DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.org/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2024-12-20 00:00 and 2024-12-21 00:00

Jupstar ✪

i wonder if someone simply reported one of your messages

Discord had some issue with germans

02:33

A ton of germans got flagged and accounts got disabled, including mine

02:34

https://www.reddit.com/r/discordapp/s/hkNuyeLil1

From the discordapp community on Reddit

Explore this post and more from the discordapp community

02:34

kek

how to force vote in f2? i dont undestend syntax

Common discord L

[ddnet-maps:master] 1 new commit

278e367 A Memento Vivere - ddnet-maps

triki

how to force vote in f2? i dont undestend syntax

vote yes or vote no

1

[ddnet/ddnet] Pull request opened: #9407 Add input type flag

was trying something for spectator cursor, then I already had this so i'll just put it here. someone can takeover this if it haven't been worked on yet and i didn't do it entirely wrong. i've split the type into three: relative, absolute and automated (where the mousepos is forcefully set in anyway). the playerflags however only has an absolute flag where i treated automated input as absolute. i didn't spent too much time checking on AB, but i couldn't tell if it syncs playerflags, but ...

GitHub

Click to see attachment 🖼️

@DeFreddie are we still sure we'd want to use a player flag for something that might as well be a netmsg ? Or is there a reason as to why this won't work

meloƞ

@DeFreddie are we still sure we'd want to use a player flag for something that might as well be a netmsg ? Or is there a reason as to why this won't work

ask heinrich actually

06:55

i don't know either

06:55

justatest

Didn't hear from Heinrich in a while NotLikeKogasa

NotLikeKogasa

meloƞ

Didn't hear from Heinrich in a while NotLikeKogasa

NotLikeKogasa

https://github.com/ddnet/ddnet/pull/9357#issuecomment-2546832457

oh! I see

meloƞ

@DeFreddie are we still sure we'd want to use a player flag for something that might as well be a netmsg ? Or is there a reason as to why this won't work

the compatibility thing i get, but i think adding more playerflags might be as easy as unclamp the protocol, it was not a byte, it is just varint but clamped as far as i can understand.

06:56

haven't tried it tho

06:57

i mean i can try it right now one sec

07:01

seems fine to me

07:01

Image attachment

07:02

a version check to make sure you don't send out of bound flags to server with lower version and we are good to go

Swarfey

A ton of germans got flagged and accounts got disabled, including mine

my account got deactivated for "suspicous activity" and I got a mail that "I got hacked" from discord. 2FA for the win

08:07

also I know this has to be bullshit

maybe discord got hacked and they won't acknowledge it

no company does, but they have to by law. But first they need to collect info about what was hacked and by what degree

08:39

this can range from the "secrataries ipad" to "all of our infrastructure can be thrown away"

<ChillerDragon> https://github.com/ddnet/ddnet/blob/b52f1a41a5890fca6da3a877221f7dce69eec60c/src/engine/shared/network.cpp#L361

ddnet/src/engine/shared/network.cpp at b52f1a41a5890fca6da3a877221f...

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

08:54

<ChillerDragon> does someone know why we shift by 2 and not 4 here?

08:54

<ChillerDragon> according to the libtw2 docs a vital chunk header consists for 3 bytes and the sequence number takes the full third byte and half (4 bits) of the second byte

08:54

<ChillerDragon> https://github.com/heinrich5991/libtw2/blob/master/doc/packet.md

libtw2/doc/packet.md at master · heinrich5991/libtw2

Some Teeworlds stuff in Rust.™. Contribute to heinrich5991/libtw2 development by creating an account on GitHub.

08:58

<ChillerDragon> the mask is a bit hard to read in ddnet pData[1] |= (m_Sequence >> 2) & (~((1 << Split) - 1));

08:58

<ChillerDragon> in teeworlds 0.6 its just pData[1] |= (m_Sequence>>2)&0xf0; and 0xf0 in binary is 0b11110000 so you can clearly see the 4 bit mask

08:59

<ChillerDragon> so shouldn't it also be a 4 bit shift?

09:00

<ChillerDragon> imo sequence numbers up to 255 should only be written to the last byte and only 256 and higher should start leaking into the 2nd byte but from what i can tell the sequence number 64 already starts writing into the second byte

<ChillerDragon> okay looking at some traffic my ddnet client sent during a 0.6 connection it seems like it actually does not pollute the 2nd byte before it reaches 256 but how

09:14

<ChillerDragon> https://zillyhuhn.com/cs/.1734685973.png

Embedded image

09:17

<ChillerDragon> @heinrich5991 also in the dissector it says 00.. .... 1111 1111 for the sequence number. Shouldn't it be 0000 .... 1111 1111 ? First 4 then 8?

09:17

<ChillerDragon> https://zillyhuhn.com/cs/.1734686251.png

Embedded image

<ChillerDragon> oh no that makes no sense the size is not decimal 49 because the second byte is 0x31 the size also leaked into the first byte because it is bigger than 4 bits (15)

09:26

<ChillerDragon> the only flag bit that is set is vital so the first byte of the header should be 0x40 but it is 0x43 (0b01000011) so the last 2 bits are size okay now i understand nothing anymore

09:30

<ChillerDragon> that shit is more encryption than compression ngl

[ddnet/ddnet] Pull request opened: #9408 Fix console command arguments

Fixes #9278 <- This pr fixed this issue.

Checklist

[x] Tested the change ingame
[x] Provided screenshots if it is a visual change
[x] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[x] Considered possible null pointers and out of bounds array indexing
[x] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](...

[ddnet/ddnet] Pull request closed: #9408 Fix console command arguments

ws-client

<ChillerDragon> that shit is more encryption than compression ngl

the line of this can be extremely blurry, I know a nintendo puzzle where they encrypted some data with bit operations and your task was to write a decoder

ws-client

<ChillerDragon> that shit is more encryption than compression ngl

More encoding than encryption, after all I'm not seeing very many large primes or elliptic curves

quantum computer doesn't care about both troll

troll

[ddnet/ddnet] Pull request opened: #9409 Add assertions to CServer::GetClientAddr functions, avoid string-based address comparison for sv_vote_kick_min

Prevent use of uninitialized memory if the functions are used for a player that is not yet ingame. The check for the ingame state is also unnecessarily restrictive, as the client address should be available immediately. Compare client addresses directly to determine the distinct client count excluding spectators. This is consistent with the IServer::DistinctClientCount function though this does not check for spectators.

Checklist

[X] Tested the change ingame
[ ] Provided sc...

github-merge-queue[bot]

[ddnet:master] 3 new commits

85e246e Add assertions to CServer::GetClientAddr functions - Robyt3 b8af61d Avoid string-based address comparison for sv_vote_kick_min - Robyt3 168391f Merge pull request #9409 from Robyt3/Server-GetClientAddr-Fixes - def-

12:43

[ddnet/ddnet] Pull request closed: #9409 Add assertions to CServer::GetClientAddr functions, avoid string-based address comparison for sv_vote_kick_min

what do you need this for?

13:18

you can always printf whatever you want, but the fact that you don't know this and directly want it for a projectile is a bit sus to me

Jupstar ✪

you can always printf whatever you want, but the fact that you don't know this and directly want it for a projectile is a bit sus to me

oh i just found this file, my original goal was just to study the laser code, i thought maybe something cool could be made out of it, but before that i tried to change the HUD, it's pretty cool to do that

13:22

There were no problems with the HUD, but with the laser I had my first problems understanding the code

laser.cpp is really mostly the laser physics

13:22

if you want to change the rendering, then the file is called items.cpp or smth like that

Jupstar ✪

if you want to change the rendering, then the file is called items.cpp or smth like that

Why change the rendering? Will something change?

I mean what part of the laser do you want to change?

what do you want in first place?

