On Sat, Apr 05, 2014 at 09:21:33PM +0200, Vincent Danjean wrote: > The disagreement comes from the fact that the maintainer does not > think that he must declare this incompatibility. > For now, if you install a r package from testing, it will pull > the r-base-core from testing (due to dependency such as > "Depends: r-base-core (>= 3.0.2-1)") > But, when r-base-core from testing is installed, the system keeps > other r related packages from stable (no conflict, break, ...) > and these packages won't work anymore. > > The maintainer think that he does not need to do anything about > that.
Then the maintainer is very wrong, if only because an upgrade from stable to testing *involves* a partial upgrade. Let's say during a dist-upgrade from wheezy to jessie, r-cran-foo is upgraded after r-base-core. This is possible, because there is no dependency relationship preventing that currently. Assume another package X which depends on r-cran-foo, and which does something in its postinst which needs r-cran-foo to work. This is allowed: at the point when the postinst of X is ran, r-cran-foo is configured and thus assumed to work. However, because of the broken r-cran-foo -> r-base-core dependency, the said postinst of X will fail. This will cause the upgrade as a whole to fail, in a pretty hard to debug way, especially for sysadmins who don't understand R. -- It is easy to love a country that is famous for chocolate and beer -- Barack Obama, speaking in Brussels, Belgium, 2014-03-26 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
