Many thanks for the replies and references all of you! I will continue to read up on this from here, and you have all boosted my interest in investigating how complex systems could be developed using functional languages. On Fri, Mar 16, 2012 at 1:46 PM, Dimitri Scheftelowitsch < d.scheftelowitsch@googlemail.com> wrote:

Hi,

it depends on what your definition of „problem“ is. If a problem is just something that is defined by „given input A, generate output B such that some constraints are met“, then you can go with anything Turing-complete.

If you think of problems in the sense of software engineering and disallow side effects, then you cannot do anything — any sensible software requires I/O.

If you want to compare the programming models of lambda calculus vs. random access machines, then you have nothing worse than a polynomial slowdown in terms of runtime. Strong Church-Turing thesis and so on.

HTH Dimitri

On 16.03.2012, at 13:23, Christopher Svanefalk wrote:

...

Dear all,

there is a question I have been thinking about a bit. In short, we could simply formulate it like this:

Are there any problems which cannot be solved a side effect-free language (such as Haskell)? In other words, are there problems that would explicitly demand semantics that can only be provided by a language allowing direct modification of external state variables, such as Java and C++?

If not, are there problems which are simply infeasible to solve with a side effect-free language?

I realize this question is very broad, and I am not expecting an exact answer by any means. I am asking since I am curious about the relation between functional and imperative/procedural languages in general. I primarily come from a Java background, but I can program Haskell and Erlang, and have recently started exploring Scala, so this would be interesting to know.

-- Best,

Christopher Svanefalk

Student, Department of Computer Science and Engineering University of Gothenburg / Chalmers University of Technology

-- Best, Christopher Svanefalk