GNU/Linux

Until only a few years ago, I had no real experience of UNIX-based systems save for some minor University-related work. However, over the last few years I have spent some time attempting to improve my knowledge of the POSIX world as part of my ongoing efforts to remain platform-independent.

At University, my UNIX experience consisted of using SuSe as part of a Networking/Computer Architecture course and using PS2-Linux for PlayStation 2 game development. Neither were very interesting - the networking coursework was largely unrelated to my degree and was made needlessly difficult by requiring us to use the foreign UNIX OS in place of the Windows environment that we were familiar with. As for the PlayStation 2 development, our coursework was to create a Pong clone using a horrible mess of UNIX and PS2 library code under a KDE that insisted on crashing at least once a day. This course module required me to sign a non-disclosure agreement, possibly in order to prevent me from informing others of how terrible the Game Console Development course was. All in all, these experiences did not do anything to improve my opinion of UNIX versus Windows NT.

Outside of University, the only UNIX-related experience I had consisted of about 10 minutes with an Ubuntu LiveCD and whatever time I could spare to experiment with my new Apple Mac Mini and it's PowerPC processor. However, during my industry placement year (2007-2008) away from the University, I decided to experiment with UNIX some more and installed Xubuntu on to a spare hard disk drive. Forcing myself to use it for a few weeks wasn't as arduous as it might have been, due to the availability of some of my favourite applications from Windows, such as OpenOffice.org, Firefox, Thunderbird, VLC and Code::Blocks.

However, I encountered more bugs and general inconveniences under UNIX than I have ever had to deal with under a standard, out-of-the-box Windows XP installation. The X Window system never seems to work correctly and installation of graphics drivers is an annoying and error-prone experience to say the least. Also, VLC refused to run after Code::Blocks was installed due to incompatibilities with the newer wxWidgets library. So-called "DLL hell" might have been an issue under the DOS-based Windows 9x design but this problem is not present under the new (and largely unrelated) NT architecture. It is disappointing to see that it is still an issue under the allegedly superior UNIX design.

During this time, I played around with a few other operating systems, including the OpenSolaris LiveCD. Those of you who have an interest in alternative UNIX systems aside from GNU/Linux and BSD may wish to download this LiveCD for yourselves.

Shortly before I completed my year working as a programmer and general IT technician, the replacement student, David, arrived. David seems to be an excellent fellow, with a much better knowledge of UNIX than myself. During our month working together, he suggested that we completely reinstall our dying Red Hat server with Debian GNU/Linux.

The end result was not one but three Debian-powered machines:

• A "test run" server, which had previously been retired from service and put into storage.
• The "real" server, which ended up being used in tandem with the "test run" server.
• A desktop machine, built by myself from spare components.

The desktop machine became my primary workstation, as David was now using the machine that had previously belonged to me. It was on this that I constructed a GUI front-end for the AMANDA backup program using Code::Blocks and the GTK+ library. I still have this on a CD, licensed under the GPL v3 but as I lack access to any sort of tape backup hardware I cannot continue to work on it.

GTK+ was interesting to work with but has some major flaws:

• It's much more limited than the Win32 API.
• It was annoyingly difficult to configure projects so that they could link and build correctly.
• GTK+ is much less efficient than pure Win32 code under Windows.
• It seems to be riddled with bugs, although this might have been caused by Debian 4.0 using an older version of the GTK+ development libraries for it's repositories. I'm fairly certain that the code I wrote was correct and bug-free and yet my application seemed to either crash or output garbage far more often than it should have.
  

Moving on to the present day, my room currently has five computers in it:

• My main tower, which is currently running Windows Server 2008.
  
• My old laptop, running Windows XP Home SP3.
  
• My Mac Mini running MacOSX 10.3.9 (currently being stored on a shelf).
  
• A spare tower, currently in a state of disassembly (previously ran Windows Server 2008).
  
• My UNIX tower, currently running gNewSense 2.2.
  

The gNewSense tower runs quite happily with a 450MHz Pentium III processor and 256MB of RAM using the integrated Intel graphics chip. I have unearthed the copy of Introduction to Unix that my University gave to me as part of the Network course, installed Code::Blocks and managed to rig up a simple OpenGL/SDL program that renders an animated .MD3 mesh at an acceptably smooth frame-rate. The same Code::Blocks project file recompiles on my main Windows tower with no alterations save for using a different build target.

