module() with _ENV

["Stuart P. Bentley" <stuart@...>]

With the proposed _ENV environment design for 5.2 that would reassign  
setfenv/getfenv to the debug library, would it also make sense to change  
the design of module() to return the table to use as the environment,  
rather than actually having it change the environment? This way, you could  
replicate the old use of module with the only slightly different statement  
"_ENV = module(...)" without having to "bend the rules" of the language  
(that non-debug functions can't change the environment of their callers),  
as well as allowing other uses of module(...) for its semantics handling  
(its table reuse/creation rules)- so, if you prefer to explicitly put  
everything inside your module's table, you could still make this table  
with "local mymod = module(...)", and it would handle all the complexities  
of creating the table (re-using any existing table in the working  
environment, creating it otherwise, and placing it in package.loaded).