org-metaup destructive when region active?

[Adam Porter <adam@alphapapa.net>]

Hi,

I noticed that using org-metaup when a region is active seems to behave
in an unintuitive and potentially dangerous way.  I would expect that it
would behave similarly to org-refile when a region is active, but it
doesn't.

Looking at the code, I see that it does this:

┌────
│ (let* ((a (min (region-beginning) (region-end)))
│        (b (1- (max (region-beginning) (region-end))))
│        (c (save-excursion (goto-char a)
│                           (move-beginning-of-line 0)))
│        (d (save-excursion (goto-char a)
│                           (move-end-of-line 0) (point))))
│   (transpose-regions a b c d)
│   (goto-char c))
└────

It was difficult for me to understand what that would do, and the
docstring of transpose-regions didn't help.  So I did a small
experiment, defining that code as a command and calling it in this
buffer (where the region is marked with brackets):

┌────
│ abcd
│ [efgh
│ hijk]
│ lmno
└────

The result was this:

┌────
│ efgh
│ hij
│ abcdk
│ lmno
└────

Now for an example with Org headings, before:

┌────
│ * A
│ ** A1
│ ** A2
│ [* B
│ ** B1
│ ** B2]
│ * C
│ ** C1
│ ** C2
└────

After:

┌────
│ * A
│ ** A1
│ * B
│ ** B1
│ ** B
│ ** A22
│ * C
│ ** C1
│ ** C2
└────

As you can see, it's corrupted the outline: Heading B2 is now another B
but on the second level, and heading A2 is now a child of the
first-level B but named A22.  And these examples don't even have entry
text…

Ideally it would act on all headings in the region, but given the
complexity of doing that correctly, would it be a good first step to
simply do nothing when org-metaup is called and a region is active?  It
would at least avoid corrupting data.  I think that's a non-trivial
risk, because when I noticed this behavior, it was in a partially
collapsed outline, and some of the headings in the region (and their
entry content) seemed to vanish altogether; I only recovered them with
undo.

Thanks.