So therefore, what happens—you take a regression test that maybe has 14 million tests and you can pinpoint the—maybe—dozen that hit the things you’re changing. Now we used that in the year 2000, because the year 2000—the year, the data field—was driving the whole thing. But you find often that the organizations producing quality software are getting it not so much as the result of a new paradigm. It’s by engineering principles, stuff like regression testing and the stuff we’re talking about.

Let me just do a little bit more here—how am I doing with time?

CORK: The good news is that we’re going to try to copy Tom’s slides—and everyone else’s—and have them to distribute tomorrow. The bad news is that I need to ask you to wrap up in about five minutes.

MCCABE: Let me first take questions. Yes?

MARKOSIAN: This question isn’t directly relevant to the application, but more of a historical question. [inaudible] And then things got better over time, and now there’s been remarkable development that seems to be sending things back to the beginning, which was to proliferate current systems. What you’re doing there is introducing two things: the enormous complexity, the syntactical kind of complexity that’s represented in your chart. And also that complexity is hidden because it’s not available for the programmer to look at, things like leaving to the operating system. So, do you have any approaches to that?

MCCABE: Yeah, there are a lot of them, and some are not pertinent to the subject here. There are a lot of just engineering principles, about testing and project management, that pertain to that. And there are collaboration tools that help with that. So there’s no silver bullet. But within that I would suggest that, within the newer systems, these issues are even more important because one of the facts of life these days is that you get different contractors and people separated geographically in different countries all working on the same thing. And, [for instance], what Microsoft does, when you recompile every day, the idea is to visualize the thing you have. Easily and frequently, and share that. And it’s probably more important now because of what you say than it was back then. And things are growing in complexity, not the other way around. And the environments are more complex, in fact, and hearing the standards is even more important.

So I think—let me just summarize by saying that you’re saying that we want to know things about data complexity, and that’s a way to view a subset of data, and then what complexity it induces, and how to test that and how to change that. And a lot of this is about visualization. It was striking to me as we discussed it this morning that the issue, it seems to me about surveys, vis-a-vis documentation and testing, requirements being together up front—are entirely analogous to what has been happening in software. And in software the joke is that you never get to



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement