Clarifying Unseeded Universe Freeze

[Scott Kitterman]

On Friday, September 30, 2011 09:16:21 AM Kate Stewart wrote:
> On Thu, 2011-09-29 at 01:50 -0400, Scott Kitterman wrote:
> > Clint Byrum <clint at ubuntu.com> wrote:
> > >Excerpts from Kate Stewart's message of Wed Sep 28 15:36:09 -0700 2011:
> > >> Dear Ubuntu release team members,
> > >> 
> > >>   On the #ubuntu-release IRC channel today the discussion of
> > >>   when
> > >
> > >should
> > >
> > >> the unseeded universe freeze came up today. A proposal was made to
> > >> freeze it 36 hours before the final images are published.
> > >> 
> > >> The precedents that exist in our current process documentation
> > >> right
> > >
> > >now
> > >
> > >> are:
> > >> 
> > >> https://wiki.ubuntu.com/ReleaseProcess
> > >> 
> > >>       * Work with universe/multiverse community to identify
> > >>       delegates
> > >
> > >to
> > >
> > >>         approve Feature Freeze Exceptions, in addition to
> > >
> > >ubuntu-release
> > >
> > >>         until the date of Final Freeze for universe.
> > >>       
> > >>       * Set the Final Freeze date (typically at release
> > >>       minus 5 days)
> > >>       
> > >>         for universe/multiverse for the packages that are
> > >>         NOT found
> > >
> > >on
> > >
> > >>         any installation media.
> > >>       
> > >>       * Universe/multiverse delegates and final freeze date
> > >>       is
> > >
> > >broadcast
> > >
> > >>         to ubuntu-devel-discuss and ubuntu-devel-announce.
> > >> 
> > >> https://wiki.ubuntu.com/FinalFreeze
> > >> 
> > >>         * For packages in universe that aren't seeded in
> > >>         any of the
> > >>         Ubuntu flavors, this final freeze is nominal;
> > >>         packages must
> > >
> > >be
> > >
> > >>         manually accepted by the archive admins, but no
> > >>         additional
> > >>         approval is required until the Unseeded Universe
> > >>         Final
> > >
> > >Freeze.
> > >
> > >> Have I missed any other references to the Unseeded Universe Freeze
> > >
> > >that
> > >
> > >> are relevant?
> > >> 
> > >> What do others think?   pro's/con's of moving from 5 days to 1.5
> > >
> > >days?
> > >
> > >This is not all inclusive, but these things come to mind:
> > >
> > >Pro's of 1.5 days vs 5 days:
> > >
> > >* Less 0-day SRU's (any bugs found late can be fixed in release,
> > >easing
> > >the burden of the heavy SRU process)
> > >
> > >Con's:
> > >
> > >* Approving packages means more duties for the release team close to
> > >the release date
> > >
> > >* Time spent fixing things in universe is less time spent testing main
> > >
> > >
> > >I think having more fixes in the released universe is a good thing for
> > >quality, but I worry about the impact, even if its small, on the
> > >release
> > >process as a whole.
> > 
> > In my experience there's no impact.  Even if a group of MOTU get fired
> > up and fix a bunch of stuff, getting things reviewed isn't an issue.  
> > Some releases we do get a motivated group at the end that really makes
> > a good bunch of changes. It's amazing what a few focused MOTU can
> > accomplish.
> There is an impact to those members of the release team who are also
> working on stabilizing the final images in terms of dealing with the
> interrupts and context switches.  The last week is pretty full with
> getting the last fixes in and tested, updating release notes, announces,
> etc.   Since the archive is frozen,  every fix still does need to be
> reviewed for sanity before being accepted.
> 
> That being said,  if the release team members who are not directly
> working on the final images (usually those without archive admin
> capability) are willing to be the handle the bulk reviews and
> interfacing in #ubuntu-motu channel during this period it could be
> workable to have the later date.   There does need to be a 'go to'
> contact on the release team identified, so folks know who to ping during
> the interval if they want something reviewed/included for the unseeded
> universe packages.
> 
> Stefano Rivera (tumbleweed) has been the MOTU representative this cycle,
> so he seems to be the logical "go to" person.  Note, this does not mean
> that he has to do all the reviews.  :)   Basically he's there to be
> contacted when someone may not have an existing relationship with
> another release team member to ask, and help balance the review load.
> I've checked with him and he's willing to help out over the next couple
> of weeks in this way.
> 
> If after the release meeting, there are no objections brought up.   I'll
> do the updates to the existing process pages this afternoon to reflect
> the new timeline, create a new one for Unseeded Universe Freeze along
> the line of what is outlined above, and send out mail to the developers.
> 
> Thoughts?

Such reviews are always on a best effort basis.  

We've never had a problem reviewing everything in the queue, so I don't see 
benefit in trying to limit the workload.

Scott K