That is correct behaviour as far as Emacs is concerned (whether it is sensible is another debate…)

Before transient-mark-mode came about Emacs would work from point “onwards” to the end of the buffer or the end of the mark. This effect is still present in Emacs hence its slightly annoying “two-tier” region system.

The solution is to either C-x h first to mark the region or alter the function to work on the entire region.

Mickey.

To save a C-p.

That is not quite the same as C-x C-o will only work on the area around point; and if there happens to only be one empty line then that line is removed, which is not what we want.

Not bad for a first attempt! I think the first one can be simplified to this:

(defun flush-blank-lines (start end)
  (interactive "r")
  (flush-lines "^\\s-*$" start end nil))

This way we use Emacs’s built-in support for working with regions by using the interactive “r” code letter which will pass in the start/end of point and mark automatically — this is important for people who don’t use transient-mark-mode. This approach also lets use pass in explicit ranges from lisp.

The second function works in much the same way as the first, although I have opted to use the interactionversion instead of the “proper” elisp regexp function:

(defun collapse-blank-lines (start end)
  (interactive "r")
  (replace-regexp "^\n\\{2,\\}" "\n" nil start end))

(defun flush-blank-lines ()
“Removes all blank lines from buffer or region”
(interactive)
(save-excursion
(let (min max)
(if (equal (region-active-p) nil)
(mark-whole-buffer))
(setq min (region-beginning) max (region-end))
(flush-lines “^ *$” min max t)
)
)
)

(defun collapse-blank-lines ()
“Collapse multiple blank lines from buffer or region into a single blank line”
(interactive)
(save-excursion
(let (min max)
(if (equal (region-active-p) nil)
(mark-whole-buffer))
(setq min (region-beginning) max (region-end))
(query-replace-regexp “^
\\{2,\\}” ”
” nil min max)
)
)
)

Any suggestions on improving these?