Friday, October 10, 2014

Building a Data-Oriented Entity System (Part 4: Entity Resources)

In the last post, I talked about the design of the TransformComponent. Today we will look at how we can store entities as resources.

Dynamic and Static Data

I’m a huge fan of compiling resources into big blobs of binary data that can be read directly into memory and used as-is without any need for “deserialization” or “reference patching”.

This requires two things:

  • First, the data must be swapped to the right endianness when the data is generated.

  • Second, internal references in the resource must use offsets instead of pointers, since we don’t know where the loaded resource will end up in memory and we want to avoid pointer patching.

Initially, this approach can seem a bit complicated. But it is actually a lot simpler than messing around with deserialization and reference patching.

Note though, that this approach only works for static (read-only) data, such as meshes, textures, etc. If data needs to change for each instance of a resource, we must store it somewhere else. If an instance needs to change color, we can’t store that color value in a memory area that is shared with other instances.

So what typically happens is that we split out the dynamic data as “instance data” and let that data refer to the static resource data. Many instances can make use of the same resource data, thus saving memory:

---------------                -----------------
|Instance of A| --------+----> |A resource data|
---------------         |      -----------------
                        |
---------------         |
|Instance of A| --------+
---------------

---------------                -----------------
|Instance of B| --------+----> |B resource data|
---------------         |      -----------------
                        |
---------------         |
|Instance of B|---------+
---------------

We typically hard-code what goes into the instance. For example, we know that the color is something that we want to modify so we add it to the instance data. The vertex and index buffers cannot be changed and thus go into the static data. When we create the instance we initialize the instance data with “default” values from an instance template in the resource data.

You could imagine doing this another way. Instead of hard-coding the data that can be modified per instance, you could say that everything can be modified per instance. In the instance data you would then use a flexible key-value store to store the delta between the instance and the resource data.

This is more flexible than hard-coding, because it allows you to override everything per instance, even texture or vertex data if you want that. It can also save memory, because the instance data will only contain the things you actually override, not everything that potentially could be overridden. So if you have many instances that use the default values, you don’t have to store any data for them.

On the other hand, there are drawbacks to this approach. Accessing value becomes a lot more complicated and expensive, because we always need to perform an extra query to find out if the instance has overridden the default value or not.

We currently don’t use this approach anywhere in the engine. But I think there are circumstances where it would make sense.

Anyway, I’m getting sidetracked. Back to the entity system.

The thing is, the way I envision the entity system, it is very dynamic. Components can be added and removed at runtime, child entities linked in and properties changed. Components that handle static data, such as the MeshComponent do it by referencing a separate MeshResource that contains the mesh data. There is no mesh data stored in the component itself.

Since everything in the entity system is dynamic, there is only instance data. The only thing we have in the resource is the template for the instance data. Essentially, just a set of “instructions” for setting up an instance. There is no need for the instance to refer back to the resource after those instructions have been followed.

Defining the Resource Format

So an entity resource should contain “instructions” for setting up an entity. What should it look like? Let’s start by just writing up what needs to go in there:

struct EntityResource
{
    unsigned num_components;
    ComponentData components[num_components];
    unsigned num_children;
    EntityResource children[num_children];
};

Note: Of course the above is not legal C++ code. I’m using some kind of C-like pseudo-code that allows things like dynamically sized structs in order to describe the data layout. I’ve written about the need for a language to describe data layouts before.

The exact binary layout of the ComponentData is defined by each component type, but let’s use a common wrapper format:

struct ComponentData
{
    unsigned component_identifier;
    unsigned size;
    char data[size];
};

Now we have a common way of identifying the component type, so we know if we should create a MeshComponent, a TransformComponent or something else. We also know the size of the component data, so if we should encounter a component type that we don’t understand, we can ignore it and skip over its data to get to the next component. (Another option would be to treat unknown component types as a fatal error.)

A quick fix to make this layout slightly better is to move all the fixed size fields to the start of the struct:

struct EntityResource
{
    unsigned num_components;
    unsigned num_children;
    ComponentData components[num_components];
    EntityResource children[num_children];
};

Now we can access the num_children parameter without having to look at all the components and their sizes to know how far we need to skip forward in the resource to get to the num_children field.

This may or may not matter in practice. Perhaps, we only need the value of num_children after we have processed all the component data, and at that point we already have a pointer into the resource that points to the right place. But I always put the fixed size data first as a force of habit, in case we might need it.

