It is an experiment to bring namespaces to ruby. It is not quite clear yet why this would be useful, but it can be interesting nonetheless.
This system is inspired by the CommonJS and Python module systems, adapted for the ruby particularities.
By loading the .rb files into a temporary module, it is possible to avoid polluting the global namespace and select what to import in the current module.
Compatible with ruby 1.8.7 and 1.9.2 (and others?)
Possible to avoid namespace collision on the module level
uses eval and other nasty ruby hacks
classes and namespaces don't mix well because `class X; <<here>>; end` is a new context that doesn't inherit from it's parent.
m = Namespace.open("hello::world") m.hello #=> "hello world"
In the context of “hello/world.rb” Namespace#import is made available as a shortcut.
X = import "another::module" def hello X::Joiner.new("hello", "world").to_s end
Classes and modules are also exported:
class Joiner def initialize(*a) @args = a end def to_s @args.join(' ') end end
Handle circular dependencies ?
Avoid eval (but I don't see how)
I just learned that Kernel#load(path, wrap=true) will load the file in an anonymous module.
A module can emulate the main object behavior by extending itself. In that way, it makes it's functions directly available to consumption.
module X extend self def test "x" end test #=> "x" end
Should ruby “namespaced” modules have their own $LOAD_PATH and file extensions ?