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 2021-07-22 00:00:00Z and 2021-07-23 00:00:00Z

For skins.tw, on upload I store the md5_checksum of files now to detect duplicates and tell the user that this skin already exists. I thought I finally had it, because it works 99% of the time. But some completely different image files seem to have the exact same hash value.

Image attachment

00:55

But what is this?

Image attachment

isnt md5 pretty bad for hash collision?

why

i dont know specifics i just remember reading it

Its bad for security

01:00

But I dont need it for security

01:00

Like hashing a password. From what I read its fine to find duplicates

Did you download both PNG and checked with a different tool?

hm no I could do that

01:02

I literally tested my duplicate check with like 20 assets. Worked nicely, but the last one was this weird one

01:02

But let me check

01:07

9db702c6cdeb9c73412a72f0f8f54d98 (chinese by whis) 877c479b2b7dbfe9b0c1c0d6bbd14f44 (eggp)

01:07

They are even different, bruh

01:07

I think my sql query is just weird

ddnet

suggesting ddnet

ddnet

add median time here

Image attachment

3

4

weird thing to add for browser considering there are other useful things not there

01:18

might work better as a server filter

01:18

like for example, there's no points tab

I mean you could see "do I have time for this map", but I agree there are better informations to display

median time is a numeric value so i think you could implement it better as a filter

1

01:20

but sucks you couldnt sort it by median (edited)

How responsive is this browser list even

01:20

on smaller screens

probably terrible

01:21

font size is not scaled

01:21

i learned this the hard way

lynn

weird thing to add for browser considering there are other useful things not there

so best thing to doing

add all this things and make it appear or hide and size Optional to players

yeah that would be the best, but pretty big lift for devs

devs have to work to die giftee_red

giftee_red

[ddnet-maps:master] 1 new commit

ceb5db4 M Sleepy, M Repentance, M 0 to 100, M Victory 3, A run_Jellyfish_Fields - ddnet-maps

NeXus

why

the entire point of a hash is to be unique justb y changing the smallest thing

06:34

and md5 is proven to be bad at that

06:34

it doesnt matter if its not for security

06:34

sha256 is ur best bett

06:34

for a fast hash

06:35

oh i guess ur query is wrong

06:35

(finding a md5 colision shouldnt be that easy i think)

all sha are rly good

06:38

oh jk sha1 sucks

Ryozuki

and md5 is proven to be bad at that

Well md5 isn't bad specifically at that. Minor changes cause wildly different md5 just fine

well yeah

10:18

xD

10:19

but i dont think there is much perf dif between using sha256 and md5 right

just not crc thanks

java

Image attachment

Especially in modern hardware they are fairly similar

10:22

Anyway iif you are going to use md5 you have to keep in mind that the user can generate two different files with the same hash. And in this case it doesnt matter that much since you reject duplicates

https://nuclearmonster.com/2021/07/native-linux-games-vs-windows-api-compatibility-layers-on-the-steam-deck/

Native Linux Games vs Windows API Compatibility Layers on the Steam...

I spoke with Ryan C. Gordon, a renowned Linux developer for the past 22 years, today over a text chat to get his thoughts on the Steam Deck and native Linux ports in the light of Valve’s call…

10:23

Ryan C. Gordon is the main maintainer of sdl2

10:23

iirc

10:26

We are no longer talking about people running Linux as their primary desktop to compete here. We are no longer talking about how many people found a machine to install SteamOS on.

>

We’re now talking about concrete sales numbers for a game console.

>

And maybe someday down the road, if this is wildly successful, we tell people that it’s a no-brainer to target 18 bazillion Linux users that aren’t Linux users so much as customers reliably running a Linux-based game console.

10:26

this is such an important remark

10:26

this can make linux be popular whithout it being known at all xd

there are already a lot of linux users, see android

10:27

but I guess they don't run desktop linux

yeah its not the same

but the comment above still seems like it's mostly trying to get upvotes from linux fans tbh

Ryan C Gordon should stop doing interviews and do my sdl patches smh

monkalaugh

10:29

the Steam Deck and native Linux ports in the light of Valve’s call for developers not to port games to Linux and instead rely on their Proton fork of WINE and other tools.

10:29

valve is actually asking ppl not to port games to linux but use proton

10:29

idk what to think of this tbh

@Learath2: @murpi identified all the dummy maps that need changing for https://github.com/ddnet/ddnet/pull/3959

Try to fix "start line skipping" by heinrich5991 · Pull Request #39...

Previously, tees could go around the start line on some maps, going directly to the finish line while leaving one tee behind at the start. By letting that sole tee start and kill itself, the team c...

I agree more with ryan here, what is this guys problem with calling out the windows layer as ruining linux?

can we get this in? we had a large increase in start line skips recently

I dont see any problems with it and I don't see any code quality issues. So on that front I'm okay with merging it

10:31

But as with other patches like this I'd really prefer if you could set some time aside and deploy it yourself/monitor it a bit

then it'd have to wait until saturday

10:33

or we could deploy it this evening maybe

Either works for me, up to you really

k

Saturday is a little less than ideal as I'd be very tired from the flight but I guess you can deploy alone

Learath2

Saturday is a little less than ideal as I'd be very tired from the flight but I guess you can deploy alone

vacation?

Sort of, going back to turkey to see family. Will still be plenty busy so not sure if it qualifies as vacation :D

You'll finally be back home again!

10:45

thats great news

iirc

@NeXus: ah ye i wanted to tell you when i saw on github that you search dupes I have some CI for my mapres also using some python lib to detect similar images because sometimes same skins get resaved and have different hash

I fixed it now, but thanks

i also check mapres for proper resolution

I do too

13:14

also support hd skins now

ah ok nice

imagehashing skins is a decent idea, you can then use a large hashmap to see if a skin that looks nearly identical exists

Yeah it doesnt check for nearly identical skins atm, they have to be the exact same. I also noticed that even opening a skin with an image editing software, changing nothing and saving it again as png results in a "different" image with a different hash. I guess it depends on the algorithm they use to save images?

13:56

Also sending skins via social networks compresses them and change the hash

It's probably a difference in the png metadata rather than the image itself unless you changed settings like more bits per pixel

13:58

So hashing images as is can be a bit of a hit and miss

13:58

A naive approach to get more correct (with non lossy formats like png or bmp) is to only hash the pixel data

the mapoptimizer for example does the SHA over the bitmap of the image and ignoring fully transparent pixels (edited)

But imagehashing takes it a bit further. They also hash loosely so similar images land in the same bucket

Sometimes people upload almost identical skins, they just make corners smoother or fix positioning of eyes [...]. Wouldnt be bad to detect these as duplicates aswell (edited)

14:02

So they have to delete their old version first

wouldnt you just allow 1 name 1 skin then?

I got vaccinated today

14:03

monkalaugh

Yeah but they just name it xx2 __fix etc

Ryozuki

I got vaccinated today

SeemsGood

then look for similar names

hm

14:05

could be false positive no

would nouis be false positive to the original skin?

Depends on how I check for similar names

i mean like if you want similar skins to be found by some algorithm

@NeXus use the power of the hip machine learning

xD

monkalaugh

Found an example for someone who uploaded updated versions. Idk this is a problem for future NeXus

I like how it is for now

Image attachment

yeah, do you want to force it btw?

Actually some crazy AI image recognition would be the best solution

or just some user specfici warning, "delete this first"

14:17

the problem is always the tollerance if i add eyelashes and call it Jake_female, it might be a different skin xd

I would force to not be able to upload I think (depends on how accurate the algorithm will be)

14:18

I force it now for exact same hashes

14:18

because im pretty confident its the same skin

14:18

Tested it pretty much now

yeah, guess thats a good idea

Strip metadata and hash

zopflipng --keepcolortype -y -m name.png name.png

14:19

but very expensive calculation

14:22

ah btw our dilate script also uses the hash of the bitmap

14:22

https://github.com/ddnet/ddnet/blob/master/scripts/check_dilate.sh

ddnet/check_dilate.sh at master · ddnet/ddnet

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/check_dilate.sh at master · ddnet/ddnet

14:22

exactly to prevent this from happening, zopfli also changes the chunks of the PNG so its not really hashable

Ryozuki

I got vaccinated today

I’m getting my second dose tomorrow. Hope to hear you on the 5g channel

NeXus

Actually some crazy AI image recognition would be the best solution

Try imagehashing first, runs fast and old enough to have a couple major mature algorithms

Ryozuki

I got vaccinated today

do you also feel it, when u lift your arm? xd

On the first dose it didnt hurt at all for 4-5 hours. Then it started to ache a bit, ached for an entire day, the next day I woke up and it was gone

for me i directly felt it

14:37

after 1-2minutes

14:38

did you got biontec or moderna btw?

