Bug #6120: Float and BigDecimal bug in remainder in corner cases

• Status: Closed
• Assignee: marcandre (Marc-Andre Lafortune)
• Target version: -
• ruby -v: r34927
• Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN
• Tags: ext, core

[marcandre (Marc-Andre Lafortune)]

Currently:

4.2.remainder(+Float::INFINITY) # => 4.2, ok
4.2.remainder(-Float::INFINITY) # => NaN, should be 4.2
# (same with all signs reversed)

Reasons the remainder should be 4.2 and not NaN:

1. foo.remainder(bar) == foo.remainder(-bar)
2. foo.remainder(bar) == foo when bar.abs > foo.abs

Similarly:
require 'bigdecimal'
bd = BigDecimal.new("4.2")
bd.remainder(BigDecimal.new("+Infinity")) # => NaN, should be bd
bd.remainder(BigDecimal.new("-Infinity")) # => NaN, should be bd
# (same with all signs reverse)

Reasons: same as float.

Finally:
bd = BigDecimal.new("4.2")
bd.modulo(BigDecimal.new("0")) # => ZeroDivisionError, probably ok?
bd.remainder(BigDecimal.new("0")) # => NaN, should be probably raise a ZeroDivisionError?

Like in #6044, this could be decided either way, as long as there is consistency. Anyone prefer NaN to raising a ZeroDivisionError?

[naruse (Yui NARUSE)]

Ruby's math should portably follow SuS. [ruby-core:28206]
You can fix it for Float/Math if it is obviously wrong and the right implementation is clear.

BigDecimal is little another world and it is up to mrkn.

[shyouhei (Shyouhei Urabe)]

• Status changed from Open to Assigned

[mame (Yusuke Endoh)]

• Target version changed from 2.0.0 to 2.6

Should it be assigned to mrkn?

--
Yusuke Endoh [e-mail]

[marcandre (Marc-Andre Lafortune)]

Didn't get around fixing it for 2.0.0. Will fix and then assign to mrkn for BigDecimal.

[naruse (Yui NARUSE)]

• Target version deleted (2.6)

[jeremyevans0 (Jeremy Evans)]

I've submitted a pull request to fix this issue for Integer/Float: https://github.com/ruby/ruby/pull/4257

I've submitted an issue to the bigdecimal repository to fix the issue in BigDecimal: https://github.com/ruby/bigdecimal/issues/187

• Status changed from Assigned to Closed

Applied in changeset git|aaab3b1de943c3317e115d623ffc7908b4c96578.

---

Fix integer/float remainder with infinity argument of opposite sign

Previously, the result was incorrect:

4.remainder(-Float::INFINITY)
Before: => NaN
After: => 4

4.2.remainder(-Float::INFINITY)
Before: => NaN
After: => 4.2

Fixes [Bug #6120]