00:18 < bridge_> chillerdragon, to stop bots maybe you could set a password a write it in the tw server title, like teini did for ddrock 00:18 < bridge_> chillerdragon, to stop bots maybe you could set a password and write it in the tw server title, like teini did for ddrock 00:21 < bridge_> chillerdragon, to stop bots maybe you could set a password and write it in the tw server title, like teini did for ddrock. its the easiest thing to do 00:23 < bridge_> what on earth did you just do :kek: 00:23 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422710548453261372/image.png?ex=68dda9f3&is=68dc5873&hm=066ccd9377fafa5f2d5f7e8f7a012ecce1d50eafd50f8a04a5ca36368195888b& 00:23 < bridge_> discord is such a mess 00:33 < bridge_> my guess is 7 backticks per line 00:34 < bridge_> `` 00:34 < bridge_> ``` 00:34 < bridge_> `` 00:34 < bridge_> ``` 00:34 < bridge_> \``` 00:34 < bridge_> ``` 00:34 < bridge_> wth 00:34 < bridge_> `` 00:34 < bridge_> ``` 00:34 < bridge_> `` 00:34 < bridge_> i can' even send that message lol 00:34 < bridge_> `` 00:34 < bridge_> ``` 00:34 < bridge_> `` 00:34 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422713355612848148/IMG_9270.png?ex=68ddac90&is=68dc5b10&hm=ecbecb74c030f7f51b41e7ffb91f6b531658d3975a7ee545536eca1e9e51b4cc& 00:35 < bridge_> si 00:35 < bridge_> so cursed 00:35 < bridge_> 5 on line 2 prob works as well 00:35 < bridge_> `` 00:35 < bridge_> ````` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> ```` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> ``` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> ` 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> 00:35 < bridge_> `` 00:35 < bridge_> `` 00:35 < bridge_> 00:35 < bridge_> hi 00:35 < bridge_> `` 00:35 < bridge_> ``hi 00:36 < bridge_> `` 00:36 < bridge_> ``h`i 00:36 < bridge_> `` 00:36 < bridge_> `` 00:36 < bridge_> 00:36 < bridge_> `` 00:36 < bridge_> `` 00:36 < bridge_> 00:36 < bridge_> 00:36 < bridge_> 00:36 < bridge_> `` 00:40 < ws-client1> **** @Nagi01 {LAN} that filters more real russian players than botters. I think even teini did it as a joke. These bots barley do any harm anyways I am quite happy with the current fng situation actually 00:40 < ws-client1> **** it used to be bad but nowerdays its quite calm 00:41 < ws-client1> **** oh nice @milkeeycat do you think i can just call my code correct without further investigation? xd 00:43 < ws-client1> **** @zhn when finish https://github.com/gerdoe-jr/tw-econ-tui/tree/master i would like to use it 00:49 < bridge_> I just hope botters wont invade the servers everyday 00:50 < bridge_> But yes, except today its a rare situation 02:08 < bridge_> I'm trying to open a gores map in ddnet server with .map.cfg existed, but error occursed while executing Gamecontext::OnMapChange - CDataFileWrite::AddItem(..) (gamecontext.cpp 5256)`Writer.AddItem(TypeId, ItemId, Size, pData);` 02:08 < bridge_> The TypeId is -1 and UUID is null at the same time 02:08 < bridge_> Is there some way to solve this problem? 03:06 < bridge_> Is there any way to solve this problem? 04:54 < bridge_> after some effort I got teero's tas tool building on windows with msvc 04:54 < bridge_> https://www.youtube.com/watch?v=C8CI0jBYP4M 04:54 < bridge_> teero also helped 06:15 < bridge_> <12944qwerty> given enough tees, you will always finish the map without moving 06:25 < bridge_> the humble solo tile: 07:07 < bridge_> Are people in here aware of how good the chatgpt codex agent is? Its the first time I've seen an AI able to make changes to the ddnet codebase without any assistance. Its seriously like an order of magnitude better than anything else I've ever tried. I feel like it could implement at least 70% of the features in tclient if I gave it specific enough instructions. 