14:38

or other stuff

I rolled biontech

yeah me2

Italy pretty much ran out of everything else and they are only reordering biontech and j&j for now

in germany they want to get away from astra xD

14:39

bcs nobody wants it

14:40

and yeah biontec seems to have higher protection against the mutants

Yeah they only recommend AZ to people that are more allergic here and people older than 60

14:41

But recommend is a strong word there, they wont give you AZ unless you are an allergic person or specifically ask for it

moderna

Ooh american spy

i think i felt it some mins after but now nothing

14:42

u have to wait 15 mins there after taking the vaxx

bill gates owns u now

kek

14:43

i feel a bit tired tho

J&J is pretty promising too with it being single dose and all but we got so little of it

monkaS

14:43

but sometimes i dont know if its me being overly conscious

14:43

lmao

I felt tired after too, then I noticed I walked 3 km back home under the sun

oh yes

14:45

i just come here home

14:45

outside was 31C

14:45

sunny

14:45

barcelona

i didnt felt tired, but my arm felt a bit weird for 2 days, but 2nd day was not alot anymore

you know when there is a bit of wind

I’m very confused how people can still claim the vaccine will kill you because it wasn’t tested long enough

but its hot af

14:46

like ur in front of the AC hot air thingy

14:46

xd

Learath2

I’m very confused how people can still claim the vaccine will kill you because it wasn’t tested long enough

every polish player claims this

@Learath2 hav useen linus torvads on the mail list against a anti vaxx

14:46

https://www.theregister.com/2021/06/11/linus_torvalds_vaccine_smackdown/

Linus Torvalds tells kernel list poster to 'SHUT THE HELL UP' for s...

'Any Linux discussion list isn't going to have your idiotic drivel pass uncontested from me'

14:47

xd

Millions of doses administered over almost 1.5 years now and only a HANDFUL of people even had large complications and even for them most people think they had a pre-existing condition

Linus Torvalds tells kernel list poster to 'SHUT THE HELL UP' for saying COVID-19 vaccines create 'new humanoid race'

for logical thinking ppl its always a pain if someone completly denies smth xD

https://lore.kernel.org/ksummit/CAHk-=wiB6FJknDC5PMfpkg4gZrbSuC3d391VyReM4Wb0+JYXXA@mail.gmail.com/

14:47

this is gold

ScReeNy

every polish player claims this

So sad to see news being so populist all over the world, italian media had these people for a couple seconds too

<lkml@metux.net> wrote:

And I know a lot of people who will never take part in this generic

human experiment that basically creates a new humanoid race (people who generate and exhaust the toxic spike proteine, whose gene sequence doesn't look quote natural). I'm one of them, as my whole family.

14:48

this is in the kernel mailing list

14:48

ppl who are supposedly smart

14:48

insane

btw this also happens in development, just read today: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/65 its about input lag, bcs wayland's perfect frames are better than tearing and latency (edited)

Tearing updates protocol (!65) · Merge requests · wayland / wayland...

This protocol is meant to be used by drivers to be able to hint to the compositor if and when it should employ asynchronous page flips for presentation....

Ryozuki

<lkml@metux.net> wrote:

And I know a lot of people who will never take part in this generic

human experiment that basically creates a new humanoid race (people who generate and exhaust the toxic spike proteine, whose gene sequence doesn't look quote natural). I'm one of them, as my whole family.

You need to be dimwitted to think mrna vaccines change your genome. Like you really have to have not read a single piece of information about it

14:49

If only it was that easy to change the genome you goofs…

monkaS

Deleted User

btw this also happens in development, just read today: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/65 its about input lag, bcs wayland's perfect frames are better than tearing and latency (edited)

lol

i really wonder if humans with gen mutation would be worth it more than machines in the end i mean maybe for a specific time range, and some specific jobs but in the end mashines are much better most of the time

everybody will have blond hair blue eyes, perfect face and no cancer

14:51

monkaS

i have most of that xd

same kek

https://en.wikipedia.org/wiki/Gattaca

Gattaca is a 1997 American dystopian science fiction film written and directed by Andrew Niccol. It stars Ethan Hawke and Uma Thurman, with Jude Law, Loren Dean, Ernest Borgnine, Gore Vidal, and Alan Arkin appearing in supporting roles. The film presents a biopunk vision of a future society driven by eugenics where potential children are conceiv...

