Eric Armstrong


Date: Fri, 21 Apr 2000 21:21:39 -0700

From:   Eric Armstrong

To:     unrev2,

Subject:   Meeting Summary for April 20, 2000 at SRI

We had a rather good meeting at SRI this month.

Highlights of the Formal Meeting

This was the official meeting, held at SRI.

  1. The (primary) development activities that go on before the product is released (designing, planning, coding, documentation, etc.)

  2. The (primary) development activities that go on after a product is released (bug tracking, suggestion lists, and enhancements).

  3. "Breaking in" a new developer (choosing a bug to work on, mining the code for rationales)

  4. User activities (reading documentation, asking questions)

In an effort to prune the list down to the "first cut" activities, it was observed that some of the activities represented "formal language" processes, or "formal processes". Those activities included project management, coding, bug management, and testing. The rest of the processes were "natural language" processes, which means that the system which is effective for one is likely to be effective for all of them.

[My inclination is to counsel a sharp focus on those activities, excluding the formal processes for the moment. The only conceivable counter argument is that the design really needs to take them into account, in order to ensure future compatibility with those systems. For some reason, my "gut feeling" is not to worry about it -- but I want to leave the door open for alternative views. It helps if we can restrict our focus, but...]

Highlights of the Informal Meeting

This was the unofficial meeting, held at the Applewood Gourmet Pizza palace, where they have salmon pizza and barbeque chicken pizza, and all kinds of good stuff...

  1. It *could* be a coding dead-end.

    That is, it might take us part of the way to where we want to go, but leave us with no good way to progress from there. I don't *know* that to be the case, but I'm concerned that it may be.

  2. The rather interesting tidbit of information that surfaced recently: That the folks who build the HTML-page-annotation system (Crit) found themselves using EMAIL to carry on discussions -- even though they tried to get each other to annotate HTML pages.

    This led to the observation that there is something seductive about the email interface -- the immediacy, the way information comes to you, and asks only for a reply. That, in turn, led to the realization that email the right interface, but the wrong data structures.

    Add good data structures to the system, and the result should be interesting...]


Eric Armstrong