On Sat, Aug 10, 2024 at 06:19:31PM +0200, Andrea Bolognani wrote: > I spent some more time testing the restructuring today, and it seems > to behave well. There is still one known issue with transferring > conffiles between packages under certain conditions, but I believe > that I have a solid plan to deal with that and I just need to > implement it. There are probably other smaller issues with > dependencies and such, but I think that the overall "shape" of the > end result is near-final. > > The git history is still a complete mess though, and I really need to > polish that up before I feel comfortable asking the poor reviewers to > wade through it O:-) > > So, here's my proposal: in two weeks' time, i.e. by August 25, I > will upload a version of libvirt that includes both the restructuring > and the usr-merge bits to experimental. > > This upload will not be prepared from the official git repository, > but from my own fork. This removes the need to have all changes > squeaky-clean and reviewed before we can proceed, and allows us to > move forward with whatever I have ready at the time. > > The idea is that this will give dumat a chance to validate the whole > ordeal and ensure that users will not run into file loss scenarios. > If any issue is detected, we'll have the opportunity to rectify it; > if not, we can feel safer about taking a bit longer to polish and > land the restructuring. > > I will of course endeavor to get the branch in a reviewable shape, or > as close to that as possible, before then. > > Does this plan sound reasonable?
Since no disagreement was raised, I'm working under the assumption that everybody is okay with this plan. I've made some more progress over the past week, including fixing the known bug mentioned earlier and integrating the usr-merge patches. Things generally seem to work fine. I have however realized that there is an obstacle to enacting the plan as outlined above: the restructuring involves introducing several new binary packages, which in turns requires a trip through the NEW queue, and I'm not (yet) a Debian Developer so I can't prepare the upload myself. Guido, would you be willing to take care of that part? I still need to clean up the git history, but given the current status I'm feeling relatively optimistic and I'm convinced I should be able to cobble together something good enough for a proper (draft) merge request by either Friday or Saturday evening. That'd give Guido approximately a day to sanity-check the changes and prepare the upload if we want to stick to the original schedule. Hopefully that's workable. Also note that the implementation of usr-merge I've imported is the same proposed by Michael, which can't be sensibly backported to older releases. As mentioned in [59] this might be a problem for Ubuntu, so I'm CC'ing Christian again in order to give him another chance to speak up. [59] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1064126#59 -- Andrea Bolognani <e...@kiyuko.org> Resistance is futile, you will be garbage collected.
signature.asc
Description: PGP signature
