01:42 < bridge> i think i’ve sorted a windows desktop out 01:51 < bridge> so i can actually do my project now 02:34 < bridge> But discord only. When bridge to irc 04:20 < bridge> Hello. I'm reacreating my own Dockerfiles for different teeworlds/ddnet modes. 04:20 < bridge> 04:20 < bridge> I noticed that after install there's kinda a lot stuff: 04:20 < bridge> 04:20 < bridge> ``` 04:20 < bridge> /tw/data # ls 04:20 < bridge> announcement.txt censorlist.txt deadtee.png game.png languages race_flag.png wordlist.txt 04:20 < bridge> arrow.png communityicons debug_font.png gui_buttons.png mapres shader 04:20 < bridge> assets config editor gui_cursor.png maps skins 04:20 < bridge> audio console.png emoticons.png gui_icons.png maps7 skins7 04:20 < bridge> autoexec_server.cfg console_bar.png extras.png gui_logo.png menuimages strong_weak.png 04:20 < bridge> blob.png countryflags fonts hud.png particles.png themes 04:20 < bridge> ``` 04:20 < bridge> 04:20 < bridge> Which files i can easily remove and server will stay fully functional? 04:20 < bridge> 04:20 < bridge> Dockerfile, if needed: https://termbin.com/t9cg 04:22 < bridge> ^ Same question for icons: 04:22 < bridge> 04:22 < bridge> ``` 04:22 < bridge> /tw/share/icons/hicolor # tree 04:22 < bridge> . 04:22 < bridge> ├── 16x16 04:22 < bridge> │ └── apps 04:22 < bridge> │ └── ddnet-server.png 04:22 < bridge> ├── 256x256 04:22 < bridge> │ └── apps 04:22 < bridge> │ └── ddnet-server.png 04:22 < bridge> ├── 32x32 04:22 < bridge> │ └── apps 04:22 < bridge> │ └── ddnet-server.png 04:22 < bridge> └── 48x48 04:22 < bridge> └── apps 04:22 < bridge> └── ddnet-server.png 04:22 < bridge> 04:22 < bridge> 8 directories, 4 files 04:22 < bridge> ``` 04:23 < bridge> Hello. I'm recreating my own Dockerfiles for different teeworlds/ddnet modes. 04:23 < bridge> 04:23 < bridge> I noticed that after install there's kinda a lot stuff: 04:24 < bridge> 04:24 < bridge> ``` 04:24 < bridge> /tw/data # ls 04:24 < bridge> announcement.txt censorlist.txt deadtee.png game.png languages race_flag.png wordlist.txt 04:24 < bridge> arrow.png communityicons debug_font.png gui_buttons.png mapres shader 04:24 < bridge> assets config editor gui_cursor.png maps skins 04:24 < bridge> audio console.png emoticons.png gui_icons.png maps7 skins7 04:24 < bridge> autoexec_server.cfg console_bar.png extras.png gui_logo.png menuimages strong_weak.png 04:24 < bridge> blob.png countryflags fonts hud.png particles.png themes 04:24 < bridge> ``` 04:24 < bridge> 04:24 < bridge> Which files i can easily remove and server will stay fully functional? 04:24 < bridge> 04:24 < bridge> Dockerfile, if needed: https://termbin.com/t9cg 04:25 < bridge> <0xdeen> The server only needs the executable itself, autoexec_server.cfg and the map you want to run. announcement, censorlist, wordlist are only required if you have set something in them 04:25 < bridge> Yes, i know, but what about that all icons, png's and other? 04:26 < bridge> <0xdeen> not required 04:26 < bridge> Sounds too? 04:26 < bridge> not required 04:26 < bridge> Thanks. 04:26 < bridge> What do you think about merging my Dockerfile in official github repo? 04:52 < bridge> <@713443740898951185> 04:52 < ws-client> i thought i saw you patch this line but i cant find it anymore https://github.com/ddnet/ddnet/blob/09203aee78dd19b8d81c037a0e45d77e875c426c/src/game/server/gamecontext.cpp#L4436 04:53 < ws-client> returning a ``new CInfClasGameContext`` here 04:53 < ws-client> did i make that up? 04:53 < ws-client> btw @ninekeem i also made a dockerfile once if you are interested https://github.com/BanBansNet/ddnet-docker/blob/master/ddnet-simple/Dockerfile 04:54 < bridge> nice one 04:54 < bridge> but i like mine more ;) 04:54 < ws-client> ye classic 04:54 < bridge> :D 04:54 < ws-client> because u have mysql? 04:55 < bridge> i don't like that fact you download maps while build image 04:55 < ws-client> thats just to get it quickstarted its designed to be simple to use 04:55 < bridge> but wget sources is interesting 04:55 < bridge> why alpine 3.18? 04:55 < bridge> ah, last year 04:55 < bridge> got it 04:56 < ws-client> ye old stuff 04:57 < bridge> and wget will not work :) 04:57 < bridge> 404 04:57 < ws-client> i rage quitted on it when i verified that i really do not like containers xd 04:57 < bridge> i prefer local downloaded sources 04:57 < ws-client> 404? woah how so 04:57 < bridge> so how do you host your own servers? 04:57 < ws-client> bare matle 04:57 < bridge> don't say you use screen/tmux 04:57 < bridge> pls… 04:57 < ws-client> i use screen/tmux 04:58 < bridge> … 04:58 < bridge> no comments 04:58 < ws-client> i mean i tried kubernetes https://github.com/BanBansNet/kubernetes/blob/master/ddnet.yaml 04:58 < bridge> i think i am only one who enough crazy to use damned kubernetes 04:58 < ws-client> i rly did not like it 04:58 < ws-client> kog does koobernetes 04:58 < bridge> damn 04:59 < bridge> did you inspired by mine dockerfiles or wrote it by yourself? 05:00 < ws-client> its been so long i think i looked at ddnet's dockerfile but its not alpine so its mostly from scratch 05:00 < bridge> ok 05:59 < bridge> yeah and it just works 07:39 < bridge> normin 0.6.5 enjoyers and others 10:18 < bridge> gm 10:31 < bridge> gm 13:17 < bridge> @learath2 There's little whoopsie daisy with da compiler, 11 registers isn't enough, right now I do smth like 13:17 < bridge> ```rust 13:17 < bridge> let register_1 = allocate_reg(); 13:17 < bridge> generate(expr.right, ®ister_1); 13:17 < bridge> let register_2 = allocate_reg(); 13:17 < bridge> generate(expr.left, ®ister_2); 13:17 < bridge> 13:17 < bridge> ...do other stuff 13:17 < bridge> ``` 13:17 < bridge> and it all breaks if I try to compile something like 13:17 < bridge> ```rust 13:18 < bridge> let a: u8 = 1 + 1 + 1 + 1 + 1 + 1 + 1; 13:18 < bridge> ``` 13:18 < bridge> 13:18 < bridge> What's the better way to generate the expressions :\ 13:53 < ws-client> is this a bug in chat.zillyhuhn.com or do you use a copyright unicode in your code? xd 13:53 < ws-client> https://zillyhuhn.com/cs/.1727178818.png 13:54 < bridge> it's `®ister_2` 13:54 < bridge> oh 13:54 < bridge> hm 13:54 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288106263812771891/image.png?ex=66f3f9f1&is=66f2a871&hm=8f87233227928f433af390074eec7c990a49c42afd16bd1f774acc01efe2def1& 13:54 < bridge> & r e g i s t e r _ 2 13:54 < ws-client> ah ok thanks patiga 13:54 < ws-client> yea i dont see images @milkeeycat xd 13:54 < bridge> too bad 13:54 < bridge> ez @milkeeycat :p 13:55 < ws-client> would have been funny if your code had that (R) symbol xxxD 13:55 < ws-client> is ® html? 13:55 < bridge> it will panic on the first line of lexer probably xdd 13:55 < ws-client> lol < test 13:55 < ws-client> ok wtf it encodes all the html thingies xxxxD 13:56 < bridge> what is ws client:thonk: 13:56 < bridge> ``` 13:56 < bridge> Failed to parse char Ð 13:56 < bridge> ``` 13:57 < bridge> you need to replace ` & ` with `& a m p ;`, ChillerDragon 13:57 < bridge> (remove spaces) 13:57 < ws-client> @oxyzo. https://chat.zillyhuhn.com/ some homemade web frontend for irc 13:57 < ws-client> @heinrich5991 yea i guess so 13:57 < bridge> :brownbear: 13:57 < ws-client> i wonder if there is any case where i want this to expand. I dont think so 14:56 < bridge> use stack then xd 14:57 < bridge> or u can also try doing constant inlining after the parsing step 14:57 < bridge> like a tree walker if it sees 2 integers `1+5` evaluates it to `6` 14:58 < bridge> Why close epic pr @milkeeycat 14:58 < bridge> noone looked at it xd 14:58 < bridge> if anyone needs 0.7 servers, they will make a pr 14:58 < bridge> if anyone needs 0.7 servers in lan tag, they will make a pr 14:58 < bridge> if anyone needs 0.7 servers in lan tab, they will make a pr 15:00 < bridge> Yea sadly ddnet pr reviews are not too active right now. But you gotta stay strong. You took so much time to debug it. Give the maintainers time to respond. 