I'm not a specialist of FP, so please take what I say with a grain of salt. Before answering your questions, here's a preliminary point:
0) The problem is that FP is not really defined anywhere, or rather, everyone has its own definition. I was wondering about the definition of FP about a month ago, and discussed it with my roommate who's a phd student in programming languages. The conclusion of this 2-hour long conversation was: well FP is kind of a fuzzy concept. The concepts of FP languages, typed languages, static languages, etc.. all tend to conflate. Some people will consider that Scheme, Python, Ruby, Scala, Haskell, OCaml, R are all FP languages. Some will say that only Haskell and OCaml really are. Whatever..
Keeping that in mind, here some attempts at answers:
1) In a very loose sense, I guess that you can say that, and to make it possible, you need some constructs in your language, the minimum requirements being higher order functions, and maybe closures. I'd say that's the minimum, but again I'm no specialist. But then if you want to push this "no globals" moto to its logical extreme, you get to a point where you can't even use a "print" function, since it has some side effects. Haskell does, and solves the problem with monads. But to have monads you need a very particular type system, which leads to the question of wether you need a type system to go FP all the way? I don't know. Someone more knowledgeable should answer that
2) I don't know about hardware. But like I said, FP can deal with state. In fact, even hardcore FP as embodied by Haskell can deal with state. It's not trivial to wrap your head around the concept, but it's worth giving it a shot.
0) The problem is that FP is not really defined anywhere, or rather, everyone has its own definition. I was wondering about the definition of FP about a month ago, and discussed it with my roommate who's a phd student in programming languages. The conclusion of this 2-hour long conversation was: well FP is kind of a fuzzy concept. The concepts of FP languages, typed languages, static languages, etc.. all tend to conflate. Some people will consider that Scheme, Python, Ruby, Scala, Haskell, OCaml, R are all FP languages. Some will say that only Haskell and OCaml really are. Whatever..
Keeping that in mind, here some attempts at answers:
1) In a very loose sense, I guess that you can say that, and to make it possible, you need some constructs in your language, the minimum requirements being higher order functions, and maybe closures. I'd say that's the minimum, but again I'm no specialist. But then if you want to push this "no globals" moto to its logical extreme, you get to a point where you can't even use a "print" function, since it has some side effects. Haskell does, and solves the problem with monads. But to have monads you need a very particular type system, which leads to the question of wether you need a type system to go FP all the way? I don't know. Someone more knowledgeable should answer that
2) I don't know about hardware. But like I said, FP can deal with state. In fact, even hardcore FP as embodied by Haskell can deal with state. It's not trivial to wrap your head around the concept, but it's worth giving it a shot.