A recent HBS article Mastering the Three Worlds of Information Technology categorizes IT projects, among other dimensions, as to how much organizational change they require. I think a large part of the problem -- ignoring incompetent programmers or users -- with IT project delivery is that few people understand (even during) or appreciate (beforehand) the amount of organizational change a project might entail.

Even though it's mainly a mathematics cheat sheet, this "Theoretical Computer Science" cheat sheet could be a useful/nice quick reference.

Martin Dengler v2.0 was released today (a week late, but for a 40 week project that's much better than the IT industry average). Migration to the new system is expected to take at least 16 years.

Bill Joy officially says emacs is better than vi -- but he sounds like a good guy, so read The History of VI in this Interview with Bill Joy.

• Geforce 6800 GT; performance within 20% of $800 cards at 1/4 of the cost.
• Athlon 64 X2 3800+ -- same dual core as P4EE at 1/2 the price
• nForce 4 430 + GeForce 6150 IGP: perfect HTPC mainboard
• ORIGENae X15e -- Best HTPC case

Posted some code on the projects page...a simple treemap generator written in python.

I've added a contributed CCRMA apt configuration example to my list of Fedora Core apt/yum/up2date repositories.

Wiped out a power supply and with it, most of my weekend. Now back online!

I've updated my list of Fedora Core apt/yum/up2date repositories to reflect the new Fedora Core 3 release. Hopefully this will be of use to some people.

Clearly had a few website provider problems. Changed providers.

I've created an implementation of Splay Trees (see Sleator, Tarjan, 1985) usable as a member of the Java Collections framework. It's available under the GPL, ASL, and BSD licenses. Check out the code and my project page for more info.

I submitted a fix to my Apache Tomcat 4/5 CGI parameter handling code to apache based on code/suggestions from X and Y. The patch was accepted by Apache and is in Tomcat 5's CVS. This fix addresses Tomcat bug #19998.

Code for exploiting some current Bluetooth-enabled phones' insecure bluetooth/OpenOBEX protocol stack (aka "bluejacking") is available in C (basic tarball here). OpenOBEX-tools required (Linux pre-compiled binary here. This is only proof-of-concept code but works "in the wild" (tested legally). It is based on the hole reported by the BBC and others. As it is just proof-of-concept it is very unpolished and some programming experience is required to build it; instructions and published reference papers available on request.

Played around with the storage and processing of tick data (stock prices). Standard OS performance metrics are useful for low-level performance stats but not immediately useful to determine how well tick capture and retrieval processing will perform. My primary goals were/are: 1) to put upper bounds on the performance of basic tick management operations; and 2) to determine relative performance benchmarks for Java and C# implementations available today.

I put together a scope document and implemented two basic benchmarks as an experiment. Initial control runs have proved that something is being done and being measured, so the next step is to determine more useful testing parameters and deploy the tests on more platforms.

Right now the tests are written in java but I will be coming up with a .NET version soon.

The scope document and JRE vs. GCJ test results are available in draft form -- comments welcome.

Fedora Core version 1 is out, so I updated my Fedora Linux repositories page to help people keep track of the latest & greatest ways to keep up to date. Hope it helps. Even got a tiny mention on slashdot.

I created a page to help Fedora Core Linux users keep track of the latest & greatest ways to keep up to date with the latest software releases (typically daily): Fedora Linux repositories page. Hope it helps.

The headline of this could be true if a new EU directive is passed. Many web site owners are shutting down their websites temporarily in a protest to demonstrate the dangers of this type of law. Sofware patents can get you prosecuted for publishing texts you wrote yourself...

• It is ethically scandalous, because in today's world, knowledge, information and ideas can not be separated from their technical representations and the software that manipulate them. It would allow patent offices to further develop the giant auctioning of the domain of ideas and knowledge, when this domain was always considered as a precious common good, that can not be turned in anyone's property.
• It is economically unjustified, because the very arguments that have been used to justify patents for mechanical and chemical industries, or more generally manufacturing, do not apply in anyway to software. No need for software of those monopolies without which one could hesitate to build a production plant.
-- 30 Computer Scientists petition against the Software Patent Directive

For more information, visit swpat.ffi.org.