supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
tokudan has quit [Remote host closed the connection]
tokudan has joined #nixos-chat
tokudan has quit [Remote host closed the connection]
tokudan has joined #nixos-chat
vesper11 has joined #nixos-chat
cole-h has quit [Ping timeout: 264 seconds]
kalbasit has quit [Ping timeout: 240 seconds]
<ldlework>
Distrotube did a Top 8 Tiling Window Managers video, and QTile was #2 and he showed the layout I contributed years ago. It was on screen for like 3 seconds, but still! :P
<hexa->
didn't consider sway
<hexa->
rip
KREYREEN has quit [Remote host closed the connection]
KREYREEN has joined #nixos-chat
<abathur>
rest in pizza
<abathur>
round beds should make a comeback, just for this joke
<abathur>
cheesy sheet set, the works
<supersandro2000>
if you are reviewing so fast that the github api does not even know that the PR exists
<ashkitten>
samueldr: re, randall munroe tweets: very good example of correct answers to incorrect questions
<ashkitten>
that's a thing i saw someone do with gpt-3 a while ago
<ashkitten>
in fact, they determined that gpt-3 can actually tell if questions are nonsense, and if you instruct it to give a certain response to nonsensical questions instead of trying to answer them, it will
srk has quit [Ping timeout: 240 seconds]
srk has joined #nixos-chat
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
<gchristensen>
lol, picking a fight with the Google's VP of Infrastructure, as you do
red[evilred] has joined #nixos-chat
<red[evilred]>
I'm all about iconoclasty
<red[evilred]>
and honestly -=- people who rise to that level of experience know that being challenged makes them better at what they do
<red[evilred]>
I'm reminded of one of my annual reviews:
<red[evilred]>
1. Positive: Is comfortable communicating with any later of the organization, including executives, CEO, and major shareholders.
* samueldr
checks twitter
<red[evilred]>
2. Needs Improvement: Is too comfortable communicating with any layer of the organization, including executives, CEO, and major shareholders
<red[evilred]>
:-)
* colemickens
bad cole. I looked at twitter and lost 25 minutes :|
<red[evilred]>
If you want to lose another hour, #SerialkillerOrScientist
<red[evilred]>
pretty funny
<red[evilred]>
At Home Depot: "Excuse me maam, can I help you find something?"
<red[evilred]>
Scientist: "I'm trying to work out if this drillbit is strong enough to cut a clean hole through bone"
<red[evilred]>
Home Depot employee walks away
<red[evilred]>
or: "I get self conscious taking my car to the Nissan dealership for service now, since last time I took it I forgot I left a frozen baboon on the back seat"
rajivr has joined #nixos-chat
<lovesegfault>
Any tailscale users around?
<gchristensen>
I dunno about users but developer(s) probably
<lovesegfault>
gchristensen: who?
<gchristensen>
da\nderson
<lovesegfault>
I have magic DNS questions :P
<lovesegfault>
Oh, nice, I'll ask them:
<lovesegfault>
danderson: is it possible to use / set up magic DNS with a local resolver like unbound/systemd-resolved running as well?
srk has quit [Remote host closed the connection]
srk has joined #nixos-chat
<lukegb>
I should really replace my horrendous mess of nix-generated etc/hosts with either magic dns or... actual dns
<elvishjerricco>
Is it possible to get hardware accelerated video in firefox with an nvidia gpu?
<lovesegfault>
lukegb: Yeah, I use actual dns right now and it's a PITA
<lovesegfault>
trying to move to magic DNS since I put all my systems on tailscale yesterday
<lovesegfault>
but I just can't get it to work
endformationage has quit [Quit: WeeChat 2.9]
Dotz0cat_ has joined #nixos-chat
Dotz0cat has quit [Ping timeout: 256 seconds]
<bbigras>
lovesegfault: probably not related since I don't use magicdns yet. https://github.com/NixOS/nixpkgs/issues/104414 . I had to reboot to fix it. I'm not sure I'm able to reproduce anymore.