Sometimes, it makes sense to add offset tables to these kinds of resources, so that we can quickly lookup the offset of a particular component or child, without having to walk all of the memory and count up the sizes:

struct EntityResource
{
    unsigned num_components;
    unsigned num_children;
    unsigned offset_to_component_data[num_components];
    unsigned offset_to_child_data[num_children];
    ComponentData components[num_components];
    EntityResource children[num_children];
};

With this layout, we can get to the data for the i’th component and the j’th child as:

struct EntityResourceHeader
{
    unsigned num_components;
    unsigned num_children;
};

const EntityResourceHeader *resource;
const unsigned *offset_to_component_data = (const unsigned *)(resource + 1);
ComponentData *data_i = (const ComponentData *)
    ((const char *)resource + offset_to_component_data[i]);

const unsigned *offset_to_child_data = (const unsigned *)
    (offset_to_component_data + num_components);
EntityResourceHeader *child_j = (const EntityResourceHeader *)
    ((const char *)resource + offset_to_child_data[j]);

The first time you encounter code like this it can seriously spin your head around with all the casting and pointer arithmetic. However, if you think about what happens and how the data is laid out in memory it is really pretty straight forward. Any mistakes you do will most likely cause huge crashes that are easy to find, not sneaky subtle bugs. And after a while you get used to these kinds of manipulations.

But, anyway, I’m drifting off on a tangent again, because actually for our purposes we don’t need these lookup tables. We will just walk the memory from beginning to end, creating one component at a time. Since we don’t need to jump around between different components, we don’t need the lookup tables.

What we do need though is some way of storing more than one resource. Storing one entity is fine if we are dealing with a “prefab” type of resource, that contains a definition of a single entity. However, what about a level? It will probably contain a bunch of entities. So it would be nice to have a resource type that could store all those entities.

Ok, no biggie, we know how to do that:

struct EntitiesResource
{
    unsigned num_entities;
    EntityResource entities[num_entities];
};

Done, yesno?

Pivot!

Working for a while in this business you get an intuitive feel for when performance matters and when it doesn’t. Of course, intuitions can be wrong, so don’t forget to measure, measure, measure. But level spawning tends to be one of these areas where performance does matter.

A level can easily have 10 000 objects or more and sometimes you want to spawn them really fast, such as when the player restarts the level. So it seems worth it to spend a little bit of time to think about how we can spawn levels fast.

Looking at the resource layout, our spawn algorithm seems pretty straight forward:

  • Create the first entity
    • Add its first component
    • Add its second component
    • Create its child entities
  • Create the second entity
    • Create its first component
    • Create its second component

This is so simple and straight forward that it might seem impossible to improve on. We are walking the resource memory linearly as we step through components, so we are being cache friendly, aren’t we?

Well, not really. We are violating one of the fundamental principles of data-oriented design: Do similar things together.

If we write out the operations we actually perform linearly instead of in an hierarchy and make things a bit more concrete, it’s easier to see:

  • Create entity A
  • Create a TransformComponent for A
  • Create a MeshComponent for A
  • Create an ActorComponent for A
  • Create entity B
  • Create a TransformComponent for B
  • Create a MeshComponent for B

Note how we are alternating between creating different kinds of components and entities. This not only messes with our instruction cache (because each component has its own code path), but with our data cache as well (because each component has its own data structures where the instances get inserted).

So let’s rewrite this so that we keep common operations together:

  • Create entity A
  • Create entity B
  • Create a TransformComponent for A
  • Create a TransformComponent for B
  • Create a MeshComponent for A
  • Create a MeshComponent for B
  • Create an ActorComponent for A

Much better. And we can go even further.

Instead of telling the EntityManager to “create an entity” one hundred times, let’s just tell it to “create 100 entities”. That way, if there is any economy of scale to creating more than one entity, the EntityManager can take advantage of that. And let’s do the same thing for the components:

  • Create entities (A, B)
  • Create TransformComponents for (A,B)
  • Create MeshComponents for (A,B)
  • Create an ActorComponent for A

Notice how we are encountering and making use of a whole bunch of data-oriented principles and guidelines here:

  • Access memory linearly.
  • Where there is one, there are many.
  • Group similar objects and operations together.
  • Perform operations on multiple objects at once, rather than one at a time.

Let’s rewrite the data format to reflect all our newly won insight:

struct EntityResource
{
    unsigned num_entities;
    unsigned num_component_types;
    ComponentTypeData component_types[num_component_types];
};

struct ComponentTypeData
{
    unsigned component_identifier;
    unsigned num_instances;
    unsigned size;
    unsigned entity_index[num_instances];
    char instance_data[size];
};

For each component, we store an identifier so we know if it’s a MeshComponent, TransformComponent, etc. Then we store the number of instances of that component we are going to create and the size of the data for all those instances.

