Hey Robert, Robert Sander <r.sander@xxxxxxxxxxxxxxxxxxx> writes: >> The advantage of that would be having tested build environments and an >> easy way to check if ceph version X still easily compiles on distro A >> version B. > > AFAIK it is not an issue with the build environment but that the Ceph > code depends on the latest C compiler and libraries and cephadm on a > recent Python. > > So the code cannot be built on older distributions. No matter how > sophisticated the build system would be. Got it. My understanding was more that maintaining the build system itself was not worth the effort, not being a hard "cannot be built" constraint. > I am asking the question: Why even build packages any more? A valid question and I think the answer is that not everyone can already deploy in containers. > The upstream project has decided that cephadm and deployment with > container images is the way to go. Wouldn't it be sufficient and > easier to just build the images? In the end I think it is a very easy question: if there is enough (>= 1) demand for packages for $OS $OSVERSION, then someone will need to build them. If that someone is then also publishing the resulting packages, any other person on the same os/version can reuse that effort. If there no one needing any os packages anymore, this is obviously not required. Best, Nico -- Sustainable and modern Infrastructures by ungleich.ch _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |