00:07 <+bridge_> <broshesss> discordapp.com/invite/J4QVHR7Z
12:00 <+bridge_> <aegisub> gpu manufacturers try not to break half of all games with every single driver update challenge
12:00 <+bridge_> <Assa> I am pretty sure this is just microslop
12:01 < ChillerDragon> i just gained a 2x fps boost by doing `git checkout a0828b6291c399aca15434df6ef39a017e637cbe`
12:01 < ChillerDragon> current master is at 80 fps which is visibly laggy and older versions are above 200fps
12:01 < ChillerDragon> almost 300 fps
12:01 <+bridge_> <Assa> what did you do, what is this commit?
12:02 < ChillerDragon> random older commit
12:02 < ChillerDragon> its not a bisect
12:02 < ChillerDragon> just saying the current master is slower than what i am used to
12:02 < ChillerDragon> rust with debug build is famously slow isnt it
12:02 < ChillerDragon> does that make the client build with debug useless now?
12:03 < ChillerDragon> i usually do gaming with debug builds so i get the traces
12:03 <+bridge_> <Assa> hmm, didn't I do a more up to date benchmark? https://github.com/AssassinTee/ddnet-benchmarks/blob/main/AverageFPS.png?raw=true
12:04 <+bridge_> <Assa> yes I did here
12:05 < ChillerDragon> on abyss i also notice no diff
12:06 <+bridge_> <Assa> I wanted to add springlobe 4 anyways
12:06 < ChillerDragon> but on 88.198.96.203:8306 i notice a big diff
12:07 < ChillerDragon> or on 213.171.25.11:8312
12:09 < ChillerDragon> i would blame the map BlmapChill because two servers are affected but on fokkos server i get smooth fps even with the new client
12:10 < ChillerDragon> maybe its the /design shenaigans that fix it idk
12:11 < ChillerDragon> local server with pure ddnet is smooth too i wonder if there is some weird edge case going on here
12:11 < ChillerDragon> but yea does not look like a gernalized regression 
12:18 < ChillerDragon> uhm sos i got a bug
12:18 < ChillerDragon> i have an std::vector
12:18 < ChillerDragon> print its size and it is 2
12:18 < ChillerDragon> i try to loop over it and it is an infinite loop but asan does not complain
12:18 < ChillerDragon> `log_error("antibot-worker", "num matches %zu", vMatches.size()); // prints 2`
12:18 <+bridge_> <chairn> show code
12:19 < ChillerDragon> `for(const auto &Match : vMatches)`
12:19 < ChillerDragon> `{`
12:19 <+bridge_> <chairn> show whole code...
12:19 < ChillerDragon> `log_info("was", "bruder");`
12:19 < ChillerDragon> `}`
12:19 < ChillerDragon> the rest is confidental xd
12:19 < ChillerDragon> shiet
12:19 < ChillerDragon> its full of nudes
12:20 < ChillerDragon> but i mean that alone should not happen no matter what i did to the vector before hand right?
12:20 < ChillerDragon> i called emplace_back() twice
12:20 <+bridge_> <chairn> maybe some UB happens, no idea
12:21 < ChillerDragon> but isnt that what ubsan should catch for me
12:27 < ChillerDragon> funny finding
12:28 < ChillerDragon> i tried to censor the code by renaming variables and types
12:28 < ChillerDragon> and that fixed it
12:28 < ChillerDragon> i tried 3 times switching it back and forth and confirmed the theory
12:28 < ChillerDragon> i have the same class name defined in another c++ file maybe that is the issue
12:28 < ChillerDragon> what a odd bug
12:31 < ChillerDragon> oke enuff debuggin that has to be it `git commit -m "Fix runtime bug by renaming a variable"`
13:06 < ChillerDragon> @heinrich5991 i slid into your irc dms
13:56 <+bridge_> <melon> #10381
13:56 <+bridge_> <DDNet> https://github.com/ddnet/ddnet/issues/10381
14:03 <+bridge_> <robyt3> Debug build might also be slower due to #11851
14:03 <+bridge_> <DDNet> https://github.com/ddnet/ddnet/pull/11851
14:54 <+bridge_> <Assa> just redid the FPS graph :owo:
14:54 <+bridge_> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1500118216385495191/AverageFPS.png?ex=69f74575&is=69f5f3f5&hm=c93edbba14ffb5f01687431eb021f90f7918d01165d8a97153dba00ddb786101&
14:55 <+bridge_> <Assa> springlobe really is that bad
15:30 <+bridge_> <bananouly> @essigautomat how do you make those graphs
15:30 <+bridge_> <Assa> <https://github.com/AssassinTee/ddnet-benchmarks>
15:30 <+bridge_> <byfox> Maybe should also calculate p99 and p95? For a complete understanding :)
15:31 <+bridge_> <Assa> I have them
15:32 <+bridge_> <Assa> well more or less
15:32 <+bridge_> <byfox> Can you show them to me and are they on github?..
