00:18 < bridge> @jupeyy_keks https://github.com/PCSX2/pcsx2/pull/10179 00:18 < bridge> > Disables Wayland, it's super broken/buggy in basically every scenario. KDE isn't too buggy, GNOME is a complete disaster. 00:18 < bridge> > 00:18 < bridge> > Stupid obsession with CSD in Gnome => inconsistency 00:18 < bridge> > Inability to position windows => window position saving doesn't work, log window attaching (not merged yet) doesn't work 00:18 < bridge> > Hacks in render-to-main because WL craps itself otherwise 00:18 < bridge> > Despite said hacks, game list still glitches after stopping emulation, happens more often in gnome 00:19 < bridge> > NVIDIA just crashes in swap chain creation under Wayland 00:19 < bridge> > Broken global menus 00:19 < bridge> > and many others 00:19 < bridge> > 00:19 < bridge> > Until they sort their s**t out, which is unlikely, since there's been very little progress over the last decade, just keep it disabled. For the Flatpaks, users can re-enable it with flatseal if they really want the crappy experience. 00:19 < bridge> lmao 00:21 < bridge> > Quoting @stenzek 00:21 < bridge> > 00:21 < bridge> > > But Wayland is just broken, and everyone would rather sit around arguing with each other instead of actually addressing the design flaws. 00:21 < bridge> > 00:21 < bridge> > > It's not the first time such a proposal has been put forward. Something that developers need for their applications to work properly on WL (particularly multi-window applications), and it gets vetoed. Every other OS manages this fine. But apparently we're in the wrong for not conforming to some warped view of how applications should be, despite our applications working fine on every other platform. 00:21 < bridge> > 00:21 < bridge> > This is so damn true. Wayland lacks a Linus-style BDFL saying that the kern... the compositor is for applications, and not the other way around. We won't have nice things until Wayland maintainers stop thinking about what their end users must or must not do, and start closing feature gaps with X and compositors/window managers on other OSes. Right now they're reinventing the wheel while making it square. 00:28 < bridge> When I have to scroll to see the ping,i know it was ryo xddd 00:28 < bridge> Xdd for every broken no joke 00:29 < bridge> xd 00:30 < bridge> is wayland that bad? 00:31 < bridge> Under kde it's quite ok. But for some reason there are driver bugs extremely often when i try it out 00:32 < bridge> And it's faster than x11 00:32 < bridge> That's ofc argument for me :brownbear: 00:38 < bridge> But it's true they always take like twenty years to implement stuff. And the maintainers are pretty based. Tearing took like two years until it was merged and then another year to be merged into the graphics drivers 08:11 < bridge> gm 08:11 < bridge> maybe 08:11 < bridge> for early birds 09:14 < bridge> Hewo 09:15 < bridge> wawa 09:38 < bridge> morning 09:43 < bridge> morning 11:23 < bridge> dashcam expensive 11:23 < bridge> grug 11:52 < bridge> hi 11:52 < bridge> https://edgarluque.com/blog/mlir-with-rust/ 11:52 < bridge> new blog post 13:48 < bridge> @learath2 rust does the abs with the asm i told u 13:48 < bridge> https://godbolt.org/z/cTr57beWx 13:51 < bridge> I think that's how llvm has it in general 13:52 < bridge> Gcc with generic x86 or amd generates it that way too 13:52 < bridge> Only for intel does gcc fall back to xor-sub for some reason 13:53 < bridge> they use cmovs 13:53 < bridge> i used cmovl 13:53 < bridge> but i guess both work 13:53 < bridge> maybe cmovs is cheaper 13:54 < bridge> @learath2 what u up to btw 13:54 < bridge> u looking for a job 13:54 < bridge> ? 