' > > So a range of nil..nil throws an error in 1.8.7 but does not in 2.0.0. > While it is accepted in 2.0.0, nothing meaningful may be done with it. > I'd highly prefer the 1.8.7 behaviour of throwing an error when > constructing such a range. > > What are your opinions? > > Regards, > > Pascal > > > --047d7b339d5b083b3f04d8fcd1a7 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable

In 1.9+, you can indeed create ranges that are not very me= aningful with any objects. The method `cover?` works, though.

=A0 =A0 o =3D Object.new

=A0 =A0 r =3D o..o

=A0 = =A0 r.cover?(o) # =3D> true

I don't see a good way to forbid `nil..nil` and rem= ain generic, though.

You should open a feature= request if you feel strongly about this.

On Thu, Mar 28, 2013 at 6:44 AM, Pascal = Hurni <hurnip@gmail.com> wrote:

Hi,

I came accross a bizarre behaviour when porting my code from 1.8.7 to 2.0.0= .

Here are irb sessions:

>> RUBY_VERSION
=3D> "1.8.7"
>> nil..nil
ArgumentError: bad value for range
=A0 =A0 from (irb):2

2.0.0p0 :002 > range =3D nil..nil
=A0=3D> nil..nil
2.0.0p0 :003 > range.include? nil
TypeError: can't iterate from NilClass
=A0 =A0 from (irb):3:in `each'
=A0 =A0 from (irb):3:in `include?'
=A0 =A0 from (irb):3:in `include?'
=A0 =A0 from (irb):3
=A0 =A0 from /Users/phi/.rvm/rubies/ruby-2.0.0-p0/bin/irb:16:in `<= ;main>'

So a range of nil..nil throws an error in 1.8.7 but does not in 2.0.0. Whil= e it is accepted in 2.0.0, nothing meaningful may be done with it.
I'd highly prefer the 1.8.7 behaviour of throwing an error when constru= cting such a range.

What are your opinions?

Regards,

Pascal

--047d7b339d5b083b3f04d8fcd1a7--