Saturday, March 3, 2012

How will the field of software analysis emerge?

The software field is still very young; I was lucky enough to enter it when it was just beginning, and I'm still alive and actively practicing. I think there are very few fields of study which are so young and immature, and yet so incredibly active.

In most other fields, new students are introduced to the field through the study of classics. Literature students study Shakespeare; physics students cut their teeth on Newton's work; artists study daVinci; musicians learn to perform and comprehend Mozart.

The software field still seems to do very little of this. There is the occasional work such as Beautiful Code, or Coders at Work or the much earlier Literate Programming, but these works are few and far between, and I don't get the feeling that they are viewed as successful, or that they are finding enduring usage in students of software.

I expect that at, some point, software education will incorporate a greater "study of the classics" aspect, although clearly the field must still mature in order to even comprehend what "the classics" might be.

When that time comes, I believe that it will be incredibly valuable to have detailed version control histories of those great works, because, in clear distinction to the works I mentioned at the start (Shakespeare, Newton, etc.), great software is almost never a single-person effort, and almost never emerges fully-formed.

Great software is written by great teams, over years or decades, haltingly and painstakingly, with trial and error, bug and fix, initial raw idea and later refined variation.

I hope that tools like those we make at my day job will be able, not only to endure long enough to be part of this eventual emergence of software study as a field, but also to attract bodies of software that are maintained in that SCM system long enough to record all those design decisions and intermediate versions.

At work, we have the entire history of Perforce in Perforce itself, and I can crawl through the history of Perforce itself, going back 17 years, and see the evolution of the product and its concepts, read the ideas, both good and bad, of those who worked on the software before me, and incorporate all of that experience into my own work.

It's an incredibly valuable tool, and I'm not sure how many people realize how valuable this can be until it's too late. Most young students, I believe, get started in programming working on homework assignments and class projects. Their work goes from start to finish in a matter of days or weeks, and they rarely if ever return to it. I rather doubt that students bother checking their work into an SCM system, but I'd love to hear otherwise.

I don't have any clever way to close this post, I just wanted to record some ideas that I was thinking about as I was lifting weights the other day...

No comments:

Post a Comment