| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Burlap and Hessian use type names in two contexts:
The mapping from a language's types to the Burlap/Hessian names is necessarily language dependent, but designed for cross-language use.
The mapping of specific Java types to Burlap/Hessian is relatively straightforward. A Java implementation will also use this mapping when it's parsing and knows the type of the object. Contrast that with the Burlap to Java mapping. Note that XML DOM is mapped to a string.
Note: Some of these conversions are lossy. For example, a Byte object serialized and then deserialized will become an Integer. This is the price paid working with other languages.
Java-based servers typically expose three overloaded methods:
In this case, all of "add" , "add__2", and "add_int_int" will map to the same method. The client will use whichever is appropriate to its capabilities. For example, many scripting languages can only handle the number of arguments as a mangling technique.
In several cases, a Java parser may know nothing about the expected type. For example, an method's argument might be Object or the serialized data may be in a Vector, or a client may not specify the type for a <map>.
v3 changes (2001-12-27)
v2 changes
|