<{^_^}>
#104414 (by bbigras, 1 week ago, open): split dns with systemd-resolved + tailscale stop working after a deploy
<lovesegfault>
bbigras: Oh, interesting
<lovesegfault>
I decided to try tailscale after seeing it in your config :D
<lovesegfault>
bbigras: did you have to tell systemd-resolved to use the 100.100.whatever address as the DNS server?
<lovesegfault>
100.100.100.100
<bbigras>
lovesegfault: hehe. It's pretty nice. I wish they would had support for multiplies networks/account. I still need to use zerotier for that but I would rather use something based on wireguard.
<bbigras>
lovesegfault: is that for magic DNS? I don't use it. I just use split DNS with networks/resolved to make *.workdomain use my work DNS
<lovesegfault>
Yeah, for magic DNS
<lovesegfault>
I just wanted to play with it
<bbigras>
I would like to use it too but I'm not sure how. Tailscale might be doing some magic with (is it resolvconf?)
waleee-cl has quit [Quit: Connection closed for inactivity]
<bbigras>
I linked you my report just in case you had a similar problem
<lovesegfault>
I have no idea how it works
<lovesegfault>
maybe danderson can help us understand
<bbigras>
With my issue I was wondering if things were not set in the right order or something. Or maybe one program took over my DNS settings.
<lovesegfault>
Right, I think unbound is my problem
<lovesegfault>
but I don't want to disable it b/c that it's own can of worms
<bbigras>
I'm going to bed. You know there a flag with `tailscale up` for the DNS stuff? I think it's on by default
<lovesegfault>
bbigras: just learned about that :D
evanjs has quit [Ping timeout: 260 seconds]
evanjs has joined #nixos-chat
supersandro2000 has quit [Quit: Ping timeout (120 seconds)]
evanjs has quit [Ping timeout: 260 seconds]
evanjs has joined #nixos-chat
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
tilpner has quit [Read error: Connection reset by peer]
tilpner has joined #nixos-chat
red[evilred] has joined #nixos-chat
<red[evilred]>
bbigras - got a sec?
<red[evilred]>
have a question about one of the packages you maintain
<red[evilred]>
oh wow - I'm two hours late
<red[evilred]>
nevermind :-)
cole-h has joined #nixos-chat
<sphalerite>
$ sudo make install PREFIX=/usr
<sphalerite>
whyyyyyyyyyyy are they recommending this aaaaaargh
<samueldr>
hi, welcome to poor practices in operating systems
<samueldr>
I'm a bit annoyed at how long it took me to figure out the reason the efi stub'd kernel didn't boot was because I didn't give much memory to the QEMU VM
<samueldr>
I don't remember what the defaults are
<samueldr>
but apparently it's not enough, so the EFI firmware can't start it
<sphalerite>
oh yeah the default is pretty tiny
<samueldr>
and it's not the first time it bites me :)
<sphalerite>
yep, I think I've run into that multiple times too…
<samueldr>
though now I have mobile nixos doing a generic UEFI build, and it took much less time than I thought to get things going
<sphalerite>
\o/
<samueldr>
Nix really is great to just get things falling into place without much worry
FRidh has joined #nixos-chat
<samueldr>
I pretty much just reused what was done for u-boot systems, but instead of copying files for u-boot, it prepares the kernel with the appended stage-1 into that efi stub'd kernel
tilpner_ has joined #nixos-chat
tilpner has quit [Ping timeout: 256 seconds]
tilpner_ is now known as tilpner
FRidh has quit [Ping timeout: 260 seconds]
FRidh has joined #nixos-chat
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-chat
thibm has joined #nixos-chat
<eyJhb>
Seriously, fuck google recaptha atm... I get a thousand images each time
<srk>
lol, same complaint on another channel few minutes ago :D
<srk>
I was raging bout that yesterday, it's so bad
__monty__ has joined #nixos-chat
supersandro2000 has joined #nixos-chat
<eyJhb>
They have started going all in srk. In the beginnig it was OK, but now it is insane
<eyJhb>
It is the same with youtube adds, there are SOOO MANY...
<__monty__>
Heh, my HM activation sometimes doesn't quite get firefox addons right, like this morning. First site that's open *immediately* triggers me to check up on that.
<__monty__>
Time to check out AoC day 2 : )
<etu>
:)
<etu>
I failed on several points in todays puzzles. I didn't read the instructions carefully enough and implemented a different validation than they asked for. Waking up extra early and only be awake for 15 minutes makes it a bit harder :p
<thibm>
I just finished them
<thibm>
algorithm: 20%. parsing: 80% :(
<thibm>
lib/strings.nix: splitString NOTE: this function is not performant and should never be used.
<thibm>
huhu
<thibm>
Both in terms of lines of code and in terms of spent time
<etu>
:D
<etu>
thibm: So nix today again? Wow :)
<etu>
thibm: I'm impressed :)
<thibm>
yes, still Nix
<thibm>
It was actually easier than yesterday (for the algorithm itself. Parsing was harder because, well, Nix.)
<thibm>
I'll try to do all of them in Nix
<thibm>
4-5 lines of (not compressed) Nix for each part
<thibm>
(without the parsing)
cole-h has quit [Ping timeout: 246 seconds]
<__monty__>
I'm always humbled by the people who get single-line solutions per part (modulo reading input and printing results).
<thibm>
Well, that's usually easy with functional programming ;)
<thibm>
But un other languages that are usually more verbose it can be impressive, yeah
<insep_>
anyone done part2 already?
<thibm>
yes
<insep_>
my program seems to work with example input but not with my input :(
<insep_>
either parser broke or something else, idk
<thibm>
insep_: the toy example does not really cover all cases.
<insep_>
the issue is that i have the same input and the only thing changed from part1 is validation algorithm
<thibm>
do you get an error or just the wrong result?
<insep_>
wrong result
<thibm>
well then your validation algorithm should be wrong
<thibm>
if it worked for part 1 the parsing should be good
<thibm>
(I should use builtins.split and/or builtins.match instead of costly lib.splitString next time)
<insep_>
i used xor
<Taneb>
insep_: did you do what I did first time and check the thing's in range by seeing if the index is greater than the length of the string
<thibm>
is your indexing corretc?
<Taneb>
^ the puzzle expects strings to be 1-indexed whereas most programming languages have 0-indexed strings so you need to take that into account
<insep_>
oh
<insep_>
i get it now
<insep_>
i messed my algorithm that i came up at school
<insep_>
indexing was incorrect
<insep_>
i used +1 instead of -1 :D
<thibm>
haha
<Taneb>
insep_: working now?
<insep_>
yes
<thibm>
(so we have a logical implication in the Nix language (which is really nice) but no xor.)
<Taneb>
:)
<Taneb>
thibm: does != work for or
<Taneb>
xor
<thibm>
sur
<thibm>
sure it works
<thibm>
although it does not ensure that the operands are boolean
<eyJhb>
Does it count making a runCommand using Nix? ;) :p
<thibm>
Actually, in my version, I wrote `!(l == r)`, not `l == r`. Not sure why
<thibm>
eyJhb: it will not be pure Nix anoymore but do as you wish ;)
<thibm>
(I meant `l != r`)
<eyJhb>
What if I do a runCommand to run Nix! \s I will stop now
<eyJhb>
:D
<thibm>
:)
<Taneb>
Ugh, cloning nixpkgs on my terribly slow wifi is... slow
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
<thibm>
Taneb: I cloned it on night last time. It takes hours and internet access is almost unusable while it's downloading…
<thibm>
You can use --depth if don't care about the history
<Taneb>
I *could* use depth but didn't think about that and I'm 64% of the way through now
<etu>
insep_: I had that problem as well, it worked with sample but not real. But in my case I hadn't read the instructions carefully enough and it was just luck that it worked with the test.
<Taneb>
I know it's the sunk cost fallacy but I'm not in a hurry
<thibm>
:)
<thibm>
Taneb: I do developpment/packaging on servers with good Internet access. It's just not doable locally
<Taneb>
Although... if someone wants to see if haskellPackages.integer-roots builds with doJailbreak = true or doCheck = false and can do so before me I would be very happy (the test suite has an outdated bound on smallcheck)
<thibm>
(Of course I need a checkout for NixOS)
<thibm>
Taneb: I can try
<Taneb>
Thanks :)
<Taneb>
Ah, looks like the issue has been fixed in the upstream repo but not on Hackage yet
<supersandro2000>
eyJhb: What are you after? Help with troubleshooting or user expierence?
<eyJhb>
Security. I am not sure how they structure it, or if they are trustworthy to use
<eyJhb>
structure it -> basically the infrastructure of the addon, which servers it contact, what info is stored, etc... Because as far as I saw, they save a lot on their servers
FRidh has quit [Ping timeout: 240 seconds]
FRidh has joined #nixos-chat
Dotz0cat has joined #nixos-chat
Dotz0cat_ has quit [Ping timeout: 240 seconds]
<eyJhb>
gchristensen: Did GH delete volth user?
<gchristensen>
yea
<eyJhb>
Is there any writing on it or anything? I'm interested in knowing why.. .:(
<gchristensen>
no
Dotz0cat has quit [Ping timeout: 265 seconds]
<bbigras>
red[evilred]: I'm here if you still have a question.
<mtr>
can't see a nixpkgs derivation for latexmk :/
<thibm>
mtr: texlive.latexmk
<mtr>
ty <3 I always forget the namespaces haha
<thibm>
it's more than a namespace here, you should read the texlive entry in the manual
<mtr>
ah, thanks for the heads up
<thibm>
you can get latexmk.pl with texlive.latexmk, but that's maybe not what you want
<mtr>
just checked the manual, the combine fn seems like a pretty neat way of doing things
<thibm>
Yeah it's neat
<thibm>
I maintain a "minimal" set of packages in the .nix file for a tex project
<thibm>
and edit the file with nix-shell --run "vim main.tex" (so that vim plugins can find the right binaries and stuff)
<thibm>
(or in continuous integration). Don't have pdflatex & co in my default profile anymore
red[evilred] has joined #nixos-chat
<red[evilred]>
re: bbigras (IRC)
<bbigras>
red[evilred]: re
<mtr>
thibm: ah nice, I haven't used latex in quite a while but if I do I might do something similar
<__monty__>
thibm: You should check out direnv and either lorri or nix-direnv. Then changing to that directory loads your environment, the shell is reevaluated when shell.nix changes and it works wonderfully with other shells, so you're not stuck with bash unless you want to be.
<thibm>
Yeah, the tools (thanks anyway) I just didn't have time to play with them
<thibm>
actually I have a lot of "pending" changes to my system for a few months
<__monty__>
It was a major QoL improvement for me.
<thibm>
Yeah, I should definitively do it
<red[evilred]>
bbigras (IRC): I do'nt recall which product it was oof the top of my head but I polled you as the maintainer about it because fixing CVE(s) for it required a bump in version
<red[evilred]>
and potential changes to configurations etc
<red[evilred]>
BUT
<red[evilred]>
here['s my confusioin - I just re-visited the ticket and now I can't find the old version at all
<red[evilred]>
and both of us were talkinga bout the older version
<red[evilred]>
but according to git it was the new version at the time that we spoke
<bbigras>
was it for keycloak?
