Issue #7352 has been updated by mame (Yusuke Endoh).


phasis68 (Heesob Park) wrote:
> I guess this bug is related with the optimization of GCC 4.7.2.
> The version built with VC 2010 or GCC 4.5.2 works fine.
> The workaround is adding a volatile qualifier.
> 
> Additionally, I found some problems in Range#bsearch method.

I've committed your patch, my MEGA thanks!  

I would really like you to have a commit bit.  Are you willing?

Matz, what do you thing?  He is a great all-round player; his
contribution to Ruby includes deadlock issues (his patches was
applied by kosaki), windows issues (accepted by usa and naruse),
algorithmic issue (accepted by mrkn and me), documentation
issues (by nobu), etc.

  $ grep -i heesob ChangeLog doc/ChangeLog-* | wc -l
  18

-- 
Yusuke Endoh <mame / tsg.ne.jp>
----------------------------------------
Bug #7352: Array#bsearch test failure on Range (32bits MinGW)
https://bugs.ruby-lang.org/issues/7352#change-32931

Author: luislavena (Luis Lavena)
Status: Closed
Priority: Normal
Assignee: mame (Yusuke Endoh)
Category: test
Target version: 2.0.0
ruby -v: ruby 2.0.0dev (2012-11-15 trunk 37656) [i386-mingw32]


=begin
Hello,

After r37655, I noticed a failing test on RubyInstaller CI for x86:

 test_bsearch_for_float(TestRange) [C:/Users/Luis/Code/ruby/ruby/test/ruby/test_range.rb:392]:
 Expected -1.7976931348623157e+308 to be >= NaN.

This do not fail on x64, both running with GCC 4.7.2

I don't have Linux test results to compare at this time, but wanted to raise awareness of this issue.

Thank you
=end


-- 
http://bugs.ruby-lang.org/