GNU bug report logs - #7430
Bug in ediff's treatment of whitespace (this time with ALL the attachments)

Previous Next

To reply to this bug, email your comments to 7430 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: "Tom Breton (Tehom)" <tehom <at> panix.com>
To: bug-gnu-emacs <at> gnu.org
Subject: Bug in ediff's treatment of whitespace (this time with ALL the   
	attachments)
Date: Wed, 17 Nov 2010 15:40:23 -0500

[Message part 1 (text/plain, inline)]

**** Explanation of the bug:

Wordwise merging in ediff doesn't handle whitespace reasonably when
merging to a blank element.  It just plops the non-blank element at the
end of whatever whitespace it ends up with, even across line breaks.

**** How it should work (IMO)

IMO it would be more correct to infer the division of whitespace from the
whitespace around the non-blank element.

**** Instructions for reproducing it

***** General orientation
Use the attached files:

    * odd-whitespace-2-file1.{abc}.txt

    * odd-whitespace-file1.{abc}.txt

I found it useful to give each set of {a,b,c} its own directory and keep
the filenames the same across directories.  I can't attach them to this
email with directory names, though.

I found this code useful in seeing this bug, so I'm including it here.  It
just starts an merge-with-ancestor with the respective files - saves time.
 It expects filenames of the form file1.{a,b,c}.txt in different
directories.

  (defun bug-ediff-merge-files-with-ancestor (dir)
     ""

     (interactive "DDirectory: ")
     (ediff-merge-files-with-ancestor
        (expand-file-name "file1.a.txt" dir)
        (expand-file-name "file1.b.txt" dir)
        (expand-file-name "file1.c.txt" dir)))

***** Explicit instructions

 * Merge file1.a.txt with file1.b.txt using ancestor file1.c.txt

 * "n" to go to first clash
 * "b" to partly merge - just so it's merging nicely and not seeing
   "<<<<<<" ">>>>>>" "#####Ancestor" etc
 * "=" to start an inferior merge
 * "a" to compare to buffer A
 * (Now in the inferior ediff)
 * "n" to go to a line that still needs to be merged to the ancestor.
   In the demo they're all of the form "Line N A".
 * "a" to try to use the version from A.
 * It doesn't merge right.  It moves "A" to another place.

Both odd-whitespace-* and odd-whitespace-2-* exhibit similar unexpected
behavior.

	Tom Breton (Tehom)

[odd-whitespace-file1.a.txt (text/plain, attachment)]

[odd-whitespace-file1.b.txt (text/plain, attachment)]

[odd-whitespace-file1.c.txt (text/plain, attachment)]

[odd-whitespace-2-file1.a.txt (text/plain, attachment)]

[odd-whitespace-2-file1.b.txt (text/plain, attachment)]

[odd-whitespace-2-file1.c.txt (text/plain, attachment)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.