DDraceNetwork - developer

nice to see theres at least some errror checking for maps

when u save a map in editor, all unused tiles in tele/speedup/switch/tune layer are replaced by 0 after reload it. (edited)

05:21

is it expected behavior?

ZerolAcqua

[ddnet/ddnet] Issue opened: #9762 Editor: all unused tiles in tele/speedup/switch/tune layer are replaced by 0 after reload it.

If you draw unused tiles in tele/speedup/switch/tune layer, they will be replaced by 0 after reloading, in this video, you can see the direction, force and max speed (i forgot to set in video) of speedup tile are kept, but the index change to 0. https://github.com/user-attachments/assets/b93c3b82-8f28-414b-81a2-ba455f90f78c

david loves Jellycar Worlds

nice to see theres at least some errror checking for maps

If you got time on your hands, you could alo try to find maps that can crash the library twmap https://gitlab.com/Patiga/twmap. It's in Rust and in general tries to catch all errors and not crash happy

(edited)

3

lmfao

def-

[ddnet-web:master] 1 new commit

b4d14a6 Add 3 € donation by CLANNAD - def-

2

6

Patiga

If you got time on your hands, you could alo try to find maps that can crash the library twmap https://gitlab.com/Patiga/twmap. It's in Rust and in general tries to catch all errors and not crash happy

(edited)

I have seen some blog about fuzzing in teeworlds map, this could maybe be applied here

Does anyone know, how I can get tuning in prediction/character? This is client sided and works a bit differently

09:06

probably something like GetTuning(GetOverridenTuneZone)->m_MyNewVar

09:07

ah this works with CTuningParams directly, I see

Yo the new update doesn't just increase max_clients to 128 right? (edited)

10:35

That would be so bad

My guess is, that it uses sixup logic for older clients, ofc just a guess (edited)

10:36

imagine being on a client which only allows 16 players when there are 128 xD what can go wrong

Teero

That would be so bad

What would you expect instead?

@Jupstar ✪

3 lines of code

Assa

@Jupstar ✪

3 lines of code

neat

PR coming soon, need to adjust the default value, set it accidently to 128 instead of 118 xD

AssassinTee

[ddnet/ddnet] Pull request opened: #9763 add a speedboost limiter tuning

As discussed in #9670, speedtiles can give so much speed, that they actually slow you down again. This PR introduces the speed_boost_max_speed tuning parameter, at which the speedtiles stop giving you extra speed. 118 is the magic number, that has the highest default speed. It is depending on velramp values, but not easily calculable. 0 simply deactivates this behavior and makes default behave like previously @Jupeyy @Patiga @KebsCS

Checklist

[x] Tested the ch...

Assa

@Jupstar ✪

3 lines of code

also @Patiga and @kebs

So much speed tuning in the past few days poggers2

Last thing left on my speedtile list is maybe clamping the slowdown of max_speed Force, so that max_speed isn't such an instant blocker

11:05

this would be changing TempVel += Direction * (TempMaxSpeed - CurrentDirectionalSpeed); to TempVel += Direction * max(-Force, (TempMaxSpeed - CurrentDirectionalSpeed));

11:05

This is just up for discussion

‍♂️

11:06

this would also change the behavior again of max_speed set for different forces

So it just caps and doesn't interfere with any existing behaviour? +1 (instead of, well, going over a specific value slowing you down) A map utilizes this, but with bluetele and grenade IIRC (edited)

a booster with max speed can currently instantly slow you down to max_speed

11:11

the other proposed way can only slow you down up to Force

Ah I see

meaning you can still be faster then max speed

11:11

but still experience a slow down

11:14

this can also (partially) prevent the reflecting up part patiga was so worried about. But you can still do it with hughe Force values (edited)

a good default value makes more sense than keeping it unclamped anyway - wonder why we didn't do that before But maybe we just didn't bother because some maps utilized this so it was considered ⭐a feature ⭐

I think now I can explain why you fly up like for a 5 year old: @Patiga when you hit a speedtile with max speed set and you are way too fast, the speedtile instead pushes you in the opposite direction in order to slow you down. The opposite direction of bottom right is top left. Right and left speed gets cancelled, but you will be pushed up

meloƞ

a good default value makes more sense than keeping it unclamped anyway - wonder why we didn't do that before But maybe we just didn't bother because some maps utilized this so it was considered ⭐a feature ⭐

Nobody was italian enough to touch the old code

ChillerDragon

[ddnet/ddnet] Issue opened: #9764 Exclude clients that are not connected from the console victim "all"

Suggested by heinrich in https://github.com/ddnet/ddnet/pull/9584 The current problem is that when passing a specific client id to a rcon command and the id is not valid the user wants to see an error message. But then passing the magic id "all" nobody wants to see up to 127 error messages for all invalid ids that are not connected to the server. Imo it would be nice if the command that was called would know if it was a single id or all ids and could then adapt its verbosity but since that ...

Jupstar ✪

What would you expect instead?

Not making the game 2x more CPU heavy? This will prob reduce my fps by a lot.

11:24

even with only 2 players on the server everything still loops over the 128size array

11:31

@Jupstar ✪ how is ddpg coming along

Teero

Not making the game 2x more CPU heavy? This will prob reduce my fps by a lot.

Agreed, but would you instead expect to use Vec or smth?

Teero

@Jupstar ✪ how is ddpg coming along

Dunno, ddrace didn't advance ig :/

Jupstar ✪

Agreed, but would you instead expect to use Vec or smth?

yes definitely. There are some problems with strong/weak no?

11:32

weakhook and bounce is a bug feature anyway

Teero

yes definitely. There are some problems with strong/weak no?

if you don't keep the order, then yes. If you keep the order then mostly removing and inserting is slower ig

can someone explain me how "weak bug" works, based on the code. I really struggle to make sense out of it

11:36

ok nice the bridge works again

11:53

is it unsafe to mem_zero a multi dimensional array?

11:53

removing this line fixed a crash in my client hm ... ``mem_zero(m_aaDbgMsgs, sizeof(m_aaDbgMsgs));

<ChillerDragon> mem_zero(m_aaDbgMsgs, sizeof(m_aaDbgMsgs));

annoying submit message on paste

11:54

fakin weechat

11:56

as far as i know multi dimensional arrays are also contiguos in memory so zeroing out the full size should not corrupt anything??

it shouldn't™

11:57

chillerdragon: can you send a link to the repo if it's not private?

yea the recent merge into ddnet broke chillerbot-ux but the crash that is more complicated to debug is only in its fork that shall not be named to not violate rule 7

<ChillerDragon> replacing the mem zero with for(auto &aDbgMsg : m_aaDbgMsgs) aDbgMsg[0] = '\0'; fixed the crash but it is still full of asan warnings of pointers with invalid type so probably the issue is elsewhere

12:05

<ChillerDragon> sad that asan does not help me find the origin ._.

12:08

<ChillerDragon> oof this is frustrating @MilkeeyCat are you interested in debugging it? :D if you leak your gitlab name i can invite you to the repo. Its only closed source so skids dont abuse its features for illegal purposes but you could build a more harmful client your self in 10 minutes.

i don't have a gitlab account xd

<ChillerDragon> wot

12:09

<ChillerDragon> blaze it netherlands alternative for microsoft crap?

12:09

<ChillerDragon> just click "sign in with github" :D

uninstalling graphics lib removed keepassxc -.-

<ChillerDragon> which distro u on again?

debian

<ChillerDragon> didnt you recently rage quit gentoo?

it was a year ago xd

<ChillerDragon> as i said