15:32 <+bridge_> <Assa> I also have worst 0.001% fps, but it looks similar but even more noisy
15:32 <+bridge_> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1500127938329448498/Worst0.01.png?ex=69f74e83&is=69f5fd03&hm=1c7e1dae1329aebc86a0aa1579a5b1b0264c95cc7a74cd6ab5715d0607ced804&
15:33 <+bridge_> <Assa> they are in the jupyter notebook
15:36 <+bridge_> <Assa> I wonder if I read my graph wrong, if this is 0.01 or 0.01½
15:36 <+bridge_> <Assa> I wonder if I read my graph wrong, if this is 0.01 or 0.01%
15:38 <+bridge_> <Assa> it's 1% 🙂
15:38 <+bridge_> <Assa> I also have worst 0.001 (0.1%) fps, but it looks similar but even more noisy
15:38 <+bridge_> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1500127938329448498/Worst0.01.png?ex=69f74e83&is=69f5fd03&hm=1c7e1dae1329aebc86a0aa1579a5b1b0264c95cc7a74cd6ab5715d0607ced804&
15:41 < ChillerDragon> i think my windows compiler is trying to tell me i have too many config variables
15:42 < ChillerDragon> something about too deep nesting whatever and it gets fixed if i delete one line from config_variables.h
15:42 < ChillerDragon> i dont have a lot of configs and on linux it builds just fine
15:43 < ChillerDragon> did anyone run into that already i smh can not believe it i feel like i have some other bug 
15:44 <+bridge_> <byfox> clang-tidy?
15:45 <+bridge_> <byfox> In any case, I have not encountered this error with code 0XF on different systems :)
18:42 <+bridge_> <heckerisprolegit4250> I saw that the master server is protected. My friend says he doesn't want to download Discord. Maybe creating a website from a ticket would be easier for him. And adding an account with a nickname and password on the website would be easier. But you're forgetting that new users won't want to download Discord.
18:43 <+bridge_> <Assa> this is a temporary patch, we (heinrich) is working on a better system
18:45 <+bridge_> <heckerisprolegit4250> okay. he will wait for website's tell me it. i tell him
18:56 <+bridge_> <heckerisprolegit4250> s
18:56 <+bridge_> <heckerisprolegit4250> why i can't write in bugs xd
19:23 <+bridge_> <_pmakin> windows compiler might have allergy on you
19:48 <+bridge_> <marmare_314> @robyt3 Could the magic constant in CScrollRegion be needed because of an off-by-one error in CUi::ClipEnable (ie ClipEnable clips one pixel less than the rect)? At least adding one to the rect passed to ClipEnable in CScrollRegion::Begin avoids the need for the constant I think
21:45 <+bridge_> <robyt3> Isn't this basically the same as you are merely adding the 1 later? Might be due to pixel alignment because the UI screen uses a different float size that gets truncated to int when calling `Graphics()->ClipEnable`. Applying this rounding https://github.com/ddnet/ddnet-rs/blob/997113bac236ab6e3f8a5be368ffa2b7903a0777/lib/graphics-types/src/rendering.rs#L140-L149 seems to work at least in one case but it probably needs testing with different lists a
21:50 <+bridge_> <robyt3> ```cpp
21:50 <+bridge_> <robyt3> void CUi::UpdateClipping()
21:50 <+bridge_> <robyt3> {
21:50 <+bridge_> <robyt3>     if(IsClipped())
21:50 <+bridge_> <robyt3>     {
21:50 <+bridge_> <robyt3>         const CUIRect *pRect = ClipArea();
21:50 <+bridge_> <robyt3>         const float XScale = Graphics()->ScreenWidth() / Screen()->w;
21:50 <+bridge_> <robyt3>         const float YScale = Graphics()->ScreenHeight() / Screen()->h;
21:50 <+bridge_> <robyt3>         const float XScaled = pRect->x * XScale;
21:50 <+bridge_> <robyt3>         const float YScaled = pRect->y * YScale;
21:50 <+bridge_> <robyt3>         const float RoundedXScaled = std::round(XScaled);
21:50 <+bridge_> <robyt3>         const float RoundedYScaled = std::round(YScaled);
21:51 <+bridge_> <robyt3>         const float ClipW = std::round(pRect->w * XScale + XScaled - RoundedXScaled);
21:51 <+bridge_> <robyt3>         const float ClipH = std::round(pRect->h * YScale + YScaled - RoundedYScaled);
21:51 <+bridge_> <robyt3>         Graphics()->ClipEnable(RoundedXScaled, RoundedYScaled, ClipW, ClipH);
21:51 <+bridge_> <robyt3>     }
21:51 <+bridge_> <robyt3>     else
21:51 <+bridge_> <robyt3>     {
21:51 <+bridge_> <robyt3>         Graphics()->ClipDisable();
21:51 <+bridge_> <robyt3>     }
21:51 <+bridge_> <robyt3> }
21:51 <+bridge_> <robyt3> ```
22:07 <+bridge_> <user_94084> discord.gg/ye9aNaPN