Fedora 31 Finally Planning To Gate Packages While Testing, More Stable Rawhide
As something that arguably should have been done long ago, developers drafting plans for Fedora 31 are planning to introduce single-package gating so packages don't actually land in Rawhide (the Fedora development repository) until they successfully pass their tests... This should help weed out broken packages in Fedora Rawhide and lead to a more usable experience for those living on Fedora's bleeding-edge while also helping along a smooth release process.
The initial plans call for this to be an opt-in process and only be done for single packages in the initial stage while multi-package updates will be handled later. The plan is to gate packages on continuous integration test results before being able to land within Rawhide, in order to prevent broken dependencies, uninstallable packages, and other headaches that come as a result of package failures.
At this stage the process though is opt-in and test coverage isn't mandatory, but hopefully those elements will change as this support matures. The gating of Rawhide packages (or holding back the packages from landing until they pass a CI workflow) has been talked about before for improving the Fedora release quality and the ending of their alpha releases while for Fedora 31 due out in late 2019 this could finally happen.
Hopefully this will encourage more open-source projects for expanding their continuous integration coverage. This is great news if successful for improving the quality of Rawhide, especially for any developers that may be consuming Rawhide directly in order to have a rolling-release developer base and ideally helping to see future Fedora releases ship on schedule.
The Fedora 31 plans are laid out as part of the feature proposal Gating Rawhide - Single package updates.
The initial plans call for this to be an opt-in process and only be done for single packages in the initial stage while multi-package updates will be handled later. The plan is to gate packages on continuous integration test results before being able to land within Rawhide, in order to prevent broken dependencies, uninstallable packages, and other headaches that come as a result of package failures.
At this stage the process though is opt-in and test coverage isn't mandatory, but hopefully those elements will change as this support matures. The gating of Rawhide packages (or holding back the packages from landing until they pass a CI workflow) has been talked about before for improving the Fedora release quality and the ending of their alpha releases while for Fedora 31 due out in late 2019 this could finally happen.
Hopefully this will encourage more open-source projects for expanding their continuous integration coverage. This is great news if successful for improving the quality of Rawhide, especially for any developers that may be consuming Rawhide directly in order to have a rolling-release developer base and ideally helping to see future Fedora releases ship on schedule.
The Fedora 31 plans are laid out as part of the feature proposal Gating Rawhide - Single package updates.
11 Comments