Pressing On (Tuesday)

This is a real "cold shower, hot shower" day. Every time I cross one risk off my list, I have to add two more. I've made quite a few discoveries today. For lunch I decided to continue work; I stay in and order a pizza because I can't interface with the database. The thing crashed because my version of the database software is too old. Also, I didn't read the API specification carefully enough. I lost an hour with tech support, then downloading the right version, then studying the documents.

I haven't made much progress, and I'm starting to think the whole thing was a bad idea. There are so many things that can go wrong, even with such a ridiculously small app!

I have built a UML model, though. Just a single class diagram and two collaboration diagrams. While I was at it, I also built a component diagram based on my architecture sheet, which I can discard now. I pin all my remaining sheets on the wall.

Since my Risk List was starting to look like a mess, I put it on my PC in an Excel worksheet.

I am not changing the Vision, but it is now accompanied by 17 Post-It notes, all with questions and issues. I start adding constraints, such as

• The code shall run on Windows NT or UNIX.

• The database runs under Windows NT 4.0 or above.

When Gary arrives for dinner with his colleague, Eric, I am putting the latest touches on a not-too-unreasonable prototype. All the data is canned; there is only one user ("Gary") and one activity ("Think"), and I was able to do an on-the-fly "suspend and resume" as an extension to the Timing an Activity use case. The database runs on my desktop with the applet on my laptop through the Net. My Risk List is now down to a few simple risks.

We spend about five minutes playing with the prototype until Eric crashes it. He's shocked, but I explain that robustness and completeness were not the objectives. Then we discuss the look and feel of the applet and reorganize the fields a bit. We look at my list of questions. Eric is very concerned about losing data if someone has to reboot a machine while a counter is running. I promise to look into it, although I was hoping he would say to forget it. Maybe I shouldn't have brought up the issue.

I end up adding a couple of new risks to my Risk List and half a dozen more requirements to my Vision Document, but that is not too bad. I decide to leave the Plan as is. To the Vision, I add more use cases for System Administration:

• Clean Up the Database

• Add a User

• Clean Up the Activity List

The good news is that Gary is happy with what he saw, and he says to move ahead with the whole project. He does not object to the constraints.