Fedora 11 Retrospective Recap

John Poelstra poelstra at redhat.com
Tue Jun 16 16:55:01 UTC 2009

Fedora 11 Retrospective Recap

Wiki version: https://fedoraproject.org/wiki/Fedora_11_Retrospective_Notes

Invitees: Cross section of people and leads from each group: 

Attendees: James Laska, Bruno Wolff, Jóhann B. Guðmundsson, John 
McDonough, Jon Stanley, Jarod Wilson, Todd Zullinger, Ricky Zhou, Steven 
Moix, Mike McGrath,Jack Aboutboul, Paul Frields, Eric Christensen, John 
Poelstra, Edward Kirk, Adam Williamson, Máirín Duffy, and Jesse Keating

Unrepresented Teams: Ambassadors, Translation

Location: Global teleconferencing system & Fedora Gobby

When: 2009-06-16 @ 14:00 UTC

== Meeting Guidelines & Agenda ==

1) Each person gets five minutes to say whatever they want.  Suggested 
questions to answer:
    a) What went well?
    b) What could have gone better?
    c) If time and resources were not an issue, what three things would 
make Fedora 12 our best release ever?

2) While the speaker has the floor nobody can interrupt or correct what 
they are saying
    --there are no "wrong" ideas
    --other participants can seek clarification from speaker after they 
are done

3) Other attendees will help capture points raised in gobby

4) Open discussion and go forward plans after each person has had an 
opportunity to speak

== Meeting Outcomes ==
o Individual feedback (see below)
o John Poelstra will create a wiki version of these notes and post link 
to in email recap to fedora-advisory-board
o Jesse suggested checking back in a month to see how we are doing on 
the things we said we said we wanted to change
   --email is fine; telephone not required
o Jesse strongly urged people on the call to take the proposals from the 
recent FAD back to their teams
   --feedback needed ASAP
   --will affect many parts of Fedora
o John Poelstra encourage attendees to blog about this meeting to give a 
wider audience an understanding of what the meeting was like and what 
they thought about it
o The Fedora CMS (Zicula) will affect a lot of groups in Fedora
    --We need a point person to coordinate these efforts for our 
greatest chance at success
    --No one has been identified
o ADDED AFTER THE FACT by John Poelstra
    --Anyone is free to tag onto the end of the wiki document to add 
their own feedback

== Legend ==

(+) Went well
(-) Could have gone better
(w) Wish-list--no limits--everyone CAN have a pony

== Jon Stanley (jds2001) - fesco ==
(+) Feature process went well
   --good contingency plans
(-) X item fell through the cracks
(-) Meeting minutes creation failed
(+) Started using a minute meeting bot
(w) Massive QA testing
(w) Earlier freezing/branching for critical packages

== Eric Christensen (Sparks) - docs ==
(-) Release notes churn at the last minute
   --some is legitimately unforeseen, other notes just not timely enough
(-) docs.fedoraproject.org page doesn't flow well
   --languages hard to find
(+) Turned out more guides and docs than past releases
(+) use of Red Hat guides that were relicensed
(w) Zikula CMS
   --will replace the existing docs.fp.o
(w) standardize and focus on release notes and meet more users point

== Jarod Wilson (j-rod) - FESCo ==
(+) Feature process went smoothly--deadlines understood and less confusion
(+) No major disagreements within FESCo
(-) No major disagreements within FESCo
(-) Meeting minutes
(-) Collection of public feedback not optimal (need more _constructive_ 
(-) Secondary architecture team updates
(w) Focused Fedora kernel QA effort--build up community

== John J. McDonough (jjmcd) - docs ==
(+) people were great
(+) use of publican went surprisingly well
(+) install guide is the best we've ever had
(-) integrating milestones better into schedule
(-) pushing to docs.fedoraproject.org is an ugly process
   --Too many formats
(-) Need more packagers on the docs team
(w) More engagement with developers/determining who to talk to
(w) PackageKit get fixed
   --problems finding application to click on
(w) Problems with wireless on NetworkManager

== Edward Kirk (tk009) - BugZappers ==
(+) Communication really good about what was going on between all the 
teams and across the project
(+) More people are joining the project and contributing
(-) Blocker bugs were not handled very well
(-) Problems at the end with Anaconda
     --identifying blocker bugs before the last minute
(w) A more solid, stable release that is technically better than 
anything else out there

