00:24 < bridge> cheaters suck :owo: 00:24 < bridge> im all for perma bans 00:24 < bridge> simply because we can teach cheaters to fear them instead of being like "yeah ill just wait it out, its only 1 or 2 days" 00:25 < bridge> sadly idfk how to deal with evaders 00:36 < bridge> perma bans suck 00:37 < bridge> the average 12 y/o cheater doesn't need a perma 00:37 < bridge> for severe offenses yeah though 00:38 < bridge> they can just make a new account 00:38 < bridge> We don't have hwid bans 01:43 < bridge> we should NOT encourage ban evasion 01:44 < bridge> I don't think this is radical information 01:45 < bridge> valve takes the same approach with CS2, you're allowed to play again if you make a new steam account, the punishment is losing everything associated with your old account. 01:46 < bridge> at that point might just make their account content deletion the actual punishment then 01:46 < bridge> it's also pretty lame to encourage people to come back under different names 01:47 < bridge> I don't think we should have policies that are impossible to enforce 01:48 < bridge> but as if you say "the punishment is losing everything", then might as well make that the official stance 01:49 < bridge> but if as you say "the punishment is losing everything", then might as well make that the official stance 01:50 < bridge> damn my engish brokey 01:50 < bridge> can't even order correctly words 01:54 < bridge> If you're perma banned I think it's implied that you've lost your account. 01:54 < bridge> 01:54 < bridge> The only way to truly prevent ban evasion is to do what osu! does, because they made the entire gamemode from scratch they can basically enforce that all account have a reasonable skill progression history, anyone who doesn't have that gets investigated for ban evasion (mostly for top players). 01:54 < bridge> If you're perma banned I think it's implied that you've lost your account. 01:54 < bridge> 01:54 < bridge> The only way to truly prevent ban evasion is to do what osu! does, because they made the entire gamemode from scratch they can basically enforce that all accounts have a reasonable skill progression history, anyone who doesn't have that gets investigated for ban evasion (mostly for top players). 01:55 < bridge> they also just auto ban your account if you improve too fast 01:55 < bridge> new accounts aren't allowed to set top ranks 02:00 < bridge> hi sjrc6 02:00 < bridge> hi 02:00 < bridge> gn sjrc6 02:00 < bridge> gn Teero 02:59 < bridge> Interesting, this would work for ddnet 02:59 < bridge> Even now, but hard to enforce 02:59 < bridge> it's been suggested 02:59 < bridge> but admins need to do the work 03:06 < bridge> guess the idea would be to try to find a formula that expresses most people's growth, and fit that to each player 03:06 < bridge> the shape of the graph is a lot more important than the absolute values since people improve at such different rates 03:07 < bridge> I think that's over complicating it 03:08 < bridge> you're never going to get such accurate data for each player that you can confidently compare it to a curve 03:08 < bridge> also we can't use this anyway because we're not osu, ddrace has already been released for 10 years without finish tracking being tied to accounts 03:19 < bridge> Issue after a pr is a new one 03:21 < bridge> I think you're underestimating how much you can get from fudging statistics with educated guesses :P 03:21 < bridge> it doesn't need to be a great model to be a useful model 03:22 < bridge> no I mean there are already thousands of highly skilled players for whom we have no gameplay history, so you can never accuse someone of ban evading when they could have already gained their skill at any point in the last 10 years 03:22 < bridge> you know they pay people lots for doing bad statistics 03:22 < bridge> they call them data scientists XD 03:26 < bridge> oh, I wasn't thinking ban evasion tracking based on skill levels, I was just thinking of detecting cheaters by suspicious finishes 03:27 < bridge> but I agree that that's a bit of a probelm 03:27 < bridge> but I agree that that's a bit of a problem 03:31 < bridge> Less