00:50 < bridge> Why are ghost files loaded when loading a map in the editor? 00:50 < bridge> Why are ghost files loaded when loading a map in the demo viewer? 01:00 < bridge> demo viewer is haunted 👻 01:01 < bridge> yea the ghosts haunt me so much that i have to wait several minutes for them to load on a specific map 👻 01:01 < bridge> that's a very haunted map 👻 01:03 < bridge> maybe my drive is just really slow. several thousand shouldn't take that long though 02:05 < bridge> demos store entire maps right? 02:12 < bridge> https://unlicense.org/ 02:12 < bridge> lol nice license 02:18 < bridge> ah it can do both i guess 10:27 < bridge> wow nice error. very helpful 10:27 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428298394677415977/image.png?ex=68f1fe09&is=68f0ac89&hm=3f94d85f57b55f5b94c398982556bfc80a9d315cdb53d726fb69abbd30837f63& 10:27 < bridge> thanks 10:29 < bridge> huh corrupted file? https://github.com/ddnet/ddnet/blob/60ac315511b3233082bf9336e3d89c45b11e71e5/src/engine/shared/demo.cpp#L868 10:31 < bridge> there are multiple things that could cause them, from invalid keyframes, invalid headers to empty keyframes 10:37 < bridge> uhhh 10:37 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428300770058571887/image.png?ex=68f20040&is=68f0aec0&hm=efe3cb45242f8f54e3a55581d5a3ef7f29b272c2860d2103ed8b27f90e94f4a9& 10:42 < bridge> why is one dialog red and the other not 🤔 10:44 < bridge> dunno 10:44 < bridge> ask robyt 10:45 < bridge> lets not annoy robyt with some minor inkonsitencies in UI 10:56 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428305682376949801/image.png?ex=68f204d3&is=68f0b353&hm=55a4fb12a9f826c1476c6285b0fda859fbdf3d1ca68ae530f9ac4157bc6a1f2f& 10:58 < bridge> @jupeyy_keks i got access to epyc 10:59 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428306220720193617/image.png?ex=68f20553&is=68f0b3d3&hm=e014d43528edfb39fa140d637b0f66134cca21487ca119c4e16c5b993a89e067& 11:00 < bridge> nice 11:09 < bridge> we have like 12 of these 11:09 < bridge> i wonder how much money is this 11:09 < chillerbot> You can see donors and how to donate here https://ddnet.org/funding/ 11:09 < bridge> xd 11:09 < bridge> good bot 11:09 < bridge> :heartw: 11:10 < bridge> I need a new heart, due to too much coffee consumption. Do you know where I find a donor? 11:10 < bridge> aww bad bot 🙁 11:10 < bridge> :C 11:11 < bridge> i would **not** like to donate 11:11 < bridge> i cash 11:11 < bridge> it's all about the money 11:11 < bridge> i forgot the triggers 11:11 < bridge> no u 11:12 < bridge> epyc 13:21 < bridge> anyone wanna review this? I am also annoyed by it. https://github.com/ddnet/ddnet/pull/9931 13:22 < bridge> oh it changes snapinput, ig that's kinda ugly 15:45 < bridge> closing integers in favour of a non-merged PR 15:45 < bridge> 15:45 < bridge> such chiller! 15:45 < bridge> now heinrich has to close the PR 15:45 < bridge> its the third law of greed 15:50 < bridge> Thx Jupey :deen_star: 15:50 < bridge> closing issuesin favour of a non-merged PR 15:50 < bridge> 15:50 < bridge> such chiller! 15:50 < bridge> closing issues in favour of a non-merged PR 15:50 < bridge> 15:50 < bridge> such chiller! 15:52 < bridge> Doesn't seem like a complex change 15:58 < bridge> It’s all mind games 17:19 < bridge> you talked about wanting to keep the quad shader values lower. Wouldn't I need to scale the values up in the shader, or would you somehow scale them up before doing the draw call? 17:21 < bridge> nevermind there is already a conversion in place 17:21 < bridge> ``` 17:21 < bridge> FinalPos.x = FinalPos.x / 1024.0 + gQuadBO.gUniEls[TmpQuadIndex].gOffset.x; 17:21 < bridge> FinalPos.y = FinalPos.y / 1024.0 + gQuadBO.gUniEls[TmpQuadIndex].gOffset.y; 17:21 < bridge> ``` 17:22 < bridge> but why is it only present in vulkan 🤔 17:23 < bridge> ~~but why is it only present in vulkan 🤔~~ 17:36 < bridge> where is account? 