00:38 < bridge> <vahemaaa> i think its time for uniting cyberpunk and ddnet 😴
00:51 < bridge> <cyberfrighter> no
02:58 < bridge> <chillerdragon> Missed opportunity for \#10000 the problem of death
03:03 < bridge> <cyberfrighter> thx
08:24 < bridge> <gratzl> https://docs.fabric.io/android/crashlytics/custom-web-hooks.html?web%20hooks#custom-web-hooks
09:57 < bridge> <ryozuki> :owo:
10:24 < bridge> <ryozuki> building llvm 20.1.2
10:25 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1356907286848737423/image.png?ex=67ee45e3&is=67ecf463&hm=baffe359984e68821f2bd0147760e474ac3134ffbb03d780973d996b90964fb7&
10:41 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1356911337787228221/vAuOow0.png?ex=67ee49a9&is=67ecf829&hm=dd2e3845d8d219464337117f20cbe1928d46efa0d938d7ebc9472e8de96e5562&
10:41 < bridge> <ryozuki> firefox
10:54 < bridge> <learath2> firefok
11:28 < bridge> <ryozuki> class CGameContext
11:28 < bridge> <ryozuki> Size: 361480 bytes, alignment 8 bytes
11:28 < bridge> <ryozuki> ok
11:28 < bridge> <totar> many cases
11:28 < bridge> <ryozuki> we have a class that is 361kb in size
11:29 < bridge> <ryozuki> im trying to reorder some fields
11:29 < bridge> <ryozuki> to reduce size
11:29 < bridge> <ryozuki> we should adopt a style of not mixing fields and function declarations
11:29 < bridge> <ryozuki> first fields at top
11:29 < bridge> <ryozuki> then functions
11:30 < bridge> <robyt3> That seems less readable. I always group variables and functions that belong together. Also, I think private variables/functions should be at the bottom.
11:30 < bridge> <totar> did you check CGameClient
11:31 < bridge> <ryozuki> how so?
11:31 < bridge> <robyt3> How do you know what a bunch of variables and functions belong together if you sort them?
11:31 < bridge> <totar> https://cdn.discordapp.com/attachments/293493549758939136/1356923998084268032/image.png?ex=67ee5573&is=67ed03f3&hm=5039498d2bbaeb09767da3d52949f252a7dac995a291d74dcdcca1720934100e&
11:31 < bridge> <ryozuki> but since c++ doesnt reorder fields like rust we need this
11:31 < bridge> <ryozuki> to reduce sizes
11:31 < bridge> <ryozuki> comments
11:31 < bridge> <chairn> is there a compilation flag or attribute we could use to allow the compiler to reorder the fields ?
11:32 < bridge> <chairn> because i prefer readability over some bytes
11:32 < bridge> <ryozuki> not that i know
11:32 < bridge> <Jupstar ✪> Yeah why does the size matter xd
11:32 < bridge> <ryozuki> its the pain of using c++
11:32 < bridge> <ryozuki> less size more cache friendly? xd
11:32 < bridge> <ryozuki> arent u the max fps guy
11:33 < bridge> <Jupstar ✪> I doubt we have a bottleneck here
11:33 < bridge> <ryozuki> i guess i wont do this then
11:33 < bridge> <Jupstar ✪> There are other parts we can optimize first
11:33 < bridge> <totar> I don't think the ordering is the main thing causing the size
11:33 < bridge> <totar> It's just that big
11:33 < bridge> <ryozuki> well ordering matters
11:33 < bridge> <ryozuki> cproj i reduced from 136 to 128
11:33 < bridge> <ryozuki> just changing order
11:33 < bridge> <ryozuki> but for the big ones yeah
11:33 < bridge> <ryozuki> they have big arrays in them
11:33 < bridge> <ryozuki> its normal
11:34 < bridge> <Jupstar ✪> We use lots of static sized arrays
11:34 < bridge> <robyt3> I don't want to comment 20 variables and functions with the same comment though, I'd rather put one comment on top. Or group those vars and functions in a separate class.
11:34 < bridge> <ryozuki> everyday i hate more c++ and its limitations
11:34 < bridge> <ryozuki> u have to choose readbility or perf
11:35 < bridge> <Jupstar ✪> xD
11:35 < bridge> <ryozuki> :nouis:
11:35 < bridge> <Jupstar ✪> That rust can reorder is defs cool
11:35 < bridge> <Jupstar ✪> 10/10 give you that
11:35 < bridge> <chairn> can you try with `__attribute__((packed))` to see if there's any noticeable difference ?
11:35 < bridge> <ryozuki> and since the visibility qualifier is "public:" and nota prefix like in rust
11:35 < bridge> <ryozuki> if u want to order fields with different visibility u need to put public protected again in many places
11:35 < bridge> <ryozuki> it sucks hardcore
11:35 < bridge> <chairn> but compatibility when you dump struct/class to file is lost
11:36 < bridge> <Jupstar ✪> You use serde for that anyway
11:36 < bridge> <chairn> google also threw me that https://manpages.ubuntu.com/manpages/bionic/man1/clang-reorder-fields-10.1.html
11:36 < bridge> <ryozuki> for cproj it also reduces it to 128 but packed is less perf
11:37 < bridge> <ryozuki> Size: 361480 bytes, alignment 8 bytes for cgamecontext
11:37 < bridge> <ryozuki> no change i guess
11:38 < bridge> <ryozuki> u shouldnt dump to a file as is anyway
11:38 < bridge> <totar> I thought if the alignment is 8 bytes then it won't waste more than 8 bytes between fields? so that's basically nothing?
11:38 < bridge> <ryozuki> isnt a cache line 64 bytes in most cpus
11:38 < bridge> <ryozuki> if so in that case it matters a lot
11:39 < bridge> <totar> what?
11:39 < bridge> <ryozuki> if a struct goes from 72 to 64
11:39 < bridge> <totar> but that's not what determines the struct packing?
11:39 < bridge> <ryozuki> by ordering fields
11:39 < bridge> <ryozuki> ?
11:39 < bridge> <ryozuki> if u have a int and a long then u have 4 bytes of padding
11:39 < bridge> <ryozuki> after the int
11:40 < bridge> <ryozuki> if u have int long int
11:40 < bridge> <ryozuki> u waste 8 bytes
11:40 < bridge> <ryozuki> long int int
11:40 < bridge> <ryozuki> nothing wasted
11:40 < bridge> <totar> well yes but that's not the same issue as CGameConext and other classes being hundreds of kilobytes
11:40 < bridge> <totar> it's a micro optimization then
11:40 < bridge> <ryozuki> yeah im not talking about the big ones
11:40 < bridge> <ryozuki> its ok if u dont care i wont do it
11:40 < bridge> <ryozuki> xd
11:40 < bridge> <Jupstar ✪> You can defs loose between a u32 to u64 member
11:40 < bridge> <Jupstar ✪> 
11:40 < bridge> <Jupstar ✪> but u32 and u32 should still pack fine
11:40 < bridge> <ryozuki> anyway readbility matters more and perf probs not a gain
11:41 < bridge> <totar> that's fine I just wanted to make sure you weren't trying to reduce the size meaningfully with struct reordering, because that's not why they're big
11:41 < bridge> <totar> if you can do it for free then that's good
11:41 < bridge> <ryozuki> i know cgame is big cuz static arrays which we abuse a lot
11:41 < bridge> <ryozuki> but well since there is just 1 context and we not allocating deallocating it a lot should be fine tm i guess
11:41 < bridge> <ryozuki> but for entities i thought it might matter more
11:42 < bridge> <ryozuki> which is why i looked into cproj first
11:42 < bridge> <Jupstar ✪> https://cdn.discordapp.com/attachments/293493549758939136/1356926722104230050/image.png?ex=67ee57fd&is=67ed067d&hm=ba87fbd4c6352d9c6a4f7da13f7cd217669a89d7d2a2b988565eafe6a37fa657&
11:42 < bridge> <Jupstar ✪> You happy now @ryozuki ?
11:42 < bridge> <ryozuki> xd
11:42 < bridge> <ryozuki> packed is less perfomant
11:42 < bridge> <Jupstar ✪> I just slammed a `#pragma pack(1)` in system.h
11:43 < bridge> <hectavoxel> hi cbat
11:43 < bridge> <totar> it does, but the entities are stored in a linked list and alloced individually so you're fighting the wrong battle
11:43 < bridge> <ryozuki> ah i see
11:43 < bridge> <Assa> I have to intervene, there is more to it, this pragma doesn't work on all OS or does it?
11:43 < bridge> <ryozuki> linked lists my beloved
11:43 < bridge> <ryozuki> can we use a generational arena?
11:43 < bridge> <ryozuki> xd
11:43 < bridge> <Jupstar ✪> ^ but look how few bytes that make
11:43 < bridge> <ryozuki> yeah cuz arrays
11:44 < bridge> <Jupstar ✪> We won't use it, don't worry
11:44 < bridge> <ryozuki> misaligned memory accesses are slower
11:44 < bridge> <ryozuki> and in rust they are ub
11:44 < bridge> <ryozuki> i guess in C maybe too idk
11:44 < bridge> <totar> I discussed with teero, in my opinion the best option in while also maintaining physics update order is a dynamic array swapback cursor list
11:44 < bridge> <ryozuki> ok lot of fancy words
11:44 < bridge> <ryozuki> let me search
11:45 < bridge> <ryozuki> but a generational arena is a dynamic array too
11:45 < bridge> <ryozuki> u just can remove elements
11:45 < bridge> <ryozuki> by keeping a generation on each index
11:45 < bridge> <Assa> I had to use `#pragma pack(push, 1)` and then `#pragma pack(pop)` explicitly, just FYI
11:45 < bridge> <totar> swapback array is another method
11:45 < bridge> <totar> it's easier, you can just use std::vector without any arenas
11:45 < bridge> <ryozuki> a arena is a fancy vector
11:45 < bridge> <ryozuki> thats what im saying
11:46 < bridge> <totar> https://miro.medium.com/v2/resize:fit:720/format:webp/1*RrVBXPmEwm6eJGCFMSBLlw.gif
11:46 < bridge> <Jupstar ✪> That is for temporary alignment changes.
11:46 < bridge> <Jupstar ✪> 
11:46 < bridge> <Jupstar ✪> But anyway, as said it's just to show how much bytes we waste with alignment
11:46 < bridge> <Jupstar ✪> We won't use any pragma like that
11:47 < bridge> <ryozuki> oh shit
11:47 < bridge> <ryozuki> i installed llvm 20 to my llvm 19 folder
11:47 < bridge> <ryozuki> i hate my life
11:47 < bridge> <Assa> I needed this in a project for binary file alignment (over multiple OS) and further more in networking to minimize send bytes
11:47 < bridge> <Assa> also other compilers might f*ck your alignment
11:48 < bridge> <Assa> also other compilers might f*ck your alignment (the widespread ones should work tho)
11:57 < bridge> <Jupstar ✪> What you not on llvm-21 yet? 😬
11:58 < bridge> <ryozuki> xd
12:03 < bridge> <learath2> It do be UB
12:04 < bridge> <learath2> That is rather large, what are we even storing in there?
12:05 < bridge> <ryozuki> shitloads of arrays
12:05 < bridge> <ryozuki> everything
12:07 < bridge> <ryozuki> ```
12:07 < bridge> <ryozuki> ield m_aaZoneEnterMsg
12:07 < bridge> <ryozuki> Type: char[256][256]
12:07 < bridge> <ryozuki> Offset: 82736 bytes
12:07 < bridge> <ryozuki> Size: 65536 bytes, alignment 1 byte
12:07 < bridge> <ryozuki> 0 is used for switching from or to area without tunings
12:07 < bridge> <ryozuki> 
12:07 < bridge> <ryozuki> // In CGameContext
12:07 < bridge> <ryozuki> public: char m_aaZoneEnterMsg[256][256]
12:07 < bridge> <ryozuki> ```
12:07 < bridge> <ryozuki> ```
12:07 < bridge> <ryozuki> field m_aaZoneEnterMsg
12:07 < bridge> <ryozuki> Type: char[256][256]
12:07 < bridge> <ryozuki> Offset: 82736 bytes
12:07 < bridge> <ryozuki> Size: 65536 bytes, alignment 1 byte
12:07 < bridge> <ryozuki> 0 is used for switching from or to area without tunings
12:07 < bridge> <ryozuki> 
12:07 < bridge> <ryozuki> // In CGameContext
12:07 < bridge> <ryozuki> public: char m_aaZoneEnterMsg[256][256]
12:07 < bridge> <ryozuki> ```
12:07 < bridge> <ryozuki> what is this
12:07 < bridge> <robyt3> Is padding guaranteed to be zeroed (in the map format)? CSpeedupTile and CDoorTile have 1 and 2 unused padding bytes respectively
12:07 < bridge> <ryozuki> why its this way xd
12:07 < bridge> <ryozuki> is u use memzero yes otherwise its undefined
12:07 < bridge> <ryozuki> if u*
12:07 < bridge> <Assa> for every tune zone you can set an enter and a leave message
12:07 < bridge> <Assa> why is this an array?!
12:08 < bridge> <Assa> like who does 256 tune zone messages, I did like 20 in this one map
12:08 < bridge> <robyt3> I assume we do a memzero when allocating tiles, so technically we could reclaim those padding bytes and use them for something else
12:09 < bridge> <ryozuki> ```
12:09 < bridge> <ryozuki>     char m_aaZoneEnterMsg[NUM_TUNEZONES][256]; // 0 is used for switching from or to area without tunings
12:09 < bridge> <ryozuki>     char m_aaZoneLeaveMsg[NUM_TUNEZONES][256];
12:09 < bridge> <ryozuki> ```
12:09 < bridge> <learath2> I mean if ever there was a good reason to heap allocate strings there it is. We rarely ever use zone enter messages, for like 99.99% of maps that struct is completely empty and never accessed 😄
12:10 < bridge> <ryozuki> its not just 12
12:10 < bridge> <ryozuki> 1
12:10 < bridge> <ryozuki> its 2
12:10 < bridge> <ryozuki> ok guys dont do this, ill change it
12:10 < bridge> <ryozuki> so i can have a pr
12:11 < bridge> <learath2> I mean if ever there was a good reason to heap allocate strings there it is. We rarely ever use zone enter messages, for like 99.99% of maps that array is completely empty and never accessed 😄
12:11 < bridge> <learath2> Hm, guaranteed idk, we can however go through all maps and check.
12:13 < bridge> <Assa> If you need a testmap with a lot of enter and leave messages, take a look at `KingsLeap` in map testing
12:13 < bridge> <learath2> I guess an array of `std::string`s would be fine, they'd even have SSO
13:06 < bridge> <astramast> Hi
13:06 < bridge> <astramast> hi
13:07 < bridge> <astramast> @astramast:
13:07 < bridge> <astramast> @astramast
13:13 < bridge> <Assa> mast means "pole" in german, so an astra mast is a pole to the starts, so a space evelator, nice @astramast
13:26 < bridge> <inv41idu53rn4m3> A mast is also a pole in English
13:26 < bridge> <inv41idu53rn4m3> Or at least that's like the main part of a mast
13:27 < bridge> <aegisub> something something contour integral around western europe
13:29 < bridge> <Assa> @soulyvevo can you create the snowflakes sprites for your proposed change to freeze lasers? Also I need a 4th sprite for the background. That background should match all snowflakes, we already have a snowflake in `extras.png` Lastly snowflakes are always hexagons and never octagons 🙈
13:30 < bridge> <Assa> maybe I just need the background 🤔
13:30 < bridge> <astramast> It's because i can bring you to heaven 😏
13:31 < bridge> <Assa> Then you need to die for my sins first
13:31 < bridge> <astramast> https://tenor.com/view/disintegrate-gif-22469816
13:44 < bridge> <ryozuki> @learath2 can integers from GetInteger be negative?
13:44 < bridge> <ryozuki> ah nvm there is a check
13:54 < bridge> <ryozuki> ty works, reminds me of the game about jumping
13:56 < bridge> <ryozuki> @jupeyy_keks class CGameContext
13:56 < bridge> <ryozuki> Size: 230456 bytes, alignment 8 bytes
13:56 < bridge> <ryozuki> with my pr
13:56 < bridge> <Assa> you mean the game this map is inspired by about jumping called JumpKing? never heard of it /s
13:56 < bridge> <ryozuki> xd
13:58 < bridge> <Jupstar ✪> neat, is cgamecontext not on the heap?
13:58 < bridge> <ryozuki> uh idk
13:58 < bridge> <Jupstar ✪> the created object of it ofc
13:59 < bridge> <Assa> you can static cast any object into size_t?