14:07 < bridge> I'm relearning how to code, after which I plan to apply to stuff yeah 14:08 < bridge> 2 years of only stupid engineering classes has taken a toll on my ability to code 14:11 < bridge> rly? 14:11 < bridge> so its true universities dont produce good coders 14:12 < bridge> but why learn python 14:12 < bridge> u looking for a easy python job? 14:13 < bridge> depends what you look for 14:13 < bridge> It's an insurance policy of sorts. See I need to practice my algorithms a bit, if I do it in Python I also will have a language with much broader job opportunities. Just in case I don't find something in a language I actually like 14:13 < bridge> i rly doubt u lost programming skills 14:13 < bridge> maybe u just need to refresh some syntax from x languages 14:14 < bridge> it's a good scripting tool 🙂 14:14 < bridge> It also depends on how you study. If you are a responsible student you'll have time to keep up with your coding too 14:14 < bridge> what i mean is 14:14 < bridge> university doesnt teach u how to "master" coding, as in what we do here, programming a game 14:14 < bridge> they focus on theorics 14:14 < bridge> so u need to do this by urself 14:15 < bridge> and im sure 80% of students still dont develop autonomy 14:15 < bridge> to study something out of their own will 14:15 < bridge> Well mostly I lost throughput. It's like riding a bike, you don't really completely forget it 😄 14:15 < bridge> its a rare skill i would say 14:15 < bridge> self learning 14:15 < bridge> though, Learath2 did engineering, not informatics study, so that's why he hadn't the opportunity to code that much 14:16 < bridge> i still think the same 14:16 < bridge> no matter the branch 14:16 < bridge> but informatics curriculum are a lot about theory, but the language is just a tool 14:16 < bridge> so they should be good in most language 14:16 < bridge> but they tend not to be 14:17 < bridge> modern universities aren't really that useful anymore anyway 14:17 < bridge> i think its like being a plane driver, u can know all theory but im sure u will be a noob without practice 14:17 < bridge> u get ur tasks u have to do 14:17 < bridge> If you pick computer engineering instead of science it's even worse. I think the courses I took are more suited for people trying to develop the first iteration of smart munitions for the second world war 14:17 < bridge> no time for self exploring 14:17 < bridge> yes that what i mean 14:17 < bridge> u dont do out of ur own 14:17 < bridge> I had a single theoretical computer course 14:18 < bridge> AND locking u to the language the uni uses 14:18 < bridge> is ANTI-exploring 14:18 < bridge> self learning 14:18 < bridge> etc 14:18 < bridge> anyway im just a undergrad kek, if thats the correct word 14:18 < bridge> The word you are looking for is "pilot" 😄 14:18 < bridge> i mean it probs depends on the uni. but to me university was like 90% like school 14:18 < bridge> xd ye i forgot the word 14:18 < bridge> and that's why i hated it xD 14:18 < bridge> Did you even do a bachelors? 14:18 < bridge> i went to 1 year of uni 14:18 < bridge> ok im bad with names 14:19 < bridge> is bachelors the 4 year degree 14:19 < bridge> then i dont have it 14:19 < bridge> what does undergrad mean then 14:19 < bridge> ah it means ur not a doctor? 14:19 < bridge> xd 14:19 < bridge> kek 14:19 < bridge> its just titles after all 14:19 < bridge> Undergraduates are bachelors students that haven't gotten their degrees yet 14:20 < bridge> at no point did school help me improve myself as a programmer 14:20 < bridge> not even 1st year of uni 14:20 < bridge> I think "student" is necessary there 14:20 < bridge> it was all myself 14:20 < bridge> but its true i lack stuff in places 14:20 < bridge> specially theories etc 14:20 < bridge> its weird cuz i llike reading all about it 14:20 < bridge> University successfully made me an objectively worse coder. Thanks university 14:20 < bridge> yeah, you either fit the system or you'll be unhappy.. 