On Wednesday 16 May 2007 09:11, Mgr. Peter Tuharsky wrote: > I'd say, half of problems with testing were connected to bugs in > installer. This statement really wants some qualifications... The official releases (beta and RC) of the installer for testing have had no really serious bugs, though there may be errata that can affect specific situations or hardware. They are tested extensively. Also, in most cases it is not the _installer_ that is broken, but that there are bugs in individual packages that are installed during an installation that can cause the installation to fail. Unfortunately such bugs are often only detected after a package migrates to testing because that is the first time someone will try to do an installation that includes the package. I also think that we will see a lot less of such problems for Lenny than we have for Etch. For Etch we've had a few really major changes (kernel, initrd generators, removal of base-config, XOrg transition) that had a high impact on the installer an installations. I doubt we'll have so many for Lenny. Installation problems when using daily built images or weekly snapshots is therefore quite possible, but we always try to get such issues fixed ASAP. However, you are also almost guaranteed to be able to install testing using a full CD/DVD image from the last official D-I release, especially if you choose to use only the CD and not use a network mirror in addition to the CD/DVD. And, as Andreas has already said, you can always install stable and upgrade to testing (though that may get harder as stable gets older, especially as there will be no release notes yet). Even though all this may not really change things from the viewpoint of an end user, it is IMO very relevant when discussing the usability of testing as a whole. Cheers, FJP
Attachment:
pgpuPlo7cjLHw.pgp
Description: PGP signature