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 2024-07-04 00:00 and 2024-07-05 00:00

<ChillerDragon> @jxsl13 all good now right?

00:00

<ChillerDragon> then ill go find some coffee

no magic today

<ChillerDragon> arent u tired jxsl?

00:01

<ChillerDragon> what time is it? 1 am?

2am

<ChillerDragon> woah

importing works just fine

00:09

https://pkg.go.dev/github.com/teeworlds-go/huffman/v2

what are you making teeworlds-go for? just the lulz?

saw chiller doing it, wanted to help a little bit

00:26

found it cool

<ChillerDragon> @Learath2 im on some mission to spread teeworlds to as many languages as i can before i rq coding

00:36

<ChillerDragon> so ppl can build tw projects in their fav language

Noble quest

<ChillerDragon> do i see that correctly? is the client really storing empty snapshots? xd

00:39

<ChillerDragon> maybe after the 100 robsti refactors ddnet isnt im using tw 0.7

who wants to play some teeworlds of any kind

00:40

bored rn

00:40

#developer lfg

<ChillerDragon> but from my dbg prints it looks like it uses the m_aSnapshotIncomingData buffer as payload for snap empty and stores it xd

00:40

<ChillerDragon> https://zillyhuhn.com/cs/.1720053493.png

Embedded image

00:42

<ChillerDragon> i wonder if the server creates deltas against empty snapshots

00:43

<ChillerDragon> i somehow do not want to store empty snapshots with some old payload xd

00:44

<ChillerDragon> also is it weird that the hexdump kinda makes sense to me xd i stared so long at debug prints i can read most (non huffman compressed) tw traffic as hex xd

<ChillerDragon> Nvm it doesnt use old data. That was a bug in the teeworlds engine xd. It uses the empty snapshot as incoming delta and then does store basically the old snapshot.

01:04

<ChillerDragon> so its wasting a bit of storage imo but at least its not doing anything weird

<ChillerDragon> https://zillyhuhn.com/cs/.1720059289.png

Embedded image

02:15

<ChillerDragon> is that a bug?

i am hitting this assert in my custom fork quite often and its pretty annoying https://github.com/ddnet/ddnet/blob/610381d29e560ec64f2f42f6be9fdddc7f6df0f9/src/engine/shared/http.cpp#L421 am i doing something wrong? its either me having network hiccups or the backend dieing. Either way i dont want my client to crash.

ddnet/src/engine/shared/http.cpp at 610381d29e560ec64f2f42f6be9fddd...

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

04:19

not sure if im just using the every changing http code incorrectly https://github.com/chillerbot/chillerbot-ux/blob/a39c3f65f65c4d90c377d68bf893526e2a3ed48f/src/game/client/components/chillerbot/chillerbotux.cpp#L43-L59

chillerbot-ux/src/game/client/components/chillerbot/chillerbotux.cp...

chillerbot-ux UserExperience based on DDraceNetwork, a mod of Teeworlds - chillerbot/chillerbot-ux

Ok got it https://github.com/chillerbot/chillerbot-ux/commit/3204a95fd7e137151db0ddbf863692a784265023 i guess i was doing the state handling wrong

Fix crash when backend is down · chillerbot/chillerbot-ux@3204a95

<ChillerDragon> so much chat.zillyhuhn.com downtime these days yikes. Im bakk!

04:41

<ChillerDragon> https://zillyhuhn.com/cs/.1720068089.png

zhn

@fokkonaut hey did you find the way around targetx/y zoom problem?

Wdym?

gm

gm

fokkonaut

Wdym?

TargetXY scales with the zoom nowadays which is very odd since you can hit less angles when zoomed in due to the scaling of integers. So when you zoom in/out your aim changes

Ah, yea that thing

05:33

No, I haven't looked into that since heinrich seems unbothered.

<ChillerDragon> @Teero holy shit that sounds bad

06:20

<ChillerDragon> so if i aim lock some nasty edge and then do a quick zooz i might fail the hook?

yes

ddnet fps on my setup sucks so bad

07:08

400-500 fps but feels very choppy

ws-client

<ChillerDragon> @Teero holy shit that sounds bad

There's an issue already

theoretically vsync is disabled and the game is running in real full-screen mode but it doesn't feel that way (edited)

https://github.com/ddnet/ddnet/issues/8452

Don't send mouse target in zoomed state · Issue #8452 · ddnet/ddnet

Precision has been changed due to this. We should revert this change where the mouse position gets sent in a zoomed state. A fix would be to keep the normal mouse pos for everything physics related...

i dunno how but i made this by shooting once in tp thingy for grenade justatest

justatest

Image attachment

MilkeeyCat

i dunno how but i made this by shooting once in tp thingy for grenade justatest

justatest

Duplicating entities?