<red[evilred]>
so - I just wanted to ask you to confirm thhat I'm not completely mad
<{^_^}>
tailscale/tailscale#683 (by xXorAa, 15 weeks ago, open): DNS Support breaks resolving for SNAP on Ubuntu
<bbigras>
lovesegfault: I'm guessing if you need to use unbound you could use `tailscale up -accept-dns false` and point unbound to 100.100.100.100. In my case I would like to avoid that since I would prefer resolving to tailscale and use resolved as a fallback so I can still use dnssec and tls
<bbigras>
but it's weird. with tailscale's /etc/resolv.conf, magicdns doesn't seem to work
<bbigras>
for me
<bbigras>
oh it's working. I had to use `bruno-job.something.beta.tailscale.net`. I thought it would work with only the short name.
<gchristensen>
....aaaaand the reason I suffered through the terraform 0.12 -> 0.13 upgrade was for naught
KREYREEN has quit [Remote host closed the connection]
<gchristensen>
silly me trying to cheat terraform's "providers can't be dynamic" goop
KREYREEN has joined #nixos-chat
<bbigras>
what do you mean? I tried terraform 0.13 recently but I so far I only let terraform install the plugins instead of using the nix way.
<gchristensen>
I want to put a server in every AWS region, but you have to statically define all your providers
<siraben>
colemickens: I used to rely on XFCE to manage audio settings, but now I don't seem to have a way to change audio settings
<bbigras>
gotcha
<siraben>
Should I use ncpamixer?
kalbasit has joined #nixos-chat
<siraben>
Interesting, it appears as Microphone (unplugged) in pavucontrol
<gchristensen>
I don't suppose anyone knows if I can contort `printf "%s %s" "mything"` in to printing "mything mything" ?
<__monty__>
Ah, apparently some printf's provide indexing `printf '%1$s %1$s' 'mything'`
<__monty__>
Fish printf doesn't though.
<__monty__>
Hmm, may only be the C function on closer inspection. It's a SUS POSIX extension.
lunc has joined #nixos-chat
<LinuxHackerman>
sounds sus.
<gchristensen>
something a hackerman would slip in
<__monty__>
Like a banana?
<eyJhb>
Among us, but with minions.
<sphalerite>
oh no, not bananas. I don't like bananas.
<eyJhb>
GET SPHALERITE WITH THE BANANAS!!!
<supersandro2000>
but what if we are a banana republic?
<hexa->
pretty sure LinuxHackerman is an imposter
<LinuxHackerman>
I haven't actually played among us yet x)
<insep_>
play among-sus
<insep_>
anywhy
<sphalerite>
nor that
<insep_>
s/hy/ay/g
<insep_>
anyone remember that website with coding challenges where you complete tasks by taking input from task page and then submitting code that works with that input?
<__monty__>
Project Euler?
<insep_>
iirc that website also had separate categories for brainfuck and regex
<insep_>
also it had blue-ish background
<__monty__>
Codewars similarly does eval.
<__monty__>
Maybe neither of those then.
<insep_>
it's neither of those
c4rc4s has joined #nixos-chat
<__monty__>
Hackerrank?
<insep_>
no
<insep_>
it required you just to register to solve tasks and website wasn't anything fancy
<tilpner>
insep_: Your description matches adventofcode.com, except for the BF/regex categories
<insep_>
it's not advent of code either, they had an archive of tasks that you could solve any time at any order
<red[evilred]>
exercisms?
<__monty__>
AoC also doesn't work by submitting code, just the solution.
cole-h has joined #nixos-chat
<insep_>
no :(
<gchristensen>
terraform is so bizarre in ways ......
<ldlework>
gchristensen: yeah but i'm thankful to have it
<gchristensen>
yeah...
<ldlework>
did they at least improve hcl in the last few years?
<gchristensen>
provider aliases, though, so janky.
<ldlework>
i think the dsl is called hcl right
<ldlework>
yep, nailed it
<tilpner>
insep_: Codingame?
<ldlework>
I stopped using it before there was supposed to be some impending major improvement to hcl
<gchristensen>
I'm trying to create a machine in many regions and it involves extensive copy-paste
<ldlework>
gchristensen: there are various extension projects
<ldlework>
that allow you do some abstractions that you can't by default
<ldlework>
sad to hear you'd be dealing with anything like that, years later
<insep_>
tilpner: no :(
* tilpner
will stop now
<ldlework>
no but you guys remember that one Python coding challenge
<ldlework>
Where you had to do things like look at the HTML source and other things?
rajivr has quit [Quit: Connection closed for inactivity]
<ldlework>
holy crap I haven't thought of that in a decade or more
<cransom>
gchristensen: yeah, i've found similar things where HCL was anti-helpful. it made me want to have some other configuration language that would spit out the json that tf would use.
<crazazy[m]>
so apperently GH actions had ben busy building my system config for 1.5 hours
<crazazy[m]>
i'm not sure how that happened
<__monty__>
Dhall could be another DRY alternative with one of the dhall-terraform projects.
<__monty__>
crazazy[m]: Maybe libreoffice or a browser wasn't cached or something?
<__monty__>
Had no idea GH Action limits were so generous though.
<crazazy[m]>
i dunno i saw virtualbox coming by
<crazazy[m]>
i hope its cached now, and that my cachix isntance didnt overflow with a single update
<bbigras>
nix-build-uncached is nice to reduce cachix's usage
<bbigras>
I have a 3h 33m gh action run from last month.
<crazazy[m]>
oof
<bbigras>
I wonder what is the limit
<bbigras>
got a 6h 5m 10s from 2 months ago
<bbigras>
I'm not sure but maybe I was trying to build firefox
<bbigras>
I think 6h might be the limit
<bbigras>
"Each job in a workflow can run for up to 6 hours of execution time. If a job reaches this limit, the job is terminated and fails to complete."
<crazazy[m]>
i scheduled my system to update every saturday
<crazazy[m]>
GH actions to update my system*
<crazazy[m]>
its downloading a lot of stuff on my laptop ho
<crazazy[m]>
maybe kernel change?
<bbigras>
I tried to switch my config to flakes this week and I had to push like 17GB to my laptop. it was weird. but my config had an old kernel pinned.
<bbigras>
but I think I didn't see the download amount before using flakes. I was using nix-build.
<colemickens>
17GB still makes me think you had a stale channel
<colemickens>
I just can't think of any other explanation that makes sense
<samueldr>
colemickens: what is a mass rebuild other than a stale channel?
<samueldr>
it could have been the previous channel, but a mass rebuild due to an stdenv change would have still rebuilt everything, no?
<colemickens>
but switching to flakes would'nt just cause a stdenv change?
<cole-h>
Well, it could dependingon the difference in revisions between the channel and the flake
<colemickens>
that's what I said though?
<samueldr>
ah, in that sense, straight up switching to the same revision
<eyJhb>
Well, if the reason for volth disappeares becomes clear, it would be nice to know
lunc has quit [Ping timeout: 240 seconds]
<bbigras>
colemickens: how stale? would 17GB be normal for a week or 2?
<bbigras>
I think the last update I did was not as bad. I'll check the next one.
<__monty__>
bbigras: Any step between revisions could entail a huge rebuild.
<__monty__>
It's not necessarily a steady increase.
<bbigras>
__monty__: yeah it was probably that. I think that I just didn't see the download size when I was using `nix-build` but I see it now with `nix build` or flakes.
<colemickens>
it'd be kinda funny if bbigras reboots and goes "wait, when did we get to kernel version 5?"
<bbigras>
haha. so there was a PR months ago to add back the zen kernel. I didn't want to wait to use it so I added the PR in my config. and I forgot to switch back to zen from nixpkgs when the PR was merged.
<crazazy[m]>
i've been trying to download intellij community and stunt rally for about an hour and my internet connection isnt that bad
<crazazy[m]>
i dont know why it wouldnt download
vesper11 has joined #nixos-chat
vesper has joined #nixos-chat
vesper11 has quit [Ping timeout: 240 seconds]
FRidh has quit [Quit: Konversation terminated!]
vesper has quit [Ping timeout: 265 seconds]
<jtojnar>
supersandro2000: I am porting nixpkgs-hammering to python now so it might be easier to integrate
<lovesegfault>
Does anyone know what happened to volth?
