To understand what object-oriented programming entails, we need to place it in the context of other styles. Historically, the first computer programs were written in an *imperative programming* style. A program was construed as a series of instructions, where each instruction performs some action: changing the value of a memory location, printing a result, and so forth. Assembly language is an example of an imperative language. As experience (and ambition) grew, programmers looked for ways of controlling the complexity of programs. The invention of subroutines marked the *algo- rithmic* or *procedural programming* style, a subclass of the imperative style. Subroutines are helpful for two reasons: breaking up the problem into small pieces makes each piece easier to understand, and it also makes it possible to reuse pieces. Examples of procedural languages are FORTRAN, C, Pascal, and Lisp with setf. Subroutines are still dependent on gloabal state, so they are not completely separate pieces. The use of a large number of global variables has been criticized as a factor that makes it difficult to develop and maintain large programs. To eliminate this problem, the *functional programming* style insists that functions access only the parameters that are passed on them, and always return the same result for the same imputs. Functional programs have the advantage of being mathematically clean -- it is easy to prove properties about them. However, some applications are more naturally seen as taking action rather than calculating functional values, and are therefore unnatural to program in a functional style. Examples of funtional languages are FP and Lisp without setf. In contrast to imperative languages are *declarative* languages, which attempt to express "what to do" rather than "how to do it". One type of declarative programming is *rule-based* programming, where a set of rules states how to transform a problem into a solution. Examples of rule-based systems are ELIZA and STUDENT. An important kind of declarative programming is *logic programming*, where axioms are used to describe constraints, and computation is done by a constructive proof of a goal. An example of logic language is Prolog. *Object-oriented programming* is another way to tame the problem of global state. Instead of prohibiting global states (as functional programming does) , object-oriented programming breaks up the unruly mass of global state and encapsulates it into samll, manageable pieces, or objects. Chapter 13, PAIP Peter Norvig --

※ 发信站: 批踢踢实业坊(ptt.cc) ◆ From: 211.150.216.198 ※ 编辑: caml 来自: 211.150.216.198 (03/27 10:07) ※ 编辑: caml 来自: 211.150.216.198 (03/27 10:11)