Discussion Making A Java Based Lineage 2 Server From Scratch

[Mobius]

How would you make a server from scratch?

Asumming there was no other project to get as a base.

 

To make it more interesting lets asume you would only want to support the Awakening client.

A kinda new client with no access to sniff info from retail or other servers.

 

Elaborate on...

-Server/Client packet handling system.

-Skills engine.

-Scripting system.

-Geodata handling.

-Database system based on MySQL.

 

What else you think it would be important to elaborate on?

 

Please keep conversation serious an on topic.

[Elfo]

How would you make a server from scratch?

Asumming there was no other project to get as a base.

 

To make it more interesting lets asume you would only want to support the Awakening client.

A kinda new client with no access to sniff info from retail or other servers.

 

Elaborate on...

-Server/Client packet handling system.

-Skills engine.

-Scripting system.

-Geodata handling.

-Database system based on MySQL.

 

What else you think it would be important to elaborate on?

 

Please keep conversation serious an on topic.

Why don't you join me? I am doing the same thing pretty much on C#.

 

There is no point to do it again in java unless if you have tons of free time, but you will end up copying a lot of code from l2j.

[Mobius]

The basic idea of this is, to theoretically use the knoledge we have gained so far from java servers to make the best software we could make, from scratch.

Theoretically... this info can help someone that might want to get involved with it. ;)

 

I would really like to see people seriously elaborate with the idea.

Edited April 21, 2016 by Mobius

[Tryskell]

How would you make a server from scratch?

Asumming there was no other project to get as a base.

 

To make it more interesting lets asume you would only want to support the Awakening client.

A kinda new client with no access to sniff info from retail or other servers.

 

Elaborate on...

-Server/Client packet handling system.

-Skills engine.

-Scripting system.

-Geodata handling.

-Database system based on MySQL.

 

What else you think it would be important to elaborate on?

 

Please keep conversation serious an on topic.

 

A server from scratch without client informations ? Better create your own game, it will be faster.

 

- Skills engine is like any XML engine out there, and like any fixed data it shouldn't be set on SQL because you will need a lot of redundancy to do the same thing (mostly 2 or 3 tables, like npcs data).

- Geodata you have to refer to an existing data too, so you're kinda limited. Where you can have more choices is about pathfinding type (a*, etc).

- Scripting engine, L2J engine handles already 99% of retail cases. You then got the choice between script system : one script per npcid, or one script by "theme". One script by npcid is extremely redundant, but you got no problem between the different scripts.

- server/client handling, better use directly an existing architecture (Putty or similar stuff, I saw some far lesser hungry stuff)

- database, well... MySQL is fine. If you mean which db pooler, the good stuff atm is HikariCP.

 

On a L2J base, I think the biggest flaws are the knownlist system, db pooler choice and the client/server packet system. In term of gameplay, I think the whole casting and movement behaviors should be rewrite from zero. Those are the main downfalls.

 

I doubt than rewriting everything from zero will lead to "transcendal" speedup. If you fix the previous, you will get already a solid architecture. All by all, I think L2J architecture is ok. There were only strange choices made about technologies, and no thoughts about handling "masses".

Edited April 21, 2016 by Tryskell

[Sdw]

The whole AI, the whole stats system, basically everything is important, even if Tryskell never saw all flaws

[xxdem]

Create client from scratch fools, we all got shit of one night stand crappy projects

[Trance]

We are in 2016. Why the hack would you spend your time making something from the scratch? You rather develop another game.

[Szakalaka]

Do you realise how expensive it is to develop client side

[xxdem]

Do you realise how expensive it is to develop client side

 

 

everything is cracked, the most expensive part is the UI, the textures, the animations and the meshes. All are cracked. This client can even be coded in java

[Szakalaka]

I rather meant creating something new and original, not using existing assets from Lineage :)

[AccessDenied]

Unreal engine -> New game faster...

 

how exactly you gonna find out the whole information from the client? you gonna copy it anyway so is not from scratch...

just build new..

[Elfo]

Didn't you wanted to leave?

You are completely out of date aren't you now?

[Tessa]

Unreal engine -> New game faster...

 

how exactly you gonna find out the whole information from the client? you gonna copy it anyway so is not from scratch...

just build new..

You should have damn good reverse engineering skills in order to copy it anyway...

I spend 3 days in order to find all packets (with some help) for Interlude in engine.dll... maybe you can do it better. :lol:

Edited April 24, 2016 by Tessa

[Szakalaka]

You should have damn good reverse engineering skills in order to copy it anyway...

I spend 3 days in order to find all packets (with some help) for Interlude in engine.dll... maybe you can do it better. :lol:

 

There is static array of simple structs like const char* packet name, const char* format, and as far as i remember callback address

 

And if u need detail info just use packets from phx data, they did it for you 10 years ago

Edited April 24, 2016 by Szakalaka

[Tessa]

There is static array of simple structs like const char* packet name, const char* format, and as far as i remember callback address

 

And if u need detail info just use packets from phx data, they did it for you 10 years ago

L2PHX doesn't contains all the packets and some of the packets listed there are not completed. I found the packets but I had hard time looking for all the parameters they accept. Anyway, I completed the list... it was a big effort, because I don't have a clue about assembler. :lol: