On Wed, Oct 16, 2019 at 8:42 PM Stephen Gallagher <sgallagh@xxxxxxxxxx> wrote: > > On Wed, Oct 16, 2019 at 8:12 PM Kevin Kofler <kevin.kofler@xxxxxxxxx> wrote: > > > > Stephen Gallagher wrote: > > > 3) We need to get the policy I described above written onto -stone > > > tablets- the Packaging Guidelines and then we need to go and make any > > > stream that isn't compliant with it a non-default stream. > > > > But then we need a policy that requires a default version (non-modular or at > > least a default stream) to be available. Otherwise we end up with packages > > that are not installable out of the box because they have no default version > > at all. > > > > Not necessarily. It may be that we have to content ourselves with some > software always requiring a module enablement to use it. For example, > I maintain a module for Review Board, a Django-based code review tool. > For complicated reasons, it cannot run against Django newer than 1.6 > (the Review Board upstream maintains the 1.6 stream of Django for > security patches). If we decide that deps of default streams are bound > to the same rules as default streams (or, alternately, Matthew's > proposal of requiring only depending on default streams), then the > necessary outcome is that Review Board can only be available via > non-default stream (since it depends on django:1.6). > > That said, I'm sure that we can come up with ways to make that more > discoverable, but today we have limitations with how DNF is able to do > `search` or `repoquery`... it can only manage this with default or > enabled streams and it doesn't "see" the non-default streams. If we go > this path, I'd want to hear some input (and get some help!) on > improving the discovery experience. > A big chunk of the problem in fixing this will go away once this is merged: https://github.com/rpm-software-management/libdnf/pull/814 > > Matthew Miller wrote: > > > How would this act in the case where a default stream depends on a > > > non-default stream? > > > > From a policy standpoint, that non-default stream then ought to be bound by > > the same rules as default streams. But allowing a default stream to depend > > on a non-default stream paves the way for version conflicts to happen, so I > > am not convinced that it is a good idea to begin with. > > > > With the notable caveat of the example above, I'm willing to accept > this restriction in Fedora. We just need to acknowledge that it *will* > mean that some software we provide is less discoverable. If that's > deemed to be a worthwhile tradeoff, I have no problems with > implementing it. > We need a way for module dependencies to not care of default/non-default and handle this gracefully, even in transitions. > > > (And how would restricting default streams to only be able to depend on > > > default streams change things?) > > > > It would solve the version conflicts issue, so it makes a lot of sense, but > > at that point, why not require the default versions to just be non-modular > > instead? The main argument for using default streams was that they can > > depend on non-default streams of other modules. So if we disallow this > > (which I think makes sense), we may as well disallow default streams > > entirely and simplify things for everyone. (We would just need a short-term > > workaround for the default streams that exist now. The problem would be gone > > in the long run.) > > > > There's still a case that you haven't considered, which is things that > work at runtime as a default but cannot *build* against the default > set of packages. For example, we might have packages whose buildsystem > still relies on Python 2 (WAF?) but doesn't require it at runtime. > There might be packages that haven't yet migrated to a new, > backwards-incompatible change to Docbook for generating documentation. > Or packages that only build properly with a newer version of golang > than shipped in that release, or any of a thousand other examples that > are easily solved by build-time-only content and dependencies in > module streams. > > Also there are yet other packages (such as in the Go and Rust > ecosystems) that could simply be built once on Rawhide with the latest > compiler and shipped to each of the other releases without needing a > rebuild because they are statically linked. Modules allow this, basic > RPMs not so much. > This is actually not true. Every language stack we have does some degree of dynamic linking, even if it's only to glibc. The *only* reason we don't allow non-modular RPMs to be built once in rawhide and shipped to all releases without a rebuild is purely policy. We can mechanically do that since Koji lets you multi-tag RPMs into multiple collections. > So even if we eliminate the version conflicts issue by restricting > what comprises a default stream, there are other benefits to module > default streams. Ehh... -- 真実はいつも一つ！/ Always, there's only one truth! _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
