Re: Some comments on CPS control flow functions in Lua 5.2

[Drake Wilson <drake@...>]

Quoth Roberto Ierusalimschy <roberto@inf.puc-rio.br>, on 2010-05-21 09:32:24 -0300:
> I think I was not clear about the "quite complex restrictions". Surely
> destroying the C frames at each callk would be simpler for the one
> programming that particular function. (As you said, there would be
> only one continuation.) The problem is for people *using* such function.
> 
> Lua can destroy a C frame only when there is a "protection point"
> (setjmp) down the stack *and* all intermediate C frames have
> continuations.

Ah, I see.  Yes, I misunderstood what you said earlier, sorry.  Now
that I look at the internals more, it seems like an unfortunate
restriction of the implementation.  My reasoning was that since Lua
controls the entrance to the C frames to start with, it should be able
to fix everything up on return without reference to any lower frames,
but it seems that it's not built that way.  Hmm.

I will have to poke around at this some more.  Thanks for the
clarification.

> -- Roberto

   ---> Drake Wilson