Generally speaking you probably won't have an easy time to understand the laser in full detail. The DoBounce method is the most interesting one

Jupstar ✪

I mean what part of the laser do you want to change?

just output information about the laser to understand the code by type of direction, energy, bounces, etc.

then simlpy print what you need in the tick function

Jupstar ✪

Generally speaking you probably won't have an easy time to understand the laser in full detail. The DoBounce method is the most interesting one

I tried to understand through chat gpt, and I was very surprised with the laser's work)

but if you already know about direction, energy and bounces. the variables are most likely also called similary to this

exosphere

I tried to understand through chat gpt, and I was very surprised with the laser's work)

yeah lasers are bit hacky, because they need to stay alive while the client renders them, so they don't directly despawn when they are gone/finished. and if you look in ddrace code, you see all the 2 trillion extra cases for laser doors, tele lasers etc

13:27

which makes it even harder to understand the whole code

13:27

maybe first try: https://github.com/teeworlds/teeworlds/blob/c56fa9e6a20cfc9d7d16502e18c7d7633acdf492/src/game/server/entities/laser.cpp

teeworlds/src/game/server/entities/laser.cpp at c56fa9e6a20cfc9d7d1...

A retro multiplayer shooter. Contribute to teeworlds/teeworlds development by creating an account on GitHub.

13:29

doors are extra code btw, so you gladly don't have to deal with that

[ddnet:18.8.2] 25 new commits

5120e79 fix spectate closest - TsFreddie 51ff510 Update Simplified & Traditional Chinese translations for upcoming 18.8 - ASKLL-STAR ada887c Fix hook collisions preview string not being localizable - Robyt3 7294013 Revert "remove self from spectate selector" - TsFreddie 0f1b253 fully allow clicking on yourself to spectate - TsFreddie 7e79e4e Update Simplified Chinese translations for upcoming 18.8 - ASKLL-STAR 72f3f6a Update Traditional Chinese translations for upcoming 18.8 - ASKLL-STAR ab0aaa1 Unite the word used in traditional_chinese.txt - By622 626c324 Update Turkish translation for 18.8 - GokturkTalha f72297f Update Azerbaijani translation for 18.8 - GokturkTalha 0381d82 don't allow SpectateClosest to select self - TsFreddie 018c64a 100% RO translation (1/2) - Sans3108 aa3087d 100% RO translation (2/2) - Sans3108 cf6309e Update brazilian_portuguese.txt - rffontenelle cc57c1c French translation by Yubel - def- 41a4d5a Update Swedish translations for 18.8 - furo321 8061d82 pr#9329 supplement - ASKLL-STAR a718956 Update Ukrainian l10n (18.8) - EGYT5453 0177660 fix style - def- d0bdc64 Update ddnet-libs to include Android libs - def- 656d179 Fix clipping of last console line, save background noise as RGBA - Robyt3 6d61f78 russian.txt translations - RRRmnik 1e06f0d Update russian.txt - RRRmnik de4bd18 Update spanish.txt for 18.8 - n0Ketchp 56340af Version 18.8.2 - def-

huhu

hai

ho ho ho

zanta

zaza

[ddnet/ddnet] Issue opened: #9410 Android mouse invertion

What happens: if you open ddnet on android with mouse connected, mouse will invert x and y How to reproduce: Open DDNet on Android. Connect mouse. Try to move mouse Video: (directions: left, up, right, down) https://github.com/user-attachments/assets/9dfdf94a-b35a-4276-93b7-33bdb9e3bebf Possible solution: add option to settings to invert x and y on mouse

[ddnet/ddnet] Issue opened: #9411 cant play

my game goes crazy and it wont let me join any of my friends https://github.com/user-attachments/assets/d6f5da6f-bf36-4a90-81ef-6676da548129 please fix it thank you

[ddnet/ddnet] Pull request reopened: #9408 Fix console command arguments

hi

@DeFreddie can you give me the static clang format 10 link you mentioned heinrich sent? i can't find it and i want to add it to the readme, new contributors often forget

I though he was Chinese

and?

kek

16:46

he a silly billy

Chinese from Europe

16:47

Maybe we shouldn’t combat blockers on non-blocker servers, maybe we should create more blocker maps? I’m tired of banning those idiots

[ddnet-web:master] 1 new commit

4697b84 18.8.2 - def-

[ddnet/ddnet] Pull request opened: #9412 extend CONTRIBUTING.md and README.md

just a small addition because it's easy to miss CONTRIBUTING.md's existence.

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UB...

17:02

[ddnet/ddnet] Pull request closed: #9408 Fix console command arguments

GitHub

Click to see attachment 🖼️

@ASKLL this could be helpful for you :P our fix_style script is a bully

@mel1x I've tried. It doesn't give me any warn.

Jupstar ✪

I think this is simply your ingame name?

