On Fri, Feb 21, 2025 at 06:27:42PM +0100, Fabio Valentini wrote: > On Fri, Feb 21, 2025 at 5:13 PM Miro Hrončok <mhron...@redhat.com> wrote: > > > > On 21. 02. 25 12:41, Fabio Valentini wrote: > > > On Thu, Feb 20, 2025 at 1:17 PM Miro Hrončok <mhron...@redhat.com> wrote: > > >> > > >> Hello. > > >> > > >> With the recent discussions about provenpackagers in Fedora, I recently > > >> got an > > >> idea. > > >> > > >> One of the common needs for provenpackagers is to simply "bump and > > >> rebuild" a > > >> set of dependencies. > > >> > > >> All packagers are already able to build anything (except a very specific > > >> and > > >> small set of specially-signed packages). However, to bump the package, > > >> they > > >> need commit rights. For that reason, provenpackager rights are often > > >> required. > > >> > > >> With the wide adoption of %autorelease, such bump commits are empty, > > >> which > > >> should be easy to verify. > > >> > > >> What if we allowed all packagers to push empty commit to any package? > > >> That > > >> should eliminate *some* need for provenpackager access. We would also > > >> communicate in our policies that such bumps do not require prior > > >> agreement with > > >> the maintainers to avoid confusion about "what are we allowed to do". > > > > > > This sounds a bit more complicated than it needs to be ... > > > I would much rather explore not having to push empty commits *at all*, > > > and have koji auto-increment the build number if it would cause an NVR > > > conflict ... (yes, this should be possible with minor changes to the > > > %dist macro and some small additions to koji). > > > > IIRC this is the direction we explicitly decided not to go with > > %autorelease. > > We wanted the rebuild reason committed. > > Those are kind of two orthogonal issues. You could still use an empty > commit if you *want* a changelog entry :)
I still think we should always require a changelog entry, so that that the reason for the rebuild is always explicit. (In the normal case, there should never be a need for "trivial" rebuilds, and rebuilds should be for cases when there really is an imporant change. If your API or ABI changes all the time in a way that is backward incompatible and no-change rebuilds are required, then _that_ is the problem and something to reconsider. We shouldn't build a system to make such cases silent.) Zbyszek -- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org 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@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