14:52

i remembered this movie

14:52

The film presents a biopunk vision of a future society driven by eugenics where potential children are conceived through genetic selection to ensure they possess the best hereditary traits of their parents.[5] The film centers on Vincent Freeman, played by Hawke, who was conceived outside the eugenics program and struggles to overcome genetic discrimination to realize his dream of going into space.

genetics are defs the hardest topic ever

i think the movie focuses on the wrong thing, if u can modify genes u should be able to become ageless right?

14:55

monkalaugh

e.g. i dont fear that plants are genetical modified if studys show a real benefit, but e.g. currently in agriculture its mostly used to have a resistence against glyphosate. So if you spray it all plants around it die And its not clear if it can cause cancer

14:57

so i'd prefer a mechanical solution for it in the meantime, e.g. machines that cut weeds

Deleted User

i really wonder if humans with gen mutation would be worth it more than machines in the end i mean maybe for a specific time range, and some specific jobs but in the end mashines are much better most of the time

Worth it for what? To your capitalist overlords? Maybe, but the main point would be eradicating genetic disease and maybe even boosting our immunity to things like cancer

Learath2

Worth it for what? To your capitalist overlords? Maybe, but the main point would be eradicating genetic disease and maybe even boosting our immunity to things like cancer

yes, i mean the classic argument against vancines "it alters my DNA to control me" I just wonder if they think they are worth enough to be controlled xD

Oh yeah complete insanity. Tech giants already control you, just check your google maps history

google tells me if i wanna know a recap of where i want

15:00

monkalaugh

15:00

i went*

The ads you are served absolutely are there to manipulate you, not just to buy, it also manipulates the way you think. Look at how cambridge analytica manipulated an entire country by serving them ads they’d hate

yeah, ads are so annoying and they get more and more xD

monkalaugh

Image attachment

Altering your dna to control you is like using a ring -2 exploit to steal your passwords

yeah

You can make it work in a very small scale inside a lab, but in the wild it’s faaaar faaaaar faaaaaaaaar off from feasible. Observing chunks of data without context to find a password there is insanity

yeah interesting point

15:03

thats also why i wonder so much how chinese ppl think xD

15:03

i mean chiense ppl vs their leaders

15:03

cant be the picture we always get

like we dont know about how they think they probs dont know about us here, they cant compare so they probs thing theirs is right

You mean “and” their leaders. There is no vs there, they all live in perfect cohesion

thats the question

15:04

especially bcs alot of chinese study outside of china

Learath2

You mean “and” their leaders. There is no vs there, they all live in perfect cohesion

well if u dont go against the leaders u probs jsut live normally

15:05

without being able to say x things

15:05

or draw human bones in games

If you are not in cohesion you take a nice long vacation

yeah but i mean under the hood

15:05

not publicly

https://dota2.fandom.com/wiki/Low_Violence

Low Violence is a series of graphical variations used on Perfect World servers. They were made to edit out or obscure blood, gore, skulls and skeletons in order to conform with Chinese censorship.

15:05

Low Violence is a series of graphical variations used on Perfect World servers. They were made to edit out or obscure blood, gore, skulls and skeletons in order to conform with Chinese censorship.

Deleted User

especially bcs alot of chinese study outside of china

I’ve had a couple chats with chinese students here, they are fairly shielded to outsiders. They hang out in their own groups

i see

its hilarious

Image attachment

15:06

china

Ryozuki

everybody will have blond hair blue eyes, perfect face and no cancer

justatest

I wonder if their time outside changes their thinking a lot at all

but in the end there is taiwan and hongkong and they are both economically not bad but dunno how much the press hides these facts

15:07

or how ppl think about it, that have a higher intellect

why does dev-channel become into polit-channel every time

its also a way of development

15:07

it doesnt say teeworlds development xd

ye, make human-dev channel :D (edited)

Chinese taipei and hong kong autonomous regions are very happy under ccp rule tyvm

make a creative room channel

15:08

creative would

gerdoe

why does dev-channel become into polit-channel every time

It’s a bunch of computer people, sometimes you just want to talk about things with people with similar backgrounds

i mean

15:09

this channel is logging out

15:09

and indexing by any search engine

the good thing about computer scientist is, that they have to accept facts, bcs that's litterally how programming works xd

your thoughts here are public (edited)

is that a problem xd

Not like we are hiding anything

xd

15:10

true

as soon as tsfreddie starts the chinese revolution i can say IT ALL STARTED HERE

I probably can’t work in the us anymore though

15:11

China might still take me

anyone else having dns issues right now?

