Re: Non-unified patches and dpkg source format ‘3.0 (quilt)’.

["Giacomo A. Catenazzi" <cate@debian.org>]

Charles Plessy wrote:
> Le Wed, Aug 05, 2009 at 03:45:00PM +1000, Ben Finney a écrit :
> > The point, rather, seems to be that unified-diff format is the de facto
> > standard format for exchanging patch information.
>
> Le Wed, Aug 05, 2009 at 10:53:21AM +0200, Michael Banck a écrit :
> > It's the preferred format for 99% of all Free Software work/projects
> > AFAICT.
>
> In my workplace's cafeteria, 99 % of the people eat curry rice with a spoon,
> and 1 % with chopsticks. But this is causing no trouble, and never the spoon
> users ask the chopsticks users to change their instrument (and I can tell you
> that I do not leave a single grain of rice when I eat my curry rice with
> chopsticks).
>
> I am all for campaigning for the unified diff format if there are arguments on
> which I can base a discussion with Upstream, but a mere cultural preference,
> be it the one of a very large majority, is a too weak argument.

I think there is few advantages on non-unified diff:
I don't think many upstreams will take advantage of it, and I think
few upstream will take patched from our source package.

IMHO the right way to sent patched upstream is send comments and patches,
in they preferred form (format of patch, format of comment, target
(like mailing list) and last upstream version).
Note also: rediff works only on unified patches.

The disadvantages: we need to build a lot of tools to test quality
of our packages. I think handling different diff format will
decrement the quality of such tests.

I really think that in our future we will have a lot of automatic testing
tools to detect problem before they happens.

ciao
	cate