0.6 Debrief
Topics
- Planning
- Design
- Development processes
- Testing
- Bug fixing, reporting and triage
- Build/Release
- Documentation
- Community
- General
More Of
- non code deliverable planning (wiki, website, service, etc.)
- close on tenets earlier
- planning efficiency
- shorter releases
- preplanning
- decoupling development and planning
- planning around architecture goals
- coordination between cosmo, chandler, scooby, hosting service
- public code review comments
- bugzilla for task tracking
- user testing of design alternatives
- public design discussions
- clear discrete questions on design list, with timeline for answering
- atomic design discussions with specific goals
- more summaries of design list discussions
- paper trail of work (more descriptive checkin comments, more discussion in bugzilla, dev list)
- design discussion about code before it gets written
- higher quality milestones
- checkpoint summaries in the last few months
- access to checkpoint summary from download
- automated testing (functional tests, unit testing, tinderbox observing)
- automated testing between components (chandler, cosmo)
- more perf load testing for chandler
- more perf testing of whole system
- more linux dogfooding
- more irc bug bashes
- test interoperability sooner
- talkback feature
- bugfixing as we go (instead of saving until the end)
- developer in loop for bug triage
- punt bugs sooner
- more clarity on bug council intention w/bug priority (p3/p4)
- more error logging for tinderbox
- tinderbox tracking automated tests, functional tests
- separate commits list for developer documentation
- schedule time for documentation writing
- css stylesheet for the documents
- more landing page-like wiki/website communication
- summaries of conversations
- complete notes in meetings
- demos at staff meetings
- staff meetings even when mitch is unavailable
- clear expectations about dependencies (chandler/cosmo)
- involve everyone in decisions about governance
- integrated help files
Less Of
- irc decisions (secret decisions)
- general design discussion in bugzilla
- manual testing
- bundling fixes into a single checkin
- fewer forks from upstream packages
- documentation and release overlap
- cross posting
- cc'ing people already on the list
- irc office hours for broadcasting information
- decisions made in hallway conversations (irc, skype, meetings)
- staticy polycom conversations in gotham
Same As
- tenets
- functional specs
- screenshots - look like what would be implemented
- specs in svn
- linking bugs, tasks to specs
- bugzilla for task tracking (apps team)
- philippe's bug and task graphs
- checkin comments
- checkpoint/milestone system
- irc sharing testing
- bug prioritization
- bug council
- fixing/reporting/triage cycle
- chandler landing page
- management style - engaged but not micromanaging
- multiple channels for feedback
Open Questions
- should we have a changelog? tag checkin comments? require someone to summarize?
- should we create a mock server for testing?
- chandler/cosmo field testing (beta) -- at universities?
- should we podcast meetings?
Feel free to add free form comments here!
--
ReidEllis - 11 Jan 2006
I made a
recording of the meeting (low quality - recorded over the phone).
I think recording meetings and making them available via podcast would:
* require little to no work, and
* serve as a valuable reference resource
As someone who joins most meetings over the phone, I would appreciate being able to refer to a high-quality (compared with phone) recording.