i can ping google fine

https://edgedns.status.akamai.com/incidents/n5zl6dythvfv

Edge DNS Service Incident

16:33

F

@murpi i recommend u not relying on ur ISP dns

16:33

and just use

16:33

1.1.1.1 or 8.8.8.8

im using public dns

use 1.1.1.1

https://twitter.com/campuscodi/status/1418245632922370048

Catalin Cimpanu (@campuscodi)

The internet is on fire again. This time courtesy of Akamai DNS: https://t.co/BUusCvwNGb

16:41

@murpi

yeah

16:43

looks like they fixed it already

[ddnet-discordbot:master] 3 new commits

f9d0603 Update database population scripts - Learath2 8e5fc1a Remove only after we are done - Learath2 99efa94 Update to use local files - Learath2

16:57

[ddnet-discordbot:master] 1 new commit

3fa04a4 Add ddnet.py - Learath2

16:58

[ddnet-discordbot:master] 1 new commit

08983b7 Tiny bit of logging to help diagnose - Learath2

16:59

[ddnet-discordbot:master] 1 new commit

59cc1f0 Make the scripts executable - Learath2

17:00

[ddnet-discordbot:master] 1 new commit

1441baa Make the scripts executable - Learath2

can someone explain to me why all the rust python-bindings crates exist instead of them being integrated into the crates they create bindings for?

example?

[ddnet-discordbot:master] 1 new commit

27ae271 Create map_backgrounds folder - Learath2

https://crates.io/crates/hyperjson

17:12

hm cant remember the second one i took a look at

looks like it has different authors

17:19

@Learath2 let's deploy the PR

[ddnet-discordbot:master] 1 new commit

a094c86 Don't forget newline - Learath2

[lib]
crate-type = ["rlib", "cdylib"]

my understanding is that this will change the crate-type from "lib" to those two. I'm not sure if this has an effect on people that want to add my crate as a dependency or just on people that build it locally

heinrich5991

@Learath2 let's deploy the PR

The startline one?

yes

Sure

17:22

Should we get rid of CodeQL btw, it keeps failing

is seem to need those options if I want to produce python bindings for my library https://pyo3.rs/v0.14.1/building_and_distribution.html#manual-builds, might that be a reason to create a new crate?

@heinrich5991 1 last question, do you know if old saves are fine with this?

untested, but added with the same backward compatibility as everything else

17:25

it assumes that every tee has hit the start line already

Patiga

[lib]
crate-type = ["rlib", "cdylib"]

my understanding is that this will change the crate-type from "lib" to those two. I'm not sure if this has an effect on people that want to add my crate as a dependency or just on people that build it locally

i think this is for making dynamic libraries

17:29

https://doc.rust-lang.org/reference/linkage.html

Linkage - The Rust Reference

hm, if you have rlib in there I thought the linkage would be fine for people using your crate too

@Ryozuki I also read those docs (maybe skimmed more than read), though I was not able to answer my question

whats your question

I'm not sure if this has an effect on people that want to add my crate as a dependency or just on people that build it locally

why all the rust python-bindings crates exist instead of them being integrated into the crates they create bindings for?

17:31

i dont get what u mean

@Patiga sometimes I find the best course of action is to just try it, it'll break pretty quick if it's a hard requirement

hyperjson is a python library made in rust i think

@Ryozuki he is asking why people are splitting off their python bindings into separate crates

an example?

17:33

i dont see hyperjson as multiple crates here

Learath2

@Patiga sometimes I find the best course of action is to just try it, it'll break pretty quick if it's a hard requirement

fair, in this case though I fear that I wouldn't catch side effects

Patiga

I'm not sure if this has an effect on people that want to add my crate as a dependency or just on people that build it locally

@Ryozuki my question that I quoted here would suffice for me :)

i think if u specify rlib

17:34

it can be statically linked

17:34

hmm

17:35

i think testing it is the best

17:35

xd

17:35

If a static library is being produced, all upstream dependencies are required to be available in rlib formats.