Now it's clear, This message is visible only to you (yourself) I'll change my nickname okay? plz dont ban me...

[ddnet/ddnet] Issue closed: #8724 Discussion: Add serverbrowser gametype icon like 0.7?

[ddnet/ddnet] Pull request opened: #9413 Always shows team size indicator

When you play a new map, sometimes the team message is not received and it is hard to tell whether teaming is required, or if there is a maximum team size. I usually join a team for team size indicator, but it is not shown for 1-tee team. In my opinion, it's not looking worse adding team size indicator to them, while providing useful infomation. Before: !图片 After: ![图片](https://github.com/user-attachment...

21:25

github-merge-queue[bot]

[ddnet:master] 2 new commits

1326deb extend CONTRIBUTING.md and README.md - BlaiZephyr 2012937 Merge pull request #9412 from BlaiZephyr/extend_readme_contributing - heinrich5991

21:25

[ddnet/ddnet] Pull request closed: #9412 extend CONTRIBUTING.md and README.md

New openai model has 2700 code forces rating... it's so over

Image attachment

22:18

I don't see how anyone will be writing code by hand in 8 years

how big are the code tasks?

2700 is top 0.01%

22:20

Rank 150

Tater

I don't see how anyone will be writing code by hand in 8 years

Hey, more bugs to fix for us who do write by hand :D pepeW

pepeW

but is that like codewars?

It doesn't matter it crushes every benchmark

yeah but software engineering isn't about small algorithms

from what I've read about AI development it writes a ton of code that LGTM

but actually has a bunch of bugs

SWE bench the one to the left is about big problems

risu

from what I've read about AI development it writes a ton of code that LGTM

but actually has a bunch of bugs

It went from 1900 rating to 2700 rating in 3 months bro

22:22

Its over

before my AI can't code teeworlds on it's own, it's not AI xd

How does this rating work? Is it the number of problems solved correctly or does it take into account the performance of the solutions?

Jupstar ✪

before my AI can't code teeworlds on it's own, it's not AI xd

I promise you're under estimating how much it can fit in the context window of new models

22:24

This isn't the same thing as it was 6 months ago

Can you get a high Elo by solving just a few problems or do you need to solve a lot of them? I'm willing to bet that the 99.9% of users lower than that rating haven't actually solved most of the problems on the platform simply because they don't want to put in the hours

I could dedicate my life and will not achieve 2700 rating

22:25

IOI gold medal is 2400 rating

Tater

It went from 1900 rating to 2700 rating in 3 months bro

When a metric becomes a target it ceases to be a good metric

Tater

I promise you're under estimating how much it can fit in the context window of new models

i am open to be replaced by an AI. But I am not really a to good to be true guy

22:25

coding often means re-iterating over existing code

22:25

and the current GPT models are not designed for such tasks

You guys are missing the point, it's no longer about if it can do everything a software dev can it's about when. The science is over now it's just money and engineering. (edited)

I am currently not sure if you are serious xD

I am totally serious 2700 cf is absurd

Also a robot being able to solve well established problems is not much more impressive than a robot rephrasing established literature. These things will still choke on novel problems because they lack proper ability to reason.

chatgpt would probably easily beat me in A-Levels exams.. I'd still say it's stupid af xD

Tater

I am totally serious 2700 cf is absurd

ok I'll trust you on that but getting an absurd score on a bunch of microbenchmarks is not super relevant...

I'm not going to pretend to be the chess players in 1960 who said a computer can never beat them at chess because it cannot beat the top 100 players

especially as it has probably read multiple articles about each of those problems

I 100% believe AI is possible

these things have basically the entire internet in their training data

But the way the current models work isn't really good for logical tasks

I absolutely believe proper intelligent AI is possible, but I do NOT believe we will reach anything like that by simply scaling up existing language models

Current models you've used get 600 on code forces or worse probably

I haven't used any models hehe

22:30

How am I such a grumpy old man when it comes to computers? I've been like this since I've had a computer basically lol

meloƞ

@DeFreddie can you give me the static clang format 10 link you mentioned heinrich sent? i can't find it and i want to add it to the readme, new contributors often forget

i guess im late since you already got the pip thing

risu

ok I'll trust you on that but getting an absurd score on a bunch of microbenchmarks is not super relevant...

