Friday, October 5, 2012

Greg Wilson on software engineering

I actually ran across this in August (yes, I have a six-week backlog on bookmarked things to post about - mostly I put them back into the stream in the same week I encountered them, but this one is really good and I want to think about it a little more than just posting a "this exists" post.)

Greg Wilson has been bopping around the software world for a long time now, and is concerned about the science of computer programming.  As in: there isn't one.  He has a fantastic slideshow here, which you can scan through in about three minutes, and it all leads up to his book "The Architecture of Open-Source Applications".

I've just grazed the surface with it, but before I spout off some thoughts, let me inject a couple more links:  http://www.neverworkintheory.org/ is a blog about software development research that is relevant in practice (says so on the header), http://software-carpentry.org/ is an organization teaching researchers how to code better, and Wilson's own blog at http://third-bit.com/ - those are the links at the end of the slideshow.

OK. So.  Architecture of open-source applications, yeah.  This is essentially a list of high-level descriptions of the shape of the code for 49 different serious open-source applications and a little rumination on choices made.  This is higher-level than Decl core is looking, but clearly a separate semantics of application architecture would be really nice to have.  To that end, I need to read this book, cover to cover.

What I'd like to do is then look at some of these applications at the code level, and build a "semantic framework" describing the code.  I'm not even sure what that means yet.  But I want to evolve towards code understanding code (for certain values of "understanding").

No comments:

Post a Comment