DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/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 2022-12-10 00:00:00Z and 2022-12-11 00:00:00Z

[ddnet/ddnet] Issue opened: #6112 Add tests for backup worker

So that we can be more confident that the backup worker keeps working like now. Followup on #6110

00:05

[ddnet/ddnet] Issue opened: #6113 Add unit test for updating teamranks

And make sure that all other write queries are covered by unit tests (insert rank, insert teamrank, save, load) Followup on #6107

00:12

[ddnet/ddnet] Issue opened: #6114 Fail more graceful when truncating a string on selecting data from mysql

A debug log with a keyword would be probably enough + still returning an error so that the query doesn't continue. It would be good to setup some infrastructure to grep through logs and notify (someone/a mailing list with developers or administrators?) about these failures. Followup on #6051

@Learath2 maybe you know a way to swap out a map but keep all player positions

00:26

We are considering splitting Adrenachill 5 into 2 parts, but people are already nearly at 50%

00:27

For team they can just save and load, but team 0 not

00:30

I think the best will be to set a few tiles as finish live with debugger

justatest

1

00:30

I'll try that on an empty server now

00:31

@heinrich5991 any idea how to access the right memory?

First time I'm trying something like this

do you have debugging symbols?

yes

you can execute C++ with the print command

00:33

print (Server()->Map()->…)

Yes, need the coordinate next

let me open the map in the editor

00:36

x:915 y:268-272

thanks

game layer width is 1950

00:37

so it should be 1950 * 268 + 915, 1950 * 269 + 915, 1950 * 270 + 915, 1950 * 271 + 915, 1950 * 272 + 915

00:37

these indices in the game layer

finish line index?

00:38

@Pipou what number is finish line?

34

00:39

https://github.com/ddnet/ddnet/blob/5290193f1fe3f9fb91345fb95d4b35a027acbd13/src/game/mapitems.h#L137

ddnet/mapitems.h at 5290193f1fe3f9fb91345fb95d4b35a027acbd13 · ddne...

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

thanks

00:39

I'll try

00:39

print m_pTiles[1950 * 268 + 915].m_Index = 34
print m_pTiles[1950 * 268 + 916].m_Index = 34
print m_pTiles[1950 * 268 + 917].m_Index = 34
print m_pTiles[1950 * 268 + 918].m_Index = 34

wait, these are different x y than I told you

oops

@deen index 34

Image attachment

@Pipou can you go left?

wow interesting

@heinrich5991 didn't work

00:43

(gdb) print m_pTiles[1950 * 268 + 915].m_Index = 34
$1 = 34 '"'
(gdb) print m_pTiles[1950 * 269 + 915].m_Index = 34
$2 = 34 '"'
(gdb) print m_pTiles[1950 * 270 + 915].m_Index = 34
$3 = 34 '"'
(gdb) print m_pTiles[1950 * 271 + 915].m_Index = 34
$4 = 34 '"'
(gdb) c
Continuing.

what did not work?

the finish

00:44

Pipou is standing on the tile

and he should get a time?

yes

I'm in practice

i.e. he's in race, hasn't cheated etc.?

oh, did you get a message?

00:44

that you would have finished?

Nope, will try again

00:45

Still nope

00:45

and this is the game layer's tiles you're modifying?

yes

00:45

oh, I hope

00:45

CCollision stops on multiple layers?

00:46

(gdb) break collision.cpp:282
Breakpoint 1 at 0x55a98b6f6b5d: collision.cpp:282. (3 locations)
(gdb) c
Continuing.

Thread 1 "DDRace64-Server" hit Breakpoint 1, CCollision::GetTile (y=3140, x=<optimized out>,
    this=0x7fa8d83ec0b0) at /home/teeworlds/src/master/src/game/collision.cpp:283
283            {

pipou did you make sure to touch start

CCollision::m_pTiles seems to be the game layer, yes

00:47

can you print out m_Width?

Yes @louis

00:47

Starkiller and :0 should reach it in 5 minutes

1950

00:49

"Your team has fewer than 2 players"

00:49

i think that's why!