This is not a micro benchmark code forces was never intended to be an ai benchmark its competitive coding for humans

22:31

It existed before chatgpt or llms

Tater

This is not a micro benchmark code forces was never intended to be an ai benchmark its competitive coding for humans

no I mean competitive programming tasks are all microbenchmarks

22:31

it's actually testing the speed of a single algorithm

meloƞ

@DeFreddie can you give me the static clang format 10 link you mentioned heinrich sent? i can't find it and i want to add it to the readme, new contributors often forget

https://github.com/ddnet/ddnet/pull/7034#issuecomment-1700805235 but nonetheless

At 2700 level it's not about speed alone

22:32

You've never read a 2700 question i promise

Isn't that type of competitive programming always about the performance of the solution code?

22:33

or, well, assuming you can produce code that solves the problem in the first place

past a point it becomes about solving the problem at all

22:34

If youre not good at competitive programming then the questions are about optimizing an easy algorithm

I've competed at a national level but that's not really impressive considering the size of the nation lol

but wouldn't it also be possible that they just have better training dataset now? could still fall apart as soon as new problem arises.

but at least I can say that I was one of like five people there who weren't from a fancy maths focused high school

i've heard somewhere that AI benchmarks are basically just cheating dataset. forgot where i've heard that tho.

DeFreddie

but wouldn't it also be possible that they just have better training dataset now? could still fall apart as soon as new problem arises.

Code forces rating can only be obtained in live competitions with new questions afaik

yeah these things have read basically the entire internet like three times over

22:36

(three hundred?)

22:36

(thousand?)

22:36

honestly no clue with how ridiculous the scale of computation is at this point

22:37

if there are solutions to these problems online these models have read them

Hi @DeFreddie

but like, how new tho. chinese olympiads trainees can also just do problems 24/7 for 6 years and get through a lot of problems not even understanding what they've learned and what algo actually uses in real applications.

risu

if there are solutions to these problems online these models have read them

There's nothing they can say that will convince you then, they could just write any number they want on the graph if they wanted. I promise you the researchers making $5million a year are aware of dataset leakage

prob not 2700 level but still (edited)

Jupstar ✪

Hi @DeFreddie

Hi juppy

heartw

Tater

There's nothing they can say that will convince you then, they could just write any number they want on the graph if they wanted. I promise you the researchers making $5million a year are aware of dataset leakage

I mean yeah, it's probably really good at some things, and definitely superhuman at many, but being superhuman at a few development related tasks doesn't really make you a software developer

You won't get to experience these models first hand for a couple years because they are too expensive until improvements to hardware or algorithm is made so you can read the benchmarks and believe them or be surprised in 2 years when they aren't lying

i knew actual people who just memorize like tons of snippets and just do problems with pattern recognitions justatest

justatest

DeFreddie

i knew actual people who just memorize like tons of snippets and just do problems with pattern recognitions justatest

justatest

This just sounds like solving rubik's cubes

that's how a lot of us can skip chinese college entry exams

22:42

i was in the same trianing camp since middle school

22:43

but i failed tho, so mom sent me to the US since i've only done "coding" without going to actual schools for like 5 years

risu

I mean yeah, it's probably really good at some things, and definitely superhuman at many, but being superhuman at a few development related tasks doesn't really make you a software developer

The benchmark next to it is called "software engineer benchmark" chatgpt 4o gets 6% o3 gets 71% cat_tired

cat_tired

where was o2 (edited)

Skipped because there's a company called o2

justatest

I guess I should just become a carpenter

22:45

(mom also tells me that from time to time lol)

The guy who invented the algorithm for o1 is saying this as well https://x.com/polynoamial/status/1870172996650053653?t=xlrABqsLiH6yLdsyX1hFMw&s=19

Noam Brown (@polynoamial) on X

We announced @OpenAI o1 just 3 months ago. Today, we announced o3. We have every reason to believe this trajectory will continue.

