My rant for Computing in Science and Engineering is finished; I've rewritten a key paragraph of the grant proposal that's due on Friday (though it's going to take most of tomorrow to re-synch the rest), and most of the changes "requested" by the publisher of my next children's book have been made (though I'm left wondering where the story has gone---I mean, if the villain isn't allowed to call his monster dogs "stupid, lazy, brutes" because someone's feelings might be hurt by the word "stupid", it's no wonder kids would rather play videogames). I should dig into the pile of CSC301 re-mark requests lurking next to my lunch, but instead I have cranked up some early Zepplin while I search for a general "what's new and cool in CS research" blog. No luck so far, but I have found this:
Hertz & Berger: "Quantifying the Performance of Garbage Collection vs. Explicit Memory Management". Using code instrumentation to create different memory management oracles, they conclude that "...with five times as much memory, a [really modern] collector...matches the performance of [one particular kind of] explicit memory management. With only three times as much memory, the collector runs on average 17% slower than explicit memory management. However, with only twice as much memory, garbage collection degrades performance by nearly 70%. When physical memory is scarce, paging causes garbage collection to run an order of magnitude slower than explicit memory management." I'd normally recommend taking results like this with a grain of salt, but their rigor is impressive.
Cheng, Desmond, & Storey: "Presentations by Programmers for Programmers". The authors build on a waypointing tool for Eclipse called TagSEA to create a system for giving guided tours of software. I haven't played with it yet (can't wait for this term to be over), but it's one of those things you don't know you've been looking for until you discover that it exists, at which point it moves directly into the "essential" category.
Finally, I'm trying to find as many academic studies of software project portals (like SourceForge, Trac, and DrProject) as I can. Lots of people have used data from these portals (particular SF) to study software evolution, open source social networks, etc.; what I'm after is studies of the portals themselves: how usable they are, how they shape projects that rely on them, what other tools are typically used in conjunction with them, etc. If you have pointers, please toss 'em my way.
comments powered by Disqus