Bug #12780: BigDecimal#round returns different types depending on argument

• Status: Closed
• Assignee: mrkn (Kenta Murata)
• Target version: -
• ruby -v: ruby 2.4.0dev (2016-09-21 trunk 56199) [x86_64-linux]
• Backport: 2.1: UNKNOWN, 2.2: UNKNOWN, 2.3: UNKNOWN
• Tags: ext

[Gat (Dawid Janczak)]

BigDecimal#round returns Integer when no arguments are given and BigDecimal otherwise. I would have assumed the result to always be BigDecimal.

BigDecimal.new('12.34').round # => 12 (Integer)
BigDecimal.new('12.34').round(0) # => 12 (BigDecimal)

Attached is a patch with possible fix.

I found a similar issue to this one (https://bugs.ruby-lang.org/issues/2662), but it's in Japanese. If this should be rejected as well could someone provide an explanation in English please? Also in that case documentation should be changed as it clearly states that the result should be BigDecimal: "Round to the nearest integer (by default), returning the result as a BigDecimal."

[noahgibbs (Noah Gibbs)]

The Japanese issue appears to be saying that there's a problem in RubySpec that will need to be fixed, and that fixing it is a problem with (I think?) early 1.9 compatibility.

[shyouhei (Shyouhei Urabe)]

• Status changed from Open to Assigned
• Assignee set to mrkn (Kenta Murata)

[mrkn (Kenta Murata)]

Float#round returns Fixnum value when the argument is 0 or omitted.

irb(main):005:0> 12.34.round.class
=> Fixnum
irb(main):006:0> 12.34.round(0).class
=> Fixnum

If we will fix BigDecimal#round, we should change the return value for the case of the argument 0 is supplied.

[Gat (Dawid Janczak)]

Good point, I didn't catch that. If it was up to me I'd rather have Float#round always return Float as well for consistency. I could work on a patch for that too if you agree.

[mrkn (Kenta Murata)]

We need matz's approval to change Float#round.

[mrkn (Kenta Murata)]

• Assignee changed from mrkn (Kenta Murata) to matz (Yukihiro Matsumoto)

[shyouhei (Shyouhei Urabe)]

We (mainly mrkn and matz) discussed why Fixnum#round returns Integer today. The answer is because (unlike other languages, namely C's round(3)) in ruby we can safely store all possible return value that round can theoretically generate, in Integer.

BigDecimal#round's returning BigDecimal instance sounds nice. But matz said he need to think a while for this historic Float behaviour.

[mrkn (Kenta Murata)]

• Related to Feature #13420: Integer#{round,floor,ceil,truncate} should always return an integer, not a float added

[matz (Yukihiro Matsumoto)]

#round with the argument less than 1 (n < 1) should return Integer value.

Matz.

[jeremyevans0 (Jeremy Evans)]

• Assignee changed from matz (Yukihiro Matsumoto) to mrkn (Kenta Murata)

matz (Yukihiro Matsumoto) wrote in #note-9:

#round with the argument less than 1 (n < 1) should return Integer value.

I've submitted a pull request that implements this: https://github.com/ruby/bigdecimal/pull/170

[jeremyevans0 (Jeremy Evans)]

• Status changed from Assigned to Closed