than 1k points and top10=investigate 03:31 < bridge> 🤯 03:31 < bridge> you can do a lot more than that 03:32 < bridge> not all cheaters are TASing top ranks 03:33 < bridge> It's definitely better than hiding record channel 💀 03:34 < bridge> oh for sure 03:36 < bridge> it's such a shame that all the best anticheat systems are largely security through obscurity 03:42 < bridge> for any simple notion of human-like behaviour you can engineer a system that maximizes human-likeness while also maximizing evil cheatyness 03:44 < bridge> if you think you're clever by picking a complex notion of human-likeness, by using machine learning or something, your adversary can simply produce their own system by training it on your system 03:46 < bridge> release your anticheat for others to make use of and it becomes half useless by the next week 03:46 < bridge> or keep it to yourself and let others either suffer through making their own or simply force them to live without since they don't have the resources to create one 03:47 < bridge> Clearly this is a great justification for SaaS 03:47 < bridge> oh how I hate SaaS 03:48 < bridge> and you can still totally use a SaaS to produce a dataset for machine learning or for simply analyzing by hand 03:52 < bridge> Are we truly doomed to fail? Should I go to bed? 03:52 < bridge> The answer to both those questions is yes. 04:00 < bridge> eeeh idk about this 04:01 < bridge> im sure classifying human vs bot is much more easy than replicating human behavior 04:03 < bridge> it's easier, but it's also much easier for the bot creator to produce training data (assuming we're at the point where we're resorting to machine learning) 04:04 < bridge> or more like, the bot creator doesn't even need training data, they can just use the anticheat as a classifier 04:06 < bridge> plus the botter has the advantage that they only need to control one or two tees, and they can use a fairly large amount of computing resources for that 04:06 < bridge> meanwhile the anticheat needs to cheaply classify hundreds of players 04:08 < bridge> strictly speaking the goal isn't human behavior, it's behavior that passes the anticheat 04:10 < bridge> the anticheat also needs to be fairly lax in order to avoid undeserved punishment 04:16 < bridge> we can do that for ddnet 04:16 < bridge> there is a huge gap in skill level gain due to age i think 04:16 < bridge> but its a thing 04:17 < bridge> we just have to wait a couple of years for everyone to transfer 04:17 < bridge> hi tooter 04:17 < bridge> im awake 2 :D 06:04 < bridge> it’s normal time for him ish 07:10 < bridge> Gm sjrc6 07:10 < bridge> gm Teero 11:02 < ws-client> @totar what is the name of your client? 11:02 < bridge> it's tater client, or tclient 11:02 < ws-client> nono 11:02 < bridge> ? 11:02 < ws-client> no or 11:02 < bridge> both are acceptable 11:03 < ws-client> and watafak is TaterClient-ddnet then? xd 11:03 < bridge> the name I picked when I forked it 11:03 < bridge> xd 11:03 < ws-client> cleanup ur branding pls 11:03 < bridge> too late 11:03 < ws-client> no 11:03 < bridge> scripts are pointed at the url 11:03 < bridge> :P 11:03 < ws-client> oh 11:04 < ws-client> go create a new org github.com/tater-client/tater-client and drop tclient everywhere 11:04 < ws-client> ez 11:04 < ws-client> @totar and what is your name? 11:04 < bridge> it's Tater 11:05 < ws-client> whats totar then 11:05 < bridge> the avaiable discord username 11:05 < ws-client> and who is sjrc6 11:05 < bridge> the available discord username 11:05 < bridge> is your bridge bad and uses "username" instead of "display_name" 11:05 < ws-client> seems like it 11:05 < ws-client> but its good for pining people 11:06 < ws-client> but solly has spaces in his name for example i assume its the display name 11:06 < bridge> for me it shows "@Tater what is your name" lol 11:06 < ws-client> https://zillyhuhn.com/cs/.1743930385.png 11:06 < ws-client> thats why you should fix it xd 11:06 < bridge> you don't get profile pictures? 