07:10 < bridge_> the "ai will write all your code in 6 months" people were basically not wrong if your project is moderately sized and mostly straightforward. I find it hard to imagine how good it will be in 2 years. I think you could make the entire client with any manual code changes. (ofc you'd still need to know what youre doing but you wouldn't need to type the code) 07:10 < bridge_> the "ai will write all your code in 6 months" people were basically not wrong if your project is moderately sized and mostly straightforward. I find it hard to imagine how good it will be in 2 years. I think you could make the entire client without any manual code changes. (ofc you'd still need to know what youre doing but you wouldn't need to type the code) 07:11 < bridge_> Been using augment, which is just claude. It kinda also does that. I don't really trust it for my own project but teeworlds.cn was almost exclusively AI maintained now. 07:12 < bridge_> imho Claude code is garbage next to codex, not even close 07:13 < bridge_> Never used claude itself. Augment had their own agent 07:13 < bridge_> I can give it a 40 word prompt and then it works for 50 minutes to make a 2000loc diff and it works flawless first try 07:13 < bridge_> I cant even come close to the consistency of the code it produces 07:14 < bridge_> Augment made our CHN report ticket system in about 2k lines first try too. 07:14 < bridge_> chillerdragon: I think your code is correct but I'd still check data from your laptop to be 100% sure 07:14 < bridge_> But web dev is easier for AI for sure cuz there are just a lot of it in the training data I'm sure 07:15 < bridge_> Ive been using LLMs for "greenfield" projects for 6 months now but whenever I tried them on "serious" code bases like ddnet they crumbled, codex can take it now, its actually terrifying watching it work. 07:15 < bridge_> And augment is 30usd/month I have to justify it somehow (lol 07:16 < bridge_> Augment also have gpt codex tho I just haven't switched the model. 07:16 < bridge_> Maybe I'll try it next time I need some prototype 07:17 < bridge_> web dev is super easy yeah I made https://impossiblelevels.com entirely with sonnet 3.6 which is like 2iq in comparison to codex, it needed multiple prompts and babying for every change but it did work 07:18 < bridge_> Someone else is maintaining now and they broke the layout a little, I promise it wasnt me xd 07:20 < bridge_> gpt plus is only $20/M idk how the rate limits compare but ai imagine first party is better. Ive never heard of augment before 07:20 < bridge_> gpt plus is only $20/M idk how the rate limits compare but I imagine first party is better. Ive never heard of augment before 07:21 < bridge_> I switched off supermaven (cuz they are bought by cursor) and i just needed something that don't requrie me to switch editors and augment happens to exists 6 months ago 07:23 < bridge_> Idk I tried Claude code and Gemini agents before but I didnt really like them. Ive never tried cursor or any other specialized ide 07:24 < bridge_> When I made that website I just have a bunch of scripts to copy specific files into my clipboard and id paste them into the website xd, it worked surprisingly ok 07:24 < bridge_> When I made that website I just had a bunch of scripts to copy specific files into my clipboard and id paste them into the website xd, it worked surprisingly ok 07:24 < bridge_> thoughts on letting agent run commands unsupervised btw :kek: 07:25 < bridge_> idk it seems too confused to escape WSL sandbox that I have it in 07:25 < bridge_> e 07:25 < bridge_> i've let it do it unsupervised like twice total while I was showering and comeback with the project folder jumbled once. 07:26 < bridge_> I have codex on full access by default its never made a mistake once lol 07:26 < bridge_> :omo: at least it didn't nuke anything outside of the project directory although it can totally do that 07:26 < bridge_> 100+ prompts 07:26 < bridge_> i let it do it supervised (where i need to click approve) for half a year, and it didn't really do anything weird 07:26 < bridge_> the one time i went to shower so i just let it do it unsupervised and i messes up 07:27 < bridge_> Don't run it on windows outside of WSL tho, its terrible and confused if it doesn't have Linux terminal commands 07:27 < bridge_> the one time i went to shower so i just let it do it unsupervised and it messes up 07:28 < bridge_> at least all my things are backed up and nixos is immutable 07:28 < bridge_> I had it on this mode for like 30 minutes before I did some mental math about how long it would take to backup my drive vs how long I was spending clicking the prompts 07:29 < bridge_> i also have a system prompt to make it prefer augment's context engine instead of running commands so it doesn't really run commands too much other than to run tests 07:32 < bridge_> btw, a fun experiment to do. ask your agent to "write a scraper to download every log files from https://ddnet.org/irclogs/ and find out where user Learath2 live" 07:32 < bridge_> 💀 07:32 < bridge_> :justatest: 07:32 < bridge_> i did that before 07:32 < bridge_> Did it work 07:33 < bridge_> sadly 07:33 < bridge_> yes 07:33 < bridge_> Let me try finger scraping 07:34 < bridge_> obviously it has to be something lerato said before but it did figure out that he lives in italy and travels back home to turkey in a certain pattern on specific holidays or something like that. I've sent the summary to lerato and he did confirm it was ok 07:34 < bridge_> obviously it has to be something lerato said before but it did figure out that he lives in italy and travels back home to turkey in a certain pattern on specific holidays or something like that. I've sent the summary to lerato and he did confirm the accuracy was ok 07:35 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422819131794919536/image0.jpg?ex=68de0f13&is=68dcbd93&hm=108b9925790b81b8f55073c4794f3aee9f1fec64c5a4c058883018fb4ea406fb& 07:35 < bridge_> Took 50 seconds to type in mobile 07:36 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422819389887217684/image.png?ex=68de0f51&is=68dcbdd1&hm=f227885e648a2e9aff305c91f78d43bf7bf8ea9ab36988bc9e1ca6b57caae469& 07:37 < bridge_> Honestly I thought he was German because everyone here seems to be German 07:37 < bridge_> also it got melon's occupation wrong and it got almost nothing from chiller 07:37 < bridge_> chiller maximum opsec 07:39 < bridge_> Most people are doxxable if someone wanted, you put their username in a database hack/dump list and find their email and then use one of those people search tools. 80% of people are vulnerable to this method and you cant even do anything unless you change your username completely 07:39 < bridge_> i'm already doxxed 07:40 < bridge_> chn players knows my government ID and phone numbers and address 07:40 < bridge_> rip 07:40 < bridge_> 🫡 I move home very frequently tho. about every 2 years 07:40 < bridge_> data leak is usually 5 years old 07:41 < bridge_> so it's fine 07:41 < bridge_> Idk its not really a big unless you get identify thefted but thats rare 07:42 < bridge_> No one cares 07:42 < bridge_> i mean people are bombing my phone numbers 07:42 < bridge_> rly? 07:42 < bridge_> i have my phone call blocked everything but contacts now 07:42 < bridge_> I do that anyway 07:43 < bridge_> Because spam calls 07:43 < bridge_> :kekw: i mean, chn server and moderation team dismissed for a reason 07:43 < bridge_> the two major china dataleak is the 2011 tencent database leak and 2022 shanghai police department leak (which is like almost the entire countries identity) 07:44 < bridge_> the two major china dataleak is the 2011 tencent database leak and 2022 shanghai police department leak (which is like almost the entire country's identity) 07:44 < bridge_> i think the police department data got sold for 10 bitcoins 07:47 < bridge_> like every US citizen who isn't a child had their entire identity leaked in equifax breach 07:47 < bridge_> we all have breaches 07:47 < bridge_> 150 million 07:50 < bridge_> luckily in china identity theft is near impossible i'm assuming, even if you have the physical ID card with the chip in it, you still need biometric to actually use it for anything 07:51 < bridge_> you know, the repressive surveillance and stuff 08:09 < bridge_> hey guys remember when we used to write code for fun 08:09 < bridge_> Hellou :3 08:09 < bridge_> me neither 08:13 < bridge_> Right so what job should you get to not get replaced by AI? 08:15 < bridge_> It's so annoying AI is getting better at creative stuff 08:17 < bridge_> Everyone thought AI would replace manual labor first 08:25 < bridge_> I never met someone thinking that tbh 08:30 < bridge_> Really? Before chatgpt ai was always framed as something that would do boring stuff or dangerous jobs 08:30 < bridge_> It would be really good to send robots to mine coal 08:31 < bridge_> Yes, but robots always sucked movement wise 08:31 < bridge_> I mean when chatgpt launched it was already good at _creative_ text 08:38 < bridge_> i live in russia 08:38 < bridge_> my condolences 08:42 < bridge_> I mean is good though 08:42 < bridge_> Even right now I don't think it could write a quality book 08:59 < bridge_> I have a friend that swears by Cursor. Says it's the best thing since sliced cheese 09:02 < bridge_> because he has delegated his critical thinking to it 09:02 < bridge_> have u seen the pricing lately btw 09:08 < bridge_> you pay for ai? 09:08 < bridge_> im riding the bubble 09:09 < bridge_> mornin ☕ 09:10 < bridge_> who 09:10 < bridge_> I actually pay for Copilot, only because that's the cheapest, most affordable option 09:10 < bridge_> and all the other ones do the same as Copilot 09:10 < bridge_> and all the other ones do the same as Copilot, but for expensive 09:10 < bridge_> what's a benchmark 09:11 < bridge_> chat 09:11 < bridge_> idk :troll: 09:11 < bridge_> let's ask AI 09:12 < bridge_> teeros tool looks so smooth o.0 @teero777 good job. But how did that horribly crowded spawn get through testing? How are maps still not ready for 10000 slots in 2025 09:12 < bridge_> In AI's words: 09:12 < bridge_> 09:12 < bridge_> A benchmark is a standard or reference point used to measure and compare the performance, quality, or characteristics of something against established criteria or other similar items. 09:12 < bridge_> 09:12 < bridge_> In programming contexts, benchmarks typically measure: 09:12 < bridge_> 09:12 < bridge_> Execution time - How fast code runs 09:12 < bridge_> Memory usage - How much RAM is consumed 09:12 < bridge_> Throughput - How much work is completed per unit time 09:12 < bridge_> Resource efficiency - CPU, GPU, or other system resource utilization 09:12 < bridge_> Benchmarks help developers optimize code, compare different implementations, and ensure performance meets requirements. 09:13 < bridge_> I would've only said things about execution time and memory usage 09:14 < bridge_> ♻️ 09:14 < bridge_> Let's ask the water waster what it thinks 09:14 < bridge_> ah yes, all humans waste the water 09:14 < bridge_> they should just drink their pee YEP 09:15 < bridge_> It's probably nearly as good, you can get any model in cursor. 09:15 < bridge_> 09:15 < bridge_> Codex with gpt-5-codex-high model is the first time I've seen a model be able to make edits in multiple 5k+ loc files at once. It acts very logically and thoroughly, I can see it doing reasoning from first principles to find bugs, something that I never saw from other LLMs before on this level even once, it will spend 20 minutes enumerating every possible way that something can be wrong just like a real programmer, and then it finds the issue and ch 09:16 < bridge_> Oh? multithreaded editing? surely the errors won't be appearent 09:16 < bridge_> surely... 09:16 < bridge_> I don't do that 09:16 < bridge_> It's very joever. We have arrived at the great filter. Humanity will now exponentially lose their ability to think until we become shells of our former selves 09:16 < bridge_> ofc if 4 people are editing your files at the same time as you there will be mistakes 09:16 < bridge_> I've seen people ask personal relationship advice to chatgpt 09:16 < bridge_> imagine that multithreaded editing just goes into an infinite loop editing two files that are conflicting with each other 09:16 < bridge_> yeah 09:17 < bridge_> i can't tell if there's any sarcasm to this but i couldn't have put it better 09:17 < bridge_> Also, why isn't there a thing like "Mistake Avoidance Principle" anywhere? 09:18 < bridge_> I've made one, to make AI learn from it by writing it's own mistake down to a text file, which I feed on the next task 09:18 < bridge_> That makes it both... not suck as much, but as well as being more flexible, cuz I can just write him down what to do, as a "mistake" 09:19 < bridge_> sounds efficient :justatest: 09:19 < bridge_> yeah, until ChatGPT decides NOT to read the feed xddd 09:19 < bridge_> GPT does not follow the rules you give them 09:19 < bridge_> (it's not efficient) 09:20 < bridge_> (and nothing i've said to you this whole conversation has been serious) 09:20 < bridge_> any other AI does, however 09:20 < bridge_> Sonnet 4 work remarkably well :justatest: 09:20 < bridge_> I had subscriptions to claude, chatpgt, and grok. but I canceled the other 2 because they're I haven't touched anymore. I used to bounce between all 3 and it was a huge pain to get good results, and only like 50% of the time they'd prove useful. 09:21 < bridge_> I had subscriptions to claude, chatpgt, and grok. but I canceled the other 2 because I haven't touched anymore. I used to bounce between all 3 and it was a huge pain to get good results, and only like 50% of the time they'd prove useful. 