We showed Lars a prototype user interface. He liked pretty much every part of it, and had some ideas for new behaviors and new GUI elements. In my (Tim's) opinion, the hardest thing he asked for was the ability to cut an image into the clipboard and paste it into Microsoft Word, preferably in some resolution-independent drawing format.
We tried to brainstorm the design for our Java application using CRC cards. The session didn't seem very successful. Perhaps we were a little past the stage where the CRC cards would be useful, or perhaps the application is too simple in structure. Many of the classes turned out to be the GUI components (already known from the prototype) and a simple read-only database interface.
At any rate, in the face of a good bit of apathy, Rob and I decided to deal the CRC cards out at random. Each class "is" the person who holds it, so if you have a card that needs collaboration from another class, you have to contact the relevant person and get him to add something to the interface. We don't know yet if this will work.
Rob's comments on this: Our attempt at CRC cards was less than ideal. Putting everything out on the table made the problem seem larger than it really was, even though I already have a good idea how I wanted to do this, having already done a simplified version for the prototype. I also wanted to hold back to get some new ideas on the table rather than just reproduce what I'd already done, but this didn't seem to work very well. The alternative of assigning everyone cards and having them decide what each class they have needs to know about the others may not be working either since I have yet to hear from anyone. I suspect this was one of the cases where as mentioned in the beginning of the class everyone agrees on something when in reality they each have a different understanding about what was agreed to. Fun stuff. :>
Rob has spent time on the prototype interface: The GUI ran into some problems with properly displaying itself. In particular, the lower rule very often wouldn't be properly colored, and the text message section was often superimposed on it. This may be a bug in the JVM. Second, menus would occasionally be displayed far from the menu bar. Removing the legend panel would fix this, but I can't think of any reason why including the legend panel would cause such a problem.
Finally, Rob struggles with image packages: I had to use a M$ image editor to build the screen shots because GIMP was uncooperative. The M$ product didn't provide a very useful interface.
Xiao spent his time: