Transparent Development, Torque 2, and You!
by Stephen Zepp · 10/18/2007 (5:26 am) · 122 comments
There has been some buzz floating around since IGC this year, and a lot of it has been around InstantAction, but we also announced a couple of additional topics that are starting to sink in, so I wanted to dive in to the deep end of the community pool and make some waves....
Transparent Development
Sounds nifty, but what is it? One analogy I've heard that is a pretty solid description is consider TD a window into the offices of the development staff working on Torque 2--see what's being worked on this week, read discussions about designs as they are being discussed internally, and most importantly be able to discuss the designs and implementations themselves, as they occur.
We're not talking marketing information for a product that has already been developed--we're talking watching, and even being able to discuss and provide feedback on the development of Torque 2 as it occurs. I'll get to Torque 2 in a moment, but I want to reinforce that this is a fundamental shift back to the way GarageGames used to interact with the community--away from keeping completely silent regarding what we're working on and why, and back to giving the community rapid and fully packed pages of information on the development as it occurs--not after the fact, and especially not after it's released. Torque 2 isn't even scheduled for beta until well in to 2008, but we're working as rapidly as we can to get infrastructure in place to set up discussions as the engine team works on development.
We're talking roadmaps, milestone lists, design write ups, expected work flow and implementation standards, the works--all up front, all the time. One of the projects that I am working on right now is designing a new set of forums, wiki, and related web tools to manage extensive and extended data flow between the Transparent Development Community (that's you guys) and the engine development team (that's us!). Once this infrastructure is in place, we're going to open it up for communication purposes to a limited set of users for beta testing for a few weeks, and then we will make the information available to all.
[edit: see below in the comments section for further clarification on content and timing]
It's going to be several weeks yet before this infrastructure is ready, and it will be a few weeks of beta testing of that infrastructure as well before we're ready to turn on the fire hose, so please be patient while we get things set up for this to work in the long term--when it's ready, we'll let everyone know and let the show begin!
I do want to make something fundamentally clear from the beginning (and trust me, I'll keep pressing this point until it sinks in fully!): Torque 2 development will be an iterative process, very similar to the SCRUM, or Agile/Spiral Development models--and one of the key aspects of this style of development is that sometimes you go down a path, and it turns out that the path is not the best solution. When this occurs, the best course of action is simply to cut out the results of that development spiral, and reset your project expectations. Sometimes this means removing features, adjusting the milestone list, or delaying (or in some cases even canceling) entire systems.
It is critical that the entire Transparent Development Community understand that this development process you will be watching and discussing is a fluid one--and it should be fully expected that features and systems being discussed one week may very well end up being cut forever the next. No projects, games, and especially no funds or contracts should ever be committed during this development process--if you need to make commitments or financial decisions regarding your projects, then the Torque 2 Transparent Development Community is absolutely not somewhere you should be.
Torque 2
So what's Torque 2? To put it simply, Torque 2 is a ground up re-factor of the entire architecture of all of our engines into a new engine with some pretty amazing architecture and features. We've learned quite a bit over the last several years of engine development, and the time has come to take those lessons and build a new engine that not only provides the maximum flexibility possible to game developers, but fundamentally change many of the designs of the engine itself--correcting design assumptions and work flows that came about over the years that we ourselves find too restrictive, and along the way providing customization and feature enhancement capabilities that were never available before.
Torque 2's fundamental changes fall under three primary areas:
Modularity of Systems: In the past, it was very difficult to leverage into Torque systems, libraries, and raw code written by 3rd parties, such as ODE/Havoc/PhysX Physics, alternate scripting languages such as Lua, Python, and even .NET languages, and many other cool things that are available to game developers. Torque is extremely integrated, with each of the systems tightly coupled to provide a powerful, but ultimately restrictive engine when it comes to "mashing up" other solutions into a coherent whole--and Torque 2 is designed from scratch to correct those issues by allowing game developers to replace major system modules in a "plug and play" style--as long as the external systems match interfaces that the Torque 2 framework expects, you will be able to use just about any 3rd party (or self-developed) system within the core Torque 2 framework.
