Re: Q: Invisible text spec + define minor mode macro

[Tassilo Horn <tsdh@gnu.org>]

Arthur Miller <arthur.miller@live.com> writes:

Hi Arthur,

> After playing a bit more: if I scroll through entire file before I
> enable my mode, then entire file is processed correctly.  But if file
> is not scrolled through, then only 3½ screen pages are processed.  Has
> nothing to do with regex I use, just with text being shown on screen
> or not. I don't find anything in docs about that. A bug? Or
> undocumented behaviour? Or something else I am missing here?

I've tried it and can confirm your observations.  Your code is correct,
though.  The funcion `hbm--update-line' is called 400 times at the right
positions and puts the 'invisible text property there.

Then I augmented that function to spit out the `buffer-substring' where
it puts the property, and that showed that lines became invisible where
the text was already fontified such as

  #("#+end_src" 0 4 (fontified t font-lock-fontified t
   font-lock-multiline t face org-block-end-line wrap-prefix #("#" 0 1
   (face org-block-end-line font-lock-multiline t font-lock-fontified t
   fontified t))) 4 9 (fontified t font-lock-fontified t
   font-lock-multiline t face org-block-end-line wrap-prefix #("#" 0 1
   (face org-block-end-line font-lock-multiline t font-lock-fontified t
   fontified t))))

and stopped working at the next line

  #("#+begin_src emacs-lisp" 0 22 (fontified nil))

which is not yet fontified.  And indeed, as soon as you disable
`font-lock-mode' your minor mode starts working fine.  (With
`font-lock-mode', you can "unhide" a line also by editing in it which
will trigger refontification of that region.)

The problem is in Org: it has tons of fontification functions such as
`org-fontify-meta-lines-and-blocks' which calls
`org-fontify-meta-lines-and-blocks-1' which removes `display',
`invisible', and `intangible' text properties.

So it seems you cannot use those properties in a sensible way in org
files...

Bye,
Tassilo