Note that now when we are walking the format, we can skip all instances of an unknown component type with a single jump, instead of having to ignore them one by one. This doesn’t matter that much, but it is interesting to note that data-oriented reorganizations often make a lot of different kinds of operations more efficient, not just the one you initially targeted.

The entity_index is used to associate components with entities. Suppose we create five entities: A, B, C, D and E and two ActorComponents. We need to know which entity each ActorComponent should belong to. We do that by simply storing the index of the entity in the entity_index. So if the entity index contained {2,3} the components would belong to C and D.

There is one thing we haven’t handled in the new layout: child entities.

But child entities are not conceptually different from any other entities. We can just add them to num_entities and add their component instances to the ComponentTypeData just as we would do for any other entity.

The only additional thing we need is some way of storing the parent-child relationship. We could store that as part of the data for the TransformComponent, or we could just store an array that specified the index of each parent’s entity (or UINT_MAX for root entities):

struct EntityResource
{
    unsigned num_entities;
    unsigned num_component_types;
    unsigned parent_index[num_entities];
    ComponentTypeData component_types[num_component_types];
};

If parent_index was {UINT_MAX, 0, 1, 1, 2} in our A, B, C, D, E example, the hierarchy would be:

A --- B --- C --- E
      |
      + --- D

Implementation Details

This post is too long already, so I’ll just say something quickly about how the implementation of this is organized.

In the engine we have a class EntityCompiler for compiling entities and a similar class EntitySpawner for spawning entities.

A component that can compile data needs to register itself with the entity compiler, so that it can be called when component data of that kind is encountered by the compiler.

Ignoring some of the nitty-gritty details, like error handling, endian swapping and dependency tracking, this looks something like this:

typedef Buffer (*CompileFunction)(const JsonData &config, NittyGritty &ng);

void register_component_compiler(const char *name, CompileFunction f,
    int spawn_order);

The compile function takes some JSON configuration data that describes the component and returns a binary BLOB of resource data for insertion into the entity resource. Note that the compile function operates on a single component at a time, because we are not that concerned with compile time performance.

When registering the compiler we specify a name, such as "mesh_component". If that name is found in the JSON data, the entity compiler will redirect the compile of the component data to this function. The name is also hashed into the component_identifier for the component.

The spawn_order is used to specify the compile order of the different component, and by extension, their spawn order as well. Some components make use of other components. For example, the MeshComponent wants to know where the enitty is, so it looks for a TransformComponent in the entity. Thus, the TransformComponent must be created before the MeshComponent.

A similar approach is used to register a component spawner:

typedef void (*SpawnFunction)(const Entity *entity_lookup,
    unsigned num_instances, const unsigned *entity_index, const char *data);

void register_component_spawner(const char *name, SpawnFunction f);

Here the entity_lookup allows us to look up an entity index in the resource data to a an actual Entity that is created in the first step of spawning the resource. num_instances is the number of component instances that should be created and entity_index is the entity index from the ComponentTypeData that lets us lookup which entity should own the component.

So entity_lookup[entity_index[i]] gives the Entity that should own the ith component instance.

The data finally is a pointer to the instance_data from the ComponentTypeData.

That’s certainly enough for today. Next time, we’ll look at a concrete example of this.

Friday, October 3, 2014

Building a Data-Oriented Entity System (Part 3: The Transform Component)

In the last post, I talked generally about the design of components. Today I will focus on a specific real-world component, the TransformComponent.

The Transform Component

The purpose of the TransformComponent is to allow entities to be positioned in the world.

It handles positioning of entities in the world and child-parent linking. For example, you may want to link a “wheel” entity to a “car” entity, so that the wheel follows the car around when it moves.

In that sense, the TransformComponent is what forms the scene graph of an engine world.

Design Decisions

Should every entity have a transform component?

In some engines, every entity has to have a transform component, even if it is just a purely “logical” entity that doesn’t really have a position in the world.

To me it seems strange to force an entity to have a position when that has no real meaning. I also want entities to be as cheap as possible. So it seems better to have the transform component optional, just as any other component. An entity that doesn’t have a transform component doesn’t have any position in the world.

Actually, talking about the world is a bit of misnomer. The Bitsquid engine does not have a single World where everything has to live. Instead you can create multiple worlds, each populated with its own objects. So you might have one world for your “main game”, one world for the “inventory screen”, one world for the “loading screen”, etc.