Abstraction and Interfaces: One of the key ways we will be able to provide this type of "swap out" modularity is by abstracting most of the core systems via the use of interfaces. Each major system will describe a set of interfaces that other systems and objects will use to communicate with the system, so instead of calling code directly within a system, external modules and objects will interact with the interfaces that system exposes--allowing other systems to replace core modules at the whim of the developer.
Torque 2 has already leveraged the previously developed GFX abstraction layer present in TGE-A, and in addition has had several iterations of abstraction of core systems as well, such as the sound layer (now abstracted through a new SFX system), lighting (fully abstracted into a separate module), collision and physics (currently under iterative refactoring towards a comprehensive interface), and many others. As the iterative development spirals of Torque 2 continue, additional systems will be abstracted and interfaces implemented for even more comprehensive modularity.
Components: The concept of a component based architecture is actually very simple--but it has some far reaching design implications, and in many cases can take some time to adapt towards. In most game engines, including Torque, you have a very deep set of classes that form a pretty static ancestry, and you use inheritance of all of these classes in your game object level classes. Unfortunately, this has several issues that become more apparent the longer you use these classes for development, especially across multiple games: your game object classes start becoming more and more bloated, and less appropriate for specific uses. Anyone that takes a good look at the Torque ShapeBase class will soon see what I mean--there are still left over methods in the ShapeBase class that are part of original Tribes code, and aren't particularly useful for most games.
Here's an example of exactly what I mean: Back when BraveTree was working on their environment pack, one of the neat things they wanted to have was butterflies with random movement about the scene. Pretty nifty stuff--they derived a new class off of ShapeBase, and their butterflies merrily wandered about the stronghold mission randomly--until one of them happened to run in to a crossbow object placed in the scene.
Suddenly, this tiny little butterfly picks up this huge crossbow, and starts swinging it back and forth madly as it continues to wander about--merrily meandering about the scene with a totally unexpected weapon of mass destruction dangling from its cute little mandibles.
It was immediately obvious what was going on, and as soon as they picked themselves up from off the floor and stopped laughing, they realized that since the butterfly inherited from ShapeBase, it also inherited the ability to collide with Items, and mount weapons--and that's exactly what it did!
So what does this have to do with components? Well, with the idea of components (based on the Object Oriented Programming concept of aggregation, sometimes called composition), instead of inheriting a series of ever more complex functionalities that your particular object may not need, you create very small units of functionality that accomplish simple tasks--and "build" your game objects from these little components. In the case of the butterfly, they might have added a render3D component, a collision3D component, a hoverFlightMovement component, and a AIRandomMove3D component--and thereby had all the functionality they needed for the butterfly, without any of the bloat that inheriting from a full class would have given them.
We'll be covering components in quite a bit more depth once the Transparent Development Community is in place, but if you are interested in learning more about how component based game development works, TorqueX also uses components, and is a great place to learn how the new design concepts can be used.
What this means for You
Well, the very first thing it means is not a lot to be honest--this is an announcement of a new development style (transparency), and a new engine that is just now entering heavy development. Torque 2 is not going to be ready for use in a major (or even minor) game project until sometime in 2008 (more accurate time line will come when we get more iteration cycles finished up). If you are currently working on a game with Torque (or another engine), or are planning on starting a new project in the next 6 to 9 months, Torque 2 is not for you at this time.
If you are someone that is interested in how an engine is developed, and want to participate in discussions of that development process as it occurs, stay tuned for further announcements about the Transparent Development Community. Once it opens to the public stop on by and hang out--at the least it should be educational and informative, and hopefully some of you might even generate ideas, designs, and implementations that may very well be factored in to the future of Torque.
I'm already well over my word count for a .blog, so I'll answer specific questions in the comment area below.
Transparent Development
Sounds nifty, but what is it? One analogy I've heard that is a pretty solid description is consider TD a window into the offices of the development staff working on Torque 2--see what's being worked on this week, read discussions about designs as they are being discussed internally, and most importantly be able to discuss the designs and implementations themselves, as they occur.
We're not talking marketing information for a product that has already been developed--we're talking watching, and even being able to discuss and provide feedback on the development of Torque 2 as it occurs. I'll get to Torque 2 in a moment, but I want to reinforce that this is a fundamental shift back to the way GarageGames used to interact with the community--away from keeping completely silent regarding what we're working on and why, and back to giving the community rapid and fully packed pages of information on the development as it occurs--not after the fact, and especially not after it's released. Torque 2 isn't even scheduled for beta until well in to 2008, but we're working as rapidly as we can to get infrastructure in place to set up discussions as the engine team works on development.