o1 has 1891 is pretty cool too

Now the real question is if it's breaking the established scaling laws or not

too bad im still using supermaven which is usually dumb af if you let it writing a thing from scratch

because that'll really tell you whether "this trajectory will continue" or not

risu

Now the real question is if it's breaking the established scaling laws or not

You can just make ASIC and it becomes 1000-2000x faster

do they still do papers for o models

no but there are scaling laws for quality of output too, everyone kinda expects a plateau

they don't tell you much

too bad

(everyone except the VCs lol)

risu

no but there are scaling laws for quality of output too, everyone kinda expects a plateau

1900->2700 doesn't look like a plateau to me

but again, i still believe if you seen enough you can get good at codeforces without understanding a lot of it

who understands, please help me how to compile the source code of the game via MSVS? I downloaded and changed the code that I need, BUT I DON'T UNDERSTAND HOW TO COMPILE?!?!?!?

Tater

1900->2700 doesn't look like a plateau to me

That's why I'm asking if it's breaking the established scaling laws or not because that's the deciding thing for me on how I'm viewing this 1900->2700 improvement

risu

That's why I'm asking if it's breaking the established scaling laws or not because that's the deciding thing for me on how I'm viewing this 1900->2700 improvement

Scaling laws only apply to pre training

22:50

This is not that

22:50

Scaling laws say nothing about RL training either

is o3 the diffusion-adjacent thing or they don't tell you like at all

Its a transformer

22:51

As usual

transformers are so blursed

i know but there was a non linear transformer thing iirc

22:51

that doesn't do next token only

They're not going to tell you anything more detailed other than "we trained it to reason"

i guess so

22:53

i can see generation quality get better if the AI doesn't need to cope with already generated nonsense due to bad luck on tempreture alone tho

DeFreddie

but again, i still believe if you seen enough you can get good at codeforces without understanding a lot of it

Can you get good at chess without understanding it if you just play thousands of games and memorize the patterns?

oh o3 plays chess now?

22:54

that's cool

It did that since gpt4 turbo

but like is it codeforces level of good

hi from europe

22:55

im from europe

Ryozuki

hi from europe

hi

DeFreddie

but like is it codeforces level of good

We already have chess robots better than humans

im from europe too

Does it get its ass kicked by Stockfish?

If it was better than humans at chess you would just say it's not impressive because it can't beat chess engines

I literally just said that lol

Tater

We already have chess robots better than humans

i doubt stockfish is transformer

Meanwhile it does your job because your job is not chess (edited)

Tater

The guy who invented the algorithm for o1 is saying this as well https://x.com/polynoamial/status/1870172996650053653?t=xlrABqsLiH6yLdsyX1hFMw&s=19

he works for openai btw so if it's true what he says, and this is as impressive as it sounds, then gg else it was just marketing and you fall for it xd

Jupstar ✪

he works for openai btw so if it's true what he says, and this is as impressive as it sounds, then gg else it was just marketing and you fall for it xd

I know he works for openai, he is the lead researcher who designed o1

22:57

It was his idea

developers please help

22:57

hom compile game source?

why

but yeah I guess I do need to consider carpentry or something if I can't make it in the business without dealing with this garbage

read cpp tutorials and you'll understand

and read this https://github.com/ddnet/ddnet/blob/master/README.md

22:59

i mean i'm pretty sure coders are losing jobs regardless how good AI gets, cuz the business people are pretty much convinced already.

Jupstar ✪

he works for openai btw so if it's true what he says, and this is as impressive as it sounds, then gg else it was just marketing and you fall for it xd

I only posted his message because I trust him more than any of the other openai employees to be honest about his own algorithm

doesn't hurt to have better tooling to playaround with i think

idk I for one enjoy doing things

Tater

I only posted his message because I trust him more than any of the other openai employees to be honest about his own algorithm

I see, so he has your trust to lose at least. Do we know when this thing is publicly tested?

Well we can all enjoy doing things but I think there's simply a time limit now. Before I was a little skeptical there was a chance they could never do it, but they've proved now that's its only time. Who knows how much

Like not even by us, just by anyone

~~can AI generate blender constraints and switch drivers and animations, cuz im still learning that~~

Jupstar ✪

Like not even by us, just by anyone

https://arcprize.org/blog/oai-o3-pub-breakthrough

OpenAI o3 Breakthrough High Score on ARC-AGI-Pub

OpenAI o3 scores 75.7% on ARC-AGI public leaderboard.

