A vision for Fedora’ development cycle

by davidnielsen

I was talking to my friend Peter Gordon the other night about how to optimize the Fedora development cycle. We both agreed that we needed to take into account that Fedora has two different kinds of users.

Users like me who aggressively track Development using Rawhide and technology preview releases. As well as core users who want a stable OS with few updates. These two goals are not immediatedly the same, sure stable users enjoy new technology but the transition of this is often times painful and timeconsuming.

We also need to take into account that our kind corporate sponsors at Red Hat need to make money and this means they need as good a product as possible to base RHEL on.

I propose a model where we keep an alternating 9 month technology focused preview release cycle followed by a 4 month polish cycle. Giving us 6 month of full on development, 3 months of bugfixing and stablisation and we release that as the FC technology preview release.

We can release these with known minor bugs as long as they are in a general decent state. The next 4 months will be dedicated to producing a system that’s polished and bugfree – if that’s all we are doing for 4 months a great product should be possible come release.

We could even release most of the polish release as updates to the technology preview release thus ensuring that long term it’s less variables to support. If we hand a polish release over to Legacy everytime a two polish releases have been released, or roughly every two years, great support is ensured for users for a decent portion of time (I don’t currently remember how long Legacy commits to support a release in addition to the time provided by Red Hat).

I believe this brings about the best possible solution for all parties, both those of us who love new technology and those who love stable and somewhat slower release cycles.