Re: setfenv on userdata

[Rici Lake <lua@...>]

On 24-Aug-05, at 9:53 AM, Chris Marrin wrote:

> This gets the desired field out of the env table. If it is nil that 
> means it does not exist in that table, which makes sense. But then you 
> test to see if that value is the same as the environment. Is that 
> right? I don't see how that can work. Wouldn't this test actually want 
> to test -2 against the LUA_ENVIRONINDEX?

Oops. Quite right. -1 is correct in setenvfield, but not in getenvfield.

> - CodeSnippetThree: It looks like you leave a value on the stack. Is 
> that intended?

Where, exactly?

Both of these were abstracted out of a more complex binding library, so 
I haven't actually tried them. I'll try to post the real code (which 
actually works, I think) later today.

> After reading all this I am still confused about what the environment 
> table in a userdata does. Is it just a place to store a table that can 
> be retrieved and manipulated, or does it actually come to play in some 
> API calls somewhere? As I mentioned, I don't see it being used 
> anywhere in the 5.1 API. Sorry about being dense, but I really think 
> this will solve a lot of problems for me, so I really want to 
> understand it well.

Me, too. :)