Re: diff-mode misinterprets empty lines.

[David Kastrup <dak@gnu.org>]

Richard Stallman <rms@gnu.org> writes:

>     When people don't want trailing whitespace warnings to trigger in their
>     editor, they should obviously amend diff-mode for that.  But changing
>     the diff format instead is insane.  And that patch happens to have
>     tolerance code intended to deal with utterly broken mailers is no excuse
>     for that.
>
> Are you suggesting I should ask the diff maintainers to revert that
> change?  That might be a good idea.

Personally, I would consider this a good idea.  The "patch" program has
apparently been made robust in the presence of mail-mangled patches,
cut&paste carnage and other damage.  But "patch works with it" is, in my
opinion, not a positive definition of the diff format (while "patch
fails" is a different issue): patch is not the only program working with
diffs, and some version control systems have a diff-based workflow
without using "patch" for it.

Here is the format definition actually delivered with diff itself for
context diffs:

<URL:http://www.gnu.org/software/diffutils/manual/html_node/Detailed-Context.html#Detailed%20Context>

And here another for unified diffs:

<URL:http://www.gnu.org/software/diffutils/manual/html_node/Detailed-Unified.html#Detailed%20Unified>

It appears quite definite from the description that no whitespace
reduction is intended, neither for content nor for formatting
characters.

So yes, I would ask the diff maintainers to revert the change.  I can't
think of a good reason for it at all, but maybe they can, so it would
probably be a good idea to keep emacs-devel in the list (after all,
Emacs is one of the applications broken by this, for now): that way we
might together reach consensus on the best course.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum