From: Kurt Stephens <ks@...>
Date: 2010-02-03T12:17:44+09:00
Subject: [ruby-core:28024] Re: RCR: RUBY_VERSION_INT

I have a Version class that implements Debian version strings.

http://github.com/kstephens/cabar/blob/c9946cdf80d2ed79dfdb902dac085cc83c6dbb45/lib/ruby/cabar/version.rb

KAS

Hugh Sasse wrote:
> On Wed, 3 Feb 2010, Shot (Piotr Szotkowski) wrote:
> 
>> Roger Pack:
>>
>>> Currently in a .rb file, to determine if you're on
>>> 1.9 versus 1.8 is not all that straight forward.
>>> Ex:  To tell you're on 1.9.
>>> RUBY_VERSION > '1.8'
>>> except that that fails
>> Ummm??
>>
>>>> RUBY_VERSION
>> => "1.9.1"
>>>> RUBY_VERSION > '1.8'
>> => true
>>>> RUBY_VERSION > '1.9'
>> => true
>>>> RUBY_VERSION > '1.9.1'
>> => false
>>>> RUBY_VERSION > '1.9.0.9'
>> => true
>>>> '1.9.0' > '1.8'
>> => true
>>>> '1.9' > '1.8'
>> => true
>>
>>> Thoughts?
>> I don?t like the ?one digit? policy, but this is precisely why it was
>> instituted ? to have string comparisons of version numbers ?just work?.
> 
> Having a Version class would solve this
>  RUBY_VERSION > '1.9.0.9'
> ArgumentError: comparison of Version with String failed
>  RUBY_VERSION > '1.9.0.9'.to_version
>  RUBY_VERSION > Version.new('1.9.0.9')
> giving appropriate results.
> I think it would be useful for all sorts of versionable things.
> It could tie in with ranges, with arrays of versions and .include?; 
> and the semantics could well defined.
> 
>> ? Shot
> 
>         Hugh
>