bug#59141: 28.1.90; Face :extend when all the line but trailing \n is invisible

[Juri Linkov]

>> I still don't understand is does magit use outline-mode?
>
> It does not; it relies on magit-section (see references a couple of
> messages prior), which does more or less the exact same job outline.el
> does: let the user navigate and un/fold a hierarchy of headings.
>
> Some recent features of outline.el have indeed been in magit-section for
> a long time: visibility cycling with TAB/S-TAB, bitmap fringe indicators
> (with fallback to outline-style ellipses e.g. on TTYs).
>
> My point with this comparison is to show that an outline-like UI with
> :extended backgrounds is obviously possible; in my previous messages, I
> tried to highlight the relevant code in magit-section that handles
> delimiting section headings vs content and setting the overlays.
>
> I did that mainly FTR, so that Someone™ with motivation and time can see
> if outline.el could grow a user option to support a similar way to
> display outlines, thus solving the problem of :extended backgrounds.

I haven't looked at the magit-section source code.  I once tried
to copy the syntax highlighting code from diff-mode to magit-diff,
but magit code is such a mess that I abandoned the attempt.

But from your screenshots it's clear what is needed to do
to achieve the same in outline(-minor)-mode:

1. to support the :extended face attribute on the outline heading lines,
newlines should be included in the match.  This is not a patch,
but only shows possible changes:

@@ -242,7 +242,7 @@ outline-font-lock-keywords
   '(
     ;; Highlight headings according to the level.
     (eval . (list (or outline-search-function
-                      (concat "^\\(?:" outline-regexp "\\).*"))
+                      (concat "^\\(?:" outline-regexp "\\).*\n"))
                   0 '(if outline-minor-mode
                          (if outline-minor-mode-highlight
                              (list 'face (outline-font-lock-face)))
@@ -486,7 +486,7 @@ outline-minor-mode-highlight-buffer
   (save-excursion
     (goto-char (point-min))
     (let ((regexp (unless outline-search-function
-                    (concat "^\\(?:" outline-regexp "\\).*$"))))
+                    (concat "^\\(?:" outline-regexp "\\).*\n"))))
       (while (if outline-search-function
                  (funcall outline-search-function)
                (re-search-forward regexp nil t))

Maybe such changes are not needed when a function in
outline-search-function could include newlines in the match.

2. not to hide the newline of the outline heading line,
overlay boundaries could be shifted forward by 1:

@@ -960,7 +960,7 @@ outline-flag-region
     ;; We use `front-advance' here because the invisible text begins at the
     ;; very end of the heading, before the newline, so text inserted at FROM
     ;; belongs to the heading rather than to the entry.
-    (let ((o (make-overlay from to nil 'front-advance)))
+    (let ((o (make-overlay (1+ from) (1+ to) nil 'front-advance)))
       (overlay-put o 'evaporate t)
       (overlay-put o 'invisible 'outline)
       (overlay-put o 'isearch-open-invisible

Maybe this could be conditional via a new option that you proposed.

3. Your screenshot shows that magit doesn't use an ellipsis.
And indeed, ellipses get in the way.  But we need to find a way
to disable them without breaking this feature.  The line

  (overlay-put o 'invisible 'outline)

either should be replaced with

  (overlay-put o 'invisible t)

or the ellipsis glyph to be disabled with something like:

  (or standard-display-table (setq standard-display-table (make-display-table)))
  (set-char-table-extra-slot standard-display-table 4 (vector))