Monkey Patching Float Infix Operator's Produces Unexpected Results

Question

Redefining Float#/ appears to have no effect:

class Float
  def /(other)
    "magic!"
  end
 end
 puts 10.0/2.0 # => 5.0

But when another infix operator Float#* is redefined, Float#/ suddenly takes on the new definition:

class Float
  def /(other)
    "magic!"
  end
  def *(other)
    "spooky"
  end
end
puts 10.0/2.0 # => "magic!"

I would love to hear if anybody can explain the source of this behavior and if anybody else gets the same results.

Ruby: ruby 2.0.0p353 (2013-11-22) [x64-mingw32]

To quickly confirm the bug, run this script.

Confirmed on Ubuntu 13.04 with Ruby 2.0.0p247. Perhaps this out to be listed as a bug? [Ruby Issue Tracker](https://bugs.ruby-lang.org/) — Dan Grahn, Dec 10 '13 at 17:29
For me both are giving `"magic"` output... I am on `kirti@kirti-Aspire-5733Z:~$ ruby -v ruby 2.0.0p0 (2013-02-24 revision 39474) [i686-linux] kirti@kirti-Aspire-5733Z:~$ ` — Arup Rakshit, Dec 10 '13 at 17:29
Interestingly, this happens only for `Float`. I just tested with `Fixnum` — Dan Grahn, Dec 10 '13 at 17:36
Interestingly, while `10.0/2.0` returns a Float, `10.0.send(:/,2.0)` returns `"magic!"` — Peter Alfvin, Dec 10 '13 at 17:41
Reported. [Ruby Issue Tracker](https://bugs.ruby-lang.org/issues/9237) — ChrisTimps, Dec 10 '13 at 17:44
I got the same result on Linux, and since all three major platforms (Windows, Mac OS, Linux) behave the same, it looks the OS is irrelevant. I removed the OS description from the question, accordingly. — sawa, Dec 10 '13 at 17:44
@screenmutt No, it is not resolved. I ran it in 2.1.0dev (2013-11-23 trunk 43807), and got the same result. — sawa, Dec 10 '13 at 17:52
@screenmutt: I was using 2.0.0p247, and confirmed bug in two separate IRB sessions (running no other code) — Neil Slater, Dec 10 '13 at 18:02
@PeterAlfvin: Indeed. The bug was in the YARV optimization code. The optimization is not used when using `send`. I talk about a similar bug in my rubyconf talk http://www.confreaks.com/videos/2894-rubyconf2013-ruby-core-dilemmas — Marc-André Lafortune, Dec 10 '13 at 18:32
@Marc-AndréLafortune Just watched your talk. Delightful! Kudos to you (for your work and the talk) and thanks for the link. :-) — Peter Alfvin, Dec 10 '13 at 19:45

score 3 · Accepted Answer · edited Dec 10 '13 at 18:48

3

This appears to be a bug in the implementation of Ruby. A bug report has been filed here.

In the mean time, you may either switch implementations or switch versions. 1.8.7 appears to be bug free.

EDIT

This bug was fixed with revision 44127

edited Dec 10 '13 at 18:48

sawa

165,429
45
277
381

answered Dec 10 '13 at 17:53

Dan Grahn

9,044
4
37
74

What is the basis of your claim that 2.1 is bug free? – sawa Dec 10 '13 at 17:54
@sawa IRB was bug free in 2.1 for me, but when I ran it as a script the bug was present. – Dan Grahn Dec 10 '13 at 17:55
1

Yup :-) Cute bug! (and fixed...) – Marc-André Lafortune Dec 10 '13 at 18:26

Monkey Patching Float Infix Operator's Produces Unexpected Results

1 Answers1