想象一个涵盖人类所有知识的圆：

当你小学毕业时，你只知道一点儿：

到你中学毕业时，你知道多点儿了：

有了学士学位后，你获得了专长：

硕士学位则加大你的专业纵深：

阅读研究论文将你领向人类知识的边沿：

一旦到达边界，选好发力点：

然后在那个点上使劲推边界几年：

直到有一天，那个边界退却：

你弄出来的那个凸起就叫Ph.D.：

当然，对你来说现在世界看上去不一样了：

因此，别忘了大图：

继续推。

SRFI-1 provides a pair of procedures proper-list? and improper-list? for testing whether a list is proper or not respectively. Thus when we write a procedure to handle proper list, we can first check to see if the passed in argument is a proper list before we actually deals with it. For example, the standard length procedure could be defined as follows:

   (define (length l)
     (cond ((improper-list? l) (print "Error: passed in an improper list"))
           ((null? l) 0)
           (else (+ 1 (length (cdr l))))))

It’s easy to see that both proper-list? and improper-list? need to traverse the list before they can tell whether the list is proper or not. To avoid this unnecessary traversal, we can embed the test inline as shown below:

   (define (length l)
     (cond ((pair? l) (+ 1 (length (cdr l))))
           ((null? l) 0)
           (else (print "Error: passed in an improper list"))))

This version of length seems better than the first one, however, it may not generally be true. For a general procedure on list, the late coming info about the proper/improper attribute of the list (we don’t till the end) may make us do much useless work on every element of the list except the last.