MathDrag'n

I don't have a lot of time right now to make a big, detailed announcement, but I have to say something.  I have released a new version of MathDrag'n.  It is available for download now on the sourceforge page.  The first thing to notice is that it now has a windows installer which includes an embedded JRE, which should make handling the system on windows straightforward and require no additional downloads to get the program running.  It is also now a executable file on windows, rather than a .jar file, which should make things simpler.  You will still need to download Maxima (or buy Mathematica) seperately in order to get full functionality.

 
The next biggest change people should notice is that it solves the problems MathDrag'n was having communicating with Maxima.  I have heard the complaints, and I've done my best to solve them.  It *should*  work with all systems now.  *Please* tell me if you are having trouble.

 
There are several other minor changes, improvements, and bug fixes that people will appreciate, but none as serious as the Maxima problems it was having before.

I've been gone for a long time.  In fact, I had rather lost interest in MathDrag'n.  My research was pulling me in other directions, and it seemed like nobody was interested.  Several things changed this.

 First was a friend of mine here at school who came back to visit and give a presentation.  He asked me how things were going, what I was getting done, etc.  I told a little about scattering signals off of chaotic cavities, which is what I'm actually paid to work on.  He said---No, no!  What's going on with that math thing you were working on?

Then I get a couple of emails, completely out of the blue, asking about MathDrag'n.  One was making a good suggestion about how to do MathDrag'n, which I unfortunately had to turn down, but which I may use in a slightly different form.  The other was just asking about how to get MathDrag'n working with Maxima, a perfectly fair question, and one which helped me debug MathDrag'n significantly.  (I think the new version should work with Maxima just fine now.  I will be reading the comments regularly now, so if you ask for help, I will help you.)

 Well, all of this interest, coming up out of (for me) absolutely nowhere got me excited again.  The other students I work with who I've shown it to also really, really liked it.  I've decided that I'm going to make some progress on MathDrag'n every week, no matter how little.  Sometimes this will have to give while I'm working on my thesis research, but just a small bug fix may be all that's needed.

 So I've made several changes.  I can't make any promises about the timeline, unfortunately.  I want to make a release version of MathDrag'n which I can put up on sourceforge for general use.  The biggest differences will be:  the much improved connections with Maxima (now much more flexible and likely to work on other systems and with different versions), the addition of derivatives and integrals, and a boat-load of minor bug-fixes.  Oh yeah.   And a windows installer and uninstaller, with a bundled JRE.  If you don't know what that is, don't worry---the installer takes care of it.

I just tested MathDrag'n on a Linux box with Mathematica installed, and it worked absolutely correctly first try.  The tester found it quite usable, and checked all sorts of things that I've done.  I must say, I was simply floored!  With all the trouble I'd gone to with the OSX, I was certain I'd have all sorts of problems with Linux, but noooooo---pure perfection.  How it did it, I'll never know, but it seems to me that my choice of Java as a platform is beginning to pay off in terms of WORE (Write Once, Run Everywhere).

Woo-hoo!  (Slaps self on back, high-fives internet in general!)  Well, I think I've officially got it working.  I finally ran MathDrag'n on a Mac and got it to automatically connect with Mathematica!  There were exactly two typos in the entire program that were causing the connecting code to fail catastrophically, both of them a single wrong character in just the wrong spot.  Amazing, really, how stupid computers are after all the work we've put into them.  All the behavior seemed to be exactly what I had come to know and love from my PC testing, so it seems likely to have worked well.  (Ironically, I think I discovered a bug in the more general Java code that I may need to track down, all while having debugged it on a Mac.  I'm going to do that at home  until I have CVS or something similar set up.)

It also reminded me just how much difference a good debugger makes.  It took an estimated five minutes to track them down with the debugger.  The basic logic was rock-solid; I had really done my research before-hand, and it was time well-spent, but I still couldn't follow the code with my eye well enough to see the tiny little mistakes.  Ah well, it's done now.  YES!

I have begun working on connecting MathDrag'n to Maxima.  The basics seem to work, although there are still some translation errors that take place, especially in regards to the names of system constants.  The derivative system was also designed with Mathematica in mind (although I like how I have partial derivatives on the functions) and so the two can't quite speak to each other yet.  I also haven't checked this in Unix or OSX, and the user interface for changing the kernel hasn't been implemented yet.  The basics, however, are there.  I won't release it until both sides can handle everything they can throw at each other, and most importantly, users can always connect to the system.  On the other hand, this is very exciting---finally everyone who gets MathDrag'n will be able to use all its features without owning an expensive external product.