07:41

you only shot once?

yes

weird

but it stops when i look away from tp thingy

07:42

even though i didn't hold shoot

07:42

thonk

07:45

MilkeeyCat

but it stops when i look away from tp thingy

antiping? probably the client trying to predict which teleporter next

fokkonaut

antiping? probably the client trying to predict which teleporter next

same happens without antiping

Classic duplication glitch. Back in 2018 we used to do that with carpets … oh wait wrong game

chillerdragon: you can simply keep an empty init function

08:55

so that there's no breaking change

MilkeeyCat

Click to see attachment 🖼️

I just held shoot button before pressing spec and it was shooting when I was in spec pepeW

pepeW

Does other player see the same thing

09:49

would be pretty crazy if we really have a duplication glitch which seems pretty unlikely

fokkonaut

https://github.com/ddnet/ddnet/issues/8452

why did anyone want to scale it with zoom as multiplier

having it as an unscaled world position is nice for some things

10:30

the server knows where the client points to

Failed to bridge media: attachment too large (48141924 > 20000000) Yum.

Replying to @MilkeeyCat Failed to bridge media: attachment too large (48141924 > 20000000)

[ddnet/ddnet] Pull request opened: #8558 Fix integer overflow when computing tilemap size

Cherry-picked from https://github.com/teeworlds/teeworlds/pull/2076/commits/d25869626a8cfbdd320929ba93ce73abed1402ce Thanks to Charly Reux for notifying us.

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
...

10:33

[ddnet:18.3.1] 25 new commits

f429ede Version 18.3 - def- e694df6 Update credits - def- bbe8722 Fix wrong context value being used for translation lookup - Robyt3 a570243 Update Ukrainian l10n (18.3) - EGYT5453 da5e551 Update russian.txt - ByFox213 b101311 Update russian.txt - ByFox213 9783d8c Update russian.txt - ByFox213 46cdd6d Update russian.txt - ByFox213 6b495a8 Update Swedish translations for 18.3 - furo321 e669803 new french translation - Emilcha 5ff306a merge into translation file + credits - Emilcha 30171b8 Changement - Emilcha bf636dc Consistance - Emilcha 401d9a5 Update Turkish translations - GokturkTalha 8f70908 Update Azerbaijanese translation - GokturkTalha 76feb50 Fix percentage not being used for console command progress spinner - Robyt3 7301fff update czech translations for 18.3 - dobrykafe df95321 update slovak translations for 18.3 - dobrykafe 619cc64 Update simplified_chinese.txt - By622 2900550 Update traditional_chinese.txt - By622 74972ab Fix demo seekbar tooltip not showing correct hovered time - Robyt3 23fa51f Update brazilian_portuguese.txt - rffontenelle 97d9384 Fix tiles being handled in same tick for dead tees in some cases - Robyt3 46efd96 Fix integer overflow when computing tilemap size - def- 7b0f30f Version 18.3.1 - def-

10:33

[ddnet:master] 1 new commit

3806f32 Version 18.3.1 - def-

Hmmm

[ddnet:18.3.1] 2 new commits

c1f8ead Fix integer overflow when computing tilemap size - def- d947280 Version 18.3.1 - def-

Do I have something to do now?

Nope

Not even updating the translation? (edited)

no, it's just a bug fix release

10:41

no translation changes

Okaay

https://github.com/teeworlds/teeworlds/issues/2071

10:42

was thought to be fixed 5 years ago. but it wasn't ^^

heinrich5991

the server knows where the client points to

is there any chance to revert this change btw?

there's a chance to fix the bug

10:44

but I think it's good that the server knows where the client points to

which one?

that your angle depends on the zoom level

heinrich5991

but I think it's good that the server knows where the client points to

couldn't we just add zoom multiplier as extended attribute in the past xd

why should the server do the calculation there?

10:46

it seems much more well-defined for the client to tell the server where exactly it is pointing

why would you need clients sending two different type of targetxy?

up-to-date clients only send one type of targetxy, no?

yeah but what about other clients huh

you can recognize them based on the version number, if you need to

anyway its strange to change players input behaviour that old servers can rely on

is there a server that relied on that?

i bet there was

10:49

at least fokkonaut was mad when this came out

I'd think the situation improved with this change

10:49

no longer needing to rely on brittle hacks to determine the actual cursor position

when will 0.7 support be dropped

no plans. why do you ask? @ReiTW

cuz there's no reason to still support a dead game

heinrich5991

no longer needing to rely on brittle hacks to determine the actual cursor position

so m_Pos + m_Target is hacky?

ReiTW

cuz there's no reason to still support a dead game

please take your flamebait elsewhere

and m_Zoom multiplier would do the trick, ddnet wouldn't accept "player input changes" if it would break (ddnet) servers behaviour but this one just got through #7512 (edited)

ig there might be features that anyone would love to be implemented that requires breaking the compat with 0.7 ?

ReiTW

ig there might be features that anyone would love to be implemented that requires breaking the compat with 0.7 ?

let's talk about it once we get that hypothetical feature idea

heinrich5991

please take your flamebait elsewhere

not really flamebaits but facts, anyone knows how dead 0.7 is sadly

this discussion isn't interesting, we've had it too many times

ReiTW

not really flamebaits but facts, anyone knows how dead 0.7 is sadly

vanilla players try to revive it though

10:52

by playing only on aeon servers and refusing to enter bridge ones xd

zhn

vanilla players try to revive it though

there's prob less than 100 or even 50 players

please take this discussion to #off-topic if you want to continue it

10:53

(unless you have new information, which I doubt)

heinrich5991

(unless you have new information, which I doubt)

I don't plan to continue, but once there was mentioned stuff like a skeleton for people who wants to use ddnet base for custom mods. without 0.7 ddnet could have easily worked on smth to have server-side content (idk like custom guns, etc..) allowing more cool mods to be developed by the community. or even when you've all talked about the networking improvement part (to allow more efficient protections to be developed around). even if Davide has a workaround the most annoying thing that makes it more difficult to improve is the support to legacy clients

ddnet could have easily worked on smth to have server-side content (idk like custom guns, etc..) allowing more cool mods to be developed by the community.

sounds cool, let's do that

even if Davide has a workaround the most annoying thing that makes it more difficult to improve is the support to legacy clients

I don't even know if Davide has a workaround for 0.7 clients or not. but if not, that'd be fine

11:03

I'm not sure what you're talking about. I don't think we had a feature declined because it couldn't work with 0.7

recently you've talked about having tcp first for conn handling

we can drop 0.7 support if it stands in the way of improvement

11:03

yes, let's do that

Is there really any benefit to that. Maybe it'll help with some firewall implementations. Do we even know if any of our hosters do it?

github-merge-queue[bot]

[ddnet:master] 2 new commits

4297772 Fix integer overflow when computing tilemap size - def- 447b44d Merge pull request #8558 from def-/pr-int-overflow - heinrich5991

11:10

[ddnet/ddnet] Pull request closed: #8558 Fix integer overflow when computing tilemap size

Learath2

Is there really any benefit to that. Maybe it'll help with some firewall implementations. Do we even know if any of our hosters do it?

davide said it'd help him tremendously

So we just accept tcp connections then close them? Or do we need it alive?

just open a tcp connection and that's it. we can close it immediately

11:27

like to https://ger10.ddnet.org/

Oh doesn't have to be the exact port?

no, just somewhere

11:28

I'd just fire and forget a https request

[ddnet-web:master] 1 new commit

cafbf00 Version 18.3.1 - def-

I think we even have an issue for it (edited)

Perhaps we can also make this more generic? Allow servers to advertise a verify link

https://github.com/ddnet/ddnet/issues/8210

heinrich5991

https://github.com/ddnet/ddnet/issues/8210

This issue says on the same port

[ddnet-scripts:master] 1 new commit

0d7c0b8 No more config_directory.sh - def-

Learath2

This issue says on the same port

@Davide does the TCP handshake need to be on the same port number as the UDP port number?

I don't think so, TCP and UDP ports are not related in any way, you can just have each server announce it's tcp-handshake-ip-port-combo

I would also not think so, but I'd also not have guessed that a TCP handshake would help at all

When I first heard it I assumed it's some bizarre firewall behaviour

heinrich5991

@Davide does the TCP handshake need to be on the same port number as the UDP port number?

No

Learath2

This issue says on the same port

fixed it

You talkin go Huffman? Well the api changed from an unitialized raw go struct to the weird go styled fake constructor new function. Since I have no users I don’t mind to do breaking changes anyways xd

Replying to @heinrich5991 chillerdragon: you can simply keep an empty init function

chillerdragon: the rules also apply to you. go to #off-topic for flamebait topics

Yo btw when bridge off-topic?

I thought it already was bridged

only to matrix

12:14

he's requesting IRC

ah, probably never to irc

Matrix cring

not on my priority list right now

12:14

although I agree that having an IRC bridge is kinda cool

When leak priority list

the bridges are literally only being used by 1 or 2 people for legitimate purposes and is only a tool to dodge moderation on discord

why

Replying to chillerdragon Matrix cring

Yes 50% of those 1 or 2 ppl is me and I am a big supporter of no moderation @Learath2

its really cool

Bugged and bloated. IRC is just so much smoother

Replying to Souly why

irc is the very best

dont tell me you hate matrix bots

Learath2

the bridges are literally only being used by 1 or 2 people for legitimate purposes and is only a tool to dodge moderation on discord

good point. but I like having at least one non-proprietrary method to access our chat. I guess matrix is good enough for that

thats the best part

When non proprietary map submissions

12:18

Matrix.org instance is basically proprietary xd

ReiTW

cuz there's no reason to still support a dead game

why support a redefined and at the same time, slightly abandoned engine? (edited)

I would setup a selfhosted matrix instance but I just don’t like it enough

chillerdragon: ddnet matrix isn't hosted on matrix.org

Well I am using matrix.org xd

that's on you

Sent from element on iOS

12:19

Yes on me

12:20

@Cellegen: wat engine xd

0.7

Yea it’s frustrating Oy is so inactive lately

so inactive "lately"

"lately"

this isn't a change

lately

He used to be more active imo

brother I havent seen him since 0.7.4

bro went missing for over a year

https://en.wiktionary.org/wiki/lately

gigachad

2

12:22

"recently" "not too long ago"

Yea I don’t recall ever waiting this long for oy return

XDDD

4 years ago is still not too long ago right?