18:13 < bridge> does ddnet cache the shaders somewhere? I have a weird bug where every vulkan shader does not load even tho I only touched one 18:14 < bridge> delete that weird file sha smth in build dir 18:14 < bridge> but normally it should noticed source changes 18:14 < bridge> did u maybe change them in build dir? 18:20 < bridge> yes, I probably did, I also did another huge nono 18:20 < bridge> deleted the build dir and full clean reconfiguring rn 18:41 < bridge> @jupeyy_keks for you :owo: 20:12 < bridge> if you want to make prs for me now start with aggounts 20:15 < bridge> Idk what aggounts are :justatest: guess I don't want to do more PRs for you if the next topic is accounts 20:15 < bridge> i know 20:15 < bridge> we never touch the important stuff 20:15 < bridge> :saddo: 20:17 < bridge> My issue is that I am out of the loop there, Learath told me there is a proposal how you want to handle it with quic and so on. Do you have a link for me so I can at least read into it? 20:17 < bridge> yeah read into quic hehe 20:17 < bridge> My issue is that I am out of the loop there, @learath2 told me there is a proposal how you want to handle it with quic and so on. Do you have a link for me so I can at least read into it? 20:18 < bridge> hm :/ 20:18 < bridge> :tear: 20:20 < bridge> no map shaders thnks to me 20:20 < bridge> no accounts thanks to lea 20:20 < bridge> no new stuff thanks to heino 20:20 < bridge> no new website thanks to deen 20:20 < bridge> no microservices thanks to robyte 20:20 < bridge> 20:20 < bridge> 20:20 < bridge> just replace all of us 20:20 < bridge> and ddnet gets some new vibe shit 20:26 < bridge> good that all of these are wrong 🙂 20:27 < bridge> no 20:28 < bridge> - I implemented a new shader part for 19.4 for example 20:28 < bridge> - no accounts is just true 20:28 < bridge> - we get new stuff, like new speedtiles for example, but we have other candidate tiles 20:28 < bridge> we wont get map shaders as long as i have write access to ddnet 20:28 < bridge> new stuff does not mean new tiles 20:29 < bridge> what do you have against new shaders/map shaders? 20:29 < bridge> if I'd write a tee shader, would it be automatically rejected? 20:29 < bridge> "new" speedtile 20:29 < bridge> we are stuck with a shit network system that 2 ppl in this project understand 20:29 < bridge> we are stuck with a brocken map format 20:29 < bridge> we are stuck with some shit rendering system that is literally from 1997 20:29 < bridge> sry 20:29 < bridge> i meant shaders provided by the map 20:30 < bridge> not new shaders for map rendering 20:30 < bridge> yeah you mean like the water shader I made 20:30 < bridge> just baked into the map 20:30 < bridge> any kind of arbritrary shaders 20:31 < bridge> and no accounts is true in multiple senses 20:31 < bridge> we most likely would not merge a chat based one either 20:31 < bridge> bcs someone is always unhappy 20:32 < bridge> and old clients... i still have not heard from any admin why we need to support clients from 2010 20:32 < bridge> that are probs full of vulnerables 20:32 < bridge> that alone would be a reason to not allow them, so ppl stop using them 20:33 < bridge> I always thought we have some usage stats and remove really unused clients? 20:33 < bridge> and most modding support sucks too, bcs it's not general enough to not make ddnet an ugly mess 20:33 < bridge> and lua sucks ass 20:33 < bridge> and wasm nobody likes bcs it's too new 20:33 < bridge> too cool 20:33 < bridge> would require a nice rewrite too 20:34 < bridge> python scripts? 20:34 < bridge> he is a python hater 20:34 < bridge> https://tenor.com/view/fbi-gif-2734175389934997010 20:34 < bridge> python scripts actually good 20:34 < bridge> better than lua 20:34 < bridge> really 20:34 < bridge> wow that is so 2099 20:34 < bridge> what if linux distro xy gets incompatible 20:35 < bridge> as much as i love linux, but i think ddnet has outgrown packet managers 20:35 < bridge> updating a game every 2 years is simply retarded 20:35 < bridge> i admit lua is on a similar level from a shit perspective.. but bro 20:35 < bridge> 20:35 < bridge> python 20:35 < bridge> that is already bottom 20:36 < bridge> u cannot go below ground 20:36 < bridge> this aint minecraft 20:36 < bridge> python is nice 🙂 20:36 < bridge> https://m.media-amazon.com/images/I/51MphRM1DIL._UF894,1000_QL80_.jpg 20:37 < bridge> id really love some wasm tbh 20:37 < bridge> :justatest: 20:37 < bridge> yeah but tbh that is also a problem 20:37 < bridge> function based modding 20:37 < bridge> not in our scope 20:37 < bridge> the mod loader has to be done by someone else 20:37 < bridge> but ... but minecraft is written in java 😮 20:38 < bridge> why can minecraft just break shit and everyone loves it 20:38 < bridge> how do they do it 20:38 < bridge> why can we not 20:38 < bridge> is that a serious question 20:38 < bridge> we absolutely could, we just have active maintainers against it 20:39 < bridge> sacret backwards comp ore something 20:39 < bridge> now reason 20:39 < bridge> convince me 20:39 < bridge> good luck 20:40 < bridge> it's hard to convice sb of something that you are not convinced in 20:40 < bridge> then strengthen my position xD 20:40 < bridge> i'd say not everyone loves that 20:41 < bridge> u gotta find rce in client, then we can break compatibility 20:41 < bridge> but they just don't have a say 20:41 < bridge> we need to go further, we need some more content in order to keep the game alive, new features, new tiles, the game is > 15 years old 20:41 < bridge> i dont want to know of how many robyte is aware alone with map validations added 20:41 < bridge> time for map fuzzing again? 20:52 < bridge> I actually really want to fuzz maps and network. I really enjoyed the lecture on fuzzing a couple weeks ago 20:53 < bridge> But I noticed I have like 5 unfinished ddnet things, really need to stop hopping from thing to thing, pick one and finish it 20:54 < bridge> @jupeyy_keks any opinion on #11079 ? 20:54 < bridge> https://github.com/ddnet/ddnet/issues/11079 20:55 < bridge> is it possible to.. finish projects? :justatest: 20:55 < bridge> yes, if you define what a finish state is before starting it and just keep pushing towards that goal 20:56 < bridge> and if you fail to reach that goal, then you're also finished. Otherwise you're just in a pending state 20:56 < bridge> The datafile reader has been fuzzed extensively and should™ be free from any bugs. Fuzzing the map format on a higher level would be pretty slow because you have to restart the entire client for it. Network fuzzing would be useful, because it seems like there's some memory corruption bug somewhere to cause #11046. 20:56 < bridge> https://github.com/ddnet/ddnet/issues/11046 21:01 < bridge> That sounds bad, thankfully probably extremely annoying to exploit for anything harmful 21:01 < bridge> Python is so much heavier and less based than lua 21:01 < bridge> Wow a lua enjoyer in gods year 2025 21:02 < bridge> Cancelled 21:02 < bridge> Based opinion frfr 21:03 < bridge> honestly im bad at both, but my nvim config is pretty clean though 21:04 < bridge> we love lua 21:04 < bridge> lua is awesome 21:05 < bridge> it's my favorite slop programming language 21:05 < bridge> Python is unbased for being object oriented but based for occasionally being functional 21:05 < bridge> Lua is unbased but also based for not being anything normal oriented 21:05 < bridge> Just write something wacky with Nth order functions and metatables 21:05 < bridge> Lua is hashmap oriented 21:06 < bridge> lua is 1 index oriented 21:06 < bridge> lua is whatever you want it to be 21:06 < bridge> no 21:06 < bridge> cpp is 21:06 < bridge> C++ is a bad language but a number of subsets of C++ are decent languages 21:06 < bridge> didnt teero literally make a c file behave like a bash script 21:07 < bridge> xD 21:07 < bridge> c supremacy 21:07 < bridge> derive macros 21:07 < bridge> i like zig, i wish i had the energy to learn it 21:07 < bridge> gimme c with generics 21:07 < bridge> how can you tell you like it 21:08 < bridge> i watched a lot of vids on it 21:08 < bridge> gimme c with type safety :lol: 21:08 < bridge> I like Rust but I'm not learning it exactly because I know I'd like it 21:08 < bridge> and I don't want to go down that path 21:09 < bridge> nothing wrong with lua 21:09 < bridge> wild take 21:09 < bridge> there are still many many things you can only do in C++ 21:09 < bridge> rust metaprogramming and compile time evaluation lags behind and nothing else even competes 21:09 < bridge> nuh uh 21:10 < bridge> I'm pretty sure you too only use some subset of C++ you consider reasonable 21:10 < bridge> not really 21:10 < bridge> i am exposed to a lot of C++ nerds 21:11 < bridge> I am pretty sure we all only _know_ a subset of c++ 21:11 < bridge> 😬 21:11 < bridge> i read about way more than i use 21:11 < bridge> only .1% of cpp users have any kind of mastery over the langauge lol 21:11 < bridge> I bet even most cpp compiler devs don't know everything 21:11 < bridge> you got numbers a bit too big 21:11 < bridge> 0.0001% 21:11 < bridge> yea 21:11 < bridge> i started