<lukegb>
infinisil: I run overwatch on nixos... inside lutris
<infinisil>
lukegb: Ohh yeah I also tried lutris, but just about everything didn't work about it
<infinisil>
The first error was about wine-mono not being installed, the second about gecko not being installed, both of which were installable from lutris itself, but that's not recommended
<infinisil>
Then the standard installation didn't work at all (no file given error or so), so I tried the manual one by downloading the battle.net launcher myself and passing it to lutris
cosimone has quit [Quit: cosimone]
<infinisil>
But that then failed to start with DXVK errors, at which point I tried to look into how to install dxvk, but gave up after trying hardware.opengl.driSupport and finding no information about it on nixpkgs
<infinisil>
(almost no info, driSupport was the only hint I found)
<infinisil>
lukegb: If you can tell me the magical incantation that made it work for you, I'd love to know!
<siraben>
Does anyone know how to change the target of a fork? I have a fork of B/proj and GitHub says it's "Forked from B/proj" but I want to change it to be a fork of another fork C/proj
<energizer>
i dont think you can do that siraben
<energizer>
just fork the other one
<siraben>
oof
<lukegb>
or file a ticket with support and ask nicely
dada_da has joined #nixos-chat
da_dada has quit [Ping timeout: 260 seconds]
bqv has quit [Quit: WeeChat 3.0]
bqv has joined #nixos-chat
kalbasit has quit [Remote host closed the connection]
kalbasit has joined #nixos-chat
<ashkitten>
omg, jellyfin is so good
<ashkitten>
i'm so impressed by this software
<ashkitten>
especially its ability to cast to any connected device and do syncplay stuff
kalbasit_ has joined #nixos-chat
kalbasit has quit [Ping timeout: 240 seconds]
Dotz0cat has joined #nixos-chat
kalbasit_ has quit [Remote host closed the connection]
kalbasit has joined #nixos-chat
<colemickens>
abathur: for me, 330MB is not even the remotest consideration compared to waiting a second every time my prompt loads
<abathur>
it was just my way of saying you'd successfully nerd-sniped me and I've blown the whole day figuring out how to save myself a few seconds a day without the memory overhead :[
<colemickens>
abathur: aha! I'm sorry for my role in that! :) and consider me interested if you wind up with some neat nix config for it
<abathur>
hehe
<abathur>
it happens
<abathur>
I'll probably fully daemonize it at some point, ironing out the tail end of half-daemonizing it with a coproc
<abathur>
my constraints are a little odd maybe
<abathur>
I don't care for most of the git-gorilla-in-a-birdcage prompts
<abathur>
I'm going to go type git status anyways in most cases, so I just focus on figuring out when to make it red or not
<colemickens>
Is there a difference in that cost though?
<colemickens>
My impression was that it was juts fundamentally hard to get dirty status on a big repo with huge checkouts
<colemickens>
s/hard/slow/
<abathur>
there wasn't before, because I was lazy and just parsed status
<abathur>
that's a bit of it, but I think I found a few wins at least
<colemickens>
huh, someone made a rust client for gitstatusd :P
<abathur>
man, this feels fast now
<abathur>
even git status feels surprisingly fast because you aren't paying the penalty twice
* colemickens
has been wondering idly if there's a way to make naive git commands actually use gitstatusd ?
* abathur
naively thinks he's ready for a screencap, and then realizes how much crap is on his desktop
<abathur>
oh, man, small enough to stick in a gh comment I think
<abathur>
ah, gah
<abathur>
guess they didn't enable video for gists
<abathur>
~ignore the reported time under the commands, it intentionally tries to avoid measuring the prompt-rendering time and such
<colemickens>
abathur: so did you write a replacement or slim down gitstatusd or?
<abathur>
I ended up writing a little thing, though I guess it's not terribly different
<abathur>
it was a little annoying, I ended up doing like half in python and then shelling back out from there to run git diff --quiet because it was night-and-day faster than the best approximation I could figure out in pygit2
<abathur>
lol
<abathur>
so, running the shell function that extracts the git stuff in that top/slow shell
<abathur>
5x
<abathur>
all between 2.19s and 2.89s
<abathur>
5x in the fast one, between 5.3ms and 11.4ms
* colemickens
looks at the rust impl of git
<abathur>
I haven't actually got this bundled up and in my profile yet, just barely got it far enough along to hack it into one shell and start testing it
<abathur>
so my shell function is echoing $PWD onto one half/fd of a coproc that is running the python script
<abathur>
and the python script reads it, uses pygit2 to load a repo object and figure out if it's detached and in sync with upstream (but doesn't freshly check...), shells out for one or two diff commands, and then prints True or False for whether it is dirty, and either the branch name or 'detached @ 11-chars-of-hash'
<abathur>
and then the bash script reads that back out of the other half of the coproc
<abathur>
the lion's share of this time was getting the latter half of that working, which was a little frustrating because even a full invocation was beating the socks off my existing profile, but the setup/import time for the python process is still hefty enough that it'd be a step in the wrong direction for other repos
<bbigras>
^ "a Java-based modernization of the original Master of Orion game from 1993"
<abathur>
colemickens: fwiw, my code's fairly small here, I would guess it isn't a huge lift to RIIR or go or whatever as long as the python batteries are readily swapped...
<colemickens>
nice, that's all I need too
<bbigras>
oh shit. nixos-unstable advanced today
<colemickens>
it's a christmas miracle
<bbigras>
yep đ
<bbigras>
uh. there's glitter or something in element when we use đ
<insep_>
and when using âī¸
<insep_>
and when using đ
<insep_>
and when using đâī¸đ
<bbigras>
nice
<bbigras>
,locate bin java
<{^_^}>
Found in packages: zulu, zulu8, graalvm8-ce, graalvm11-ce, adoptopenjdk-bin, adoptopenjdk-jre-bin, adoptopenjdk-openj9-bin-8, adoptopenjdk-hotspot-bin-8, adoptopenjdk-openj9-bin-11, adoptopenjdk-openj9-bin-13, adoptopenjdk-hotspot-bin-13, adoptopenjdk-jre-openj9-bin-8, adoptopenjdk-jre-hotspot-bin-8, adoptopenjdk-jre-openj9-bin-11, adoptopenjdk-jre-openj9-bin-13, adoptopenjdk-jre-hotspot-bin-13
<abathur>
heh, it can even run in 1.5s with nix-shell shebang overhead :P
cirno-999 has quit [Ping timeout: 265 seconds]
cirno-999 has joined #nixos-chat
waleee-cl has quit [Quit: Connection closed for inactivity]
<abathur>
lulz, oof
<siraben>
âī¸
<abathur>
colemickens: packaging was going somewhat well until osh & thus resholve fell over on the coproc syntax :P
<abathur>
anyways, got a vaguely-working small package that I can load with nix-shell -p and source; even on the commitless repo it's like 120ms vs 20ms
ashkitten has quit [Quit: WeeChat 3.0]
<abathur>
time to turn into a pumpkin or fruitcake or whatever, though
<red[evilred]>
Data is delayed, but that's almost certainly operational
<bbigras>
and there's gaps. it's censored! /me put tinfoil hat /s
<red[evilred]>
Except there's gaps in the exact same place every other day :)
<red[evilred]>
*removes tin foil hat*
<bbigras>
"Except there's gaps in the exact same place every other day" to make it less suspicious! but I mean if they would fake it, they could fake the data without gaps.
<viric>
__monty__: sad enough I'm writing my tool to find duplicates. None fit what I wanted
<viric>
rmlint almost, but it deadlocks and fails often for me (glib thread things)
<__monty__>
It's a really hard problem.
<__monty__>
I implemented one in ranger.
<red[evilred]>
So, I've been learning a new programming language and it's kicking my behind.
<pie_>
whatcha been learnin
<red[evilred]>
I was all like "I'll do advent of code with it, that will help with my fluency"
<red[evilred]>
25 days later and I've still not completed day one
<red[evilred]>
ponylang.io
<pie_>
ah hehe
<pie_>
me, yesterday, meanwhile: oh boy maybe i should learn about/try to do lua sandboxing (which seems relatively easy)
<red[evilred]>
Oddly, it's not the reference capabilities that's kicking my butt
<red[evilred]>
It's remembering how to write OO code since I've been mostly functional the last 5 years or so
<red[evilred]>
How that working out for you?
<pie_>
:'D i kind of never really learned how to oo in the first place
<pie_>
i can do python with some basic inheritance and im like ok, but none of those design patterns
<pie_>
well, i talked to some people just now and it seems like its not completely hopeless. lua seems to be reasonably secure if you remove everything from its environment, and at least on linux i should be able to use seccomp to drop all privileges, dunno about windows...and then have the host side inject the code into the guest interpreter, and use some sort of communication channel
<pie_>
so then its a matter of having safe libraries availible, and making my APIs secure which is apparently the bigger problem
<pie_>
red[evilred]: i was actually thinking about asking you for input once id distilled things a bit more
<pie_>
(i also do more intellectual masturbation than ever actually writing any code that actually does anything...)
<pie_>
*and not calling native code
<pie_>
this all started from me wanting some manners of addons with a capability system in my haskell app
<pie_>
Which might be kind of questionable in the first place because it's probably not a good idea to generalize before you have a better idea of the problem space, but eh.
<viric>
__monty__: what is 'ranger'?
<pie_>
I'm still wondering what exactly I like about interpreters. And thining about live coding, and the actor model, and code reloading. Really need to get mo into erlang.
<bbigras>
isn't it the cli filemanager with the vim control?
<red[evilred]>
Yeah, I love and adore the BEAM
<red[evilred]>
There really is nothing better for server side imho
<red[evilred]>
But I'm looking for something I can make single executables from
<red[evilred]>
Whereas the BEAM is a whole VM
<red[evilred]>
If only I could compile that into a single executable
<pie_>
Its "just an RTS" man :PPP
<pie_>
cant you? ("compile")
<viric>
mh something may have changed in ffmpeg that my prev commands don't work for hw decoding/encoding
<red[evilred]>
RTS?
<pie_>
runtime system
<bbigras>
real time strategy game
<bbigras>
I prefer turn-based
<pie_>
bbigras: :P
<pie_>
i gues not really usually used as a term for the whole vm but, in the sense that libc is an RTS like haskells RTS is an RTS and the jvm probably has an rts :P ....so I guess its usually used to denote the C-side implementation
<pie_>
i mean people dont really say libc is an rts
<bbigras>
beware that systemd-resolved is broken in the latest nixos-unstable. Fixed with #107572 it seems
<pie_>
question is really what single-exe gets you i gues
<pie_>
sure making a self-extracting thing might be cheating but it might also be fine? depending on your size
<red[evilred]>
Just ease of installation and packaging.
<pie_>
or maybe not even depending on your size. just unpack erlang next to you, and for multiple instances youd just build in some logic for handling that?
<pie_>
sure you dont want users to have to install erlang separately
* supersandro2000
wasted another hour with fixing a bug that was already fixed by me a couple of weeks back....
<red[evilred]>
Well, erlang does have the concept of OTP packages
<red[evilred]>
Which works flasks flawlessly on windows
<red[evilred]>
But each Linux distribution does its otp packages differently
<red[evilred]>
And supposedly, nixos has issues
<red[evilred]>
But that's off topic for this channel
lejonet has quit [Quit: WeeChat 2.4]
<pie_>
supersandro2000: nooooo xD
<pie_>
red[evilred]: i meant not passing the packge handling to the OS or whatever but rather bundling everythng your app needs in its self extracting installer or whatever
lejonet has joined #nixos-chat
<red[evilred]>
Right
<red[evilred]>
BRB, family stufffs
Hm7000 has joined #nixos-chat
rajivr has quit [Quit: Connection closed for inactivity]
cosimone has joined #nixos-chat
<manveru[m]>
got nix working on my remarkable 2, now i just need to build some kind of nixos-rebuild for it... there are so many one-off hacks of random people that'd be a pain to maintain otherwise :P
<manveru[m]>
but i think deploy-rs might work for it... finally a chance to try that
<manveru[m]>
sphalerite: need to manage more than $HOME :)
<samueldr>
AFAIK nothing that's not underpowered or way too overpowered really
<sphalerite>
elvishjerricco: hardware? A number of us have ordered the honeycomb lx2k
<sphalerite>
most of us haven't got it yet though
<elvishjerricco>
samueldr, sphalerite: Cool, I'll take a look.
<manveru[m]>
I started making a profile and activate script and noticed I'm going to slowly implement a crappy version of nixos-rebuild
<lovesegfault>
bbigras: i hit that resolved issue :P
<lovesegfault>
Was thoroughly confused
<bbigras>
yeah. I tried to bump my "channel" to the fix but it didn't change a thing. ah maybe the rollback kicked again. I'm disabling the rollbacks for now. one of the deploy-rs dev is supposed to add a log soon.
<supersandro2000>
github blocks emoji reacts for 50% of the commenters
<__monty__>
As an A/B?
<samueldr>
energizer: uh, colour only in chinese and without play store
<samueldr>
that's odd
<samueldr>
knowing that the source code for their vendor kernel tree would be available, I'd smash the buy button so hard I'd break through the floor
<supersandro2000>
I can react to the 2nd, 3rd commenter but not the author
<supersandro2000>
and can't follow those people. So people started to block me or what?
<supersandro2000>
but already found a workaround: I quote there comment and put my emoji under that
<sphalerite>
supersandro2000: want me to block/unblock you to test?
<supersandro2000>
sphalerite: whats your github name?
<sphalerite>
lheckemann
<supersandro2000>
found a PR
lunc has quit [Ping timeout: 256 seconds]
<sphalerite>
blocked
<sphalerite>
supersandro2000: have you tested?
