Re: setfenv on userdata???

[Rici Lake <lua@...>]

On 22-Aug-05, at 10:35 AM, Chris Marrin wrote:

> Mike Pall wrote:
>
> > Hi,
> > Chris Marrin wrote:
> > > I see in 5.1-work6 that you can now set environments not only on 
> > > both C and Lua functions, but also on userdatas. But I don't see 
> > > anywhere in the code where this environment is being used. What can 
> > > you do with this environment? Does it get used when you make 
> > > metamethod calls with this userdata? Or is it just there so it can 
> > > be retrieved or something?
> > http://lua-users.org/lists/lua-l/2005-04/msg00213.html
>
> Thanks for the pointer. This brings up a question:
>
> - I assume that luaopen_queue() is what gets implicitly called when 
> you do the require "queue", right? If so, then it looks like it sets 
> the environment for your test program to the environment for the 
> queue, as evidenced by lua_replace(L, LUA_ENVIRONINDEX). Looking at 
> the code for this, it gets the current function (which is the main lua 
> program at this point?) and sets its environment to the passed 
> table(which is the metatable you are constructing for queue). Doesn't 
> that change the environment for the whole program?

Yeah, that threw me as well. However, the idea is that you don't just 
call luaopen_queue; in 5.1, you lua_call library open functions. So 
your code would probably have looked like:

lua_pushcfunction(L, luaopen_queue);
lua_call(L, 0, 0);

in which case the instance of the luaopen_queue function which is 
setfenv'd is promptly discarded.

It still seems a little odd to me but I'm sure I'll get used to it.

Rici