Re: table.el

[Stefan Monnier]

> 01 Dec 2001 18:21:32 +0900: Miles Bader <address@hidden> wrote:
> 
> > Tak Ota <address@hidden> writes:
> > > (defun kill-region-offset-by-5-wrapper ()
> > >   (kill-region (+ beg 5) (+ end 5)))
> > 
> > It's bad to require that a `wrapper' know the argument names of the
> > function that invokes it; better to simply pass the arguments to
> > `run-wrappers' somehow, and have them passed each wrapper as ordinary
> > arguments.
> 
> No, I am intentionally making the normally discouraged conduct.  As
> the name `kill-region-offset-by-5-wrapper' implies this wrapper
> function is the extension of the kill-region.  It is perfectly
> legitimate the extender knowing exactly what it is extending and
> knowing how the original is implemented.

If the wrapper is intimately linked like that, then why not
just use a simple`kill-region-function':

        (defun kill-region (babla)
          (if kill-region-function
              (funcall kill-region-function blabla)
            blibli))

> Since I am trying to create an extension mechanism I am fully taking
> advantage of http://www.gnu.org/software/emacs/emacs-paper.html#SEC18

I'm interested in a wrapper kind of extension mechanism (I need
something like that in PCL-CVS so that I can use the same keymap
in two buffers but with different behavior (in the *cvs* buffer
it operates on the marked files, while in the *cvs-info* buffer
it operates on the current file and current version of that file).

But I want a single wrapper for several functions, so
the wrapper can't be so intimately linked to the wrapped function.


        Stefan