00:31 < bridge> aoc will be interesting 00:35 < bridge> https://stability.ai/news/stable-video-diffusion-open-ai-video-model 00:35 < bridge> @jupeyy_keks gg 00:44 < bridge> <_voxeldoesart> :( 07:48 < bridge> Nice 08:43 < bridge> morning 08:44 < bridge> its finally time i become a llvm contributor 08:47 < bridge> xd you have 20k issues to choose from 08:49 < bridge> nono 08:49 < bridge> i have my own issue 08:50 < bridge> https://github.com/llvm/llvm-project/issues/73008 08:58 < bridge> Epic 08:59 < bridge> Now do the pr to immortalize yourself 09:13 < bridge> yes, i only need to do a setup that takes shitload and then code 09:27 < bridge> btw maybe interesting for you (run on intel skylake i7 6700k): 09:27 < bridge> https://pastebin.com/jpRXghp2 09:28 < bridge> one of the reasons i linked hashlink. even tho it's a linked list, it's pretty fast for my use cases 09:28 < bridge> (source: https://github.com/Jupeyy/dd-pg/blob/test5/game/shared-base/src/indexmap_tests.rs) 09:31 < bridge> kinda funny that it beat std vec in the second benchmark in iteration 09:31 < bridge> 09:31 < bridge> is ofc bit luck based, but shows that a efficiently packed linked list can still be very powerful 09:35 < bridge> @jupeyy_keks why u not using criterion and black_box for benches? 09:36 < bridge> 1. i never heard of that 09:36 < bridge> 2. i wrote these benchmarks when i started with rust. so i defs didnt heard of them back then xdd 09:36 < bridge> i tried cargo bench, but was unstable 09:36 < bridge> like anything in rust 😬 09:36 < bridge> https://doc.rust-lang.org/std/hint/fn.black_box.html 09:36 < bridge> cargo bench is not unstable?? 09:36 < bridge> then it was back then, i could not use it without nightlies 09:37 < bridge> "Note: The #[bench] attribute is currently unstable and only available on the nightly channel. " 09:37 < bridge> anyway everybody and their mother uses https://github.com/bheisler/criterion.rs 09:37 < bridge> you can also use https://github.com/bheisler/iai which i think uses callgrind under the hood 09:38 < bridge> the blackbox function is important to prevent const time optimizations on input 09:38 < bridge> ah ur looking at the attribute 09:39 < bridge> next time i'll look into it xd 09:39 < bridge> but it doesn't look trivial to use these libs to me 09:39 < bridge> i prefer to have some fancy ass macro that hides anything from me 09:41 < bridge> ? 09:41 < bridge> use criterion 09:41 < bridge> it is like a cargo bench impl 09:41 < bridge> when u setup criterion u just use cargo bench 09:41 < bridge> ok 09:42 < bridge> neat thing is it can make html reports with graphs 09:42 < bridge> im seting up ccache before i venture on llvm dev 09:42 < bridge> yeah, for a more important benchmark i can look into it xdd 09:43 < bridge> how does it feel as cpp dev? 09:43 < bridge> wdym 09:43 < bridge> isnt llvm in cpp? 09:43 < bridge> ye 10:40 < bridge> is there a command for a server to make it load maps in which you always must join a team? 10:44 < bridge> Smth Like teamsize exist 10:46 < bridge> i was looking for sv_team command but it gets overwritten by map commands or smth :thonk: 10:53 < bridge> it doesnt seem to work even if a map has no sv_team coomand set :p 10:55 < ChillerDragon> reset.cfg is your friend for sv_team 10:55 < ChillerDragon> https://github.com/ddnet/ddnet/blob/8b0fccc6a3992084bda4bbfb8b957ca3b92f871d/src/engine/shared/config_variables.h#L356 10:56 < ChillerDragon> @milkeeycat this and inside the map are the places where sv_team can be set. Setting it in the regular autoexec.cfg gets troled away 11:04 < bridge> thanks :D 11:42 < bridge> i got wow working on linux 11:42 < bridge> with lutris 11:42 < bridge> linux is amazing 11:56 < bridge> Wow 12:03 < ChillerDragon> wow 12:34 < bridge> wow 12:47 < bridge> ```cpp 12:47 < bridge> LLVMModuleRef mlirTranslateModuleToLLVMIR(MlirOperation module, LLVMContextRef context) { 12:47 < bridge> Operation *moduleOp = unwrap(module); 12:47 < bridge> 12:47 < bridge> llvm::LLVMContext *ctx = reinterpret_cast(context); 12:47 < bridge> 12:47 < bridge> std::unique_ptr llvmModule = mlir::translateModuleToLLVMIR(moduleOp, *ctx); 12:47 < bridge> 12:47 < bridge> LLVMModuleRef moduleRef = reinterpret_cast(const_cast(llvmModule.release())); 12:47 < bridge> 12:47 < bridge> return moduleRef; 12:47 < bridge> } 12:47 < bridge> ``` 12:47 < bridge> guys 12:47 < bridge> do u see ub here 12:47 < bridge> xd 12:47 < bridge> a c api for the c++ api 12:51 < bridge> i tested it and it works 12:51 < bridge> kek 13:37 < bridge> epic 13:37 < bridge> i even added a test 13:39 < bridge> Wat u hacking ryo 13:43 < bridge> llvm 13:43 < bridge> Contribute to the project? 13:47 < bridge> He's basically the maintainer at this point 13:53 < bridge> ye 14:12 < ChillerDragon> wow 14:35 < bridge> https://github.com/llvm/llvm-project/pull/73117 14:36 < bridge> :justatest: 14:45 < bridge> Wow 14:45 < bridge> wew 14:45 < bridge> @heinrich5991 btw did u know 14:45 < bridge> llvm clang format options 14:45 < bridge> they rename ur variables to the correct format 14:45 < bridge> xd 14:46 < bridge> no, did not 14:46 < bridge> ❯ cat .clang-format 14:46 < bridge> BasedOnStyle: LLVM 14:46 < bridge> lmao 14:46 < bridge> what's "access"? how can it be slower than "iterate"? 14:46 < bridge> ``` 14:46 < bridge> ❯ cat .clang-tidy 14:46 < bridge> Checks: '-*,clang-diagnostic-*,llvm-*,misc-*,-misc-const-correctness,-misc-unused-parameters,-misc-non-private-member-variables-in-classes,-misc-no-recursion,-misc-use-anonymous-namespace,readability-identifier-naming' 14:46 < bridge> CheckOptions: 14:46 < bridge> - key: readability-identifier-naming.ClassCase 14:46 < bridge> value: CamelCase 14:46 < bridge> - key: readability-identifier-naming.EnumCase 14:46 < bridge> value: CamelCase 14:46 < bridge> - key: readability-identifier-naming.FunctionCase 14:47 < bridge> value: camelBack 14:47 < bridge> # Exclude from scanning as this is an exported symbol used for fuzzing 14:47 < bridge> # throughout the code base. 14:47 < bridge> - key: readability-identifier-naming.FunctionIgnoredRegexp 14:47 < bridge> value: "LLVMFuzzerTestOneInput" 14:47 < bridge> - key: readability-identifier-naming.MemberCase 14:47 < bridge> value: CamelCase 14:47 < bridge> - key: readability-identifier-naming.ParameterCase 14:47 < bridge> value: CamelCase 14:47 < bridge> - key: readability-identifier-naming.UnionCase 14:47 < bridge> value: CamelCase 14:47 < bridge> - key: readability-identifier-naming.VariableCase 14:47 < bridge> value: CamelCase 14:47 < bridge> - key: readability-identifier-naming.IgnoreMainLikeFunctions 14:47 < bridge> value: 1 14:47 < bridge> - key: readability-redundant-member-init.IgnoreBaseInCopyConstructors 14:47 < bridge> value: 1 14:47 < bridge> - key: modernize-use-default-member-init.UseAssignment 14:47 < bridge> value: 1 14:47 < bridge> ``` 14:47 < bridge> with this 14:47 < bridge> i think 14:47 < bridge> uhhh my game crashed when i reloaded my server while i was still uploading a map, should i put the crash dump 14:47 < bridge> Access By value 14:47 < bridge> For vector it's find 14:47 < bridge> That's why it's slow 14:47 < bridge> @jupeyy_keks btw this uses the default hasher for hashmap 14:47 < bridge> there are faster hash functions 14:47 < bridge> wtf?? 14:48 < bridge> Ok but I doubt it will beat indexmap 14:48 < bridge> https://github.com/tkaitchuck/ahash 14:48 < bridge> https://gitlab.redox-os.org/redox-os/seahash 14:49 < bridge> what does access mean for the different containers? probably not access by value, but by key somehow, for hashmaps? 14:49 < bridge> uh oh my server is crashing me whenever i try to join it 14:51 < bridge> Yes. But since i also need delete it needs to be able to have hashmap lookup perf anyway. Don't over interpret the vec results. It's mostly in there to check iteration speed 14:51 < bridge> It's original designed to compare index map to a hash queue 14:52 < bridge> Ofc u can use a vector and never ever delete any element to have index performance 14:52 < bridge> But then u have the limitations of ddnet code 14:53 < bridge> there are also hashmaps that can preserve insertion order 14:53 < bridge> Indexmap 14:53 < bridge> But not deleting order 14:53 < bridge> At least not fast 14:54 < bridge> yes, also deletion order 14:54 < bridge> I think python has one 14:54 < bridge> I think you do it with a linked hash map, basically 14:54 < bridge> Yes that's what I'm using here too 14:54 < bridge> That's the whole point xd 14:55 < bridge> Linked lists are used 14:55 < bridge> but apparently thel inked hashmap you use cannot maintain insertion order on deletion 14:55 < bridge> I just wanted to say that this is possible 14:55 < bridge> It can 14:56 < bridge> Only indexmap can not 14:56 < bridge> I use hashlink 14:56 < bridge> Indexmap basically used a vec internally 14:56 < bridge> So it should normally outperform a internal linked list 14:56 < bridge> In iteration 14:57 < bridge> what do u use this map for 14:57 < bridge> Lot of stuff 14:58 < bridge> E.g. skins container 14:58 < bridge> Which makes it easy to remove unused skins 14:58 < bridge> A.k.a unhot unloading 14:58 < bridge> And hot loading if used again 15:01 < bridge> https://github.com/Jupeyy/dd-pg/blob/test5/game/client-containers/src/container.rs#L98-L110 15:01 < bridge> 15:01 < bridge> i'm basically moving used stuff to the end, which is _almost_ free with a linked list 15:01 < bridge> then i know that all stuff in the front _might_ be old 15:02 < bridge> it's also nice for network queues 15:02 < bridge> 15:02 < bridge> and entity containers 15:04 < bridge> did u know the maintainer of the crate is/was the lead dev of starbound 15:05 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1176886068855967854/image.png?ex=65707f90&is=655e0a90&hm=2828311a53ddf13a9d491a5aec736975fec7810573a090a4d7f564a437428a91& 15:05 < bridge> ChillerDragon 15:05 < bridge> what are u up to 15:05 < bridge> little troll xd 15:05 < bridge> https://store.steampowered.com/app/211820/Starbound/ 15:05 < bridge> this game xdd 15:06 < bridge> ok enough talk, cya 15:06 < bridge> 15:06 < bridge> chiller stop troll on important projects lmao 15:06 < bridge> nice 15:06 < bridge> she even had a rustcon talk xDD 15:07 < bridge> when c++20 on ddnet 15:08 < bridge> chillerdragon: what did u even try to write that it was auto detected by some bot 15:08 < bridge> i wanna know too 15:08 < bridge> in 30 years 15:09 < bridge> im amazed how well lutris works 15:09 < bridge> and well wine proton 15:09 < bridge> ill try starcraft 2 15:09 < bridge> did you maybe get an email about it? 15:10 < bridge> lemme check 15:10 < bridge> yes 15:10 < bridge> he said this 15:10 < bridge> ``` 15:10 < bridge> wow ryOwO is llvm dev 15:10 < bridge> ``` 15:10 < bridge> no wonder xd 15:10 < bridge> typical chiller ngl 15:11 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1176887690784931890/image.png?ex=65708112&is=655e0c12&hm=fbc26fc2ab43d845886aa9b4898f5910c15fff3e096279aa6ba42e799e7a1e88& 15:11 < bridge> 113mb/s rn 15:11 < bridge> linux rocks 15:12 < bridge> i need to buy a new 2tb ssd 15:12 < bridge> xd 15:12 < bridge> 15:12 < bridge> i won lottery => linux rocks 15:12 < bridge> i kinda wanna change the m2 ssd which i think is 500gb 15:12 < bridge> but i would need to move my OS 15:12 < bridge> and im too lazy 15:12 < bridge> maybe it would work if i copy it with dd 15:12 < bridge> and then change fstab? 