2010/4/21 Tanaka Akira <akr / fsij.org>:
> 2010/4/21 Yusuke Endoh <redmine / ruby-lang.org>:
>
>>> 1) Seeding sometimes ignores the high bit of seed data.
>>>
>>> =A0 =A0Random.new((1<<64)-1).rand =3D=3D Random.new((1 << 65) -1).rand
>>> =A0 =A0 # =3D> true, should be false
>>
>> I really agree that this is uncool.
>
>> Incidentally, I prefer removing MSB to current implementation.
>
> I guess it is worse in the sense of the reporter.
>
> Probably it causes
> Random.new(2**32).rand =3D=3D
> Random.new(2**33).rand =3D=3D
> Random.new(2**34).rand =3D=3D
> ...
> Random.new(2**63).rand.

You are right :-)

--=20
Yusuke Endoh <mame / tsg.ne.jp>