Primary Proposal
The initial proposal for this approach was put forth on Mark's Blog 'Not convinced by rolling releases'
The essence of the proposal is to use what is essentially our current release methodology, but to increase the rate of release dramatically, resulting in increased drive to streamline our release process while releasing fresh software to users faster. Note that the "monthly" cadence is implied, but the heart of the proposal is to release more frequently.
- Make the update process from point to point really bulletproof. Upgrading today is possible, but to keep the system clean over multiple successive upgrades requires an uncommonly high level of skill with APT.
- Strengthen the definition of point releases in the LTS so that interim releases are obviously less relevant.
- Do a reasonable amount of release management on, say, MONTHLY releases that they are actual releases rather than just snapshots.
Benefits
- Fresher software in the hands of users
- Fewer stabler releases to support conserves engineering efforts for current development
Drawbacks
- Cost of doing release management each cycle
- Users must update each month to receive support
- Impact on community flavors and documentation and quality
Sub Proposals