[ddnet:master] 4 new commits

35402b0 Try to fix "start line skipping" - heinrich5991 4c2bd53 /save each tee's starting state - heinrich5991 9f57293 Warn teams after 30s if not every tee has started - heinrich5991 5ff7b7f Merge #3959 - bors[bot]

17:35

[ddnet/ddnet] Pull request closed: #3959 Try to fix "start line skipping"

If an rlib file is being produced, then there are no restrictions on what format the upstream dependencies are available in

I think(tm) by doing rlib you are sort of forcing the compilers hand, so if another rust project includes your project they need to link statically

f a dynamic library or an executable that is being dynamically linked is being produced, then the compiler will attempt to reconcile the available dependencies in either the rlib or dylib format to create a final product.

Ryozuki

If an rlib file is being produced, then there are no restrictions on what format the upstream dependencies are available in

so has it an effect on someone who uses my library as a dependency?

17:39

okay that reply is misleading

17:39

I still don't know the answer to my question tho

[ddnet-discordbot:master] 1 new commit

4b322ab Line endings... - Learath2

Check the linkage

17:40

Info

@Ryozuki the things that you are copying from the docs are lines I already read, if you think that they still answer my question, it would help me if you could formulate it differently

It tells it there

17:40

Hmm

Learath2

I think(tm) by doing rlib you are sort of forcing the compilers hand, so if another rust project includes your project they need to link statically

ah, thats interesting :)

If they use ur lib as a dep and compile statically then urs is too

17:42

@Learath2 if i lift my arm from vaccine i can actually feel some pain monkalaugh

monkalaugh

17:42

justatest

cammostripes

justatest

Ryozuki

If they use ur lib as a dep and compile statically then urs is too

ah, so the crate that uses my lib as a dep can overwrite that for my lib, or does that require my lib to include the 'rlib' crate-type

I dont know if it requires ir

17:43

It

oh, it's merged

17:47

gonna deploy it

for me the docs kinda sound like they don't want me to touch it

it seems the only way to find out is ask on some rust channel

[ddnet-discordbot:master] 1 new commit

773194b Use Path - Learath2

@Learath2

17:50

that commit looks like it changes behavior

17:50

map_backgrounds/abc.png → map_backgrounds/abc

[ddnet-discordbot:master] 1 new commit

16feb5c Early error - Learath2

heinrich5991

that commit looks like it changes behavior

oh wow, how did you even catch that

I looked at the commit because it popped up in the chat

[ddnet-discordbot:master] 1 new commit

5c1124a Oops - Learath2

you used the path in two places ^^

[ddnet-discordbot:master] 1 new commit

db8f7fa Missed one - Learath2

thumbs up

@heinrich5991 btw did you deploy the patch?

./build.sh                                                                 ~/src
15016 (process ID) old priority 0, new priority 19
[…]
[100%] Built target map_optimize
Copying to servers

17:57

still in progress

leaks

17:57

calling the nsa

hmm hmm

17:58

theoretically it should have output for each server

17:58

but I don't see a single one yet

17:58

I'll restart the script

17:58

ah, apparently my connection broke?

in an ideal world I'd make these scripts parallel with multiprocessing, but I really don't have the energy for it

18:01

so, it takes like 5 minutes to parse the releases file

I don't understand your last message

18:02

which releases file?

https://ddnet.tw/releases/releases

18:02

A list of all maps, when they were released, who their mappers are, etc

ah

18:03

turns out the script is already parallel

18:03

and the status output doesn't go to stdout

18:03

but to $$.log

Ah, that makes sense

:o

18:04

well parallel with 2 cores

18:04

its that

18:04

monkaS

doesn't matter, it's mostly IO

ah

18:05

https://github.com/tokio-rs/loom

GitHub - tokio-rs/loom: Concurrency permutation testing tool for Rust.

Concurrency permutation testing tool for Rust. Contribute to tokio-rs/loom development by creating an account on GitHub.

18:05

TIL you can test concurrent code like this

18:05

xd

loom is indeed a cool idea

Loom provides a way to deterministically explore the various possible execution permutations without relying on random executions. This allows you to write tests that verify that your concurrent code is correct under all executions, not just “most of the time”.

deployed on ger2:8303

18:06

anyone up for quick testing?

This ofc is not very cheap to test, with there being n! permutations

heinrich5991

anyone up for quick testing?