00:50

confirmed, it worked

be careful on the live server

00:51

you're pausing it with the debugger, I think

yes, i know

I think you have everything sorted, no? I'll go to bed then

yes, good night

00:55

and thanks

good night

[ddnet-maps:master] 1 new commit

74dc13b A Adrenachill 5 - ddnet-maps

[ddnet/ddnet] Issue opened: #6115 cl_show_others not working on servers set to solo

Tees are always shown, even if cl_show_others is set to 0.

Zwelf

Please do not update to newest version (master) on ddnet server, until https://github.com/ddnet/ddnet/issues/6107 is fixed: updating existing teamranks doesn't work (edited)

I applied https://github.com/ddnet/ddnet/pull/6111 yesterday evening

09:36

Ah sorry, this one https://github.com/ddnet/ddnet/pull/6110

@nori so i am on wayland now. there is also a "allow tearing options" as expected starting with xwayland vulkan also does not warn that there is no immediate mode fps were good, around 13k but wayland clamped my FPS to around 1k and also vulkan warned that immediate not available so i build mesa from source now its completely broken xD xwayland: it shows 1k fps for few seconds then 10k fps then 1k agian etc. wayland: basically like before so i assume some packages are not ready yet. i dunno if the wayland protocol is only for the compositor, or under the hood also is a direct kernel call and i need to wait for a kernel update for now i cannot really test it yet, im not really sure if there is tearing with xwayland, but on 240hz with good fps tearing is hard to notice anyway :/ I'll try again next year xD

10:34

but pure wayland there should be no tearing bcs vk falls back to mailbox

@heinrich5991 i do not understand your comment here :( https://github.com/ddnet/ddnet/pull/5949#discussion_r1001743270 The inheritance is needed so it can be passed as a cnetbase pointer to the connection and the connection then can call netbase functions such as SendControlMsg like here

Just enough 0.7 client support to join a server by ChillerDragon · ...

While keeping 0.6 fully working and untouched This adds the option to connect via 0.7 This is the first working version there is still work todo There are a few known bugs and missing featues to na...

11:16

https://github.com/teeworlds/teeworlds/blob/26d24ec061d44e6084b2d77a9b8a0a48e354eba6/src/engine/shared/network_conn.cpp#L166

teeworlds/network_conn.cpp at 26d24ec061d44e6084b2d77a9b8a0a48e354e...

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

11:17

not sure how that relates to the m_aRequestTokenBuf variable the way i understood it is more about the method calling not a variable. Also what is so bad about the inheritance? I just implemented the changes from the upstream

https://stackoverflow.com/help/gpt-policy

Why posting GPT and ChatGPT generated answers is not currently acce...

Stack Overflow | The World’s Largest Online Community for Developers

12:08

Ops wrong ping

12:08

@Jupstar ✪ for u

Ryozuki

@Jupstar ✪ for u

lets just change reality then ai always right

If only we had the technology

Image attachment

justatest

1

12:53

100% confirmed

12:54

the whole story xD

Image attachment

12:54

i tried to let it act as if it knows everything but no chance xD

Jupstar ✪

@nori so i am on wayland now. there is also a "allow tearing options" as expected starting with xwayland vulkan also does not warn that there is no immediate mode fps were good, around 13k but wayland clamped my FPS to around 1k and also vulkan warned that immediate not available so i build mesa from source now its completely broken xD xwayland: it shows 1k fps for few seconds then 10k fps then 1k agian etc. wayland: basically like before so i assume some packages are not ready yet. i dunno if the wayland protocol is only for the compositor, or under the hood also is a direct kernel call and i need to wait for a kernel update for now i cannot really test it yet, im not really sure if there is tearing with xwayland, but on 240hz with good fps tearing is hard to notice anyway :/ I'll try again next year xD

did u apply this? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18268

vulkan/wsi/wayland: add support for IMMEDIATE (!18268) · Merge requ...

Use the tearing-control-unstable-v1 protocol to indicate to the Wayland compositor that tearing is acceptable. See

nori

did u apply this? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18268

no good to know

there are also xwayland merge requests too so

13:04

not complete yet i guess :D

but we are close ^^

[ddnet:master] 2 new commits

b7205f5 Fix updating teamranks - Zwelf 63187f5 Merge #6111 - bors[bot]

13:18

[ddnet/ddnet] Issue closed: #6107 Teamranks not inserted at all

13:18

[ddnet/ddnet] Pull request closed: #6111 Fix updating teamranks

ladies and gentlemen, it is with great pleasure to inform you that i am now officially a doctor

3

3

2

16:21

Nonetheless, i still have plenty of stuff to do, so i probably won't be working on ddnet's code soon

16:21

Being a fully grown adult is a never ending task

Hats off!

gg wp no re

congrats

congratulations mr phd

16:25

well im not sure what degree it is but congrats regardless

it is PhD. Thank you all

16:35

https://discord.com/channels/252358080522747904/757720336274948198/1051175280036806706

16:35

maps2.ddnet.org down?

Chairn

Nonetheless, i still have plenty of stuff to do, so i probably won't be working on ddnet's code soon

Do a phd in ddnet

what's the research thematic?

Chairn

what's the research thematic?

Enable QUIC support & route everything through cloudflare

Chairn

what's the research thematic?

The efficacy of layer 3 filtering of traffic in online games

Wouldn't it be easy to write a server plugin that detects griefing attempts by players on other players?

default

Wouldn't it be easy to write a server plugin that detects griefing attempts by players on other players?

bruh what

Alexander

bruh what

what am I supposed to reply to that? that a player that applies force to another player which results in them going into the freeze and ending up respawning isn't a factor you could go after? If such a player doesn't get reported manually, a set threshold could be reached within a specific time. that eliminates false positives as well. (edited)

what about people joking, people playing in playground, people respawning for other reasons

people joking wouldn't keep them in the freeze, would they? also when in team 0, and you are surrounded by other players, "joking around" is basically blocking. could be applied to a specific area. people respawning for another reason won't have force applied to them by griefer A.

I think what you are asking for would be hard to implement and could give many false positives for various reasons

I mean that's what you got smart minds for to think out very specific logic that would only apply to people grieving people. a threshold is what should prevent legits from getting banned. the aimbot detection got a threshold as well. you will get warned, not banned on the first offense. it could also be used to not punish people first hand, but letting it inform humans to check it out.

as soon as a human has to check it, its not an improvement and fully automated is hard bcs it depends also on where in the map you are

17:27

if someone trolls during hammer hits its basically impossible to detect

17:27

also some noobs might not even block but are simply too bad

17:28

train a AI that studies behavior of bad and good players

it's not about blocking persé, but about a single person killing half of the server. getting banned multiple times by vote, etc.

then u have a slight chance it would work out

by all means if you think that you have the knowledge to program what you want then go ahead

Alexander

by all means if you think that you have the knowledge to program what you want then go ahead

might check out the server code for that. but I focus more on the client-side atm.

i mean it would definitely work on that one multeasy spot

How do you tell apart a genuine fail from a block though?

17:49

There are many faily parts where a tee can hook another tee into freeze and that the other tee has to reset because of a fail

Chairn

Being a fully grown adult is a never ending task

Indeed

Learath2

There are many faily parts where a tee can hook another tee into freeze and that the other tee has to reset because of a fail

yea but we are talking about something reoccurring. not a single "kill". (edited)

18:12

say, some guy on Kobra for example, that keeps killing people at the start of the map

Learath2

How do you tell apart a genuine fail from a block though?

If his name is Learath2 then we know its a block. Ez prediction.

^

Just needs an if statement to do the job.

default

say, some guy on Kobra for example, that keeps killing people at the start of the map

Start of the map makes sense to you and me. Computers, not so much

I mean if it was that easy then it would already be implemented

Also I see a lot of novices failing on the second part of kobra trying to help eachother

Learath2

Also I see a lot of novices failing on the second part of kobra trying to help eachother

But I think that's in the nature of the game. One needs to invest time to increase skill. I remember how bad I was, still am!

nah that doesn't contribute to the discussion

18:15

you basically are saying: get good or get banned by some new anti-grief system

So it's not very very easy to write an algorithm to catch this. The best I can think of is just keeping track of how many times someone killed after being put in freeze by a player and mods can use that counter as more information

18:16

Atleast, I can't imagine an algorithm consistent enough to be trusted to ban people

Love to chat but gtg ice skating. Cheers!