13:59 < bridge> <ryozuki> its an int
14:00 < bridge> <Assa> oh, let me keep thinking
14:00 < bridge> <Assa> did you check skipping a tune zone?
14:01 < bridge> <ryozuki> i didnt rly change the code at all
14:01 < bridge> <Assa> nevermind found the resize
14:01 < bridge> <ryozuki> logic wise
14:01 < bridge> <ryozuki> the resize probably never happens anyway
14:01 < bridge> <ryozuki> its just a safeguard
14:02 < bridge> <ryozuki> because i init the vectors with a resize to num tunezones
14:02 < bridge> <ryozuki> hm
14:02 < bridge> <Assa> yeah I've seen that, is that necessary  or what benefit do you see in that?
14:02 < bridge> <ryozuki> hm i think its more predicatable
14:02 < bridge> <Assa> othwise 99.7% of the maps will initialize 256 empty strings 😄
14:02 < bridge> <ryozuki> but is oninit called more than once?
14:03 < bridge> <ryozuki> a string is a ptr and 2 ints
14:03 < bridge> <ryozuki> and a empty string should not allocate at all
14:03 < bridge> <ryozuki> i guess i can remove it but i doubt it matters much
14:03 < bridge> <ryozuki> actually il lchange this to a reset
14:03 < bridge> <ryozuki> to make sure its cleaned
14:04 < bridge> <ryozuki> if for some reason context is inited again
14:05 < bridge> <Assa> waiting for the refresh then, also wanted to add that it's a full refresh and not only a resize
14:05 < bridge> <ryozuki> pushed
14:06 < bridge> <ryozuki> > IGameServer *CreateGameServer() { return new CGameContext; }
14:07 < bridge> <ryozuki> @essigautomat https://github.com/ddnet/ddnet/pull/10003#discussion_r2024688675
14:07 < bridge> <ryozuki> where do u mean by here
14:07 < bridge> <Jupstar ✪> @ryozuki u never initialize your vectors
14:07 < bridge> <ryozuki> u dont need a cast to compare to zero
14:07 < bridge> <Jupstar ✪> ah you do
14:07 < bridge> <ryozuki> it works tho
14:07 < bridge> <Jupstar ✪> but why do you include array?
14:08 < bridge> <Jupstar ✪> clangd?
14:08 < bridge> <ryozuki> ehm
14:08 < bridge> <Assa> I mean that line: `if(m_TuneZoneOld >= 0 && static_cast<size_t>(m_TuneZone) < GameServer()->m_vZoneEnterMsg.size() && !GameServer()->m_vZoneLeaveMsg[m_TuneZoneOld].empty())`
14:08 < bridge> <ryozuki> vscode likes to include everything
14:08 < bridge> <Assa> you wrote one time `m_TuneZone` instead of `m_TuneZoneOld`
14:08 < bridge> <Jupstar ✪> @ryozuki do you think a vector is really better?
14:08 < bridge> <Jupstar ✪> 
14:08 < bridge> <Jupstar ✪> an array of 256 and u8 directly makes sure the index is always in bound
14:08 < bridge> <ryozuki> oh
14:08 < bridge> <ryozuki> a typo
14:08 < bridge> <ryozuki> copy pasting xd
14:09 < bridge> <Assa> yes :owo:
14:09 < bridge> <ryozuki> nice catch
14:09 < bridge> <ryozuki> i love git commit --amend
14:11 < bridge> <ryozuki> idk, i think its fine to not waste memory since tune zones are rarely used anyway
14:11 < bridge> <Jupstar ✪> i think it's better to waste memory than overcomplicate code
14:11 < bridge> <ryozuki> :justatest:
14:11 < bridge> <ryozuki> @learath2 whats ur opinion on this matter sir
14:11 < bridge> <Assa> I don't think that code is overcomplicated like this xD
14:11 < bridge> <Aoe'nt> Is it really overvomplicated tho?
14:12 < bridge> <Aoe'nt> Is it really overcomplicated tho?
14:12 < bridge> <Jupstar ✪> it's much harder to read, you cannot assume the index (if u8) is not out of bounds
14:12 < bridge> <ryozuki> the added benefit is u can have tune zone msgs larger than 256
14:12 < bridge> <ryozuki> actually not
14:12 < bridge> <ryozuki> abuf is 256 still
14:13 < bridge> <ryozuki> ah no
14:13 < bridge> <ryozuki> i removed abuf
14:13 < bridge> <ryozuki> yeah u can have bigger xd
14:13 < bridge> <ryozuki> omg i hate my c++
14:13 < bridge> <Aoe'nt> :monkalaugh:
14:13 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1356964751498022912/image.png?ex=67ee7b67&is=67ed29e7&hm=564b445d3632b91fa8ed31e237b209a01b1a0f24be5319456b606c433aea9aec&
14:13 < bridge> <ryozuki> ok no
14:13 < bridge> <ryozuki> github likes to hide code in git diffs
14:13 < bridge> <Jupstar ✪> yeah but we don't have tune_zones over u8
14:13 < bridge> <Jupstar ✪> the std::string can also be used with an array
14:13 < bridge> <Jupstar ✪> that is independent
14:13 < bridge> <Assa> this is not true for linebraked strings ... wtf
14:14 < bridge> <ryozuki> yeah just realized
14:14 < bridge> <Assa> using std::string makes the code less complicated there xD
14:14 < bridge> <ryozuki> imho yeah xd
14:14 < bridge> <ryozuki> no str copy
14:15 < bridge> <Assa> does std::string have a split function we could use there?
14:15 < bridge> <Assa> certainly
14:15 < bridge> <ryozuki> anyway i have no strong opinion on this matter, feel free to decide and merge or close
14:16 < bridge> <ryozuki> https://en.cppreference.com/w/cpp/string/basic_string/getline
14:17 < bridge> <kebscs> Please open an issue before pr!!!
14:17 < bridge> <Assa> I suggest actually fixing the linebreak issue now, since it wasn't one before
14:17 < bridge> <kebscs> What is this Ryozuki
14:17 < bridge> <Assa> this is a refactoring, as I mentioned these should be excluded from issue creation
14:18 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1356966094677479506/image.png?ex=67ee7ca8&is=67ed2b28&hm=748da5c407ece9b1630728c7a6d8b1f1b792ce80103b782db5e5f7e69f0313e9&
14:18 < bridge> <ryozuki> should work tm
14:19 < bridge> <ryozuki> @jupeyy_keks is this clean
14:19 < bridge> <ryozuki> let me test
14:19 < bridge> <ryozuki> works
14:20 < bridge> <Jupstar ✪> honestly i forgot how the cpp streams work xD
14:20 < bridge> <ryozuki> pushed
14:20 < bridge> <Assa> also do we have a limit for server tuning lengths in maps?
14:20 < bridge> <ryozuki> idk
14:20 < bridge> <Assa> found the same solution on stack overflow for this
14:21 < bridge> <Jupstar ✪> some of the cpp apis manages to be so counter intuitive
14:22 < bridge> <ryozuki> which is what u used kek
14:22 < bridge> <ryozuki> i*
14:22 < bridge> <ryozuki> my keyboard is crazy
14:23 < bridge> <Assa> does getline return the full string if no linebreak is found?
14:23 < bridge> <Jupstar ✪> @ryozuki now you got the heinrich ban hammer on the vec approach anyway
14:23 < bridge> <Jupstar ✪> 🫠
14:23 < bridge> <ryozuki> xdd
14:23 < bridge> <ryozuki> i leave it for a day for more opinions
14:23 < bridge> <ryozuki> then if not ill change
14:24 < bridge> <Assa> > Extracts characters from is and stores them into str until the delimitation character delim is found (or the newline character, '\n', for (2)).
14:24 < bridge> <Assa> okay
14:24 < bridge> <ryozuki> or eof
14:27 < bridge> <learath2> Which part of it do you require an opinion on? Not wasting memory is good. Wasting memory is bad
14:28 < bridge> <cellegenrih> Yup, what did I expect from a guy who banned me because of an emote
14:28 < bridge> <cellegenrih> https://cdn.discordapp.com/attachments/293493549758939136/1356968556218028122/image.png?ex=67ee7ef3&is=67ed2d73&hm=2a2333be0c9349a3a647831b3d5f9cbd3fa5e400a2fd07b76989da83380196e8&
14:28 < bridge> <learath2> I'd have used an array of `std::string`s though not a vec, they are small enough
14:29 < bridge> <Jupstar ✪> How should I read that?
14:29 < bridge> <Assa> I need an "approve that code is working and to my standards, but I don't know if the concept is the right one to use"-button
14:29 < bridge> <Jupstar ✪> I didn't follow
14:29 < bridge> <learath2> Or if you want to avoid the modernity and embrace teeworlds, a CHeap of CStrings? 😄
14:29 < bridge> <Jupstar ✪> Use a dynamic arena ryo
14:30 < bridge> <cellegenrih> All I wrote in the "less toxicity in the repo" post is
14:30 < bridge> <cellegenrih> - 10/10 approve
14:30 < bridge> <Jupstar ✪> dynamic bulk arena
14:30 < bridge> <Jupstar ✪> Oh is that PR relevant, lemme check
14:30 < bridge> <cellegenrih> #9999
14:30 < bridge> <DDNet> https://github.com/ddnet/ddnet/pull/9999
14:31 < bridge> <Jupstar ✪> Ok was just a april pr i see
14:31 < bridge> <Jupstar ✪> and heinrich didnt get it xD
14:31 < bridge> <cellegenrih> Like it's just one comment removing profanity, but even if it wasn't an april fools joke, well
14:31 < bridge> <cellegenrih> can I not comment on this? xd
14:31 < bridge> <Jupstar ✪> dunno how it's related to your comment tho
14:31 < bridge> <Assa> no it was an april fools joke, maybe I deserve an outtime for this one .. xD
14:31 < bridge> <Jupstar ✪> You wasted the holy number 9999 for aprils fools joke
14:31 < bridge> <learath2> It seems you did comment. I don't follow
14:31 < bridge> <Jupstar ✪> How dare you
14:31 < bridge> <cellegenrih> I just simply commented "10/10 approve" as a positive thing
14:31 < bridge> <cellegenrih> and removed xd
14:32 < bridge> <Assa> Yes XD brilliant, the stars aligned
14:32 < bridge> <learath2> I still see your comment fwiw
14:32 < bridge> <Jupstar ✪> me too
14:32 < bridge> <Jupstar ✪> I am very confused xD
14:32 < bridge> <cellegenrih> eh?
14:32 < bridge> <learath2> I am very very confused
14:32 < bridge> <cellegenrih> same as you guys
14:33 < bridge> <Jupstar ✪> https://github.com/ddnet/ddnet/pull/9999#issuecomment-2770312616
14:33 < bridge> <cellegenrih> didn't check my github yet, but this message just arrived now
14:33 < bridge> <cellegenrih> 3 hours before
14:33 < bridge> <cellegenrih> idfk dude
14:34 < bridge> <Assa> This one is even more infuriating:
14:34 < bridge> <Assa> > Nice to see you're working on yourself 😉
14:34 < bridge> <Assa> Thanks I guess? xDD
14:34 < bridge> <cellegenrih> Or perhaps I'm a huge retard and the whole repo's name is "remove comment"?
14:34 < bridge> <cellegenrih> if that's the case, then I got april fooled idk :greenthing:
14:35 < bridge> <Jupstar ✪> Chairn defs nailed it
14:35 < bridge> <Jupstar ✪> The only comment that made the joke better
14:35 < bridge> <learath2> The name of the pr was "remove comment"
14:35 < bridge> <cellegenrih> Bro i wanna kms I really thought heinrich removed my comment XD
14:35 < bridge> <cellegenrih> cuz it was the repo's name
14:35 < bridge> <cellegenrih> and not an action from github
14:35 < bridge> <cellegenrih> 😄
14:36 < bridge> <cellegenrih> fuck man
14:36 < bridge> <Jupstar ✪> xD
14:36 < bridge> <Jupstar ✪> you got april fooled too
14:36 < bridge> <cellegenrih> I was like: why the fuck
14:36 < bridge> <milkeeycat> technologies are hard
14:36 < bridge> <cellegenrih> I bet he changed the name of the post's header
14:37 < bridge> <learath2> My eye is extremely itchy, but itching it makes it itch more
14:37 < bridge> <cellegenrih> wash it with water
14:37 < bridge> <Assa> heinrich renamed the PR, so that it doesn't appear in the git history
14:37 < bridge> <cellegenrih> no fucking way
14:37 < bridge> <Assa> which is understandable
14:38 < bridge> <cellegenrih> he wins
14:38 < bridge> <cellegenrih> got fooled even if I didn't want to
14:38 < bridge> <Assa> get fooled :nouis:
14:38 < bridge> <cellegenrih> :nouis:
14:40 < bridge> <Assa> your pipeline failed btw 🙈
15:06 < bridge> <ryozuki> oh
15:10 < bridge> <kebscs> why not use std::map?
15:11 < bridge> <kebscs> right now if you have only msg for zone 255 itll have size 255 even though only 1 item
15:11 < bridge> <kebscs> right now if you have only msg for zone 255 itll resize to 255 even though only 1 item
15:11 < bridge> <kebscs> @ryozuki
15:14 < bridge> <Assa> std::map is logarithmic in access time
15:14 < bridge> <Assa> std::map is logarithmic in access time, to be pedantic
15:15 < bridge> <learath2> Yeah if you truly want that space saved too an associative container is the way to go since the array is sparse. But usually people will use the tune zones in order
15:15 < bridge> <kebscs> but pr reduces size
15:15 < bridge> <ryozuki> cuz hash?
15:15 < bridge> <learath2> I would just use a normal array
15:15 < bridge> <Assa> yes, but this one is about tune_zone_enter and tune_zone_leave, this is not guaranteed to be in order!
15:15 < bridge> <Jupstar ✪> if at all use unordered_map
15:15 < bridge> <Jupstar ✪> which is a hashmap
15:16 < bridge> <kebscs> unordered_map yea
15:16 < bridge> <Jupstar ✪> But I'd say if 256 is your upper limit for indices, using a static sized array is often a nice way to go
15:16 < bridge> <learath2> Cuz it's ordered and an ordered map is usually implemented as a tree and tree access is log
15:16 < bridge> <Assa> actually no idea why, could be hash, could be binary tree
15:17 < bridge> <kebscs> std::map is red-black tree usually
15:17 < bridge> <ryozuki> i miss rust
15:17 < bridge> <ryozuki> isnt unordered_map slow
15:17 < bridge> <ryozuki> cuz std
15:17 < bridge> <kebscs> no
15:17 < bridge> <ryozuki> it is
15:17 < bridge> <kebscs> std good 👍
15:17 < bridge> <ryozuki> no lol
15:17 < bridge> <ryozuki> the standard library's maps are, indeed, inherently slow. Google's Chandler Carruth explains this in his CppCon 2014 talk; in a nutshell:
15:17 < bridge> <ryozuki> std::map maintains its data's sort order, leading to a tree-based implementation; and trees are slow: non-contiguous with a lot of pointer chain chasing.
15:17 < bridge> <ryozuki>     std::unoredered_map is not as-bad, but it's still cache-unfriendly because it uses linked lists as buckets.
15:17 < bridge> <learath2> It is slower than it can be because of the standards de-facto requirement that the hashmap is implemented with closed addressing
15:18 < bridge> <ryozuki> yeah cuz the standard
15:18 < bridge> <ryozuki> no way around it
15:18 < bridge> <learath2> But this is just over over over optimizing 😄
15:18 < bridge> <ryozuki> open addressing is faster
15:18 < bridge> <Jupstar ✪> the problem with unordered_map is that you might not save lot more RAM in the end for 256 maximum
15:18 < bridge> <Jupstar ✪> i dunno what the default bucket size is
15:18 < bridge> <Assa> std::unordered_map is constant or linear in the worst case
15:18 < bridge> <Jupstar ✪> But probably not like 1 xd
15:18 < bridge> <learath2> Something that isn't even a huge issue to begin with. I would just put it in an array and call it a day
15:19 < bridge> <ryozuki> perfect hashing
15:19 < bridge> <ryozuki> lets over over optimize
15:19 < bridge> <ryozuki> for fun
15:19 < bridge> <ryozuki> and no profit
15:19 < bridge> <kebscs> that is for initalizing it
15:19 < bridge> <kebscs> you initalize once in this case and then its o(1) for unordered map
15:20 < bridge> <ryozuki> i meant that the std impl is inherently slow
15:20 < bridge> <ryozuki> compared to state of the art
15:20 < bridge> <kebscs> dont think its an issue when it runs only once
15:20 < bridge> <ryozuki> rust uses hashmap with quadratic probing and simd lookup
15:20 < bridge> <kebscs> rust :cammostripes:
15:20 < bridge> <ryozuki> c++  :cammostripes:
15:21 < bridge> <learath2> Not quite O(1), it degenerates to O(n), but it is on the average O(1)
15:21 < bridge> <ryozuki> what is the default hashing algo for std
15:22 < bridge> <kebscs> good hashing algo for sure
15:22 < bridge> <kebscs> c++ 💪
15:23 < bridge> <ryozuki> like c++ regex, which is slower than starting php
15:23 < bridge> <learath2> Anyway, a sparse data structure might not even help here since as jupman noted the default buckets can be too large or the hash space can be too large. Which for 256 elements is probably the case
15:24 < bridge> <kebscs> no map has 256 tune msgs
15:24 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1356982515654459544/ORL3Z.png?ex=67ee8bf3&is=67ed3a73&hm=b72f22cffb398a2f9355302455f7df4b0454206fbd64884102854f079b361551&
15:24 < bridge> <learath2> If you really want to turbooptimize this for some reason, you need to make sure to design your own sparse datastructure, maybe a btree could work
15:24 < bridge> <learath2> But that's logn, not o1
15:24 < bridge> <learath2> But that's logn, not constant
15:24 < bridge> <Jupstar ✪> Well it tested it, seems like it allocs 2 buckets only
15:24 < bridge> <Jupstar ✪> https://cpp.sh/?source=%2F%2F+Example+program%0A%23include+%3Ciostream%3E%0A%23include+%3Cstring%3E%0A%0Aint+main()%0A%7B%0A++std%3A%3Astring+name%3B%0A++std%3A%3Acout+%3C%3C+%22What+is+your+name%3F+%22%3B%0A++getline+(std%3A%3Acin%2C+name)%3B%0A++std%3A%3Acout+%3C%3C+%22Hello%2C+%22+%3C%3C+name+%3C%3C+%22!%5Cn%22%3B%0A%7D
15:24 < bridge> <learath2> Re2 beats rust regex, wow, c++ best
15:25 < bridge> <Jupstar ✪> I assume it switches it's internal algorithm quite often
15:25 < bridge> <Jupstar ✪> oh
15:25 < bridge> <Jupstar ✪> that didnt work
15:25 < bridge> <Jupstar ✪> https://cpp.sh/?source=%2F%2F+Example+program%0A%23include+%3Ciostream%3E%0A%23include+%3Cstring%3E%0A%23include+%3Cunordered_map%3E%0A%0Aint+main()%0A%7B%0A++std%3A%3Aunordered_map%3Cunsigned+char%2C+std%3A%3Astring%3E+name%3B%0A++name.insert(%7B0%2C+%22%22%7D)%3B%0A++name.insert(%7B255%2C+%22%22%7D)%3B%0A++%0A++std%3A%3Acout+%3C%3C+%22Hello%2C+%22+%3C%3C+name.bucket_count()+%3C%3C+%22!%5Cn%22%3B%0A++for+(int+i+%3D+0%3B+i+%3C+name.bucket_count
15:25 < bridge> <ryozuki> not in many cases
15:25 < bridge> <Jupstar ✪> now
15:25 < bridge> <kebscs> c
15:25 < bridge> <kebscs> c++ always on top
15:25 < bridge> <ryozuki> btw the image is 8y old
15:25 < bridge> <ryozuki> 3
15:25 < bridge> <ryozuki> *
15:25 < bridge> <ryozuki> 3y
15:26 < bridge> <kebscs> rust definetly slower in these 3 years
15:26 < bridge> <ryozuki> isnt it impl defined
15:26 < bridge> <Jupstar ✪> yes, but i only care about linux perf ofc
15:26 < bridge> <Jupstar ✪> which this tool probs uses
15:26 < bridge> <ryozuki> well linux has 2 compilers
15:26 < bridge> <ryozuki> gcc and llvm
15:26 < bridge> <Jupstar ✪> i assume if u compile with clang u rarely use the clang cpp runtime
15:26 < bridge> <Jupstar ✪> but yeah if you want to test
15:27 < bridge> <Jupstar ✪> go ahead
15:27 < bridge> <ryozuki> xd
15:27 < bridge> <ryozuki> i wonder the lib std from llvm is faster
15:27 < bridge> <Jupstar ✪> afaik it's often equal, sometimes slower
15:27 < bridge> <Jupstar ✪> But things change
15:28 < bridge> <Jupstar ✪> especially since android does not really support gcc anymore
15:28 < bridge> <Jupstar ✪> google pushes
15:30 < bridge> <Jupstar ✪> where u have that from btw?
15:30 < bridge> <ryozuki> so
15:30 < bridge> <Jupstar ✪> ah too bad
15:32 < bridge> <Jupstar ✪> @ryozuki i think for hashmaps it was even worse or not?
15:32 < bridge> <Jupstar ✪> the cpp std one is rather slow.
15:32 < bridge> <Jupstar ✪> 
15:32 < bridge> <Jupstar ✪> 
15:32 < bridge> <Jupstar ✪> But let's face it, u rarely notice a diff in real world anyway xdd
15:40 < bridge> <Jupstar ✪> Look at this non scientific benchmark:
15:40 < bridge> <Jupstar ✪> ```
15:40 < bridge> <Jupstar ✪> ctf1, move_box Vec:
15:40 < bridge> <Jupstar ✪> 7.049902133 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 141845.94071442267 call/s
15:40 < bridge> <Jupstar ✪> 6.935108404 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 144193.8527483182 call/s
15:40 < bridge> <Jupstar ✪> 6.815435359 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 146725.76986288454 call/s
15:40 < bridge> <Jupstar ✪> 
15:40 < bridge> <Jupstar ✪> ctf1, move_box FxHashMap:
15:40 < bridge> <Jupstar ✪> 12.252110956 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 81618.58830622886 call/s
15:40 < bridge> <Jupstar ✪> 11.270511348 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 88727.11886115569 call/s
15:40 < bridge> <Jupstar ✪> 11.329539075 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 88264.84408413588 call/s
15:40 < bridge> <Jupstar ✪> 
15:40 < bridge> <Jupstar ✪> ctf1, move_box std HashMap:
15:40 < bridge> <Jupstar ✪> 34.203735266 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:40 < bridge> <Jupstar ✪> 29236.57291296028 call/s
15:40 < bridge> <Jupstar ✪> 33.750160829 s - pos: vec2_base { x: 640.00037, y: 433.02246 }, vel: vec2_base { x: 320.0, y: -0.0 }
15:41 < bridge> <Jupstar ✪> 29629.488436118645 call/s
15:41 < bridge> <Jupstar ✪> ```
15:41 < bridge> <Jupstar ✪> 
15:41 < bridge> <Jupstar ✪> 
15:41 < bridge> <Jupstar ✪> The hash function is what makes it super slow xdd
15:41 < bridge> <Jupstar ✪> 
15:41 < bridge> <Jupstar ✪> I changed the map game layer tiles from vec to hashmap.
15:41 < bridge> <Jupstar ✪> 
15:41 < bridge> <Jupstar ✪> The FxHashMap is not even 2x slower.
