Re: Ruby 1.[89].\d+ and beyond.

On Nov 16, 2006, at 9:18 AM, Hugh Sasse wrote:
> Yes: much of the code that supports this uses the RUBY_VERSION  
> constant,
> which is a string.  Thus string comparisons are done in most cases  
> where
> it is used, at present, rather than
>   major, minor, teeny = RUBY_VERSION.split(/\./).map{|x| x.to_i}
> or similar.   Lexicographic comparisons will fail for more digits
> between the dots.

   It seems like something like this would maybe work:

	class Version
		def <=>(other)
			other = Version.new(other) if String === other
			... componentized compare ...
		end
		def to_s
			... build X.Y.Z ...
		end
		def method_missing
			... invoke on self.to_s for other String methods ...
		end
	end

	class String
		... alias <=> to old_compare ...

		def <=>(other)
			return other <=> self if Version === other
			old_compare(other)
		end
	end

   The hack on String is fugly, though; just a lame multi-method, but  
it'd slow down every string compare... :(

   A simpler option might be to change string comparison to do  
numerical sorting for characters
that are ranges of numbers, but this would probably have too many side  
effects ... and it would be "interesting"
trying to work with non-Arabic digits.

   Anyway, this solution might be worse than the problem.

-tim