Tuesday, March 28, 2006
Last post, October 8th. Wow. Well here's a few updates of what's gone on with chungles the past few months:

Development has been slow. I got back in school and back to work. Every now and then I sneak away from benchmarking strange systems and helping fix mac clusters to write a few lines for Chungles. There have been some things done, just still far too much to do.

Firstly, I'm migrating from CVS to Subversion. I should've probably done this a long time ago. Subversion is insanely cool and CVS pisses me off a lot. I took down the CVS link from the project page. Everything should be working on SVN now, but if it's not, feel free to email me. I'm going to try and have everything laid out so that using subclipse you can import everything from the repository and build it easy.

Now in the repository, you might notice two new directories - stateless and jnistateless. This is what I've been coding up recently (instead of finalizing little things in the main Chungles... sorry). So one of the things I said I'd like Chungles to be is a framework. Part of the reason why is this subproject - Stateless Chungles (I haven't thought of a better name yet -- suggestions?). I was talking with a guy a while back about various projects and he talked about his primary focus of work - stateless applications. He mentioned the idea of being able to record a state and move around programs to other machines and pick up where they left off. This is a cool thing in grid computing I'd like to bring to desktops. Idealy, I want Stateless Chungles to be able to drag a program on a local machine to a remote machine; that is, freeze a program on one computer, ship it off to another, and thaw it out. I want to make Chungles the swiss army knife of networking on the desktop.

As far as what's done with Stateless so far, there's not a whole lot there. Some initial code is written and committed. The jnistateless folder contains a nifty little bridge for Chungles and native applications wanting to use Stateless. It's got a configure script and will build a jnilib and install a header file for applications wanting to implement a stateless app. The functions right now are primitive and lacking, but they're there and working. I have classes written to successfully load both native and java stateless apps. It's not finished, however, but I may have something to demo off using it soon.

The last thing is some more GCJ talk. GCC 4.1 is released and I've been working on getting a good set of cross-compilers installed on my system. I got a mingw32 and regular linux one going, but theoretically things should play well if I start producing some builds in GCJ with the same compiler for the big 3 platforms. I was reading some news about performance of GCJ versus JVMs. It's definitely something I want to have done. I think it was IBM that did a benchmark crunching prime numbers and the stats that stuck out the most to me were the time it took and the amount of memory used. Sun's JVM used something like 150mb where GCJ used 8mb. I'd really like to see how GCJ performs against python (so I can win the language battle at the office). Some have clocked Sun's new 1.6 JVM to be twice as fast or something of the like. Anyway, I'd definitely like to have nice native built versions of Chungles and expect to see some. I think a lot less people will complain about my choice in java when they need absolutely no dependencies or virtual machine to run it.
posted by Alex at 10:35 PM 0 comments



Archives:
04/06/2005   04/08/2005   04/11/2005   04/12/2005   04/17/2005   05/08/2005   08/04/2005   08/06/2005   08/11/2005   10/08/2005   03/28/2006   04/19/2006   05/02/2006   09/29/2006   10/12/2006   11/01/2006   11/03/2006   11/05/2006   11/16/2006   11/21/2006   12/01/2006   12/03/2006   12/04/2006   12/07/2006   12/09/2006   12/11/2006   12/26/2006   01/28/2007   10/27/2008   08/08/2009