Timeouts and Handlers

This document explains how Scarpe manages timeouts and event handlers within its cooperative evented approach.

Timeout Management

Scarpe implements timeouts with these key considerations:

1. Cooperative Nature
   • Timeouts don’t interrupt execution
   • They’re processed when the event loop is ready
   • Timing is approximate, not exact
2. Implementation

   def after(milliseconds, &block)
     register_timeout(Time.now + (milliseconds / 1000.0), block)
   end
   

Event Handlers

Event handlers in Scarpe follow these principles:

1. Registration
   • Handlers are registered with specific events
   • Multiple handlers can exist per event
   • Order of execution is preserved
2. Execution
   • Handlers run in the main event loop
   • They should be non-blocking when possible
   • Long operations should be broken up

Restrictions

The cooperative approach has some important restrictions:

1. No Preemption
   • Long-running handlers block other events
   • Timeouts may be delayed
   • UI updates wait for handlers to complete
2. Best Practices
   • Keep handlers short
   • Use async for long operations
   • Break up complex tasks

Error Handling

Handlers include error management:

1. Exception Catching
   • Errors don’t crash the event loop
   • Exceptions are logged appropriately
   • Error handlers can be registered
2. Recovery
   • System can recover from handler errors
   • State remains consistent
   • Other handlers continue to work