Hi, On Mon, 2008-04-28 at 00:53:15 +0200, Raphael Hertzog wrote: > On Sun, 27 Apr 2008, Matthias Klose wrote: > > gpc-4.1 (2.1-4.1.2-22~dummy) unstable; urgency=low > > > > * Upload as gpc-4.1. > > > > -- Matthias Klose <[EMAIL PROTECTED]> Sun, 27 Apr 2008 13:44:14 +0200 > > > > gcc-4.1 (4.1.2-22) unstable; urgency=low > > > > [...] > > > > If you don't have this dummy entry, then dpkg-genchanges complains: > > > > $ dpkg-genchanges > ../gpc-4.1_2.1-4.1.2-22_i386.changes > > dpkg-genchanges: warning: the current version (2.1-4.1.2-22) is smaller > > than the previous one (4.1.2-22) > > dpkg-genchanges: including full source code in upload > > > > and the upload to ftp-master is rejected. including the full source > > in this case sounds like a bug. > > It's not a bug, it's a change in the heuristic used to determine when > the source code has to be uploaded. Please read dpkg-genchanges man page > (and the -si option in particular).
I guess the problem is that people were used to the old heuristic and this changes their expectations, or might imply changes in their workflow. (Which might not be inherently bad.) > If you don't want to upload the orig.tar.gz then the previous entry > has to contain the same upstream version. Or you override this with > -sa. In this case it's not only the version but also the source package name. I guess any changelog parser might get a bit confused with that as well. This also is a problem for stuff like apt-listchanges, it lists all gcc changes (up to gcc-snapshot) on each frontend sync upload. > This bug should be just closed unless you have a better suggestion > that handles correctly all cases. This heuristic is better than the > previous one IMO. Matthias would you find it unacceptable to sort the changelog entries in version incremental order? regards, guillem -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
