DistributedApplicationsOnTheSunSPOT
From SPA Wiki
Photographs 1
Movie 2
Contents |
Retrospective on Sun SPOT Session
Things that went Well
- Cheat sheet was useful
- CLDC is more useful that previously imagined
- Working with Eclipse junkies
- "Whole thing was fun"
- Ant build "ecologically sound" -- didn't use too many cycles
- The tool chain was quick and simple
- Programming in pairs
Things that went no so well
- Test-driven design is not do-able on J2SE with the SPOT
- Didn't understand capabilities of the SPOT from the exercises we gave them to do
- Setting up took way too long, would have been good to prep participants more
- The documentation on the installer CD is not simple enough
- Too much of a learning curve to figure out how to discover API of sensors, needed to get going more quickly
- COM port busy problem when using eclipse (only sporadically, and only on Windows?)
Puzzles
- CD installer is too hard.
- Installing FTDI on Mac is not obvious
- What is the accelerometer (does it measure acceleration)?
- What went wrong with some of the transducer lib installs?
- Polling sensor state on battery-powered device will drain battery
- How do I uninstall?
- Mysterious process build (what actually happens when I type ant deploy and ant run?)
- How to run unit tests on SPOT
- When selecting participant pairs base it on tool experience rather than Java experience
Ideas/Learning
- Need a SPOT emulator.
- Code-along instructions might be useful
- Install-along instructions would be useful
Actions
- More precise instructions for the installation
- More specific storied. For example, define the UI rather than getting participants to dream it up
- Better Eclipse instructions, or a template project
- More examples using I2C, or 1-Wire
- Produce Test-driven examples, show how to do test-driven development
Other observations
Of the 14 participants, 10 used eclipse, 3 IntelliJ, and 1 emacs. 13 brought Windows laptops and one Mac OS/X.