There has to be some small or tiny set of common concepts that can be used to implement all of those. Perl comes, in my opinion, close. Perl provides packages, which are really just namespaces, and you can import and export names from other packages. It is possible to bind a scalar (variable) reference into a particular package by "blessing" it into the class, after which you can use the familiar arrow syntax to call functions on it, and fake it's really an object (in the object-oriented sense). If a function foo() is defined in package A, and you bless variable $v to A, $v->foo() then calls foo() in class A with an implicit first argument $v (A::foo($v)).

Perhaps it's not the best way to do it. I have only briefly read about mzscheme modules, but they seem to support importing and exporting at least. Maybe it's enough; maybe the only thing needed is light-weight namespaces.

Yes, you could do all this with macros and closures, and as I understand it this is how it's implemented in many Scheme implementations (since the standard up until R6RS didn't provide a module mechanism). But it feels this is a deeper design issue that should be taken in to account. Are we missing something?

I certainly hope Graham is not going to include something as large as CLOS, or any particular object-oriented framework, in the language. That would miss the goal entirely.