---
tags:
- CAGE_dev_log
- CAGE
- rust
- bevy
- game_dev
- road_system
---

# CAGE Dev Log - Day 5

Today I implemented the path locking system allowing car merging into one lane without crashing into each other.

Each car needs to propose its *intent*, and a *god* (temporarily called `schedule_intents`) will review each intent, and approve, trim, or reject them.

This implementation is a little bit naive and only scheduling by priorities.

Here's the simplied psuedocode for those systems:

### schedule_intents

pub fn schedule_intents(
    mut intents: Vec<PathLocks>
) {
    for intent in intents.iter_mut() {
        // for each segments
        for lock in intent.locks {
            for other_lock in other_already_locks {
                    if collision(lock, other_lock) {
                    // try to trim lock
                    if trim_lock(&mut lock) {
                        intent.path_locks.truncate(j + 1);
                    } else {
                        // lock with `lock_together` cannot be trim
                        // since some path needs to be lock together
                        // e.g. paths in the junction.
                        intent.path_locks.truncate(j);
                        pop_locks_until_no_group(&mut intent.path_locks);
                    }
                    break;
                }
            }
        }
        
        for other_lock in other_new_locks {
            // trim lock or other_lock which has lower priority.
            
        }
        
    }
}

We should create a index to speedup the collision query later.