RE: [External] : Re: Change default behavior of some commands that act on region?

[Drew Adams]

> > Breaking long-standing behavior.  I'll let others figure
> > out how to minimize this (e.g. by testing for
> > `transient-mark-mode` or something).

There's nothing to figure out.  The proposal said
clearly from the outset that it would have NO effect
when `transient-mark-mode' is turned OFF.

So yes, `transient-mark-mode' would be tested, and if
nil then the long-standing behavior would be UNCHANGED.

Similarly, the behavior of Lisp code that invokes the
command non-interactively would be UNCHANGED.

(if (or (not (called-interactively-p 'interactive))
        (not transient-mark-mode))
    (apply COMMAND ARGS) ; NO CHANGE FROM PAST BEHAVIOR

  ;; Do what user wants for such a command
  ;; when region is inactive.
  ...
  )

The only change in "long-standing behavior" would be for
interactive use when `transient-mark-mode' is turned ON.

And even that could be configurable.  E.g.:

* A list (defvar) of the commands that would be affected.
  Users could change this, but mostly would not bother.

* A defcustom to choose command behavior for inactive region:

  1. Do nothing (ignore - command is a no-op on inactive region).

  2. Prompt to confirm action.
     (e.g. "Region is inactive.  Invoke `foobar' on it anyway?")

  3. Raise an error.
     (e.g. "Cannot use `foobar' interactively with inactive region")

  A good default behavior would be to do nothing (#1),
  in keeping with the aim to use `transient-mark-mode'
  to govern things.  These are essentially commands that
  never learned about `transient-mark-mode' and active
  regions.  In general, there's little sense in turning
  on `transient-mark-mode' and wanting commands to act
  on the INactive region.

* For named commands at least, we could add an ability to
  override the defcustom behavior for particular commands:

  (put 'foobar 'inactive-region-command-action 'error)