This is a lot better than having an “inventory room” at some secret hidden place in the main game world.

Each world has its own TransformComponent manager, and an entity can create transform components in several of these managers, if it so desires. So the same entity can exist and be positioned at different places in different game worlds. Since a MeshComponent manager also exists in each world, the entity can have different graphical representations in each world.

This is a bit esoteric, and I don’t expect many entities to make use of this, but there are situations when it could be interesting. For example, a player’s pickaxe could exist both in the “game world” and in the “inventory world” and still be managed as the same entity.

Entity scene graphs and model scene graphs

In the entity system there are really two kinds of “scene graphs” that we need to deal with.

The first is the one we have already talked about, the graph formed by entities and their linked child entities.

The second is the graph of nodes within an entity. For example, a character entity may have a model with hundreds of bones that can be individually animated.

What should the relationship be between these two graphs?

In previous engine code, I have always treated these two graphs as parts of the same system. The model scene graphs were linked to nodes in the entity scene graphs and computed their transforms in world space. This creates an update order dependency. We can’t compute the world positions in the model scene graph until we have computed the world position in the entity scene graph. This limits what kinds of things we can do in parallel.

For the entity system I’ve decided to decouple these two concepts. The model scene graph won’t compute world space poses, instead it will compute poses relative to the entity pose. This means that we can evaluate the animations and compute the model pose without knowing anything about the entity pose. (Ignoring world space constraints, of course, but they will be handled in a later pass.)

Of course it also requires us to multiply the model node transforms with the entity transform to get the actual world position of the model nodes.

I have not completed the design of the model scene graph component yet, but maybe I’ll get a chance to return to this in a future post.

Immediate or deferred updates

In previous engines I have always used deferred updates of the world transforms. I.e., changing the local transform of a node would not immediately update its world transform (or the world transforms of its children). Instead it would simply set a “dirty” flag in the entity. Later, I would compute the world transforms of all the dirty nodes (and their children) as a single step.

This has the advantage that we never have to compute the world transform of a node more than once.

Consider the worst case scenario, a long chain of nodes:

[ node_1 ] ---> [ node_2 ] ---> [ node_3 ] ---> ... ---> [ node_n ]

With a deferred update, changing the local pose of every node will still just require O(n) computations to compute all the world transforms. With an immediate update, where we compute the world transforms of all children as soon as the parent transform changes, we will need O(n^2) computations.

On the other hand, there is a drawback to using deferred updates. Whenever we ask for an object’s world position we won’t get its actual world position, but its world position from the last frame (unless we ask after the world transform update). This can lead to a lot of confusion and subtle bugs. Solving them often requires ugly hacks, such as forcing graph updates at different times.

So what should we choose?

I think that with the decision to decouple the model scene graphs from the entity scene graphs the performance problems of immediate updates are a lot less serious. Long chains of nodes that are all moving can certainly exist in the model scene graph. (Consider an animation of a character swinging a whip.) But I would guess that long chains of objects that are all moving at once are a lot less common in the entity scene graph.

Note that the performance problems do not appear if it is just the root entity that is moving. In that case, both the immediate and the deferred update will be O(n). It is only when the parent and the children are moving that the immediate update does worse.

I don’t expect there to be very long chains of entities (n <= 5 ???) and I don’t expect all of the objects in those chains to be moving simultaneously. So I have decided to go with immediate updates so that we always have accurate world transforms.

Note: If we run into performance problems as a result of this, we can always create an API function that allows us to set multiple local transforms at once while doing a single world transform update, thus getting back the O(n) performance.

A side note on deferred updates

Note that if you want to do deferred updates, you want to keep the entity array sorted so that parents always appear before children. That way you can just walk the array from beginning to end and compute the transforms and be sure that the world transform of a parent has been computed before you compute the world transform of its children.

Also, you don’t want to loop over the entire array to look for dirty objects:

for (int i=0; i<n; ++i) {
    if (dirty[i])
        transform(i);
}

Typically, in a scene, only a small percentage of the objects are moving at any one time (maybe as little as 1 %). So looping over all objects, even just to check a flag, can waste a lot of time.

A better solution is to sort all the dirty objects to the end of the array, so we can loop over just them:

for (int i=first_dirty; i<n; ++i)
    transform(i);

Since we only need a partial sorting of the array, we don’t have to run an expensive O(n log n) sorting algorithm. (It would kind of defeat the purpose to run an O(n log n) sort to avoid an O(n) update.) Instead, we can achieve this by judicious swapping.