11:07 < ws-client> bro its irc 11:07 < bridge> ok 11:07 < ws-client> there are no profile pictures 11:07 < bridge> sry idk which of the 3 bridges you are using 11:07 < ws-client> fair 11:08 < ws-client> ok good talk i will also consistently refer to your client as some undefined mess of a branding 11:08 < bridge> tbh I didn't name my client something more interesting because I didn't plan to make a client. If I knew it would have thousands of users I would have given it a name that wans't my name lol 11:08 < bridge> I thought I would just give the exe to a few friends 11:08 < ws-client> i like tater better than tclient because letter-client is so overused there are duplicates already. For example f-client is that FallenKN client or fokkonauts client who knows 11:09 < ws-client> i think having your name in there even with a big user base and contributors is ok 11:09 < bridge> now people copy me and call their clients {their name} client lol 11:10 < ws-client> it always have been a thing kinda no? 11:10 < bridge> idk maybe? 11:10 < ws-client> as i said f-client f stands for Fallen or fokko 11:10 < ws-client> chillerbot-ux chiller stands for chiller 11:10 < ws-client> kclient k stands for krisko 11:11 < ws-client> didnt botox also name a client after him self? 11:11 < bridge> yeah but they usually don't use the full name like people do with tater client 11:11 < ws-client> a ye 11:12 < bridge> I mean tbh I'm not unhappy with the name, because it's fun to go ingame sometimes people are like "Tater? like the client?" xdd 11:12 < ws-client> fame 11:12 < bridge> they don't even know I'm a person tho xd 11:12 < ws-client> i usually get "chiller like the gores mapper? YOUR MAPS SUCK" 11:12 < bridge> fallenkn client is old af, no one will use it nowadays anyway. but yeah, that makes client names context-based 11:13 < bridge> hahaha, I have some nice memories on your maps. but they do suck 11:13 < ws-client> ._. 11:13 < bridge> f3 ngl 11:13 < ws-client> cyber bullying 11:13 < bridge> the suck is part of the fun tho 11:13 < bridge> gitgud 11:14 < ws-client> sucking can be fun ;) 11:14 < bridge> the new maps are perfectly balanced and flawless. the old maps you never know what will happen ahead 11:14 < bridge> suddenly the worst part you've ever seen in your life 11:14 < ws-client> there might be dragons 11:14 < bridge> yk they're like oldschool parts where you fail the worst possible way but its fun to fail like that and not like on any hh map xd 11:14 < bridge> just bumping into bottom freeze 11:17 < ws-client> oof i regret updating arch.. ddnet launches slower than csgo now 11:17 < bridge> gg 11:18 < ws-client> @zhn can you help me ping Cans? is he here? 11:20 < bridge> > i usually get "chiller like the gores mapper? YOUR MAPS SUCK" 11:20 < bridge> fair 11:20 < ws-client> ._. 11:23 < ws-client> hm i would blame some sdl update or something but other sdl games still launch instant 11:24 < ws-client> no idea how to debug this 11:41 < ws-client> i can consistently reproduce this backtrace during the black loading screen https://paste.zillyhuhn.com/Xj ``libusb_handle_events_timeout_completed`` looks sus 11:46 < ws-client> hmm i can not reproduce on my laptop??? 12:13 < bridge> @cansulat 12:13 < ws-client> what issue did you have with your client Cans? 12:47 < bridge> new linux-zen is buggy af 12:47 < bridge> don't update 12:48 < bridge> there is a memory leak somewhere and sometimes kernel threads start but don't stop causing cpu to be at 50% until i restart consistently 12:48 < bridge> they appear individually 13:49 < bridge> torvalds pls fix 13:49 < bridge> man 13:49 < bridge> 4gbs on idle?.... almost like windows 13:49 < bridge> 4gbs and 50% cpu on idle?.... almost like windows 13:56 < bridge> Modernizing 13:56 < bridge> im downgrading to paper and pencil 14:07 < bridge> Don't do it 14:07 < bridge> Don't validate schools which make us do that 14:30 < bridge> i hate ddnet... 