#off-topic-tech
1 messages · Page 234 of 1
yup
it's also stupid because if it's installed through Steam, they already have the basic info needed to determine if you're at least 18
Well this whole thing appears to be asinine
it took em 17 minutes to even mention that detection depends on freequency as well
it should be just question of sensitivity and sophisticated filtering.
like, radio signals are fairly weak, but we amplify them and use clever tech to find specific signal in the sea of noise and other signals.
still a hard stretch to get human heartbeat...
on april 22nd, 2026, the Secure Data Act was introduced in congress. the act pre-empts every state privacy law in the country, replacing California's CPRA, Illinois' BIPA, and Colorado's CPA with a single weak national standard that explicitly forbids states from enforcing anything stricter. the reason this happened is simple: major AI companies...
people are ok with this
Over 50% of people are for the banning of phones for kids so ofc they are
its over
hiiiiiiiiiiiii
how is linux perf with nvidia 580 driver
for this game in particular
Don't know if anyone's tested, most anyone gaming on linux in here is on amd graphics
Mostly about the same as on Windows AFAIK.
But just general hearsay.
Nvidia linux drivers are still quite iffy compared to amd in my experience(using nvidia gpu with kubuntu for work)
had a weird issue today with a customers server
would POST normally but after handing off to any bootloader it freezes
didn't matter if installed OS or off a USB drive
secure boot issue ?
graphic hand off issue ?
SB was off
GPU was only one installed, also it instantly turned off when I pressed the power button
Corrupted bootloader?
they said it doesn't matter what boot loader
sounds like ram instability then, or acpi/pre-setup stuff corrupt
did you try booting tools like memtest ?
would make sense memtest being hardened against these kind of issues
This guys website seems to currently be struggling but it's an interesting writeup
www.thatprivacyguy.com/blog/chrome-silent-nano-install/
Possible bootrom is broken (e.g failed bios update?)
would be weird it crashes right at chain loading
Also, there seems to have been a significant infra outage of some sort this morning
Anyone have details?
I did a BIOS/UEFI update as a troubleshoot step but no change
Didn't try memtest specifically but my money's on a memory problem too.
Ventoy itself loaded fine, but then booting from a ISO froze it too
Anyway the server was 10ish years old and slated for a replacement anyway so we just do it earlier.
Gonna use another system as a backup host in the meantime
😭 i want to have issues
I have to admit. I watched some science videos from Kurzgesagt.
But this is ridiculous.
How clickbaity can you make a title.
Who?
? :))
Who is gerfew?
ger many
ger few
🙃
no worries
you know Kit is now part of my Ori cult
[obligatory Nigeria vs Germany "who the hell didn't pre-check the scoreboard before airing this" pic]
hey guys I want to reinstall windows for the first time in way over a year ... is there an ideal version to install these days?
Yeah, first you measure the wall and then you make sure you can install windows 
*jk
There is no ideal version afaik but clean windows install usually worked better in my case
People at work migrated their laptops from windows 10 to windows 11 and we had so many issues which could only be solved with a clean windows 11 installation
the latest cause if you go back enough they cant update themselfs
and then you will end up having to full install again
cool, thanks
truth is im pretty sure as long as your atleast on late 22H your good
- download official windows iso
- delete it
- download and install linux instead
🙃
iv been thinking about this for a while actually but what do you do if you want to go back to windows after swapping ot linux if you have no other device on windows?
since you cant download the ISO on firefox, and chromium and stuff obviously dosent exist on most ditros
???
download in FF works fine and in most mainstream distros there should be some kind of chromium based browser in the repos
hmm, let me think...
oh, right
always has been.jpg
sometimes i wonder who coded the 7th and 8th symbol to the sg1 computer.
oh it does work now
for atleast 2 years now trying to download the win11 ISO on firefox would just not work and detect you as a bot
Just learned a network engineering joke: "switchport trunk allowed vlan 30"
People who got burned will understand
so many jokes recently on noone explains
yea that one is too far for me lol i dont get it
port is set as trunk but only vlan 30 can go on it, and it's tagged ?
you know one does now simply be evil 🙁
Yeah, this is apparently common to Cisco
Since you didn't put "add" before 30, it means only vlan 30 is allowed while every other vlan is not
That's how people brought down the network by accident
An example of explanation: "For the uninitiated, there's a command to restrict a VLAN-tagged trunk to carry only specific VLANs. The command to restrict it to, say, VLANs 5, 6, 7, and 9 is switchport trunk allowed vlan 5-7,9. As you can see, you can use ranges or comma-separated values.
If you want to add VLAN 8 to the list above, you use the command switchport trunk allowed vlan add 8. What you definitely don't do is forget the add keyword and just type switchport trunk allowed vlan 8—because that replaces the existing list with whatever you included in the command. You could alternatively type switchport trunk allowed vlan 5-9, but who does that?"
"Nice" syntax from Cisco.
Sane design would have something like "set" as keyword for what they do without any keyword currently.
But sane and Cisco don't really go together.
www.ip.network/blog/de-domain-dnssec-outage-may-2026
Weeeeee, German .de TLD websites were unavailable yesterday evening
was a fun read to follow along as people figured it out 😄
any1 here know delphi, i refuse to elarn it and my friend needs help 😭
what?
you mean people are still turbo pascaling around
damn i grew up on pascal and delphi those were the times
and no i no longer remember ... tho help files are around
Has it? 🤔 I don't think you know what you are talking about.
Most titles. You know more or less what it's about.
Except the first one ofc.
Or atleast explain to me how "Tattoo Removal is Insane" is as clickbaity as "GERMANY IS OVER" (in capslock btw!)
"This is the scariest place in the universe" tells me more about the video than the Germany one.
And the german one is about the decline in population, but I didn't watch because I cba anymore with the channel.
2 red arrows out of 6
but anyway, was talking about about them being ridiculous and clickbaity for .... years?
perhaps more about how I think that their substance to 'presentation' ratio is 1/3 of late Veritasium
walk through their titles and covers and highlight every clickbait-ish word:
weirder
than you think
the most
insane
megaproject
you never heard about
the scariest
in the universe
super[voids]
insane
Ok, you don't understand me. And that's fine. No point that I keep explaining myself 🤣
you point is that they become clickbait channel
I say that well they always were
but it doesn't mean that they didn't become more clickbait
No, I said "how clickbaity can you make a title".
If we could put a % on clickbait, I would say it was 50% on most titles (so not denying they make clickbait titles, but most of them gave some insight into the topic). But the last one is nearing 90% (subjective).
Making a title full capslock with fire background, it's next level to me 🤣
is it logarithmic scale?
no, the letterdief scale.
oh, how to calibrate it?
Only I can do that 
It's my opinion. I know, it's subjective.
And I fixed it. By not watching the videos or the channel anymore.
I still think titles should provide insight into the topic and also provoke curiosity.
But no insight into the topic and full throttle on the curiosity/drama is a bit much for me.
But it's no news that YT content has come to this point. Most channels have to resort to this to keep relevant.
well...
"how gyroscope works?"
is only curiosity and no insight
but isn't bad by any measure
How is it not insight into the topic? It's about gyroscopes. And how they work. It's in the title? Am I missing something here?
well, in my book, if I say "tree" I don't explain how "tree" works
okay I probably understood you too literally
please don't judge by the cover
I think YT music thinks I am depressed?..
anyone on here that solders have any suggestions for home use hot air stations? I have a project with large ground planes so i need a high power station, but am skeptical to buy a clone, not to mention stations like the atten 862D look different in every listing i see online
i would never judge you
i am always happy being the evil one
Factory in my pocket, my dreams have been achieved
Crazy
Note: Post-quantum protection is being rolled out gradually in Proton Mail. If you do not see the option in your account yet, there is nothing you need to do. It will become available soon.
The quantum era is no longer a distant thought experiment. Quantum computers — advanced computing systems designed to solve complex problems that classical computers cannot — are not yet capable of breaking the encryption that protects email today. But the risk is real enough that security teams, standards bodies, and technology providers are already preparing for what comes next by adopting post-quantum encryption.
Post-quantum protection is now an optional upgrade in Proton Mail, available on all plans including free. Once enabled, Proton Mail can generate and use post-quantum-ready keys for new encrypted emails to protect your personal messages and business communications against today’s threats and a future where current public-key cryptography may no longer be enough.
ON MOBILE????
Yessir
Legit how- I am astonished
What hardware are you using?
Phone, probably
If I bought the whole phone, im gonna use the whole phone
Gotta love the 99% GPU
how in tf did you get a steam game to run on ARM lmao
screenshot is literally watermarked "Redmagic 11 pro"
because advertissing through screenshots is just the average android experience these days it feels like
oneplus does it redmagic does it
like bich i dont want your watermark in my pics
GameNative app, just log into steam, download and play
you cant play games on the steamapp tho 😐
It's using wine emulator to run windows, it's literally the desktop steam client
Lol u can turn it off I leave it on because a lot of people ask what phone, rather let the screenshot do the work of telling.
Alternatively: Steam Link, Parsec or similar remote desktop technology
I like the games to be running native on the device, streaming takes way to much data when I'm not home
Iv never gotten steam link to work on mobile it only works pc to pc for me
Steam link also only works on a local network, if if 300 miles away from my pc and it's off, it's useless
Works across different networks but they have to support it
Like i used to use steamlink on my school laptop to save on battery and that was with one pc on a network and the other on another
I guess you can, but I still don't wanna have to keep my pc on all the time, it's a 1200watt pc, it adds to the electric bill lol
The same way valve is running games on the frame
fex+proton
Please tell me you're just referring to your PSU's wattage and not the actual power draw of the PC
cuz actually drawing more than 1kW on a desktop PC is a tough thing to achieve
i draw around 800-1000 when gaming. overclocked 7900xtx + overclocked 14900k + 3 monitors and a 60 inch tv, measured at the outlet using a volt meter. pc is prob around 600ish of that power
Wait. I have questions.
But I might be dumb.
Monitors and TV, they don't draw power on PC do they?
no, but they are used for using the pc
All 4?
yup
Damn. I turn of my monitors and screen when not in use etc 😄
oh ofc i turn off when not used lol
I leave mine on nowadays
Can I ask what you do with all the monitors? What information you put on them?
since the button is behind and with multiple monitors forget it
Or is it 3 monitors acting as "one" ? Not sure what the technical term is. But I've seen it on the internet sometimes.
Ultrawide?
Something like that.
Making a bad tv out of monitors like that seems so inefficient
I agree.
And the edges of the screen would annoy me.
the 60 inch tv is a 4k 120hz as my main, 1 ultra wide for youtube and or other media, the other is used for satisfactory calc, and the other is for discord and communication
My bad. I was walking home when looking at the image
Can't see shit in daylight
So cursed. Also makes screensharing impossible :)) (trust me we had one dude at a previous place that did this...)
i have to say my setup is a lot more efficient that i would have tought
3 main monitors ethernet switch + pc and peripherals only 700W peak
Not bad
not gonna lie i was sure it would break 1000 but i guess not
with both my pc's definitely but i expected just my main to blow 4 digits
how old is it now?
i have that one HDD from 2013 which is probably the oldest hardware i own now
that still works anyway
i still daily a HDD but its getting rough now that Win11 stops disks from spinning if they arent in active use so anytime something pulls from it it crashes for 10 seconds while the disk spools up
i still feel like a lottery winner every day with this thing
Disable spindown in BIOS or win power options.
But w11 letting disks not spin down by default sound like a bug, because even back with w10 it did that
Or maybe it did, but MagicZ didn't notice until the HDDs dying got worse.
He has known about it for 1+ years, but why do anything until it fails completely?
it’s something that started from a windows update for sure so i doubt its anything i did
I can’t tell if im seeing things cause its 4am or because this controller is all kinds of wrong
I'm kind of at a loss rn, hopefully someone here can help me out.
I've recently got myself a new laptop off the used market (Fujitsu Lifebook E5511) and put CachyOS on it.
And for over a week now, I've been struggling to get it connected with the free wifi at work.
Connecting to the wifi network is easily done, but before receiving internet access, there's a captive portal to accept the terms and conditions which just won't open. This site is hosted on hotsplots.de btw, so not a locally hosting captive portal
It works perfectly fine on my phone and the laptops we use at work, so I know that it's not the network itself.
Things I've tried:
- disabling ufw
- disabling DNS over HTTPS in Firefox
- disabling HTTPS-only mode in Firefox
- opening hotsplots.de in Firefox, both https and http
- getting help with my system time settings, see below
The only thing I could find out was a difference in timedatectl status between my home wifi and the one at work.
Result at work can be seen in the image, result at home is this:
Local time: Mo 2026-05-04 23:55:28 CEST
Universal time: Mo 2026-05-04 21:55:28 UTC
RTC time: Mo 2026-05-04 21:55:28
Time zone: Europe/Berlin (CEST, +0200)
System clock synchronized: yes
NTP service: active
RTC in local TZ: no```
Does anyone have any ideas on what else I could try or how to get this captive portal to open?
Because that line needs working internet to be yes.
So it is just result, not cause.
That turns to yes when timesyncd has NTP connection to internet time server.
Seeing what the WLAN IP settings are while needing to open that site, before accepting.
IP and routes.
IP is usually some 192.168. thingy with a /22 subnet mask
If I get a free minute, I'll set up the laptop and see if I can get you those infos in more detail.
How do I check routes?
"ip route"
And also DNS server(s).
Can that domain the site lives on be resolved.
Limited number of things that could be wrong, in sequence where they matter:
DHCP not working, in which case you wouldn't get right IP, subnet, DNS server etc.
ARP (local network segment) & Routing (other network segments) not working, so you cannot talk to anything.
DNS not working and you cannot convert the domain name to IP for connecting.
The actual connection to that IP not working.
With local firewalls being able to cause issues with any step if too restrictive for outgoing traffic or related incoming traffic.
And remember that any manually configured DNS servers aren't working at that point either.
Need to use that DHCP given one, or manual mapping in /etc/hosts file.
And here we go again: https://thehackernews.com/2026/05/linux-kernel-dirty-frag-lpe-exploit.html
Might be time for me to make a list of all linux kernel modules, remove the actually loaded ones, and blacklist the rest...
Somebody pushed (part?) of fix into public kernel repo too soon.
Output:
192.168.44.0/22 dev wlan0 proto kernel scope link src 192.168.45.152 metric 600```
So there is default route via that 192.168.44.1 router.
I've actually been there before I just realized
When I go to 192.168.44.1, it forwards me to the hotsplots.de website that usually shows the captive portal, but for me I just get this error message (using a pic from a different website cuz it was easier to find this one than to take a photo of the exact same error message on the portal website)
Oh, and in case it matters, I also did this as part of my troubleshooting endeavors
And next thing was to see if DNS is working and that hotsplots.de resolves to an IP.
Try pinging it or like, where you get specific failure for hostname resolution failing.
ping: hotsplots.de: Name or service not known
So DNS is not working.
So now you would need to check what you have configured currently as the DNS server(s).
Which depends on what DNS client software is in use.
How do I check that?
What distro?
And /etc/resolv.conf might help determining that.
CachyOS is by default set to always use Cloudflare DNS, which doesn't work until that other thing has been passed.
So either need to change that, or add right /etc/hosts entry.
search hotspot
nameserver 127.0.0.53
options edns0 trust-ad```
So NetworkManager is configuring that, and there is local DNS proxy in use.
Does "resolvectl" output anything?
Because that would suggest that SystemD ResolveD is in use.
But the DNS settings are inside NetworkManager.
Check that WiFi connections settings in there for DNS.
So ResolveD is in use, with external OS config.
With right DNS gotten via DHCP, but also bunch of other static ones configured.
Moment.
Does "resolvectl query hotsplots.de" return any results?
And check /etc/systemd/resolved.conf and /etc/systemd/resolved.conf.d/ for configuration of that fallback DNS.
hotsplots.de: Name 'hotsplots.de' not found
There is no resolved.conf.d
resolved.conf is basically empty, it's just a bunch of comments and one line being [Resolve]
Ok, so it is getting those fallback DNS servers via some other way than its own config files.
So ResolveD is jumping to the fallback servers too quickly.
So need to either get rid of them, or add that response into that /etc/hosts file.
You can try putting fallback.conf with following contents into that /etc/systemd/resolved.conf.d/
FallbackDNS=```
And then "systemctl restart systemd-resolved.service"
And then check with that resolvectl if the fallback entries are gone.
Well, the name doesn't matter as long as it ends in .conf and is in that directory.
What does the .d mean?
There is only resolved.conf and no resolved.conf.d
config directory.
For users override configs.
Just mkdir /etc/systemd/resolved.conf.d/
Huh. Funny
Yes, they are in there as example, but commented out.
