[Merad]

> I am curious as to where you heard this. Modifying the stack in place seems like an implementation detail.

You are correct, but as the other reply mentioned, when we're talking about abstract specifications a stack typically doesn't allow reads and writes to arbitrary indices, therefore a stack machine can't perform an in-place swap. The implementation will almost certainly optimize the operation as you're describing, but I'd still expect to see a swap operation defined as "pop pop push push". Likewise I would usually expect to see an add described as popping two arguments and pushing one result, even thought the implementation probably pops one argument, then reads the second and overwrites it with the result.

In short, it's the abstract description of the thing vs the actual implementation.

[lacampbell]

I don't see why the abstract specification should mention pushing or popping at all. The abstract specification for swap is simply "swaps the first and second elements of the stack". How that's achieved doesn't seem important, and it seems odd to me to define it in terms of primitive operations it's not actually defined in terms of - unless it's some ultra minimalist Forth thing where swap isn't a primitive.

[frjalex]

I can only partially agree with this - having a stack exposed to modification would virtually make it not a stack.