Game #000: Astrorix

So after a fair bit of work scraping together the basis of our rube goldberg machine of a framework , I finally got it limping along enough to build the first game of the challenge:

Game #000 is Astrorix. A little top-down massive multiplayer space shooter thing inspired by Subspace.

Link to play it is at the bottom of this post.

Making some version of this is basically a tradition at this point. It's the "hello world" of multiplayer for me. It's simple enough to not have to give much thought to the game design, but is fast paced enough to stress test that multiplayer netcode and sync is working at least reasonably.

Here's me failing to control four ships at once:

It's not going to win any awards, but it's a good start and good test that the multiplayer parts are in a decent state, which is where all the effort has gone so far.

Everyone joins the same periodically repeating sessions, every 3mins. It should be able to handle up to 16 players ... but I have not really tested that.

Some kind of proto rollback networking is in place, although it does not do any actual prediction, so it's effectively lockstep with a bit of guessing for packet loss.

How did Claude do?

As a reminder; this little project is me building out some rails to see how AI does building multiplayer games. I'm spending the effort on the framework side, then basically vibe coding the game and seeing where it fails - then focusing on that failure for the next task.

The actual game implementation was super quick, Claude took my basic design, plus the pile of notes for how to write the simulation code in rust and the renderer in react, and it ... mostly worked.

The bigest failures were:

• Constant confusion about which way is "up" and how rotations work - I'd say most of the follow up prompts were some form of "Remember this is a Y-up, right-handed coordinate system"
• Cameras - I can see this being an ongoing issue, but AI can't see anything - maybe I need to bake a hand made camera system in so it doesn't get confused
• Physics - it's clear we should bake some kind of physics into framework rather than let AI invent it's own half-baked solutions, so that will be good candidate for next thing to fix

Give it a try!

Excuses Disclaimers

• Barely tested it on my own
• I've only got one relay (in Europe) so apologies for the lag on the other side of the planet for now
• WebRTC connection takes AGES to connect ("Loading...") - not found out quite why yet
• No sound ... ran out of time
• No mobile support - may have to make this a priority soon

#000

Astrorix

Multiplayer space shooter

Making some version of this game is basically a tradition for me at this point.

Read more→

PLAY

51 to go.

With networking basics and the general build structure in place, I want to start working on something I'm less familer with for the next one: third-person camera controllers.