CachyOS is setting them to that somewhere else.
Having config with empty assignment in that .d/ should override any previous assignments, unless it is being done at runtime via D-BUS API.
For that, you would need to look into inside Network Manager if being done.
SystemD stuff usually has multiple levels of config options.
Starting from compile set defaults
Then bunch of different config files, for that ResolveD:
/run/systemd/resolved.conf
/usr/local/lib/systemd/resolved.conf
/usr/lib/systemd/resolved.conf
/etc/systemd/resolved.conf.d/*.conf
/run/systemd/resolved.conf.d/*.conf
/usr/local/lib/systemd/resolved.conf.d/*.conf
/usr/lib/systemd/resolved.conf.d/*.conf```
And then D-BUS API for runtime changes.
Uhhhhh
It didn't work? I think?
So you added that contents?
And then restarted resolved?
You are just creating an empty file in that and not restarting.
Ah, sorry, the restart was there..
And the file was filled with this exact text
Ok, then some later point is adding them.
That was just to override those compiled defaults and the config locations before on that list I put.
So check inside that Network Manager then.
Possibly this place:
That was resolvectl query hotsplots.de?
Or just check that basic "resolvectl" to see if there are still any fallback DNS entries or not.
Nothing in the Other DNS Servers field
The global Fallback DNS Servers are gone
Then check if the page works now.
Because those two DNS check commands I gave were ones that forced use of specific DNS server, which was the DHCP given one.
That ResolveD wasn't using for some reason while the fallbacks existed.
Rest is the same
Nope
So "resolvectl query hotsplots.de" doesn't work?
Nope, still doesn't work
Still Name not found
And just "resolvectl" still shows that 192.168.44.1 DNS server in the listing, and no others now?
Yup
Which is just bizarre, when direct queries to that server do work...
It's the same output as this, but with two changes:
- The Fallback DNS Servers line under Global is gone
- Link 2 (enp0s31f6) and Link 3 (wlan0) have switched places, now wlan0 is listed last
And DNS works when using some other connection?
To bypass that DNS for now, add following to /etc/hosts, amount of whitespace doesn't matter:
185.238.219.194 hotsplots.de
Still don't work
I can browse the Internet just fine when I'm at home
Which is wifi to Fritzbox 6490
As in ping doesn't work, or doesn't work in browser?
Forgot to check, because current browsers also allow to force DNS inside their own settings now.
Holup, ping does work, tho browser still doesn't
So it is now that browsers own DNS settings thing.
I've searched for DNS in Firefox settings and managed to get all results into one screenshot this way
So ping works, but resolvectl query doesn't?
Or does that resolvectl query now work too?
Query works too now
Try restarting that browser completely, or rebooting?
I'll just reboot
Resolvectl and ping still work, but opening the page or resolving/pinging other URLs doesn't
Opening 192.168.44.1 in Firefox also doesn't work
Which is just weird, because that default OS DNS resolver is set to be the stub-resolver of ResolveD, which is now resolving that specific IP because of that hosts file...
as that resolvectl query works.
Yeah, other DNS queries aren't expected to work, you would need to see with those DNS server specifying commands if that DNS server responds to anything else right now.
See nslookup above
Didn't specify that DNS server to use like earlier.
To bypass that local stub-resolver.
*above the resolvectl commands in this screenshot
nslookup google.com 192.168.44.1
So ResolveD isn't using that 192.168.44.1 for one reason or another.
Point was that that one IP is gotten from that hosts file, so it works.
You can try.
But no idea what is going on, or how to diagnose why ResolveD is failing like that.
When it does list it as one of the DNS servers available.
Did you configure alternative DNS in cachy hello?
There was by default, overridden and off.
And confirmed that resolvectl isn't listing anything but that one DNS server.
So the DNS server given by DHCP responds when queried directly.
But SystemD ResolveD isn't querying it for one reason or another, even when it is the only DNS server it lists as being available.
What even is 127.0.0.53?
One of the IPs for local computer.
That is IP that that SystemD ResolveD caching stub-resolver listens on.
So that queries that local programs run, hit ResolveD, which then queries the actual DNS server.
that resolvectl is the interface to ResolveD
127.0.0.1 is the most common, but the whole 127.0.0.0/8 is localhost.
Ah
There's no place like 127.0.0.0/8
So we now know that queries to that DNS actually work but somehow, ResolveD just decides not to
Yes.
ResolveD answers to any queries it can.
But it doesn't make any queries to that DHCP configured DNS server when it should.
Direct queries to that DNS server do work.
But ResolveD is dropping the ball, hard, on the local computer.
So either ResolveD just doesn't see the queries on 127.0.0.53 (but why would my home wifi work then?) or it does see them but doesn't process them through 192.168.44.1
It does, when those queries not via resolvectl work right to the hosts defined name.
And resolvectl query bypasses all that, and directly tells resolved to do a query with its current config.
Stub-resolver not involved.
I got another idea, I could launch a mobile hotspot on my phone and connect the laptop to that
CachyOS seems to have some DNS settings in "Hello App"...
Which people complain about.
But AFAIK that fallback.conf bypassed that.
Yeah, ResolveD dropping the ball.
Or whatever state combination of CachyOS, NetworkManager and ResolveD are in.
Because NetworkManager is bypassing parts of ResolveD etc.
Normally ResolveD would be in control of that /etc/resolv.conf, but currently for you, NetworkManager is.
But just has put the stuff that ResolveD would put into there, but not as the symlink that ResolveD would do, so it is in that "foreign" mode.
But basically configuration state I have no idea about, on distro that I have no idea about.
I do use ResolveD, even when it isn't Debian default.
Still massive thank you, I've made more progress today than in my previous attempts
Thanks
What I am currently wondering: why did it work with my home wifi?
But why can't I do so on my phone's hotspot?
Because I've overwritten them?
You can try if they work now via the phone.
Or connecting to other DNS can be blocked at intervening firewall.
So comment out that fallbackdns override and try with phone?
I've tried nslookup with the DNS server that is listed here for wlan0, it worked.
I've tried nslookup without, it didn't work
Yeah, first bypasses everything DNS related on the local computer and does direct query to specified server.
Second uses the local config etc.
Okay, I'm starting to make sense of the commands
Tells that the remove DNS server works, but something broken at the local OS level.
And that resolvectl query doesn't work rules out some local issues.
Should I also try removing the comment here?
Or wait, no, it doesn't matter, it's configured elsewhere
If you wanted to force those, copy the line into that .d/fallback.conf instead.
Editing the "base" configs causes issues at package upgrades sometimes.
So better to use that .d
Nah, don't wanna force em, I just forgot that CachyOS would enable them automatically if they weren't overwritten
So I thought I would need to enable them manually
And remember you can diagnose if you can reach those other DNS servers the same way with nslookup or dig, just replacing that DNS server IP with whatever.
1.1.1.1 for Cloudflare, 8.8.8.8 for Google etc.
Dafuq
I did the ResolveD restart after commenting out the fallback.conf, the two nslookups before it were done with the override still active
Bruh, phone turned off my hotspot
Try that nslookup <something> 9.9.9.9
Turned it on, connected laptop to it, ran nslookup again without specifying DNS server, got nxdomain again
works
So something broken at local level still.
Probably related to NetworkManager and that work WiFi AP interacting.
But I have no idea about NetworkManager, never used and always stayed away intentionally.
Personally using SystemD NetworkD (again not Debian default).
But it is the one common GUI side interacting tool for Linux network configuration.
Because something has broken on that local DNS stack when it connected to that work AP, that persists even after connecting to something else.
Deleting the config completely related to it, and rebooting, might help.
Do note that here you said you were on phone hotspot, but were still able to directly connect to that work AP set DNS server on private network.
So the computer or phone was still connected to that work AP.
The phone was in between, because those resolevctl DNS entries were for the phones caching DNS proxy.
Wait, yeah, true. It shouldn't have worked
But because I use Debian & NetworkD & ResolveD combo, you use Arch/CachyOS & NetworkManager & ResolveD combo, I don't have enough idea what might be going on.
From whatever messing that NetworkManager does, as I have no idea about it.
Just that for stuff that it touches, it has to be in control or stuff just doesn't work.
And it touches basically everything networking related when in use.
Hm, let's see. I've removed the OpenWifi connection from my laptop and am now rebooting it
Okay, yeah, it just works now
On my phone's hotspot
Heh, I still had the T&C webpage in my browser history and just accepted them
So as long as the OpenWLAN is configured, stuff just breaks
So something that that work WiFi DHCP returns as settings is most likely confusing your stack somehow.
But no idea more closely.
Those Arch pages might contain clues, but don't have time today to read through.
The general Arch DNS page, Arch ResolveD page and Arch NetworkManager page.
But cannot read through first and third today, and IIRC second doesn't contain anything relevant over what I already said (IIRC I read through it when switching to ResolveD myself).
All good, I don't expect you to.
It's quite high level stuff for me, but I've narrowed it down a whole lot now
Remember to remove that /etc/hosts line too.
Correction: it being configured is just one part of the puzzle.
It being configured and available is what's relevant.
Tho idk if I've restarted the laptop with it connecting to my phone's hotspot
But something that NetworkManager does when configuring for that work AP breaks DNS server usage inside ResolveD.
As the working hypothesis at this point.
Which stays until probably reboot or resolved restart (the restarts didn't help as NetworkManager just redid the configuration when it was there)
But I have to head to grocery store now.
I did connect to my home wifi after configuring the work wifi, so it's not just about the network being configured
Aight, thanks for all the help
Don't feel the need to answer btw, I'll just be using the chat to keep track of some of my tests
But you probably turned the machine on between, and didn't move it while it was in sleep or like.
Yeah, I shut it down at work and turned it on at home
Okay, so rebooting while connected to the phone hotspot also results in the internet working
Which again points to that the config existing doesn't matter, but when the computer actually connects and gets that DHCP data, NetworkManager messes something up in ResolveD.
Which lasts until ResolveD is restarted without that AP being connectable.
So when I connect to work wifi, something breaks inside NetworkManager (*or ResolveD).
And it only goes back to working after rebooting the PC
One test you could do would be:
Connect to work AP.
Disconnect and delete the config.
Connect to phone AP.
Test that stuff is still broken with DNS.
Restart just systemd-resolved.
See if DNS works now.
To differentiate between NM keeping broken stuff, and ResolveD getting broken by NM.
Tested some more stuff.
Rebooted with phone hotspot connected and work wifi configured, stuff worked instantly.
Connected to work wifi, loaded a webpage, didn't load.
Connected to phone hotspot again, loaded another webpage, worked.
Have fun
Connected to work AP.
Removed work AP network (deleted the config) while connected.
Connected to phone AP.
Pinged muensteresports.de, Name or service not known.
Probably just had the other sites cached then
It gets weirder
Connected to work wifi, shit broke.
Disconnected from work wifi, connected to phone, removed work wifi config, shit still broken.
Restarted ResolveD, shit still broken
Wtf, it's getting really strange now
Deleted OpenWLAN config.
Rebooted.
Automatically connected to phone wifi.
DNS still broken
wtf
This is straight after reboot
Rebooted again, works without issues now
It just won't make sense
Next point would probably be NetworkManager logs, and ResolveD debug logs.
But no idea where first would be or if there would be anything usable.
And the default level on which ResolveD logs doesn't probably contain anything relevant, and don't currently know/remember how to make it log more.
journalctl is probable place for NM logs, but not sure.
That is the systemd common logging system, and that is where resolved logs to.
Also, to earlier.
That NXDOMAIN is return that means No such domain.
So it isn't that ResolveD isn't able to connect to a DNS server.
It thinks it connected and got response that says no such domain exists...
If it couldn't get response, it should respond with SERVFAIL instead.
But as the actual direct queries to that configured DNS server did return valid responses...
But might also be partially too aggressive caching of negative entries in ResolveD.
To disable:
Cache=no-negative```
Or just "no" to disable local caching completely.
resolvectl show-cache allow to view current cache.
resolvectl monitor allows to view the queries hitting that stub-resolver at real-time.
I've decided it's weekend time for me, already shut down the laptop and getting ready to head home
If the logs persist, I'll check them when I'm back.
If not, testing will probably continue on Monday
And that resolvectl has options to override DNS server etc. settings.
resolvectl --help
https://man.archlinux.org/man/resolvectl.1.en
I've got a macintosh Plus
If you can get satisfactory running on it you can have it
why are you @'ing me?
dis idk if its true sorry
Should i ping moderator
Lol idk
i'm not a mod

<@&387163995947270144> idk if its true can you see if it is please 🙂
Im still not sure why your pinging a mod but ok
its very true
how about giving the link to the part list then? 🙂
im not gonna reccomend any PSU's just use the tier list is all il say
I wonder why people do this.
Ask for help and just intentionally withhold information that would undoubtedly be relevant to the helpers
i more hate when they ask for help and then disapear off the face of the earth
i think it is the "can i ask a question?" syndrom already asking a question 🙂
lowkey fire website
@sharp matrix 29.99 is pretty optimistic but im hyped anyway lol considering anything releasing anytime soon is 80
Thats the announced price🤷♂️
but also don't buy it
Im going to buy it and you cant stop me mwhahaha
Yea lowkey secretly hope for it to be no mans sky 2 
my issues with it is that i want it to be good enough to buy. when it's complete. but also, i want the devs out from under Kratom so that they don't have to send them a dime after the shit that went on
Oh krafton, i have words for that but it would probably get me banned
there's a reason i didn't use their name
Where that second isn't going to happen, as Krafton still owns the studio, and there are no court plans to try to change that.
Just force return to the old studio heads having the operational control like the original agreement stipulated.
sometimes I am blown away by how fast pandas can be
othertimes I use profanity when it eats memory without any control
"eats memory without any control"
you're talking about web browsers right?
not even close
wish i could easily yank the battery out of my "dead" mouse and just plug it into this one for some stupid long life
Hi everyone,
I’m looking for someone interested in helping with an experimental UI/UX prototype.
The ideal person has experience with Unity, interactive interfaces, spatial UI, tool design, or polished prototypes. I’m looking for someone curious, self-driven, fast at testing ideas, and comfortable working with early-stage concepts.
I’m keeping the project details broad for now.
If you’re interested, please DM me with a short intro: who you are, what you can do, what tools you use, your realistic availability, and any examples of your work.
How unsafe is 246V ac? For PSU
I don't like the fact that I heard faint high-pitched noise like someone was screaming but from within my pc
Check the specs if your PSU is rated for it
i mean generally there only speced to 240V so your gambling but for short periods ehh
it is not
thats quite a bit over tho so thats a dodgy outlet level issue for sure
outlet?
more like power grid of entire country...
thats when one of those battery thingies comes into play UPS's i think there called?
we have 120V here and rn my thingy is reading shifts between 120.6V-121.3V
after second power shutdown,
voltage dropped a calm 228V
and frequency, for the first time in last couple weeks (of my infrequent measurements) is a nice clean 50.0Hz (and not 49.9(
so a whole 6V over is a lot
they don't like being used that much... sadly
in fact, the ones I have are more picky to the input than PC's PSU
they jsust don't charge if V is below 190V
i should probably get one of those but im too lazy and batteries are expensive
the shortlasting quiet period has ended...
If you get online UPS, their inverter will fix any dirty voltage or frequency input into nice and healty output for PC and other sensitive equipment
For today
Not that input voltage in range 227-232 would ne anyway noticeable
Or bad in rural area
terrible discord change i just noticed
you can no longer search for messages by a speciffic person
as well as the other ones but thats the main one i recall
oh wait nevermind im stupid
streamermode being on for no reasson as usual
Nice they broke search again 
Works fine, MagicZ did say that the cause was streamer mode being on for some reason.
Ah, on mobile app?
I was just testing site version.
Works on mobile but it didnt work on desktop when I tested this morning.
Well turning it off brought it back but they have been messing with it lately so it wouldn’t surprise me if it was broke
But yea no mobile works i can confirm that
This a good glossary? (For a home server user/installation manual)
All my usb devices keep disconecting at random (on both my desktops) like montly and everytime it throws this warning in event viewer at the time but i dont really get how im meant to do anything about that
I would reccomend checking USB power saving/sleep settings
USB bluetooth chipset doesn't support Bluetooth LE.
Or otherwise errored out when it should have replied with capability mask.
Seems like it errored out, as the mask is wrong length and all 1-bits.
More likely to be result of whatever is the actual issue than the issue itself.
That system USB functionality glitches, and that is just error from during that glitch where the USB BT driver cannot talk with the BT chipset, or from when the functionaly is starting to recover, but the chipset is in buggy state.
And in case you are confused, most combined WLAN & BT adapters have WLAN via PCIe and BT via USB, even when they are M.2 cards etc.
Which is just weird in my eyes when both are in single chip.
so something actually broke and its not just a software issue?
iv had back io issues with the ethernet not working at all if you remember so starting to have issues with usb more and more frequently doesn’t surprise me much
AMD has responded to what gamers have been most vocal about: bringing its FSR 4.1 upscaling technology to older RDNA 3 and RDNA 2 GPUs. Today, we learn that this dream is becoming a reality, as AMD is officially introducing its FSR Upscaling 4.1 technology to millions of Radeon RX 7000 Series graphi...
No idea why mixing 7000 and 6000 in the headline together:
Finally, FSR Upscaling 4.1 is also coming to older RDNA 2 GPUs, but that port will have to wait until early 2027
I found one thing AI is EXTREMELY bad at: electrical installations
Example:
Should have become an electrician, AI will never replace that job considering the shit above lol
The first thing that caught my eye is grounding, neutral and live wires connected to the same terminal block on the right
There are also some dangerous mistakes too besides the terminal block
cant break if it aint connected
Ai has several mental problems when talking about warthunder 😭
It’s BAD
its bad at games in general
its bad in general
still probably safer than our north american jerry rigging
it talks about snail?
It says uh let’s see
Yeah a 20mm cannon can frontally penetrate 800+ mm of armour
Current LLMs are bad in anything that needs to be factual and the answer isn't VERY common on internet data it was trained on.
It's a linguistic predictor.
Its primary objective is to sound as much as a human as possible
LLMs are just pretenders, they look at a bunch of word data and try to guess what the next word in a sequence will be
It's like those "what number comes next?" puzzles but a lot more complicated
It's like going up to a random guy on the street and asking him about stuff, except he doesn't have the capability of realizing he doesn't know the answer but he DOES have the ability to come up with a very convincing answer on the spot
Ultimate realistic sounding bullshit generation algorithm.
Which for things that are old enough to be included in the training material, and common enough in the training material, with specific answer that hasn't been swamped by wrong answers in the training matrial, can give usable answer.
But if you know enough to be able to differentiate between right and wrong answers, you don't need to ask the LLM.
I basically use it as a machine for "I can't think of the right solution rn, give me a dozen possible solutions and I'll recognize the correct one"
Which rarely happens to begin with but even in those scenarios, it often just doesn't work
I've tried using Copilot once to bulletproof an argument I was preparing but it ended up just agreeing with me on everything, so I got basically nowhere.
The idea was for the LLM to throw random shit at my reasoning to try and disprove it so that I would be prepared when others will try to argue against me. Didn't work at all
It's use should have been restricted to "search" from the beginning.
It can search for stuff thanks to more sophisticated embeds than usual indexing.
But companies wanted this hype and made it in form of chat
Using LLMs for search is dumb, there's better ML tech for that.
The point of LLMs is to produce text that's indistinguishable from human writing (or human speech in general).
That's what they are built for and chatbots are the most obvious usecase that's more complex than a basic text generators
The term "AI" has been buzzworded so much that people started confusing the different technologies that make up the field of Machine Learning.
And with LLMs (especially ChatGPT) gaining a whole lot of popularity, people wanted to use this "AI" stuff for other things they wanted automated - except they lacked the proper knowledge to realize that slapping an LLM onto something doesn't make it automated
And that's actually where most of the issues lie when people want to put AI into something.
Instead of asking themselves "how do we best automate this?" they try to squeeze an LLM into it in hopes that it will produce the desired automation - because they just don't know any better than to assume an LLM can do everything because LLM companies advertise their products as if they were all-capable
Well, that and the fact that generative models primarily produce slop and can't really create anything of good quality like a human could
Can you use the other ml tools to look for "what people are saying about X"?
I can use Google for that, just extend the search prompt with
site:reddit.com
Jokes aside, yeah, obviously. You can train a search model to value random people's comments higher than authored articles or an official web presence.
It actually makes less sense to use an LLM for that, because as explained above, LLMs are generative.
They don't search, they predict.
This is exactly what I mean, assuming that an LLM is the solution, then trying to make it work somehow.
Instead of taking a step back, looking at the problem and asking yourself "what's the best way to properly solve this?"
LLMs are a tool, and by themselves - not very useful.
Yet in tandem with other tools, like in RAG - it's a good way to improve search as a feature.
The difference is that LLMs encode tokens (words) differently, in a much better way when search engines did.
Llms are sensitive to context.
You can make a "search" in a form of question, where meaning (usage) of other words will change what the tool is looking for.
You could say that search engines are predicting what you were looking for by pulling entries that were indexed with same 'tokens'
Just that out of 2 approaches to encode, "understand" and predict desired outcome, one is more deterministic, and one uses richer embedding,
Also it's kind of calming that LLMs are going to doom themselves the more they are used for writing text that gets puplished on the internet and fed back into them
The amount of articles I read weekly "X company fires 25% to be replaced by AI".
Last one I remember was Gitlab.
And in many cases then tries to rehire most back 6-12 months later, for the earliest ones.
Yup.
Another one was a dutch company, called Bird (20% replaced by AI).
I wonder if this is the same as the industrial revolution.
Finally got rid of iPhone and switched back to Android
Got Galaxy S25 for only €200 through company although the company owns it for the first 2 years but I can buy it for a small amount after amortization
gratz!
I've been looking into the fairphone and linux OS.
But it doesn't support my bank app 🙁
Or maybe it does, but that's what community says about it.
and the topic is updated weekly
android is like trading one evil with another (google vs Apple)
Good thing modern Android devices are now smooth and fast, including budget series like Galaxy A
I remember how laggy Android was 5 year or more
Don't want to even mention year 2007-2010 where you had to use microsd card because internal memory was not large enough for apps, those were fun times lol
Back then, iPhone was actually worth it compared to Android which was in early stage of development
I think iPhone is still better privacy/security wise. But it has other issues.
Yeah, some iPhone features are driving me insane
There are no non-Android based OS options for Fairphone?
Older ones have unofficial Sailfish ports and Ubuntu Touch as option.
Apple released update for 10 year old devices (devices that were EOL), because FBI got access to data because of notification bug or something. Apple fixed it. Kudos.
Euh?
If I can't sell my iPhone, might as well send it to you since I don't want it lol
Difference is that that Android is with google apps.
And the other is Android fork without google apps.
Fairphone 6.
It's a bit more complex than that.
Yes, like I said, LineageOS forked from mainline Android, and that /e/OS forked from LineageOS.
I mean... so it's not android?
There are many forks. But you don't name them the same as the original?
More like distribution of Android.
Still Android base, just with some higher level stuff swapped.
The base is the same, the add-ons have been changed from Google stuff to other things.
It's focus is security and privacy?
So I wouldn't call it android. Which is google infested?
Same Android apps don't even work on it. So it's not that simple.
Not really.
Base Android doesn't contain most of the Google stuff.
Google sets strict limits on you when you want the Google stuff in the phone.
And that the apps don't work is because they expect those extra Google APIs.
And telemetry should also be nuked.
I don't trust google on their word.
And the fairphone itself has other benefits too.
There is the mainline open source Android.
Then there is the "Android (c)" with Google stuff in it.
And then there are forks of that mainline Android.
My point was just that FairPhone 6 can only currently have various flavors of Android as the OS.
Because you seem to think that "Android" means the version of the OS that includes Googles closed source app package.
That they use to control phone manufacturers.
When it is just (almost must) add-on to the actual OS that just provides specific apps and their extra APIs.
Android is an operating system owned by Google which is based on a modified version of the Linux kernel and other free and open-source software,
So with your logic, it's actually linux then?
When that is just the Google "distribution" of Android.
Which is why I talked about the mainline open source Android, which is the base.
And that one is not owned by google? Or heavily steered by them (contributions made by mainly google?)
In a way, but also not.
But point is to differentiate between that by itself.
That + Google closed source stuff.
And forks of the base OS, where the actual OS code is usually not much changed, the various forks just edit small parts and then add their own stuff on top to fill in for those missing Google closed source bits.
Like that /e/OS having MicroG inbuilt to provide replacement for some of the Google closed source APIs etc.
LineageOS 23 and 23.2 are for example based on Android 16.
Linux distributions and their fork trees are very close approximation about what is going on.
Well, I can live with that. I just don't want anything big tech anymore.
Just with much more commonality on Android forks.
but like I said, my bank app wont work on it. So I am stuck with Apple.
