Do we want extras/testing/{4, 5} repos (was Re: Packaging review guidelines clarification)

[Jeff Spaleta]

On 2/18/06, Dan Williams <dcbw at redhat.com> wrote:
> Whee, more layers of complexity.

A mandatory pre-publish testing pretty much demands the complications,
in a way that a scratch area does not. But even a scratch area will
need to be integrated into the buildsystem for maximum benefit. Anyone
wanting to test that a set of packages that make up a dependancy chain
work.. will need to turn on the scratch area to fill deps as they walk
through the chain.
Example:
if a new update to application bar needs an update to package libfoo
and the current release of application bar wont work with the new
libfoo. As a maintainer you'd like to use a scratch build on the
buildsystem to make sure the new set of foo and bar both build as
expected so you can push both packages to the release tree so noone
experiences am unrecoverable dep problem if they are not pushed
together.  Testing the new libfoo by itself isnt useful because if you
push it but the new application bar doesn't build.. you have broken
the deps for the old version of bar. And you can't scratch test the
new version of bar with the released version of libfoo.  You'd have to
scratch test the set or you don't get value from the scratch.

-jef