On Tue, 2008-09-09 at 09:30 -0500, Mike McGrath wrote: > On Tue, 9 Sep 2008, Bastien Nocera wrote: > > > I'm pretty sure it was discussed before, and I forcefully ignored it > > thinking people would realise it was a bad idea to drop it. > > > > It's just that it's making us look like idiots in the changelogs, having > > to bump release for something like that: > > http://cvs.fedoraproject.org/viewvc/rpms/gvfs/devel/gvfs.spec?r1=1.69&r2=1.70 > > > > In fairness... you should be testing things before you commit them, then > tag them. If we keep tagging things that don't work then I'm afraid we > are idiots. I know the code compiles, I was the one who made the release and used "make distcheck". The upstream configure creates a .tar.gz, not a .tar.bz2 (which the maintainer usually picks up from the GNOME FTP). I forgot to change the suffix because all my projects dist to .tar.bz2 files. So I'm supposed to run every single build in mock before committing them? That's not going to happen. The CVS keeps the sources and all the commits done. What's wrong with using it for safe keeping? Fine if you make "force-tag" fail when a build has already been successful, but removing the command altogether is the wrong thing to do. I'd rather "make force-tag" told me to cancel the previous build (if I know it'd gonna fail), just lets me do it if nothing got built on the previous run, or forbids me from using the command if a successful build was done. -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
