Re: Question about display engine

[Eli Zaretskii]

> Cc: Eli Zaretskii <address@hidden>, address@hidden
> From: martin rudalics <address@hidden>
> Date: Fri, 9 Aug 2019 10:59:33 +0200
> 
>  > You are right, I ignored those use cases, but I still don't think that
>  > the faces are the right place to flag that. The line extension maybe
>  > needs to be decided based on another text property. Maybe there are
>  > already some conditions we can check dynamically. Because adding a flag
>  > is a bit error prone when there are already some conditions.
>  >
>  > There is also the case when the face to use comes from FACE_FOR_CHAR or
>  > another is merged over that. Or when there is a highlight inside the
>  > region at the end of the line text... That, in the display engine, I am
>  > not clear yet how are handled.
> 
> OTOH, the display engine could simply delegate some design decisions
> to the face specification apparatus and attribute any faults to the
> latter.

I don't think this is workable, because of face merging.  The actual
face used to display the last character on a screen line can (and
frequently does) come from merging several faces, and there's no
meaningful answer to the question: which face did this attribute come
from?  For a face merged from 2 or more faces defined via defface, how
do you tell whether or not to extend it?  Thus, such delegation can
only yield inconsistent behavior and more bug reports.