Re: Question about display engine

[Ergus]

On Mon, Aug 19, 2019 at 07:50:36PM +0300, Eli Zaretskii wrote:
> > Date: Mon, 19 Aug 2019 18:13:05 +0200
> > From: Ergus <address@hidden>
> > Cc: Eli Zaretskii <address@hidden>, address@hidden
> >
> > Extending the face is a detail that most users won't even notice a
> > change while it works somehow. Actually we have had different behaviors
> > between gui and tui for years and nobody complained up to now.
>
> I think you over-simplify the situation.  First, both GUI and TTY
> frames behave the same when extension of face background color is
> concerned, they only differ in how they handle extension of other
> attributes (of which only the underline is relevant to TTY frames).
> (And I did hear over the years a couple of complaints about how TTY
> frames extend the underline attribute.)
>
> And second, I refer you to the renewed discussion of bug#15934 a day
> or two ago, from which my take is that users will notice and do care
> about such changes in at least a couple of important use cases, as a
> soon-to-be-pushed changes will prove.  So I don't think we can change
> this behavior at will on the assumption that "no one will notice".
Hi Eli:

If soon-to-be-pushed means that you have had already some time to work on
this and it will be fixed before emacs 27 then I'll be happy with that
(I'll be allowed to fix the dfci issue). I was actually wondering
if the discussion was not going anywhere as I didn't see any comment
in a couple of days.

Unrelated with this I wanted to ask you if you think we should continue
with the indentation highlight implementation... because that discussion
never ended. Or you think it does not worth the effort.

I have seen that elpy already have something like what we want to
implement in the lisp level. So they will actually switch for sure to
ours if available. 

What do you think?

> > I am wondering about over-specifications and over-engineering for such a
> > detail, when most of the users only need to extend the background color.
>
> This discussion established that even supporting just the background
> color non-extension as a user option will require to have almost all
> of the machinery in place: the extend bit, the generation of a special
> face without the background color, etc.  And once we have that, adding
> other face attributes to the soup is relatively easy and won't require
> any design changes.