VersionedFile.walk deprecated?

[Martin Pool]

On 20 Apr 2006, John Yates <jyates at netezza.com> wrote:
> The matching algorithm repeatedly eliminates maximal regions of
> identical text from both files.  The work list invariant is that for any
> line on the list there remains in each file exactly one "unmatched"
> instance.  That invariant is incrementally maintained during the growth
> of a matched region.  I fail to see how the order of selecting lines
> from the work list can alter the final result.

suppose you are comparing

  1x2y3z4
  1 2x3y4z

[1234xyz] would all be on the work list; it would be plausible to choose
either sequence as common but you get different results either way.
We try to look for the longest subsequence to get the best result.

-- 
Martin