15:01 < bridge> you can make it 10x better :p 15:01 < bridge> Look at poor bubble butt who stayed strong for 9 months without a single response comment https://github.com/ddnet/ddnet/pull/7737 15:02 < bridge> :pepeW: 15:02 < bridge> Milk, we need them lan servers 15:02 < bridge> chiller waited two years to get his pr merged 15:03 < bridge> and more 15:03 < bridge> in teeworlds og repo 15:04 < bridge> @milkeeycat reopen the pr, I believe in you. 15:08 < bridge> it's BubbleBut 15:08 < bridge> :kek: 15:10 < bridge> Bubble butt! 15:11 < bridge> 🫧 15:25 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288129062312743012/RDT_20240924_1524522140989270181646160.jpg?ex=66f40f2c&is=66f2bdac&hm=c6a1afbbb9b2458d784a4217af1ed48a9bcf9c4b30d7f5d22fb15615778fffe9& 15:27 < bridge> Let me think a little. So the issue is that for a long expression for each op you descend into you allocate 2 registers, running out of registers? 15:28 < bridge> The naive way out of it is to get stack scratch space. If no registers left, allocate on the stack 15:28 < bridge> leave the notion of registees for after, just do allocations and at a later stage do "register allocation" and the others on the stack, loading as needed 15:29 < bridge> &bull &ndash $bull :greenthing: 15:29 < bridge> and a simple opt is to analyze the most ysed allocas 15:29 < bridge> rank them 15:29 < bridge> and put top on reg 15:29 < bridge> easier to do in SSA 15:29 < bridge> ok cool 15:29 < bridge> 😬 15:29 < bridge> and a simple opt is to analyze the most used allocas 15:30 < bridge> He is generating raw assembly, it's important where his data lies because not all instructions can operate at every width and both with mem and regs 15:30 < bridge> But yes, that is the actual proper solution. You do one pass with just abstract allocations, then in another pass you put stuff into registers if available 15:31 < bridge> But now you can't just emit assembly as you go like you used to 15:31 < bridge> this is why its so important so hace a simple ir in SSA form, cuz thats where u optimize this way easy, indeed in the register allocation pass if you see a alloca gets its addresd taken for example then it cant be put in a reg 15:31 < bridge> this is why its so important so to have a simple ir in SSA form, cuz thats where u optimize this way easy, indeed in the register allocation pass if you see a alloca gets its addresd taken for example then it cant be put in a reg 15:31 < bridge> shit mobileee 15:32 < bridge> @milkeeycat another copout is reducing constant expressions at compile time 15:32 < bridge> ill do constant folding later 15:32 < bridge> i just don't like that it takes bajillion of registers for nothing 15:32 < bridge> i beg u, do ssa form ir 15:32 < bridge> u will love optimizing it 15:33 < bridge> Sorry boss, only janky engineering allowed here 15:33 < bridge> xd 15:33 < bridge> :pepeW: 15:34 < bridge> with ssa you can find the canonical form and see patterns easily 15:34 < bridge> casually introducing the word Tee to a sentence. 15:34 < bridge> this reminded me i have to retake edlang 15:35 < bridge> but i stopped at multiple packages xd 15:37 < bridge> @milkeeycat you have an AST right? 