<supersandro2000>
yep. can't react
<supersandro2000>
you know what. I am just skipping those PRs.
<sphalerite>
unblocked
<sphalerite>
don't skip mine ;)
sorear has quit [Ping timeout: 260 seconds]
spacekookie has quit [Quit: **aggressive swooshing**]
spacekookie has joined #nixos-chat
sorear has joined #nixos-chat
<ashkitten>
psa: add the nofail option to any non-critical filesystems (especially bind-mounts) or you'll be stuck in rescue mode if those filesystems disappear
<ashkitten>
i spent 4 hours last night after bedtime figuring out why my server wasn't coming back up, and it was because i moved a directory in my nextcloud
<bbigras>
damn
<bbigras>
btw nixos-unstable advanced again 17 mins ago
<ashkitten>
i mean, realistically it would've been easy to figure out why it wasn't coming back up if i had console access
<ashkitten>
but kimsufi is an awful hosting provider
<ashkitten>
so i had nothing but an ancient netboot rescue environment based on debian stretch
<ashkitten>
that couldn't even read the newer journal files
<samueldr>
ashkitten: I generally start qemu-kvm with the whole disk on kimsufi, while tunneling the display as vnc, if I need to unbreak things
<samueldr>
I mean, ssh tunnel to qemu's own VNC
<ashkitten>
samueldr: i would, but zfs didn't like qemu's disks
<ashkitten>
maybe a newer version of qemu than 2.2 or whatever would work :)
<ashkitten>
the debian environment was so ancient it was hard to do anything with it
<genevino>
don't remind me of that debian woody box i've seen at a customer a year ago or so
<samueldr>
last time I used that was when I setup the server though, but I wouldn't think their image has been updated since
<genevino>
THAT was something special
<ashkitten>
heh
<ashkitten>
anyway i also found out that znapzend stopped working sometime back in november
<ashkitten>
er
<ashkitten>
september
<ashkitten>
because of a mismatch in ssh known_hosts
<ashkitten>
somehow...
<infinisil>
ashkitten: Oh that's one of my main motivations for creating a zero-setup ssh module for nixus
<infinisil>
Because now I could write a znapzend module (haven't done it yet though), which sets up that the ssh connections works, without having to do anything manually
<ashkitten>
that'd be great
<ashkitten>
could you also make it so it sets up the receiving end with delegated permissions correctly?
<infinisil>
delegated permissions?
<infinisil>
In Nixus, you can write modules that configure multiple machines, so pretty much anything is possible
<ashkitten>
i gave the server a special no-password ssh key that unlocks the znapzend user on my desktop, which i have `zfs allow`ed permissions enough to receive znapzend sends under a specific parent dataset
<ashkitten>
(server backs up to desktop)
<infinisil>
Oh actually Nixus doesn't do any key management yet, so I guess you need to at least manually create keys for now
<genevino>
and i have 5 machines with nixos, 3 of them with usb audio devices.
<genevino>
no problems here.
<genevino>
what hardware?
<hexa->
# reboot
<hexa->
Failed to talk to init daemon.
<hexa->
genevino: this is one of your issues, right?
<hexa->
also maybe to a lesser degree (but due to the same issue) this:
<hexa->
# systemctl status
<hexa->
Failed to read server status: Failed to activate service 'org.freedesktop.systemd1': timed out (service_start_timeout=25000ms)
<joepie91>
genevino: Gigabyte Aorus Pro x570 motherboard, on-board audio, old pair of speakers but issue does not occur under Windows
<hexa->
joepie91: have you tried an older generation?
<joepie91>
I have not
<joepie91>
I may have already GC'ed them :P
<hexa->
ah neat, so much free space!
<joepie91>
welllllll the reason I GC them is that I don't :D
<joepie91>
my PC, too, suffers from the "every disk is always 95% full" problem
<hexa->
rough
<joepie91>
and I have like 10TB in here by now
<hexa->
we're in the same boat then
<hexa->
I'm looking for a NAS upgrade but can't find proper prosumer hardware for it.
<joepie91>
my NAS also contains 95% of 10TB :D
<hexa->
96% of 9.0T :)
<joepie91>
ha
<joepie91>
that 95% rule is so nicely universal
<genevino>
joepie91: what chipset is that onboard thing exactly?
<genevino>
joepie91: anyways, your problem is 100% solvable, it's possible that the defaults will do what you described, but it's still definitely solvable.
cole-h has joined #nixos-chat
<joepie91>
genevino: Realtek ALC1220-VB - basically the same as nearly every other x570 motherboard