15:44 < bridge> <hectavoxel> blazingly slow? :troll:
15:44 < bridge> <Jupstar ✪> only trust benchmarks
15:46 < bridge> <kebscs> https://cdn.discordapp.com/attachments/293493549758939136/1356988268620550205/image.png?ex=67ee914e&is=67ed3fce&hm=90204456e87bf9cfb8f75d36c5170cff48ff8534f793cb10396575c5bdfdd8fe&
15:49 < bridge> <Assa> I always wondered if ddnet would benefit from a quad tree, as we can optimize out so many collision checks
15:49 < bridge> <ryozuki> yeah
15:49 < bridge> <Jupstar ✪> we already talked about that lot of times
15:49 < bridge> <Jupstar ✪> quad tree is probs too slow
15:49 < bridge> <Assa> I know
15:49 < bridge> <ryozuki> we cant change physics too
15:49 < bridge> <Jupstar ✪> but spatial hashing
15:50 < bridge> <Jupstar ✪> i used a spatial hashing in the vanilla impl of ddnet-rs and it easily beats all other implementations for many tees
15:50 < bridge> <Jupstar ✪> well i used a optimized form of spatial hasing i'd say
15:50 < bridge> <Jupstar ✪> the definition sounds more generalized
15:52 < bridge> <Jupstar ✪> 64-tee benchmark:
15:52 < bridge> <Jupstar ✪> vanilla
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 32534 t/s
15:52 < bridge> <Jupstar ✪> 33871 t/s
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 
15:52 < bridge> <Jupstar ✪> twgame:
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 9725 t/s
15:52 < bridge> <Jupstar ✪> 10219 t/s
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 
15:52 < bridge> <Jupstar ✪> 512-tee benchmark:
15:52 < bridge> <Jupstar ✪> vanilla
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 3766 t/s
15:52 < bridge> <Jupstar ✪> 3771 t/s
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 
15:52 < bridge> <Jupstar ✪> twgame:
15:52 < bridge> <Jupstar ✪> ```
15:52 < bridge> <Jupstar ✪> 346 t/s
15:52 < bridge> <Jupstar ✪> 299 t/s
15:53 < bridge> <Jupstar ✪> ```
15:53 < bridge> <Jupstar ✪> on a 9950x3d, which i'll not be able to reproduce anymore
15:53 < bridge> <Jupstar ✪> 😬
15:53 < bridge> <ryozuki> what is t
15:53 < bridge> <ryozuki> tick?
15:53 < bridge> <Jupstar ✪> yeah
15:53 < bridge> <Jupstar ✪> _random_ input
16:18 < bridge> <matodor> who has access to master server bans? heinrich only?
16:19 < bridge> <learath2> Legacy masger only heinrich, the httpmaster any ddnet admin
16:19 < bridge> <learath2> Legacy master only heinrich, the httpmaster any ddnet admin
16:22 < bridge> <matodor> for httpmaster i can open ticket for report server?
16:22 < bridge> <learath2> Yep an admin-mail ticket would be appropriate
16:24 < bridge> <Assa> can you give me info, after you got an admin answer?
16:26 < bridge> <matodor> I think not, the problem has already been resolved, after talking to the server administrator, they stole the map and made the same server name (some schoolboy) :nouis:
16:37 < bridge> <_qey> Been there…
16:38 < bridge> <_qey> Was there ever a DDNet version 18091?
16:38 < bridge> <_qey> On a legit official client?
16:39 < bridge> <_qey> Also 14062?
16:41 < bridge> <_qey> 18091 translates to 18.9.1? 14062 to 14.6.2?
16:41 < furo> 18091 is 18.9.1, yes.
16:44 < bridge> <_qey> Though they usually close the server within a week or two due to lack of player activity.
16:44 < bridge> <Assa> @jupeyy_keks before we discuss much on github, want to do it quickly here?
16:45 < bridge> <Jupstar ✪> You have few minutes then my package arrives
16:46 < bridge> <Assa> I answered:
16:46 < bridge> <Assa> > This is just Step one, next step would be to add functions taking LayerType as an argument and generalizing over all of this layer checks. Also, these bools are exclusive, you can't have a tileLayer, which is a switch AND a tune layer for example.
16:46 < bridge> <Assa> &
16:46 < bridge> <Assa> > Additionally in cases like LayerType <= LAYER_FRONT i collabs 3 times duplicated code to 1
16:46 < bridge> <Assa> In the mock I made I had like 8 functions, and 2 with switch cases for LayerType, the rest was generalized out
16:47 < bridge> <Jupstar ✪> I am ok with collapsing the if stuff
16:47 < bridge> <Jupstar ✪> it's more about readability
16:47 < bridge> <robyt3> Reminder to check performance in release mode, evaluating the performance difference in debug build is not meaningful. The FPS in both screenshots look pretty low to me
16:48 < bridge> <Assa> This was build in release mode
16:48 < bridge> <Assa> the debug console works in release mode.
16:48 < bridge> <robyt3> How many FPS do you have on a small/empty map?
16:49 < bridge> <Assa> I capped my game to 990, so I guess about 2K?
16:49 < bridge> <Jupstar ✪> Honestly 20 more fps for that change is defs not true
16:49 < bridge> <Jupstar ✪> What should be so much faster?
16:49 < bridge> <Assa> I never claimed that
16:49 < bridge> <Assa> > I checked if this impacts performance, and I can't tell if it does or not, which is good.
16:49 < bridge> <Jupstar ✪> Ok
16:49 < bridge> <teero777> @jupeyy_keks package arriving???
16:49 < bridge> <Jupstar ✪> Got baited
16:49 < bridge> <Jupstar ✪> Yes
16:49 < bridge> <teero777> when
16:50 < bridge> <Jupstar ✪> I dunno, probs next 10-15mins
16:50 < bridge> <Jupstar ✪> it only gives an estimate
16:50 < bridge> <teero777> cam?
16:50 < bridge> <Assa> `Configure->Release` FYI
16:50 < bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1357004215108505761/Unbenannt.PNG?ex=67eea028&is=67ed4ea8&hm=356707753e462e9bb964f40e08aa972156650e9941ad2e13eb4cf3bd7d2132b6&
16:50 < bridge> <Jupstar ✪> I dunno how I should hold the cam with only 2 hands, which i need to build @teero777 soz xD
16:50 < bridge> <teero777> use headlight
16:50 < bridge> <Aoe'nt> leak address i come over and hold
16:50 < bridge> <Jupstar ✪> Alr
16:50 < bridge> <Jupstar ✪> Fly over
16:51 < bridge> <teero777> @jupeyy_keks
16:51 < bridge> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1357004404007370904/iu.png?ex=67eea055&is=67ed4ed5&hm=48e826d9ae4ae3c1366f284ee114b9905538f95f26d51679485699614ebc4da6&
16:51 < bridge> <Jupstar ✪> ok ok, i'll quickly search smth xDD
16:52 < bridge> <Assa> 4900 - 5xxx .... can't tell
16:54 < bridge> <robyt3> So rendering on this map is just very slow in general
16:54 < bridge> <Assa> this is an extreme case, yes
16:54 < bridge> <Assa> With my PR
16:54 < bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1357005357087723732/screenshot_2025-04-02_16-53-26.png?ex=67eea139&is=67ed4fb9&hm=00552dafb0a960ccbac453170060b2e300b88bee43a642ee3c14b82b4835c5f6&
16:55 < bridge> <Assa> Wihtout my PR
16:55 < bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1357005405498118415/screenshot_2025-04-02_16-54-03.png?ex=67eea144&is=67ed4fc4&hm=0b58c1b677dec1c442a76f430f13eeb4e3b676d0fe7cc59b09aafc44dcab6c1e&
16:55 < bridge> <Assa> this looks bad, but there is no difference, like my FPS is heavily impacted by other things and flying around like crazy
16:55 < bridge> <_qey> I have 67 fps!
16:56 < bridge> <_qey> Never crossed my mind DDNet could run at 5k fps, lol.
16:56 < bridge> <Assa> easily with a last gen graphics card on a 2K monitor
16:58 < bridge> <Assa> I don't know how to do proper fps benchmarking for ddnet
17:00 < bridge> <Solly (bun bun)> outlining a function shouldnt affect performance at all
17:02 < bridge> <Assa> only outlining it, no, optimizing it might
17:02 < bridge> <Solly (bun bun)> doesnt look like much changed
17:02 < bridge> <Assa> might be slightly faster as it can early exit and needs like 7 comparisons less on average, witch is nothing on 3.7GHz
17:03 < bridge> <Assa> and then cache effects and measurement is in the mud anyway
17:03 < bridge> <Solly (bun bun)> this will be rearanged to an if else chain
17:03 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1357007504978215023/image.png?ex=67eea339&is=67ed51b9&hm=5cbeb70bc686d9a05365b601b5ac7538acea727b0fef6c77cfda04788d55ef32&
17:03 < bridge> <Solly (bun bun)> given compilesr are smart
17:03 < bridge> <Solly (bun bun)> well im also basing my "not much has changed" on the benchmark you did, so a bit circular assumptions lmao
17:03 < bridge> <Assa> I did anyways, why not
17:04 < bridge> <Solly (bun bun)> why not :)
17:04 < bridge> <Assa> nothing has changed in terms of FPS
17:04 < bridge> <Solly (bun bun)> ik
17:04 < bridge> <Assa> Like how should I know if I have an fps on average more or not? xD
17:04 < bridge> <_qey> I wonder how everyone’s day going while I’m compiling the server build?
