On Wed, 30 Jul 2008, Cyril Brulebois wrote:
> Don Armstrong <email@example.com> (30/07/2008):
> > There's no reason not to increment the version. You've made a
> > release, and have made changes to that release. Whether it
> > actually hits the archive is immaterial. You just build with the
> > appropriate -v option, and you're done.
> And when one checks in which revision this or that happened, that's
> quite misleading not to see any upload in the archive matching those
> revisions (yeah for QA work!).
In QA work you're always comparing revisions. There's no difference
between dealing with versions which skipped the archive because they
were QAed elsewhere, or versions which are no longer in the archive
because they have been superseded. Furthemore, in this case, there was
an upload, it just never made it out of NEW.
> > Moreover, a new revision makes it easier for everyone to follow
> > what had changed between a version which was reviewed by the
> > ftpmasters and the next version that was uploaded.
> That was already discussed at length in another mentors.d.n thread
> (other threads actually).
There's a significant difference between uploads to mentors.d.n and
uploads to debian.org. If you actually uploaded to the archive, even
if it never made it out of NEW, you should never elide changelogs.
Three little words. (In descending order of importance.)
-- hugh macleod http://www.gapingvoid.com/graphics/batch35.php