15:13 < bridge> the uuid 15:13 < bridge> https://www.amazon.es/Crucial-CT2000P5PSSD8-S%C3%B3lido-Interno-6600MB/dp/B098WKQRDL/ 15:13 < bridge> its amazing 15:13 < bridge> 100€ 15:13 < bridge> pic4 15:13 < bridge> so cheap rn 15:13 < bridge> you can clone disks/partitions with gparted 15:14 < bridge> also iirc UUID gets also copied 15:15 < bridge> rly? 15:15 < bridge> i thought uuid was made up from some hardware info 15:15 < bridge> xd 15:15 < bridge> ill buy that m2 15:16 < bridge> gparted also has an option to regenerate UUID for a filesystem/partition 15:18 < bridge> So if you can have both SSDs connected to your PC at the same time you can just copy+paste partitions with gparted 15:19 < bridge> you can resize them also etc 15:20 < bridge> @ryozuki how the hell do you live with 500GB xdd 15:20 < bridge> 15:20 < bridge> compiling 10 rust projects = full 15:20 < bridge> i have a normal ssd 15:20 < bridge> with 2tb 15:20 < bridge> but shared with winshit 15:20 < bridge> ok 😄 15:20 < bridge> urgh 15:20 < bridge> damn 15:20 < bridge> amazon is smth else 15:20 < bridge> i will get the ssd today 15:20 < bridge> XD 15:21 < bridge> big city 15:21 < bridge> ye 15:22 < bridge> oh i see my setup now 15:22 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1176890554836058213/image.png?ex=657083bd&is=655e0ebd&hm=ffa9fb27ffc25a08232a6424e890c9c1cd5ae6e59b2d4b3a9526b61133811cab& 15:22 < bridge> this is how i have it 15:23 < bridge> 2 m2s 15:23 < bridge> windows stupidly takes lot of space 15:24 < bridge> https://askubuntu.com/questions/57908/how-can-i-quickly-copy-a-gpt-partition-scheme-from-one-hard-drive-to-another 15:27 < bridge> https://wiki.archlinux.org/title/Rsync#Full_system_backup 15:27 < bridge> hmm 15:28 < bridge> https://man.archlinux.org/man/xfs_copy.8 15:28 < bridge> looks pog 15:46 < bridge> damn i already got a review 15:54 < bridge> Rip 15:54 < bridge> More review than code xd 15:54 < bridge> Good that u paid for that 15:54 < bridge> nah its good 15:55 < bridge> its mostly style and doc conventions 17:21 < ChillerDragon> https://zillyhuhn.com/cs/.1700670082.png 17:21 < ChillerDragon> watfak 17:21 < bridge> don't spam people's issue trackers 17:22 < ChillerDragon> ofc banhammer5991 agrees with censorship axaxax 17:22 < ChillerDragon> i was cheering for my frint 17:22 < bridge> https://lucisqr.substack.com/p/why-i-think-c-is-still-a-very-attractive 17:41 < bridge> chillerdragon: ~600 ppl are watching that repo 17:41 < bridge> nice mass ping 17:41 < bridge> 😬 17:48 < ChillerDragon> oh no shiet 17:48 < ChillerDragon> but wait they dont watch pr comments do they? 17:48 < bridge> https://github.com/llvm/llvm-project/pull/73117#pullrequestreview-1744884956 17:48 < bridge> ez 17:48 < bridge> my pr got approved 17:48 < ChillerDragon> epic 17:48 < bridge> if you watch a repo, you get all comments 17:49 < ChillerDragon> okay i fucked up then 17:49 < ChillerDragon> wopsi daisy 17:49 < ChillerDragon> Im not used to github comments being red by more than 2 people 17:49 < ChillerDragon> @ryozuki thats big flex bro congrats 17:50 < bridge> chiller 17:50 < bridge> they