15:37 < bridge> ye 15:40 < bridge> I guess the issue is your recursive approach. How about the shunting yard algorithm to first figure out the actual order you'll evaluate in? 15:42 < bridge> Nah that's not good enough. I'll think about it more over tea 15:42 < bridge> i also can use passed register for one side 15:42 < bridge> instead of allocating a new one 15:43 < bridge> and that's what i actually have rn in my codebase 15:43 < bridge> but there were problems with dat 15:47 < bridge> Yeah the issue is recursiveness. Travel down the tree first, do the operation upside down, then 2 registers will be enough for all expressions 15:48 < bridge> Will it solve the problem if I just return a register instead of passing it 15:50 < bridge> As in let generate return a register? It doesn't change much, the issue is that your code travels all the way down the tree allocating registers 15:52 < bridge> How about passing 2 registers all the way down and info about which side of the expression we are currently generating? 15:53 < bridge> Allocate 2 registers at an expression root. LHS, RHS. Leaf expressions don't allocate their own of these but use either rhs or lhs as their result depending on which side they are on 15:53 < bridge> it's actually not a skill issue, my language just doesn't allow long mafs expression to not ruin code readability 15:55 < bridge> Yeah no this doesn't work either. You just cant generate assembly traversing a tree in in order because that's not the order of operations 15:55 < bridge> Yeah no this doesn't work either. You just cant generate assembly traversing a tree in order because that's not the order of operations 15:55 < bridge> Well you can if you have infinite registers 16:02 < bridge> @milkeeycat oh one more idea, idk if it works but tea inspired me. The issue is that you keep creating extra registers for the result of the expression but that result doesnt come until much later. How about you pass down the result register as the lhs or rhs? 16:03 < bridge> That wouldn't fix the issue, but it would be less wasteful 16:04 < bridge> I think that's what I have now 16:04 < bridge> https://github.com/MilkeeyCat/meraki/blob/6633eddd08db542002dd010b76df7a33f3c5a83d/src/codegen/codegen.rs#L396-L401 16:04 < bridge> Can't be lhs or rhs since I'm guessing generate generates an immediate load to that register right now for constants, so you can reuse only on the expression side 16:07 < bridge> Eh, yes that's what you are doing indeed. So I guess that's not enough 16:08 < bridge> I just had a problem that the destination size could be different than expression type size and it would generate not valid assembly 16:09 < bridge> so I started looking maybe there's another way 16:09 < bridge> but I can just simply change the size of it and that's it 16:10 < bridge> and it wont be as bad as this 16:12 < bridge> I was thinking you were doing this, this is far too wasteful of registers 16:22 < bridge> :pepeW: 16:23 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288143873184628927/20240924_162346.jpg?ex=66f41cf7&is=66f2cb77&hm=7f3cb64b60dc68810ba5377c44b018fdc81477af677aa00d0117afdd76bf81a7& 16:35 < bridge> @milkeeycat here you go https://en.wikipedia.org/wiki/Sethi%E2%80%93Ullman_algorithm, I had to dig up some very old notes to remember whomst wrote this 16:37 < bridge> This is the olden way of doing this properly, you traverse the expression, figure out which side needs how many registers in total, then traverse down the side that needs the most registers all the way down, emit assembly on your way up 16:39 < bridge> Does ddnet-saves.txt only contain savegames created on official DDNet servers? 16:58 < bridge> Any server. 17:05 < bridge> furo: Ah, okey! Do you have any idea why the file isn't being created for me? 17:42 < bridge> No clue, it's works fine for me. Is there any errors? 17:48 < bridge> wow they actually did it 17:48 < bridge> https://github.com/WinampDesktop/winamp 17:49 < bridge> dont have to rely on the leaks anymore 17:50 < bridge> wait LMAO 17:50 < bridge> 17:50 < bridge> "open source" 17:50 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288165661193015296/image.png?ex=66f43142&is=66f2dfc2&hm=b8bf866ae76c62248c1e998e69b2038d8e4f86e6e8422de4199794ca7d3836fe& 17:51 < bridge> lmao - why publish it to github then 17:51 < bridge> clout 17:52 < bridge> this is the same group responsible for the NFT fiasco 17:52 < bridge> and for inflating their beta tester numbers 17:52 < bridge> lots of shady things going on 18:03 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288168835706257549/Screenshot_2024-09-24_at_16.png?ex=66f43437&is=66f2e2b7&hm=05fbf6a3d398ed3e31c99e8c93a1b3d55439a93ef42e3df9dbfc3fadb2768d1e& 18:03 < bridge> lol 18:03 < bridge> huuuuh why 18:03 < bridge> i've been using llama for a while now 18:05 < bridge> ??? There is a fork button ??? Do I get picked up by the FBI if I click it? 18:06 < bridge> https://github.com/WinampDesktop/winamp/issues/6 18:14 < bridge> No 18:14 < bridge> some weird shady group will have to sic their lawyers on you 18:14 < bridge> with all their $4.53 18:15 < bridge> though if this was valid under GitHub's TOS then it may be taken down via. DMCA notice 18:15 < bridge> which it isn't. so... 😁 18:49 < bridge> i thought compilers have dynamic amount of registers while generating 18:50 < bridge> https://llvm-mos.org/wiki/Imaginary_registers 18:50 < bridge> ah lol 18:50 < bridge> they even have wiki page for it 18:51 < bridge> i have as many registers as amd64 architecture 18:51 < bridge> I'll check this, thanks a lot! 18:53 < bridge> *while cdoe generating* 18:53 < bridge> u can have as many as u need to 18:53 < bridge> i have nothing fancy 18:53 < bridge> i just raw dawg assembly 18:54 < bridge> and I can use only amd64 registers :p 18:54 < bridge> and i can crash ur compiler by giving this as source code or what 18:54 < bridge> yep :greenthing: 18:55 < bridge> perfect combination xd 18:55 < bridge> you can also crash it by using non ascii chars 18:56 < bridge> gg 19:01 < bridge> This approach requires you to optimize after, he has no optimization passes at all for now 19:03 < bridge> It requires more than just normal optimization passes, it requires an intermediary step like an SSA where you can merge all of these unnecessary assignments, then another where registers are allocated all before emitting any assembly 19:05 < bridge> It's first time in a long time I understood what's written on wikipedia page xD(probably because there isn't a wall of mafs) 19:07 < bridge> These olden algorithms were so much more pleasant to understand. Nowadays it's all post quantum black magic 19:08 < bridge> Yep, so such neanderthals like me can't understand anything after 2nd sentence :\ 19:09 < bridge> I heard the modern approach to this is graph coloring. Good luck digging through dozens of pages of abstract mathematical theory to figure out how to properly color a graph with 11 colors optimally 19:09 < bridge> I love when there's a simple example of how the thing works xd 19:41 < bridge> no 19:42 < bridge> @learath2 can you help with this one pls 19:42 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288193878456799324/image.png?ex=66f44b8a&is=66f2fa0a&hm=528d27fa1d5932122d993993d2577edb6694307e6208d7b73a56e38e481205ea& 19:42 < bridge> https://github.com/Jupeyy/ddnet-web/tree/testing_nextjs 19:43 < bridge> so