Is calling rawset ot the same table in __index valid?

["Dimiter \"malkia\" Stanev" <malkia@...>]

	Hi people,
	
	I'm making my slow progress using luajit and ffi, and my goal was to 
have in the same table both the ffi symbols from a library, and new ones 
added. But the goal is if one wants to add a symbol that's from the 
library it should report an error.

	Anyway I'm not getting something right here, as my __index callback is 
not hit everytime, although I've installed it with setmetatable. I 
wonder if that's not because I'm using rawset in __index to the same table.

	Here is the code (luajit), and it loads "curl", not entirely

	https://gist.github.com/898171

	And the output

	https://gist.github.com/898172

	The problem is that in the following three lines (one is only needed 
really, but this is to demonstrate the problem)

	print( lib["curl_easy_init"] )
	print( lib["curl_easy_init"] )
	print( lib["curl_easy_init"] )

	I get the output

	cdata<void **()>: 0x00043820
	cdata<void **()>: 0x00043820
	cdata<void **()>: 0x00043820

	But I would've expected to get

	__index: returning from table <...>
                           the symbol curl_easy_init = ...

	because I cache it with rawset( t, k, v ), although that happens in _index

	The way things are is that luajit has no "portable" way of knowing what 
symbols might be in a .dll/.so/.dylib/etc, so as soon as someone asks 
for them, they get "GetProcAddress/dlsym/dyld_whateveritis" - e.g. they 
come in lazy.

	And I'm trying to make lazy "loaded" symbols to have their "seats" 
preserved, if someone calls their name, but still allow for others. This 
is simply to wrap (hence wraplib name) symbols that might be coming out 
of the dynamic library, and locally defined lua ones.

	I'm not entirely sure whether that's good idea at all, but wanted to 
try and see.
	
	Thanks,
	Dimiter "malkia" Stanev.