We’re all over the map today, from general theories of software development to low-level optimized bit-banging. Well, all over the software map, I guess.
From Ars Technica, check out Jon Stokes’ Cache and memory in the many-core era. Despite the infelicitous phrase “As Moore's Law increases” appearing early on (no, it’s the transistors which increase, the law stays the same), it explores usefully what the system component builders are up to in a fashion quite accessible to this software weenie.
Here’s the promised general theory, two of ’em in fact, presented by Reg Braithwaite in Which theory fits the evidence? He names his theories “D” and “P”; the contrast between which best explains observed outcomes, as opposed to that which does better in the marketplace, is kind of saddening. This is an important piece but the people who need to read it probably won’t.
The population of people who will enjoy Bit Twiddling Hacks by Sean Eron Anderson is pretty small, but those people will really enjoy it. I did. The title is accurately descriptive.
I also enjoyed David Ing’s C# 3.0 Considered Rubenesque? It’s amusing and instructive, and of course much of it applies to modern-day Java as well.
Somehow last month I missed Joab Jackson’s Ruby’s easy but Java is quicker, which is surprising since the Java One paper was by Sun people. The data is interesting, but this is totally a moving target, with lots of Ruby implementations being cooked up and Java not standing still either. I wonder if it would be a good idea to take the Daly/Cooke model apps and have someone run them on a regular basis every six months or so against the latest Java and Ruby builds?
Saving the best for the last: Elliotte Rusty Harold’s North and South proposes an amusing, and to my eye crushing, metaphor to describe the relationship between the WS-* and REST approaches to the world.