Re: 22.1.50; newsticker and buffer-invisibility-spec:

[Stefan Monnier]

> I've not used newsticker in a while, and I noticed today that there seems
> to be  a problem with the way buffer-invisibility-spec is getting handled
> in newsticker buffers.

> Could someone who can see the display look at it and confirm that
> there is a problem, or that it is indeed doing the right thing so
> I can chase this down further?

I don't know how to run the newsticker code to try it out, but the code
looks odd indeed: it presumes that the invisibility property can be a list
of values and that if one of those values is a member of
buffer-invisibility-spec then the text is invisible.  But in reality, the
invisible property is not treated as a list, its value (which could
presumably be a list, but that would be asking for trouble) is directly
looked up in buffer-invisibility-spec.
I.e. newsticker--lists-intersect-p (which is always called with
(get-text-property FOO 'invisible) as first arg and buffer-invisibility-spec
as second, so it should really be replaced by a new function
newsticker--invisible-p) should be newsticker--list-member-p or somesuch.

But fixing this function is not enough because the rest of the code makes
use of the nonexistent feature.  I've looked at the XEmacs-21.4.20 docstring
for buffer-invisibility-spec and it doesn't seem to document such an
extension either, so I have no idea in which circumstance this code
may work.

The problematic code was part of the 1.1 revision (checked in on 12 Sep
2005), so there's no change log to explain the problem.


        Stefan