Re: Fill column indicator functionality

[Ergus]

On Fri, Mar 15, 2019 at 09:52:53AM +0200, Eli Zaretskii wrote:
> > Date: Thu, 14 Mar 2019 22:13:15 +0100
> > From: Ergus <address@hidden>
> Can you solve this problem by using the :height attribute whose value
> is 1.0 (a floating-point number) for the face of the character?  If
> that doesn't work, we might introduce a special feature for this use
> case.

I'll try this in a while to see how the experience changes. If we set
that this way the font will autoresize depending of the height of the
current column?

Anyway I think the discussion in progress can give a different answer. I
will wait a bit more to get a recommendation. Or a solution that promise
to be better.

> In any case, I think this use case is rare enough to leave it for
> subsequent improvements.  It doesn't sound as show-stopper to me.
>
> > Or use a different type of glyph in graphical mode... maybe an image as
> > the fci package does? Is that possible?
>
> It should be possible to use an image, yes.  Does the fci package
> solve the use case of lines that are taller than default, as in Info
> or EWW?  If so, what do they do -- use a larger slice of an image
> according to the line height or something?  Because using an image
> still needs to solve the problem of displaying a taller image when the
> line's height is larger than the default.

The XPM image seems to be a good enough approach according to Alp's
comment/experience, but it only solves a part of the problem. But I
haven't go in the details of what they do and I don't want to over
specify the functionality or produce too much overhead.

> One more potential issue that I think you should test is when the
> fringes are disabled on GUI frames, and we usurp the last column of
> the text area to display continuation and truncation glyphs there, as
> we do on TTY frames.  If the fill-column is equal to the last column
> of the window, or to one before the last, I think you will need to
> augment your comparison with it->last_visible_x in this case.
The code that adds the information looks to run later, so the line is
hidden in that case, which for me is the expected behavior. Otherwise I
didn't understand the issue.

Apart from that there, are some corrections I made in order to fix some
background issues I observed in graphical mode.

In graphical interfaces the space after the line is always filled with
the background color of the face of the last produced glyph, even if I
reset it to the saved value after the generation. The only solution I
found so far was to add an extra glyph after reset the face to the
default (saved) value, but hopefully there is a better way?