17:10 < bridge> <chairn> troll pr, troll comment 😄
17:10 < bridge> <chairn> it felt like a linus torvalds moment, kinda
17:13 < bridge> <milkeeycat> @ryozuki i will add `icmp` instruction to my compiler backend and will try to use it in my language
17:13 < bridge> <chairn> is it time or throughput ? why are some cases highlighted and others aren't ?
17:15 < bridge> <ryozuki> nice
17:16 < bridge> <Jupstar ✪> @teero777 discord too buggy. Doesn't see my camera
17:16 < bridge> <ryozuki> doesnt load
17:23 < bridge> <Jupstar ✪> https://call.element.io/room/#/jupstarbuild?password=y9B3WMnToUCsIK6Og3y2TA&roomId=%21KTrwlQsuOmfuZSVEpe%3Acall.ems.host
17:23 < bridge> <Jupstar ✪> @teero777 go
17:23 < bridge> <Jupstar ✪> jez oder nie
17:50 < bridge> <kondensator> Hey guys, where are skins like `horsouly` `4kultrahdskin` etc coming from?
17:50 < bridge> <DDNet> Hellou :3
17:51 < bridge> <melon> we blame @soulyvevo for that
17:51 < bridge> <soulyvevo> im gonna punch you
17:51 < bridge> <melon> yes please
17:51 < bridge> <kondensator> hehe, where are those downloaded from?
17:51 < bridge> <soulyvevo> 😳
17:52 < bridge> <soulyvevo> those skins were in the database as a part of april fools joke, skins already got removed from the database and arent visible on the website
17:53 < bridge> <soulyvevo> but due to how skin downloading works it will take some time till those skins will be completely gone
17:53 < bridge> <melon> can we add them if i make them saddo?
17:53 < bridge> <soulyvevo> yes
17:53 < bridge> <kondensator> ahhh crap
17:53 < bridge> <kondensator> that's why I couldn't find them
17:53 < bridge> <kondensator> but they still have to be somewhere
17:53 < bridge> <kondensator> like I cannot download what's not there
17:54 < bridge> <kondensator> like if `https://skins.ddnet.org/skin/community/yomaheart_yomasoul.png` does not exist then how the hell I am able to have it ingame
17:55 < bridge> <patiga> because you have it as a file locally
17:55 < bridge> <soulyvevo> its probably still in your downloadedskins folder
17:55 < bridge> <patiga> $config
17:55 < bridge> <DDNet> This command was not found.
17:55 < bridge> <patiga> $config_dir
17:55 < bridge> <DDNet> This command was not found.
17:55 < bridge> <melon> $configdir
17:55 < bridge> <patiga> :c
17:55 < bridge> <melon> poor patiga
17:55 < bridge> <melon> xD
17:55 < bridge> <soulyvevo> removing skins from the database doesnt remove it from that directory
17:56 < bridge> <soulyvevo> anyway if you want to check out all of those skins there you go
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020876369039461/4kultrahdskin.png?ex=67eeafad&is=67ed5e2d&hm=0d40895ffcfdac31edcb5980075720e560130d36e79261e71635e7f39439c5c6&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020876830281728/me_when_fine_shyt.png?ex=67eeafad&is=67ed5e2d&hm=92b660683011a00446500c9347a62fba372f2fd787e64c7b9aff21b483657315&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020877174079728/nahbruhnowayhellnah.png?ex=67eeafad&is=67ed5e2d&hm=95473d67a471d2eebbd06bc73fd77a6c8d706a85be4bb0f5e69a49e72a61c029&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020877526405160/ilovesellingcopper.png?ex=67eeafad&is=67ed5e2d&hm=0b3e2c2125aec842d61b3d2414034260b868f7d93ef47af43789c4b75f513369&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020877887373493/horsouly.png?ex=67eeafad&is=67ed5e2d&hm=6892571661f02ae65c21afdfd3ae680ce1be51f6c192e3ccd2330bd38af1e90c&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020878369587331/muchlovefromddnetteam.png?ex=67eeafad&is=67ed5e2d&hm=e613410dc66f7f7da824bdf24237c9f978dc88eeeea8c802b184cfb4f4c2cb37&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020879086944527/78fwa9r983.png?ex=67eeafad&is=67ed5e2d&hm=86ddfd709091f4d65dc6cfe3e376043b8e0c8ddb6dc01c9f503df3e13ea215da&
17:56 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357020879422357574/yomaheart_yomasoul.png?ex=67eeafad&is=67ed5e2d&hm=8901d2d5db1820675df45a717fce5fe387f89a97e7295d3f36f2a252f4780c2f&
17:56 < bridge> <soulyvevo> true work of art
17:56 < bridge> <kondensator> ohh, that's interesting, I did not realise I had them there
18:01 < bridge> <soulyvevo> so i made something like this
18:01 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357022204159594546/snowflake_icons.png?ex=67eeb0e9&is=67ed5f69&hm=abe83bb8449fba81fdd5498eacaef5677aaa73abcf1dac96d2a11d2ed37db313&
18:01 < bridge> <soulyvevo> but im not really sure about it
18:02 < bridge> <soulyvevo> because it looks weird with that fast changing animation
18:02 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357022376088571935/coverage_2025-04-02_17-59-36.mp4?ex=67eeb112&is=67ed5f92&hm=b39f7f7d5a7f7b4bef24b5a01a593d0287030ad1aee09859163e3e5226286495&
18:02 < bridge> <Assa> I think I wrote above, that we might only need the background for the one existing snowflake 🤔
18:03 < bridge> <Assa> one thing looking weird is the background ofc
18:03 < bridge> <soulyvevo> ye could be worth checking how it looks with that hexagon shape as bg
18:04 < bridge> <Assa> I think a single snowflake sprite rotating fast might be enough
18:04 < bridge> <soulyvevo> how are those outlines even generated in game
18:04 < bridge> <soulyvevo> it looks so weird lol
18:04 < bridge> <Assa> usually they are the original sprite, bug bigger and darker color
18:04 < bridge> <Assa> (it's the otherway round but whatever)
18:04 < bridge> <soulyvevo> aah i see
18:04 < bridge> <Assa> usually they are the original sprite, but bigger and darker color
18:05 < bridge> <Assa> scaling is not the same as adding a pixel on each site, so you'll have issues on any non concarve shape (like a snowflake)
18:05 < bridge> <Assa> and that's why it needs it's own background sprite
18:07 < bridge> <Assa> yet another crazy idea, use a distance shader or something ... nah
18:07 < bridge> <soulyvevo> @essigautomat i tried to make it match that snowflake in extras
18:07 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357023709910663199/hexagon.png?ex=67eeb250&is=67ed60d0&hm=4966e4bfa7c1dff40ebc2b856ca7034ee2d652facdf868edb8760c439b0f3a39&
18:08 < bridge> <Assa> nice :3
18:08 < bridge> <soulyvevo> the issue i have is that snowflake is kinda big
18:08 < bridge> <Assa> yes don't worry, I'll have to scale it down
18:08 < bridge> <Assa> but don't expect immediate results, the door brick isn't even merged yet 😦
18:08 < bridge> <soulyvevo> aight
18:10 < bridge> <soulyvevo> i have a few more designs if thats ever needed
18:10 < bridge> <soulyvevo> https://cdn.discordapp.com/attachments/293493549758939136/1357024385147601097/image.png?ex=67eeb2f1&is=67ed6171&hm=62cf86ee5dc543ec58600573a0b988f8aa5e1104dc16aaca23d2e02933db6885&
18:11 < bridge> <inv41idu53rn4m3> generate snowflakes at runtime
18:11 < bridge> <inv41idu53rn4m3> because no two snowflakes are the same
18:14 < bridge> <Assa> hmm
18:16 < bridge> <louis> wait yeah why is it animated its probably better without changing
18:35 < bridge> <Jupstar ✪> new pc again
18:35 < bridge> <Jupstar ✪> https://cdn.discordapp.com/attachments/293493549758939136/1357030585062789160/image.png?ex=67eeb8b7&is=67ed6737&hm=5dfda1bc920b12cd63958ab478b5e064ba67073c28be674456ef9093b126bdfe&
18:35 < bridge> <Jupstar ✪> :deen_star:
18:35 < bridge> <milkeeycat> buy me a new pc :deen_star:
18:36 < bridge> <melon> We big frens now
18:50 < bridge> <Jupstar ✪> for much money u have?