<lovesegfault>
cc. gchristensen
<cole-h>
Graham probably has more info, but it boils down to "GitHub disappeared them"
<colemickens>
um what
<samueldr>
not the first time it happened
<colemickens>
weren't they doing like nix-windows work and stuff. that's frustrating.
<samueldr>
there is a "process" at github that can end up disappearing a user's data
<colemickens>
I mean, its frustrating regardless of what they were doing,...
<energizer>
samueldr: what happened before?
<samueldr>
not clear, but likely it was that the usage pattern was "adversarial"
<samueldr>
like, being from "undesirable" origin countries, maybe using proxies and such
<energizer>
like, some abuse detector screwed up?
* colemickens
stops ircing and gets back to trying to selfhost matrix and a git forge so he can stop being a hypocrite
<lovesegfault>
Right, I wanted to know _why_ they were banned
<samueldr>
or worked as designed
<lovesegfault>
or nuked more like
<samueldr>
but it's extremely bad that banning the user means DESTROYING any integrity and trust in github
<samueldr>
they already have a process for account deletions
<samueldr>
where they re-attribute the issues, PRs and such to the "ghost" github user
<samueldr>
why the flipping flipetty heck do they just disappear everything here?
<colemickens>
I've seen it before and then gone to correlate to github emails, and then find that the user has a new github account anyway?
<colemickens>
I think sometimes maybe it's self-initiated?
<colemickens>
Idk
<samueldr>
anyone has an orange site account and wants to make some noise?
<colemickens>
lol who's gonna volunteer for that privilege
<samueldr>
because apparently in twenty twenty you have to create enough outrage before anything is done for anything
<bbigras>
colemickens: you saw https://radicle.xyz/ this week? I hope it's not based on a blockchain. it seems to mention some crypto currencies
<colemickens>
Oh, I'd seen it a while back, glossed over it when I saw it pop up this week. Seems like it's grown a lot or they've just put a lot more effort into marketing.
<colemickens>
wow this is night and day different, and I definitely love the pitch.
<colemickens>
whoa, glad you pointed it out again
<colemickens>
bbigras: I assume you promptly packaged it for nixpkgs ;)
<cole-h>
oh no
<cole-h>
My GPU is artifacting
<cole-h>
This is a bad time for my GPU to start artifacting
<bbigras>
colemickens: haha not yet. I didn't try it yet. It's not the first p2p git thing I saw since youtube-dl-gate but it might be the one with the better brand/marketing. I'll check if I can try it.
kalbasit has quit [Remote host closed the connection]
kalbasit has joined #nixos-chat
<samueldr>
yay github is not totally a pain in the bum
<samueldr>
github gists will be silently truncated if they're too big to render
lassulus has quit [Remote host closed the connection]
<__monty__>
That sounds... helpful >.<
lassulus has joined #nixos-chat
vika_nezrimaya has joined #nixos-chat
endformationage has joined #nixos-chat
<vika_nezrimaya>
While working on cloud-init tests in NixOS, I think I understood one thing. cloud-init and NixOS aren't best friends, and if stateful configuration at runtime is needed it's best to manage with plain old Systemd units, ordered before and having a wantedBy on the unit whose stateful data is being managed (like SSH keys)
<vika_nezrimaya>
But the thing with cloud-init is the fact that a lot of clouds expose a cloud-init-compatible interface at 169.254.169.254, right?
<vika_nezrimaya>
everyone uses cloud-init because everyone uses cloud-init
<vika_nezrimaya>
Isn't it weird how things become standard just because they get enough traction and everyone uses them and if you're not using them you're weird?...
<gchristensen>
that is how the internet works, vika :) RFCs are nothing more than 'here's a thing I made' and then it is up to the rest of the internet to decide if they care or not
<gchristensen>
it is really cool
<gchristensen>
also totally annoying sometimes!
<vika_nezrimaya>
oh hi gchristensen
<vika_nezrimaya>
didn't I see you like three days before watching a recording of a presentation on NixCon?
<vika_nezrimaya>
something about NixOS at tumblr, was that you?
<gchristensen>
that was me!
<vika_nezrimaya>
yay!
<vika_nezrimaya>
do you still work there?
<vika_nezrimaya>
do they still use NixOS?
<vika_nezrimaya>
if you can answer that of course
<gchristensen>
I left at the end of 2017 I think, and i don't know
<supersandro2000>
jtojnar: I got it working but it does not work well with multiple PRs
<supersandro2000>
and the sorting rule is a bit annoying
<jtojnar>
the python rewrite will allow making it more flexible (including disabling rules)
<vika_nezrimaya>
Back on topic of cloud-init, I've been thinking about provisioning NixOS instances with something similar to cloud-init?
<jtojnar>
supersandro2000: what do you mean about multiple PRs?
<vika_nezrimaya>
like, a NixOS image boots, then curls 169.254.169.254 for a derivation it needs to deploy as system closure, `nix copy`-es it to its hard drive, `switch-to-configuration boot`'s it and reboots
<vika_nezrimaya>
it feels like a good idea
<vika_nezrimaya>
I don't know if it's a good idea
<vika_nezrimaya>
I don't know how to test it... wait, test... I could use NixOS tests for that! They automatically provision a virtual switch to which machines get connected, right?
<vika_nezrimaya>
I'm not sure if these machines have real writable /nix/store though
<vika_nezrimaya>
also I think I'm moving on topic of #nixos more
<hexa->
[root@dione:~]# dmesg
<hexa->
dmesg: command not found
<hexa->
wc
<vika_nezrimaya>
,locate bin dmesg
<{^_^}>
Found in packages: toybox, busybox, klibcShrunk, libuuid.bin, utillinux.bin
<vika_nezrimaya>
right, utillinux
<hexa->
same for ls, htop, etc.
<hexa->
the root fs apparently went missing
<vika_nezrimaya>
hexa-: are you on NixOS? is your /nix/store ok or did it fail too?
<vika_nezrimaya>
you might need a live USB...
<hexa->
rebooted and it is back up
<hexa->
it's an rpi4 with a sata ssd attached over uas
<hexa->
so, pretty sure the usb subsystem/controller went AWOL
<samueldr>
maybe the disk asked for too much power at once
<samueldr>
while power was already constrained
<hexa->
hm
<supersandro2000>
jtojnar: I like that
<hexa->
original rpi usb-c power supply
<hexa->
i have lan connected, a zigbee stick and the ssd
<samueldr>
yeah
<samueldr>
they also tell you to use a powered hub if you're using anything that requires usb power :)
<hexa->
brrrr
<hexa->
shitty.platform
<hexa->
unfortunately that tld does not exist yet
<samueldr>
it could do better if it didn't skimp for the power input
<samueldr>
e.g. it could use _actual_ power delivery to get the power it needs
<hexa->
aye
<samueldr>
or not use USB for power
<supersandro2000>
for file in logs/*.log; do nixpkgs-hammer "${file%.*}"; sleep 0.1; done
<samueldr>
but instead it's using the legacy usb interface for power
<hexa->
hence shitty.platform
<ajs124>
can't you just put a few amps directly into it thorugh one of the gpio pins?
<hexa->
I could
<samueldr>
I don't know
<samueldr>
but it's a bad design either way :)
<hexa->
^
<samueldr>
though since the 4th it's now not out of spec, since IIRC for type-c they increased the power a device can ask/receive
<ajs124>
not sure if that's break it, but you can probably do that at least one
<samueldr>
I think the GPIOs have fewer protections for power stuff
<samueldr>
not sure how it matters in that case
<samueldr>
and I don't know if it would help getting the power to the USB devices