Sure

@Ryozuki no, not hard restart

18:12

restart when empty

ok

lol, didn't know the bot kept the entire database of finishes aswell

18:30

It will take the rest of my human life to wait for this script to copy it all

only once per day

Yay, finally it's all there

18:36

I'll probably hack in a mysql connection there so it can use the actual record_race table

postgres better

ye postgres ftw

Definitely but ddnet is using mysql for now

18:39

unless one of you wants to take the immense task of rewriting sql_score for postgres and taking the risk of migration it'll be like that for a while

Can somebody fix/restore build push for #bugs bug

@heinrich5991 um, something broke

19:23

But I specifically asked you be here when we deploy it

I'm here

19:23

(but I was away, I'll give you that)

19:24

@ScReeNy can you tell me which map?

@heinrich5991 flw into my server

all Maps but not on all server it seems @heinrich5991

19:24

but not sure

@heinrich5991 you can roll back to f23475665f6cff947d1cdafdf73f77c18396892c I guess

19:27

It's probably @Deleted User's C++ patch, not the startline thing

19:27

spawning random entities sounds insane for your patch

I think so, too. but can never know in C++

what map can i test it in?

some servers were running ubsan, this is definitely a ub issue

@Deleted User check Nostalgia

@Deleted User I highly doubt it happens consistently

nostalgia is consistent according to Patiga

ah perfect, ok we rollback and we can do asan + ubsan to check it out on nostalgia

doing rollback rn

oh or maybe 74ca331e

19:29

the gamecontext destructor stuff

where is the entity in nostalgia

[9:26 PM] Knuski: it looks like it only appears after a map change

19:30

it's probably 74ca331e

near teleto 1

ah

why not both

map change yeah

19:30

that makes more sense

2a74ca331e sorry

@Learath2 is the commit I built bad?

No

f23475665f6cff947d1cdafdf73f77c18396892c

19:31

good

19:31

then I'm deployingh

I'm definitely adding the commit the game was played on to records and saves

all the random tiles that are getting spawned into other maps are from this map

Image attachment

19:33

watch your time by welf

19:33

i saw the smiley bottom left on each server

19:33

and other patterns

that's extremely weird, I hope it was happening when you switch from watch your time and not everywhere

i dont know i could go on any server, switch and those things were there

19:35

so not only servers with watch ur time are affected

19:36

but it still choses the switch tiles of that map

@heinrich5991 take a note of the time when you finish restarting servers, we'll have to nuke the records between deployment and rollback

reload only affected servers :/ could finish voidof* normally

maybe if the teehistorian file contains the server commit hash I can try to only remove affected records

[ddnet/ddnet] Pull request opened: #3962 Go back to constructor destructor

checked runtime with valgrind, found one uninitialized value

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[x] Tested the change with [ASan+UBSan or valgrind's memchec...

i guess saves and all will also be needed to be deleted then if they arent already included

@Learath2 ^

19:51

and @heinrich5991 that should fix it

@Deleted User have you actually checked that it was that commit though?

it must be

19:52

it was bcs the user declared contructors initialize values, instead of putting it in oninit and onshutdown and i forgot to check about that

19:52

it also fixed it locally for me

delete my "what" save on getspeed

19:54

and record too xd

@Learath2 it does

19:57

also saves also contain the gameid

[ddnet/ddnet] Issue opened: #3963 Log git revision hash to stdout

Makes it easier to find the git revision hash of running servers.

19:59

[ddnet/ddnet] Issue opened: #3964 Pressing Home or End in server address field changes the selected server, overriding the field

20:04

[ddnet:master] 2 new commits

c10a74c Go back to constructor destructor - Jupeyy 8c86336 Merge #3962 - bors[bot]

20:04

[ddnet/ddnet] Pull request closed: #3962 Go back to constructor destructor

@Learath2 okay, with this PR in, next try?

20:07

I'm not sweating enough as is

eh, I'm fine with trying it again

I'll try to restore the gamecontext.cpp file to something similar as before before trying again

@Deleted User possibly a draw order bug? (only noticable on tall skins)

Image attachment

@heinrich5991 so you are not confident about jupeyyys pr?

