Ludum dare 40 compo was a hard time.
For this event I wanted to make an online multiplayer game. Thing is done but I’m not really enjoying the result but it basically works.
https://ldjam.com/events/ludum-dare/40/super-childcare-assistant
My base code ,which was done the week before the event, was inspired by the wwx2017 workshop on multiplayer game in haxe. I spent the week to make this agar.io clone able to run with flixel client framework and change the NodeJs only target for server compile to cpp and neko. Moreover I added a master server where each game server registers. This way the client can easily get a list of all the game running online.
What went wrong :
I spent a lot of time on technical issues instead of focusing on the game, so I didn’t had time to make cool graphics and sounds.
No premade collision system so I did one very quickly and was a bit buggy. I wasn’t able to implement map collision. So no walls in the game.
Desynchronization between server and client take me lot of time to figure out that client was not able to consume server states despites he was running 2x faster. This stacking of states must be due to network : Sometime all pending messages arriving together. I must check this.
No client side prediction sot the game seems laggy.
What went right :
Code sharing between client and server is very efficient, save lot of time, no need for another language, no need to write same logic twice. Easy to make a solo version with bots.
Efficient workflow. Can running everything locally without any special setup.
Conclusion :
Despite/thanks to it was technically hard I learned a lot on fast paced multiplayer. I used to work with smartfoxserver . But hand in the engine is really interesting.
I need to work on a better system collision before a jam.
I don’t think I will make another online game for a 48h game jam.