bug#25777: 25.1; [PATCH] `rectangle--pos-cols' should not move point

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

> > I checked the code of `rectangle--pos-cols' and saw that
> > it now does `goto-char' (and it did not do so before
> > Emacs 25.1).
> 
> I'm confused as to what you mean by this, since `rectangle--pos-cols'
> did not exist before 25.1.

Sorry, my bad about that expression.  I meant to say that
_there is no such problem in Emacs 24.5_.  In Emacs 24.5,
following the recipe DTRT: the C-p moves point up one line.
Emacs 24.5, like 25.1, has rectangle support.

Nothing changes in that regard for my code: I take advantage
of exactly the same functionality offered by rect.el as before.
What has changed is how rect.el obtains and returns the columns
delimiting the current region rectangle.  I changed my code
accordingly.

Perhaps I'm not being sufficiently clear about this regression.

I think it's great that a function such as `rectangle--pos-cols'
was added.  (But it has no business being considered "internal".)

Just as, for Emacs 24.5, I reused some code from rect.el to
do what I need to get the rectangle columns, so I reused
the corresponding code from Emacs 25.

Look at function `apply-on-rectangle'.  In Emacs 25 it calls
`rectangle--pos-cols' to get the rectangle columns - and so
do I.  In Emacs 24.5 it also does just what I do (I stole
the code) to obtain those columns.

Do you really think that a function whose only purpose is
to let you know what the rectangle columns are should move
point and leave it in a position that is not one of the
rectangle corners?

If so, then you will say that all callers of this function
should wrap with `save-excursion' if they don't want point
moved.  I think that's the job of a function whose only
purpose is to return those columns.