C++ has support for nested namespaces; however, KL does not (yet) support namespaces. As such, Kludge maps C++ namespaces to KL nested names using a Prefixed_With_Underscores_ syntax.
The root ext object that is provided to the API description language also represents the root namespace. A nested namespace can be added with the ext.add_namespace(method). Its parameters are:
def add_namespace(self, cpp_name, kl_name=None):
The cpp_name is the name of the namespace in C++, and is required. kl_name is the name that should be used for prefixing in KL, and if not provided defaults to the cpp_name.
The resulting namespace object can in turn have types added to see as described in How Kludge Wraps Types, have functions added to it How Kludge Wraps Functions, or even have sub-namespaces created below it.
This is illustrated with an example. The following API description:
ns = ext.add_namespace('MyLib')
ty = ns.add_owned_type('SomeType')
ty.add_const_method('Foo', 'int')
sns = ns.add_namespace('ChildNS')
sns.add_alias('IntType', 'int')
would result in the KL API:
struct MyLib_SomeType {...};
SInt32 MyLib_SomeType.Foo?();
alias SInt32 MyLib_ChildNS_IntType;
Note that wrapping a type (for example, using .add_owned_type()) also creates a child namespace. This is important functionality that allows, for instance, nested types, enums and type aliases to be scoped under classes or structs:
ty = ext.add_type("SomeType")
ty.add_enum("SomeEnum", ['ValueOne', 'ValueTwo'])
produces:
struct SomeType {...};
alias SInt32 SomeType_SomeEnum;
const SomeType_ValueOne = 0;
const SomeType_ValueTwo = 1;
It is important to note that C++ types are looked up in the namespace where they are referenced. This works exactly as you would expect in C++.