Bug #15312: Ruby gives me wrong result for plus operation

• Status: Rejected
• Assignee: -
• Target version: -
• Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN

[kot-begemot (Max Zet)]

Run into this weird issue today. Tried that on rubies 2.5.1p57 and ruby 2.5.3p105 (i believe this is latest stable at the moment).

Here is how to reproduce it

$ irb
2.5.3 :001 >  2.6 + 0.1
 => 2.7 
2.5.3 :002 >  2.7 + 0.1
 => 2.8000000000000003 
2.5.3 :003 >  2.8 + 0.1
 => 2.9

Here some more cases

$ irb
2.5.3 :001 > 2.7 + 0.01
 => 2.71 
2.5.3 :002 > 2.79 + 0.01
 => 2.8 
2.5.3 :003 >  2.8 - 0.1
 => 2.6999999999999997 
2.5.3 :004 > 2.79 + 0.01
 => 2.8 
2.5.3 :005 > 2.81 - 0.01
 => 2.8000000000000003 

Haven't discovered other cases yet, if will catch them, will post them here

This is not a bug, that's how floating point arithmetic works in accordance with the IEEE 754 standard.

You can read more about that on Wikipedia or in this article: https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

To get mathematically correct results you should be using BigDecimal or Rational.

require 'bigdecimal'
BigDecimal('2.7') + BigDecimal('0.1') #=> 0.28e1

[marcandre (Marc-Andre Lafortune)]

• Status changed from Open to Rejected

See also https://0.30000000000000004.com/

[nobu (Nobuyoshi Nakada)]

You can write rational numbers by r suffix.

p 0.1r + 0.2r #=> (3/10)

As I found these literals (r and i) have not been mentioned in doc/syntax/, added them now (r65773).

[nobu (Nobuyoshi Nakada)]

• Description updated (diff)