Jesse Keating wrote:
We do separate testing per release, because each release is
different.
Different library sets, different kernels, glibc, some different desktop
environments, etc... Assuming that testing on one release means that
it'll work on other releases is grossly irresponsible.
In practice it's almost always true, with really extremely rare exceptions,
which are not worth headaches such as broken upgrade paths.
Kevin Kofler