Owen O'Malley wrote:
Rather than set a date, I think it is time to move to feature-based releases.

The problem with feature-based releases is that one feature can delay releases arbitrarily, so a single feature can hold all others hostage. Rather I think we need to be willing to remove or disable a feature in a branch if it proves problematic. If an addition causes regressions that are not resolved promptly then that addition might be removed or disabled. This levels the playing field: features that doesn't cause regressions can be made available in a release in a timely, predictable manner and are not held back by more problematic features.

Doug

Reply via email to