00:23 < bridge> @heinrich5991 classic bash is the best for me, but heard alot about zsh, only thing is it looks slow af 00:23 < bridge> with all the shitty stuff it comes with 00:23 < bridge> (with also o-my-zsh) 00:23 < bridge> (with also oh-my-zsh) 00:33 < bridge> yeah I don’t love zsh 01:07 < bridge> fish is almost bash with good defaults 🙂 try it some time 01:10 < bridge> use the rust-lldb took 01:10 < bridge> tool 11:42 < bridge> average dev 11:42 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1157975798142218250/image.png?ex=651a9080&is=65193f00&hm=2740f3c22e131cd1af08aca2e82da33970e2b040130585d18c5a5dc4e94ae7b1& 17:45 < bridge> today out of nowhere my fps went again up to 600 fps 17:45 < bridge> :feelsgoodman: 17:57 < bridge> why isnt it possible ;(( 17:57 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1158070208599183410/WHYYYYYYY.mp4?ex=651ae86d&is=651996ed&hm=0919c648d8b940bae0a95af5e38b34b509829b96ccb4f092d9d2b6d228f23d94& 18:29 < bridge> Because tee collision physics are not deterministic as far as I tested it. Maybe it’s floating magic or some networking delay. But it seems that when tees collide random things happen. 18:44 < bridge> I'd guess some networking randomness 18:48 < bridge> it's the order in which the characters get updated 19:10 < bridge> oh true, these are bounces afterall 19:12 < bridge> tee collision are deterministic 19:13 < bridge> only the network is not deterministic in ddnet 20:25 < bridge> Oh yea that also crossed my mind. But would that mean that if a bounce is recreated with the same clientIDs it’s always gonna bounce the same? 20:29 < bridge> Im honestly curious but too lazy to test. @teero777 do you get the very same outcome if you redo the move with the same tees? Possibly without killing not sure what affects the physics order. 20:53 < bridge> Tw physics are 96% deterministic 20:54 < bridge> Only random thing is the tele 20:54 < bridge> not anymore 20:54 < bridge> Or networking thingies 20:55 < bridge> Did taters PR get merged? 20:55 < bridge> idk 20:55 < bridge> but ddnet uses PRNG 20:55 < bridge> what's the seed? 20:55 < bridge> server start time? 20:55 < bridge> idk xd 20:55 < bridge> i just heard this from hein 21:10 < bridge> I am 100% sure it doesn't have anything to do with client ids. Else my whole TAS thing wouldn't work 21:11 < bridge> Btw there shouldn't be an order at all right? 21:12 < bridge> That would be stupid 21:13 < bridge> Every new tick for an entity should be based on the previous tick and without having to rely on order 21:13 < bridge> Hmm 21:13 < bridge> Maybe I should fix that 21:14 < bridge> Random bits from platform dependent entropy source 21:14 < bridge> So it's pretty much random 21:14 < bridge> The seed is then recorded in teehistorian files so that things can be reproduced 21:15 < bridge> tw physics is very much order dependent, it's the cause of weak and strong 21:15 < bridge> Entity order that is, not client id order 21:15 < bridge> yea 21:15 < bridge> not cool 21:18 < bridge> should be like 21:18 < bridge> 1. make a copy of the world 21:18 < bridge> 2. tick all entities in the copy independently based on the original world 21:18 < bridge> 3. copy the copy to the original 21:19 < bridge> Maybe I'll implement that and check what diffrences will occur 23:49 < bridge> to make interactions between those two tees completely symmetrical in all cases I think you need both order independent physics (which is really hard to achieve for bounce) and fixed point math 23:51 < bridge> yea 23:51 < bridge> and those things would surely affect the trout population 23:51 < bridge> + existing map physics 23:52 < bridge> yes, doing all that without breaking any maps and tricks is an even harder problem