From: Ross Bamford <rossrt@...>
Date: 2006-03-25T19:47:20+09:00
Subject: Re: require a certain version of the ruby interpreter

On Sat, 2006-03-25 at 19:08 +0900, Robert Klemme wrote:
> Ross Bamford wrote:
> > On Sat, 2006-03-25 at 11:06 +0900, dblack@wobblini.net wrote:
> >> (What, no inject? :-)  I don't think Array has >= though.
> 
> Actually, a version with #inject occurred to me after I sent the other 
> one out when I thought about reducing redundancy in the code above - 
> you're also right about Array (strange though as it does implement <=>):
> 
> def ev(ver)
>    raise "Requiring at least #{ver}" unless
>      [RUBY_VERSION,ver].map! {|v| v.scan(/\d+/).map! {|x|x.to_i}}.
>        inject {|v1,v2| (v1<=>v2)>=0}
> end
> 

Ahh, there it is - I knew inject would fit in here somewhere :) Array
implementing <=> is a nice little thing, I guess then (again given no
two digit versions as Matz has promised) it could be:

def ev(v)
  raise "need #{v}" unless (VERSION.split('.') <=> v.split('.')) >= 0
end

> > :) Even I couldn't find an excuse for inject on this one. The best I
> > could do was:
> > 
> > def ensure_version(v)
> >   raise "need #{v}" if VERSION.split('.').zip(v.split('.')).any? {|a,b| a < b}
> > end
> 
> Nice, too!  I like the approach with any?.
> 
> > (given that we don't need to worry about two-digit versions of course).
> 
> You could still to the .to_i when comparing.

True, but at 78 characters and the limit of whitespace decency, I
couldn't make it fit ;)

-- 
Ross Bamford - rosco@roscopeco.REMOVE.co.uk