14:31 < bridge> why must the physics be so bad..... 14:31 < bridge> you shouldnt have started the C rewrite 14:31 < bridge> your sanity is slowly dropping teeto 14:32 < bridge> it's not about this 14:33 < bridge> im testing some new ddnet techniques and they are not always possible 14:33 < bridge> for example you can climb a 5x1 staircase with max speed with just djs 14:33 < bridge> but not for infinite time 14:33 < bridge> at some point it breaks and you bump 14:33 < bridge> so unsatisying 14:33 < bridge> so unsatisfying 14:37 < bridge> when did the shutdown times get so slow? 14:39 < bridge> there is an open issue about it, i dont remember what actually introduced it 14:40 < bridge> also is anyone else having the bug where you have to wait endlessly after rendering a demo? 14:40 < bridge> the client just sleeps for like 30s 14:40 < bridge> I told you how to reproduce it 14:40 < bridge> what? how 14:40 < bridge> i don't know how 14:40 < bridge> I dunno, look dms 14:41 < bridge> nice, gonna look through the 10k msgs 14:41 < bridge> https://discord.com/channels/@me/1151255660965658675/1317932605227929631 14:41 < bridge> xDD 14:41 < bridge> https://github.com/ddnet/ddnet/issues/8445 :c 14:41 < bridge> nice 14:41 < bridge> is that a club i'm missing? 14:41 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1358421428034080999/image.png?ex=67f3c80a&is=67f2768a&hm=fa557eb573af65dc4b8de104949d0c476d8894f03011407ea7fe1962e846d14a& 14:41 < ws-client> lmao i pinged some dude from red hat to fix my ddnet client issue xd 14:41 < bridge> dms xd 14:41 < bridge> it's private msg link 14:41 < bridge> :kek: 14:42 < bridge> ??? XD 14:42 < bridge> poor dude 14:42 < ws-client> what poor dude 14:42 < bridge> "hey man, can you fix this indie game for me pls ty" 14:42 < ws-client> i spent my entire sunday debugging because of him 14:42 < bridge> and did u fix it? 14:42 < ws-client> i found multiple fixes and the bad commit 14:43 < bridge> and is it their fault? 14:43 < ws-client> idk 14:43 < ws-client> i found no fix in his repo 14:43 < ws-client> just know that downgrading his package is one of the fixes 14:43 < bridge> was the bad commit in their software? 14:43 < ws-client> yes 14:43 < bridge> i have almost every single time 14:43 < bridge> Ok nice 14:43 < ws-client> https://github.com/vcrhonek/hwdata/commit/fa1ccf4e4944bdb4110025c5c40e4fee8c8de29b 14:43 < bridge> Always blame others 14:43 < ws-client> yes! 14:44 < bridge> ddnet has no bugs 14:44 < bridge> ddnet is perfect 14:44 < ws-client> i would say poor dude doing open source for free and i am some ass complaining 14:44 < ws-client> but he is red hat so he gets paid 14:44 < ws-client> why is launching ddnet not part of his pre release test flow? 14:45 < ws-client> @milkeeycat it was hwdata OMG 14:45 < bridge> true 14:46 < bridge> can you take my steps and try to reproduce the exact thing? - because i can then reopen the issue 14:46 < bridge> select a demo 14:46 < bridge> render cut to video and set the playback time to 0.1 14:46 < bridge> pause the demo rendering 14:46 < bridge> forward a bit 14:46 < bridge> Close the rendering 14:46 < bridge> chillerdragon: pro debugger 14:46 < bridge> bug hunter 14:46 < ws-client> ikr 14:47 < ws-client> @Jupstar ✪ my favorite hotfix is this one ``sudo rm /usr/lib/libusb-1.0.so*`` (dont try at home) 14:47 < ws-client> increases ddnet launch times by 10000x 14:47 < ws-client> decreases* 14:47 < bridge> watafak 14:47 < ws-client> :D 14:47 < ws-client> i giggled too 14:47 < ws-client> but it makes sense 14:48 < ws-client> this red hat dude with his usb id database ddosed libusb 14:48 < bridge> xDDD 14:48 < ws-client> so if you delete libusb ddnet just says no joystick found and carries on 14:48 < ws-client> if you have libusb it gets stuck there 14:48 < bridge> I tried it.... 14:48 < bridge> I am now writing from phone 14:48 < bridge> me who did rm -rf ~/.c and tab enter, without realizing .config is autotabbed before .cuts 14:48 < bridge> Delete it on your phone too 14:49 < ws-client> f @melon 14:49 < bridge> There is .cuts? xd 14:49 < bridge> glad i'm on nixos so most of my config directory is declaratively in my dotfiles 14:49 < ws-client> @teero777 my keyboard kept working :D 14:49 < bridge> I dont have that 14:49 < bridge> i created .cuts myself 14:49 < ws-client> giga chad keyboard does not need libusb 14:50 < bridge> it were cutstamps of my songs 14:50 < bridge> xD 14:50 < bridge> chiller has a apple keyboard. it's soldered on the soc directly 14:50 < ws-client> bluetooth keybord goat 14:51 < bridge> sounds like some floating point shenanigans once again 14:51 < ws-client> teeworlds far lands when 14:52 < bridge> i respect your willingness to alpha test my 3 month late debian packages on arch tho chiller. 14:52 < bridge> 14:52 < bridge> Thanks 14:52 < bridge> I'll not have this issue 14:56 < ws-client> you are welcome :) 14:56 < ws-client> you might not be afected my laptop isnt either it seems hardware dependent 15:03 < bridge> exactly :D 15:03 < bridge> and i hate it :D 15:03 < bridge> 😀 15:04 < bridge> @jupeyy_keks send gpu 15:04 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1358427187296079902/image.png?ex=67f3cd67&is=67f27be7&hm=042aec0356345e9b2b7f01539f83514d6ee202d5454a562527bd3be3a8b3395b& 15:04 < bridge> my gpu is slower in encoding than my cpu 15:05 < bridge> copying framebuffers around is simply not efficient 15:05 < bridge> i could this faster by hand 15:05 < bridge> @jupeyy_keks give some chill music, 1min 15:06 < bridge> 1min and chill that is hard 15:06 < bridge> i could do this faster by hand 15:07 < bridge> also should be instrumental 15:07 < bridge> i cna send you a sad ost 15:08 < bridge> https://www.youtube.com/watch?v=eWws1VM7iW0 15:08 < bridge> this??? XD 15:08 < bridge> that's not chill 15:08 < bridge> that's epyc 15:08 < bridge> send link 15:08 < bridge> https://youtu.be/YknzjncfHnY?t=63 15:09 < bridge> it ends at like 2:14 or smth 15:10 < bridge> too much 15:10 < bridge> to high quality 15:10 < bridge> like relaxing ones 15:10 < bridge> ?? 15:11 < bridge> anyway can't help you xd 15:12 < bridge> https://youtu.be/lr1fTEi9EMA?si=ZurtsOBEH4DVjmIV&t=2217 @teero777 check this timestamp 15:12 < bridge> the start is good 15:12 < bridge> thx 15:17 < bridge> fuck this 15:17 < bridge> no music today 15:17 < bridge> even if jupstar won't watch 15:18 < bridge> you need to collect your music from other sources anyway. i already hear my own music anyway :lol: 15:18 < bridge> the jump sounds overtone the music anyways xd 15:19 < bridge> teeworlds sounds are ASMR anyway 15:20 < bridge> https://youtu.be/HKYE1HVRPBw 15:20 < bridge> here you go 15:20 < bridge> thats what i wanted to show 15:20 < bridge> @chairn 15:22 < bridge> and due to some dubious floating point behaviour you can't keep the speed forever 15:22 < bridge> uh, i don't see anything weird ? 15:22 < bridge> where's the red circle when you need it ? 15:23 < bridge> well, at some point the speed decreases for 3/4 of the runs because there is no possible solution to keep going at the same speed 15:23 < bridge> the speed in the parentheses is the unramped value 15:51 < bridge> you cut the footages too soon 15:51 < bridge> bottom right start at 75, ends at 63 i think 15:52 < bridge> bottom left starts at 36, ends at 36... 15:54 < bridge> look at the unramped values.... 15:55 < bridge> for ddnet 2 please remove the stupid velramp lol, it is pretty useless. 15:57 < bridge> bottom left starts at -50 and ends with -43 15:57 < bridge> bottom right starts with -111 ends with -64 15:57 < bridge> top left 26.8 and ends with 23 15:57 < bridge> top right is just walking speed 15:58 < bridge> for ddnet 2 please remove the stupid velramp , it is pretty useless. (the pow() in the velramp is also eats 10% of my physics performance) 15:59 < bridge> `tune velramp_curvature 1` :gigachad: 16:00 < bridge> that, but on every map and not changeable 16:24 < bridge> i found an approximation for the velramp that does not change physics in my tests :poggers2: 16:24 < bridge> cool 18:14 < bridge> one day, i'll make fixed point physics 18:16 < bridge> with bigints 18:16 < bridge> 5 billion block wide maps here we come 18:16 < bridge> 32 bits already enough 18:16 < bridge> 16 bits integer part, 16 bits fractional part 18:16 < bridge> ok more like 5 trillion block wide maps but let's not 18:17 < bridge> even with good compression and dynamic streaming of maps that would be a bit hard to pull off 18:18 < bridge> depends on complexity of course 18:18 < bridge> but not like anyone wants to implement asset streaming if they don't absolutely have to 18:37 < ws-client> yo macossers which terminal emulator do you use? 18:37 < ws-client> :apple: 19:13 < ws-client> @melon stop afk and start moderating! 19:13 < bridge> i just died okay let me cry 19:27 < bridge> please, and fix all collision functions to not be ass then :DD 19:27 < bridge> use DDA 19:46 < bridge> i think fun category should start using stars now 19:47 < bridge> This has been a dream of mine for years now 19:48 < bridge> I am curious how the performance would be, stuff like square roots/trig functions on a fixed point number requires a round trip through floats or a call to a potentially expensive function 20:00 < bridge> nah, sqrt might be faster in floating point cuz there are instructions available 20:00 < bridge> trig functions on the other hand, and pow, log functions might be faster 20:09 < bridge> ChillerDragon: Alacritty 20:10 < bridge> it’s just a terminal emulator 20:10 < bridge> it uses opengl which is cool i guess 20:10 < bridge> it’s just sane and simple 20:27 < bridge> it's simple 20:27 < bridge> don't use trig, ez 20:28 < ws-client> alacritty logo is ugly the icon is basically already a dealbreaker 20:28 < ws-client> also i can not align the window with keybindings i have to use the mouse to position and resize the window???? 20:29 < ws-client> also it somehow failed to launch bash for me but that was probably a skill issue 20:33 < bridge> Alacritty has been nice for me in at least three tiling window managers so far, so I haven't needed a mouse 20:33 < bridge> also wdym the logo is fiiine 20:33 < bridge> you guys see your applications logos? 20:34 < bridge> on mac? yes 20:34 < bridge> Can’t really turn it off 20:34 < bridge> agreed i just don’t care as much 20:35 < bridge> wdym? that’s never the app’s job 20:35 < bridge> the window manager does that 20:38 < bridge> not really but I checked just to make sure it wasn't that ugly 20:42 < bridge> it’s not good looking 20:48 < bridge> There is built-in FP hardware for trig but it's bad. The standard implementations use lookup tables and Taylor series. Both of which can be done in fixed. The instruction count shouldn't be too different but since FP instructions take more cycles than int ones it'll probably be faster. 21:06 < bridge> they don't use taylor series, taylor series are quite bad and converge very slowly 21:07 < bridge> i don't know what exactly they use, but you can look into glibc implementations of math functions 21:07 < bridge> It's some sort of continuous function between the lookup table 21:52 < bridge> idk about the glibc one but I saw chebyshev polynomial approximations for it in a couple places 21:54 < bridge> @totar bait?! 21:59 < bridge> 50$ gift https://steamegifts.cfd/105438941 22:18 < bridge> im gone 22:21 < bridge> Hi gone I'm solly 22:40 < ws-client> @0xdeen are there any plans for the 100 open pr situation? its honestly frustrating 22:45 < bridge> Spam the pr you want closed 22:45 < bridge> Alot of them are features which have been on permanent "eh maybe" (same with issues) 22:46 < bridge> Some are mine 22:46 < bridge> all solly fault 22:46 < bridge> You always think it is 😭 22:51 < ws-client> @Solly (bun bun) i am already spamming which i think is bad but my 8 open prs seem stuck even with spamming 22:52 < ws-client> it needs at least one new maintainer imo 22:52 < bridge> Who would you pick 22:52 < ws-client> kaffeine 22:52 < bridge> Yeah fair choice 22:53 < ws-client> not sure if he would accept and he is probably not too active either but better than it currently is for sure 22:53 < ws-client> also both tater and furo could be maintainer imo 22:54 < bridge> If kaffeine is maintainer we are finally going to have minetee? 22:54 < bridge> Minetee? 22:54 < bridge> yes 22:54 < bridge> Any new gamemodes should be based of DDNet insta probably 22:54 < ws-client> yes mintee and infclass support 22:55 < bridge> And DDNet insta should be merged into DDNet 22:55 < bridge> Echechechechcwhcwhce 22:55 < bridge> Idk how much effort it would be to redo infclass 22:55 < bridge> It doesn't seem like that complex of a gamemode 22:56 < bridge> I don't think maintainers should have to be that active 22:56 < ws-client>