<theseriousadult>
Can mirror at same resolution, but if I try to do anything other than mirroring or if I try to make the two screens have different resolutions, external monitor becomes unresponsive and laptop screen gets that weird wraparound effect
Supersonic112 has quit [(Disconnected by services)]
Vorpal has quit [(Ping timeout: 250 seconds)]
Supersonic112_ has joined #nixos
Supersonic112_ is now known as Supersonic112
srdqty1 has joined #nixos
theseriousadult has quit [(Quit: Leaving)]
<tejing>
when I try to move from wine to wineStaging, for some reason, nixos-rebuild is trying to downgrade my version of samba to something old that won't install due to security issues. why would this even be happening? how can I stop it?
jb55 has quit [(Ping timeout: 268 seconds)]
srdqty2 has quit [(Ping timeout: 260 seconds)]
lambdamu has joined #nixos
dhess has quit [(Ping timeout: 258 seconds)]
lambdamu_ has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 2 new commits to master: https://git.io/vF48M
<NixOS_GitHub>
nixpkgs/master dd18e06 Peter Hoeg: sqlitebrowser: 3.10.0 -> 3.10.1 and make it work with Qt 5.9
<NixOS_GitHub>
nixpkgs/master 376516f Peter Hoeg: Merge pull request #31120 from peterhoeg/f/sqlite...
NixOS_GitHub has left #nixos []
Vorpal has joined #nixos
mbrgm has quit [(Ping timeout: 248 seconds)]
asuryawanshi has joined #nixos
mbrgm has joined #nixos
Arcaelyx has joined #nixos
asuryawanshi has quit [(Ping timeout: 268 seconds)]
endformationage has joined #nixos
Leira has joined #nixos
theseriousadult has joined #nixos
erictapen has quit [(Ping timeout: 240 seconds)]
Leira has quit [(Quit: Page closed)]
iqubic has joined #nixos
<iqubic>
I am super pissed right now.
<iqubic>
My new Dell Latitude 5480 refuses to boot from the nixos 17.09 graphical installer live usb.
<iqubic>
And that means I can't install nixos on my new laptop.
<iqubic>
Anyone have anything I can try to get this to work?
<iqubic>
I spent a large amount of money on this.
<theseriousadult>
Did you disable secureboot?
<theseriousadult>
iqubic
dieggsy has quit [(Ping timeout: 250 seconds)]
<iqubic>
Yeah, I think so. Let me just check one more time.
<iqubic>
Secureboot is disabled. I have no idea what's up with this.
<theseriousadult>
Huh, that's really weird. Maybe there was an error writing the installer to your usb?
<iqubic>
Yeah, it's just not booting into linux from my usb. The bios recognizes that the USB is a valid thing to boot from. It just boots to windows instead.
<theseriousadult>
wat
<iqubic>
Yeah. This really is weird.
<theseriousadult>
Ok, I actually have no idea what's up with that
<theseriousadult>
Worthwhile troubleshooting step would be to make another live usb with a different distro, try booting from that
<iqubic>
I'll keep trying things to see if I can't get this to work.
<iqubic>
I think Ubuntu or Fedora should work, since Microsoft let's those boot, even with secure boot enabled.
<maurer>
Have you ensured that the USB you created is using the same of UEFI/BIOS boot as your machine is set to use
<maurer>
*?
<maurer>
(BIOS boot may be referred to as "Legacy" boot in the bios menu)
<iqubic>
maurer: I have not. I have no idea how to check if they are the same.
<iqubic>
This Dell Latitude 5480 has like, it's own bios/uefi system. It's like a propritary thing that Dell made for this machine.
drakonis has quit [(Read error: Connection reset by peer)]
<iqubic>
It's weird as heck.
snikkers has quit [(Ping timeout: 268 seconds)]
kkini has quit [(Ping timeout: 268 seconds)]
<iqubic>
Well, I found my issue.
<theseriousadult>
The weird proprietary BIOS thing?
<iqubic>
"Note: If the Fastboot option is set to 'minimal' [default setting] the 'Enable USB Boot Support' setting is ignored, and the system will not boot from any Pre-boot USB devices."
<iqubic>
theseriousadult: That is part of the reason.
<iqubic>
Why even let me select the thing, if you know I can't boot from it in my current configuration?
<adisbladis>
iqubic: Huh. That is super strange.
<adisbladis>
You'd imagine that the option is greyed out rather than ignored
<laertus>
"the system will not boot from any Pre-boot USB devices." but will it boot from Post-boot USB devices?
<iqubic>
adisbladis: The other options I can't select are greyed out, but not this one.
<iqubic>
It's just customized to my particular machine.
<iqubic>
And in the top left corner it says: "Dell Latitude 5480" instead, making me absolutely sure that this is tailored to my machine.
<iqubic>
What I mean to say is that my BIOS has more menus for hardware on my machine than this example has.
fragamus has joined #nixos
alhariel has quit [(Remote host closed the connection)]
takle has quit [(Remote host closed the connection)]
takle has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
arianvp2 has quit [(Ping timeout: 260 seconds)]
RayNbow` has quit [(Ping timeout: 240 seconds)]
theseriousadult has quit [(Quit: Leaving)]
RayNbow`TU has joined #nixos
orivej has quit [(Ping timeout: 246 seconds)]
boomshroom has joined #nixos
<boomshroom>
Hello!
arianvp2 has joined #nixos
<boomshroom>
I'm looking at reinforcement learning which led me to the OpenAI Gym and accompanying library. The library looks like it's meant to be installed with pip. Is there a recommended way of handling this? (I'm not very familiar with python)
laertus has left #nixos ["WeeChat 1.9.1"]
ylwghst has joined #nixos
asuryawanshi has joined #nixos
Anton-Latukha has quit [(Ping timeout: 260 seconds)]
<universalpackage>
how would i make a .so file portable
iqubic has quit [(Ping timeout: 250 seconds)]
justanotheruser has quit [(Ping timeout: 268 seconds)]
justanotheruser has joined #nixos
Fare has quit [(Ping timeout: 260 seconds)]
<srk>
boomshroom: just look at the examples of python packaging, it's actually quite easy in most cases
<srk>
also tools like pip2nix can help with larger projects, but if your package doesn't have too much dependencies just do it by hand
dhess has joined #nixos
ylwghst has quit [(Quit: Lost terminal)]
leat has joined #nixos
timokau has joined #nixos
Fare has joined #nixos
takle has joined #nixos
<hyper_ch>
Mic92: how to use a more up-to-date zfs version than zfsUnstable? It seems with zfs receive an encrpyted dataset I hit a bug that causes panic, however there was a fix commited 20 days ago
Ivanych has joined #nixos
takle has quit [(Ping timeout: 250 seconds)]
pie_ has joined #nixos
ma27 has joined #nixos
ma27 has quit [(Ping timeout: 250 seconds)]
asuryawanshi has joined #nixos
dhess has quit [(Remote host closed the connection)]
<srhb>
hyper_ch: Override the zfs package
<srhb>
If it's packaged in a kernel agnostic way at least
<srk>
yeah, I know but when I've tried last time there it wouldn't compile with 4.14
<srk>
needs never zfs
<srk>
I guess
<srhb>
srk: Was that with unstable enabled?
<srk>
probably not
<srk>
gonna try that, thanks
<srhb>
boot.zfs.enableUnstable for reference.
<hyper_ch>
4.14 didn't work for me niether... but it was due to virtualbox
<srk>
btw, I now have hydra running and tried building nixos from source but it won't build libuv due to some HRtimer test failing
<gfixler>
just saw in a video someone went to edit a file and did $ vi release.nix, and nothing happened, then did vi actualfile and vi opened with the file
<hyper_ch>
wanted to test if wifi works finally since 4.14 should have drivers
<gfixler>
what was that $vi release.nix bit doing?
<srk>
how would you solve that? builder is a kvm vm
<hyper_ch>
so question is: can I get 4.14 with wifi drivers to run sooner than a compatible inter wifi is being sent to me (it's being delayed currently)
<gfixler>
srhb: wondering if he just hit ctrl+c at the end of the first one after realizing it was the wrong file
<srhb>
gfixler: Looks like it. :P
<gfixler>
srhb: good - having vi[m] not open a file was creeping me out
<srk>
:D
* hyper_ch
heard only people with strong coding-fu can handle vi/m
<srhb>
Nah, I suck at coding and do just fine with Vim. :P
<srhb>
Is there a good way of creating a custom error message when I transform a previous types.string to a types.submodule? I want to give a custom deprecation warning (instead of foo I now have foo.safe and foo.unsafe, and I want users of foo to be notified)
endformationage has quit [(Quit: WeeChat 1.9.1)]
<srhb>
I could do it by simply keeping foo around, but that's a bit ugly..
pxc has quit [(Ping timeout: 240 seconds)]
MercurialAlchemi has joined #nixos
<universalpackage>
is it possible for multiple dynamic linkers/loaders to co exist
<universalpackage>
with different configurations and cache's
<srhb>
universalpackage: What are you actually trying to do?
<universalpackage>
exactly that
<srhb>
The reason I'm asking is that that sounds like a means to an end, and the end could possibly have other solutions. :)
<universalpackage>
not if the end has its environment wiped then set up to its likings
<universalpackage>
effectively making LD_PRELOAD or LD_LIBRARY_PATH useless
<srhb>
Eek.
proteus-guy has quit [(Remote host closed the connection)]
<universalpackage>
cus it IS entirely possible
<universalpackage>
so
<universalpackage>
exactly that
<universalpackage>
is it possible for multiple dynamic linkers/loaders to co exist with different configurations and cache's *
<srhb>
I don't know exactly. One could imagine setting up some shell wrappers for reintroducing the two different envs.
<srhb>
They wouldn't be coexisting as much as being "easily loadable" I guess.
<universalpackage>
as obvioudly the application would need to be compiled to use the modified linker/loader
<universalpackage>
instead of the system one
<srhb>
Right, but that should be as simple as referring to the ld wrapper than wraps LD_PRELOAD1 or whatever
<srhb>
Instead of just ld
<universalpackage>
except it would be more difficult to impliment
<srhb>
How so?
<universalpackage>
for example, an executable could somewhere in its code clear the env then fork a process with a pre defined hardcoded script
<srhb>
OK, sure. You'd have to patch it then.
<universalpackage>
and the same could happen for that hardcoded script
<universalpackage>
it could even intentionally dissallow the use of LD_*
<universalpackage>
wich is why a ld linker/loader would be more apropriate
<srhb>
I mean, normally there isn't even a "system linker"
<srhb>
So this sort of thing needs to be handled whether or not the misbehaving application clears its env.
<universalpackage>
and it would be easier to manage in events that the libs folder is moved as u would only need to re point it then rebuild the cache
<universalpackage>
instead of trying to manually patch every executaable and .so
<srhb>
Hmm. I think this problem is too vague for me to give any helpful advice, sorry. :)
<universalpackage>
"So this sort of thing needs to be handled whether or not the misbehaving application clears its env" yes
<universalpackage>
although the main problem is getting two linkers/loaders to be independant and not rely on /etc/cache
<universalpackage>
or at least for the 2nd
justbeingglad has left #nixos []
asuryawanshi has joined #nixos
<universalpackage>
nix os does not have /etc right?
<srhb>
universalpackage: It has /etc, but not /etc/cache
<universalpackage>
ok
<srhb>
But usually, wrapping up things with LD_LIBRARY_PATH=${lib.makeLibraryPath [ libfoo libbar ...] } isn't a problem.
reinzelmann has joined #nixos
<universalpackage>
nix doesnt have /etc/ld.so.cache either
<srhb>
Nope.
<universalpackage>
does nix have a ld.so
<srhb>
universalpackage: Not a global one, certainly.
<universalpackage>
ok
<universalpackage>
ldconfig
<universalpackage>
ldconfig: Warning: ignoring configuration file that cannot be opened: /nix/store/xzx1bv1d7z4mgg6sg6ly0jx609qvka4x-glibc-2.25-49/etc/ld.so.conf: No such file or directory
<srhb>
Yes?
<universalpackage>
ok so ldconfig -p works
<universalpackage>
how is that implimented
<universalpackage>
eg
<universalpackage>
libBrokenLocale.so (libc6,x86-64, OS ABI: Linux 2.6.32) => /nix/store/xzx1bv1d7z4mgg6sg6ly0jx609qvka4x-glibc-2.25-49/lib/libBrokenLocale.so
<universalpackage>
/tmp/glibc-2.26/build/elf/ldconfig: Can't open cache file /etc/ld.so.cache
<universalpackage>
: No such file or directory
boomshroom has quit [(Ping timeout: 260 seconds)]
dhess has joined #nixos
periklis has joined #nixos
hotfuzz_ has joined #nixos
<universalpackage>
where is gcc located
jacob has joined #nixos
<srhb>
universalpackage: As with all other things in the store, it depends on its input arguments.
hotfuzz has quit [(Ping timeout: 240 seconds)]
<srhb>
universalpackage: You might have several of them around, all in the form /nix/store/hash-gcc-version
<srhb>
universalpackage: Where hash depends on the inputs.
<srhb>
In other words, there is no single system-wide gcc.
p4cman has joined #nixos
<universalpackage>
ok
<tilpner>
sphalerite - Tried again, xvfb even comes with an "xvfb-run" that executes your command with an X server. Needed -s "-screen 0 640x480x24" for color profiles
<srhb>
Is there a logical xor somewhere?
<universalpackage>
is there any way to see what -l libraries are available to link against
<sphalerite>
tilpner: and it worked??
<tilpner>
Yes! \o/
<sphalerite>
Yay!
<srhb>
universalpackage: None! That's the trick. You spin up environments for linking with only exactly the packages you want.
<srhb>
Alternatively the answer is "everything in nixpkgs"
<universalpackage>
ok
scribbler has quit [(Remote host closed the connection)]
<etu>
srhb: I just realized that I already had apache installed and didn't notice :D
<srhb>
Ah. :-P
<etu>
srhb: In general, on other distros. I avoid installing apache because of all the deps and it's huge. But on Nix it's less noticable :p
<srhb>
Indeed.
<universalpackage>
ohhh its cus glibc is in the ld.so.cache
kuznero has joined #nixos
scribbler has joined #nixos
<tilpner>
sphalerite - X wasn't enough though, it wants to connect to pulse too. There's an open PR that might fix this
<sphalerite>
tilpner: oh joy...
<sphalerite>
What is this awful installer for?
ertes has quit [(Ping timeout: 250 seconds)]
ylwghst has quit [(Remote host closed the connection)]
<sphalerite>
universalpackage: there shouldn't be an ld.so.cache on nixos
<tilpner>
It's Godot, a game engine with a graphical editor. There is a flag to generate API definitions as JSON, and I wanted to do that as part of my build process, in case they ever change
<sphalerite>
And... It needs... Pulseaudio and X for that...
<tilpner>
There's no need to open the editor to generate them, it's just a bug. Since the definitions are slow-changing, I'll generate them by hand now, and hope the PR is merged quickly
<universalpackage>
[17:50] <sphalerite> universalpackage: there shouldn't be an ld.so.cache on nixos why?
asuryawanshi has joined #nixos
ylwghst has joined #nixos
<srk>
$ cat
<srk>
Segmentation fault
<srk>
oops :D
<universalpackage>
rip
roosemberth has joined #nixos
<srhb>
srk: ... ouch
<tilpner>
How did you manage that...?
<srk>
dunno, some commit from master
<universalpackage>
how did u get ld to use a hardcoded cache path
<sphalerite>
universalpackage: because it would break the guarantees made by nix. We only ever want to load the library a given executable needs from one specific location
<sphalerite>
And that varies by executable/library
<roosemberth>
Hello, I'm trying to run pulsemixer (https://github.com/GeorgeFilipkin/pulsemixer), so I downloaded the file, but when I want to run it, it says it can't find libpulse.so.0. I tryed installing the package and I see the lib under /nix/store, but it still can't find it
<roosemberth>
I tryed running it on nix-shell, but same problem
<vaibhavsagar>
nix keeps libraries in unusual places and this doesn't play well with software that assumes a standard setup
<sphalerite>
the nix packaged version works for me though
<universalpackage>
libpthread is provided by /nix/store/xzx1bv1d7z4mgg6sg6ly0jx609qvka4x-glibc-2.25-49/etc/ld.so.cache: libpthread.so.0 (libc6,x86-64, OS ABI: Linux 2.6.32) => /nix/store/xzx1bv1d7z4mgg6sg6ly0jx609qvka4x-glibc-2.25-49/lib/libpthread.so.0
<srk>
cool, that's like two machines with ssd corruption in one week
<sphalerite>
vaibhavsagar: the nix expression contains a bit where the full path is substituted in
<universalpackage>
ok
<vaibhavsagar>
roosemberth: you can set LD_LIBRARY_PATH or use patchelf to point that package at the right place
<sphalerite>
srk: ouch
codedmart has quit [(Ping timeout: 255 seconds)]
<sphalerite>
roosemberth: you said the nix version didn't work for you either?
<tilpner>
vaibhavsagar - I don't think patchelf is any use here. Either patch the script, or LD_LIBRARY_PATH
<srk>
need to invest in better ssds I guess, or put them in a raid
<roosemberth>
patch python?
lukec has quit [(Ping timeout: 255 seconds)]
<srk>
they can't handle that much compilation :D
MinceR has quit [(Ping timeout: 240 seconds)]
<roosemberth>
sphalerite: I didn't know there was a nix version of pulsemixer x)
<vaibhavsagar>
tilpner: you're right, patchelf is unnecessary since this isn't a binary
<sphalerite>
roosemberth: nah, patching the script should be enough and that's what the expression in nixpkgs already does
<sphalerite>
yeah
<tilpner>
roosemberth - Yes, either load the CDLL from an absolute path, which would require changing the script, or set the appropriate LD_LIBRARY_PATH
<sphalerite>
it's, uh, recent though :D Date: Sat Oct 21 11:21:24 2017 +0200
ocharles has quit [(Ping timeout: 255 seconds)]
<roosemberth>
sphalerite: what's the name of the nixos package?
MinceR has joined #nixos
<vaibhavsagar>
pulsemixer
<sphalerite>
pulsemixer. You'll need to get it from the unstable channel or backport it though
<roosemberth>
ah, that's why x)
<srk>
path ‘/nix/store/8yd19hm3gq4vs4hwp6s2ydhzff5cnnv5-coreutils-8.28’ was modified! expected hash ‘82ae6b78aac243a142af92691190ca155316bef885b0aacf03bcdf7c15eae5f9’, got ‘3819a7f0b800760be9c548d7ffad715bf369fc9835c2126b0925f33d0cf76573’
<srk>
here's the culprit :D
<sphalerite>
or if it's more of a one-off thing you can nix-shell -I nixpkgs=https://nixos.org/channels/nixos-unstable/nixexprs.tar.xz -p pulsemixer
codedmart has joined #nixos
<roosemberth>
sphalerite: thanks, that's what I was about to google x)
jtojnar has quit [(Ping timeout: 264 seconds)]
ocharles has joined #nixos
<srk>
it's even worse if I count 4 nvme ssds in one of our servers that are pretty much useless due to corruption
<srhb>
sphalerite: Depending on python at runtime might be a bit much :-P
<sphalerite>
<insert equivalent perl here>
<srhb>
Heh
<sphalerite>
:p
<adisbladis>
sphalerite: python -m json.tool
<adisbladis>
Oh wait, I get it :P
<adisbladis>
It's not the same.
<sphalerite>
adisbladis: not quite, since we're escaping a string, not pretty-printing
<sphalerite>
yeah
<srhb>
tilpner: I mean... I'm causing a systemd prestart to run some substitutions on a locked down file. Essentially the target is a yaml file of key: secret, with one secret being multiline
<sphalerite>
srhb: ooooh how about `jq -R .`
<srhb>
sphalerite: That's probably a good idea.
<etu>
jq <3
<srhb>
jl <3
<vincent_vdk>
srhb: indeed i was also going to suggest jq
<srhb>
For the record, I'm attempting to shore up the leaky secrets in the gitlab service.
<srhb>
It's... Fugly.
__monty__ has joined #nixos
<srhb>
(If I didn't have to preserved compatible with store leaky paths, it would be slightly less fugly.)
<srhb>
preserve*
asuryawanshi has joined #nixos
<srhb>
compatibility*. Dear god. More coffee.
universalpackage has quit [(Quit: Konversation terminated!)]
civodul has joined #nixos
universalpackage has joined #nixos
<__monty__>
rycee: I'm trying to set up a binary cache cache but having some trouble figuring out the nginx config, could you share yours?
<sphalerite>
!m srhb
<[0__0]>
You're doing good work, srhb!
<srhb>
sphalerite: Thanks. :-P
asuryawanshi has quit [(Remote host closed the connection)]
<sphalerite>
__monty__: a binary cache is just a bunch of files that can be served statically by nginx, isn't it?
universalpackage has quit [(Client Quit)]
<__monty__>
sphalerite: Yes but I have almost no experience configuring nginx.
<srhb>
OK, here's what I'm thinking. I'll actually make the compatibility layer writeFile the "bad/old" secrets. Then the interface for substituting in the secrets at runtime is the same for both the unsafe and secure version (since every secret is a file, only some of them are unsafe nix stored files)
<__monty__>
You run your own? Isn't the real problem not getting identified as spam?
<sphalerite>
__monty__: haven't had any problems with that actually
thc202 has joined #nixos
Fare has quit [(Ping timeout: 268 seconds)]
<sphalerite>
srhb: idk its config feels like a nightmare maze to me
<srhb>
sphalerite: Well, okay, I'll give you that... :-P
ylwghst has quit [(Quit: Lost terminal)]
<sphalerite>
srhb: I've heard some other thing is infinitely worse (maybe it was exim?) but postfix is bad enough
asuryawanshi has quit [(Remote host closed the connection)]
<srhb>
exim is pretty terrifying, yes. I can't configure it without all the deb helpy bluescreens.
<adisbladis>
No one using qmail any more? ;)
<sphalerite>
I get the impression that postfix is too flexible, you have to have read the entire manual to not accidentally forward emails from anywhere without auth and become an inadvertent spam source
* srhb
eyes mailserver warily
<sphalerite>
adisbladis: is qmail less scary to configure?
<adisbladis>
sphalerite: No I'd say more
<sphalerite>
oh ok
<adisbladis>
My qmail server was a very special snowflake
<sphalerite>
qmail is djb's mailserver, right?
<adisbladis>
Yep
Mateon3 has joined #nixos
<edofic>
I cannot use nix-serve as a binary cache if I use "nix build" (1.12) to build my derivation. I get "substituter ... does not have a valid signature for path ...". Any tips? Cannot find anything in documentation on this topic
arianvp2 has quit [(Remote host closed the connection)]
Mateon1 has quit [(Ping timeout: 240 seconds)]
Mateon3 is now known as Mateon1
asuryawanshi has joined #nixos
pie_ has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
crapwifi has quit [(Read error: Connection reset by peer)]
crapwifi has joined #nixos
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
tv has joined #nixos
<LnL>
you might need to generate/configure a singing key
<Mic92>
hyper_ch: both zfs and spl needs to be updated
<sphalerite>
edofic: what LnL said, or for a (less secure!) option, you can do the nix build as root or add your user to trusted users. Wouldn't recommend that option though
<edofic>
thanks woffs, looks like exactly what I needed
freusque has joined #nixos
<goibhniu>
hiya sphalerite!
mbrgm has quit [(Read error: Connection reset by peer)]
nh2 has joined #nixos
<sphalerite>
Sorry to bother you personally about this — I tried ardour the other day and it refused to connect to my audio interface
takle has quit [(Ping timeout: 264 seconds)]
Truewar_ has joined #nixos
<sphalerite>
just saying "Failed to open audio device"
<sphalerite>
My first guess was that it wasn't getting along with pulseaudio, but stopping pulseaudio didn't help either
<Truewar_>
Hello, nixoids :)
asuryawa_ has joined #nixos
pie_ has quit [(Ping timeout: 250 seconds)]
asuryawanshi has quit [(Ping timeout: 240 seconds)]
asuryawa_ has quit [(Ping timeout: 260 seconds)]
<sphalerite>
srhb: oh yeah, for the thing you asked earlier you probably also need to give --slurp to jq
cfricke has joined #nixos
loz has quit [(Quit: Leaving.)]
<srhb>
sphalerite: Yeah, -Rs seems like it will work in every case :)
catch22 has joined #nixos
<crapwifi>
if i change this line '# define LD_SO_CACHE SYSCONFDIR "/ld.so.cache"' in ./sysdeps/generic/dl-cache.h what would change when recompiled
asuryawanshi has joined #nixos
<neonfuz>
What info is actually hashed for the package hash?
<neonfuz>
do you need to build a package to create it's hash?
<sphalerite>
neonfuz: the derivation file, I believe
<sphalerite>
No, you don't need to build it to get the hash
<sphalerite>
The build needs the output path as an input so that would be difficult ;)
<neonfuz>
lol true
<neonfuz>
so is it like, the actual text of the derivation file? or some sort of hash of the result of computing the derivation?
<sphalerite>
I might be wrong about the derivation file bit, I haven't actually looked at the code or the paper in question
<sphalerite>
But intuitively that would make sense to me
<sphalerite>
Oh actually the output name as well, because a single derivation can produce multiple outputs
periklis has quit [(Ping timeout: 250 seconds)]
<neonfuz>
hm, do you know where this would be documented?
matthewbauer has quit [(Ping timeout: 260 seconds)]
cfricke has quit [(Ping timeout: 268 seconds)]
nschoe has joined #nixos
<LnL>
you can get the hash from the drv
<LnL>
nix-store -q --binding out /nix/store/....drv
catch22 has quit [(Quit: Leaving)]
matthewbauer has joined #nixos
<fadenb>
Ffpletz, globin: I was just told by AWS that they completed the rollout for the IPv6 mtu fix on cloudfront. Do you still experience issues with cache.nixos.org in the office?
typeggzero has quit [(Quit: leaving)]
takle has joined #nixos
LangeOortjes has joined #nixos
<goibhniu>
Hi kgz, how was freenode #live? Was there much interest in NixOS?
mbrgm has joined #nixos
kiiiiii has quit [(Ping timeout: 250 seconds)]
asuryawa_ has joined #nixos
takle has quit [(Ping timeout: 250 seconds)]
betaboon has joined #nixos
asuryawanshi has quit [(Ping timeout: 260 seconds)]
asuryawa_ has quit [(Ping timeout: 240 seconds)]
asuryawanshi has joined #nixos
derchris has quit [(Ping timeout: 250 seconds)]
Itkovian has joined #nixos
bennofs has joined #nixos
VLetrmx has quit [(Quit: leaving)]
MP2E has quit [(Remote host closed the connection)]
<kgz>
goibhniu: it was great!
<kgz>
lots of people took stickers, and were interested in nixos
<kgz>
someone gave a talk about guix as well
<hyper_ch>
so, had a breakin at work this night.... server gone, various other electronic devices gone.... some money missing... they even took toilet paper and paper handkerchiefs.... they didn't take some.... they didn't take some other, expensive electronic things like yealink t48g phones but an ancient microwave.... they even took wireless keyboard and mouse... so annoying to type on the notebook
<goibhniu>
that's brilliant!
derchris has joined #nixos
<goibhniu>
(not the break-in, sorry to hear about that hyper_ch)
fuzzy_id has joined #nixos
<srhb>
Urghh. What's the right sed way to do this. I have JSON=$(jq -Rs . somefile) which produces a json string with escaped newlines. I then want to substitute it into another file, but sed -e "s,placeholder,$JSON,g" -i someotherfile seems to unescape those newlines.
<srhb>
Or am I misunderstanding shell magic again?
<fuzzy_id>
I'm debugging `services.cassandra` atm. Can someone explain me how I'm supposed to run the tests?
<fuzzy_id>
the module where the service is defined is not imported into system
<fuzzy_id>
i know how to do it, when i want to run cassandra in container
<betaboon>
is blender working correctly for anyone ?
<fuzzy_id>
but i don't know where to put the import statement in the test file…
<LnL>
import in the test?
<fuzzy_id>
srhb: 'error: The option `services.cassandra' defined in `<unknown-file>' does not exist.'
<srhb>
fuzzy_id: Yeah, so I see.
<fuzzy_id>
LnL: yeah, how do i do that?
<goibhniu>
betaboon: what do you mean by correctly?
<LnL>
fuzzy_id: you shouldn't have to, is the module missing in module-list.nix
matthewbauer has quit [(Ping timeout: 268 seconds)]
<fuzzy_id>
yeah
<betaboon>
goibhniu: i can build and run blender, but from time to time it just gets semi-stuck (sometimes recovers sometimes not). when blender is in that state, strace shows that all the binary is doing is repeatedly (relatively fast) ioctl(DRM_IOCTL_I915_GEM_SET_DOMAIN).
<betaboon>
goibhniu: and by correctly i mean, without that issue xD
<fuzzy_id>
should i simply add it there and make a pull-request?
<LnL>
fuzzy_id: add it there, all modules should be in there AFAIK
<crapwifi>
glibc comes with nix-os right?
<fuzzy_id>
ok
<goibhniu>
betaboon: ah ok, I haven't used it in about a month, and with nvidia. It was fine then FWIW
<crapwifi>
if so where could i view its compilation script
asuryawanshi has quit [(Read error: Connection reset by peer)]
asuryawanshi has joined #nixos
matthewbauer has joined #nixos
bennofs has quit [(Ping timeout: 260 seconds)]
<betaboon>
goibhniu: maybe i should mention that I'm on a 4K display with an intel-integrated graphicscard
<goibhniu>
yep, all that info would be useful, and the channel you're using ... did you check for similar issues on other distros?
<sphalerite>
hyper_ch: oh wiw
<sphalerite>
Wow*
bennofs has joined #nixos
<betaboon>
now that i mention it, i just noticed that /sys/class/graphics/fb0/device/uevent mentions "DRIVER="i915". which pretty much matches the ioctl call that i mentioned earlier. so that seems to be a driver-related issue ...
LexLeoGryfon has joined #nixos
<sphalerite>
betaboon: yeah I've had some issues with blender on Intel in the past too
<goibhniu>
yep, i915 is a good indicator
<betaboon>
goibhniu: i looked for issues regarding that ioctl-call. not quite sure how to search for "sometimes gets stuck sometimes doesnt" issues on the internet without going crazy reading through internet-hate xD
<neonfuz>
so would even something as simple as whitespace change in the nix file cause the hash to be different?
<sphalerite>
neonfuz: no
<sphalerite>
The DRV file it instantiates to is what matters
<neonfuz>
ah
<neonfuz>
I don't know much about drv files
<LnL>
if it's in the builder script then yes
<neonfuz>
that's the output of the nix expression, right?
<neonfuz>
and it has a canonical form?
<sphalerite>
The expression can be completely different, as long as it evaluates down to the derivation built-in being called on the same set
<MoreTea>
neonfuz: the hash is computed on the derivations you want to build. So you can change your entire implementation in nix, but if it results in the same calls to builtins.derivation, it won't change the hash
<neonfuz>
alright
<goibhniu>
betaboon: yeah, I'd try searching with your kernel version, and blender version
<goibhniu>
betaboon: it's highly unlikely to be NixOS specific ... perhaps
<goibhniu>
... fiddling with versions with help
jensens has joined #nixos
<crapwifi>
how do i re-compile an archive on nix
<crapwifi>
file*
<crapwifi>
like recompile bash but not install it
<betaboon>
goibhniu: just going through the google-results for "blender i915 drm" its a starting point :)
<betaboon>
crapwifi: nixos-rebuild build ?
<betaboon>
goibhniu sphalerite thanks for the thoughts. gotta solve that problem later on. work to do on another project first :)
<goibhniu>
betaboon: cool, good luck!
<sphalerite>
crapwifi: what are you trying to achieve?
<betaboon>
another question tho: in nixos-modules for bool-options that are default=true. do you prefer them to be named "enableFooOption" or "disableFooOption" ?
<crapwifi>
i want to rebuild a package that was built when nix os was installed
<betaboon>
sphalerite: i sometimes get confused what is turned on by default and which options i have to enable explicitly. just thought if disableOption would be more natural for the thought-process
<sphalerite>
crapwifi: yes, but why
<LnL>
betaboon: enableFoo or withFoo are most common I think
<crapwifi>
so i can attempt to modify it
<neonfuz>
crapwifi: wouldn't recompiling just give you exactly what you already have?
<neonfuz>
oh, so you need modifications
<sphalerite>
LnL: isn't that mostly for packages, not moduels?
<crapwifi>
yes but i do not know HOW it gets compiled
<sphalerite>
crapwifi: then you use an override
<neonfuz>
so you aren't asking how to recompile, you're asking how to change a package
<sphalerite>
crapwifi: the nix pills would be helpful for you to read
<sphalerite>
crapwifi: they go into detail about how all that works
<LnL>
ah for modules enable or enableFoo
<crapwifi>
also nixos-rebult build should work in a chroot right
<crapwifi>
rebuild build *
scribbler has quit [(Remote host closed the connection)]
<sphalerite>
betaboon: nah, disable is confusing
<hyper_ch>
who has experience with qemu? How would you start qemu vm without graphical interface?
<sphalerite>
betaboon: because of multiple negation
<sphalerite>
hyper_ch: qemu -nographic
<hyper_ch>
sphalerite: so simple?
<sphalerite>
Yep :)
crapwifi has quit [(Read error: Connection reset by peer)]
<hyper_ch>
sphalerite: well, this far I used proxmox
crapwifi has joined #nixos
<hyper_ch>
sphalerite: but I'd love to put nixos on the server now
<__monty__>
sphalerite: I could see a use for disable for options that are enabled by default.
<sphalerite>
There's a bunch of finer-grained options which -nographic is "composed" of but -nographic is a handy shortcut
<betaboon>
sphalerite: where does multiple negation come into play here ?
<crapwifi>
Error in tempdir() using /run/user/1000/nix-build.XXXXXX: Parent directory (/run/user/1000) does not exist at /nix/store/b8xy8bdxh1r5yjda332qrxj1vvvsy4fq-nix-1.11.15/lib/perl5/site_perl/5.24.2/x86_64-linux-thread-multi/Nix/Utils.pm line 45.
<sphalerite>
monty: betaboon: disable = false;
<hyper_ch>
I need to manage a vms for: vpn, samba, mail, voip....
<crapwifi>
warning: don't know how to get latest Nix
<crapwifi>
building the system configuration...
<hyper_ch>
and how would I assign them "static" lan ips.... that's the other thing I don't really know howto do
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<hyper_ch>
as said, with proxmox it was all quite easy
<betaboon>
sphalerite __monty__ : having a disableOption option makes the default value visible at the API, level. which can be seen in the code - without looking the default value up in documentation, and relying on the default-value staying the same over time.
<sphalerite>
betaboon: I've been confused by disable options (and got them wrong) quite a few times before
<betaboon>
sphalerite: yeah disable = false. but that 'problem' is only visible in the implementation of a service, not on the Interface
crapwifi has quit [(Read error: Connection reset by peer)]
universalpackage has joined #nixos
<sphalerite>
I prefer to look up some docs to work out what the default is than to get confused by multiple negation personally
<__monty__>
betaboon: If the default changes to off then there wouldn't be a disableOption any more. For non binary options neither come into play imo.
<betaboon>
sphalerite: then maybe we're just arguing about the several ways individual brains associate information. like eg visual-vs-logical
<betaboon>
XD
<sphalerite>
And nixos module docs are very easy to access and query
universalpackage is now known as crapwifi
c0bw3b has joined #nixos
scribbler has joined #nixos
<betaboon>
__monty__: exactly. having the disableOption requires an API-change when changing the default. which would break the code that relies on the default value. i find that more explicit that with enableOption. which for me is a good thing
<crapwifi>
can 'nixos-rebuld build' work in a chroot?
<betaboon>
sphalerite: you're right on the docs. nix realy excells on that front. but removing the need for a developer (in a team of developers) to look into the docs is something i found to be an advantage
<__monty__>
sphalerite: I don't like the "you need to read the docs to do simple things" mentality. It just makes things harder to use. The double negation wouldn't be an issue if nix was statically typed so now we know what everyone needs to do ; )
<crapwifi>
im gonna attempt to boot up my Nix OS
<LnL>
you can also see the default with nixos-option, etc
<betaboon>
anywho, don't get me wrong, i love to work with nix. it raised my life-quality by quite a bit xD
<betaboon>
LnL: ouh i didnt see nixos-option yet. thank you :)
ThatDocsLady has joined #nixos
ixxie has joined #nixos
<sphalerite>
betaboon: and for prior examples, boot.enableContainers, boot.hardwareScan, boot.grub.enable, boot.loader.systemd-boot.editor, fonts.fontconfig.antialias.enable, fonts.fontconfig.hinting.enable, hardware.Bluetooth.powerOnBoot, ...
<betaboon>
and i generally don't have a problem with reading the docs. docs are great. but as i mentioned a while ago I'm in the process of introducing a team to nix, nixos, nixops, hydra so my thoughts tend to go to topics that might raise cons from a closed-protective-developer ;)
crapwifi has quit [(Quit: Konversation terminated!)]
roberth has quit [(Ping timeout: 248 seconds)]
<sphalerite>
\_\_monty__: it's not "you need to read the docs to do simple things". It's "sensible defaults and easy-to-understand options"
ma27 has joined #nixos
<betaboon>
sphalerite: i think i will stick with enableOptions in my modules for now (for the sake of consistency ;) )
<sphalerite>
__monty__: if someone knows that they care about such an option they can enable it explicitly even if that matches the default — no harm done
Itkovian has joined #nixos
<sphalerite>
In fact I'd even say it's beneficial. It documents "we specifically want this option"
<LnL>
yeah, I probably have some stuff like that in my configuration
matthewbauer has quit [(Ping timeout: 240 seconds)]
kai_w has joined #nixos
<betaboon>
sphalerite: that last argument about documentation sold me ;)
fragamus has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<__monty__>
rycee: Thank you.
<srhb>
Can anyone jq/sed wizards help out? I want to read a multiline file in as a JSON string (which I assume means escaped newlines) and THEN substitute that exact literal string (escaped newlines and all) into another file verbatim.
takle has joined #nixos
__Sander__ has joined #nixos
takle has quit [(Remote host closed the connection)]
takle has joined #nixos
<__monty__>
sphalerite: Documenting this is what we want is the dual of documenting this is what we don't want. Having options that are enabled by default indicate this in their name is useful imo. You basically said, if you want to be sure, check the docs. And, disable options are confusing and make me check the docs. The difference being in the former case everyone has to check the docs while in the latter just
<__monty__>
the people who get confused about disable flags do.
<sphalerite>
srhb: I think stdenv's substitute functions can do that? You could either use those or look up how they're implemented
p4cman has quit [(Quit: Connection closed for inactivity)]
<LnL>
srhb: not sure if sed works with multiline strings like that
<srhb>
LnL: It feels like it doesn't, indeed.
<srhb>
Though I don't want to substitute in a multiline string
<universalpackage>
i gonna try using menuentry "NixOS - Default" --class chakra --class gnu-linux --class gnu --class os {
<LnL>
our substituteInPlace doesn't use sed, maybe that works
<sphalerite>
\_\_monty__: The problem with disable flags isn't that what they're for is hard to understand, it's the multiple layers of negation
universalpackage has quit [(Quit: Konversation terminated!)]
<srhb>
sphalerite: I can use those at runtime?
takle has quit [(Ping timeout: 240 seconds)]
<sphalerite>
srhb: don't think so, but you could copy them
<srhb>
Yeah, okay..
<__monty__>
sphalerite: And I disagree on that point. DisableOptionA = true/yes, = false/no.
<__monty__>
Completely intuitive to me.
<sphalerite>
monty: disabling is a layer of negation
sigmundv has joined #nixos
<LnL>
srhb: source $(nix-build '<nixpkgs>' -A stdenv)/setup
<LnL>
I think
<__monty__>
I understand that, I see no reason to have false mean "yes, disable this."
<srhb>
LnL: *boggle*
<LnL>
note that it probably sets -e in your shell
<srhb>
I think this is a little too scary to include in a preStart script.
<srhb>
You'd think substituting in strings would be one of the simplest things in UNIX, but, alas...
<__monty__>
sphalerite: It's also not really an issue, the options could be called defaultOption and default to true, while enableOption's would default to false. Now we're rid of the extra negation.
universalpackage has joined #nixos
<LnL>
oh in that case you can use . ${stdenv}/setup
<sphalerite>
__monty__: but that's based only on the use case where someone wants to disable the option
<sphalerite>
The point of having the option is that both cases are possible
pie_ has joined #nixos
<sphalerite>
And "do/don't disable the option"
<sphalerite>
Isn't as clear as "do/don't enable the opyion"
<__monty__>
sphalerite: No, it's based on the use case where anyone can see what options are enabled by default just by looking at their names.
python476 has joined #nixos
<universalpackage>
it didnt work ;-;
<__monty__>
sphalerite: There is no disable anymore. It's all "enable" semantically, just with different names that reflect the default enabledness.
johnw has quit [(Ping timeout: 246 seconds)]
<universalpackage>
how do i manually add nix os grub menu
<universalpackage>
to my grub menu
<sphalerite>
__monty__: well yes there is disable, because now you've named the options like that
<__monty__>
sphalerite: Have you actually read what I said? I said the name is a non-issue and doesn't have to be disable.
kiiiiii has joined #nixos
johnw has joined #nixos
<goibhniu>
__monty__: the default could change too, it sounds like it's overloading the parameter IMO. FWIW I'd say it would be a good policy to use enable* everywhere.
<sphalerite>
__monty__: oh right. But doesn't that then mean you need to know the default to set it at all?
<__monty__>
goibhniu: We're not talking about non-binary options here. And if the default changes then the name for the option changes from defaultOptionA to enableOptionA if OptionA is now disabled by default.
patrl has joined #nixos
<__monty__>
sphalerite: I don't see why? Deliberately setting it on/off would be exactly the same as for enableOption, you just gain the information that the default is on.
<sphalerite>
monty: which requires everyone to change their config.
<__monty__>
Yes, if defaults change everyone should know about it.
<sphalerite>
Not really, if they've made an explicit decision that doesn't care about the defauly
<__monty__>
Why would someone who doesn't care have the option set at all?
roberth has joined #nixos
asuryawanshi has quit [(Read error: Connection reset by peer)]
<LnL>
if changing the default has significant brhaviour changes stateVersion should probably be used to avoid breaking stuff
asuryawa_ has joined #nixos
<__monty__>
Oh no, I see what you mean, sphalerite.
johnw has quit [(Ping timeout: 258 seconds)]
pie_ has quit [(Ping timeout: 258 seconds)]
<universalpackage>
how do i manually add nix os to my grub menu
<sphalerite>
I think a better solution here would be a more advanced editing environment
<sphalerite>
universalpackage: you don't. Let nixos handle it for you.
<sphalerite>
universalpackage: it needs to create a menu entry for each system generation, so you can't really do it manually
srdqty1 has quit [(Ping timeout: 248 seconds)]
johnw has joined #nixos
<sphalerite>
universalpackage: The impression I have is that you really don't want nixos, you want Linux From Scratch where you cobble your whole system together yourself
srdqty1 has joined #nixos
cfricke has joined #nixos
damke has joined #nixos
<sphalerite>
__monty__: encoding the default in the name feels like the wrong solution for the problem. Like Hungarian notation. It's a hack around poor tooling.
<__monty__>
: o Polish notation is great, how dare you.
<universalpackage>
but i tried copying its grub config to mine but nix didnt appear in the menu ;-;
<__monty__>
sphalerite: Oh, hadn't heard of it. I agree thats terrible. And I understand the tradeoff better now. I'd rather read once when setting up than each time the defaults change. Still don't like that I have to read at all though : )
hellrazor has joined #nixos
<sphalerite>
__monty__: also, how do you find the option name in the first place? Maybe… the docs? ;0
<sphalerite>
;) *
<ma27>
is there anybody else having issues when trying to build tensorflow using `nix run nixpkgs.pythonPackages.tensorflow`?
<ma27>
I'm currently on 18.03pre119245.5cfd049a03 (Impala)
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
asuryawa_ has quit [(Ping timeout: 250 seconds)]
patrl has quit [(Client Quit)]
patrl has joined #nixos
patrl has quit [(Client Quit)]
patrl has joined #nixos
patrl has quit [(Client Quit)]
patrl has joined #nixos
<pie_>
im having a weird printing issue
<pie_>
im using a network printer and it works but its really slow because it takes the job, winds up, then prints a page, winds down for a moment, winds up, prints another page, winds down, repeat
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edolstra pushed 1 new commit to release-17.09: https://git.io/vF4j8
<NixOS_GitHub>
nixpkgs/release-17.09 d62f29a Eelco Dolstra: Set $NIX_DEBUG_INFO_DIRS when environment.enableDebugInfo is enabled...
NixOS_GitHub has left #nixos []
ylwghst has joined #nixos
ylwghst has quit [(Client Quit)]
patrl has quit [(Quit: WeeChat 1.9.1)]
ylwghst has joined #nixos
patrl has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vF4jh
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
<sphalerite>
Composability <3
<srhb>
sphalerite: Points for creativity. nc is such a handy tool...
<sphalerite>
:D
ylwghst has quit [(Remote host closed the connection)]
erictapen has quit [(Ping timeout: 250 seconds)]
<universalpackage>
does nix of have a version of itself wich is incompatible with ELF executable
patrl has quit [(Quit: WeeChat 1.9.1)]
<universalpackage>
like a version that specifically uses another executable format
<goibhniu>
hi universalpackage, what are you trying to do?
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
patrl has joined #nixos
<universalpackage>
the fact that a statically compiled executable might not run on every linux distro
ertes-w is now known as supercynic
erictapen has joined #nixos
patrl has quit [(Client Quit)]
<universalpackage>
i want to see if "the fact that a statically compiled executable might not run on every linux distro" is true *
patrl has joined #nixos
<tilpner>
How is a successful run defined?
<universalpackage>
idk
JosW has joined #nixos
<universalpackage>
by it executing without errors?
<tilpner>
Well, then it's easy to just go to any ancient distro and see that it doesn't execute well
<tilpner>
But that makes the whole question useless
<universalpackage>
true but i dont know of any that it wont run on
patrl has quit [(Quit: WeeChat 1.9.1)]
takle has quit [(Remote host closed the connection)]
<universalpackage>
assume both distro's are for the same archetecture
takle has joined #nixos
nh2 has quit [(Quit: Leaving.)]
kiiiiii has left #nixos []
Acou_Bass has quit [(Ping timeout: 248 seconds)]
nslqqq has quit [(Ping timeout: 240 seconds)]
<tilpner>
Google for any "kernel too old" error questions, compile their program statically, and run on an ancient kernel
fuzzy_id has joined #nixos
<__monty__>
Can you reasonably expect any binary to run on an ancient kernel?
<tilpner>
No
<etu>
It's easier to run things compiled on an old kernel on a newer one
Vorpal has quit [(Ping timeout: 250 seconds)]
<etu>
Linux is pretty good at keeping things compatible in that way afaik
periklis has joined #nixos
Acou_Bass has joined #nixos
<tilpner>
musl specifies 2.6.39 as minimum version, with low chance to work on earlier versions
nslqqq has joined #nixos
<srhb>
How's this? Backwards compatibility really needs some sort of warning for the user, but I'm loathe to add in all sorts of hoops to jump through when people could just add databasePassword = writeText "dbpassword" "totallynotsecret" https://github.com/srhb/nixpkgs/commit/c8c35f949c02de70fc8ab53d767fffd8ae55f9d8
<tilpner>
universalpackage - Compile a small binary that uses the statx syscall added in 4.11
patrl has joined #nixos
<tilpner>
universalpackage - It shouldn't be hard to find a system that's still not on 4.11, and it should fail on that
Isorkin_ has joined #nixos
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
Isorkin has quit [(Ping timeout: 240 seconds)]
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
asuryawa_ has joined #nixos
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
asuryawanshi has quit [(Ping timeout: 250 seconds)]
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
ixxie has joined #nixos
asuryawa_ has quit [(Ping timeout: 250 seconds)]
ixxie has quit [(Client Quit)]
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
proteus-guy has joined #nixos
Acou_Bass has quit [(Ping timeout: 240 seconds)]
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
Acou_Bass has joined #nixos
anderslundstedt has quit [(Quit: leaving)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 2 new commits to master: https://git.io/vFBkK
<NixOS_GitHub>
nixpkgs/master b9a4279 Tim Steinbach: Merge pull request #31315 from Lassulus/bitcoin-abc...
NixOS_GitHub has left #nixos []
cfricke has quit [(Quit: WeeChat 1.9.1)]
nslqqq has quit [(Client Quit)]
nslqqq has joined #nixos
damke_ has joined #nixos
Acou_Bass has quit [(Read error: Connection reset by peer)]
periklis has quit [(Ping timeout: 240 seconds)]
damke has quit [(Ping timeout: 240 seconds)]
Acou_Bass has joined #nixos
joehh has quit [(Ping timeout: 250 seconds)]
rsa has quit [(Ping timeout: 248 seconds)]
<fearlessKim[m]>
fetchgitLocal /home/toto can't "cd /home/toto" : "permission denied". The directory is readable by the world. How is one supposed to use fetchgitLocal then ?
<sphalerite>
srhb: I definitely like the concept, we need fewer passwords in the nix store :) I'll have a look at the implementation later but probably today
<sphalerite>
fearlessKim: is the x bit set on it as well?
<srhb>
sphalerite: I might just throw a wip pr up there for others to comment on too, think that's OK?
<srhb>
sphalerite: I updated the NixOS test to respect those changes as well, but of course, it still times out due to qemu weirdness...
<sphalerite>
srhb: I think it's a good idea, PR comments are a little easier to navigate than IRC logs ;)
<sphalerite>
regnat: I have question or two about your nix typing project, mind if I PM?
<fearlessKim[m]>
Linus: I try to package tegaki. The repo is in ~/tegaki . I have buildPythonApplication { ... src = fetchgitLocal /home/teto/tegaki;
<fearlessKim[m]>
...} then I run "nix-env -iA python3Packages.tegaki-python" x bit is on /home/teto/tegaki is set for everyone
bgamari has joined #nixos
<bgamari>
steveeJ, it's not clear
chreekat has quit [(Quit: quitting)]
<bgamari>
steveeJ, almost all of these issues are easy to fix
<bgamari>
what's hard is figuring out what needs to be fixed ;)
<sphalerite>
fearlessKim: hm, I don't get how fetchgitLocal actually works at all
<sphalerite>
fearlessKim: I think it just won't work at all with sandboxing enabled
<sphalerite>
It's a delightfully hacky implementation as well
<gerschtli>
Hey guys, I'm a little desperate because I need eclipse for java development with maven support, but I cant get it.. Is somebody, who can help me out? :)
<fearlessKim[m]>
I had to disable sandboxing because fetchgit wouldn't work on local filesystem so it's not enabled at the moment. fetchgitLocal looks pretty sexy for my usecase since tegaki is a bunch of different python modules but nixos only allows one to be indevelop mode
<tilpner>
gerschtli - What doesn't work, and how did you get there? (An error maybe?)
<gerschtli>
tilpner: I have eclipse-sdk running, everything is working fine, but I dont know, how I can enable maven support..
<gerschtli>
there is no marketplace in the toolbar, where I could have just downloaded it
<sphalerite>
fearlessKim: then I don't know why it wouldn't work. Both r and x are enabled for all users on /home/teto and /home/teto/tegaki?
bgamari has quit [(Ping timeout: 240 seconds)]
<tilpner>
gerschtli - "Help > Install New Software" does not exist?
<fearlessKim[m]>
arf not on /home/teto. I added x it fixes but then I have another permission problem. I can fix it though. I guess nixbuild group is not omnipotent.
<steveeJ>
this is a funny one "Compiler error reporting is too harsh for ./configure (perhaps remove -Werror)." :-D
<makefu>
what is the current way (after the package split into lib,out,etc) to `make menuconfig` a linux kernel? i get an error that it is unable to find `curses.h` . `export NIX_LDFLAGS="${NIX_LDFLAGS} -lncurses"` does not work
<steveeJ>
makefu: pkg-config should solve all that for you
<steveeJ>
makefu: how recent are the changes you mention? I ought to have an environment that allows "make menuconfig" for linux, I can test it in a bit if you need feedback
nslqqq has quit [(Ping timeout: 260 seconds)]
nslqqq has joined #nixos
<fearlessKim[m]>
Linus: thanks for the help. It should have been obvious. I had to chmod -R a+wx to tegaki folder so that fetchgitLocal can write the new git index into it. Still seems like the best solution.
<sphalerite>
makefu: I use nix-shell -E 'with import <nixpkgs> {}; linux.overrideAttrs (o: {nativeBuildInputs=o.nativeBuodlInputs ++ [ pkgconfig ncurses ];}'
Orbstheorem has quit [(Remote host closed the connection)]
<sphalerite>
Minus typos, sorry, on my phone
<sphalerite>
fearlessKim: uh won't that have marked every file in the repo as executable as well?
<fearlessKim[m]>
Linus: yep I went berserk mode x)
FRidh has joined #nixos
orivej has joined #nixos
<sphalerite>
fearlessKim: you may not want to commit those changes though ;)
<fearlessKim[m]>
true haha I reset teh change
<sphalerite>
fearlessKim: for future reference, +X is useful for that sort of thing
<fearlessKim[m]>
chmod a+rw ~/tegaki/.git seems to be enough
patrl has quit [(Ping timeout: 268 seconds)]
<sphalerite>
(Only adds the bit if it already exists for some other user or it's a directory)
<steveeJ>
bgamari: I managed to fix the ncurses issue, I'll post a PR when I'm done fixing all issues. currently I'm stuck with zlib which I *think* fails to find gcc
<sphalerite>
fearlessKim: you'll probably also want to undo making all the git objects writable... :)
reinzelmann has quit [(Quit: Leaving)]
mithsv has quit [(Quit: mithsv)]
mithsv has joined #nixos
lukec has quit [(Ping timeout: 252 seconds)]
<fearlessKim[m]>
Linus: didn't know +X nice. making git objects writable doesn't seem to be too much a problem, doesn't reflect in git status and I trust fetchgitLocal. your solution is better though. for next time ;)
jensens has quit [(Ping timeout: 264 seconds)]
lukec has joined #nixos
chreekat has joined #nixos
betaboon has quit [(Quit: WeeChat 1.9.1)]
<makefu>
sphalerite: awesome your nix-shell worked. i will add this knowledge to the wiki :)
patrl has joined #nixos
<sphalerite>
makefu: I usually have the contents of the -E as shell.nix in my kernel checkout :)
<sphalerite>
And thanks, docs are good
<sphalerite>
I think there's also a github issue about it (keywords kernel hacking), maybe link to it from there while you're at it
<sphalerite>
I'd do it myself, but phone...
<bgamari>
steveeJ, are you using my branch?
<bgamari>
steveeJ, if not I would strongly recommend it
<bgamari>
there is no reason to redo this work
<bgamari>
it literally has been weeks of effort off and on
nslqqq has quit [(Ping timeout: 268 seconds)]
<makefu>
sphalerite: the issue had been closed (without a valid solution however)
<makefu>
there is some documentation in the manuals, however it is for speeding up the write/compile/eval loop of modules
<sphalerite>
Right
<patrl>
hey ho, has anyone successfully gotten the weechat matrix script working?
<patrl>
I get an error about a missing Lua package cjson
<fearlessKim[m]>
patrl lua is not handled well in nixos, most likely you have to export LUA_CPATH yourself
<patrl>
I'll give that a try. I tried including luaPackages.cjson as an extraBuilInput in a weechat override; I used such a method to get wee-slack working, but not joy
<patrl>
no joy*
<fearlessKim[m]>
LUA_CPATH delimiters are of the form LUA_CPATH="?.so;/toto/?.so" (ie ; is the delimiter)
<patrl>
fearlessKim[m]: thank you, that probably would've tripped me up
bbarker has joined #nixos
jtojnar has joined #nixos
mithsv has quit [(Ping timeout: 250 seconds)]
<bbarker>
When I install gcc6, I see there is a gcc-6.4.0/lib directory, which I can a access in a shellHook like ${gcc6}/lib. However, I need to get to a gcc-6.4.0-lib/lib (iirc) directory - not sure how to reference that.
<NixOS_GitHub>
[nixpkgs] makefu opened pull request #31364: linux.kernel: enable build of stk1160 modules (master...pkgs/linux.kernel/stk1160) https://git.io/vFBBB
NixOS_GitHub has left #nixos []
erictapen has quit [(Ping timeout: 248 seconds)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
p4cman has joined #nixos
<hyper_ch>
hmmm, how can I install grub onto a usb thumbdrive.... it keeps complaining when I do that
takle has joined #nixos
erictapen has quit [(Ping timeout: 268 seconds)]
erictapen has joined #nixos
ma27 has joined #nixos
<srhb>
How does one use the --print-dead switch of nix-collect-garbage? I was assuming it would print rather than delete, but I just get "unrecognized option"
erictapen has quit [(Ping timeout: 248 seconds)]
<Dezgeg>
seems to be: nix-store --gc --print-dead
timokau has joined #nixos
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
<sphalerite>
yeah I don't like nix-collect-garbage
<oleks>
Hi, when trying to execute nix-env -iA nixos.jetbrains.jdk, I get the error: attribute ‘jdk’ in selection path ‘nixos.jetbrains.jdk’ not found, but I can see the package under https://nixos.org/nixos/packages.html
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<hyper_ch>
ha, figured it out :)
<hyper_ch>
ain't I awesomst :)
fuzzy_id has quit [(Ping timeout: 240 seconds)]
erictapen has quit [(Ping timeout: 268 seconds)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
LangeOortjes has quit [(Remote host closed the connection)]
erictapen has joined #nixos
<tilpner>
oleks - jetbrains.jdk isn't that old yet. What channel are you on?
bts- has joined #nixos
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<oleks>
apperently, I was at 17.03
<tilpner>
It's available in 17.09
ssmike has joined #nixos
roosemberth has quit [(Ping timeout: 268 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 3 new commits to master: https://git.io/vFB2V
<NixOS_GitHub>
nixpkgs/master c7e3555 James Earl Douglas: jenkins: 2.87 -> 2.88
<NixOS_GitHub>
nixpkgs/master 0e5014d James Earl Douglas: jenkins: add earldouglas as a maintainer
<NixOS_GitHub>
nixpkgs/master 5b84cdf Tim Steinbach: Merge pull request #31362 from earldouglas/jenkins-2.88...
<taaperotassu>
Have being using while the unstable version but I feel like that the differences are not that huge.. so stable sounds better for now
<stph>
before launching the first nixos-install on a fresh partition, i want to define on which partition / should be. But looking at https://nixos.org/nixos/options.html#filesystem leaves me puzzled, why isnt the filesystem (ext4 for example) not indicated for / ?
<MoreTea>
taaperotassu: you can have a stable base operating system, and then include stuff from unstable when you want some updated part.
roosemberth has joined #nixos
acowley_away is now known as acowley
<taaperotassu>
more
<taaperotassu>
MoreTea: How?
<taaperotassu>
It certainly sounds very nice to have it like that.. I suppose u can somehow define it in configuration.nix
<MoreTea>
I do this for projects mostly, and use pinning of nixpkgs
nschoe has quit [(Quit: Program. Terminated.)]
davidak has joined #nixos
asuryawa_ has joined #nixos
nh2 has joined #nixos
iyzsong has quit [(Ping timeout: 250 seconds)]
asuryawanshi has quit [(Ping timeout: 248 seconds)]
asuryawa_ has quit [(Ping timeout: 240 seconds)]
snikkers has quit [(Ping timeout: 246 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bjornfor opened pull request #31367: keepassxc: rename from keepassx-community (master...keepassxc-rename) https://git.io/vFBrD
NixOS_GitHub has left #nixos []
<jlle[m]>
hi, how can escape ${TMP} in a nix string, so it gets expanded by bash?
<vaibhavsagar>
hi, how do I get the nix .drv file corresponding to a result?
guillaum2 has joined #nixos
ssmike has quit [(Ping timeout: 264 seconds)]
asuryawanshi has joined #nixos
freusque has quit [(Quit: WeeChat 1.7.1)]
fuzzy-id has joined #nixos
<srhb>
jlle[m]: In a multiline string? ''${TMP}
<acowley>
jlle[m]: Put two single quotes before the dollar-sign. E.g. ''Hey ''${PEOPLE} how are you?''
tazjin has quit [()]
<srhb>
How does that happen several minutes after the fact...
<srhb>
vaibhavsagar: I don't think you do.
<jlle[m]>
acowley srhb thanks!
endformationage has joined #nixos
Sonarpulse has joined #nixos
dejanr has joined #nixos
ma27 has quit [(Ping timeout: 260 seconds)]
<acowley>
vaibhavsagar: Do you mean you just want to find the file? nix-repl is handy for that.
pie_ has joined #nixos
fendor has joined #nixos
fuzzy-id` has joined #nixos
fuzzy-id has quit [(Ping timeout: 250 seconds)]
ma27 has joined #nixos
mithsv has joined #nixos
<fuzzy-id`>
is there some function to populate a directory in the nix-store with several files?
<fuzzy-id`>
i'm writing a service module for cassandra. cassandra accepts a environment variable CASSANDRA_CONF which is supposed to be a directory holding _all_ the configuration files
<srhb>
fuzzy-id`: Why not just use a standard mkDerivation?
ThatDocsLady_ has joined #nixos
<fuzzy-id`>
so, how would i do that?
<fuzzy-id`>
simply write a script which generates the configuration files?
<srhb>
Essentially, yeah.
ThatDocsLady has quit [(Ping timeout: 264 seconds)]
periklis has quit [(Ping timeout: 250 seconds)]
<srhb>
fuzzy-id`: Though be careful about putting secrets in the Nix store (makes them world readable) :)
erictapen has quit [(Ping timeout: 250 seconds)]
ryantrinkle has joined #nixos
<ryantrinkle>
has anyone seen a situation where nix thinks all the build users are in use, even though nothing is actually building (confirmed by ps | grep)?
<fuzzy-id`>
hmm, looks ugly
<aminechikhaoui>
ikwildrpepper or domenkozar : anyone knows why we're using the weird names "/dev/disk/by-id/scsi-0Google_PersistentDisk_" for block device mappings attr names instead of using /dev/sd[a-z] like EC2 ?
<aminechikhaoui>
this is in Google cloud backend btw
zarybnicky has joined #nixos
damke_ has quit [(Ping timeout: 240 seconds)]
<aminechikhaoui>
(in nixops)
__monty__ has left #nixos []
bbarker has quit [(Ping timeout: 260 seconds)]
mithsv has quit [(Ping timeout: 260 seconds)]
<domenkozar>
aminechikhaoui: these are more deterministic
damke_ has joined #nixos
ma27 has quit [(Ping timeout: 250 seconds)]
roberth has quit [(Ping timeout: 250 seconds)]
lars__ has joined #nixos
<lars__>
hi, im trying to enable python and lua support in vim, and i think i must use vim_configurable for that
jb55 has joined #nixos
<lars__>
but how do i know what attributes there are to override?
damke_ has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 2 new commits to master: https://git.io/vFBXt
<NixOS_GitHub>
nixpkgs/master d0c8c66 Andreas Rammhold: cvs: fix CVE-2017-12836...
<NixOS_GitHub>
nixpkgs/master cbd458a Tim Steinbach: Merge pull request #31365 from andir/fix-cvs-CVE-2017-12836...
NixOS_GitHub has left #nixos []
zarybnicky has left #nixos []
patrl has joined #nixos
fendor has quit [(Ping timeout: 250 seconds)]
takle has quit [(Remote host closed the connection)]
<rnhmjoj[m]>
it appears to work but i have mostly no idea what i'm doing
fragamus_ has joined #nixos
<ldlework>
clever so I guess you're saying with regards to library code, they have those dependencies but that's sort of different than observing that all three tools utilize Nixlang as their implementation language. I mean even if you just look at Nix the package manager without the others, the name collision is still there.
<clever>
yeah
ma27 has quit [(Ping timeout: 268 seconds)]
fragamus has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
fragamus_ is now known as fragamus
<rnhmjoj[m]>
disasm: Lisanna can you see the snippet from IRC?
<Lisanna>
rnhmjoj[m]: you're redirecting the stdio *after* you execute the command that generates output?
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
patrl has joined #nixos
<rnhmjoj[m]>
Lisanna: before the actual script is executed
<Lisanna>
well, your'e running systemd-cat asynchronously, so that makes me ):
bts- has joined #nixos
<Lisanna>
but it's a pipe, so probably okay
timokau has joined #nixos
<Lisanna>
I don't understand io redirection very much, so if I were writing that I'd make extensive use of tee ^^
<sphalerite>
rnhmjoj: I think I know what I'm doing with this stuff, I'm just a bit busy now, I can look at it later though
<rnhmjoj[m]>
Linus: thank you
stph has quit [(Ping timeout: 260 seconds)]
<Lisanna>
anyways, per my question: I'm thinking the answer is no, which would make me very sad, but anyways: can a nix expression that is downloaded with fetchgit / fetchurl in a nix expression be imported from that same nix expression?
<ldlework>
clever I have no idea what Disnix is, but I just read the front page - is it basically container orchestration like kubernetes and so on, but with nix?
<Lisanna>
ldlework: I was just reading about disnix too ^^ yeah, it builds distributed systems on networks, kind of like kubernetes
<ldlework>
"distributed systems" is so vague :)
fuzzy-id` has quit [(Read error: Connection reset by peer)]
<clever>
ldlework: i havent looked into it either
lebel has joined #nixos
goibhniu has quit [(Ping timeout: 268 seconds)]
kai_w has quit [(Quit: WeeChat 1.9.1)]
<rnhmjoj[m]>
Lisanna: are you trying to create a paradox in nix with that expression?
<Lisanna>
No, I'm trying to distribute a build recipe with a source code, so that you always get the right build logic when you download the src
takle has joined #nixos
<Lisanna>
otherwise they have to be kept separate, and then there's nothing keeping them in-sync for major version changes
<Lisanna>
like if a new release needs another dependency
<Lisanna>
I haven't seen anything that solves this nicely in nixpkgs
<clever>
Lisanna: the only real solution is import from derivation, where you fetchFromGithub, then import a default.nix in that fetched result
<clever>
Lisanna: but it has the downside, that something as simple as nix-env -i hello, has to download your source always
fragamus_ has joined #nixos
<clever>
because it wants to check if yourpackage.name contains "hello", and it has to download the source to get the default.nix
bts- has quit [(Ping timeout: 260 seconds)]
<clever>
so it will never scale to all of nixpkgs
<Lisanna>
oh, so a binary cache is out of the question, since you don't know what you're building until you've downloaded the src?