== James Laska (jlaska) - QA ==
(+) 20 test days--way more than F10
(+) Documentation around test days and how to create live images, etc.
(+) Test results reporting
(+) Common F11 Bugs
    --multiple teams were invested in success
(+) Community testing
(-) No specific test days for Sound
(-) No concept of capacity
(-) Clarify hand off procedures between Release Engineering & QA for RC
    --"who has the ball?"
(-) Transparency around release readiness decision and whether to slip
    --public discussion & publish minutes
(-) Automation of testing
(-) No metrics around health of QA community
(w) Micro-communities around each of the features
     --specific dashboard
     --blocker bug tracking
(w) Community test case management system

== Adam Williamson (adamw) - QA/BugZappers ==
(-) reading several reviews that Fedora 11 is un-installable
(-) unclear how to get in late release notes additions in a cooperative way
(+) Test Days were fantastic
(+) BugZappers and QA wiki spaces were substantially revised
(w) A massive testing lab
(w) Improving kernel triage for future releases

== Bruno Wolff - Games Spin ==
(+) Help from Jeremy Katz to enhance livecd-tools for larger games iso size
(+) Rawhide updates come quickly from the mirrors and Koji
(+) test days were helpful and gave focus to people
(+) New Spin process documentation got developed
    --specific test instructions are now available
(-) Broken installer and video got in the way of testing
    --painful when rawhide was broken
(-) We don't have a process for maintaining recurring spins
(w) Less broken stuff
(w) Spins process document finished
(w) Lots of test days
(w) More with live USB with persistent storage

== Jack Aboutboul & Steven Moix - Marketing ==
(+) Lots of progress restructuring the marketing team
(+) Release schedule helped cross team coordination
(-) Need to identify task inter dependencies and add them to the schedule
(-) Better capture things that get accomplished
(+) Inter group communication with Art, Translation, etc.
(w) Host mini-meetings with different teams throughout the release cycle
(+) Continuous content production
    --building up and telling stories
(w) Fedora Magazine
(+) Tracking press reports in different languages
(w) Finding a better way to do metrics to identify how far our reach is 
and understand audience
(w) More marketing to younger audiences and greater public awareness

== Ricky Zhou - Websites==
(+) two or three people helping consistently with Web team tasks
(-) Late changes from marketing and docs team which had minor impact
(w) Add a review period so that issues can be identified and corrected 
before release

== Mairin Duffy - Design ==
(-) Need to see feedback to the Design list as opposed to sprung on blogs
(w) Six more designers on the team!
(w) pink pony!!!!11!!

== Jesse Keating - Release Engineering ==
(+) Better involvement with QA and Triage
    --improved confidence in QA beating on the trees
(+) Did a better job of managing the blocker lists
(+) Good help from others verifying blocker bugs
(+) Test Days ... more feature owners are requesting targeted testing
(+) No last minute compose-related surprises
(-) Better communication around what testing needs to be done
    --improve capture of testing outside of the matrix (mailing lists)
(-) Clearer guidelines around code changes
(+) FAD Proposals - 
(w) earlier review blocker bugs
    --maybe weekly after we branch
(w) automated QA which provides automatic feedback

== Jóhann B. Guðmundsson - QA Team==
(w) Taking test days to the next level
    --improved advertisement
    --re-use live images for additional testing
(w) coordinate with upstream testing
(w) litmus testing system added to Fedora
    --litmus is coded to be quite mozilla-specific (according to jlaska)
(w) Use of Fedora QA Trac system to track request from maintainers
(w) Host test day without requiring maintainer present
(-) lack consensus amongst maintainers if a reporter is supposed to 
report to our bugzilla or upstream one
(w) Daily scheduled testing on components.

== Todd Zullinger -- Websites Team ==
(+) more complete translations than we've had before
(+) Infrastructure team syncing release content earlier than past releases
(w) relocate everyone to a tropical island where we can all work 
together in person

== Mike McGrath - Infrastructure ==
(+) team is growing
(-) multiple release slips hindered work in infrastructure due to freezes
     --one big slip would be nicer
(w) Feature freezes in earlier
(w) Contingency plans for core distro functionality
(w) Look for key ways that Infrastructure can enable success of QA Team

== Paul Frields ==
(+) Well coordinated for the releases including slip announcements
(+) Ambassadors did a great job behind the scenes
    --media production
    --release parties
(-) Feedback mechanisms are unclear and point to too many different places
    --documentation and marketing materials
    --could anaconda bug reporting mechanism be helped by design team?
(w) Marketing, Docs, Website, Design team work on coordinated approach 
to the Zicula (CMS)
    --organized and well coordinated fashion
(w) More forceful in not allowing late/invasive changes to rawhide and 
putting them off to next release

More information about the fedora-test-list mailing list