From: "H.Yamamoto" <ocean@...2.ccsnet.ne.jp>
Date: 2006-02-04T17:20:00+09:00
Subject: Re: [ ruby-Bugs-3438 ] (2**32).to_s(2) fails with exception on Redhat Opteron with gcc 3.4.3 compile

Hello.

>>Assigned to: Nobody (None)
>>Summary: (2**32).to_s(2) fails with exception on Redhat Opteron with gcc
>3.4.3
>>compile
>>
>>Initial Comment:
>>Sorry I don't have the complete exception to provide here.
>>The computer is at work and not connected to the Internet.
>>
>>The Summary pretty much says it all.
>>
>>(0..64).each {|i| print i,"\t",(2**i).to_s(2),"\n" rescue next}
>
>   I get same failure with solaris 10, sunstudio 11 compiler. My ruby
>  is 64 bit as well.
>>
>>If I was a bettin' man, I'd bet that you'd find some 64-bit integer
>>somewhere being incorrectly cast or squeezed into a 32-bit integer.
>>When Bignum takes over, all the conversions work properly.
>
>            I will not bet against you. There is probably some 'long'
>problem.
>
>>
>>This problem also makes me wonder what other surprises are waiting
>>for 64-bit platform users.
>
>      At least sun studio 11 spits many warnings about interger overlfows
>      and other messages. I can post them here.


I have fixed this problem on CVS. Please try latest ruby_1_8 branch.

# I think we should clarify the type of interger. rb_int2big is misleading.
# (this actually deals with `long` not `int`)
#
# And maybe this should be fixed at same time?
# http://rubyforge.org/tracker/index.php?func=detail&aid=1823&group_id=426&atid=1698