Re: ediff and longlines-mode

[Roland Winkler]

On Sat Jun 11 2005 Richard Stallman wrote:
>     Visit the files foo and bar. Enable longlines-mode for buffer foo.
>     Then apply ediff-buffers.
> 
> What aspect of the resulting behavior do you consider incorrect?

I noticed that my bug report was slightly screwed up in the
emacs-pretest-bug archive. The file `foo' from my example should
contain only one long line (no line break with a second line of
text).

If the buffer foo is displayed without longlines-mode being enabled,
ediff says correctly "Only white-space differences in region 1".

If longlines-mode is enabled for buffer foo (so that foo is
displayed in two lines), the second line of this buffer is not
highlighted at all by ediff, as if this line was the same in the
buffers foo and bar. On the other hand, in the buffer bar, both
lines are highlighted, so there is no unhighlighted part in bar
matching the unhighlighted part in foo.

There are two possiblities for ediff I could imagine when comparing
buffers foo and bar while longlines-mode is enabled for foo:

- If hard and soft newlines are treated the same by ediff, then
  ediff should not find any differences between these buffers.

  This could be confusing because the result of ediff would depend
  on how the buffers are displayed while the corresponding files can
  be different.

- If ediff distinguishes between hard and soft linebreaks, then it
  should highlight both buffers completely, saying "Only white-space
  differences in region 1" (as it is the case when longlines-mode is
  not enabled for buffer foo).

Roland