When a node becomes dirty we move it to the start of the dirty list by swapping it with the element before the dirty list and decreasing first_dirty:

                                 =============== dirty ==============
|   |   |   | D |   |   |   | X |   |   |   |   |   |   |   |   |   |

                             ================= dirty ================
|   |   |   | X |   |   |   | D |   |   |   |   |   |   |   |   |   |

We do the same for all children of the node and the children’s children, etc.

As we process the items in the dirty array, whenever we find a child that has its parent at a later position in the array, we swap the child and the parent.

                             ================= dirty ================
|   |   |   |   |   |   |   |   |   |   | C |   |   | P |   |   |   |
                                          ^

                             ================= dirty ================
|   |   |   |   |   |   |   |   |   |   | P |   |   | C |   |   |   |
                                          ^

This guarantees that parents are always processed before their children.

We also need a way to move items off the dirty list, or it will continue to grow indefinitely. We could clear the list every frame, but that might lead to a lot of swapping as items are moved in and out of the list. A better approach might be to check if an item hasn’t moved in five frames or so, and in that case we move it off the dirty list. This avoids swapping those items which are always moving.

When using the immediate update strategy, sorting the list is not as important, but we can employ similar swapping strategies to make sure that a parent node and its children are kept close together in the array, so that the immediate update is cache friendly.

Implementation

With the design thought through, there is really not that much to the implementation.

Just as in the last post, we store the transform component data for all instances in a single big memory block:

struct Instance {int i;};

/// Instance data.
struct InstanceData {
    unsigned size;              ///< Number of used entries in arrays
    unsigned capacity;          ///< Number of allocated entries in arrays
    void *buffer;               ///< Raw buffer for data.

    Entity *entity;             ///< The entity owning this instance.
    Matrix4x4 *local;           ///< Local transform with respect to parent.
    Matrix4x4 *world;           ///< World transform.
    Instance *parent;           ///< The parent instance of this instance.
    Instance *first_child;      ///< The first child of this instance.
    Instance *next_sibling;     ///< The next sibling of this instance.
    Instance *prev_sibling;     ///< The previous sibling of this instance.
};

The parent, first_child, next_sibling and prev_sibling arrays all store instance indexes. We can find all the children of a particular entity by following the first_child link and then the next_sibling links of that link.

We can use that to do the immediate transform update:

void TransformComponent::set_local(Instance i, const Matrix4x4 &m)
{
    _data.local[i.i] = m;
    Instance parent = _data.parent[i.i];
    Matrix4x4 parent_tm = is_valid(parent) ? _data.world[ parent.i ] :
        matrix4x4_identity();
    transform(parent_tm, i);
}

void TransformComponent::transform(const Matix4x4 &parent, Instance i)
{
   _data.world[i.i] = _data.local[i.i] * p;

    Instance child = _data.first_child[i.i];
    while (is_valid(child)) {
       transform(_data.world[i.i], child);
       child = _data.next_sibling[child.i];
    }
}

Note: I’ve written this as a recursive function for easier reading, but you might want to rewrite it as an iterative function for better performance.

Note that when you swap two instances in the array (to do swap-erase or to sort the array as described above), in addition to swapping the entries in the array you also need to take care to keep all the parent, first_child, next_sibling and prev_sibling references intact. This can get a little hairy, especially when you are changing references and trying to walk those lists of references at the same time. My suggestion when you want to swap two instances [A] and [B] is to use the element at the end of the array [size] as a temporary storage slot and instead of trying to do everything at once, use three steps:

// Move element at A (and references to it) to size.
[size] <--- [A]

// Now nothing refers to A, so we can safely move element at B (and references
// to it) to A.
[A] <--- [B]

// And finally move the element at size to B.
[B] <-- [size]

In the next post I’ll look at compiling entities into resource files.

Monday, September 8, 2014

Building a Data-Oriented Entity System (Part 2: Components)

In the last post, I talked about the design of the Entity Manager and how we handle creation and destruction of game entities.

In this post we will look at how components can be implemented.

A quick recap: Components in our system are not individual objects, instead all components of a particular type are handled by a component manager for that type. The component manager has full control over how the component data is stored internally and how updates are applied.

A Component Example

To have something to talk about we will consider a fictitious component that handles point mass objects. For each component instance we want to store the following data:

Entity entity;          ///< Entity owner
float mass;             ///< Mass of object
Vector3 position;       ///< Object's position
Vector3 velocity;       ///< Object's velocity
Vector3 acceleration;   ///< Object's acceleration

The component needs functions for accessing this data and simulating physics.

