bug#21072: 24.5; inconsistent behaviour of `C-M-h (mark-defun)' in Emacs Lisp

[Drew Adams <drew.adams@oracle.com>]

I agree that the behavior is not particularly consistent, and it
does not completely correspond to the doc.  What the best fix is,
I don't know.

It's been this way for a long time, so there might be people
who expect or like it to do what it does, in which case the
doc should probably be fixed somewhat.

On the other hand, I'd bet that few, if any, would complain if
better behavior were implemented.

In any case, the behavior of being able to repeat to keep selecting
more defuns further down should be kept.

I'd suggest that Someone (TM) (Marcin?) implement a better
behavior and propose it to emacs-devel. ;-)

What might be better?

1. At least consistency wrt which defun gets selected, when
betweeen defuns.  The doc suggests a general rule (the next
defun), but that is not always respected.

2. Something consistent also wrt a comment before the defun
that will be selected.

3. It could be good for a numeric prefix arg to select that
many defuns.

4. It could be good for a negative prefix arg to select in
the opposite direction.  This is the main improvement I'd
like to see.  E.g. `M-- C-M-h' selects the previous defun;
`M-2 C-M-h' selects the two previous defuns.

Someone should play around and dream up something useful.

Wrt #2, I'm not sure what the best approach might be.