Re: Patch format (was: [med-svn] r1888 - in trunk/packages/glam2/trunk/debian: . patches)
Le Sun, May 18, 2008 at 12:27:10PM +0200, David Paleino a écrit :
>
> Can we standardize the patch header? Be it quilt, dpatch, $foo, the header
> might be something like:
>
> Author:
> Forwarded:
> Description:
>
> I'm currently starting to use this format:
>
> Author: Foo Bar <foo@bar.com>
> Forwarded: no | http://$url_of_upstream_BTS_with_patch
> Reason: foo
> another line
> .
> Another paragraph
Hi David,
I think that the Reason: field is not necessary, as one could use:
Forwarded: no, because blablablah…
I just read one of the messages of Pierre Habouzit, in which he says
that for the glibc packages the forward status is also encoded in the
patch name. It can be a good idea as well, although it would generate
bigger diffs on our commit list.
I also like the idea that a bug is fixed by a patch, but would be closed
only by a new upstream release that makes the patch unnecessary. In our
case, we could for instance retitle the bug and downgrade its severity
to wishlist when we apply a patch. But I wonder if it would blur the
information in the changelog.
Let's experiment with real work and turn it into a Debian-Med policy
before discussing it on -devel. We are not far from the freeze (do
not forget that 9 days before the freeze it will be already too late
for urgency=low tasks!), and we have many uploads to do; each of them is
a good opportunity to build a Policy by practice.
Ah, I just realised that I forgot to answer to your question.
Standardising the fields and their order is definitely a good idea :)
Have a nice day,
--
Charles Plessy
http://charles.plessy.org
Wakō, Saitama, Japan
Reply to: