Newbe question on function environment

Reading the discussion of _ENV (in 5.2) I got the impression that local variables were stored in the _ENV of that function and upvalues to defined functions were therefore stored in a copy of the _ENV of the function that did the definition . I wrote the following test snippet to check this:

function f(x)

print'in f'

local fvar = x

print(_G, _ENV)

print(_G.fvar,_ENV.fvar)

function g()

print'in g'

local gvar = 888

print(_G, _ENV)

print(gvar, fvar)

print(_G.fvar,_ENV.fvar, _G.gvar,_ENV.gvar )

end

but this gives the result:

> f(1000)

in f

table: 0x100100e50 table: 0x100100e50

nil nil

> g()

in g

table: 0x100100e50 table: 0x100100e50

888 1000

nil nil nil nil

Obviously things do not work the way I imagined. _ENV is always equal to _G and I don't understand where either local variables or upvalues live. This despite the fact that g() clearly is getting the correct upvalue.

I feel that I will not make any progress until I understand this critical point. Can anybody give some advice to a new student?

Thanks