Hi everybody,
following the vote on critical and blocker priorities, I propose that we change the way we handle old regressions as in practice, we’re already not blocking releases for old regressions.
My proposal is the following: Replace the sentence
- The “Priority” field must be set to
Blocker
by
- The “Priority” field must be set to
Blockerif the regression was first introduced less than two years ago, otherwise it should be set toCritical.
The idea behind this is that any issue that is older than two years is not only present in the current LTS branches but also in the previous year’s LTS branches. As we expect admins to upgrade at the very least from LTS to LTS, any regression that was introduced more than two years ago won’t appear when an admin upgrades now. I’ve chosen two years as one year might be too short, when an admin upgrades in March, a regression introduced on the master branch in January of the previous year would still appear during the upgrade. I thought about explicitly mentioning the previous LTS, but this seemed to me too difficult and not applicable to contrib extensions to which the same rule would apply.
I’ve formulated the rule as “should” with the idea that very important regressions (e.g., leading to data loss) could still be blocker issues. The idea is that the default should be Critical, but depending on the severity of the regression, it could be higher or even lower in some cases.
This vote is open until June 16, 12:00. Thank you very much for your votes!
