Re: Auto Backporting (Was: Backports of scientific packages)
On Sat, Sep 26, 2009 at 05:40:58PM +1000, Russell Coker wrote:
> On Sat, 26 Sep 2009, Rene Engelhard <firstname.lastname@example.org> wrote:
> > > Shouldn't checking if Build-Depends are satisfiable in stable be enough?
> > > And if it doesn't build that way, I'd say there's a bug in the package
> > > anyways, because it should bump some build dependencies.
> > build-deps are not necessarily runtime deps. Especially if
> > stuff changed in the fs or for policy reasons and that is not reflected in
> > the build-deps because the change is not in the build-deps.
> The runtime dependencies should either be based on the versions of the
> packages that were compiled against (in which case they should be correct
> automaticall) or they will be based on some specific features (EG a new
> version of a package containing /bin/foo adds a new command-line option
> to /bin/foo) in which case an explicit versioned dependency should be
> sufficient (failure to do so would be a bug in any case). It should not be
> THAT difficult to make a back-port build daemon refrain from building
> packages if the runtime dependencies can never be satisfied.
But runtime dependencies can be the result of shlibs bumps, where the
runtime dependency will be different depending on the build-dep version
used at build time. So the selection should be done *after* having tried
at least once to build a package that has all its build-deps