[MERGE] Bzr diff produces -p1 diffs

[Michael Ellerman]

On 3/8/06, Matthew D. Fuller <fullermd at over-yonder.net> wrote:
> On Tue, Mar 07, 2006 at 08:53:42AM +0100 I heard the voice of
> Jan Hudec, and lo! it spake thus:
> >
> > The situation is not symetric though -- while you can't generate -p0
> > diff with diff, VCS should have no problem to generate -p1 diff.
>
> See, to here we're together...
>
> > Therefore -p1 is strictly better candidate for a standard.
>
> But this steps just doesn't seem to fit at all.  It would if -p1 were
> *good*.  I find it *bad*.  Neither of our assertions is going to do
> much to move the other, I suspect.  VCS's X, Y, and Z use one, A, B,
> and C use the other, blah blah blah.

You find -p1 *bad*, I find it *good*, but Jan actually made a very
good point, independant of which VCS uses -p1 vs -p0.

That is, that _only_ a VCS can produce a recursive -p0 diff, there is
_no way_ to do it by hand[1]. On the other hand it is trivial and
intuitive to create a -p1 diff using _just_ diff.

cheers

[1] Don't even think about using sed ;)