fsateler at gmail.com
Wed May 26 09:45:11 UTC 2010
On Wed, May 26, 2010 at 05:22, Dmitrijs Ledkovs
<dmitrij.ledkov at ubuntu.com> wrote:
> On 26 May 2010 02:24, Felipe Sateler <fsateler at gmail.com> wrote:
>> On Tue, May 25, 2010 at 18:02, Jaromír Mikeš <mira.mikes at seznam.cz> wrote:
>>>> Od: Felipe Sateler <fsateler at gmail.com>
>>> RT> > The last consensus was that yes, under the condition that the clean
>>> RT> > target of debian rules restores the working copy to a state that is
>>> RT> > ready for git diff/commit operations. Most commongly this means to
>>> RT> > unpatch in the clean target.
>>> RT> > The intend behind this is to have the same workflow for both Format 1.0
>>> RT> > and 3.0 packages.
>>> FS> Be careful to only unapply patches when in a git repository! Otherwise
>>> FS> you will get misterious FTBFS in the buildds.
>>> FS> Also, you need to apply them in the build rule (again, only if in the
>>> FS> git repository).
>>> So if I understand well something like this should be in short dh 7 rules file:
>>> or something more should be done?
>> More. The unpatch should be done only if within a git repository.
>> Something like:
>> if [ -d .git ] ; then \
>> quilt pop -a \
> dh_quilt_unpatch ?
> or even easier
> dh --with quilt $@
I think (but I'm not sure, I'd love to be proven incorrect) that this
would result in unwanted behaviour when doing a NMU with a 3.0 source
package, because patches would be applied _after_ the NMUers
modification to the package, not before.
More information about the pkg-multimedia-maintainers