Re: Re[2]: require ignores setglobals()

[Luiz Henrique de Figueiredo <lhf@...>]

Peter Hill wrote:
>It did. Now you access the global table _G directly! So Lua4
>  old = setglobals(sb) ; blah ; setglobals(old)
>
>becomes in Lua5
>  old,_G=_G,sb ; blah ; _G=old

Not true. Try

  a=1
  sb={a=2}
  old,_G=_G,sb ; print(a) ; _G=old

If things were as you said, you'd expect it to get 2, but actually you'd
get an error, because print is not present in sb. Of course, you get 1.

_G is not magical variable. It's not part of the core; it's provided by the
base lib for convenience. In other words, global names are *not* resolved by
indexing _G. They're resolved by indexing whatever getglobals return at the
time (of course, the core knows about the current global table of the current
function being executed and does not call getglobals).
--lhf