Open crazy2be opened 11 years ago
(see #67 re that last point)
We may want to consider fixing our timesteps to counteract this: http://gafferongames.com/game-physics/fix-your-timestep/
We really just need to do our collision detection continuously, it would be faster anyway.
I'm not sure about that, that would require a pretty substantial overhaul of how things work. It also sees like it would be a really hard problem to solve generally, which is probably why like no other physics engine does it.
Although if you really want to attempt that feel free :P.
Our behaviour when the client is lagging (or just time-synced) and thus giving us very large simulation dts is quite bad, and can actually allow you to walk through walls (with sufficient lag). Rather than capping at 1s, we should probably opt instead to run ten movement simulations, each one 0.1s long (or something).
This also causes annoying console spam.
This is aggravated by no gradual application of ntp-sync packets.