七度&光 2013-01-24 07:01 采纳率: 22.2%
浏览 204
已采纳

Guile Scheme 解释器中奇怪的乘法行为

I was practicing Scheme in Guile 1.8.8 interpreter on OS X. I noticed something interesting.

Here's expt function which is basically does exponentiation expt(b,n) = b^n :

 (define (square x) (* x x))
 (define (even? x) (= (remainder x 2) 0))
 (define (expt b n) 
      (cond ((= n 0) 1)
        ((even? n) (square (expt b (/ n 2))))
        (else (* b (expt b (- n 1))))
      ))

If I try it with some inputs

 > (expt 2 10)
 1024
 > (expt 2 63)
 9223372036854775808

Here comes the strange part:

 > (expt 2 64)
 0

More strangely, until n=488 it stays at 0:

 > (expt 2 487)
 0
 > (expt 2 488)
 79916762888089401123.....
 > (expt 2 1000)
 1071508607186267320948425049060....
 > (expt 2 10000)
 0

When I try this code with repl.it online interpreter, it works as expected. So what the hell is wrong with Guile?

(Note: On some dialects, remainder function is called as mod.)

转载于:https://stackoverflow.com/questions/14495636/strange-multiplication-behavior-in-guile-scheme-interpreter

  • 写回答

2条回答 默认 最新

  • 笑故挽风 2013-01-24 09:58
    关注

    I recently fixed this bug in Guile 2.0. The bug came into existence when C compilers started optimizing out overflow checks, on the theory that if a signed integer overflow occurs then the behavior is unspecified and thus the compiler can do whatever it likes.

    本回答被题主选为最佳回答 , 对您是否有帮助呢?
    评论
查看更多回答(1条)

报告相同问题?

悬赏问题

  • ¥15 #MATLAB仿真#车辆换道路径规划
  • ¥15 java 操作 elasticsearch 8.1 实现 索引的重建
  • ¥15 数据可视化Python
  • ¥15 要给毕业设计添加扫码登录的功能!!有偿
  • ¥15 kafka 分区副本增加会导致消息丢失或者不可用吗?
  • ¥15 微信公众号自制会员卡没有收款渠道啊
  • ¥100 Jenkins自动化部署—悬赏100元
  • ¥15 关于#python#的问题:求帮写python代码
  • ¥20 MATLAB画图图形出现上下震荡的线条
  • ¥15 关于#windows#的问题:怎么用WIN 11系统的电脑 克隆WIN NT3.51-4.0系统的硬盘