[GROOVY-1826] Stop requiring exceptions in MetaClass to indicate method/property does not exist

• Type: Sub-task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.0
• Component/s: groovy-runtime
• Assignee: Unassigned
• Reporter: Marc Palmer
• Created: 11/Apr/07 08:45
• Updated: 13/Apr/15 08:58

Description

Currently the MetaClass mechanism has no clean way for a descendent metaclass or runtime code to know if a property or method exists - including those dynamically intercepted/provided by the metaclass.

This means you have to rely on exceptions and this is very costly in performance terms in a multi threaded environment.

See http://jira.codehaus.org/browse/GRAILS-1036

This is a Grails issue with code attached for a workaround for the lack of a hasProperty, hasMethod and hasClosure method on MetaClass. It is ugly but it works, by caching the result of invocations / pickMethod and encoding some special behaviour related to closures. This duplicates/mutates logic currently in MetaClassImpl.invokeMethod() which is a bad thing.

We have seen a 30-50% speed increase in serving Grails pages in concurrent tests after using this new metaclass, instead of catching MissingMethodException and MissingPropertyException.

Also I believe that MetaClass should have invokeClosure on it, and remove some of the closure calling logic from invokeMethod, as this makes for a cleaner separation.

Issue Links

• depends upon: GROOVY-2503 MOP 2.0 design inflluencing issues