bug#29721: 26.0; doc string of `invisible-p'

[Eli Zaretskii <eliz@gnu.org>]

> Date: Fri, 15 Dec 2017 09:41:55 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> 
> The doc string is unclear.
> 
>   "(invisible-p POS-OR-PROP)
> 
>    Non-nil if the property makes the text invisible."
> 
> What property?  What text?

Thanks, fixed.

>   "The non-nil value returned can be t for truly invisible text or
>    something else if the text is replaced by an ellipsis."
> 
> Huh?  What is "truly invisible"?

"Truly" because text with 'invisible' property could be still visible,
depending on buffer-invisibility-spec.  I added the reference to that.

> What "something else"?  Does "something else" include `nil'?  [...]
> What does the (unspecified!) value signify in the case of "the text"
> (?) being "replaced by an ellipsis"?

The value itself is of no significance, besides the fact that it's not
nil and not t, so I see no reason to document what it is, exactly.  If
you are really interested, you can read the code, or try calling that
function on invisible text replaced with an ellipsis, then you will
see that the value is truly of no importance.

> What does it mean for "the text" to be replaced by an ellipsis?

That is a well-known feature, see (elisp)Invisible Text.  I hope a
reference to buffer-invisibility-spec makes it even more clear.