I just don't fully understand how far you want to go back.
You say E/OS is Android based. Which is linux based?
So why not say E/OS is linux based? How far do you go back in the fork tree?
Android is very customized Linux kernel + lot of very specific user mode software on top that is common to all Android forks.
Where that user stuff has very little commonality with general Linux userland.
So Android can be seen as Linux distro, with its own fork-tree.
But it is way more separate from most other Linux distros as its own ecosystem.
Including all the various subdistros like LineageOS, /e/OS etc.
Which are much more like for example most Ubuntu forks that just replace desktop manager when comparing to base Android.
As there have been other Linux kernel based mobile OS attempts, Nokia Maemo, Meego, whatever the Intel one was called, Ubuntu Touch, SailfishOS etc.
That don't share the user mode and GUI layers of Android.
Think about Chromium and Chrome.
Chromium is the Google managed open source project that is the actual browser.
Chrome is the Googles own version that is Chromium + stuff.
And then there are plenty of other browsers based on Chromium and forked from it.
But keeping up to date with that mainline base.
And not complete forks where the codebases start to diverge over time.
But fork where the fork tries to keep in sync with the mainline, just with limited patchset on top of things that mainline isn't willing to take in or other similar reasons.
Android Open Source Project (AOSP) (and forks) vs. Android© (with Google closed source stuff in it)
I think even discord runs on Chromium.
Oh, sorry. Confusing it with electron.
But Electron uses Chromium?
Electron App Framework, based on Chromium.
Basically JS app running in dedicated Chromium.
So yeah. How far down the rabbit hole can we go 😅
Electron just packages Chromium and NodeJS and possibly other minor things into easy to deploy package to use for locally running HTML+JS "native" app.
This is only the beginning https://youtu.be/E_I_N_ODzmA?is=IOBZi-fqpbaw6oQZ
A new report reveals NV Energy, which supplies power to some 50,000 residents in Lake Tahoe, between Nevada and California, could soon be without any electricity after the company announced it's pulling all its wattage and giving it to a cluster of incoming data centers in Northern Nevada. CBN's Tré Goins-Phillips and Raj Nair discuss the lates...
Our town major recently raised water bills by 50-70%. Imagine if there was a data center nearby
Basically, water will cost as much as electricity here which was unthinkable 5 years ago
My bad: Lake Tahoe residents will not loose access to electricity but they need to find a new energy partner which almost certainly means higher prices
it's a way to nullify a contract without fault so no penalties or lawsuits will show up
my bet is that the datacenter bribed someone to get the old elec rates locked in for themselves, and now the power companies get to absolutely screw over the homes in the area with fresh negotiations
Krafton CEO must be so pissed right now because of Subnautica 2 excellent sales 🤣
Wouldn't have bought today, as early early access title, without the fuckery and reversal.
Thing is iv put in about 12 hours 90% of which has been multiplayer and at no point has it felt like EA it feels done
damn, two million copies old in less than 24 hours.
I guess "Bad Publicity is Still Publicity" holds true.
there was bad publicity?
Subnautica 2 got a lot of coverage with the lawsuit but I can't think of anything that I would consider to be negative towards the game or its devs
only positivity towards game and devs and negativity towards Krafton's CEO
Better pull out the wallet and prepare that $250 million bonus
I wouldn't call it possitive either.
A gaming CEO asked ChatGPT how to avoid paying a $250 million bonus. It didn’t work
Some dude trying to fuck over the developers isn't possitive news in my book.
And the court descision was possitive news.
it's not positive news but it's not bad publicity for the game either
because none of the reporting presented the game or its devs in any negative way
Firing Ted Gill isn't good publicity.
and even the proceedings and result of the court case don't paint the devs and the game in a bad way
it's mainly the CEO of the parent company that received bad publicity
And I still think the whole ordeal, influences the popularity/sales.
And I don't mean that in a bad way. Just an observation.
yeah, all good
it's just that the saying "there is no such thing as bad publicity" refers to the fact that even when you report negatively on something, you're still reporting it which gives it more reach than it would've gotten otherwise - which ultimately means people hear about it who either don't care about the negative aspects, miss it or it gets omitted from them
in this case, "no such thing as bad publicity" would apply to Krafton and its CEO if they somehow end up being better off because of the media coverage
which they might because Subnautica 2 doing well still makes them money
you know what would be the absolute cherry on top to krafton on top of them losing 44$ per sale of subnautica 2
unknown worlds using the 250M bonus to seperate from krafton by buying themselfs back
the problem is that krafton won't accept it, they'll set the price too high for a buyout
would be nice to see tho
New version of this thing, with new default config file included.
Still need to do full test pass on all the errors and corner cases, but should be about ready now.
Pylint is happy except for missing documentation.
Flake8 doesn't like some of the whitespace I did, to align stuff on different lines for readability.
Both of course complain about some of the line lengths unless using modified settings.
Edit: New version of the script in message: #off-topic-tech message
Again wanting any and all feedback.
Like before, will not implement all, but want to hear it and respond.
And implementing any kind of unit testing is way out of my league right now.
Tried to rejigger some things so it should be now closer to possible.
github is your friend.
also a readme would be handy telling what script does and how it works. And with how I mean "how the user can use it" not what lines of code do.
Again, major rewrite of: https://github.com/azlux/nft-dns
And will be sending it back to the original author at some point.
Rewrite can still be documented and forked on github.
Chances are bigger someone picks it up and finds a bug and make pull request 🙂
"PR are accepted on the source only."
Yeah, just that that github page is better documented than the other place.
I wont accept a full rewrite. Just too much headache. But maybe the developer would.
Which is linked in the readme.
And if he/she accepts it or not is up to them.
I just wanted actually working version for myself.
And I'm not willing to publish that rewrite publicly myself at this point.
🤷♂️
Goes against how open source stuff works but ok.
class ConfigException(Exception): pass

Just making new exception where only change is the name.
So I don't have to catch generic one or like.
Pretty common thing AFAIK.
debatable.
That I'm willing to send the stuff back upstream, but not publish it myself where I would need to put my actual name up tied to the thing?
case _: logger.setLevel(DEBUG) config_exception = ConfigException( "Unsupported logging level", [] )
ConfigException... Unsupported logging level?
You made a seperate exception, if someone parse a logging level that doesn't exist?
That's some ... next level stuff I wouldn't be bothered by.
I have to go out for a walk. But I will test it later. I am pretty sure logger itself shoudl handle that stuff. The package.
But I am not sure for python what happens if you do logger.setLevel(WTFBRO?). It should default or throw error.
Point is to stop the config parsing on errors.
Used to be just exit(1):s, now it throws exceptions, and catches in main and exits.
Yes, it would throw exception and crash the program.
But point is to handle the expected settings, and quit and inform the user if something isn't right.
Which is the point of that simple inherited exception class, to not swallow any exceptions I didn't want to handle.
Which just throwing Exception:s would cause.
So if you try except, you can just print the error and continue script.
I entered a not existing level. And then the script seems to continue? Else it wouldn't print hello world.
Yes, but point is that I don't want to continue if there are errors in the config.
As by default the thing will be running as service.
And point was to move to exceptions from that, for both testing purposes if someone in future wants to do unit testing.
And to allow using the stuff as includes in other code.
You also said you didn't want to share the script with anyone else?
It used to be log error and exit.
So why go through all this trouble, if nobody else is going to use it besides you?
If the upstream doesn't accept it, not for this time.
Hobby thing.
Or is this to protect yourself in the future for setting non existing logging levels.
Which, like I said, the package does handle.
Yeah, I get that. Just from personal experience. I wasted shitton of time on that kind of stuff.
Point is to give different error report than what the logging.setLevel() would give itself.
And I wouldn't do it again. Because I learned I was doing it wrong. The packages are so mature that if you need to spend so much code on it, you probably not doing it right.
I have no regrets of doing the stuff you did, because I learned and enjoyed it. But now if I write a script, I just focus on the actual goal.
ERROR: Config Error: Unsupported logging level. Exiting.
DEBUG: Need to be one of: 'debug', 'info', 'warning', 'error' or 'none'. Current: 'notsupported'
If I am really bored and have too much time on my hands, yeah I do the extra fluff.
Like I am thinking of making my own progression bar for a script 😏
I remember doing years ago something with \r to keep pasting on the same line, and made a progress bar that way.
