On Fri, 25 Mar 2011 13:42:09 -0400, Kevin wrote: > On 03/25/2011 05:39 AM, Michael Schwendt wrote: > > You haven't shown any such '=' dependencies. The thread only quotes > > unresolvable SONAME dependencies. And those are "equal to" by definition. > > Now, your mixing up semantic meaning. Being dependent upon a particular > .so library release is the same as an = depends. When that library > version gets updated to a newer version, it *will* break the dependency, > regardless of whether the program will work or not work with the updated > library. That is one of the reasons named dependencies exist. You show a clear misunderstanding of automatic SONAME dependencies. > If the > dependency were to be written as Requires directfb >= 1.4.3 then this > problem wouldn't exist. False. Due to the changed SONAME, the program would not even start. The dynamic linker would fail to find the "old" library. That's why the dependency is on a specific SONAME and not ">=". > You're fixated with RPMFusion. B^) (Given the subject of this thread, > I can't blame you for that.) False, too. RPM Fusion just serves as an example of a 3rd party repository when discussing [broken] dependencies. > > That would require the third party repos to prepare their test updates > > based on Fedora's updates-testing repo contents. With no guarantee that > > all of the packages in updates-testing, which would be built against, are > > good and will be released actually. > > I thought that's why it was a -testing repo. Sure, but if you've built against _all_ of -testing, how do you know you can release part of -testing and/or withdraw part of testing without pulling anything that has been built against? You can't. It needs a well-defined "stable" buildroot of packages to build against. -testing can't be included. > > Alternatively, they would need to find a way how to build against Fedora's > > koji buildroot package set. Dunno whether there's a reliable way to mirror > > that (and quickly enough in order to be up-to-date with regard to builds > > being added to [and removed from] it either automatically or by Fedora > > releng). > > I don't think that's the way to go. Too complicated, and koji stuff is > less stable than -testing. Perhaps its more equivalent to the atrpms > -bleeding stuff? Huh? Again you show that you don't know what you're talking of. "koji" is just the name of the software that is Fedora's build system. It cannot be "less stable than -testing" -- that doesn't make any sense at all to say that. It seems you refer to downloading _unpublished_ builds found within koji. That's a completely unrelated topic. > > Not "3 for, 3 against". That's just the karma threshold configuration > > for this update ticket. The update needs either +3 points to publish > > it before 7 days have passed or -3 points to remove it from updates-testing. > > It has received +1 from the packager (and +1 from an anonymous voter). > > That wasn't immediately clear to me from glancing at the URL you > provided. Yes, I understand its intended for developers who probably > understand the system better than I do, but if you're going to start > passing it around to non-developers, perhaps it should be clearer.... Not true. The Fedora Updates System does not target developers. It targets the entire Fedora community. > And +2 after 7 days still sounds like insufficient testing to me. Don't generalise. -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