with a few more 0s 21:13 < bridge> sus definition of mastery i.g. 21:14 < bridge> someone got a cool project to work on? 21:14 < bridge> I know this is an oversimplification but: 21:14 < bridge> python - Installed Size: 67.54 MiB 21:14 < bridge> lua - Installed Size: 1533.21 KiB 21:14 < bridge> i consider mastery having some knowledge or experience with every piece of the language 21:14 < bridge> https://github.com/ChillerDragon/teeworlds-asmr 21:14 < bridge> write a good tee skin shader for 0.6 and 0.7 skins and look if it improves performance :justatest: 21:15 < bridge> it would 21:15 < bridge> don't need to write anything 21:15 < bridge> opensrs ran 5g lte 21:15 < bridge> 100% would improve performance 21:15 < bridge> most cpp devs are either writing pseudo-C or just using smart pointers/algorithms/containers/basic templates 21:15 < bridge> :justatest: 21:15 < bridge> a custom hook shader too 21:15 < bridge> it's like 10% of the language 21:15 < bridge> a better particle shader too 21:15 < bridge> according to a dictionary: 21:15 < bridge> > comprehensive knowledge or skill in a particular subject or activity. 21:15 < bridge> You don't need to know everything of a language in order to accomplish this 21:15 < bridge> why are we arguing about my definition of mastery 21:15 < bridge> c++ is so huge xd 21:16 < bridge> wat xd 21:16 < bridge> sowwy 🙁 21:16 < bridge> Maybe we're arguing about your definition of comprehensive instead 21:16 < bridge> assa stop depression 21:16 < bridge> not today, I have ear pain 21:17 < bridge> k 21:17 < bridge> i was at The conference last month, and my local meetup (<5m from my high school) has a lot of people that write all sorts of different kinds of cpp, have done it by trade for years and years, and still learn new things from a few particular speakers 21:17 < bridge> I love how brittle everything is 21:17 < bridge> you can get along perfectly fine knowing some 10% of the language 21:17 < bridge> but mastery is a different thing 21:17 < bridge> i kinda want to write a x86 -> x86_64 compiler xD 21:17 < bridge> 21:18 < bridge> but dealing with ABI stuff is nearly impossible 21:18 < bridge> but would be cool to just transform a 32bit into 64bit 21:18 < bridge> you can't know everything a language can offer, or do you know everything about the boost library in c++ for example? 21:18 < bridge> lua 2.3M users python 11.3 mil 21:18 < bridge> i absolutely have not mastered the language 21:18 < bridge> Let's look at how many users JavaScript has... 21:18 < bridge> but boost is not a part of the standard 21:18 < bridge> neither is WinAPI or any of the build systems etc 21:19 < bridge> still very useful and common info to have 21:19 < bridge> i think scripting support should have players in mind, not filesize 21:19 < bridge> but doesn't it help you master your task? Rewriting everything from scratch instead is also not what I understand as mastery 21:19 < bridge> Lua is traditional for games! Also DDNet should stay based 21:19 < bridge> boost is shit xd 21:19 < bridge> hot take 21:19 < bridge> you are a C dev 21:19 < bridge> u hate what u dont understand 21:20 < bridge> try out mp11 21:20 < bridge> i hate that it's so bloated xD 21:20 < bridge> it's very modular 21:20 < bridge> and you can get the modules thru cpm 21:20 < bridge> this is actually a hot take ngl xD 21:20 < bridge> you seem to be reading between the lines and interpreting things i am not saying 21:20 < bridge> mastering *the language* is not the same as mastering the task you do that leverages the language 21:21 < bridge> oh btw luajit install size on my system is somehow even smaller?? 21:21 < bridge> and luajit go brr so that's always a good reason to pick it 21:21 < bridge> of course, but comprehensive knowledge about a language includes things besides it, outside of the tellerand 21:21 < bridge> luajit is good but is it up to spec with normal lua yet 21:21 < bridge> it's a few minor versions behind but who cares 21:22 < bridge> your binary size explodes according to my experiences xd 21:22 < bridge> maybe i did something wrong 21:22 < bridge> all depends on what you are doing 21:22 < bridge> templates and such are absolutely gonna balloon binary size 21:23 < bridge> https://github.com/kostya/benchmarks 21:23 < bridge> https://programming-language-benchmarks.vercel.app/ 