14:20 < bridge> That simple 14:21 < bridge> but on university i hated everything 14:21 < bridge> im always sharing stupid wiki articles xd 14:21 < bridge> time for a super AI 14:21 < bridge> that helps you to go you own way 14:21 < bridge> @learath2 did u learn this at uni btw? https://en.wikipedia.org/wiki/ABA_problem 14:21 < bridge> Ofc not. That's too complicated 14:21 < bridge> lol 14:22 < bridge> The only thing we learned about multithreading was semaphores and mutexes 14:22 < bridge> i see 14:22 < bridge> im glad i didnt do uni 14:22 < bridge> xd 14:22 < bridge> that's why they don't focus on any language 14:22 < bridge> the projects u do are probs 14:22 < bridge> and its where u learn to apply what u learnt 14:22 < bridge> They won't teach you one but they will usually pick one and use it 14:22 < bridge> its where u learn most 14:23 < bridge> i was forced to use the languages my profs wanted 14:23 < bridge> u could take optional classes for other languages 14:23 < bridge> no rust right 14:23 < bridge> c 14:23 < bridge> xD 14:23 < bridge> xd 14:23 < bridge> i find rust can be a very good educational language 14:23 < bridge> it makes explaining some stuff easier 14:23 < bridge> i take anything that isnt java and pyson 14:24 < bridge> C is the very best educational language if you intend to make actual software people 14:24 < bridge> Python is great if you want to make theory people 14:25 < bridge> as a teacher, we have no choice to teach C so that students know what's going on under the hood 14:25 < bridge> we also teach some assembly 14:25 < bridge> i think a good route would be: python (to teach basics like loops) -> C (to learn basic memory management, pointers) -> Rust (to learn from all the errors u made coding in C) -> assembly (to learn the true low level bit,s calling conventions, stack frames, etc) 14:25 < bridge> especially regarding memory safety 14:25 < bridge> more over, in engineering school, you have to teach what companies require 14:25 < bridge> Assembly is practically a must imo if there is any course about architecture 14:25 < bridge> yeah 14:25 < bridge> so we don't have 100% freedom either 14:26 < bridge> i learnt so much re-learning assembly lately 14:26 < bridge> imo reverse engineering should be a must have. 14:26 < bridge> 14:26 < bridge> That teaches you how higher level languages implement stuff like vptrs, structures etc. 14:26 < bridge> This requirement practically forced several of my courses to use MATLAB 14:26 < bridge> revverse engineering with debug symbols* 14:27 < bridge> reverse engineering obfuscated software is a different beast 14:27 < bridge> its like a detective magician 14:27 < bridge> This is imo too niche for an undergraduate course. More a graduate elective imo 14:27 < bridge> but fun too xd 14:27 < bridge> well, matlab is nice for control theory, signal processing and image filtering 14:27 < bridge> i only want the elite coders 14:27 < bridge> not for anything else i think 14:27 < bridge> same 14:27 < bridge> i have yet to find someone as motivated as me at work xd 14:27 < bridge> without*, put them in real life situation 14:27 < bridge> and use ghidra 14:27 < bridge> Well what else would you want? 14:28 < bridge> but he wants to do RE without having learn assembly first 14:28 < bridge> that sadly only works if you identify with your work 14:28 < bridge> thats why i said with debug 14:28 < bridge> to learn assembly from generated code 14:28 < bridge> then it's useless? 14:28 < bridge> i guess most of your coworkers are simply not nerdy enough 14:28 < bridge> Even with debug symbols you can't reverse engineer shit if you don't know atleast some assembly 14:28 < bridge> 99% of coders just code at work 14:28 < bridge> i know very few people who look at the assembly generated for their code 14:28 < bridge> its fun to do! 14:28 < bridge> godbolt is godsent 14:29 < bridge> Ghidra with it's amazing decompiler isn't even close to generating perfect C 14:29 < bridge> actually, it's not that hard, the assembly is very human friendly 14:29 < bridge> add is self explanatory 14:29 < bridge> isnt IDA the true state of the art 14:29 < bridge> but its paid 14:29 < bridge> 1 year of university should be reimplementing teeworlds 14:29 < bridge> not enough time 😄 14:30 < bridge> teeworlds was the best educational project for me 14:30 < bridge> there are other courses too 14:30 < bridge> Well compiled code rarely has these very simple instructions. You need to learn patterns to be good at reverse engineering 14:30 < bridge> i bet with you in 10 years u forgot most stuff u know today xDD 14:30 < bridge> RE is detective work 14:30 < bridge> even from your doctor stuff 14:30 < bridge> i admire any good RE engineer 14:30 < bridge> i think i may do RE when im older 14:30 < bridge> xd 14:30 < bridge> and got more time 14:30 < bridge> or if a job makes me do it 14:30 < bridge> -O3 is undecipherable imo, -O0 for literal translation, so starts with easy level 14:31 < bridge> -O3 + LTO + BOLT 14:31 < bridge> why 10 years, i've already forgotten most of my control theory classes 😄 14:31 < bridge> ez 14:31 < bridge> -Ofast -flto 14:31 < bridge> 14:31 < bridge> no grace 14:31 < bridge> lmao 14:31 < bridge> We do some reverse engineering here at ddnet inc. come work with us 14:31 < bridge> i wish 14:31 < bridge> when make ddnet company 14:31 < bridge> there's a job for that in a french cybersecurity company 14:31 < bridge> ah u RE bot clients right 14:31 < bridge> i should do it somewhen 14:31 < bridge> -Ofast -flto 14:31 < bridge> 14:32 < bridge> no mercy 14:32 < bridge> i tried to RE final fantasy 14 14:32 < bridge> actually got a bit out of it 14:32 < bridge> @chairn u probs already told me, but are you a teacher in school or university, or where exactly? 14:32 < bridge> im in an engineering school, but it's part of university 14:32 < bridge> this reminds me i have to write a blog post about my nasm learnings 14:33 < bridge> i c 14:33 < bridge> If you want to sew how annoying -O3 code can get look at how std::to_chars handles an int. It's absolutely unreadable 14:33 < bridge> and i teach, assembly, computer architectures, and chip design 14:33 < bridge> see* 14:33 < bridge> it's probably badly optimized, no? 14:34 < bridge> Iiit's actually not too bad, it far outperforms any naive attempt at it 14:34 < bridge> It's absolutely undecipherable assembly though 14:34 < bridge> did u do arm64 asm? its the same as aarch64 right 14:34 < bridge> after i do a bit ox x64 ill learn arm 14:34 < bridge> after i do a bit of x64 ill learn arm 14:35 < bridge> What flavour of assembly do you teach? 8086? 14:35 < bridge> nah, RISC-V 😄 14:35 < bridge> @chairn https://www.zachtronics.com/exapunks/ 14:35 < bridge> show this game to ur students 14:35 < bridge> to motivate them about assembly 14:35 < bridge> but i could do arm64, x86 or atmel 14:35 < bridge> :justatest: 14:36 < bridge> If I were teaching a beginners class in assembly I probably would go for something like 6502 14:36 < bridge> 8bit 14:36 < bridge> Nice and simple, good old days 14:36 < bridge> No pipelining, no weird cache coherency mess 14:36 < bridge> assembly is lab work part of computer architecture class 14:37 < bridge> do u use a white coat when programming assembly in the lab? 14:37 < bridge> sry i just find it too funny 14:37 < bridge> :owo: 14:37 < bridge> And an amazing platform to teach computer architecture on IMO 14:38 < bridge> Risc-v also sounds like a good idea though 14:38 < bridge> is risc-v 64bit? 14:38 < bridge> it touted having a simple arch right 14:38 < bridge> isa 14:38 < bridge> * 14:38 < bridge> language 14:38 < bridge> whathever 14:38 < bridge> idk the word 14:39 < bridge> RiscV iirc doesn't mandate any size of the address bus 14:39 < bridge> im always confused cuz in nasm a quad word is 64bit 14:39 < bridge> but in risc-v its 128? 14:39 < bridge> i guess its cuz x86 has 8bit "words" 14:39 < bridge> x64* 14:39 < bridge> I remember there is even 128bit variants of riscv but chairn would know better 14:39 < bridge> oh yeah 14:39 < bridge> hwo was it named 14:40 < bridge> CHIRP 14:40 < bridge> I only dabbled in riscv to mess around with the ISA 14:40 < bridge> mentioned in rust pointer provenance 14:40 < bridge> ok maybe its not chirp? 14:40 < bridge> i forgot name 14:41 < bridge> What are you looking for? 