basically 3*5 =15 then 15+1 =16 19:44 < bridge> thx homie xD 19:46 < bridge> like dis? 19:46 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288194816353636485/image.png?ex=66f44c69&is=66f2fae9&hm=aaca6525aaee257955a94290ff610bc60e90b8a9d46062635fedbf5db746eb76& 19:46 < bridge> no matter what you do, if u dont use shaders and no texture arrays it won't look the same. 19:46 < bridge> 19:46 < bridge> The reason the pixels are cut off is bcs mipmaps change the texture size. 19:46 < bridge> OpenGL 1.x is really only there for some ancient pcs, otherwise it should not be used 19:48 < bridge> yeah that's fair enough. I have no idea why it was the default for me 19:54 < bridge> we mostly wait for the secure server connection rn, once that is done we can do the rest ig 19:56 < bridge> hey @mpft and @blaiszephyr sorry for ping but we spoke yesterday about me getting ddnet ported to xbox somehow. as i said i completely new but i want to do it as a project and hopefully learn off of it. i managed to get my hands on a windows pc and can start. what shouldi start off with learning? 19:57 < bridge> hey @mpft and @blaiszephyr sorry for ping but we spoke yesterday about me getting ddnet ported to xbox somehow. as i said i completely new but i want to do it as a project and hopefully learn off of it. i managed to get my hands on a windows pc and can start. what should i start off with learning? 19:57 < bridge> ah fuck i think it was learath no melon 19:57 < bridge> ah fuck i think it was learath not melon 19:57 < bridge> i can’t remember 😭 20:04 < bridge> LGTM 20:05 < bridge> You should start by learning some directx 20:05 < bridge> okay awesome thank you 20:05 < bridge> Well actually, if you are completely completely new, you need to learn C++ first 😄 20:06 < bridge> oh yeah that’s probably makes sense 😭 20:06 < bridge> luckily i’m in no rush for this 20:06 < bridge> and will happily learn 20:07 < bridge> https://www.learncpp.com/ this is a pretty acceptable resource 20:38 < bridge> UWP is a lot more involved than just C++. You may be working with IDL and a bunch of other random Microsoft bullshit. I would recommend making a basic UWP in C++ that just runs on your desktop to get a feel for things 20:39 < bridge> I have to say considering there is no current DirectX backend (and such a thing is not really trivial to implement well) maybe you should start off with another project. But if you have free time and willpower I think you can do it for sure 20:39 < bridge> i have a lot of free time at the moment lmfao 20:39 < bridge> The UWP community Discord server is full of nice folks who would be happy to help you as long as you’re composed, honest, and don’t get mad 20:40 < bridge> You will need to learn a lot 20:40 < bridge> ChatGPT will not be able to help you with any large-scale problems (the biggest of which I anticipate being a DirectX backend) 20:44 < bridge> @learath2 btw, I still load integer literals into a register before using it, so "Sethi–Ullman algorithm" is not really possible, do you think it's worth to calculate bigger subtree and generate code for it first? 20:46 < bridge> my plans been foiled 20:46 < bridge> joking 20:46 < bridge> but cool i’ll have to join their server 20:46 < bridge> sounds good 20:51 < bridge> my roomba has started speaking german 20:52 < bridge> wtf ? why ? 