Years passed, and now I am tired of listening to the whine of PC fans that saturate my living space. I have built out a home network with multiple machines as a kind of lab playground. It’s been fun and educational, but now I’d like less noise and clutter. The FitPC is an interesting little system running on AMD Geode and Linux, with a small profile and no fans. The mini-ITX systems at Mini Box are also interesting. Everyone seems to be getting in on the action, with the “mini computer” category on NewEgg.

And then a friend pointed out that I have the perfect hardware under my nose. It just needs an upgrade of its software. So I took the plunge and installed the open source DD-WRT on my old Linksys router. Setup is easy. The user interface is more sophisticated and informative. And there are enough knobs and dials to keep any hacker happy. It simply blows away the old software.

Now that I have saved a little power and made things a little quieter, I wonder what can I do about this noisy computer sitting on my desk? We’ll have to see about that.

My plan is to go gold — to store everything on archival gold DVD and CD media. They actually use gold in the reflective layer, which is naturally inert and resistant to the environment. The dye used to write the data is also specially formulated to be very stable and accurate. Some archival media come with a scratch-resistant coating on the surface as well. Using N.I.S.T.’s (National Institute of Standards and Technology) accelerated aging process to test these media, they are expected to last 100 years for DVDs and 300 years for CDs. Based on those tests, my idea is to convert movies to DVD and store photos and other files on CD. I bought a spindle of Vertbatim Ultralife Gold Archival discs and a solvent-free pen for labeling them. Now, if I can just keep the kids from treating them like frisbees, I’ll be golden.

I bought a XE/XM-1541 adapter on E-Bay that was originally made by VintageComputerCables.com. One end plugs into a Commodore 1541 floppy drive, the other end into the PC parallel port. Using Star Commander in DOS or OpenCBM on Linux, it’s possible to transfer a disk image (a D64 file). It also works with a 1581 drive and 3.5 floppies (a D81 file). I made disk images of all my programs. Remarkably, everything was still readable. I tried to copy some old commercial games, but the copy protection on the disks prevented it. It made sense after I thought about it. Anyway, my personal stuff was more important.

They run great under VICE, which is a very complete Commodore emulator. It even emulates the slow load times, or you can turn on warp mode when you’re feeling impatient. A friend pointed me to FC64, a Commodore 64 emulator written in Flash that can already play some programs. That’s crazy.

Even after making disk images to complete my C-64 codeography, I still can’t bring myself to pitch the stacks of old floppies. I feel like one of those people on Style Network’s Clean House who live in clutter because they can’t part with anything that carries a memory. I’ll work myself up to it eventually.

I’m slowly documenting the steps and materials to build it on my MAME cabinet page. The controls are all working, although I still need a light to make the trackball glow. The coin doors light up and accept quarters for games. (Coin entry is also simulated by holding down Player-1 and tapping Button A.) It makes me feel like a kid again.

If you’ve been wondering what I’ve been up to lately, I’m proud to have spent my time co-authoring SymmetricDS with Chris Henson and others.

It all started while implementing a commercial Point of Sale (POS) system for a large retailer in the area. The POS came with software to “trickle back” transactions to a central office for financial posting and reports. We found that the software wouldn’t handle the load generated by the 1400 stores in the chain. It also didn’t solve other problems we had, such as:

• Loading a store with an initial set of item, tax, and employee data
• Synchronizing data changes down to stores
• Conditional synchronization and filtering for subsets of data
• Routing sensitive credit card and customer data to a separate, protected database
• Supporting one database platform at the store and another at general office

Since we couldn’t find software that did everything we needed, we wrote our own. It was deployed to half the chain the first year and made the project a success. We learned a lot and gained experience with synchronization issues.

Some of the developers were unsatisfied with the implementation and wanted to modify the architecture of the program to gain improvements. The POS project had already moved on to other areas of the system, so there was not enough momentum to support such large modifications. Instead, the developers used their spare time and created an open source project to implement a new architecture from the ground up, still leveraging the knowledge and experience of the original system. We named the software “Symmetric,” which later became “SymmetricDS” (short for data synchronization) to help identify it from other uses of the term.