😄 14:41 < bridge> let me look at my urls 14:42 < bridge> found it 14:42 < bridge> its CHERI 14:42 < bridge> https://www.cl.cam.ac.uk/research/security/ctsrd/cheri/ 14:43 < bridge> > CHERI is a hybrid capability architecture in that it is able to blend architectural capabilities with conventional MMU-based architectures and microarchitectures, and with conventional software stacks based on virtual memory and C/C++. This approach allows incremental deployment within existing software ecosystems, which we have demonstrated through extensive hardware and software prototyping. 14:43 < bridge> > 14:43 < bridge> > CHERI interacts with the design of the full hardware-software stack. We have developed: 14:43 < bridge> > 14:43 < bridge> > An abstract CHERI protection model that introduces architectural capabilities, hardware-supported descriptions of permissions that can be used, in place of integer virtual addresses, to refer to data, code, and objects in protected ways; 14:43 < bridge> > A set of ISA extensions to 64-bit MIPS, 32-bit RISC-V, 64-bit RISC-V, and (in collaboration with Arm) 64-bit Armv8-A, showing that the model is applicable to a range of contemporary ISA designs. 14:43 < bridge> > New microarchitecture demonstrating that capabilities can be implemented efficiently in hardware, including capability compression and tagged memory to protect capabilities in memory; and 14:43 < bridge> > Formal models of these ISA extensions enabling mechanised statements and proofs of their security properties, automatic test generation, and automatic construction of executable ISA-level simulators. 14:43 < bridge> RISC-V is both 32 and 64 bits 14:43 < bridge> risc-v in itself is just the isa, it does not say anything about the implementation 14:43 < bridge> > We may end up with 128-bit pointers even without a "true" 128-bit architecture. For example, CHERI has 128-bit pointers with 64-bit address space and 64 bits of metadata to crash harder on memory bugs in C and C++: 14:43 < bridge> Why no RISC-V for 8 and 16bit? 14:43 < bridge> you are free to do whatever you want as long as you respect the isa 14:43 < bridge> @learath2 cheri has pointer metadata 14:44 < bridge> What would one even have in there? Type information? 14:44 < bridge> no idea, probably more niche and harder to fit the requirements of the isa 14:44 < bridge> do u know about pointer provenance 14:44 < bridge> https://www.ralfj.de/blog/2020/12/14/provenance.html 14:44 < bridge> Hello rust devs in here, 14:44 < bridge> 14:44 < bridge> just a quick question. I would like to move some parts of our network from golang to rustlang. Is tokio-rs still the state-of-the-art tech? 14:44 < bridge> https://www.ralfj.de/blog/2018/07/24/pointers-and-bytes.html 14:44 < bridge> No, let me see 14:45 < bridge> > This “extra information” that distinguishes different pointers to the same address is typically called provenance. This post is another attempt to convince you that provenance is “real”, by telling a cautionary tale of what can go wrong when provenance is not considered sufficiently carefully in an optimizing compiler. The post is self-contained; I am not assuming that you have read the first one. There is also a larger message here about 14:45 < bridge> > just because two pointers point to the same address, does not mean they are equal in the sense that they can be used interchangeably. 14:45 < bridge> Ahh, this thing. Yeah I know. I forgor the name 14:45 < bridge> It's the same problem `std::launder` solves 14:46 < bridge> TIL 14:46 < bridge> @ryozuki would know better but as far as I know, yes 14:46 < bridge> yes 14:46 < bridge> but beware 14:47 < bridge> with golang u can do half cooperative preemption 14:47 < bridge> or smth like that 14:47 < bridge> tokio is fully coop preemptive 14:47 < bridge> Async in rust is still a bit, rusty 🥁 14:47 < bridge> ok i used bad wording 14:47 < bridge> golang is partially preemptive 14:47 < bridge> cuz the gc 14:48 < bridge> anyway unless u need to do weird stuff its fine 14:48 < bridge> Just for our internal networking 14:48 < bridge> Huh, why is tokio different in that aspect? 14:48 < bridge> Both can only preempt at function calls, no? 14:48 < bridge> > // A goroutine can be preempted at any safe-point. Currently, there 14:48 < bridge> > 14:48 < bridge> > // are a few categories of safe-points: 14:48 < bridge> > 14:48 < bridge> > // 14:48 < bridge> > 14:48 < bridge> > // 1. A blocked safe-point occurs for the duration that a goroutine is 14:48 < bridge> > 14:48 < bridge> > // descheduled, blocked on synchronization, or in a system call. 14:49 < bridge> > 14:49 < bridge> > // 14:49 < bridge> > 14:49 < bridge> > // 2. Synchronous safe-points occur when a running goroutine checks 14:49 < bridge> > 14:49 < bridge> > // for a preemption request. 14:49 < bridge> > 14:49 < bridge> > // 14:49 < bridge> > 14:49 < bridge> > // 3. Asynchronous safe-points occur at any instruction in user code 14:49 < bridge> > 14:49 < bridge> > // where the goroutine can be safely paused and a conservative 14:49 < bridge> > 14:49 < bridge> > // stack and register scan can find stack roots. The runtime can 14:49 < bridge> > 14:49 < bridge> > // stop a goroutine at an async safe-point using a signal. 14:49 < bridge> i think tokio can only preempty on await points 14:49 < bridge> i think tokio can only preempt on await points 14:49 < bridge> You'll be fine, these are all very small niche things anyway 😄 14:49 < bridge> yeah xd 14:49 < bridge> Ah. That would make sense with how rust implements async 14:49 < bridge> Okay, thanks! 14:49 < bridge> @avolicious u will be fine, faster, more memory efficient ,and above all, lightning fast and cool 14:50 < bridge> if u need a http server 14:50 < bridge> i recommend axum 14:50 < bridge> if u need to do queries 14:50 < bridge> sqlx 14:50 < bridge> @avolicious how was your experience with go? Is it enjoyable to use? 14:50 < bridge> go is the modern java, but a bit better 14:50 < bridge> its SO simple u can learn it in 1 day 14:50 < bridge> Its still in use for gRPC stuff, but I wanted to try out rustlang with gRPC 14:50 < bridge> its the perfect language for a corporate company drone 14:51 < bridge> We use this library https://gnet.host 14:51 < bridge> I'd bet it's much better at concurrency than Java 14:52 < bridge> probs idk 14:52 < bridge> modern java is going at fast pace 14:52 < bridge> but yeah golang is known for good asynnc 14:52 < bridge> but yeah golang is known for good async 14:52 < bridge> doesnt protect u from data races tho 14:53 < bridge> I don't mean performance btw. I just meant as in usability 14:53 < bridge> ah yeah 14:53 < bridge> its built in the language after all 14:53 < bridge> goroutines 14:53 < bridge> https://github.com/hyperium/tonic 14:54 < bridge> Yeah, in our case we need some sort of gRPC proxy 😄 14:54 < bridge> Because the teeworlds server is communicating over gRPC with a local endpoint to exchange data 14:54 < bridge> ```toml 14:54 < bridge> [profile.release] 14:54 < bridge> lto = true 14:54 < bridge> codegen-units = 1 14:54 < bridge> ``` 14:54 < bridge> add this to ur cargo toml 14:54 < bridge> when making release binaries 14:54 < bridge> for max perf 14:55 < bridge> also 14:55 < bridge> ```toml 14:55 < bridge> [target.x86_64-unknown-linux-gnu] 14:55 < bridge> linker = "clang" 14:55 < bridge> rustflags = ["-Clink-arg=-fuse-ld=mold"] 14:55 < bridge> ``` 14:55 < bridge> on .cargo/config.toml if u wanna use mold 14:55 < bridge> and 14:55 < bridge> `RUSTFLAGS='-C target-cpu=native'` 14:55 < bridge> to use native march 14:57 < bridge> of its for kogh 14:57 < bridge> i thought it was for ur company 14:57 < bridge> sadge 14:57 < bridge> Yeah its for KoG 14:58 < bridge> become a rust monk at ur company 14:58 < bridge> add it 14:58 < bridge> :justatest: 14:58 < bridge> No, company builds stuff with PHP 😄 14:58 < bridge> xD 14:58 < bridge> They wont change 14:58 < bridge> But I would like to showcase them how we did gRPC on KoG 14:58 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1178696388876456046/image.png?ex=6577158e&is=6564a08e&hm=ea4b2928e3b14f0a4740cc73e6c8b362b686cef16ac9ed172bfe112cfb450a64& 14:58 < bridge>