I've not been without problems, however - Debian 5.0.1, which I attempted to use before trying gNewSense, didn't seem to like my graphics chip and X.org refused to run. The gNewSense GNOME GUI runs fine, but there seems to be a fault with the hardware OpenGL acceleration that causes corruption around the mouse cursor. This is not usually a problem for me, since SDL windows do not draw the standard X window system cursor and so corruption does not occur.

Yesterday, the X window system crashed when I attempted to create a simple OpenGL/SDL window. The CTRL-ALT-BACKSPACE combination that I have become so familiar with had no effect and attempting to CTRL-ALT-F# to a terminal display was equally useless. After forcing the machine to power off, I restarted it, logged in and ran the program again. It worked flawlessly. What, then, was the cause of the X server crash?

Lastly, my router loses it's connection to the Internet whenever the gNewSense tower is online. This might be a problem with the router, a problem with the network card or a problem with the driver or version of the network card driver or Linux kernel used by gNewSense but for now I have simply disconnected the gNewSense tower from the network.

An overdue update.

It's been a while since I posted something here, so here's an update on my progress.

Firstly, the bad news: The SURGE game engine project has now been cancelled. It's a pity, because it got quite far, as can be seen from this features list:

• Milkshape 3D .MS3D support for ragdoll actors. Animation data was ignored. Automatic texture loading was implemented.
• Quake III .MD3 support for non-ragdoll actors. Automatic texture loading was implemented.
• Incomplete Quake III .BSP support for the game world. Bezier surfaces were neither drawn nor added to the world collision data (though Polygon and Mesh surfaces were) and the actual BSP visibility culling data was ignored by the renderer. "Flare" (or "Billboard") surfaces were not drawn either. Automatic texture loading was implemented.
• Simple .TGA support (only uncompressed 24- or 32-bit .TGA files were allowed).
• A basic actor system that could handle either skeletal or morph actors transparently.
• Realistic physics via ODE.
• Cross platform. Uses only APIs that are available on several platforms (OpenGL, SDL, ODE, etc.).
• Optional multithreading (to speed up physics calculations on multi-core machines).
• Some basic audio.
  

There are a few reasons why I stopped working on this project but one of the main reasons was that ODE 0.8 and 0.9 seem to have a bug that causes the physics system to crash under certain scenarios. This bug seems to be quite rare with any sort of collision other than trimesh-trimesh. However, it happens almost instantly whenever two trimeshes touch each other - for example, when a ragdoll touches another ragdoll, or the world (I represented the world BSP data as an ODE trimesh). As my engine was designed on the basis that ODE would always be the physics system used, switching to another physics engine would require me to essentially remove about 50% of the code from the engine.

Now that I come to look back on the project a few months later, the coding wasn't very good. Graphics, logic and physics code were all intermixed and not seperated out very cleanly. The fact that ODE was so integral to the engine working correctly is testament to this.

However, after talking a break for a few months, I'm now working on a completely new project that will hopefully be much better, codenamed the "Tidal" project. The name was chosen because it continues a sort of theme - Edd has been working on the Drift engine and my previous engine was called the Surge engine, words which I think have a sort of oceanic quality to them (to be set adrift, the surge of the tides, etc.), so I decided to continue my naming along these lines.

Tidal is currently almost at the same point that Surge was before I abandoned it:

• Extremely modular design. This is to avoid the problems I had with Surge being too reliant on ODE. The motto for Tidal is "Don't rely on the API." It also makes the code cleaner.
• Cross platform. The engine's .h files don't need any specific APIs (unlike Surge), meaning that their partner .c files can implement them however they want. For example, graphics.c could use OpenGL, DirectX or even software rendering to implement the core rendering functions specified by graphics.h.
  
• Written entirely in strict ISO-C90 compliant code. There's no real reason for this aside from I feel that it'll make development more interesting.
  
• Milkshape 3D .MS3D support (although it's not actually used for anything yet). Most of the code for this was taken from Surge.
• Quake II .MD2 and Quake III .MD3 support. Automatic texture loading has been implemented in a much better way. MD2 files have a small issue with some texture coordinates being wrong due to the way I've loaded the MD2 data into structs designed to hold MD3 models but this is not really a problem right now, especially since MD3s should be the main "morph" mesh format used anyway. Most of the MD3 code was taken from Surge.
• Incomplete Quake III .BSP support for the game world. Most of the code for this was taken from Surge.
• Simple .TGA support (only uncompressed 24- or 32-bit .TGA files are allowed). Most of the code for this was taken from Surge.
• Realistic physics.
• New ActorClass-based actor system. Complete, aside from scripting (which I'm working on now). Currently, ragdoll actors are not supported.
• Very incomplete features include: Shaders, audio, networking and a GUI system.
  