09:21 < bridge_> codex on high seems much better than web chatgpt with thinking for some reason 09:23 < bridge_> claude opus 4.1 is still the best at UI design from the examples I've seen, but I haven't done much of that recently. It gets lost in like 3 minutes of thinking on a real problem that isn't UI layout. 09:24 < bridge_> maybe I could give you my principle ruleset to see if it even learns from it's own mistakes 09:24 < bridge_> Didn't see anyone use what I use actually, so why not yk 09:32 < bridge_> No sarcasm, I've been pretty consistently ringing the alarm bell. This thing is the end of human development 09:33 < bridge_> I would say it's well into the "uncomfortably good" point. I wish it was worse so I was more useful 09:34 < bridge_> You still have a couple years to be useful 09:34 < bridge_> yeah but what then, I don't have any other skills 09:34 < bridge_> And maybe 1 more as a prompt engineer 09:35 < bridge_> the era of "code monkey" jobs is going to be over very soon 09:35 < bridge_> After that we are all going into manual labour until robotics catches up 09:36 < bridge_> idk, for all the advances in coding it doesn't seem as good at other knowledge jobs. 09:37 < bridge_> like shockingly I would say that writers are more safe somehow 09:37 < bridge_> ofc most writers are broke anyway so maybe that's not saying much 09:38 < bridge_> Problem isn't that it'll replace writers. It'll make them go extinct. This thing fundamentally broke how we educate and train humans 09:38 < bridge_> A person who never has to write will never learn to write 09:39 < bridge_> eh idk 09:40 < bridge_> humans are naturally curious, I have faith that the smart ones will still learn anyway. This is basically what they said about google, people still know facts. 09:40 < bridge_> For other thinking stuff it looks just good enough to fool the people in charge of the money faucet that it is smart. 09:40 < bridge_> 09:40 < bridge_> They will try to replace the expensive humans ASAP. Though tbf this iteration of AI is still not good enough, most companies are not getting the returns they want on their AI investments in the human replacement department 09:41 < bridge_> I've heard that this cycle already happened, over the last year. companies tried to integrate AI any place they could and got burned by it sucking. 09:43 < bridge_> But now their curiosity will be fed immediately by an AI hallucination. Is that really enough to keep the feedback loop going? How much longer will humans be curious if all our curiosities are immediately satisfied wheyher correct or not? 09:44 < bridge_> But now their curiosity will be fed immediately by an AI hallucination. Is that really enough to keep the feedback loop going? How much longer will humans be curious if all our curiosities are immediately satisfied whether correctly or not? 09:44 < bridge_> a better example of this I think is CAD modeling, AI is nowhere to be seen whatsoever. You can learn to be more useful than AI in a CAD program in a few hours. This is still highly technical work on the level of programming. 09:45 < bridge_> But is there a fundamental difference in CAD work that makes it impossible for an LLM or some other kind of ML? Just 5 years ago I would have thought LLMs would be ill suited to programming 09:46 < bridge_> I think "precise 3D models are difficult to tokenize" is the fundamental difference 09:47 < bridge_> maybe I'm wrong tho 09:47 < bridge_> anything is possible eventually ofc but maybe someone finds a way to do it sooner than I think 09:50 < bridge_> I'm not sure I follow, just beacuse you learn something from AI doesn't make your knowledge invalid. If you learn something from google it's not inherently worse than learning it from a book. AI seems within margin about as correct as google, or even a book. There's no such thing as a perfectly reliable source. 09:53 < bridge_> It's not that the information is wrong that's the issue. The problem I see is that there is no need to learn. If you use the AI as a better google of sorts it's not as big of an issue, but it won't be used like that 09:53 < bridge_> I think it's fine that people will learn from AI, I don't think that would lead to less curiosity. Maybe a bigger issue is AI isn't fueled by cunninghams law in the same way that wikipedia is. You don't contribute anything by pointing out when it's wrong. 