MotomaSTYLE

Alice Release

Tue, 15 May 2007 00:00:00 -0400

I have made an official release of the graph traversal engine, codenamed Alice, today. Alice is an experimental graph traversal algorithm, based loosely on the Breadth-First Search, and having the properties listed in one of my previous posts. For you readers who are curious, you may pick up the path-finding library, its source code, and a small demonstration from the MotomaBASE Sourceforge Page. If anyone takes the time to try it out and play with it, please post your thought here so I can know what you think of it.

New Project: Slipstream

Tue, 24 Apr 2007 00:00:00 -0400

I have been recruited to work on a new Sourceforge.net project called Slipstream. We are still in the design and layout phase, but we have some hard-hitting experts in on the project and there are great ideas kicking around.

Character generation, and rendering engines are slated, the UI is going through some heavy construction, and I am tackling an AI path-finding engine which is approximately 35% complete.

Stay tuned for more on this project as we get our modules up in the repository.

Slipstream: Pathing

Tue, 24 Apr 2007 00:00:00 -0400

I completed the path-finding algorithm today. At its simplest level it is merely a Breadth-First Search on an undirected graph. This has a number of benefits:

The code is short: The entire traversal algorithm takes up no more than 20 lines of code. The code is fast: Due to the method in which parental assignment occurs, the traversal takes place in O(n) time. Another benefit of the BFS algorithm is that the time to find one path is based on the number of vertexes, not the complexity of the graph itself; therefore, a difficult maze will be solved in the same amount of time as a wide open field. Vertex adjacency can be arbitrarily assigned: The system has no concept of dimensional space, only vertex adjacency. Further, there are no limitations on the number of adjacent vertexes any particular vertex can have. Additionally, these vertexes could conceivably be in an entirely different section of a map (i.e. a teleporter) or in the exact same position allowing two entities to occupy the same location in space at the same time. The path-finding engine will correctly handle these situations. Mapping is not limited to a set number of dimensions, nor is it restricted to the Cartesian coordinate system: The path-finding system and the logical representation of the nodes are not reliant on each other. The path-finding algorithm works as well in a Cartesian coordinate system as it does in a Polar system. It can be as easily applied to an Isometric tile set as it can to the 4 dimensions of a hyper-cube; the concept of space, as well as its relationship to paths, are up to the programmer's design. One traversal of the graph finds the shortest path from any location in the map: My implementation of the BFS algorithm will traverse all vertexes in a map, assigning pointers to parent nodes to denote how any particular vertex was reached. At the end of a traversal, the graph is left in such a state that from any given node you can follow the parent path back through the graph and arrive at the origin. The benefit of this feature is that for any given target on any given graph, the traversal needs only to happen once. The immediate benefit of this is that if I were making a number of entities move to the same position, I would need only to give them the path map, and by looking up their position on the map, a parent traversal for each entity would land them at their destination.

With all that said, I am now in the process of rewriting the code. The major pitfall of the engine as it is right now is the lack of support for weighted graphs. To add to the robustness of the path-finding procedure, I am going to implement Dijkstra's Shortest Path Algorithm. This will allow the engine to take into account the difficulties of passing through certain terrains.

Gumstix Verdex has finally arrived!

Fri, 08 Jun 2007 00:00:00 -0400

I received my Gumstix Verdex XL6P and the matching Console-VX board from the friendly UPS man last night. After 15 minutes of set-up and tinkering, I had my PuTTY terminal greeted me with with a friendly login prompt:

Welcome to the Gumstix Linux Distribution!
 
gumstix login: root
Password:
Don't eat me!
#

So far I have had quite a bit of fun playing with it. I am using USB to power the tiny machine, and connecting to it over a USB to Serial connection cable from my laptop. Two things have caused me frustration, however: the Verdex board (according to the meager documentation I have found) does not support USB networking (a.k.a USBnet), and the USB host connector is a USB-B female connection.

Now, neither one of these would be much more than a nuisance alone, yet together they put a heavy damper on my intent for the board. USB Host is a very welcome (and in my opinion necessary) addition for a HL programmer like myself, but I have been unable to find the proper connectors at my local hardware stores. At this point I am sure that you are asking me "Why didn't you pick up the Gender Changer that they supply for just that purpose?" The thing is, I never expected that this would be an uncommon part, and for $8 I figure I would be better off picking one up at the local Radio Shack. I walked in and asked the two men at the counter for a USB-A gender changer, and the smacktards cracked up giggling. BOH! I will not be able to connect with this wirelessly until I am able to get one of these damn parts to plug my B/G adapter in to.

The lack of USBnet ability is another frustrating thing (this is probably related to the fact that the Console-VX board is a USB Host, rather than client); I wanted to be able to set up a bridged network connection and try out the portage system that the Gumstix supports. I have always been a big fan of Gentoo (link at the bottom of the menu) and I figured this would be a great little fun way to get the system set up. I had big plans for this too: create a samba share on my laptop to act as the primary storage for the Gumstix, keep X11 running on the laptop, and use X11 forwarding over the USBnet to run native Linux GUI applications from the device. This would be the ultimate merging of the Windows and Linux environment; finally I would have both the Linux and Windows interfaces running on the same data side by side.

One last note: I have been unable thus far to get Cygwin to properly build the buildroot system, leaving me feeling quite useless at the moment. I do not feel like getting a full Fedora virtual machine running in VMWare (the VMWare image is huge!) just to do a little cross compiling. I was hoping to use Microsoft's Virtual PC 2007 to boot up a LiveCD distro, download the proper maketools and the gumstix-buildroot toolkit to a tiny vitual partition, and use that system to run my cross compiling. The problem with this is that most LiveCD distros are severely lacking as far as developer tools are concerned (Backtrack2 might have gcc) and the package systems are mostly broken, as these are LiveCD distros, and not intended to be used in the perverse ways I do.

If anyone has a solution to any one of my problems, or wants an update on how my Gumstix adventure is going, send me an email or give me a post.

Gumstix == Success

Thu, 14 Jun 2007 00:00:00 -0400

This weekend I spliced together a couple of USB cables to allow me to attach my USB-A devices to the USB-B port on the Console-VX. I have successfully connected my USB thumbdrive, 60Gb iPod, and Vimicro webcam to the tiny board. I still need to compile the video drivers for the camera, but it was quite exciting to see the Verdex mount and search my iPod's 60 gig hard drive.