Upon reading about Scheme I came across this statement.
"Scheme's equivalent of an activation stack is really a chain of partial continuations (suspension records)."
I'm a little confused on what this actually means. What differentiates scheme stacks from, say, C's?
On
In C, your stack would be little more than a series of memory pointers telling you where you were when you left off.
In Scheme, since everything is a list, you're really just moving up a list. You could actually look at it like your program IS the stack.
On
The basic difference is that in Scheme (among others) the current state of the stack can be/is a first class object. You can (for example) create a variable that holds an entire stack state. You can then use that continuation to call different functions from the same starting state, or you can walk the (Scheme view of) "the stack" about like a normal list.
Scheme supports the construct call-with-current-continuation (read about it in R5RS). The construct can be used to return to a previous continuation. This implies that the activation records no longer form a stack. Instead the activation records form a tree (note that a stack is special form of a tree). In order to get a feel for the concept of continuations, I can recommend
Daniel P. Friedman. "Applications of Continuations: Invited Tutorial". 1988 Principles of Programming Languages (POPL88). January 1988. http://www.cs.indiana.edu/hyplan/dfried/appcont.pdf
If you get hooked on continuations, then you'll find lots of interesting papers here.