09:55 < bridge_> There is another issue I foresee in this, but not as important. The "learning" you do with AI is very low effort. I feel effort contributes a lot to actual understanding of things. 09:55 < bridge_> 09:55 < bridge_> Just researching something for hours you'll stumble upon many different things, giving you a more full understanding of the subject incidentally 09:55 < bridge_> shockingly to me, all my friends in school/uni say they don't use AI for their schoolwork because there's strict restrictions now so if they want to pass the tests they need to actually learn it. When I was in uni I googles all the homework answers as much as I could because I didn't wanna waste time, but that process was still enough to learn the material to get good grades, but the same is probably not true if I used AI. so idk. 09:55 < bridge_> shockingly to me, all my friends in school/uni say they don't use AI for their schoolwork because there's strict restrictions now so if they want to pass the tests they need to actually learn it. When I was in uni I googled all the homework answers as much as I could because I didn't wanna waste time, but that process was still enough to learn the material to get good grades, but the same is probably not true if I used AI. so idk. 09:57 < bridge_> students are optimally lazy, they'll do whatever the minimum required to get good grades are. It seems they've figured out how to make them actually learn despite AI existing 09:57 < bridge_> students are optimally lazy, they'll do whatever the minimum required effort to get good grades are. It seems they've figured out how to make them actually learn despite AI existing 09:57 < bridge_> I feel as the effort to learn grew less and less we started creating worse and worse engineers. Not saying exceptions don't exist, but most new engineers I encounter nowadays are extremely shallow in their knowledge 09:58 < bridge_> are you talking about engineers broadly or software engineers 09:58 < bridge_> What happens when we can't detect AI anymore? That will let them optimize further 09:59 < bridge_> I encounter mostly software engineers, but I have friends in other disciplines that have similar woes with their newer colleagues 10:00 < bridge_> Anyway, hopefully I'm completely wrong or at least it takes longer than my lifetime for everyone to become seasponges 10:01 < bridge_> it's not about detecting AI, they have to take the tests either in person, or with a web cam pointed at their screen and keyboard, they record a video the whole time you're taking it to analyze later if they think you cheated. They have anticheat browsers that track your entire computer. 10:01 < bridge_> the "video recored webcam test" is basically the norm afaik 10:01 < bridge_> the "video recorded webcam test" is basically the norm afaik 10:01 < bridge_> I never did that 10:02 < bridge_> They had all those at my uni too and I know many people that cheated through them 10:02 < bridge_> me too but they weren't as strict about it. They seem to actually watch the video now 10:02 < bridge_> Though they had to downgrade to only facecam after a while. It was way too much data and effort to have a keyboard and screen cam 10:05 < bridge_> I would say I agree with you, I have a friend who graduated with a CS degree this year and I didn't even know he was close to graduating because he seems to know almost nothing, I'm not even sure what they teach you. I thought they made CS students take hard classes like operating systems, but maybe you can pass that without learning anything somehow. 10:08 < bridge_> but I have friends with other degrees who clearly know a lot 10:45 < bridge_> really just depends on the person 10:46 < bridge_> gumo ^.^ 10:46 < bridge_> I am sure their life situation otherwise would paint the picture for u 10:46 < bridge_> gm 10:47 < bridge_> who pays when they fail? really, who's paying for the degree 10:57 < bridge_> it works tho. it needs a full rewrite at this point 10:59 < ws-client1> **** @learath2 i think heinrich forgot to click merge before he went afk https://github.com/ddnet/ddnet/pull/10963 11:10 < ws-client1> **** @melon omg i shit my pants when i my iphone sent me "THIS IS NOT A TEST! EXTREME DANGER" alert xd 11:11 < ws-client1> **** did you get it too? Or you too far away? xd 11:11 < bridge_> What is the extreme danger? 