Once the scripting system is finished, it should be possible to make a simple game. I'll be adding support for more file formats as the project goes on.

Progress on ragdolls.

Introduction.

I've made some good progress with my MilkShape-ODE Ragdoll project since I last posted here. Currently, the project can load a MilkShape3D model, find and load any associated textures and create a set of linked joints and bodies to form a skeleton. When the model is rendered, the positions of it's vertices are transformed based on the skeletal data.

Terminology.

ODE Bodies are simply points in space with a "mass" value. They can be joined together with other bodies using "ODE Joints". To illustrate how this works, imagine that your upper arm is one "body" and your lower arm is another, with your elbow being a "joint".

Bodies cannot collide with each other, as they are used to simulate dynamics, not collision. However, they can be paired with an "ODE geom". Geoms have no dynamics data (such as mass or inertia) but do have collision data. So if you were to create a bowling ball in ODE, the ball's body would be what gravity pulled down on - but the ball's geom would be what prevented gravity from pulling it through the floor.

Please note that ODE joints are not the same thing as MS3D joints. In ODE, vertices are associated with ODE bodies. Two ODE bodies can be linked together with an ODE joint. In MilkShape3D, there is no such thing as a body - every vertex in a model is linked to an MS3D joint. An MS3D joint may have a parent joint or may be independant. When an MS3D joint is moved or rotated, all vertices associated with that MS3D joint move with it (see my last post for more details on how MilkShape3D joints work). So, MS3D joints are roughly a combination of both ODE joints and ODE bodies.

How the skeletal data is generated.

The first step is to generate a list of ODE bodies. This is done fairly simply: the number of ODE bodies created for a model is equal to the number of MS3D joints specified in the mesh file. ODE bodies are positioned by adding the position vector of every associated vertex together and then dividing the result by the total number of vertices used, so they are generally positioned around the center of a "vertex cloud". Any ODE bodies which do not have any associated vertices are simply positioned at [0, 0, 0], to avoid a divide-by-zero error.

The next step is to join these bodies together using ODE joints. This is not quite as simple as generating ODE bodies. There are two reasons for this:

• MS3D joints are joined to other MS3D joints, while ODE joints are connected to ODE bodies.
• MS3D joints do not need to have a parent, whereas if an ODE joint is not attached to anything then it will join itself to the environment. This has the same effect as nailing something to a wall, because the environment is the world - meaning it doesn't move.

The answer here is to create an ODE joint for every MS3D joint that has a parent, join it to the ODE body with the same index value as that MS3D joint (so if you were working with MS3D joint #3, you would join the current ODE joint to ODE body #3) and that MS3D joint's parent MS3D joint's ODE body. This is not as complex as it sounds - it only becomes difficult because there are less ODE joints than there are MS3D joints.

Current work.

There are still a few things missing from the simulation yet. Firstly, the ODE bodies do not have any mass values set. I think that the best solution would be to assume a uniform density value for every body in the mesh (for example, 0.4) and then approximate the total area covered by the vertices of an ODE body using a cuboid or capped-cylinder shape.

Secondly, no ODE geoms are generated yet and so there is no collision. Originally, I was planning on approximating ODE collision geoms in the same way as I would approximate mass but, while I can get away with comparatively slight inaccuracies in mass approximation, collision errors are more "visible" to the end user. My current thoughts are to generate a separate trimesh for each ODE body in the mesh. This would be very accurate, but it could be computationally expensive.

Possible collision problems.

Neither of the above approaches really solve the problem of triangles which span vertices which are associated with different bodies. While most triangles in a mesh belong to a single ODE body, triangles which are used to join the vertices of two ODE bodies together (such as "elbow triangles" which stretch from the upper arm body to the lower arm body) will not have any collision data.

Possible solutions are to create an ODE sphere geom for each ODE joint or to generate a trimesh for the entire model every time it changes. The first approach could lead to redundant ODE geoms being created for joints in locations such as at the top of a string, several feet away from a "puppet" model below and would also require me to find a way to guess what the radius of the sphere should be. The second approach could have issues if it interferes with the "temporal coherence" data that ODE uses for trimesh collision.