Readers/Web/Team/Retrospectives/Q2 Retrospective 2013-2014

Mobile Web Q2 Retrospective; 18 Dec 2013, 2pm - 4pm PST

Agenda

• Working agreements
• Gather data - Timeline
• Gather insights - Timeline analysis
• Decide what to do
• Close

Working agreements

Brainstorm Examples: One person speaks at a time, Start & finish on time, Be honest, It’s OK to challenge

Take the time to understand the goal/motivation of a commit regardless of whether you agree with it before you criticise it to death in a public setting If you want to object to something stylistic in a code commit and it is not part of the project conventions, suggest it be added to the conventions or discuss with the committer rather than just -1'ing it Write more verbose commit messages

only those below?

Have everyone the opportunity to speak about something they're passionate about Come to resolution on one topic before moving on to another one No personal attacks Provide specific examples rather than speaking in generalities When we provide examples, they are to make things better - not personal attacks! Timebox discussions Bathroom breaks once per hour Moar snacks If things are getting heated take a timeout No weapons

Timeline of Q2

Iteration 0 - Grasshopper (30 Sept - 14 Oct)

Defined metrics for upcoming iteration+ Animations to stable+ Desktop typography changes launched Switchover from Vibha to Kaity+ Deployment train[+-++-] Moving to LESS php [+++] Kenan starts paying more attention to beta and alpha and we get better at avoiding feature rot [+++] Velocity ~22 points

What did we notice?

Mixed reactions to deployment train Retrospective seemed useful since two of the positive things came out of it

Iteration 1 - Honey Badger (14 Oct - 28 Oct)

Beginning of User profiles [+-] Rob Moen joins the team [++] Issues trying to get QA running - meetings begin with Michelle to rectify our QA mess [--] Quarterly review meeting+ Lots of followup deployments needed [--] Velocity ~35 points Moved TOC stripping to core

What did we notice?

This iteration has issues Resource sharing helped us get VE much quicker Good velocity! Why? Still getting used to the deployment train Goals for user profiles were confusing

Iteration 2 - Iguana (28 Oct - 11 Nov)

Jon discusses external link icons being useful on wikitech - no one [citation needed] from mobile web involved in conversation [-] - I think I replied there Notifications overlay moved to stable [+++] Abuse Filter supported added - errors go down! [++++] Near this page to beta Mobile and designers get together to discuss how we can push overlays forward [+++] MediaWiki UI meetings begin [++-+-] Central notice user survey campaign [+] Velocity < 10 points Diff engine refactoring+

What did we notice?

Launched awesome features - but really low velocity? Is velocity an iteration delayed Lots of bugs fixed, spikes, moving features to stable

Iteration 3 - Jackal (11 Nov - 25 Nov)

Kaldari and Juliusz going to India and us having a huge code review backlog [-] Jenkins now runs QUnit tests! [++] Limn graphs stopped updating [---] New overlay designs get kicked off beginning with Editing [++] Edit guider A/B test+-(lots of confusion around this one) Dario's help with AB tests [++] Velocity ~50 points QA failure emails [-] One direction apocalypse [++]? Git deployment logs begin again [+]

What did we notice?

This graph has issues - Limn problems, highlight importance of the Limn graphs High velocity, but due to items being accepted during iteration Iguana - this is confusing! Felt really focussed and that we achieved a lot

Iteration 4 - Kickassaroo (25 Nov - 9 Dec)

Arthur/I asked mobile-tech if someone could help me write a template RFC and there was no response [-] Mobile Webfonts merged (unused currently) [+-] more modal overlays Added intro editing to User Profiles [--] Kaldari sick for a week! [----] Lots of stories were not be accepted, had to be moved to next iteration [--] disagreements about microcontributions - should mobile surface article issues? [+] Profile design changes again [-] Velocity < 10 points More QA failure emails Michelle leaving

What did we notice?

