bug#81375: diff-refine-hunk over-refines large unrelated replacement blocks

Umar Ahmad [14/Jul  7:33pm +0530] wrote:
> I can explain or simplify any part of the demo, although understanding
> its implementation should not be necessary as long as the steps for
> running and cycling the policies are clear.  If you can tell me what
> made it difficult to run or work with, I can adjust it accordingly.

The demo is effectively very large.  Looking at it all at once would
require a wall of monitors.  Even looking at two of them side-by-side is
difficult.  So I'm just not sure what to do with it.

I think that it is fine for us to just go ahead and implement
normalized-median-or-longest-run in Emacs, so long as we have a link
back to their discussion with the other policies.  I want it to be that
if we later identify a very concrete problem with that strategy, we can
quickly find the alternatives that you've taken the time to come up
with.

-- 
Sean Whitton