It is perhaps not self-evident why we want to store the entity that owns the component, but it will come in handy later.

Note that this is not a real world example. We don’t actually have a component like this in the engine, and perhaps it’s not the best or most interesting design, but it gives us something to talk about.

Component Data Layout

When considering how we should layout the data in the component manager we have two goals:

  • Given an entity we want to be able to quickly look up the component data for that entity.
  • We want the component data to be packed tightly in memory for good cache performance.

Let’s tackle the second question first.

Actual cache performance depends on how your CPU works and what the data access patterns in the code are. You can spend a lot of time trying to bend your mind around those things, but I would recommend going with a simple rule of thumb instead:

Pack the data in arrays that you access sequentially.

Only get more fancy than that when you are trying to fix a diagnosed performance issue.

A generally good approach is to use a structure-of-arrays. I.e., each field is stored in an array in memory, with one entry for each component instance:

[entity_1]  [entity_2]  [entity_3] ...
[mass_1]    [mass_2]    [mass_3]   ...
[pos_1]     [pos_2]     [pos_3]    ...
[vel_1]     [vel_2]     [vel_3]    ...
[acc_1]     [acc_2]     [acc_3]    ...

The advantage of having each field stored separately is that code that only processes some of the fields don’t have to waste precious cache space on the others.

You could go even further and put each x, y and z component of a Vector3 into its own array. An advantage of that is that you can do more efficient SIMD calculations, if you want to go down that route. But for this example, let’s keep things a bit simpler and store the Vector3s together. Since the layout of the data is entirely encapsulated in the ComponentManager class we can always go back and redesign that later if we need some extra performance.

The simplest way of implementing this data layout is to use an Array for each component:

class PointMassComponentManager {
    struct InstanceData {
        Array<Entity> entity;
        Array<float> mass;
        Array<Vector3> position;
        Array<Vector3> velocity;
        Array<Vector3> acceleration;
    };
    InstanceData _data;
};

That works well enough, but it does mean that the data gets stored in five separately allocated memory buffers. So I use a different approach. I allocate the entire memory buffer as a single allocation and then just let entity, mass, etc, point to different parts of that buffer:

struct InstanceData {
    unsigned n;          ///< Number of used instances.
    unsigned allocated;  ///< Number of allocated instances.
    void *buffer;        ///< Buffer with instance data.

    Entity *entity;
    float *mass;
    Vector3 *position;
    Vector3 *velocity;
    Vector3 *acceleration;
};
InstanceData _data;

void allocate(unsigned sz)
{
    assert(sz > _data.n);

    InstanceData new_data;
    const unsigned bytes = sz * (sizeof(Entity) + sizeof(float) +
        3 * sizeof(Vector3));
    new_data.buffer = _allocator.allocate(bytes);
    new_data.n = _data.n;
    new_data.allocated = sz;

    new_data.entity = (Entity *)(new_data.buffer);
    new_data.mass = (float *)(new_data.entity + sz);
    new_data.position = (Vector3 *)(new_data.mass + sz);
    new_data.velocity = new_data.position + sz;
    new_data.acceleration = new_data.velocity + sz;

    memcpy(new_data.entity, _data.entity, _data.n * sizeof(Entity));
    mempcy(new_data.mass, _data.mass, _data.n * sizeof(float));
    memcpy(new_data.position, _data.position, _data.n * sizeof(Vector3));
    memcpy(new_data.velocity, _data.velocity, _data.n * sizeof(Vector3));
    memcpy(new_data.acceleration, _data.acceleration,
        _data.n * sizeof(Vector3));

    _allocator.deallocate(_data.buffer);
    _data = new_data;
}

This avoids any hidden overheads that might exist in the Array class and we only have a single allocation to keep track of. This is better both for the cache and the memory allocation system.

Side note: I’m tempted to write a memory system with a 4 K allocation granularity. I.e. there is no traditional heap allocator, just a page allocator and you have to design your systems so that they only work with large allocations.

Accessing Data

Let’s consider the second issue, how we map from an entity to its component data. For the sake of simplicity, let’s assume for now that we don’t support multiple components per entity.

In the data layout, we refer to a particular component instance by its index in the mass, position, etc arrays. So what we need is a way to map from an entity to an index.

You may remember from the previous post, that Entity itself contains a unique index. So one alternative would be to just use this index.

