[BUG] Fix for inlinetask visibility cycling causes infinite loop for "degenerate" inlinetasks

[BUG] Fix for inlinetask visibility cycling causes infinite loop for "degenerate" inlinetasks

[Anders Johansson]

[-- Attachment #1: Type: text/plain, Size: 1115 bytes --]

Hi,
I use "degenerate" inlinetasks (without an "*** END" line) quite a lot (for
example for "coding" snippets in my orgqda package
https://git.sr.ht/~andersjohansson/orgqda/).

Degenerate inlinetasks are stated as allowed in the documentation at the
beginning of org-inlinetask.el, but are maybe not so common since
org-inlinetask-insert-task always inserts the END-line.

Since April, an infinite loop is caused when cycling an entry containing
degenerate inlinetasks by commit:
5f184b org-inlinetask.el: Fix visibility cycling for inlinetasks

Sample for reproducing:
-----
* Heading
*************** Degenerate inlinetask
-----

Cycling "Heading" will call org-inlinetask-hide-tasks where the last calls
in the loop of `inlinetask-goto-end` takes us to the line below the
inlinetask, but `backward-char` takes us back to the point at the end of
the inlinetask, so `inlinetask-at-task-p` is still t and the loop continues
infinitely in this fashion.

I don’t have a good suggestion for a solution that would still solve the
problem addressed in that commit.

Best,
Anders Johansson

[-- Attachment #2: Type: text/html, Size: 1438 bytes --]

Re: [BUG] Fix for inlinetask visibility cycling causes infinite loop for

[Anders Johansson]

Something like this could work. Just ensuring that we haven’t ended back at the start of the current (degenerate) inlinetask. I don’t know if this would be the most elegant solution though. I also considered whether org-inlinetask-goto-end should really move to the point after the last line of an inlinetask (is this consistent with org-element BTW?), but that would require changes across a big part of org-inlinetask and probably complicate other things.



diff --git a/lisp/org-inlinetask.el b/lisp/org-inlinetask.el
index 07aa94f32..c28f5e7b4 100644
--- a/lisp/org-inlinetask.el
+++ b/lisp/org-inlinetask.el
@@ -334,12 +334,15 @@ (defun org-inlinetask-hide-tasks (state)
 	   (org-inlinetask-goto-end)))))
     (`children
      (save-excursion
-       (while
-	   (or (org-inlinetask-at-task-p)
-	       (and (outline-next-heading) (org-inlinetask-at-task-p)))
-	 (org-inlinetask-toggle-visibility)
-	 (org-inlinetask-goto-end)
-         (backward-char))))))
+       (let ((startpoint-at-bol -1))
+         (while
+	     (or (and (org-inlinetask-at-task-p)
+                      (not (eq startpoint-at-bol (point-at-bol))))
+	         (and (outline-next-heading) (org-inlinetask-at-task-p)))
+           (setq startpoint-at-bol (point-at-bol))
+	   (org-inlinetask-toggle-visibility)
+	   (org-inlinetask-goto-end)
+           (backward-char)))))))
 
 (defun org-inlinetask-remove-END-maybe ()
   "Remove an END line when present."

Re: [BUG] Fix for inlinetask visibility cycling causes infinite loop for "degenerate" inlinetasks

[Ihor Radchenko]

Anders Johansson <mejlaandersj@gmail.com> writes:

> Since April, an infinite loop is caused when cycling an entry containing
> degenerate inlinetasks by commit:
> 5f184b org-inlinetask.el: Fix visibility cycling for inlinetasks
>
> Sample for reproducing:
> -----
> * Heading
> *************** Degenerate inlinetask
> -----

I am unable to reproduce on main and on bugfix.
What is your Org version (M-x org-version)?

-- 
Ihor Radchenko,
Org mode contributor,
Learn more about Org mode at https://orgmode.org/.
Support Org development at https://liberapay.com/org-mode,
or support my work at https://liberapay.com/yantar92

Re: [BUG] Fix for inlinetask visibility cycling causes infinite loop for "degenerate" inlinetasks

[Ihor Radchenko]

Anders Johansson <mejlaandersj@gmail.com> writes:

>> I am unable to reproduce on main and on bugfix.
>> What is your Org version (M-x org-version)?
>
> Sorry I didn’t bother to reproduce with emacs -Q. I stepped through with
> edebug and thought the problem was quite clear that way. I have reproduced
> with Emacs -q (+ loading org from git) now and the problem persists on my
> side, using a sample file consisting of just a heading with one degenerate
> inlinetask below, as posted before:

Thanks for checking again!
I forgot to load org-inlinetask when checking, which is why I did not
observe the problem.

Fixed on main now via a0b8b7303.
https://git.savannah.gnu.org/cgit/emacs/org-mode.git/commit/?id=a0b8b7303b9683fae4503c57ee8ae191cb89a297

Best,
Ihor