Re: extend diff3 to ignore whitespace differences - here's a patch

[Paul Eggert]

The problem I see is that diff3 currently uses a mishmash of lines
from the three input lines, and it takes advantage of the fact that
when there is no difference, then the input lines have to be
identical.  Once you make it possible for the input lines to be
considered to be the same even when they differ, this implementation
model breaks down.  It's very hard to explain what the resulting
output will look like.

I don't see an easy answer to this problem, unfortunately.