Unfocussed iteration - half the team was out, Name was not an animal -- Things that got stuck in ready for signoff got moved over

Iteration 5 - Liger (9 Dec - 23 Dec)

Firefox OS app kick off [+] ...but no one gets excited about it seemingly on the mailing lists [-] Diffs [+?] (new difference engine) Jon gets angry with Kenan and Juliusz for interrupting the firefox os app and apologises [-] 3rd floor runs out of Capri Sun [--] New KeepGoing Workflow (add links) [+] Vibha starts working only on VE, can't work on mobile [---] and Moiz becomes the secondary designer [++] Visual Editor prototype working in mobile on betalabs [+++-] Oh no not more QA failure emails!1

What did we notice?

Also a fake animal, real (but can't breed) Jon felt burned out, partly because previous iteration felt unfocussed Kaldari had to get clarification on a lot of things

Connections and links:

• Not one positive QA thing; QA issues spanning all of the iterations+++
• Velocity is unstable+
• A lot of instability (resourcing), people coming and going
• Doing a lot of experiments, but not delivering a lot value to users - but then again, some of this stuff may make it out to the users eventually+
• Lots of confusion! Chaos!
• All the iterations had awesome names (possibly with the exception of kickassaroo)
• We could have done with One Direction, but we had many directions

Patterns

• Starting to move stuff into core, which is good! ++++
• Communication problems/collaboration problems +

Shifts:

• Resourcing
• Changes in direction of focus +
• Working in a less isolated fashion++

How do these affect our current issues?

• Resourcing shifts really hurt us+
• Changes in direction of focus are contributing to communication problems, and vice versa (having themes for each sprint would help)++++++
• Could have helped if designers could have made prototypes for things like Profies (with clear goals/hypothesis)+
• Saying NO would have helped more +++
• Need to get better at giving designers a path to productizing experimental work++

Goal brainstorming

SMART goals! Specific, measurable, achievable, relevant, timely Starting to move stuff into core, which is good! ++++

• RfC for some things
• Avoid duplicating features that already exist in core
• Lack of database models that can be shared
• Get someone from platform that could be on call for core reviews/sign-offs <-- unrealistic:)
• Goal: move X number of features into core by end of quarter
• Can we help other people understand what Mobile is doing? Showcases with other teams?

1. Start using mediawiki.ui in MobileFrontend (by end of quarter) 2. Move javascript templates to core (by end of quarter)

Not one positive QA thing; QA issues spanning all of the iterations+++

• TDD workshop
• Encourage tests in code reviews, especially for user-facing features
• Come up with requests and guidance for the QA team on how they can help us

1. A dedicated QA point of contact (by end of quarter) 2. At least one browser test for every feature that goes to stable (by end of quarter)

Changes in direction of focus are contributing to communication problems, and vice versa (having themes for each sprint would help)+++++ Saying NO would have helped more +++

Internal GOALS for Q3

• Start using mediawiki.ui in MobileFrontend (by end of quarter) - Jon
• Move javascript templates to core (by end of quarter) - Kaldari
• A dedicated QA point of contact (by end of quarter) - Arthur
• At least one browser test for every feature that goes to stable (by end of quarter) - Juliusz
• Limit # of projects we take on in a quarter to 2 projects - Kenan
• Every iteration, remind ourselves of two projects/goals - Kenan
• Norm: Anything that doesn't map to those high level goals remains in alpha - Arthur
• Loosely schedule how those goals will fit into the iterations during quarterly planning - Arthur

Retrospective of this retrospective

What worked well?

• End result feels good - concrete goals!
• Iterative process to get to the goals felt good
• Using SMART was a good idea - we came up with goals that have a chance of being completed
• Having lunch!
• Gave us concrete examples of things to discuss - rather than hypotheticals; helped steer conversation better
• Winky the wiki parrot was good

What do you want changed in the future?

• Could have used more time to discuss the goals
• Didn't get to some of the big picture items - like code review
• Not sure if having a precise timeline was valuable
• COuld have done some work done in advance for this - like coming up with stuff for the timeline

Puzzled?

Who is going to own the goals? [Done]

Parking lot

• Start tracking how much code gets merged in an iteration
• It would be nice next quarter to set a goal for the end of the quarter of features being launched