If I have to produce code with transformer assistance to make it in the workplace I might seriously consider switching profession

They tested it on semi private benchmark

23:03

Also it won their competition

23:03

But that's small news compared to cf2700

risu

If I have to produce code with transformer assistance to make it in the workplace I might seriously consider switching profession

idk what that means, im like 99% sure managers and business does not care about how you get things done (in the workplace) and im pretty sure a lot of people can get things done better than an average new grad with copilot.

23:07

the main concern is probably whether the workplace still exists if business people is convinced a $500/month devin is better than a $3000/month engineer

DeFreddie

idk what that means, im like 99% sure managers and business does not care about how you get things done (in the workplace) and im pretty sure a lot of people can get things done better than an average new grad with copilot.

no but I mean if these things are seriously good I'd imagine matching the productivity of someone with actual skills using AI without using assistance at all would be really hard

DeFreddie

the main concern is probably whether the workplace still exists if business people is convinced a $500/month devin is better than a $3000/month engineer

In 5 years you will get 100 Devin's for $500

DeFreddie

the main concern is probably whether the workplace still exists if business people is convinced a $500/month devin is better than a $3000/month engineer

They won't be convinced for long as their shit breaks down

Devin is bad tho

23:09

Its not even worth $0

23:10

Its just a scam to trick executives

risu

no but I mean if these things are seriously good I'd imagine matching the productivity of someone with actual skills using AI without using assistance at all would be really hard

software engineering is not just coding you know... and i'm like pretty sure a lot of senior roles don't even have time to code cuz they get so much werid tnings to do

Need to make it to senior first for that :)

i mean

23:11

i got senior on day 1

fun

kek

kek

kek

23:11

ye, and i bankrupted the first company and lead a failed product in the second

23:11

so don't do that i guess

justatest

23:12

Don't put that on your resume

honestly, probably not my fault but it's hard to tell

Well yeah they wanted something that was probably always going to fail

23:13

Common story

Tater

Don't put that on your resume

but hey, a failed product is still a shipped product, it is hard to get this in games

I feel like I might be a bit safer from the AI scourge as a technical artist?

like for games?

yeah

remind me to hire you if i got money

People said creativity would be hard to copy, now they're saying that logic is hard to copy, and that role is the intersection of those two :P

Probably yes tbh

so it should last at least a few years longer

i wouldn't say software eng does not have any creativity.

you can't exactly run diffusion on your shader code to make it look prettier lol

I am biased because I do tech art but I think depending on which type you do it's safer than regular software development

23:15

Only a little tho

risu

you can't exactly run diffusion on your shader code to make it look prettier lol

o1 can write pretty good shaders

if only game industry is not running down hills

And even when were required to use assisting software there will still be (some) jobs

but it's not very good at making shader soup which looks good like a real tech artist

Tater

o1 can write pretty good shaders

can't really beat a shader graph wizard in terms of variety and just pure directivity from designers i think

Tater

o1 can write pretty good shaders

for standard things of course

Ehh

23:17

Its pretty good

23:17

I wouldn't say it can't do anything specific enough to describe to it

It'll be a long while before an AI can understand an art director well enough to make required changes to your shader code or whatever

sometimes it puts the functions in the wrong order but it usually takes like 4 seconds to fix that and it compiles first try

unless the art director lacks strong vision I guess and just says okay to whatever the bot produces (edited)

honestly, now that i think about it, probably still down to how well it can take directions. like it is really often the shader graph person gets a director just pointing at their screen and say things

Shader Graph is cammostripes

cammostripes

kinda agree

but shader graph is artists job

23:20

not tech arts

Some artists write raw shader code

nahh I hand write shader code screw the artists they can go through me

Tater

Some artists write raw shader code

Good artists*

risu

nahh I hand write shader code screw the artists they can go through me

try that when you got 3 days to optimize the renderer and 5 artists are pushing new stuff and the game is set to submit for review in 8 days

justatest

23:22

I like to pretend I won't have to deal with real life

i like to run everything by me

DeFreddie

try that when you got 3 days to optimize the renderer and 5 artists are pushing new stuff and the game is set to submit for review in 8 days

You're optimizing the render because of all the garbage shader graphs

don't know if I'll actually end up doing game dev professionally, it sounds kinda nightmarish

then i get 19hr days like regularly for 6 months

Tater

You're optimizing the render because of all the garbage shader graphs

nah. we were cpu bottlenecked

23:24

but to be fair i am not tech art but i still do tech art job cuz we don't have one

23:25

4 artists 3 dev 1 designer, 6 of them are intern on their first job

23:25

i really tried

23:27

and they gave us a "hard" deadline of 6 months to do a full (mobile-adjacent) game with client, server and web components and we do our own SA.