Meanwhile, the POS project continued to maintain the old code base for synchronizing data, but some infrequent bugs hinted at larger problems in the code, raising concerns for rolling out the rest of the chain. The issue got visibility and became part of the next release for the POS. The development group proposed bringing in SymemtricDS as a replacement, which would further refine the software and put it through rigorous quality assurance testing. After putting in place an “Open Source Policy,” the business agreed to use the software and contribute back improvements made during implementation. This move gave credibility and “polish” to the product as production-ready software.

After the retailer deployed SymmetricDS to stores in production, we announced a 1.0 release of the software publicly. It made a lot of hard work feel worthwhile and satisfying. Currently, we’re planning the next release and completing the User Guide documentation. We’re very happy and excited that we’ve delivered something unique and useful to the community, so we hope you take the time to check it out.

The factory firmware on the Sansa is pretty and worked quite well at first, but I soon experienced problems. First of all, and strangely enough, the minimum volume was uncomfortably loud. There was actually a setting to fix this. Next, advancing to the next song required hitting the button twice, sometimes. If the LCD is on, hit the button once to advance, if the LCD screen is sleeping, hit the button once to wake it, then again to advance. How annoying! The final straw came after I added more songs to the player, but was unable to find them and play them.

Installing Rockbox was much easier than I had imagined. All I had to do was download and unzip a file onto the player, then run a small installer that writes a boot loader to the player. The original Sansa firmware is still there and can be booted. But now Rockbox boots by default.

The default Rockbox is plain-looking, but it is feature-rich. I quickly found font and theme files that can be unzipped to add Rockbox eye-candy. I can’t believe how fast it boots, and I am convinced the player actually sounds better now. How long before factory players start shipping with this on it?

Back to that section of the population that responds to surveys. It turns out that it’s typically 2 percent. That statistic alone isn’t the worst problem for a marketer. The bigger problem is that the people who do fill out the survey are the kind of people who fill out surveys. In other words, it’s not a good sampling. So, marketers experimented with different incentives, like sending a cash incentive in the initial mailing, or entering respondents in a prize drawing. Of course, this affects the cost of running the survey. Experiments with cash incentives (for example, $1 and $2 vs $5 and $5 vs $10 vs $20) have been done and compared to a lottery of $250. The incentives drew more reponses, and the higher cash incentives drew more responses. But the biggest pop was from, you guessed it: a one-dollar bill.

Ah, those clever German engineers have whipped up a grand user experience. There’s no better way to keep me safe than inducing panic as I hurtle across asphalt in a metal object surrounded by other metal objects.

They could have just informed me, and let me make the decision on how to handle it. How about telling me, “Antilock brake fault, using manual brakes.” Then I might think, “I better get the ABS fixed, but until then, I’ll give myself some more stopping distance.”

So I pull my car into a local shop, and they hook it up to a fancy computer that just got several hundred dollars of new software installed. But the car won’t talk. It will only talk to an authorized VW technician. I have to schedule an appointment. So I have a car with bad brakes telling me to stop and it will only tell a dealer what is wrong. I need an open source car.

This means that the J2SE, J2ME, and J2EE source code will be released under GPL2 sometime in March 2007. It includes a special clarification, ala GNU Classpath, so you can write closed source applications that run in the VM. One immediate effect of all this is that Java can now be included in Linux distributions. In the longer term, this is going to have repercussions for .NET.

For a while, I had some interest in .NET, being no less than a ECMA/ISO standard, and even bolstered by open source hacker Miguel de Icaza (whose passion is infectious). But with recent events, the outlook for .NET has changed. Now that Novell has partnered with Microsoft and both have agreed not to sue for patent infringements, it is obvious that Mono cannot be considered safe from patent encumberment and royalties. And Sun has chosen GPL for Java’s new license, effectively granting free use of any patents to every programmer who uses it. The choice for me is very easy and obvious now. Kudos to Sun!