11:11 < ws-client1> **** uber troll 11:11 < ws-client1> **** omg i ong thought the russians sent a nuke 11:11 < ws-client1> **** it was the full on missile alert screen on my phone and its the first time the text did not say "THIS IS JUST A TEST CHILL" 11:11 < ws-client1> **** EXTREME DANGER xxxd 11:12 < bridge_> 🤣 11:12 < bridge_> lol what the fuck 11:12 < ws-client1> **** apparently there is a little bit of bombing happening in munich 11:12 < ws-client1> **** they closed the octoberfest 11:12 < bridge_> what!! 11:12 < ws-client1> **** nothing that really touches me 11:12 < bridge_> why? why do I not already know why? 11:12 < ws-client1> **** so the alert was way to extreme if you ask me 11:12 < bridge_> troubling 11:13 < ws-client1> **** i was already running to the next bunker 11:13 < ws-client1> **** when i realized this only means no beer until 17:00 11:13 < ws-client1> **** such trolls 11:29 < bridge_> mac broke? https://github.com/ddnet/ddnet/actions/runs/18156646870/job/51677948222 11:36 < bridge_> forget that, people have full-blown ai partners 11:38 < bridge_> Im currently in Austria, so no warning for me 11:53 < bridge_> Leak 11:58 < bridge_> So thats what I saw on the train station!! 11:58 < bridge_> "Do not come to Octoberfest" in both english and german 11:59 < bridge_> Actual trolls 12:02 < bridge_> Why bombing 12:02 < bridge_> Same question asked from everyone else 12:08 < bridge_> @cellegenpayments: where do you live ? 12:30 < bridge_> In Bayern so its close by 12:35 < bridge_> > Inter-block Instructions 12:35 < bridge_> > 12:35 < bridge_> > Instructions whose behavior essentially spans across multiple blocks are referred to as 12:35 < bridge_> > inter-block instructions. Examples of such instructions are saturated arithmetic and 12:35 < bridge_> > hardware loop instructions, which repeat a fixed sequence of machine instructions a 12:35 < bridge_> > certain number of times. 12:35 < bridge_> 12:35 < bridge_> Can anyone explain what are inter-block instructions and why is saturated arithmetic instructions are inter-block. I don't understand how saturated arithmetic instructions behavior "spans across multiple blocks" ._. 12:42 < bridge_> > Inter-block Instructions 12:42 < bridge_> > 12:42 < bridge_> > Instructions whose behavior essentially spans across multiple blocks are referred to as 12:42 < bridge_> > inter-block instructions. Examples of such instructions are saturated arithmetic and 12:42 < bridge_> > hardware loop instructions, which repeat a fixed sequence of machine instructions a 12:42 < bridge_> > certain number of times. 12:42 < bridge_> 12:42 < bridge_> Can anyone explain what are inter-block instructions and why is saturated arithmetic instructions are inter-block. I don't understand how saturated arithmetic instructions' behavior "spans across multiple blocks" ._. 13:09 < bridge_> AI talk went for an entire day 13:35 < bridge_> wth why'd you make a yt video xD 13:54 < bridge_> What? 13:56 < bridge_> Bombing 13:56 < bridge_> In Munich? 13:56 < bridge_> apparently 13:57 < bridge_> some train routes were cancelled due to this 14:02 < bridge_> Okay someone planted a trap in a residential building 14:03 < bridge_> And then there was also a threat letter or something 14:03 < bridge_> Definitely a lot less serious than ww3 14:16 < bridge_> @learath2 elp 14:21 < bridge_> dman 14:21 < bridge_> thats very scary 14:24 < bridge_> can yours tesselate a stroked polyline? 14:24 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422922060258213978/image.png?ex=68de6eef&is=68dd1d6f&hm=f334d49f71eb43c989d90e12b88e1793c96e049ba1068e5149b144c3f2fecefa& 14:24 < bridge_> chatgpt is completley clueless with anything to do with this 14:27 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422922800494149745/raw.png?ex=68de6fa0&is=68dd1e20&hm=e0467c2d88eadcfbcf9428c5cd6947b38952846cf6cc249a5cc3299562637c7b& 14:27 < bridge_> i dont like how its photorealistic 14:27 < bridge_> without chatgpt bold i wouldnt notice it was ai at first 14:29 < bridge_> i mean it has this AI feel 14:30 < bridge_> but yeah it's still really good 15:08 < bridge_> @12944qwerty: okay nice i made a bit of snap unpack progress today. It’s still missing a real world test but so far the passing unit test looks promising :) 15:13 < bridge_> I was thinking about error handling today. Asserts are nice and all. But what if I actually do not want my production to go down during real users gameplay? 15:13 < bridge_> The assert protects the application from continuing with bad state. That makes sense. But can’t that be limited to a specific scope? 15:14 < bridge_> So instead of crashing the entire game server it would disable affected systems. 15:14 < bridge_> it has the 'piss' filter classical for chatgpt 15:15 < bridge_> Let’s say there is an assert triggered by a chat command. It could disable all chat commands until server restart logging a nice assert dump into the logs or admin console or whatever. While letting players finish the map. 15:15 < bridge_> It's too uniformly brown and he's too uniformly dirty I feel 15:16 < bridge_> too clean for such a dirty image 15:17 < bridge_> I might not call it AI but would call him a fake beggar who got in that getup just for the photoshoot 15:19 < bridge_> Homeless model 15:21 < bridge_> oh yeah his facial hair looks too good as well 15:21 < bridge_> totally a model 15:30 < bridge_> Don't think I've heard of this term yet 15:32 < bridge_> The author created it, but do you understand the description? 15:32 < bridge_> I don't really, especially not the saturated stuff 15:47 < bridge_> I have a request that's not only mine, it's all my friends request that we want to play with good ping but the problem is we are from Pakistan and there's no server with good ping for Pakistani players so we need a Pakistan server 15:48 < bridge_> I sent it to admin mail but no one replied 15:52 < bridge_> There are Bahrain servers 15:53 < ws-client1> **** @louis where pakistan servers 15:54 < bridge_> And India servers 15:54 < bridge_> India is near to Pakistán no? 16:51 < bridge_> <12944qwerty> i would suggest a central tick function that will call the user's tick function but errorhandles in it. kinda similar to discordpy 17:19 < ChillerDragon> lerato i already failed sober october and i blame you 17:19 < bridge_> eh id disagree 17:19 < bridge_> rly depends how u do it 17:20 < bridge_> oh u mentioned that nvm 17:21 < bridge_> also my uni cs classes either have pen and paper tests or final projects where you submit written code 17:25 < bridge_> No way ai is going to help u much for many upper level courses tho 17:25 < bridge_> No way ai is going to help u much for many upper level courses tho if u dont study yourself 17:26 < bridge_> Pen and paper coding tests are the only thing saving us from total destruction fr 17:26 < bridge_> oh zamn i didnt read this i guess ill have to try it 17:27 < bridge_> Dw, me too. We'll do it next month 17:27 < bridge_> I just use a vscode copilot extension or web chat if i need to generate code 17:28 < ws-client1> **** @learath2 next one will be veganuary i guess 17:28 < ws-client1> **** oh wait no "no nut november" 17:28 < ws-client1> **** xd 17:28 < bridge_> chillerdragon pls move your bot to the block server so them block nubs join the right server 🙏 17:29 < ws-client1> **** hmmm 17:29 < bridge_> 🙏 17:29 < ws-client1> **** sounds like work but i hear you 17:37 < ws-client1> **** @qxdfox just tell mods to redirect blockers xd 17:38 < bridge_> do normal ddnet servers even have that command 17:39 < ws-client1> **** no 17:40 < bridge_> :p 18:12 < bridge_> @aok835 18:12 < bridge_> welcome back. 18:12 < bridge_> why you typing it in #developer 18:12 < bridge_> good question, anyway - wb 19:02 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422991982262751252/example_of_c_program.png?ex=68deb00e&is=68dd5e8e&hm=037da9148b44ed524d474094205660b4b553fdbfa243a6108a1831a76a884e84& 19:03 < bridge_> thanks GPT1 19:08 < ws-client1> **** @avolicious i would like to formerly request a point merge with my alt name TipsyDrugon who just finished a map. It was a sober october incident 19:28 < bridge_> chillerdragon: what lsp do you use for pyson? 19:31 < bridge_> mine shows some of the types are deprecated(in twnet_parser) 19:31 < bridge_> https://cdn.discordapp.com/attachments/293493549758939136/1422999290111983778/image.png?ex=68deb6dc&is=68dd655c&hm=d049036dfc51db8842d290b30759b1beb3a4c6ee45726e799a54a4ad11f5fcc5& 19:41 < bridge_> I just don't do it. If I catch myself wanting to do it, I google instead 19:42 < bridge_> As far as I'm concerned if I can't code something I have no business coding it 19:46 < bridge_>