This could be a good approach if almost every entity in the game had this component. But if that is not the case our arrays will contain a lot of “holes” corresponding to entities that lack the component. This will waste memory, but also performance, because we will fill our caches with unused data.

We can improve this somewhat by using a level of indirection:

Array<unsigned> _map;

Here, the _map allows us to look up a component index based on the entity index. This is a lot better, because now it is just the _map array that has holes, not the _data array, which means that the holes are fewer and smaller.

Still, I would only use this if I was certain that the component was almost universal and that lookups where performance critical. In most cases, I think a hash index is a better approach:

HashMap<Entity, unsigned> _map;

This uses less memory and lookups are still pretty fast.

Since the lookup from Entity to instance index involves an extra step we want to reflect that in the API and not force the user to do multiple lookups when she wants to access different fields of the same component. Something like this:

/// Handle to a component instance.
struct Instance {int i;};

/// Create an instance from an index to the data arrays.
Instance make_instance(int i) {Instance inst = {i}; return inst;}

/// Returns the component instance for the specified entity or a nil instance
/// if the entity doesn't have the component.
Instance lookup(Entity e) {return make_instance(_map.get(e, 0));}

float mass(Instance i) {return _data.mass[i.i];}
void set_mass(Instance i, float mass) {_data.mass[i.i] = mass;}
Vector3 position(Instance i) {return _data.position[i.i];}
...

To support multiple component instance per entity, you can add a next_instance field to the component data that allows you to traverse a linked list of component instances belonging to the same entity. This is left as an exercise to the reader.

Component Updates

Since the component data is laid out sequentially in memory, writing a function that simulates physics for all entities is simple:

void simulate(float dt)
{
    for (unsigned i=0; i<_data.n; ++i) {
        _data.velocity[i] += _data.acceleration[i] * dt;
        _data.position[i] += _data.velocity[i] * dt;
    }
}

This function traverses memory in-order which gives us good cache performance. It’s also easy to profile, vectorize and parallelize, should the need arise.

Side rant: I’m somewhat allergic to methods being called update(). That is a bad remain from bad inheritance-based designs. If you take a second to think about it you can almost always come up with better, more informative names than update().

Destroying Components

When destroying components, we want to make sure that we keep the _data array tightly packed. We can achieve that by moving the last element to the position of the component we want to remove. We must also update the _map entry for the corresponding entity.

void destroy(unsigned i)
{
    unsigned last = _data.n - 1;
    Entity e = _data.entity[i];
    Entity last_e = _data.entity[last];

    _data.entity[i] = _data.entity[last];
    _data.mass[i] = _data.mass[last];
    _data.position[i] = _data.position[last];
    _data.velocity[i] = _data.velocity[last];
    _data.acceleration[i] = _data.acceleration[last];

    _map[last_e] =  i;
    _map.erase(e);

    --_n;
}

Another question is how we handle destruction of components when an entity is destroyed. As you may recall, the entity does not have an explicit list of components that it owns. Also, it seems onerous to require of the user of the API to manually destroy the right components when the entity dies.

Instead, we use one of two approaches.

Components that need to be destroyed immediately (perhaps because they hold external resources) can register a destruction callback with the EntityManager and that callback will be called when the entity is destroyed.

However, for simpler components, like the point mass component, there is nothing that require components to be destroyed at exactly the same time as the entity. We can take advantage of that and use garbage collection to lazily destroy components instead of spending memory and effort on storing callback lists:

void gc(const EntityManager &em)
{
    unsigned alive_in_row = 0;
    while (_data.n > 0 && alive_in_row < 4) {
        unsigned i = random_in_range(0, _data.n - 1);
        if (em.alive(_data.entity[i])) {
            ++alive_in_row;
            continue;
        }
        alive_in_row = 0;
        destroy(i);
    }
}

Here, we pick random component indices and destroy them if the corresponding entity has been destroyed. We do this until we hit four living entities in a row.

The nice thing about this code is that it cost almost nothing if there are no destroyed entities (just four passes of the loop). But when there are a lot of destroyed entities the components will be quickly destroyed.

In the next post, we will look at the Transform Component that handles links between parent and child entities.

Wednesday, August 27, 2014

Building a Data-Oriented Entity System (part 1)

We have recently started to look into adding an entity/component system to the Bitsquid engine.

You may be surprised to learn that the Bitsquid engine isn't already component based. But actually there has never been a great need for that. Since the gameplay code is usually written in Lua rather than C++, we don't run into the common problems with deep and convoluted inheritance structures that prompt people to move to component based designs. Instead, inheritance is used very sparingly in the engine.

