Package: apt Version: 0.5.4 Severity: normal Though not explicitly documented, the following command do not work in a logical way (explained hereafter):
# apt-get --target-release testing upgrade IMHO, the expected outcome of this command is that a system would be upgraded to 'testing'. I.e. all packages that are to be upgraded would be upgraded to the version available at the 'testing' distribution, and packages that are already up-to-date with the version at 'testing' would not be upgraded at all. Currently, if, say, 'stable', 'testing' and 'unstable' are present in the sources.list, the given command would upgrade packages that are not up-to-date with 'testing' (if any) properly, but for packages that are already up-to-date with 'testing', the system would upgrade them to 'unstable' (if a newer version exists). E.g. (from my actual system): my system is completely up-to-date with 'woody' (with some packages from 'unstable'). I, then, installed stable's version of 'pacman', and run this command (with a "-s", of course :^>). The system wants to upgrade all packages to 'unstable', with the exception of pacman, which it wants to upgrade to 'testing'. As for "dist-upgrade", a similar reasoning can be applied. In other words, these commands should behave as if the 'unstable' and 'stable' entries of the sources.list do not even _exist_. The only "sticking point" is when -- if ever -- to downgrade a package if it is newer than the target-release. I think this would only be appropriate if the user explicitly requests this. Thanks, Zorzella -- System Information Debian Release: testing/unstable Kernel Version: Linux purple 2.2.18 #1 Sun Jan 14 08:53:46 PST 2001 i586 unknown Versions of the packages apt depends on: ii libc6 2.2.4-1 GNU C Library: Shared libraries and Timezone ii libstdc++2.10- 2.95.4-0.01090 The GNU stdc++ library
