sal/aiprompts/rhaiwrapping_best_practices.md

Best Practices and Optimization

When wrapping Rust functions for use with Rhai, following these best practices will help you create efficient, maintainable, and robust code.

Performance Considerations

1. Minimize Cloning: Rhai often requires cloning data, but you can minimize this overhead:

   // Prefer immutable references when possible
   fn process_data(data: &MyStruct) -> i64 {
       // Work with data without cloning
       data.value * 2
   }
   
   // Use mutable references for in-place modifications
   fn update_data(data: &mut MyStruct) {
       data.value += 1;
   }
   

2. Avoid Excessive Type Conversions: Converting between Rhai's Dynamic type and Rust types has overhead:

   // Inefficient - multiple conversions
   fn process_inefficient(ctx: NativeCallContext, args: &mut [&mut Dynamic]) -> Result<Dynamic, Box<EvalAltResult>> {
       let value = args[0].as_int()?;
       let result = value * 2;
       Ok(Dynamic::from(result))
   }
   
   // More efficient - use typed parameters when possible
   fn process_efficient(value: i64) -> i64 {
       value * 2
   }
   

3. Batch Operations: For operations on collections, batch processing is more efficient:

   // Process an entire array at once rather than element by element
   fn sum_array(arr: Array) -> Result<i64, Box<EvalAltResult>> {
       arr.iter()
          .map(|v| v.as_int())
          .collect::<Result<Vec<i64>, _>>()
          .map(|nums| nums.iter().sum())
          .map_err(|_| "Array must contain only integers".into())
   }
   

4. Compile Scripts Once: Reuse compiled ASTs for scripts that are executed multiple times:

   // Compile once
   let ast = engine.compile(script)?;
   
   // Execute multiple times with different parameters
   for i in 0..10 {
       let result = engine.eval_ast::<i64>(&ast)?;
       println!("Result {}: {}", i, result);
   }
   

Thread Safety

1. Use Sync Mode When Needed: If you need thread safety, use the sync feature:

   // In Cargo.toml
   // rhai = { version = "1.x", features = ["sync"] }
   
   // This creates a thread-safe engine
   let engine = Engine::new();
   
   // Now you can safely share the engine between threads
   std::thread::spawn(move || {
       let result = engine.eval::<i64>("40 + 2")?;
       println!("Result: {}", result);
   });
   

2. Clone the Engine for Multiple Threads: When not using sync, clone the engine for each thread:

   let engine = Engine::new();
   
   let handles: Vec<_> = (0..5).map(|i| {
       let engine_clone = engine.clone();
       std::thread::spawn(move || {
           let result = engine_clone.eval::<i64>(&format!("{} + 2", i * 10))?;
           println!("Thread {}: {}", i, result);
       })
   }).collect();
   
   for handle in handles {
       handle.join().unwrap();
   }
   

Memory Management

1. Control Scope Size: Be mindful of the size of your scopes:

   // Create a new scope for each operation to avoid memory buildup
   for item in items {
       let mut scope = Scope::new();
       scope.push("item", item);
       engine.eval_with_scope::<()>(&mut scope, "process(item)")?;
   }
   

2. Limit Script Complexity: Use engine options to limit script complexity:

   let mut engine = Engine::new();
   
   // Set limits to prevent scripts from consuming too many resources
   engine.set_max_expr_depths(64, 64)     // Max expression/statement depth
         .set_max_function_expr_depth(64) // Max function depth
         .set_max_array_size(10000)       // Max array size
         .set_max_map_size(10000)         // Max map size
         .set_max_string_size(10000)      // Max string size
         .set_max_call_levels(64);        // Max call stack depth
   

3. Use Shared Values Carefully: Shared values (via closures) have reference-counting overhead:

   // Avoid unnecessary capturing in closures when possible
   engine.register_fn("process", |x: i64| x * 2);
   
   // Instead of capturing large data structures
   let large_data = vec![1, 2, 3, /* ... thousands of items ... */];
   engine.register_fn("process_data", move |idx: i64| {
       if idx >= 0 && (idx as usize) < large_data.len() {
           large_data[idx as usize]
       } else {
           0
       }
   });
   
   // Consider registering a lookup function instead
   let large_data = std::sync::Arc::new(vec![1, 2, 3, /* ... thousands of items ... */]);
   let data_ref = large_data.clone();
   engine.register_fn("lookup", move |idx: i64| {
       if idx >= 0 && (idx as usize) < data_ref.len() {
           data_ref[idx as usize]
       } else {
           0
       }
   });
   

API Design

1. Consistent Naming: Use consistent naming conventions:

   // Good: Consistent naming pattern
   engine.register_fn("create_user", create_user)
         .register_fn("update_user", update_user)
         .register_fn("delete_user", delete_user);
   
   // Bad: Inconsistent naming
   engine.register_fn("create_user", create_user)
         .register_fn("user_update", update_user)
         .register_fn("remove", delete_user);
   

2. Logical Function Grouping: Group related functions together:

   // Register all string-related functions together
   engine.register_fn("str_length", |s: &str| s.len() as i64)
         .register_fn("str_uppercase", |s: &str| s.to_uppercase())
         .register_fn("str_lowercase", |s: &str| s.to_lowercase());
   
   // Register all math-related functions together
   engine.register_fn("math_sin", |x: f64| x.sin())
         .register_fn("math_cos", |x: f64| x.cos())
         .register_fn("math_tan", |x: f64| x.tan());
   

3. Comprehensive Documentation: Document your API thoroughly:

   // Add documentation for script writers
   let mut engine = Engine::new();
   
   #[cfg(feature = "metadata")]
   {
       // Add function documentation
       engine.register_fn("calculate_tax", calculate_tax)
             .register_fn_metadata("calculate_tax", |metadata| {
                 metadata.set_doc_comment("Calculates tax based on income and rate.\n\nParameters:\n- income: Annual income\n- rate: Tax rate (0.0-1.0)\n\nReturns: Calculated tax amount");
             });
   }