Re: SPAM: ediff possible bug

[Peter Lee]

>>>> Michael Kifer writes:

    MK> There are many ways in which the heuristic for selecting files
    MK> in ediff could be improved. The question is whether the
    MK> implementation effort and the complications that would arise
    MK> are justified when the existing heuristic already supports
    MK> idioms that require the same or fewer number of keystrokes as
    MK> the suggestion. (e.g., ediff-buffers)

I use ediff a lot and, for me, using ediff-buffers and browsing for
the right buffers is not faster than marking the files I want and
hitting M-x ediff RET RET RET.

    MK> However, I am open to the idea of looking at patches that
    MK> implement better heuristics.

The following works for me with the added bonus that I don't have to
confirm the files I already marked.  Maybe this would be better
implemented as a seperate dired function (dired-do-ediff?), or maybe
I'm the only one doing this type thing and it should simply go in my
init file.

diff -w --context=1 "d:/src/ediff.el" "e:/cvs/emacs/lisp/ediff.el"
*** d:/src/ediff.el     Mon Jul 26 14:24:15 2004
--- e:/cvs/emacs/lisp/ediff.el  Sat Jun  5 21:44:14 2004
***************
*** 196,200 ****
    (interactive
-    (if (and (eq major-mode 'dired-mode)
-             (> (length (dired-get-marked-files)) 1))
-        (list (car (dired-get-marked-files)) (car (cdr 
(dired-get-marked-files))))
       (let ((dir-A (if ediff-use-last-dir
--- 196,197 ----
***************
*** 221,223 ****
                                       (ediff-get-default-file-name f 1)))
!              ))))
    (ediff-files-internal file-A
--- 218,220 ----
                                   (ediff-get-default-file-name f 1)))
!          )))
    (ediff-files-internal file-A

Diff finished.  Mon Jul 26 14:57:34 2004