Yeah can do that all in the except clause I wrote before. That's what it's for.
print(error)
print(please parse 'info', 'warning' ....)
And if you note, I do stuff that info into the exception, throw it, and catch in Main(), outside of stuff that external use would do, print and exit.
Using same thing in multiple places for different config errors.
The logging package has the error "unknown level: 'test' so it's already a "feature" and a exception they (developers of package) are aware off.
Than its up to developers to give extra info if they want or not. Some don't want to give extra info for obvious reasons.
also, as I am OOP programmer, split functionality into classes 🤣
I know it's a python thing to make one script. But if you doing it as a hobby, it's fun thing to start with classes in seperate files and get it all to work.😏
makes testing easier too (visually).
For example main.py and customLogger.py or customException.py. Than you already know that those 3 can be tested separately.
But would make deployment much more pain, and the file isn't that large.
Euh, yeah no. My lead would not allow it.
Even a time there was a rule no more than 10 - 20 lines of code per method.
And no classes larger than 100 lines.
Any of those two cases and he said; "You done something wrong".
And with wrong, the architecture/structuring of the code.
It would all work fine. But it was not up to clean code practices they wanted.
Splitting that Update() would be real pain, with the continues etc.
Could be done, but would make it more complex, not simpler.
If you do it right, it wont be more complex.
It's getting the "do it right" part that is a PITA.
Don't get my wrong, I 100% understand you 😛
But since you said it's a hobby, I am just throwing a few more challenges your way haha.
This book, was kind of the bible for a project I worked.
But in the end, it's just one way of coding. And doesn't mean that other types are "bad". There are some pretty bad code bases out there. But if it's only for you? Who cares. If you work together. Decide together how you want to code etc.
Yes, it could be done, by restructuring more.
But that is first Python thing I have ever done, and it started as minor rewrite to fix some bugs in the original.
And then ballooned from there.
It is going into production use, but lot of the rewrite wouldn't have been strictly necessary.
Is the word I am looking for "be pragmatic?" Not sure.
Just see how it goes and fix/adjust along the way.
Also, it's not about how large it is. It's 756 lines 🤦♂️
already giving me a headache scrolling up and down.
Yeah, organization.
I know.
But in some cases having it in one file is easier for the actual use.
For development multiple files would be better.
While so much could have been put in seperate files so one can focus on the things.
But a problem for small stuff when you don't compile the thing in the end to one executable, but that file is the "executable" as itself.
Downsides of scripting languages.
Original had worst of both worlds.
It has one main file, and additional file that only contained single dataclasses definition...
Well, Pydantic version, stripped Pydantic dependencies out, only that dataclass equivalent and then IPvAny class instead of just strings.
And when that DNS resolver just outputs strings, and NFTables takes strings, it was kind of useless as intermediate step with validation.
And then added more dependencies by moving to those NFTables Python bindings instead of original just calling nft command in shell.
@tribal kraken Finally joined the club 😁
I am only allowed to turn on the solar power plant during testing and keep it turned off until I get final permit
Ugly roof 
I need to remove moss from the chimney btw
Dust on the roof tiles are from the panel installation and they will be washed away by rain hopefully 😅
At least I am no longer an energy slave lol
The installer freaked out when he saw 63 amp fuse though
Mentioned a family house should have 30 amps instead of 63 amps but this is what we got installed from the utility company 40 years ago when the house was built so I have no idea why I have 63 amps
There is no "should".
Just what is norm to save on costs.
Just that traditionally bigger feed costs more as both install time charge, and as higher base pay after.
If you don't have enough, you get issues.
If you have too much, you just pay more, even if that, in base charge.
And upscaling the feed to the house is very expensive, often needing completely new install, costing as much as getting the bigger one straight away.
Downscaling should only need new smaller main fuses and resealing.
He probably got confused because the utility company granted me max power of 13,80 kw from the grid when the house was built and those fuses above mean I could theoretically get even more kw although I doubt it
Since when can you control downloads remotely on desktop steam????
People who want to install solar power plants in the future might not get a permit because the grid is almost overloaded from solar power plants
thats actually really nice
My coworkers who live in apartment buildings have this problem. When you have a house, a single mains supply cable is only for you and the power is not split. When you live in a residential buidling, a single mains supply cable is split between building apartments and the power meters often reset due to limiters
Meaning, if you turn on multiple devices with high power draw like AC or oven, you will trigget the power meter limiter and will have to reset the meter
Never had this case in my home because I have 13,8 kw limit and it's really hard to reach that limit unless you literally turned on every single appliance in home at the same time
Those fuses can be anything up to whatever current the actual cables used can take.
But using larger diameter cables is costly in materials etc.
And usually people get the minimum size to be able to handle that kW limit they pay at the start, for when original installation is done.
And then often the owner 30+ years later curses the original cheapscates.
Because that extra cost from going to double limit is just drop in a bucket compared to the cost of the whole house build.
Guess I don't have to worry about that 
Too bad the electrician couldn't install huawei smart power sensor since there was not enough space in the breaker box. I will have to install a larger breaker box later I guess....
I am a bit... lagging on recent graphic advancement
was Lumen / ray tracing / whatever that (family of ) lighting model is called been used more cently?
I am replaying TTP2 and just can't help myself but like every time they placed light emitter to add some artistic tough to the scene
(so sad it's no longer part of SF...)
totally not just dumping my screenshots...
or could point a finger and say that it's a cheap illusion
what game is this
the Talos Principle 2
cant really group them all together like that
this is form a small segment that you run through in like 5-10 minutes.
it's actually mostly under open air
but sometimes thry drip you with some heavy atmosphere goodness
theres some great atmosphere there
but foliage isnt really shaded properly there..
yeah, shadows are missing here for some reason.
AO too from the foliage
im starting to understand more and more why U+ discord praises megalights so much..
im pretty certain it'd fix those shading issues..
yea the talos principle remake and second game used UE the right way
which is weird, given that same game does great detailed shadows from foliage moving under wind under sun
(I think some screenshots are missing... could windows or steam clean them? sad)
terain is pretty mid in that tho to be fair
jesus
would make CSS jealous 
U+?
ultra +
graphics tweakers/modders
yeah that happens... dithering in water reflection looks a bit questionable at closer look.
but if you just™ defocus a bit... hehe
UE SSR my beloved
they used megascan, some rocks are very detailed
this is just first one I looked at and saw shadow from trees hehe (btw tree is not in the shot)
i dident even catch that until you mentioend it so it passes
The shadows underneath the leave is because of ray tracing I think?
doing shadows from direct lighting is a lot easier than shadows from indirect lighting
and yeah, UE5 is heavy on the megascan models. Easily accessible.
yeah
its also much easier to do noise free direct lighting than to do the same with indirect
better rocks, more fine shadows
that post processing is killing me
www.fab.com/sellers/Quixel Megascans (if you want to browse actually models).
I might have added film grain...
now this i think looks good
though i still think the grass could use some more AO
and CA, sharpening..
how big is install size? 😏
more TTP2
the vine on the tree trunk on the right has individual leaves in and out of shadow
leaves above don't just get in shadow but also glow because of the angle
idk about reawakaneed, but TTP2 is around 100GB
not much
since when???
i remember when i played it originally it was less than 60GB
unless they made some more improvements since
I thought so too...
maybe DLC added a few dozen GB (given it has a lot of world stuff)
I can't look right now because steam is again blackscreening me...
jesus
that noise looks like PT without a denoiser
you cant escape nosie on UE to be fair
iv been noticing it a bunch of subnautica 2 lately aswell
so much noise on reflective surfaces
it also doesnt have to be any where near that bad
when you look at it on 4k screen you don't notice it a lot
wtf did the devs do smh
--or standing still
there framegen trailing that stays for a while if you only stay still for a sec or 2
but thats not framegen slop at all since the numbers are clear
nope
frame gen artifacting should be gone the next frame after you stop moving
there shouldnt be any delay
if there is its not FG
sadly it just trails for a second..
that is just straight up wrong tho
in reality it trails for a while
pretty sure SF has that too
if you walk in dark room and notice reflections from tiny lights on machinery - it has tendency to flow slowly
but that shouldn't be for shadows...
ok, but what is then?