watch prs 17:50 < bridge> and they have subscribers 17:50 < bridge> on another platform 17:50 < bridge> @llvm/pr-subscribers-mlir-llvm 17:50 < bridge> 17:50 < bridge> @llvm/pr-subscribers-mlir 17:50 < bridge> also on phabricator 17:50 < bridge> it's ok chiller, u'll be fame now 17:52 < bridge> <_voxeldoesart> wait, github can delete comments? LOL 17:52 < bridge> <_voxeldoesart> i thought everything there was just kinda permanent 17:52 < ChillerDragon> nono github is like a regular forum 17:52 < ChillerDragon> repo owners can delete and edit comments 17:52 < bridge> u can generally remove sensible data 17:52 < bridge> edits show their diff tho* 17:53 < ChillerDragon> ddnet also uses it to censor bot clients etc 17:53 < ChillerDragon> bot ads 17:53 < bridge> you mean yours? 17:53 < bridge> bcs u always cross link to your repo 17:53 < ChillerDragon> oh that too axaxax 17:53 < bridge> https://www.youtube.com/watch?v=0PK9RzSl9O0 17:53 < bridge> @jupeyy_keks 17:54 < ChillerDragon> i saw latlely something on ddnet github where it got edited saying "dont mention bot clients" or sumsin 17:54 < bridge> the average apple user takes that as granted 17:55 < bridge> you can also delete edit revisions 17:55 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1176928891521208422/image.png?ex=6570a772&is=655e3272&hm=a9f2edc112cfa32d93b84e3a1d4bd8de67fc922c4a5873f99348b7e5b086b152& 17:55 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1176928968478310420/image.png?ex=6570a784&is=655e3284&hm=014f16a000250622a8046fb5e2d2f7343f8664621be26ea608b126b6dd3fa0bc& 17:55 < bridge> XD 17:55 < bridge> <_voxeldoesart> lol 17:55 < bridge> apple in a nutshell 17:56 < bridge> <_voxeldoesart> when open source hardware stuff 17:56 < bridge> open hardware? 17:56 < bridge> why u carte 17:56 < bridge> care 17:56 < bridge> u use windows 17:56 < bridge> :justatest: 17:56 < bridge> pssst 17:56 < bridge> dont demotivate him 17:57 < bridge> i think rpi is open source 17:57 < bridge> @ryozuki if there would be open hardware pcs (that are useful) 17:57 < bridge> would u also built it your own, just like you build gentoo your own? 17:57 < bridge> else u not hipster enough 17:57 < bridge> idk 17:57 < bridge> sounds more complicated 17:57 < bridge> but ifg i had spare money i would try 17:58 < bridge> <_voxeldoesart> damn so im suddenly not allowed to mention j 17:58 < bridge> <_voxeldoesart> hardware now?? LOL 17:58 < bridge> <_voxeldoesart> classic linux user 17:58 < bridge> i always wonder how high voxels pulse is in this chat 17:58 < bridge> either he player, or he raging af 17:59 < bridge> <_voxeldoesart> im literally laying in bed rn 17:59 < bridge> bom bom 17:59 < bridge> bom bom 17:59 < bridge> im joking xd 17:59 < bridge> voxel u need to come to windows 17:59 < bridge> ur the last one left 17:59 < bridge> you can still use windows and care about open hardware 17:59 < bridge> well and robyt 17:59 < bridge> just like you can use closed hardware and still use linux 17:59 < bridge> yeah 18:00 < bridge> please don't do this shitting on windows users… 18:00 < bridge> please continue, my popcorn is not empty yet 18:01 < bridge> <_voxeldoesart> he said he was joking, plus this is ryozuki so, ive gotten used to it 18:01 < bridge> idc if people are "joking" 18:01 < bridge> in real life you have to make really sure that the recipient gets that, in order to still be friendly 18:01 < bridge> over the internet, it just doesn't work 18:02 < bridge> feels low quality and typical anti rust 18:02 < bridge> > Conclusion 18:02 < bridge> > Given that the performance benefits are either inconclusive, non-existent or more likely negative, and the safety benefits are not really that pressing for most applications, is that really worth moving to a completely new language, with its infamous long and hard learning curve as Rust? 18:02 < bridge> xd 18:02 < bridge> What did chiller write in the comment? 