<clever>
Lisanna: what i do, is i keep my own default.nix in the root of the project, with src = ./.; and then whenever nixpkgs has a version bump, i copy the default.nix from the project over, and change it to src = fetchFromGitHub
<disasm>
rnhmjoj[m]: maybe I'm missing something but that essentially looks to me like both if branches will be skipped all the time, which means their worthless code, right?
<Lisanna>
so the only thing you might get to skip is the build phase?
Itkovian has joined #nixos
<clever>
the binary cache would be able to give you a copy of the source, which you then eval, and find that the binary is also on the cache
<noobly>
what's the proper way to "set" things in nixos config rather than call it as a function?
<clever>
but you would have to download the source and binary, every time
<clever>
noobly: can you gist an example of what your doing now?
fragamus has quit [(Ping timeout: 240 seconds)]
fragamus_ is now known as fragamus
<rnhmjoj[m]>
disasm: of course, don't mind the conditions. those will be selected at "compile" time from nix
ma27 has joined #nixos
davidak has quit [(Quit: Leaving.)]
<Lisanna>
okay, but you *can* in fact import nix expressions from derivations
<disasm>
rnhmjoj[m]: guess I'm not quite understanding what the problem is
bts- has joined #nixos
<Lisanna>
so, if the nix expression was hosted separately from the source ,but with a mechanism to tie the version number / git rev / wahtever together, that could work?
<noobly>
clever: It's pretty short so I'll just type it: { nixpkgs.config.allowBroken = true; } was recently put at the top of my config, but when I tried to rebuild switch it failed, since it was apparently trying to call a set as a function
<clever>
noobly: you need to insert that key=value into your existing set, not put 2 sets in the file
erictapen has joined #nixos
<rnhmjoj[m]>
disasm: i was asking if there is a cleaner way to write this
<rnhmjoj[m]>
and if it's actually doing what i think it does. because i don't really understand i/o redirection
ixxie has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] disassembler pushed 3 new commits to master: https://git.io/vFBdV
<NixOS_GitHub>
nixpkgs/master b2d6c63 Raymond Gauthier: geckodriver: init at 0.19.1
<NixOS_GitHub>
nixpkgs/master f07ae89 Raymond Gauthier: pythonV.V-selenium: from 2.52.0 to 3.6.0
<NixOS_GitHub>
nixpkgs/master 3641da1 Samuel Leathers: Merge pull request #31114 from jraygauthier/jrg/python_selenium_update...
ghostyy has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
sigmundv has quit [(Ping timeout: 268 seconds)]
Itkovian has joined #nixos
ghostyy has joined #nixos
bts- has quit [(Ping timeout: 258 seconds)]
bts- has joined #nixos
scribbler has quit [(Remote host closed the connection)]
cfricke has joined #nixos
vcunat has joined #nixos
takle has quit [()]
<codedmart>
I don't understand how to use plv8 with postgres. I am using archlinux now with nix-shell. I have postgresql and plv8. But when I try to `create extension plv8;` in psql. I get `ERROR: could not open extension control file "/nix/store/2665zz2rnf1vqglm3vabvxn9xw72a01p-postgresql-9.6.5/share/extension/plv8.control": No such file or directory`.
<manveru>
codedmart: you need to put plv8 into the postgresql expression
<codedmart>
manveru: Is there some docs about this somewhere?
<codedmart>
How do I do that. I am using the latest nix-channel unstable, nix-shell, with a basic default.nix.
<Lisanna>
well, the builder lives closer to the source code,
<Lisanna>
idea is that src code author maintains both
scribbler has joined #nixos
fragamus_ has joined #nixos
scribbler has quit [(Remote host closed the connection)]
scribbler has joined #nixos
fragamus_ has quit [(Client Quit)]
bts- has joined #nixos
Neo-- has joined #nixos
hotfuzz has quit [(Ping timeout: 248 seconds)]
bts- has quit [(Ping timeout: 260 seconds)]
bts- has joined #nixos
hotfuzz has joined #nixos
noobly has joined #nixos
<noobly>
does nixos store infinite .bash_history somewhere?
noobly has quit [(Client Quit)]
Neo-- has quit [(Remote host closed the connection)]
Fare has joined #nixos
georgiy has joined #nixos
pie_ has quit [(Ping timeout: 240 seconds)]
<c0bw3b>
noobly it's in your $HOME/.bash_history but i don't thunk it's infinite :p
<c0bw3b>
ah.. he left already :p
chreekat has quit [(Ping timeout: 240 seconds)]
<georgiy>
Hey folks. I'm trying to get use Python with Protobuf on NixOS. While everything works fine on Fedora when I try to import the python file output by protoc I get an error saying that the google module wasn't found. I'm probably doing something wrong but an not sure what. Please help.
pie_ has joined #nixos
<fakenullie>
georgiy: what's the exact error
<georgiy>
>>> import MarketDataMsg_pb2
<georgiy>
Traceback (most recent call last):
<georgiy>
File "<stdin>", line 1, in <module>
<georgiy>
File "/home/fricfrac/src/fricfrac/python/MarketDataMsg_pb2.py", line 6, in <module>
<georgiy>
from google.protobuf import descriptor as _descriptor
<georgiy>
ModuleNotFoundError: No module named 'google'
<georgiy>
let me know if this rendered correctly
<georgiy>
i shpould probably also switch over to pastebin for this stuff
ryanartecona has quit [(Quit: ryanartecona)]
<fakenullie>
you need to install google protobuf python library
<georgiy>
i think i did
<georgiy>
how do i chech for sure
<georgiy>
?
<georgiy>
check*
<fakenullie>
how did you install it?
patrl has quit [(Quit: WeeChat 1.9.1)]
<georgiy>
via nix-env i think
<fakenullie>
well, then we have to wait for nixos guys
<georgiy>
btw this was from a semi-recent checkout from unstable
<timokau>
Does anybody have experience with nixos tests and qemu? I'm trying to simulate a monitor hotplug.
<timokau>
And you can appaerntly do it with spice, but I didn't get it to work either and all the documentation I found about that used GUIs
patrl has joined #nixos
bts- has quit [(Ping timeout: 260 seconds)]
noobly has joined #nixos
<noobly>
does nixos have infinite bash history?
<noobly>
and if so where might I find it?
<noobly>
and sorry for just asking this and leaving earlier, my internet is absolute garbage
romildo has quit [(Quit: Leaving)]
<c0bw3b>
noobly your session history is in the usual $HOME/.bash_history but i dont think it is infinite
nh2 has joined #nixos
<noobly>
c0bw3b
<noobly>
ok thx, I found that but was hoping I could find a longer history
JosW has quit [(Quit: Konversation terminated!)]
dhess has joined #nixos
pie_ has quit [(Ping timeout: 240 seconds)]
acarrico has quit [(Ping timeout: 255 seconds)]
<cransom>
you can set it to retain longer in your bashrc, but it wouldn't have anything beyond that saved anywhere
<copumpkin>
anyone remember the final outcome of the "how best to pin nixpkgs" discussion at NixCon? someone mentioned Tekmo had a good solution but I don't recall what it was
<sphalerite>
rnhmjoj: and remove/comment/conditionalise lines as appropriate
<sphalerite>
That starts one more tee process tthan strictly necessary. But I think it's neat because it expresses the "split the stream into this as well" concept neatly
rsa_ has joined #nixos
rsa has quit [(Ping timeout: 248 seconds)]
<bgamari>
how does one go about building a set of derivations with debug symbols?
noobly has quit [(Quit: Page closed)]
orivej_ has joined #nixos
orivej has quit [(Ping timeout: 250 seconds)]
<Fare>
copumpkin, what is wrong with using a git checkout in your NIX_PATH ?
asuryawa_ has joined #nixos
<copumpkin>
something has to run the git checkout for you, so I can't just evaluate nix in isolation
<copumpkin>
1.12 has some useful builtins for this but I can't assume 1.12 yet
<copumpkin>
even builtins.fetchGit
<copumpkin>
or fetchTarball
asuryawanshi has quit [(Ping timeout: 240 seconds)]
tanonym has joined #nixos
asuryawa_ has quit [(Ping timeout: 240 seconds)]
ixxie has quit [(Quit: Lost terminal)]
<LnL>
somebody made some stuff around <nix/fetchurl.nix>
dhess has quit [(Remote host closed the connection)]
<kyren>
I would like to update netatalk to a newer version, and I'm pretty sure I can figure it out by patching nixpkgs, and I want to do this anyway and make a PR, but in the meantime I need to run the service, so I'd like to patch it for myself
<kyren>
I know I can check out nixpkgs and rebuild using that source, but that seems heavy handed, I want to apply an.. overlay? to normal nixos haha
<c0bw3b>
Idlework not nixlang specific but reproducibilty is a bigthing for nix pkg and nixos I believe? next to atomicity
scribbler has quit [(Remote host closed the connection)]
hellrazor has quit [(Ping timeout: 268 seconds)]
scribbler has joined #nixos
hellrazor has joined #nixos
<tanonym>
If I'm using sources from git in an expression, can I point the expression at the /latest url where the project puts the latest builds so that a nix-env -i <package> would automatically build from the latest rather than having to rewrite an expression each time?
asuryawanshi has joined #nixos
<sphalerite>
pierron: one thing I don't understand about the overlay stuff: callPackage is a function so it should be taken from super, not self; but doesn't that mean all the dependencies will come from super as well…?
<manveru>
tanonym: yeah... sha is optional for fetchTarball, it just makes your expression impossible to include in pure builds and will cause a lot of downloads
<tanonym>
It seems I need to update the efl on my system to see if it fixes the enlightenment file manager from segfaulting the DE on me all the time.
<tanonym>
I'm still having difficulty wrapping my head around the documentation >,<
<manveru>
tanonym: the question is... why do you want to update so often?
<c0bw3b>
kyren : for the package alone, just DL the latest default.nix defining netatalk and edit it on your machine
<c0bw3b>
kyren: after edition you can check it's building and working with : nix-build --expr 'with import <nixpkgs> {}; callPackage ./default.nix {}'
<tanonym>
Hypothetically, bugtesting for a project. Or updating to a later value to test if a serious error has been corrected.
<c0bw3b>
kyren : you will a symlink named 'result' in your working directory and pointing to the resulting package in /nix/store
bts- has joined #nixos
<c0bw3b>
you will *get
Ivanych has quit [(Ping timeout: 250 seconds)]
<kyren>
c0bw3b: that makes sense, but what I want to do is try it out at a system level, it has a nixos service.. attached, sorry I'm probably using the wrong words here
<c0bw3b>
no you're not I understand :)
<kyren>
so that part I do understand, but that doesn't make the netatalk shares in configuration.nix use the new version
<c0bw3b>
but testing the package update alone first seems a good first step
<CrazedProgrammer>
spheralite: yes, super is the result of the previous "stages" of overlays and self is a fixed point of that and the new overlay. here the author of overlays explains it a bit better: https://youtu.be/W85mF1zWA2o
<c0bw3b>
as for testing the service module that goes with it.. I couldn't tell you myself the best way to do that actually
<manveru>
tanonym: well, it depends on your usecase, but you could make it optional and have a stable version for normal use
<tanonym>
That would be a good approach.
