On 22.11.2012 18:29, Branko Čibej wrote: > On 22.11.2012 17:01, Julian Foad wrote: >> Just use 'float' and [0..1]? I'm thinking we may want to use this for >> comparing diff hunks, and there may be more than 500 characters in each hunk >> and it matters whether just one character has been changed. > Could do that, too. Harder to test though. :) It's trivial to extend the > range to [0..10**6] or even 10**9 if that's a problem. Frankly it feels > nice to keep the whole thing integral and not leave floating-point > messes around.
Actually I'm being silly. The function already provides the actual LCS length, so if 3 significant digits aren't enough, you can always compare that to the length of the hunk. -- Brane -- Branko Čibej Director of Subversion | WANdisco | www.wandisco.com
