On Thu, Jun 8, 2017 at 5:10 PM Nicolas Goaziou <mail@nicolasgoaziou.fr> wrote:

> Test test-org/custom-properties condition:
> (ert-test-failed
> ((should
> (let
> (...)
> (org-test-with-temp-text "* H
> :PROPERTIES:
> <point>:FOO: val
> :END:
> " ... ...)))
> :form
> (let
> ((org-custom-properties ...))
> (org-test-with-temp-text "* H
> :PROPERTIES:
> <point>:FOO: val
> :END:
> "
> (org-toggle-custom-properties-visibility)
> (org-invisible-p2)))
> :value nil))
> FAILED 646/734 test-org/custom-properties
>
> (ert-test-failed
> ((should
> (org-test-with-temp-text "#+BEGIN_CENTER
> P1
>
> P2
> #+END_CENTER
> P3"
> (org-hide-block-toggle)
> (org-forward-paragraph)
> (looking-at "P3")))
> :form
> (let
> ((inside-text ...)
> (org-mode-hook nil))
> (with-temp-buffer
> (org-mode)
> (let ... ...)
> (org-hide-block-toggle)
> (org-forward-paragraph)
> (looking-at "P3")))
> :value nil))
> FAILED 668/734 test-org/forward-paragraph

These one are related to invisible text. I don't what is going to change
in this area in next Emacs release.

This commit in emacs master is causing this failure: http://git.savannah.gnu.org/cgit/emacs.git/commit/?id=9cc59ffbbb2f20fbbf1c72d2e0c9dc47c7906a99

Earlier outline-invisible-p simply returned the value of (get-char-property (or pos (point)) 'invisible).

But now (in emacs master), it does a check if that value is equal to 'outline.

That test fails as (get-char-property (or pos (point)) 'invisible) returns 'org-hide-block, not 'outline.

Now outline-invisible-p is used at many places!

lisp/org-archive.el

227: cl (if (outline-invisible-p) (org-end-of-subtree nil t))))

425: cl (if (outline-invisible-p) (org-end-of-subtree nil t))))

573: cl (if (outline-invisible-p) (org-end-of-subtree nil t))))

594: cl (if (outline-invisible-p) (org-end-of-subtree nil t))))

lisp/org-clock.el

2965: (when (outline-invisible-p) (org-show-context))))))

lisp/org.el

6926: (when (outline-invisible-p) (org-flag-heading nil))))

6960: (when (outline-invisible-p) (org-flag-heading nil)))

7134: (when (and (not (outline-invisible-p))

7136: (goto-char (point-at-eol)) (outline-invisible-p)))

7489: (when (or (outline-invisible-p) (org-invisible-p2))

7530: (when (outline-invisible-p)

8387: (setq folded (outline-invisible-p)))

8483: (setq folded (outline-invisible-p)))

8536: (let* ((visp (not (outline-invisible-p)))

8593: (when (and (outline-invisible-p) visp)

10441: (when (outline-invisible-p) (org-show-context)))

11144: (when (or (outline-invisible-p) (org-invisible-p2)) (org-show-context 'mark-goto))))

12395: cl (when (outline-invisible-p) (org-end-of-subtree nil t))))

13323: (lambda () (when (outline-invisible-p) (org-end-of-subtree nil t))))

13340: (lambda () (when (outline-invisible-p) (org-end-of-subtree nil t))))

14887: '(when (outline-invisible-p) (org-end-of-subtree nil t))))

23850: ;; Early versions of noutline don't have `outline-invisible-p'.

23852: (outline-invisible-p)))

23858: ;; Early versions of noutline don't have `outline-invisible-p'.

23859: (outline-invisible-p)))

24129: (not (outline-invisible-p

24265: ((outline-invisible-p (line-end-position))

24352: (when (outline-invisible-p (point)) (beginning-of-visual-line))))

24429: (when (outline-invisible-p (line-end-position)) (org-cycle))

24910: (or (outline-invisible-p)

24912: (outline-invisible-p)))

24918: (outline-invisible-p))

lisp/org-list.el

158:(declare-function outline-invisible-p "outline" (&optional pos))

2260: (outline-invisible-p)))

lisp/org-crypt.el

179: (let ((folded (outline-invisible-p))

207: (outline-invisible-p))))

contrib/lisp/org-drill.el

1458:;; (when (and (not (outline-invisible-p))

1483: (when (and (not (outline-invisible-p))

One solution would be to have a new function "org-invisible-p" that restores the old definition of outline-invisible-p. Interestingly I find that an "org-invisible-p" did exist back then, but it was replaced with outline-invisible-p in 2011 ( http://orgmode.org/cgit.cgi/org-mode.git/commit/?id=84d7165d74a5061413168af435d61453be217933 ). Looks like that might need reverting.

I also find an org-invisible-p2 function, but it's not clear why it does a (backward-char 1) before doing invisibility check.. and that too eventually relies on outline-visible-p. (Turns out this function has a long history: http://orgmode.org/cgit.cgi/org-mode.git/commit/?id=93a4128a6fe47b4e05c0a7cb3ec14878b41d6000 )

I am copying Bastien and Paul as they would know the history behind the above referenced emacs master commit.