Then another question, why wait for Oy?

12:22

we have deen right here owo

owo

12:22

and heinrich

because no one can be bothered to maintain teeworlds, including oy

12:23

or chillerdragon

deen and Heinrich don’t merge my 0.7 pr

or chiller ye

what 0.7 pr

vote chiller for next engine creator 2k24

you mean 0.7 features in ddnet?

0.7 support for ddnet client

chillerdragon

He used to be more active imo

He had a sudden burst of activity there for a sec

Which sec?

12:24

You mean last time when he looked at 30% of the prs?

I don't remember exact dates but I put in some work into editor2 and added the chat command menu thing

oh cool, how would it work? (nvm) (edited)

for something to change, you need an active maintainer

Bru that was like 2016

if there's no one, there's no point in debating this

12:25

do you volunteer to maintain teeworlds? chillerdragon?

Oy could assign some

we would need a person first

No

Replying to @heinrich5991 do you volunteer to maintain teeworlds? chillerdragon?

find a volunteer and if I think they're even half-qualified I'll help you bring teeworlds back alive

You guys need someone who knows the engine inside out and is willing to change certain ways the engine could function

robsti

Obv the 2 creators have different ideas in mind, they would rather just redo the engine from scratch

chillerdragon

robsti

have you asked whether he wants to maintain teeworlds, chillerdragon?

12:27

this needs to be a volunteer

12:27

not someone you assign

I don’t recall I think we had that discussion before so I guess we don’t have anyone

well thats fucked

see. there's the problem. try to fix that

No need for new teeworlds stuff anyways imo

how so? to improve the way the engine could work, it needs to be done

The game works well

chillerdragon: if you mention porting ddnet to 0.7 without any new content one more time, I'm going to ban you for flamebiat

Yes it does work, however it doesn't support things which can make it out of this era (0.6)

12:30

0.7 ain't some revolutionary aspect, where we can freely mod things or get more gamemodes without limitations

12:30

The engine is no exception for this and modders or server devs have a harder time developing their own ideas without reinventing the wheel

0.7 isn’t about revolution or features it’s just keeping up with compatibility

I swear to god, if I waited 6 years for a not revolutionary client,

12:31

then it deserves not to be called a newer "teeworlds"

Are you actively waiting?

12:31

Yea the version breaking changes are a bit unfortunate

Still, yes, but not from ddnet developers naturally

porting ddnet to 0.7 in the current state of 0.7 is never going to happen

12:32

I can guarantee you that. no current maintainer wants that

12:32

it has been discussed with no ends

Im seriously interested in a approximate priority list of yours @heinrich5991

I plan to make a new map editor, another is making a 2d to 3d render on their own engine using server data, one still tries to make a fully functional rust compatibility (edited)

Where can I vote? Xd

chillerdragon: I'm currently extracting the compatibility code of ddnet

look at this PR I just made at work, negative line count, still adds an entire feature

Image attachment

even though my number one priority should be implementing QUIC

12:33

and then accounts

I’d like to give libtw2 snaps, ddnet proxy, ddnet ip redirects, #7777 some priority votes hehe

heinrich5991

chillerdragon: I'm currently extracting the compatibility code of ddnet

this is honestly not worth your time imo, the other two are far more important, we worked around the compatibility code for a decade now, we can handle a couple more years

I told you to proxy support instead -.-

Replying to @heinrich5991 chillerdragon: I'm currently extracting the compatibility code of ddnet

Learath2

this is honestly not worth your time imo, the other two are far more important, we worked around the compatibility code for a decade now, we can handle a couple more years

I know, but we're currently getting more compat code which I don't like. on two fronts, even

add another decade to the list boys

also, this could be reused should we want to integrate it into Jupstar's ddnet

10x developer

Replying to @Learath2 look at this PR I just made at work, negative line count, still adds an …

chillerdragon: I'm accepting 20€/h to temporarily change my priority list

btw, how are you planning on doing it? 0.7 and 128p support all require pretty intimate knowledge of the gamestate, will you be hooking in for that or will you be trying to recreate the state on a separate module?

Xd

12:36

Im broke

Learath2

btw, how are you planning on doing it? 0.7 and 128p support all require pretty intimate knowledge of the gamestate, will you be hooking in for that or will you be trying to recreate the state on a separate module?

separate module

But nice

wouldn't that in theory kinda double the work? That's why I avoided it, packing stuff up into packets, then unpacking and trying to figure out what state led to that packet sounded like such a waste of our cpu

Learath2

wouldn't that in theory kinda double the work? That's why I avoided it, packing stuff up into packets, then unpacking and trying to figure out what state led to that packet sounded like such a waste of our cpu

yes, it'll double the work (edited)

12:38

but it's the only clean solution I see without introducing lots of work on maintenance

Have you considered more precise hooks where the module can directly read into the gamestate?

optimizations can be done later. I'd prefer to start from a clean slate at first

but I feel most of the work you'll be doing can be avoided if we can cleanly use the real state instead of trying to recreate it

I feel like this way my work can more easily be reused

I mean it's your project, feel free to go about it however you like. I just attempted this before and it really annoyed me

because it's just like 5 hooks or so, all straightforward

heinrich5991

optimizations can be done later. I'd prefer to start from a clean slate at first

Sounds like rust it is, no?

it's indeed rust

poggers2

https://cxx.rs/ is actually nice to work with

CXX — safe interop between Rust and C++

Never used tho

12:42

But I thought aswell of rewriting some of our gRPC/protobuf stuff into rust in ddnet source itself

12:42

Currently its cpp

it's already integrated into ddnet

12:42

that made it easy to start

Yeah I've seen that

Yea it really is easy to use

But I havent had time to do so, still working on other stuff like new website & dnsbl

12:43

I will aswell open source the dnsbl server & api on KoG Github

12:43

Maybe someone with rust experience wants to contribute then

Kog GitHub wot

12:44

URL or didn’t happen

I have one thing I would contribute to your dnsbl implementation

Learath2

I have one thing I would contribute to your dnsbl implementation

In the DNS answer itself?

Yes, technically you are supposed to be returning an NXDOMAIN, you aren't

I can change that if you need

It's not important, it's just something I noticed

Learath2

It's not important, it's just something I noticed

But we are not handling it, no?

12:47

DDNet Source only checks if there is an A record set

12:47

or empty

I added a small transparent proxy that allows us to whitelist some ips. While implementing that I assumed you would be sending an NXDOMAIN, took me a couple minutes to figure out what was going on

Oh

12:48

Now I am returning NXDOMAIN

lol

Avolicious

Now I am returning NXDOMAIN

hopefully I handled that properly

hi @Pimble

Learath2

hopefully I handled that properly

Tbh I wanted to contribute to DDNet dnsbl for a long time

12:50

But had no time to writeup my changes & why

12:50

The server-side implementation

You could also create an issue for it and maybe one of us would pick it up

12:51

What are we lacking?

Adding a more fine-grained control for dnsbl returns

12:52

https://www.spamhaus.org/resource-hub/dnsbl/using-our-public-mirrors-check-your-return-codes-now./

DNSBL | Using Spamhaus' legacy DNSBLs? Check your return codes now

12:52

Big DNSBL provideers like Spamhaus, they add specific 127.xxx.xxx.xxx return codes in A records to give more control in mailserver setups for example

I think they also have TXT records right?

So currently I only return the reversed ip, which is in fact compliant. But instead I want to return specific A addresses, so you can say Okay I want to block only cloud providers, but universities I want to allow

Learath2

I think they also have TXT records right?

You can add status messages to TXT, but A is used for ips

I mean to add something like a reason. I remember it being part of "standard"

Mhmmm, ad hoc I cant remember any standard for TXT status reasons

12:55

But I can add a reason if needed

12:55

But there aswell, where do I get the reasons from?

Well idk, it is for if you want to include more information than you can convey in an ip address. The A Record is pretty much allowed to point to any ip so you can use that as a reason/category too

Learath2

Well idk, it is for if you want to include more information than you can convey in an ip address. The A Record is pretty much allowed to point to any ip so you can use that as a reason/category too

Back in the days, the standard was to use 127.0.0.0/8

Avolicious

Mhmmm, ad hoc I cant remember any standard for TXT status reasons

RFC5782 2.1 paragraph 2 btw. It's technically a SHOULD even

This RFC was published on the Internet Research Task Force (IRTF) stream. This RFC is not endorsed by the IETF and has no formal standing in the IETF standards process.

12:59

13:00

But yeah, I can add TXT aswell

13:00

We already support IPv6, but its not used yet

13:01

If we support TXT records, shall we output it to the console?

13:01

But this could get spammy, no?

Richard Stallman is mad at us

Replying to chillerdragon When non proprietary map submissions

Hm, it might be fine depending on the amount of blacklisted ips we encounter

its a few millions a month

Guess what I am using

Replying to @heinrich5991 that's on you

it feels good when u fix a segfault

13:23

(yes i get segfaults with rust, but cuz doing unsafe stuff with llvm compiled programs)

Ryozuki

it feels good when u fix a segfault

you should write C then

well i would better say, its the compiled program that triggers a segfault not rust

13:23

so i fixed a compiler bug

MilkeeyCat

you should write C then

nein

Ryozuki

nein

my language will be advertised to you as C sprinkled with some convenient shit

MilkeeyCat

my language will be advertised to you as C sprinkled with some convenient shit

does it use llvm as backend

zhn

does it use llvm as backend

n..no

gg ryo will not use it

13:33

wait why did you choose to make it translating into nasm and then compiling (edited)

zhn

wait why did you choose to make it translating into nasm and then compiling (edited)

how did I have to do it?

using llvm as backend and something like bison as lexer and so on xd

13:35

llvm would give you a lot of out of the box optimizations and you could concentrate yourself into design of language

i wanted to make all the stuff myself

using nasm is not doing all urself

13:36

but its good tho xd

i can make generator which will be spiting out zeroes and ones

@MilkeeyCat nasm is not made to be used as a library but

13:38

i have a suggestion that is

13:38

https://www.keystone-engine.org/

The Ultimate Assembler

13:38

it looks rly good

13:38

and it can be used as a library

13:38

https://www.keystone-engine.org/docs/tutorial.html

Programming with C & Python

The Ultimate Assembler

13:38

The following sample code presents how to compile 32-bit assembly instructions of X86 in C language.

Ryozuki

@MilkeeyCat nasm is not made to be used as a library but

i can use it however i want

ok

Ryozuki

using nasm is not doing all urself

Generating machine code by hand is just a little too much unnecessary work

was just saying

Learath2

Generating machine code by hand is just a little too much unnecessary work

yeah but he probs uses nasm in a hacky way

13:40

like making the file etc

13:40

i think keystone is made more to fit what he wants

13:40

 /* test1.c */
  #include <stdio.h>
  #include <keystone/keystone.h>
  
  // separate assembly instructions by ; or \n
  #define CODE "INC ecx; DEC edx"
  
  int main(int argc, char **argv)
  {
      ks_engine *ks;
      ks_err err;
      size_t count;
      unsigned char *encode;
      size_t size;
  
      err = ks_open(KS_ARCH_X86, KS_MODE_32, &ks);
      if (err != KS_ERR_OK) {
          printf("ERROR: failed on ks_open(), quit\n");
          return -1;
      }
  
      if (ks_asm(ks, CODE, 0, &encode, &size, &count) != KS_ERR_OK) {
          printf("ERROR: ks_asm() failed & count = %lu, error = %u\n",
                 count, ks_errno(ks));
      } else {
          size_t i;
  
          printf("%s = ", CODE);
          for (i = 0; i < size; i++) {
              printf("%02x ", encode[i]);
          }
          printf("\n");
          printf("Compiled: %lu bytes, statements: %lu\n", size, count);
      }
  
      // NOTE: free encode after usage to avoid leaking memory
      ks_free(encode);
  
      // close Keystone instance when done
      ks_close(ks);
  
      return 0;
  }

heinrich5991

no longer needing to rely on brittle hacks to determine the actual cursor position

yeah but now i can't map players cursor to their screen surface xd

what's the use case for that?

ingame interface using motd/broadcasts

ah

ye that's why i was yapping xd

@zhn so no interface for blockworlds™️ ? :(

14:05

couldnt follow chat, was working

in case of broadcasts i can use m_Direction but m_Target is necessary for working motd ui

what kind of interface are you building?

14:06

a selection, like a list box? (edited)

yep

perhaps you could add a proper UI to ddnet instead?

14:07

i.e. have the server send a list box, let the client display the UI

wanted to support both 0.6 and 0.7 but yeah i think its a proper way of implementing an actual interface

you can still hack your way around the 0.7 client like you planned to

correct

14:08

i'll drop it anyway xddd

14:08

if ddnet will do so

catxplosion

heinrich5991

I can guarantee you that. no current maintainer wants that

makes sense

heinrich5991

chillerdragon: I'm currently extracting the compatibility code of ddnet

chillerdragon wanting 0.7 port while heinrich making 0.7 compatibility code more maintainable. That's more or less the closest one can get to having 0.7 support.

15:03

Took me back to 2012

Trollouaa

2

probably from 2012 actually

15:45

kek

[ddnet/ddnet] Issue closed: #8165 Add new net message for message box

zhn

ingame interface using motd/broadcasts

I'd support work on #2875 but should probably be discussed first which use cases we want to support

https://github.com/ddnet/ddnet/issues/2875

Custom text window(s) for interactive help for mods · Issue #2875 ·...

Problem: complex mods like infClass have a steep learning curve and a new player have a hard time to get game mechanics and even the basics. Currently most of players leave because they don't g...

Robyt3

I'd support work on #2875 but should probably be discussed first which use cases we want to support

yeah its actually mind blowing amount of cases you can think of

15:56

currently im adding server side quad render request and textbox render request with ids so client can send net message with id and action

15:56

small poc

zhn

yeah its actually mind blowing amount of cases you can think of

yeah, using any of our basic UI elements could be possible (text input, checkbox, scrollbar, number input, color input etc.)

You'll need to figure out a wireformat to describe a ui aswell, it's a very annoying and tough problem

oh yeah

probably would have to differentiate between messages that are places on the map or non-interactive, and interactive UI forms, since you can only have one active form/popup at a time (edited)

about ingame quads: what do you think of map-patch net object which is actually map with deltas to change "overlaying" server-side managed map

15:59

that will allow us to reuse map render and it will have less impact on number of netex messages

16:00

so ui net code will come as completely another part

I actually can't imagine anything simple that would allow laying out ui over the wire. Maybe a small bytecode that is intentionally not turing complete with commands like vsplit hsplit dotextbox docheckbox

justatest

1

Do you really need to layout UI manually? I would think a linear layout of rows of UI elements would be enough

ye sounds better than exploiting netexes xd

Robyt3

Do you really need to layout UI manually? I would think a linear layout of rows of UI elements would be enough

manual ui will allow server modders make all kind of things and so its better imo

You block traffic from millions of ips a month based on some list of known bad IPs? Are you sure there are no false positives? What kind of traffic is that even? Actual players with a real client?

Replying to @Avolicious its a few millions a month

Using json to describe more complex UI layout seems possible justatest

justatest

Would the maximum packet size be a concern then?

vsplit 2 4
    dotext "Login" center
end
    hsplit 1 3 1
    end
        vsplit 1 1 1 1 1
            hsplit 1 2
                dotext "Username:" left
            end
                dotextbox <username>
            end
        end
            hsplit 1 2
                dotext "Password:" left
            end
                dotextbox <password>
            end
        end
        end
            dobutton "Login"
        end
        end
    end
    end
end

(edited)

Robyt3

Using json to describe more complex UI layout seems possible justatest

justatest

Would the maximum packet size be a concern then?

not if server/client will translate it to bytecode :P

16:14

does ddnet have some doxygen docs instance available online?

zhn

does ddnet have some doxygen docs instance available online?

https://codedoc.ddnet.org/

1

16:16

It's not very documented though

anyone aware of an issue related to loading points/ranks ?

16:17

because right now it seems like we can't do /points nor check the ranks

hm the db server looks fine this time

rank won't even be displayed when joining a new empty server with a map you finished before

yeah it seems like nothing loads

also indicating finish in browser doesnt work

OH, the db server is not fine

16:24

Jul 04 18:14:16 ddnetdb2 systemd[1]: mariadb.service: Main process exited, code=killed, status=9/KILL
Jul 04 18:14:16 ddnetdb2 systemd[1]: mariadb.service: Failed with result 'oom-kill'.

were you connected via ssh by chance on accident?!

Kek

@deen perhaps you were a tad too generous with innodb_buffer_pool_size

16:32

Also why didn't systemd in it's infinite wisdom restart this service?

systemd can be setup to do so Restart=on-failure

Why do I need to set that up? It never refrained for making decisions for me before

https://www.freedesktop.org/software/systemd/man/latest/systemd.service.html

Restart=

    Configures whether the service shall be restarted when the service process exits, is killed, or a timeout is reached. The service process may be the main service process, but it may also be one of the processes specified with ExecStartPre=, ExecStartPost=, ExecStop=, ExecStopPost=, or ExecReload=. When the death of the process is a result of systemd operation (e.g. service stop or restart), the service will not be restarted. Timeouts include missing the watchdog "keep-alive ping" deadline and a service start, reload, and stop operation timeouts.

    Takes one of no, on-success, on-failure, on-abnormal, on-watchdog, on-abort, or always. If set to no (the default), the service will not be restarted. 

If set to on-failure, the service will be restarted when the process exits with a non-zero exit code, is terminated by a signal (including on core dump, but excluding the aforementioned four signals), when an operation (such as service reload) times out, and when the configured watchdog timeout is triggered. If set to on-abnormal, the service will be restarted when the process is terminated by a signal (including on core dump, excluding the aforementioned four signals), when an operation times out, or when the watchdog timeout is triggered. If set to on-abort, the service will be restarted only if the service process exits due to an uncaught signal not specified as a clean exit status. If set to on-watchdog, the service will be restarted only if the watchdog timeout for the service expires. If set to always, the service will be restarted regardless of whether it exited cleanly or not, got terminated abnormally by a signal, or hit a timeout. Note that Type=oneshot services will never be restarted on a clean exit status, i.e. always and on-success are rejected for them.

(edited)

16:38

i guess its just that restart is off by default, thats why you have to explicitly tell it to

Learath2

@deen perhaps you were a tad too generous with innodb_buffer_pool_size

ok, reducing

@ChillerDragon, any thought on adding two repos: varint and/or econ ? would like to split up twapi if you are fine with having those in your org?

question, is the distance between the map border (white border which you are limited to build in) and the endless death tiles exactly 200 tiles?

bool CEntity::GameLayerClipped(vec2 CheckPos)
{
    return round_to_int(CheckPos.x) / 32 < -200 || round_to_int(CheckPos.x) / 32 > GameServer()->Collision()->GetWidth() + 200 ||
           round_to_int(CheckPos.y) / 32 < -200 || round_to_int(CheckPos.y) / 32 > GameServer()->Collision()->GetHeight() + 200;
}

17:33

yep

ty for confirming owo

owo

https://www.teeworlds.com/forum/viewtopic.php?id=2339

[MOD] real RCON - with gnuTLS security (Page 1) — Fan Art and Other...

[MOD] real RCON - with gnuTLS security (Page 1) — Fan Art and Other — Teeworlds Forum — Everything Teeworlds!

Learath2

Jul 04 18:14:16 ddnetdb2 systemd[1]: mariadb.service: Main process exited, code=killed, status=9/KILL
Jul 04 18:14:16 ddnetdb2 systemd[1]: mariadb.service: Failed with result 'oom-kill'.

LOL

meloƞ

were you connected via ssh by chance on accident?!

systemctl status

Learath2

Why do I need to set that up? It never refrained for making decisions for me before

systemd has some restart options in ur unit .service file, some things would rather be restarted by their own means (like the transmission daemon prefers to be restarted with a specific signal) instead of just killing & spinning the same process up again

19:42

also sorry for double ping if I did that

19:42

didn't mean it

Ewan

systemctl status

pepeangryping

systemctl poweroff

I really need to add decorations to my ssh'd windows before I cause an incident. I almost did rm -rf /boot/* on ddnet.org

pepe_holy

1

Learath2

I really need to add decorations to my ssh'd windows before I cause an incident. I almost did rm -rf /boot/* on ddnet.org

°-°

19:50

use a terminal emulator just for ddnet's ssh sessions

Idk wym, I use a terminal emulator for all my terminals. Who even has physical terminals in 2024?

19:52

Image attachment

ah i wasnt specific enough, my bad :D i meant like, dont use your default emu for ddnet

19:53

(e.g default -> alacritty, ddnet -> gnome's term) or smth xD

I'll try to change the background color when I ssh

19:53

I cba to maintain two different configs

fair

excellent idea

Robyt3

Do you really need to layout UI manually? I would think a linear layout of rows of UI elements would be enough

even that sounds more complicated than what I would go for. a simple dialog with a number of buttons would be my start

heinrich5991

even that sounds more complicated than what I would go for. a simple dialog with a number of buttons would be my start

My only issue with that would be that every half measure we introduce has to be supported forever

1

I think sending the client a list of options and the client selecting one (or escaping from the dialog) is straightforward enough that there's no problem in supporting it forever… I think

[ddnet/ddnet] Issue opened: #8559 RFC: Introduce random map feature on empty servers

By default server has it's set sv_map variable and so newbie player enters the server thinking "oh, sure, I have never played this map, I guess i should try it" in scenario when all empty servers have 3-10 different maps for same region. Mocking empty servers with "Random" map would be a better option because newbie will not know that servers and maps are not tied. On first player join on empty server, server will behave like someone called random_map.

21:28

[ddnet/ddnet] Issue opened: #8560 RFC: Auto "Random map" vote start on all teams/running tees finish

Imagine group of newbies who don't know how to properly change maps. Server should start a "Random map" vote after all of them finish initial map, kind of soft map rotation

Learath2

My only issue with that would be that every half measure we introduce has to be supported forever

a bad "full" measure might be even more annoying to support in the long term

<ChillerDragon> @jxsl13 the import paths for teeworlds-go look cool huh? xd I invited you to https://github.com/teeworlds-go/econ

teeworlds-go/econ

Contribute to teeworlds-go/econ development by creating an account on GitHub.

awesome

indeed they do

<ChillerDragon> its your repo go nuts! push to master or heck rename to main. Use snake_case and import third party libraries xd

D:

<ChillerDragon> idk what one would use varint for

might be interesting outside Go. like the one in the standard library that's used for protobuffers.

23:08

was my first thought

23:09

is fine

<ChillerDragon> outside tw you mean?

yes

23:10

anyway. gotta sleep

<ChillerDragon> ok sure i can create you the repo too but i cant promise i will use it in protocol or something like that

23:10

<ChillerDragon> you want it to be called "varint" ?

yeah

<ChillerDragon> the main thing was a joke btw xd

23:11

<ChillerDragon> pls use master -.- i did a little trap for you

xD

23:11

I will keep it consistent

<ChillerDragon> the setup instructions in varint say master https://github.com/teeworlds-go/varint

teeworlds-go/varint

Contribute to teeworlds-go/varint development by creating an account on GitHub.

I prefer consistency over imaginary people's feelings

<ChillerDragon> the one in econ say main justatest

justatest

23:12

<ChillerDragon> so be careful

xD

23:12

thought you were chillerdragon

23:13

that doesn't look chill to me D:

<ChillerDragon> ?xd

stay calm, keep cool, chill

<ChillerDragon> e sure

we can do that

23:14

worst case, in the second try

23:14

on

<ChillerDragon> or 4th

23:14

<ChillerDragon> like 2.0.0 release

23:14

<ChillerDragon> or was it 5?

I did not count

<ChillerDragon> did you get emails for every rls?

cuz I stayed cool

23:15

my github account is linked to my spam mail account which i don't read

23:16

so

23:16

dunno

23:17

mainly use the fasthub android app

23:17

for checking for updates on followed projects

Xd

gn8

Gn8

jxsl13

cuz I stayed cool

was a trap

Exported 498 message(s)

Timezone: UTC+0