These are the "raw" retro points from November's Craftsmanship Day at Gale that I captured. Some more should be on the way from other sessions.
Note how the good/bad/improve poitns evolved over the course of the 4 code retreat iterations.
The "final retro" at the bottom was from the whole group across all sessions.
Code Retreat Retro 1:
- cross team pairing
- understood rules
- project, code, tests
- passing tests
- design emerged through tests
- Not enough time to do anything
- didn't get here early
- late start
- not sharing keyboard
- more time for first iteration ("sprint 0")
- pre-canned Eclipse project w/ JUnit
- pair with someone off-team
Code Retreat Retro 2:
- More code
- More tests this time
- Ping-pong worked well
- Tried different approaches
- People kept to-do lists
- Keyboard failure
- Not enough talking between pairs
- Codnig under red
Code Retreat Retro 3:
- First rule coded
- Built on previous experience
- Liked suggestions in pre-coding presentation
- Tried different approaches
- Duane is brilliant
- Refactored tests
- Variable naming was poor (Lions players?!?)
- Bink broke keyboard
- Easy to grab keyboard rather than explaining verbally what you're going to do
- Some people stuck in the pattern used in the previous iteration
- Bad test names
- Too much up-front design
- Build from where you left off
- More time
Code Retreat Retro 4:
- We were on the same page
- We had bad code to refactor
- Dan got something to run
- Many people got 4 rules done
- Low hanging fruit first
- More meta conversations
- Failed faster with code in test
- Amber loves her pair
- Easier start with "TDD like you mean it"
- Using a timer helped write good tests
- Too aggressive ping-pong hurt Duane & partner
- Tim talks too much
- Reluctant to delete a test
- Could not complete the exercise
- Deleted code before final iteration
- Clarify that we're not here to solve the problem
- Different stories
- Refactor prior iterations if we're practicing clean code
- Offer problem-specific options
- More than one right way to do it
- Experienced the horror of the EI team
- Safe way to work on ugly code
- Enthusiasm was high
- Group size was initially good (around 15)
- Shared many points of view openly
- Duration was good (shouldn't be longer)
- Awkward chair arrangement
- Font size on Erik's laptop
- Too large a group @ the end
- Loss of focus @ the end (people filtering through)
- Unclear goals / guidelines
- Too many side conversations
- Coding before thinking5
- Refactor should have good tests before we start
- Mike S. says heckle when you're pairing
- Do a review first & plan refactors.
- Drivers should solicit more feedback
- Clarify the navigator's role (filter / collector)
- Have a scribe to capture what refactors we did
- Mike S. should do then next one.
Lead-in points for final retro:
- Get a wristband if you don't have one!
- What epiphany did you have today? Any "I never thought of it that way before!" moments?
- What should we do again next time?
- What was incomplete? What should we do that we didn't?
- What else can we do differently next time?
- Who wants to do a session next time? (Drew and...)
- Continue at Roosevelts...
- JQuery / new technologies
- What vs. how tests
- Writing stories
- Legacy code (but spend more time on it)
- Code Retreat
- Craft day as a whole!
- Where are the delivery leads? Include them.
- Solve a real problem (but don't keep the code)
- Different tracks on different days - people missed out on the story & estimation track.
- Better preparation by presenters (John & Jason called out)
- JSP testing
- More overlap with QA team
- Where are operations people?
- Share the retrospectives after this!
- Get off-site employees involved
- Legacy support sessions
- Pair with different people
- Drew - What vs How tests
- Mike Gantz: anonymize some of our code prior to next day.
- TC & Bernard: Definition of Done
- Jason: JQuery again, with prepared hands-on code
- Mudassir Hameed: Legacy Code