I'd prefer not to have this again

Yeah I wasn't really confident with it when deen merged it the first time around either, idk just looked a tad odd to me

Do we want to make an announcement on running servers that the ranks and saves will not be saved?

might be a good idea

Ravie

@Deleted User possibly a draw order bug? (only noticable on tall skins)

can u hint me, i'm blind

looks like the next "team stripe" goes over the previous tee's feet a bit

scoreboard?

yeah in tab

yeah the tees are generally rendered too big in the scoreboard

and I guess on full server so it's the narrow version?

and yes its probably the order of draw commands

i think you would have to draw all team stripes first before drawing any tees

20:34

because no matter the individual order you're going to clip the top or bottom of tees

20:35

this method appears to be drawing the team stripe, then the tee, then the next pair below it (edited)

they shouldnt leave the stripes

20:35

then its fine

it's only noticable on really tall skins which there aren't many so adjusting the order should be good enough

but then skins can still overlap

20:40

also not really nice

yeah but only in rare cases and still looks not as buggy

20:41

maybe making them even smaller is not worth it

Deleted User

they shouldnt leave the stripes

just removing the color background entirely?

20:42

i don't see why we can't paint all the backgrounds then all the tees

you can

20:42

im too lazy

lol

@Learath2 last server dead

okay, I read Jupeyy's patch now and am reasonable certain that it won't break anything now

cant u just test them on a testing server

I don't think I would have caught this bug by testing tbh

no money to open test servers @louis :p, and no profit

louis

cant u just test them on a testing server

look at RC, 2 ppl use them

22:22

the devs are pretty much alone with testing most of the time

me and who else monkalaugh

monkalaugh

hussain xd

I'm also on RC

What i personally like, but e.g. deen dislikes is to deploy it to a certain % of players for server it would just restart a certain % of servers then e.g. google's play console has a tool for this

ye or just have a dedicated group of ppl that rigorously test but its a small game and not many ppl would volunteer to do that

yeah :c

louis

ye or just have a dedicated group of ppl that rigorously test but its a small game and not many ppl would volunteer to do that

ill volunteer

but keep in mind, e.g. heinrichs pr requires fast testing, maybe max of 1 day so in that time you'd need to be here and test everything and that can be time consuming, boring stressy But generally defs a possibility

22:36

for other prs there could be a nightly server, but maybe unranked, or any better idea?

nightly server is a good idea + a role for dev testing

22:36

then everyone can join and test

but anyway thats a decision by the admins

@louis @lynn @Deleted User follow me for "nightly server"

22:41

(ingame)

22:41

no one coming? ^^

a

22:42

u lied

no lies @lynn

22:43

louis joined me

i didnt see u at first

was making food xD

Did some new Method for filtering Chat got implemented with the latest server upgrade?

explain

im pretty sure the n word wasnt converted to ** before

22:55

not saying thats bad tho lol

22:55

or wait

heinrich dm pro

22:56

or im too laggy

he's oldschool player

22:57

they always good in vanilla

oh its china only i see

22:58

then it can be older nvm

yeah

22:58

government

@heinrich5991 later today or if it's too late this weekend, should do dm on USA

Deleted User

look at RC, 2 ppl use them

use what

release candidate

ah xd

@heinrich5991 apparently the startline thing broke for someone

ye check up #bugs

can you tell me what the symptoms are @louis? which map?

https://discord.com/channels/252358080522747904/757720336274948198/867836734166007850

^should be fixed already

its not the government its ddnet censoring it for them

their servers, their laws i guess

why there is no free vaccine on rus servers then

23:11

our government forcing it xd

xD wow

23:12

come to germany and i give u vaccine

xd

@heinrich5991 m_TeeStarted is not always set

23:17

to true

23:19

const int PlayerDDRaceState = pChr->m_DDRaceState;
    // start
    if(((m_TileIndex == TILE_START) || (m_TileFIndex == TILE_START) || FTile1 == TILE_START || FTile2 == TILE_START || FTile3 == TILE_START || FTile4 == TILE_START || Tile1 == TILE_START || Tile2 == TILE_START || Tile3 == TILE_START || Tile4 == TILE_START) && (PlayerDDRaceState == DDRACE_NONE || PlayerDDRaceState == DDRACE_FINISHED || (PlayerDDRaceState == DDRACE_STARTED && !GetPlayerTeam(ClientID) && g_Config.m_SvTeam != 3)))

23:19

PlayerDDRaceState is started already

23:19

DDRace.cpp

[ddnet/ddnet] Pull request opened: #3965 Fix save codes (symptom: always jumping after load)

"\t" was missing for one variable.

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 if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/...

Exported 551 message(s)