See http://dataturbine.org/content/new-dell-1435-server-calit2 for cameraphone snaps, server racked and plugged. Next step is OS reinstall.

*Fix for kernel source found on http://www.linux-mag.com/id/4829

 aptitude install build-essential linux-headers-`uname -r`

So now I have VMWare tools...much better mouse behavior!

Larry Freudinger made a comment about virtualization that gave me the idea of a 'canned' VMWare image for DataTurbine. You can already download zillions of prebuilt images from VMware; why not one for DataTurbine? (The images are used via the free VMWare Player program.)

The image could contain
*Linux operating system
*Java, with Ant and subversion (See this page for instructions)

Looking at Larry's code example, GPSSource, it looks like you just have to add these to your source to get the KML plugin working:

protected static String[] coordinateChannelNames =
		{"Alt", "Lat", "Lon", "TrackID", "Type", "Classification", "Speed", "Heading"};

Not too bad, think I'll try and add this to SMS-RBNB as a simplest-codebase exercise.

This morning I forced myself to sit down and grind out a brochure that we had just realized that we had to have. Before tomorrow 3PM local, no less. I used Pages to generate a simple tri-fold brochure, with lots of pictures, and it came out pretty well. Its on its way to the NSF and elsewhere, so hopefully the errors are tolerable!

I also linked it from http://dataturbine.org/content/documentation the documentation page.

Direct URL is http://dataturbine.org/files/dt%20brochure.pdf

That took a lot of work! DataTurbine source code is now hosted on Google Code with their free project hosting. Project page is at http://code.google.com/p/dataturbine

I was able to svnsync the repositories after NEESit staff upgraded the neesforge svn server from 1.1 to 1.4 - thanks!

This gets us quite a few things - number one is Google's unmatched infrastructure. We've had uptime issues with neesforge and this should solve that.

We also get a wiki, bug tracker, much improved interface, and the added visibility. Win all around.

So I'm thinking about how to add GPS positioning to my hacked-up Apple accelerometer code, and vaguely remember gpsd from freshmeat. (Aside: Surf the RSS feeds from freshmeat now and then; there's always something interesting to see.)

Pablo Bryant asked about mesh networks for dataloggers,

I've just added to Subversion a couple of new matlab routines to read and write (sink and source, respectively) JPEGs to and from DataTurbine. The code is known-buggy work-in-progress, but I wanted to commit so that others can try it out.

NI has a protocol and code for efficient messaging over TCP/IP. The idea is to send a dictionary when you first connect, then binary messages subsequently for speed. Seems quite decent. I'm trying to decide if we should rework our NI DAQ protocol, and STM is a candidate. Since its an open protocol, we could use this for other interfaces too.
Current issues: Need a complete protocol spec so we can determine issues with other languages. In other words, it's easy in LabVIEW, but how much is a Java protocol engin?
STM intro: http://zone.ni.com/devzone/cda/tut/p/id/4095