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 2019-09-28 00:00:00Z and 2019-09-29 00:00:00Z

[ddnet-maps:master] 1 new commit

90e8ba6 A balega - ddnet-maps

Embedded image

11:07

xd

11:07

cover last years

https://gist.github.com/Learath2/59e510af2a3475b7863d9477f0296b25

generate_plugin_hooks.py

GitHub Gist: instantly share code, notes, and snippets.

11:16

I worked a little on getting a cleanly integrating plugin system for teeworlds, kinda hit a roadblock with not knowing how to translate the macro call before it runs through the actual compiler preprocessor

1

feelsamazingman

2

11:17

The script above takes in a source file, and generated the required hook variables and the required registering functions

11:17

s#generated#generates#

11:19

I did it like this so as to not limit developers to only c++, but I'm kinda regretting that decision and thinking of doing an abstract class based approach instead

11:22

even after that @Deleted User's words kinda make me feel like not doing a plugin system at all but to make a framework to let people modify the actual source itself, but that I have no good approach for

11:23

either way, libclang is fun to mess around with

xush had this funny approach of injecting code into the binary xD search the correct places and then it would also be version and even sourcecode independent

11:23

https://github.com/GutZuFusss/TW-CDXX

GutZuFusss/TW-CDXX

External teeworlds stuff using memory editing & pattern scanning. - GutZuFusss/TW-CDXX

but not architecture independent

tru

imagine doing all this work to write an aimbot injector...

xxD

:/

he wanted to be able to bot instantly when he returns in a few years to 0.8

If any of you have opinions about how a plugin system should work or look like, please let me know

Can someone have a look into the DDNet file score system and debug it?! I really tried everything, but I cant find why it wouldnt save every rank...

13:04

Also I dont get error messages like failed for reading or something like that

Does not really explain why ranks are lost... but this line might cause horrible lags: https://github.com/ddnet/ddnet7/blob/cb7a6cc1a0f4f1bcfaa5d260810c472916e3f156/src/game/server/score/file_score.cpp#L93 It's waiting 1 second instead of 1 millisecond (like in ddnet 0.6)

DDraceNetwork server modification for Teeworlds 0.7, ported by fokkonaut - ddnet/ddnet7

no

14:53

ah wait

In tw 0.7 this function uses milliseconds by default

yes

14:53

i thought the other way round rightn ow

14:53

but anyways, this is a problem in ddnet 0.6 too (the rank not saving always)

Okay. The whole file scoring should be rewritten

why?

The whole locking there is not optimal. When the score file gets too big and multiple people finish this can cause delays

14:56

And in general... Using a database would be better

of course it would be

14:58

But not everyone is able to do that

14:58

So a file system is good, also ddnet uses this on official servers if the databse connection fails, the scores are written in a file for now

14:58

until an admin then manually merges them

14:58

its not always stupid to have some "old" stuff in there

A few more tiny things These lines leak memory: https://github.com/ddnet/ddnet7/blob/cb7a6cc1a0f4f1bcfaa5d260810c472916e3f156/src/game/server/score/file_score.cpp#L138-L139 https://github.com/ddnet/ddnet7/blob/cb7a6cc1a0f4f1bcfaa5d260810c472916e3f156/src/game/server/score/file_score.cpp#L203

DDraceNetwork server modification for Teeworlds 0.7, ported by fokkonaut - ddnet/ddnet7

DDraceNetwork server modification for Teeworlds 0.7, ported by fokkonaut - ddnet/ddnet7

15:00

This looks useless: https://github.com/ddnet/ddnet7/blob/cb7a6cc1a0f4f1bcfaa5d260810c472916e3f156/src/game/server/score/file_score.cpp#L217-L222

DDraceNetwork server modification for Teeworlds 0.7, ported by fokkonaut - ddnet/ddnet7

You dont have to tell me, better make an issue or pull requets at ddnet/ddnet

15:00

ddnet7 uses the same system as ddnet for file scores

15:01

and in ddnet6 its also broken

I guess on ddnet usually MySQL is used And the fallback thing appends sql commands to a file and does not use the filescoring class

might be, but there are definitely servers running the ddnet mod using file score systems

15:02

basically every server except ddnet and maybe kog

15:02

all small servers use file score, because why would you use MySQL for example for a single server?

Well it does not have to be mysql... Sqlite would be perfect for something like this But yeah... Something as simple as the file scoring is still useful

15:04

No config and easy to handle

Yea

Learath2 i think minecraft forge is a rly good example

15:04

idk how would u apply to c++ tho

15:05

@Learath2 test

Wouldn't webassembly be a good choice for plugins? When you use native plugins you have to take care about all platforms

@redix I looked into web assembly, sadly no wasm vm has good linkage to c++

15:24

Exposing host functions is a pain when there is no stable api

15:25

I also looked into having a native plugin api that allows you to also run something like a lua interpreter, but the overhead seems too much

15:26

Something like modlua and modperl in weechat

Well there is a proposed standard api. Not sure how stable it is though. Wasmtime implements it I think

15:27

https://github.com/WebAssembly/wasm-c-api

WebAssembly/wasm-c-api

Wasm C API prototype. Contribute to WebAssembly/wasm-c-api development by creating an account on GitHub.

C api is workable but the c++ one seems to be pretty incapable

Any one on debian10? @jao ? And got a fix for the graphics bug when tabbing back into ddnet after latest debian update. Resolution gets worse and it tabs in and out 3 times.

@ChillerDragon have you tried debugging it?

[ddnet/ddnet] Pull request opened: #1922 Allow to hammer directly on unfreeze

Previously, this was only doable with frame-perfect input if you were going to get frozen again the next tick. Now, you only have to hold your hammer to immediately hammer once you get unfrozen. This fixes that "deepfly" (flying with deep-frozen dummy, binding both main and dummy tee hammers onto the same key) does not only work for good connection.

Exported 64 message(s)