Re: Identifying the face between STRETCH and right fringe.

[Eli Zaretskii <eliz@gnu.org>]

> From: Robert Pluim <rpluim@gmail.com>
> Date: Tue, 27 Nov 2018 20:38:05 +0100
> 
> (showing my ignorance of redisplay)

Everybody is "ignorant" about redisplay, including your truly.  E.g.,
I learned just a couple of weeks ago something quite important about
it that I succeeded to miss in the last 10 years I'm actively hacking
that code: that try_window_id calls display update routines directly.

> Is there a reason we need to compare with the background of the
> frame rather than with the background of the default face of the
> frame?

Basically, because we clear rectangles on display using the frame's
background color.

Historically, once upon a time, the frame's background was not
necessarily identical to that of the default face.  And given that we
already have the color recorded in the frame, it is less expensive to
access that than look up the default face, which might involve
accessing face-remapping-alist etc.  Finally, the frame's face cache,
which is where we look up faces, is reset from time to time, whereas
the frame's background color is always valid.