Re: compare-windows - synchronize points

[Juri Linkov <juri@jurta.org>]

Richard Stallman <rms@gnu.org> writes:
> I can see how that would happen.  It suggests an idea to me.  The idea
> is to search for a larger matching substring first, and then try
> smaller matching substrings.  It's the same kind of idea as first
> trying smaller search ranges and then larger ones.

The latest code I posted 2003-08-19 has an implementation of exactly
same idea.  It first doubles search ranges, and if differences are
still not found, then it starts to halve a matching substring size.
So there are no problems with finding differences anymore.  It works
correctly in all cases I tested.

However, I think that the size of matching substring should be
customizable.  I found that current default value 32 (with which a
search is started) is optimal for most cases.  But sometimes user
may want to start with a lesser value to get more fine-grained results.

-- 
http://www.jurta.org/emacs/