My C++ algorithms have gone very rusty from all the Go use pepeW

pepeW

Learath2

My C++ algorithms have gone very rusty from all the Go use pepeW

pepeW

no way

get it?

how long have it been

I actually forgor how std::remove worked completely there for a second

23:28

7 months of nothing but golang

Do you like golang

but i mean... just for a second is not that bad

Tater

Do you like golang

It's honestly not bad

23:29

The GC is a shame though

Learath2

It's honestly not bad

This feels like the common opinion

Learath2

The GC is a shame though

How so?

I expected it to be much much worse at the very least from all the talk inside this channel in the past

at least you only works in one lang.

23:30

justatest

Tater

How so?

It's a decently modern GC but it's still GC. You need to be careful not to leak things around all the time and you have to accept that your "real time" worries can't really be satisfied by this language very well

23:31

Say you have 16ms budget for a frame, gc randomly kicks in, wastes a random amount of it

23:31

Now you get a frame skip

You're using it for game or something else realtime?

I'm not, that's why it's been pretty enjoyable, but it's still a weak point of the language and gc in general

ye, if it is not realtime i wouldnt worry about gc like that much. (edited)

Yeah fair

There are other GC algorithms that are more suitable for that kind of stuff too, but go's one isn't customizable iirc

23:33

You can have a MarkAndSweep GC that gets a fixed budget of time, so you can make it constant overhead

what lang has customizable gc

nim has multiple options

23:33

nim has 7 options to be exact

i've never seen nim oof

Clearly the solution is writing code that never dynamically allocates so you don't need the GC ;)

Learath2

Say you have 16ms budget for a frame, gc randomly kicks in, wastes a random amount of it

meanwhile unity's .net GC is still garbage

risu

Clearly the solution is writing code that never dynamically allocates so you don't need the GC ;)

Which is possible but pretty hard to do in Go, since everything will accidentally leak stuff unless you are extremely careful

It's not like avoiding allocations and deallocations isn't a good idea in general

Learath2

Which is possible but pretty hard to do in Go, since everything will accidentally leak stuff unless you are extremely careful

This is really annoying in many languages

If Rust was a little easier to use I'd honestly just flat out replace almost all of my languages with Rust

Before I knew cpp I was blind to the garbage I was making, now it's like paranoia

Learath2

If Rust was a little easier to use I'd honestly just flat out replace almost all of my languages with Rust

I need to stop getting heebie jeebies from Rust so I can enjoy it and stop complaining

C++ is really getting much better recently. I love that we have proper std::atomic<std::shared_ptr> now

justatest

maybe i've been working too long, i can cope with terrible langs

Also std::ranges::contains

I like esoteric and difficult to use languages but I can't accept JS

Also std::erase_if

risu

I like esoteric and difficult to use languages but I can't accept JS

TS makes it a little more bearable

23:40

Also std::format (edited)

I just wish JS had real number types

TS gets points for being funny jank workaround but immediately loses them for being related to JS

Now it's your turn to tell your favourite C++20/23/26 feature is

ever since we managed to ship a VR game that runs at 72hz locked on a 5 year old soc in unity, i just stopped worry about terrible gc altogether. cuz no way the intern's code would survive me for the second project if it doesn't just run ok with all the jank we put in it.

People actually know C++ features? justatest

justatest

Learath2

Now it's your turn to tell your favourite C++20/23/26 feature is

string_view/span are cool

1

risu

People actually know C++ features? justatest

justatest

not me

23:44

i barely know c++

arch1t3cht

string_view/span are cool

also modules in theory but lol modules

all those views

C++26 is probably getting std::linalg with portable linear algebra. So when it's implemented in 2 decades maybe it'll be great

justatest

Exported 358 message(s)

Timezone: UTC+0