apparent visibility bug: cannot achieve canonical form

[Samuel Wales <samologist@gmail.com>]

When I do isearch-forward-regexp, things are revealed in a
non-confusing way.  But when I stop the isearch with an
arrow key, I get only the following heading, and not the
headings below that at the same level.  That is confusing
because it looks like there is nothing there.

Instead I want canonical form, which is a visibility state
that can be achieved by using arrow keys and TAB.  This is
in contrast to sparse trees.

So isearch is working until I am done with it, at which
point it collapses inexplicably to non-canonical form.

What is very strange is that (org-reveal t) does /not/
repair the tree to canonical form.  I think it should.

I have org-show-siblings set to t, and by default it is on
for the isearch context.  I think this should repair the
tree to canonical form also (via org-show-context which is
called in outline-isearch-open-invisible-function).

My usage: I only want sparse trees when I explicitly run
sparse tree, otherwise various forms of canonical form.  The
only exception is sometimes to hide the parent headline's
body (with or without drawers) in show-branches or
show-children form.

I have already fixed going to org from the agenda, figured out how to
fix it from Magit, etc., but isearch is still broken.

I am not sure if the variables used by org-show-context can
get the visibility that I am looking for.  I have tried many
settings for those variables (setting hierarchy above to t,
showing next, showing siblings) but there seem always to be
invisible sibling headlines and possibly similar apparent bugs.

Ideally, we would have a function to restore to canonical in
case something makes the tree non-canonical.  Like
(org-reveal t) is AFAICT documented to do.

Are these bugs?

Thanks.

Samuel


-- 
The Kafka Pandemic:
  http://thekafkapandemic.blogspot.com/2010/12/welcome-to-kafka-pandemic-two-forces_9182.html
I support the Whittemore-Peterson Institute (WPI)
===
I want to see the original (pre-hold) Lo et al. 2010 NIH/FDA/Harvard MRV paper.