20:52 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288211367215759474/image.png?ex=66f45bd3&is=66f30a53&hm=a4cacfccb9e1bab0606f2d3bd54fd952b31a30ddaf573d1c918d4367d6842449& 20:52 < bridge> "BATTERIE LEER" 20:52 < bridge> Same reason they invalidated password auth over git years ago 20:52 < bridge> Security 20:52 < bridge> nvm, ill do it the proper way 20:55 < bridge> it's just getting boring to have heavy authentication everywhere 20:58 < bridge> <01000111g> I, goomba also speak german 21:02 < bridge> Well you can augment it, you just won't have 0 nodes 21:37 < bridge> yey, unpacking of integers work 22:21 < bridge> furo: do you own ddstats.tw? 22:21 < bridge> https://github.com/ddnet/ddnet/issues/9032 22:33 < bridge> Yeah, I do. The skins are fetched from other sources like skins.tw, but I'll try resizing them. 22:34 < bridge> Thanks 23:01 < bridge> Bro who is making 4k skins 23:01 < bridge> it's time for 8k skins 23:02 < bridge> I mean they are fine for demo renders or smth like that 23:02 < bridge> i want 10^80 skin 23:02 < bridge> The universe as skin? 23:02 < bridge> epyc 23:02 < bridge> exactly 23:02 < bridge> 4k skins are fine if u have them saved locally 23:02 < bridge> skin even contains all the maps that can exist 23:02 < bridge> the loading is what causes lags 23:03 < bridge> nah, they hog vram like crazy as well 23:03 < bridge> hm would it? 23:03 < bridge> well yea 23:03 < bridge> but many ppl now have cards with 12-24GB vram 23:03 < bridge> it really isn't a concern for most gamers 23:03 < bridge> definitely not right 😅 isnt the max map size 32000^2 23:03 < bridge> cry in 4gb and 10y old card 23:03 < bridge> rip 23:03 < bridge> zip bomb ❌ 23:03 < bridge> skin bomb ✅ 23:03 < bridge> 2gb gamer here 😬 23:04 < bridge> i saw 23k or 20k sth not long ago 23:04 < bridge> cat laughing emoji 23:04 < bridge> but yeah, it's due to int32 limit 23:04 < bridge> that's tough lil bro. 23:04 < bridge> :kek: 23:04 < bridge> for years i had just a 2GB 1050 23:04 < bridge> but then u need 4k mapres too 23:05 < bridge> 64mb of ram 😎 23:05 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1288244914383224992/4qMTpfnntq9hsyzE1wFA_ZjgIwUrTey0i7cLZxSk0II.jpg?ex=66f47b12&is=66f32992&hm=a2e2b12cafcdc59777abcce055267ab437806ab256c157930ca92ccdafb92e3d& 23:05 < bridge> :justatest: 23:06 < bridge> that guy cannot play maps bigger than 100x100 😄 23:06 < bridge> cant even install ddnet 23:08 < bridge> mac os 9 ddnet port 23:08 < bridge> :justatest: 23:08 < bridge> <01000111g> we chillin 23:08 < bridge> <01000111g> https://cdn.discordapp.com/attachments/293493549758939136/1288245733069160449/image.png?ex=66f47bd5&is=66f32a55&hm=9cf92fb15af903fe8169c3c8292b132fd3e157b2630ff1b4c3987dbf1e31142c& 23:08 < bridge> <01000111g> Can i run TW with that? 23:10 < bridge> Hmmmm I'm gonna go with No 23:11 < bridge> WTF 23:11 < bridge> Where did you get that screenshot from 23:11 < bridge> 😬 23:13 < bridge> <01000111g> from my terminal! (its a cluster i have access to for research) 23:13 < bridge> are u foldin 23:13 < bridge> You mean researching ddnet.. nice 23:13 < bridge> researching ddnet max fps 23:13 < bridge> RIP SLI 23:13 < bridge> i dont think it would be that good 23:14 < bridge> yeah defs not 23:14 < bridge> <01000111g> yeah 23:15 < bridge> Maybe with a very high resolution monitor 😄 23:15 < bridge> <01000111g> Its not made for that task, but its banger for PyTorch etc 23:15 < bridge> right 23:15 < bridge> <01000111g> I bet i could solve the optimal way through linear with it kek 23:15 < bridge> do it 23:16 < bridge> is there a way i can change my server config to allow for connections from older clients? 23:16 < bridge> blud is running 700b llama 23:17 < bridge> <01000111g> would be fun, but rather then bruteforce i would do some kind of evolutionary algorithm, tbh way too lazy to implement it, maybe if i have a lot of free time and get skilled smh xdd 23:18 < bridge> specifically 0.7 versions of teeworlds 23:29 < bridge> Playing DDNet on integrated GPU (although seems fine) 23:31 < bridge> when doom in ddnet