Monday, June 18, 2012

Let's Roll with 12.10


As a consequence of our daily quality efforts, some very interesting developments have taken place for 12.10.

First, while knocking around UDS, it occured to me in a bit of a flash that all of the effort that we invest in freezing the archives to make Alphas and Beta releases for Ubuntu is wasted work that slows down our velocity. We have daily quality and we have started using -proposed in the development release, so the chance of having an uninstallable image is greatly reduced.

(you can read the discussion on @ubuntu-devel)

So, why do we have Alphas and Betas? After some discussion, it seems to come down to:

  1. Because we want to encourage widespread testing by community members on a variety of hardware at a regular cadence
  2. Because we want targets for features and bug fixes
  3. Because we need to test our ISO production capabilities
  4. Because we always had them

Does all the effort in freezing the archive actually help? I don't think so. In fact, I think it is counter-productive.

  1. We can do the same testing with daily images. Furthermore, we can do that testing at a cadence of our liking, or even out of cadence if we want to squeeze in a special test run at some point. The ISO tracker nicely accomedates this now.
  2. Freezing the archive, by definition, *stops* packages and therefore bug fixes and features from getting uploaded. 
  3. Surely we don't need to slow down everyone's work so that we can try producing ISOs, and surely we don't need to do it so often and early.
  4. Of course, "because we always did" is not much of a reason.

It seems that what is needed is a regular cadence of deep and broad testing by the community to augment our automated tests, along with trial runs to ensure that our ISO building tools and process are working. Thefore, I propose we:

  1. Stop with the alphas and betas and win back all of the development effort
  2. Increase the cadence of "ISO testing" to whatever we want or whatever the community team can manage
  3. Spin a trial ISO near what is not beta time
  4. Spin ISOs for release candidates

3 comments:

  1. Sounds a good approach to keep the work flowing.

    Its good to still have goals and targets, so people can judge what can get done in time. However, those goals & targets should be flexible enough to allow valuable work to get in to the distribution.

    Thanks

    ReplyDelete
  2. One solution would be to declare 12.04 or 12.10 a rolling release. Updates and fixes could then be released when they were ready, not on a set date. The users would get far more stability, and all the pressure and shortcomings endured by the developers would dissolve overnight.

    ReplyDelete
  3. +1 for rolling release... it seens the best way of delivering things for me.

    ReplyDelete