21:23 < bridge> any loops that get vectorized by the compiler or excessive use of inlines 21:23 < bridge> it's about how you write the code for sure 21:25 < bridge> lmao is this not just C? xD 21:25 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428463850738683934/image.png?ex=68f29821&is=68f146a1&hm=963057fcc4fd4984ad181c90443d9a047eb11038951d6b5982525c6804c42d57& 21:25 < bridge> nimrocks 21:25 < bridge> that benchmark purposely does not consider vm startup times 21:26 < bridge> so yes, python using 100MB RAM can beat c 21:26 < bridge> not impossible that c looses if you make a language that is good in exactly one thing 😉 21:27 < bridge> do we pretend that numpy does not use c bindings? xD 21:27 < bridge> also where are my Fortran benchmarks when I need them 21:28 < bridge> nah you gotta trust the benchmark, python is def faster than C as you can clearly see 21:28 < bridge> pysoron 21:28 < bridge> it whos name shall not be named 21:29 < bridge> Kinda implying that NumPy is actually a good enough library that the algorithms could be valuable outside just Python 21:30 < bridge> on the other side python def. is easier and faster to write than C++ or rust. If you don't care about performance, go with python 21:30 < bridge> hello 21:30 < bridge> before learning python, get a vibe coder 21:31 < bridge> too late, learned python before vibe coding was a thing 21:31 < bridge> digga 21:31 < bridge> jupstar 21:31 < bridge> I get uncomfortable writing anything not statically typed so I'd go for the secret fourth option 21:34 < bridge> it's hard to say bcs python definitely has solid applications but it's not a great first language for people hoping to get intimate low level knowledge, nor does it have that much commercial application 21:34 < bridge> lol python has a looot of commercial application 21:34 < bridge> it's kind of an outcast 21:34 < bridge> you'd be suprised how many shit is running on a fastapi or flask server in python. 21:34 < bridge> it's Kinda used for trading and Kinda used for web backends and aside from data science stuff im' drawing a blank 21:34 < bridge> it's not the first pick for any of that 21:35 < bridge> ml/ai 21:35 < bridge> still? 21:35 < bridge> tensorflow pytorch, yes still 21:36 < bridge> i figured that was all obsolete by now 21:36 < bridge> i want to make teeworlds ai 21:36 < bridge> but i am not vibe enough 21:36 < bridge> yea 21:37 < bridge> i want to make a rust desktop app but they are not making the gui stuff easy 21:37 < bridge> yea 21:37 < bridge> i want that too 21:37 < bridge> but i have no project 21:37 < bridge> i really want to use xilem but it's under such heavy development that anything i build with it i will have to constantly maintain until stability 21:38 < bridge> as much as i love the decoupled model/appstate and view/rendering (with actual gpu) 21:38 < bridge> it seems like the perfect way conceptually 22:22 < bridge> luajit is great though I've made bad experiences with depending exclusively on it 22:23 < bridge> the app I maintain has a plugin system that uses luajit in lua 5.2 compat mode (not my decision, I just took over maintainership) and that's hell because almost no distro ships that 22:23 < bridge> not to mention you're screwed on any platform that luajit doesn't support 22:24 < bridge> so, kids, use lua itself for the api and luajit for the performance 23:40 < bridge> Is the client supposed to crash if it comes across an invalid .demo file? xdd 23:40 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428497841952657490/generated_demo.demo?ex=68f2b7c9&is=68f16649&hm=832f6a11b2c27f4ad749d8419f6647ce8fa8188003c807a34204d7b76bf11583& 23:41 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428497996642914346/image.png?ex=68f2b7ee&is=68f1666e&hm=76388b405db1a3b22323d6a5f9d6498b2ef8780c73a7920ccf91a1f29c317a4a& 23:42 < bridge> definitely not 23:42 < bridge> it doesn't look like the full assert msg 23:43 < bridge> it is 23:43 < bridge> well go try it yourself 23:46 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1428499259509969037/image.png?ex=68f2b91b&is=68f1679b&hm=e0ba12365046a3164b01f9c71fcd31668774f7e54bbc480dbe95fccf3bd9c8fe& 23:46 < bridge> just because it crashes for me does not mean it is supposed to crash 23:46 < bridge> ideally it never crashes 23:46 < bridge> idk what your question is