what you described is lumens temporal denoiser
dat occluded subsurface scattering tho
temporal accumilation*
doesnt necessarily have to be TAA
idk man in terms of a full frame I think this is pretty acceptable, you had to zoom in on this pretty aggressively to find this noise and let's not forget this is a realtime application, not an offline render
I most likely was using FSR
in cases where you can't match ground truth perfectly you at least want to try matching a ground truth histogram which that frame looks like it does pretty competently
that shot doesn't have UI - so it was made in photo mode
but idk if it was before they made it more reliable
the zoom was only enough to make the screenshots 4k pixelgrid fit on my 1080p display
FSR 2+ is TAA
but like FSR isnt capable of the level of smearing lumens temporal denoiser is
I think a 2x zoom is pretty aggressive when you're evaluating visuals, especially in realtime rendering apps
bc that's not how you're actually consuming the content
here, lights that cast shadows, and there's plenty of reflection
thats a lot more than 2 thats dang near 4
its absolutely reasonable
fitting a 4k pixel grid to a 1080p display is a 2x zoom
and yes, shadow does change depending on the viewing angle
which also applies some spatial upscale
which ends up blurring stuff
that's pretty
downloading the image and using ICAT's zoom / discord zoom
either way, my eye is like "okay, there's some shadow there"
it looks like they're approximating the area lights with point lights which is fair, I don't think LTC shadows are figured out yet, but it looks very nice
pheh
yep. that one grass model has no shadow
oops
super heavy shadows
it's in cutscene where... there's are jsutified post-processing noise filters (hence the grid and broken pixels)
but oh man the heat of the star is immense
definitely partially but uhh FG amplifies it (i spoilered it since its a subnautica 2 example since iv been noticing it a lot in that)
Im not seeing anything but moyion blurr
and those blobs of extra noise are likely droplets on helmet
@soft bloom and @mental oriole
If you want to take a look at the new version and comment.
No time at work (:
No hurry, as in just wanted to notify that that existed.
If it takes few weeks, no problem.
quickly can sa y that... ruff would reformat for you 🙂
it's not just extremely easy - it is written in rust and is very fast
I did use black to reformat, then reformat some stuff back.
As Ruff is not included as package in debian stable yet.
Only Testing.
Just that flake8 doesn't like the realigning of the early assignments etc.
With extra whitespace.
Or aligning multiline strings so that the contents align, instead of the start of the line.
When some of the strings start with just ", some with f", some with + " and some with +f".
Same with the line length thing, we aren't using old style text terminals anymore where you could only fit 80 characters on the screen at same time sideways.
So excessive multilining to keep lines that short is IMHO just stupid.
Needs to be short enough to be readable without excessive head movement.
I will need to look into code closer once I am more awake...
Where that multilining makes it less clear, not more.
Yeah, no hurry, same as for Kit, I posted that yesterday once without pings, but as there were no acknowledgement at all, and newer stuff pushed it up, wanted to make proper notification now.
Like said, I haven't even done proper testing of that code yet.
Just that it runs on the normal track and seems to work the same as before on quick look.
But unless I find bugs when testing bad configs or longer term actual running, that should be it.
Still need to finalize some SystemD unit files too.
So for example flake8 doesn't like this:
class Defaults:
log_format: str = "%(asctime)8s,%(msecs)03d %(levelname)5s: %(message)s"
log_datefmt: str = "%X"```
Because there is extra spaces to align the stuff for readability.
readability in my mind, some do prefer it to be with minimal whitespace, which is kind of better on one line only view.
log_datefmt: str="%X"```
With 13 such lines in that class.
The points are based to have single space on the longest one.
One of the classic differences-in-opinion style things in coding, to align with extra whitespace or not in that kind of blocks.
This in turn was a thing that pylint didn't like by default, too many members in a class.
Which I mentioned in the previous message on the channel?
Complains on defaults about the missing docstrings for everything.
That defaults class having 13 members when default limit is 7.
configure_sets() having 16 local variables when default limit is 15 and being 55 statements long when default limit is 50.
And then 12 lines being too long, when default limit is 100, with longest being 130 characters.
For 8.88 of 10 as rating without any overrides.
That configure_sets could be split up, but that would just be lipstick and kind of make it harder to read.
When you would just need to jump into another part of the file to continue reading then jump back.
ok
Flake8 in turn on defaults just complains about extra or missing whitespace and too long lines.
Doesn't like +f"<string>" and would want space between + and f
+f"Current: '{parser.get(section, "domains")}'",```
When I have done that kind of sacrilage to multiline string.
Would want it to be:
+ f"Current: '{parser.get(section, "domains")}'",```
Like black autoformatted it to, and I then manually changed after.
pylint rating goes up to 9.53/10 when I tell it to ignore that docstrings are missing.
And to 9.91/10 if I just tell that and that max line length is 130 and not the default 100, leaving just that one instance of too many attributes, one too many locals and one too many statements.
Where the latter two could be easily fixed in code by putting in subfunctions, but that would IMHO make the actual function harder to read.
But just got into editing the exception stuff by creating actual modified init and changing how the additional lines are being added...
This kind of insanity:
raise ConfigException(
"Invalid line(s) in main config file",
[
f"File: '{exception.source.absolute()}', "
+f"Line {entry[0]}: '{entry[1].rstrip('\n')}'."
for entry in exception.errors
],
) from exception```
But seems I need to reformat them all after taking example with black.
Edited the previous to match.
Full new version of the script file with those exception changes, will link in the original message:
Which also made pylint not complain about too many local variables or too many statements in that one function anymore.
So just documention, line-lengths and that defaults class having too many variables/attributes.
This ends up printing:
10:47:20,682 DEBUG: File: '/usr/local/sbin/nft-dns.conf', Line 4: 'jotain'.
10:47:20,682 DEBUG: File: '/usr/local/sbin/nft-dns.conf', Line 5: 'ihan'.
10:47:20,682 DEBUG: File: '/usr/local/sbin/nft-dns.conf', Line 6: 'muuta'.```
And exiting. Because of those three lines of random finnish crap. "Something completely different"
Just converted that ", Line X: " to lowercase.
I wouldnt be shocked if they were interpolation artifacts but like bruh, using moyion blurr with FG ? Smh
Itd be so much more interesting to see these artifacts happenin on actual details
DLSS FG is kinda garbage in sn2 tho tbf
so what ima not and leave the 25 pixel textures and shadows unsmeared?
Amazing shadow detail indeed
Well cant really save this games optimization issues with it being a ue5 game, but heck its ea1, so maybe performance will improve in the future, not that fg in sn2 doesnt look like ass, alot of artifacts with that
Maybe this the sad future, where we just fg and upscale and try to believe a engine isnt complete trash
Remember monster hunters set the sad standard, fg required just to reach 60 fps
Oh yea no im playing at 40fps with framegen on Sn2
Feels fine surprisingly but its not good at all
So 20 fps without framegen, and im getting 40 fps without framegen, welp
Im not suprised i saw a 5090 cant reach 60 fps at 4k without upscaling, classic ue5 moments
no
you'll use my tweaks 
i thought you had a B580 💀
Tweaks wont save sn2 performance issues, or maybe if you go hard enough on the tweaks hmmm
No thats on my 4070ti
wtf man
what rez and settings ?
I havent tried it on arc and i dont plan on it
Lol what thats crazy, how bad are sn2 performance issues
please hold
This might be even worse if its 20 fps at 1440p, not good
DLAA too ?
5090 4k native was in the 40’s iv i recall
Yea
Ok i can believe this is even worse than i thought, tested max settings to, getting between 20 and 30 fps, so very similar results
Not suprised this gpu im testing has 4070ti performance, so i would be getting similar results, sadly it makes sense
native without FG / with FG / FG + quality upscaling mode
my 9060 XT is doing better than your 4070 TI 
doesnt drop that low even in my base
okay it does if i disable FG
anyway the only reasonable explaination i have is that FG isnt working on your end
or my tweaks are just OP
Who knows i just tried 1440p max settings, got around 30 fps, not a great experience
whats your card ?
Mobile 4090, performance of a 4070ti i think, i know its right around a 3090ti
Runs most games great, just normally not ue5, because optimization issues, like sn2 has to
first half of the map runs even better as well
Satisfactory somehow doesnt but i think the devs used magic with that game
Satisfactory for me still remains one of the only ue5 games i have played that doesnt run like ass
yeah i wouldnt call satisfactory optimized
SF runs no better in practice than sn2 for me
while looking a lot worse
anyway thats my unique cope
Sn2 for me runs so much, it looks very similar to sn1, while running much worse to
Well all hope isnt lost its still ea1
im still hungup on the grass casting proper shadows
one of my favourite unique copes fullfilled
peak shading : )))