<kyren>
I know that I can fork nixpkgs and use nixos-rebuild -I <my_nixpkgs> right
<kyren>
that just seems very extreme
<tanonym>
Right now I need to get EFL updated to 1.20.5 from 1.20.3 to see if I can use the Enlightenment File Manager without it causing my Enlightenment Desktop Manager to segfault and crash and require either a recovery attempt or a logout.
<kyren>
I guess it does make sense though, especially if you want to edit nixos services it seems like forking and then merging changes is not a *bad* way to do it, I just wanted to make sure there wasn't a simpler way
<c0bw3b>
there must be, by overriding the service and tell it to use a local pkg instead of the nixpkgs one
jacob has quit [(Quit: Lost terminal)]
<c0bw3b>
but I don't know the actual way to achieve that sry :/
<kyren>
yeah, that's what I would like to do definitely
<kyren>
are overlays.. that thing?
mithsv has quit [(Quit: mithsv)]
<CrazedProgrammer>
sphalerite: sorry for misspelling your name. i made a reply a few messages above
mithsv has joined #nixos
<sphalerite>
ooh, it went "bink"! I'll ahve a look :)
davidak has joined #nixos
Vorpal has joined #nixos
Vorpal is now known as Guest29637
jb55 has quit [(Ping timeout: 246 seconds)]
<sphalerite>
CrazedProgrammer: yeah I know, my question was actually in response to that talk
<sphalerite>
CrazedProgrammer: since "callPackage" is coming from super, won't it pass all the non-overridden dependencies to the packages defined in the overlay?
<sphalerite>
(and that's why I highlighted pierron, because that's Nicolas Pierron 😉 )
<tanonym>
For now I just manage my files through cli when required. But having the filemanager crash my environment is grating my nerves. ^,^
georgiy has joined #nixos
dieggsy has joined #nixos
<sphalerite>
kyren: yes, overlays are what you'd use for that. But if it's a change that would make sense to upstream, like a version bump, doing it on a nixpkgs checkout is the better way since from there you can easily commit and make a PR :)
<kyren>
oh, definitely, and I will do that
<kyren>
but I kind of want to know how to do it with overlays as well
<kyren>
I felt like I should have been able to figure it out from the documentation, but I might be too noobish still to get traction with it, I suspect that I can do something like download the default.nix, modify it, stick that file in /etc/nixos/, then include it in an overlay in configuration.nix right?
<sphalerite>
kyren: the overlay itself would look something like `self: super: {netatalk = super.netatalk.overrideAttrs {name = "netatalk-x.y.z"; src = super.fetchurl {url = …; sha256 = …;};}; }`
<kyren>
it's the last part that I don't super get
<kyren>
ohhh, I don't need to actually download and modify the whole file.. I mean I suppose that makes sense
<sphalerite>
Then you'd set `nixpkgs.overlays = [ ./netatalk-update-overlay.nix ];`
Guest21796 has quit [(Ping timeout: 258 seconds)]
<sphalerite>
You could also copy the whole default.nix. It's a more verbose option, but also more robust, as it doesn't rely on netatalk's build process staying the same
<kyren>
oh, okay this is using overrideAttrs, I see it now
<kyren>
is that a function that gets added as part of mkDerivation?
<sphalerite>
oh whoops I actually got the overrideAttrs call wrong
<sphalerite>
yes, it is.
<sphalerite>
But rather than `overrideAttrs {…}` the example should be `overrideAttrs (orig: {…})`
<sphalerite>
one day we will have SOS and no longer need override and overrideAttrs. That will be a good day.
<sphalerite>
It's probably still quite a way off though :(
<kyren>
okay, I think it was not so much overlays that I needed to understand as the existing override mechanisms
<kyren>
and that it was automatically available from mkDerivation
patrl has quit [(Ping timeout: 252 seconds)]
<sphalerite>
yeah. There's overrideAttrs, which changes what's passed to mkDerivation and is provided by mkDerivation; and there's override, which is provided by callPackage and changes what's passed to the pakcage function
<sphalerite>
overlays are simple really, it's the overriding bit that's a bit more confusing :)
<kyren>
okay, sorry again for needing handholding, the overrideAttrs approach definitely works and I got that working and understand it, the problem is is that there's an accompanying patch file that needs updating, I can update that patch file no problem, but I wanted to copy the default.nix to netatalk.nix and update it that way, and I'm unsure of how to load that in configuration.nix, these are possibly basic nix questions though
hiratara has quit [(Ping timeout: 250 seconds)]
<georgiy>
thanks for the link
<MoreTea>
kyren: no worries.
<MoreTea>
you can definitely just copy the .nix of the package you want to modify
<clever>
kyren: run pkgs.callPackage on netatalk
<clever>
kyren: for example: pkgs.callPackage ./netatalk.nix {};
<MoreTea>
then load it from configuration.nix via environment.systemPackages = [ other_packages (pkgs.callPackage /path/to/nettalk {}) ];
<sphalerite>
kyren: it's fine, really, especially since you're asking sensible questions ;)
<kyren>
yesssss, pkgs.callPackage is what I needed to understand
fragamus has joined #nixos
<MoreTea>
it does some magic to inspect the function arguments in your nettalk.nix, and injects stuff from pkgs.$ARG_NAME into there.
hiratara has joined #nixos
<sphalerite>
oh god I just realised it's kind of like spring
jb55 has quit [(Ping timeout: 258 seconds)]
<gchristensen>
sphalerite: take that back!
pxc has joined #nixos
<sphalerite>
Well, detecting what it needs and fitting hte pieces together
<sphalerite>
that's sort of the core of what spring does
<sphalerite>
of course it's a lot more horribly stateful, and does tonnes of other (very magical) shite on top of that
darlan has joined #nixos
<c0bw3b>
Spring that old java thingy?
darlan has quit [(Client Quit)]
<rnhmjoj[m]>
Linus: thank you! it's perfect
<sphalerite>
c0bw3b: yes. I first used it this summer. Hopefully never again :)
<sphalerite>
rnhmjoj: glad to hear it :)
dywedir has joined #nixos
jensens has joined #nixos
ma27 has quit [(Ping timeout: 268 seconds)]
<c0bw3b>
rnhmjoj: you tested PR 31252? compton module extraOptions switching to types.lines
<c0bw3b>
it's ready2merge right?
dywedir has quit [(Ping timeout: 240 seconds)]
mithsv has quit [(Quit: mithsv)]
mithsv has joined #nixos
radvendii has quit [(Ping timeout: 240 seconds)]
radvendii has joined #nixos
<rnhmjoj[m]>
c0bw3b: yeah. the only difference is that multiples definition of extraOptions will be merged. I checked again just to be sure but it's ok.
p4cman has quit [(Quit: Connection closed for inactivity)]
<kyren>
I'll submit a PR as well, since the old version of netatalk is super broken with time machine and the new one works just fine
bts- has quit [(Ping timeout: 250 seconds)]
simukis has quit [(Ping timeout: 260 seconds)]
georgiy has quit [(Ping timeout: 250 seconds)]
mithsv has quit [(Quit: mithsv)]
mithsv has joined #nixos
mithsv has quit [(Client Quit)]
mithsv has joined #nixos
jb55 has joined #nixos
georgiy has joined #nixos
tanonym has joined #nixos
<tanonym>
Ooh, NixOS gives me a very stripped down Plasma 5. All snappy and pretty, me likey. ^,^
arianvp2 has quit [(Read error: Connection reset by peer)]
<tanonym>
And I have a spiffy connman system tray icon as well. :D
arianvp2 has joined #nixos
arianvp2 has quit [(Remote host closed the connection)]
chessai has quit [(Ping timeout: 258 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Acou_Bass has joined #nixos
cement has quit [(Ping timeout: 268 seconds)]
nh2 has quit [(Ping timeout: 248 seconds)]
jb55 has quit [(Ping timeout: 240 seconds)]
dieggsy has quit [(Remote host closed the connection)]
dieggsy has joined #nixos
<sphalerite>
Eww connman
<sphalerite>
:p
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
bts- has joined #nixos
<tanonym>
Well, it likes my computer the best. I don't have my wifi connections drop randomly with connman.
<tanonym>
For me it has been behaving itself better than anything else.
<tanonym>
So I use it. Because that's the beauty of running a nixos rig. I can run whatever the hell I want to run to get the job done :D
sigmundv has joined #nixos
<cransom>
i must be a phenomenon that i've only ever just used plain old wpa_supplicant and i haven't had any issues
bts- has quit [(Ping timeout: 240 seconds)]
<tanonym>
wpa works too, sort of. I've had to fight with it more often than connman.
ertes has joined #nixos
mithsv has quit [(Quit: mithsv)]
asuryawanshi has joined #nixos
<tanonym>
I tried, but I kept having to redo my wpa config files fairly regularly. Was like dealing with HL2 era beta-Steam... or modern Steam for that matter.
<tanonym>
Maybe it's my personality. Connman likes me better than wpa_supplicant and it's cli or networkmanager...
mithsv has quit [(Client Quit)]
mithsv has joined #nixos
dieggsy has quit [(Remote host closed the connection)]
MrCoffee has joined #nixos
asuryawanshi has quit [(Ping timeout: 268 seconds)]
dieggsy has joined #nixos
Wizek_ has quit [(Read error: Connection reset by peer)]
MP2E has quit [(Read error: Connection reset by peer)]
<kyren>
I have a version bump PR for netatalk but I'm *afraid*
Sonarpulse has quit [(Ping timeout: 248 seconds)]
Wizek_ has joined #nixos
Piece_Maker has joined #nixos
edef has quit [(Ping timeout: 255 seconds)]
<kyren>
I've tested it and it seems to work, but I haven't tested it with sandboxing, and I haven't run 'nox' on it and I'm not super sure how to even do that
Acou_Bass has quit [(Ping timeout: 268 seconds)]
Piece_Maker is now known as Acou_Bass
edef has joined #nixos
dieggsy has quit [(Ping timeout: 250 seconds)]
<ee1943>
has anyone else noticed their laptops being unable to suspend when they set up qemu/kvm virtualization?
<ee1943>
i.e. close lid, but keyboard LEDs remain on and fan keeps spinning
jgertm has quit [(Ping timeout: 240 seconds)]
<rnhmjoj[m]>
does `systemd-inhibit --list` show anything?
jgertm has joined #nixos
georgiy has joined #nixos
bts- has joined #nixos
<ee1943>
just network manager
<ee1943>
it's weird, everything worked fine prior to getting the virt setup going. has happened on other distros as well
<c0bw3b>
meeeh fetchgit or fetchfromgithub get a zip archive on some repo (not a tgz) and returns an empty src dir ..?!
davidak has quit [(Quit: Leaving.)]
<ee1943>
only way to exit sleep is to force power off by holding down the power button until it dies
mithsv has quit [(Quit: mithsv)]
mithsv has joined #nixos
bts- has quit [(Ping timeout: 268 seconds)]
jensens has quit [(Ping timeout: 240 seconds)]
<c0bw3b>
does defining a builder script also override the unpackPhase ?
mithsv has quit [(Client Quit)]
mithsv has joined #nixos
jb55 has joined #nixos
<clever>
c0bw3b: it overrides everything
<clever>
c0bw3b: its best to not touch builder 90% of the time
<c0bw3b>
clever: kk thanks
ryanartecona has quit [(Quit: ryanartecona)]
<c0bw3b>
clever: makes sense yes, that's an old package already done this ways...