
linuxdaemon has quit [Max SendQ exceeded]
feepo has joined #nixos
<neonfuz>
sweet, just cloned 17.09 and did 'nix-build . -A pkgs.sauerbraten'
<neonfuz>
built fine
<MichaelRaskin>
I think I managed to keep using MBR until UEFI became stable enough, then switched.
mk-fg has quit [Ping timeout: 256 seconds]
ambro718 has quit [Ping timeout: 256 seconds]
hamishmack has joined #nixos
<samueldr>
"stable enough"
<samueldr>
:D
<sphalerite>
neonfuz: the "pkgs." bit is redundant FWIW :)
<samueldr>
the end-user software side sure is, but wow do hardware vendors don't care
<ixxie>
If I am packaging X, which in offical documentation seems to require Y and Z, but Z is also a dependency of Y, I should explicitly still mention Z as a dependency right?
<MichaelRaskin>
Well, using the early Asus FullHD laptops as a point of comparison…
<ixxie>
because at the moment my package builds because of this, but I didn't put Z because it built
<sphalerite>
ixxie: for python packages it doesn't really matter since dependencies are always propagated build inputs
mk-fg has joined #nixos
mk-fg has quit [Changing host]
mk-fg has joined #nixos
<MichaelRaskin>
(Asus actually ships something sensible, it is just that UEFI was in flux back then)
<sphalerite>
For other languages, yes you'll typically need to include Z too
<samueldr>
there are still recent reference-platform builds of UEFI firmwares that can't handle adding EFI programs manually in the firmware interface >:(
<MichaelRaskin>
Like, Lenovo
<ixxie>
sphalerite: it is possible one day Y will no longer depend on Z
<samueldr>
but yes, ~2014 asus laptop has one of the best uefi implementation I own
<MichaelRaskin>
I bought a slightly used Thinkpad because like nobody else ships actual batteries and not these toys, but how I would prefer Zenbook blown up to 15'' with all free space used for battery
<sphalerite>
ixxie: well, it is also possible that one day X will no longer depend on Z :)
leapingf1 has quit [Ping timeout: 276 seconds]
<ixxie>
sphalerite: sure xD
<MichaelRaskin>
Optimus in Thinkpads is just how not to do things ever. Boot setup UI, too.
<ennui>
ixxie, yeah, explicit is better than implicit. that's the zen of python, by the way.
ssmike has quit [Remote host closed the connection]
leapingf1 has joined #nixos
ssmike has joined #nixos
<MichaelRaskin>
It might be that Z _is_ propagated build input of Y anyway, even outside of Python
suvash_away has quit [Ping timeout: 240 seconds]
<ixxie>
so this package seems to have an optional 'dependency'
suvash_away has joined #nixos
<ixxie>
if I omit it and cancel tests, the package builds, but if I run tests it fails
<ixxie>
I guess I can add it to buildInputs and run tests?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] markuskowa opened pull request #34093: rdma-core: init at 16.1 (master...rdma-core-pr) https://git.io/vNgRy
<ennui>
ixxie, i've run into the same issue. buildPython* is really missing a 'testInputs' attribute. just add the test deps to the regular deps, maybe with an extra comment or via propagatedBuildInputs = [ ... ] ++ testInputs
ent has quit [Ping timeout: 264 seconds]
ent has joined #nixos
swayf has quit [Remote host closed the connection]