We're talking roadmaps, milestone lists, design write ups, expected work flow and implementation standards, the works--all up front, all the time. One of the projects that I am working on right now is designing a new set of forums, wiki, and related web tools to manage extensive and extended data flow between the Transparent Development Community (that's you guys) and the engine development team (that's us!). Once this infrastructure is in place, we're going to open it up for communication purposes to a limited set of users for beta testing for a few weeks, and then we will make the information available to all.
[edit: see below in the comments section for further clarification on content and timing]
It's going to be several weeks yet before this infrastructure is ready, and it will be a few weeks of beta testing of that infrastructure as well before we're ready to turn on the fire hose, so please be patient while we get things set up for this to work in the long term--when it's ready, we'll let everyone know and let the show begin!
I do want to make something fundamentally clear from the beginning (and trust me, I'll keep pressing this point until it sinks in fully!): Torque 2 development will be an iterative process, very similar to the SCRUM, or Agile/Spiral Development models--and one of the key aspects of this style of development is that sometimes you go down a path, and it turns out that the path is not the best solution. When this occurs, the best course of action is simply to cut out the results of that development spiral, and reset your project expectations. Sometimes this means removing features, adjusting the milestone list, or delaying (or in some cases even canceling) entire systems.It is critical that the entire Transparent Development Community understand that this development process you will be watching and discussing is a fluid one--and it should be fully expected that features and systems being discussed one week may very well end up being cut forever the next. No projects, games, and especially no funds or contracts should ever be committed during this development process--if you need to make commitments or financial decisions regarding your projects, then the Torque 2 Transparent Development Community is absolutely not somewhere you should be.
Torque 2
So what's Torque 2? To put it simply, Torque 2 is a ground up re-factor of the entire architecture of all of our engines into a new engine with some pretty amazing architecture and features. We've learned quite a bit over the last several years of engine development, and the time has come to take those lessons and build a new engine that not only provides the maximum flexibility possible to game developers, but fundamentally change many of the designs of the engine itself--correcting design assumptions and work flows that came about over the years that we ourselves find too restrictive, and along the way providing customization and feature enhancement capabilities that were never available before.
Torque 2's fundamental changes fall under three primary areas:Modularity of Systems: In the past, it was very difficult to leverage into Torque systems, libraries, and raw code written by 3rd parties, such as ODE/Havoc/PhysX Physics, alternate scripting languages such as Lua, Python, and even .NET languages, and many other cool things that are available to game developers. Torque is extremely integrated, with each of the systems tightly coupled to provide a powerful, but ultimately restrictive engine when it comes to "mashing up" other solutions into a coherent whole--and Torque 2 is designed from scratch to correct those issues by allowing game developers to replace major system modules in a "plug and play" style--as long as the external systems match interfaces that the Torque 2 framework expects, you will be able to use just about any 3rd party (or self-developed) system within the core Torque 2 framework.
Abstraction and Interfaces: One of the key ways we will be able to provide this type of "swap out" modularity is by abstracting most of the core systems via the use of interfaces. Each major system will describe a set of interfaces that other systems and objects will use to communicate with the system, so instead of calling code directly within a system, external modules and objects will interact with the interfaces that system exposes--allowing other systems to replace core modules at the whim of the developer.
Torque 2 has already leveraged the previously developed GFX abstraction layer present in TGE-A, and in addition has had several iterations of abstraction of core systems as well, such as the sound layer (now abstracted through a new SFX system), lighting (fully abstracted into a separate module), collision and physics (currently under iterative refactoring towards a comprehensive interface), and many others. As the iterative development spirals of Torque 2 continue, additional systems will be abstracted and interfaces implemented for even more comprehensive modularity.