18:50 < bridge> <Assa> github just gave me a unicorn page
18:50 < bridge> <Assa> lol
18:58 < bridge> <Solly (bun bun)> @robyt3 with your pr i still get loading skins at the start
18:58 < bridge> <Solly (bun bun)> 
18:58 < bridge> <Solly (bun bun)> ```cpp
18:58 < bridge> <Solly (bun bun)> // load skins
18:58 < bridge> <Solly (bun bun)>     Refresh([this]() {
18:58 < bridge> <Solly (bun bun)>         GameClient()->m_Menus.RenderLoading(Localize("Loading DDNet Client"), Localize("Loading skin files"), 0);
18:58 < bridge> <Solly (bun bun)>     });
18:58 < bridge> <Solly (bun bun)> ```
18:58 < bridge> <Solly (bun bun)> 
18:58 < bridge> <Solly (bun bun)> i assume that is this
18:58 < bridge> <Solly (bun bun)> 
18:58 < bridge> <Solly (bun bun)> i dont think any skins should be loaded by default
18:58 < bridge> <Solly (bun bun)> 
18:58 < bridge> <Solly (bun bun)> with the really slow network drive this still takes a decent chunk of time to load
19:01 < bridge> <milkeeycat> $0, it will be a birthday gift :kek:
19:01 < bridge> <DDNet> This command was not found.
19:03 < bridge> <robyt3> It only loads the default skin otherwise you wouldn't have a placeholder while other skins load. And the list of skins is loaded so the skins list isn't empty and the client knows if it has to load a skin from skins or downloadedskins.
19:04 < bridge> <robyt3> Also, lazy loading isn't implemented for 0.7 skins and parts
19:05 < bridge> <Solly (bun bun)> its probably that then
19:05 < bridge> <Solly (bun bun)> is that on the todo
19:08 < bridge> <robyt3> For the future, I guess, but probably not in that PR, it's already large
19:10 < bridge> <robyt3> 0.7 skins and parts are pretty small so it should be almost negligible. Listing all the files over the network probably already takes some time for you.
19:12 < bridge> <Solly (bun bun)> im using sshfs
19:12 < bridge> <Solly (bun bun)> listing files *feels* O(1) regardless of size
19:12 < bridge> <Solly (bun bun)> loading a file has a large initial delay but is then very fast
19:12 < bridge> <Solly (bun bun)> so loading lots of small files is very slow
19:22 < bridge> <milkeeycat> @ryozuki do you know why there're signed and unsigned versions of comparison? https://llvm.org/docs/LangRef.html#icmp-instruction
19:24 < bridge> <patiga> less than and greater than should be different for signed/unsigned
19:24 < bridge> <patiga> wait no
19:25 < bridge> <milkeeycat> are there different instructions for signed/unsigned less/greater than? :thonk:
19:25 < bridge> <milkeeycat> i know only of `cmp`
19:26 < bridge> <patiga> ah negative numbers need to be compared differently
19:28 < bridge> <patiga> arg why am I confusing myself so much :D
19:28 < bridge> <patiga> ah yes, negative numbers would be greater than positive numbers
19:28 < bridge> <patiga> because their most significant bit is set in the two's complement
19:31 < bridge> <patiga> hrmpf ignore everything I said, cmp does math and is smart
19:31 < bridge> <patiga> (https://www.felixcloutier.com/x86/cmp)
19:31 < bridge> <patiga> thanks for the refresher
20:18 < bridge> <aegisub> you use cmp for both but check different flags afterwards
20:18 < bridge> <milkeeycat> ye ye, I got it, thanks :happy:
20:28 < bridge> <learath2> (btw just looking at the examples below would have told you why)
20:29 < bridge> <milkeeycat> didn't go that far xd
20:31 < bridge> <learath2> ((or compiling a very small piece of C to see what changes when you do an unsigned vs signed compare 😛 ))
20:32 < bridge> <milkeeycat> flags are magic to me
20:35 < bridge> <milkeeycat> > The terms "above" and "below" are associated with the CF flag and refer to the relationship between two unsigned integer values. The terms "greater" and "less" are associated with the SF and OF flags and refer to the relationship between two signed integer values.
20:35 < bridge> <milkeeycat> Mhm
21:48 < bridge> <louis> nice my fedora broke
21:48 < bridge> <Ewan> rip
22:01 < bridge> <louis> my cursor is huge when hovering over chrome 😭
22:05 < bridge> <oxyzo.> its excited
22:40 < bridge> <louis> if the client expects more fields in a netobject than the server is sending, will it just be unable to join?
22:41 < bridge> <Jupstar ✪> i still cannot deal with pr & issue number being over 10k xD
22:41 < bridge> <Jupstar ✪> looks so verbsoe
22:54 < bridge> <chillerdragon> It’s over 9000
23:03 < bridge> <Teero> halo
23:05 < furo> @louis It will drop the message, assuming the fields don't have a default value specified.
23:06 < bridge> <Jupstar ✪> hey
23:06 < bridge> <Solly (bun bun)> ```js
23:06 < bridge> <Solly (bun bun)> f=a=>{lines=a.split("\n").filter(a=>-1==="/*".indexOf(a.trim()[0])).map(a=>a.trim()).filter(a=>a).filter(a=>!a.toLowerCase().startsWith("function:")),out={param:[],return:[],remark:[],see:[]},head="desc";for(const b of lines){if(b.endsWith(":")){head=b.slice(0,-1).toLowerCase(),head=head.replace("parameters","param"),head.endsWith("s")&&(head=head.slice(0,-1));continue}out[head]=[...(out[head]??[]),b]}out.param=out.param.map(a=>-1===a.index
23:06 < bridge> <Solly (bun bun)> ```
23:06 < bridge> <Solly (bun bun)> average js script
23:06 < bridge> <Solly (bun bun)> ```js
23:06 < bridge> <Solly (bun bun)> f(test)
23:06 < bridge> <Solly (bun bun)> /**
23:06 < bridge> <Solly (bun bun)>  * Compares up to num bytes of two utf8 strings case insensitively.
23:06 < bridge> <Solly (bun bun)>  *
23:06 < bridge> <Solly (bun bun)>  * @param a String to compare.
23:06 < bridge> <Solly (bun bun)>  * @param b String to compare.
23:06 < bridge> <Solly (bun bun)>  * @param num Maximum bytes to compare.
23:06 < bridge> <Solly (bun bun)>  *
23:06 < bridge> <Solly (bun bun)>  * @return `< 0` - String a is less than string b
23:06 < bridge> <Solly (bun bun)>  * @return `0` - String a is equal to string b
23:07 < bridge> <Solly (bun bun)>  * @return `> 0` - String a is greater than string b
23:07 < bridge> <Solly (bun bun)> */
23:07 < bridge> <Solly (bun bun)> ```
23:07 < bridge> <Solly (bun bun)> converts old style docs to jsdoc
23:07 < bridge> <Solly (bun bun)> :)
23:08 < bridge> <chillerdragon> solly
23:11 < bridge> <Solly (bun bun)> yes
23:12 < bridge> <Solly (bun bun)> oh there was a mistake
23:12 < bridge> <Solly (bun bun)> ```js
23:13 < bridge> <Solly (bun bun)> f=a=>{lines=a.split("\n").filter(a=>-1==="/*".indexOf(a.trim()[0])).map(a=>a.trim()).filter(a=>a).filter(a=>!a.toLowerCase().startsWith("function:")),out={param:[],return:[],remark:[],see:[]},head="desc";for(const b of lines){if(b.endsWith(":")){head=b.slice(0,-1).toLowerCase(),head=head.replace("parameters","param"),head.endsWith("s")&&(head=head.slice(0,-1));continue}out[head]=[...(out[head]??[]),b]}out.param=out.param.map(a=>-1===a.index
23:13 < bridge> <Solly (bun bun)> ```
23:13 < bridge> <Solly (bun bun)> average js script
23:13 < bridge> <Solly (bun bun)> another one but its non fatal so to spare chillers chatlog im not going to edit it again
23:14 < bridge> <chillerdragon> Cute solly Moment
23:15 < bridge> <Solly (bun bun)> im always cute
23:29 < bridge> <Solly (bun bun)> `out.desc.at(-1) += ".";`
23:29 < bridge> <Solly (bun bun)> this is invalid and it took me ages to figure it out
23:39 < bridge> <Ewan> 😳
23:39 < bridge> <Ewan> what the fuck is that
23:54 < bridge> <Assa> eww frontend