Joe's main motivation of this, it seems, is to unify the shell and the modules. Whereas you could not define a module in the shell, and whereas you could not d

Joe's main motivation of this, it seems, is to unify the shell and the modules. Whereas you could not define a module in the shell, and whereas you could not define multiple modules in a single file, with Erl2, both are possible. And the fact that it won't generate .beam file without passing the unit tests first, is just icing on the cake :) I find this rather interesting, and I'm curious what kind of traction this might get, or if any of the paradigms introduced here might be mimicked in Erlang.