On Fri, 2002-05-03 at 20:03, Eduard Bloch wrote: > #include <hallo.h>
(BTW it's fantastic that you can post to a Debian mailing list without being subscribed) > Adam Warner wrote on Fri May 03, 2002 um 06:06:45PM: > > > How did this bug almost get into the offical release of Woody? Was the > > Pretty simple - nobody tests base-config completely unless it gets into > _Woody_. So the unstable check doesn't work for this package because the install disks only download woody anyway. But when something goes wrong we still have to wait for the package to move from sid-->woody even though no one typically tests it in sid. > > package tested before being distributed? If it wasn't for the security > > infrastructure holding back the release date of Woody, Debian could be > > in the midst of its worst PR fiasco with hundreds or thousands of CDs > > being recalled and the bad news completely overshadowing the release. > > Yes, this is one reason why IMO the general timings of the Woody release > are very, very bad. This must change in Woody+1. On my local mirror base-config 1.33.17 just became available to install. It will probably be at least another two days before the fixed package filters through to this mirror. I see your point about the general release timings being very bad if such fundamental packages are still being changed. "On the upside, woody itself is ready to be released." http://lists.debian.org/debian-devel-announce/2002/debian-devel-announce-200204/msg00020.html But what _would_ be good timing for the release Eduard? Perhaps everybody is simply feeling pressure because it has been so long since the last release. > > If there is no mechanism to roll back to earlier packages that are know > > to work when something really bad goes wrong, why not? [Ouch. A hastily written sentence] Translation: Why is there no mechanism to roll back packages? > The whole packagement system depends on the package versions. You cannot > just make the old package appear as an update without rebuilding it. And > exactly the build prozess was faulty in this case. [It's not immediately apparent why the package indices can't link back to the older file] I guess it really comes down to whether it is proper for no one to be testing packages until they hit woody, especially around the time of its release. We know that the biggest problem with testing Debian installers is that hardly anyone needs to use them. The idea of doing a new install instead of using the package management system to perform an upgrade is foreign to most users and developers. BTW thanks to the install disk developers. The disks are of very high quality. The best of any distribution I have used. Lots of options. Lots of freedom to do tasks out of order and no rampant paternalism (like refusing to proceed if a swap partition is not created). Regards, Adam -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
