bug#23823: 25.0.95; Reset between highlight buffer/file comparisons

[Tino Calancha <f92capac@gmail.com>]

On 06/23/2016 12:16 AM, Eli Zaretskii wrote:
> What I see is 'f' in a "deleted" face.  Which seems reasonable to me.
>
> Whether a change that countermands another change should count as a
> change is a philosophical question that AFAIK doesn't have a
> one-fits-all answer.
The way i understand my example is as follows:

1) An user compare current buffer (buf-a) with file-b.
;; buf-a content: boo
;; file-b content: foo
;; they are different: new faces shown their differences.

2) The user update buf-a and perform a new comparison with file-b.
;; buf-a content: foo
;; file-b content: foo
;; same content, but is not obvious because the result of the
;; previous comparison is not removed.

My point is: why don't we perform a fresh comparison in 2?