18:03 < bridge> ``` 18:03 < bridge> wow ryOwO is llvm dev 18:03 < bridge> ``` 18:03 < bridge> he wrote this 18:03 < bridge> ^ 18:03 < bridge> that indeed sounds bullshit 18:03 < bridge> xDD 18:03 < bridge> on my pr on llvm 18:03 < bridge> @learath2 im a llvm contributor now 18:03 < bridge> :EZ: 18:03 < bridge> :green_tick: rust 18:03 < bridge> :green_tick: llvm 18:03 < bridge> ??? 18:03 < bridge> <_voxeldoesart> LOL 18:03 < bridge> <_voxeldoesart> trollerdragon 18:03 < bridge> khronos 18:03 < bridge> @learath2 im an llvm contributor now 18:03 < bridge> what do u contribute to khronos? 18:03 < bridge> a spec? xD 18:04 < bridge> i contributed a bug report for the vulkan loader 18:04 < bridge> 😏 18:04 < bridge> I still think X is as good as Y. that mostly tells me that Y is generally accepted to be good 🙂 18:04 < bridge> "I still think X is as good as Y". that mostly tells me that Y is generally accepted to be good 🙂 18:04 < bridge> what is x and y here? 18:05 < bridge> im not getting what u mean exactly 18:05 < bridge> I'm quoting the title of the blog post I sent 18:05 < bridge> ah 18:05 < bridge> I have a feeling anything anti rust would feel low quality to you immediately anyway 18:06 < bridge> not rly 18:06 < bridge> i always said 18:06 < bridge> if a new lang better than rust pops up 18:06 < bridge> ill jump ship 18:06 < bridge> and i hope one appears 18:06 < bridge> rust nightly > rust 18:06 < bridge> ryo can accept that 18:06 < bridge> true xd 18:06 < bridge> unfortunately, rust has a lot of movement behind it at this point 18:06 < bridge> so the new language is going to face an uphill battle, even if it is better 18:07 < bridge> true 18:07 < bridge> maybe rust has also easier time adopting new features from new languages 18:07 < bridge> @jupeyy_keks polonius rust > rust 18:07 < bridge> wat 18:07 < bridge> new borrow checker, I think? 18:07 < bridge> polonius is the next gen borrow checker 18:08 < bridge> is it faster? 18:08 < bridge> -Zpolonius 18:08 < bridge> is it less sensitive? xd 18:08 < bridge> less sensitive 18:08 < bridge> it allows more 18:08 < bridge> interesting 18:08 < bridge> i think its not faster or slower 18:08 < bridge> https://github.com/rust-lang/polonius 18:10 < bridge> there is also plans for a better type solver 18:10 < bridge> https://github.com/rust-lang/chalk 18:11 < bridge> well im not sure if its being used already 18:11 < bridge> i read rust-analyzer uses it 18:12 < bridge> It's not exactly less sensitive but it's smarter and can prove more, no? 18:12 < bridge> yeah, i guess by less sensitive we mean it is less restrictive in what u can do 18:12 < bridge> since it can prove more yes 18:14 < bridge> I think I'll just merge #7171 and we can fix the ubsan warnings some other way 18:14 < bridge> https://github.com/ddnet/ddnet/pull/7171 18:15 < bridge> @ryozuki have u already tried NVK? 18:15 < bridge> it's now vk 1.0 conform 18:15 < bridge> whats was nvk xd 18:15 < bridge> nvidia vk driver 18:15 < bridge> oss ofc 18:16 < bridge> not from nvidia 18:16 < bridge> ah idk 18:16 < bridge> enjoy your 3 fps 18:16 < bridge> 😬 18:28 < bridge> Is anyone even making anything using the rust integration we have in the code? 18:29 < bridge> Every compilation is paying the rust tax and we still don't have anything in :/ 18:30 < bridge> yea, maybe we should rip it out again 18:30 < bridge>