Components: The concept of a component based architecture is actually very simple--but it has some far reaching design implications, and in many cases can take some time to adapt towards. In most game engines, including Torque, you have a very deep set of classes that form a pretty static ancestry, and you use inheritance of all of these classes in your game object level classes. Unfortunately, this has several issues that become more apparent the longer you use these classes for development, especially across multiple games: your game object classes start becoming more and more bloated, and less appropriate for specific uses. Anyone that takes a good look at the Torque ShapeBase class will soon see what I mean--there are still left over methods in the ShapeBase class that are part of original Tribes code, and aren't particularly useful for most games.Here's an example of exactly what I mean: Back when BraveTree was working on their environment pack, one of the neat things they wanted to have was butterflies with random movement about the scene. Pretty nifty stuff--they derived a new class off of ShapeBase, and their butterflies merrily wandered about the stronghold mission randomly--until one of them happened to run in to a crossbow object placed in the scene.Suddenly, this tiny little butterfly picks up this huge crossbow, and starts swinging it back and forth madly as it continues to wander about--merrily meandering about the scene with a totally unexpected weapon of mass destruction dangling from its cute little mandibles.It was immediately obvious what was going on, and as soon as they picked themselves up from off the floor and stopped laughing, they realized that since the butterfly inherited from ShapeBase, it also inherited the ability to collide with Items, and mount weapons--and that's exactly what it did!
So what does this have to do with components? Well, with the idea of components (based on the Object Oriented Programming concept of aggregation, sometimes called composition), instead of inheriting a series of ever more complex functionalities that your particular object may not need, you create very small units of functionality that accomplish simple tasks--and "build" your game objects from these little components. In the case of the butterfly, they might have added a render3D component, a collision3D component, a hoverFlightMovement component, and a AIRandomMove3D component--and thereby had all the functionality they needed for the butterfly, without any of the bloat that inheriting from a full class would have given them.
We'll be covering components in quite a bit more depth once the Transparent Development Community is in place, but if you are interested in learning more about how component based game development works, TorqueX also uses components, and is a great place to learn how the new design concepts can be used.
What this means for You
Well, the very first thing it means is not a lot to be honest--this is an announcement of a new development style (transparency), and a new engine that is just now entering heavy development. Torque 2 is not going to be ready for use in a major (or even minor) game project until sometime in 2008 (more accurate time line will come when we get more iteration cycles finished up). If you are currently working on a game with Torque (or another engine), or are planning on starting a new project in the next 6 to 9 months, Torque 2 is not for you at this time.
If you are someone that is interested in how an engine is developed, and want to participate in discussions of that development process as it occurs, stay tuned for further announcements about the Transparent Development Community. Once it opens to the public stop on by and hang out--at the least it should be educational and informative, and hopefully some of you might even generate ideas, designs, and implementations that may very well be factored in to the future of Torque.
I'm already well over my word count for a .blog, so I'll answer specific questions in the comment area below.
#122
I have recently put off working with Torque hoping to one day come back and find that better efforts have been made to make it easer to work with Torque (documentation, tutorials). Currently I feel like it's a mission looking for the bits and peaces in order to get my project going. Though, I don't expect anybody to make my game for me. It really does feel like sink or swim. I use TGB. It is great news to hear that GarageGames is taking steps to better the Docs and releasing more tutorials. As long as the community has these resources I don't see why T2 wouldn't crush the competition. And I'm excited about the polysoup collision. If I'm not mistaken, that means we can simply create our worlds / levels in say 3d max without having to put up with the restrictiveness of BSP?
12/04/2007 (11:35 pm)
Quote:With the resources we now have available after the IAC partnership, we have stood up an entire team who's sole focus is providing a cohesive product to our customers--tools, documentation, tutorials, examples, demos, and related materials to make working with Torque 2 a well thought out process instead of a "jump in and sink or swim" type of experience.
I have recently put off working with Torque hoping to one day come back and find that better efforts have been made to make it easer to work with Torque (documentation, tutorials). Currently I feel like it's a mission looking for the bits and peaces in order to get my project going. Though, I don't expect anybody to make my game for me. It really does feel like sink or swim. I use TGB. It is great news to hear that GarageGames is taking steps to better the Docs and releasing more tutorials. As long as the community has these resources I don't see why T2 wouldn't crush the competition. And I'm excited about the polysoup collision. If I'm not mistaken, that means we can simply create our worlds / levels in say 3d max without having to put up with the restrictiveness of BSP?
Jonathan
where can I fine T2 Features page?