Learath2

Atleast, I can't imagine an algorithm consistent enough to be trusted to ban people

you would have to test one first to say that

18:21

i believe it should be pretty clear with a big enough threshold

18:21

legit people get flagged rarely for aimbotting as well

18:21

so less accurate systems are already in place that CAN punish people. but you haven't heard of any complaints. because the system currently in place for that is so good that this false-positive happens too rare to bother. and you won't get banned if you get triggerbot flagged once in 10 minutes or something. (edited)

nor should such a grief system check be active on players that haven't been tried to get kicked by the others, eliminating legit "accidents" or how you would want to call it.

default

you would have to test one first to say that

I mean, how am I supposed to test one if I can't come up with one?

18:33

Unless you are suggesting you'll implement it, in which case I wouldn't mind doing a trial run of it

Chairn

maps2.ddnet.org down?

there is only maps.ddnet.org

18:51

congrats Chairn!

off to industry or doing a postdoc now?

Learath2

So it's not very very easy to write an algorithm to catch this. The best I can think of is just keeping track of how many times someone killed after being put in freeze by a player and mods can use that counter as more information

1. maybe keeping track of how many hammers / hooks there is between 2 players? That's usually how fights are (even tho players you are just "playing around" would also be here). 2. If a player gets hooked into freeze, there are players alive close the him, but he stays in freeze for X (a lot) of seconds, that would be blocking? Anyways, I feel like any "blocking algorythm" would still a human vote (with less votes to be successful) to confirm if a player is really blocking. Unless it's a really good trained ai xD. But yeah, as long as we have mods... it not too much of an issue?

ChillerDragon: if you move that variable out of the class and inside the function, the methods don't have to be class methods anymore and can be static instead

Devinci

1. maybe keeping track of how many hammers / hooks there is between 2 players? That's usually how fights are (even tho players you are just "playing around" would also be here). 2. If a player gets hooked into freeze, there are players alive close the him, but he stays in freeze for X (a lot) of seconds, that would be blocking? Anyways, I feel like any "blocking algorythm" would still a human vote (with less votes to be successful) to confirm if a player is really blocking. Unless it's a really good trained ai xD. But yeah, as long as we have mods... it not too much of an issue?

So you'd just teach blockers to hide their behavior better. I prefer blockers who just stand around start and block everyone over those who play map normally and kill everyone 1 part before finish

Devinci

1. maybe keeping track of how many hammers / hooks there is between 2 players? That's usually how fights are (even tho players you are just "playing around" would also be here). 2. If a player gets hooked into freeze, there are players alive close the him, but he stays in freeze for X (a lot) of seconds, that would be blocking? Anyways, I feel like any "blocking algorythm" would still a human vote (with less votes to be successful) to confirm if a player is really blocking. Unless it's a really good trained ai xD. But yeah, as long as we have mods... it not too much of an issue?

Dummy and Pseudofly players: justatest

justatest

murpi

Dummy and Pseudofly players: justatest

justatest

TRUE LOL. Well, there are a lot of things wrong with what I said anyways. so yeah feelsbadman

feelsbadman

deen

off to industry or doing a postdoc now?

im doing an ATER which is postdoc but with teaching

cool

im now looking for permanent position and also postdoc in europe

I always see cool academic jobs at EMBL here in Heidelberg, but they always ask for PhD at least, so not for me

23:34

so France also has this curse where researchers only get temp positions by default?

one of the jury member told me about aachen university for system architect position

deen

so France also has this curse where researchers only get temp positions by default?

you can get ATER wich is like 1 year contract with same work as permanent position (my current status), or postdoc where you only do research

RWTH Aachen is one of the best for compsci in Germany, so go for it

Exported 164 message(s)