But as we are expanding our plugin system, we need a way for C++ plugins to bestow game objects with new functionalities and capabilities. This makes a component architecture a very natural fit.

Entities and Components

In the Bitsquid engine, we always strive to keep systems decoupled and data-oriented and we want to use the same approach for the component architecture. So, in our system, entities are not heap allocated objects. Instead, an entity is just an integer, a unique ID identifying a particular entity:

struct Entity
{
 unsigned id;
};

A special class, the EntityManager keeps track of the entities that are alive.

A component is not an object either. Instead, a component is something that is handled by a ComponentManager. The task of a ComponentManager is to associate entities with components. For example, the DebugNameComponentManager can be used to associate debug names with entities:

class DebugNameComponentManager
{
public:
 void set_debug_name(Entity e, const char *name);
 const char *debug_name(Entity e) const;
};

Two things are interesting to note about this decoupled design.

First, there is no DebugNameComponent class for handling individual debug name components in this design. That is not needed, because all component data is managed internally by the DebugNameComponentManager. The manager could decide to use heap allocated DebugNameComponent objects internally. But it is not forced to. And usually it is much more efficient to lay out the data differently. For example, as a structure of arrays in a single continuous buffer. In a future post, I'll show some examples of this.

Second, there is no place where we keep a list of all the components that an entity has. It is only the DebugNameComponentManager that knows whether an entity has a debug name component or not, and if you want to talk about that component you have to do it through the DebugNameComponentManager. There is no such thing as an "abstract" component.

So what components an entity has is only defined by what has been registered with the different component managers in the game. And plugins may extend the system with new component managers.

It is up to the component manager to decide if it makes sense for an entity to have multiple components of its type. For example, the DebugNameComponentManager only allows a single debug name to be associated with an entity. But the MeshComponentManager allows an entity to have multiple meshes.

The manager is responsible for performing any computations necessary to update the components. Updates are done one component manager at a time, not one entity at a time, and when a component manager is updated it updates all its components in one go. This means that common calculations can be shared and that all the data is hot in the caches. It also makes the update easier to profile, multithread or offload to an external processor. All this translates to huge performance benefits.

The EntityManager

We want to be able to use the entity ID as a weak reference. I.e., given an entity ID we want to be able to tell if it refers to a living entity or not.

Having a weak reference system is important, because if we only have strong references then if the entity dies we must notify everybody that might possibly hold a reference to the entity so that they can delete it. This is both costly and complicated. Especially since references might be held by other threads or by Lua code.

To enable weak referencing, we use the EntityManager class to keep track of all live entities. The simplest way of doing that would be to just use a set:

class EntityManager
{
 HashSet<Entity> _entities;
 Entity _next;

public:
 Entity create()
 {
  ++_next.id;
  while (alive(_next))
   ++_next.id;
  _entities.insert(_next);
  return _next;
 }

 bool alive(Entity e)
 {
  return _entities.has(e);
 }

 void destroy(Entity e)
 {
  _entities.erase(e);
 }
};

This is pretty good, but since we expect the alive() function to be a central piece of code that gets called a lot, we want something that runs even faster than a set.

We can change this to a simple array lookup by splitting the entity ID into an index and a generation part:

const unsigned ENTITY_INDEX_BITS = 22;
const unsigned ENTITY_INDEX_MASK = (1<<ENTITY_INDEX_BITS)-1;

const unsigned ENTITY_GENERATION_BITS = 8;
const unsigned ENTITY_GENERATION_MASK = (1<<ENTITY_GENERATION_BITS)-1;

struct Entity
{
 unsigned id;

 unsigned index() const {return id & ENTITY_INDEX_MASK;}
 unsigned generation() const {return (id >> ENTITY_INDEX_BITS) & ENTITY_GENERATION_MASK;}
};

The idea here is that the index part directly gives us the index of the entity in a lookup array. The generation part is used to distinguish entities created at the same index slot. As we create and destroy entities we will at some point have to reuse an index in the array. By changing the generation value when that happens we ensure that we still get a unique ID.

In our system we are restricted to using 30 bits for the entity ID. The reason for this is that we need to fit it in a 32 bit pointer in order to be able to use a Lua light userdata to store it. We also need to steal two bits from this pointer in order to distinguish it from other types of light userdata that we use in the engine.

If you didn't have this restriction, or if you only targeted 64-bit platforms it would probably be a good idea to use some more bits for the ID.

We've split up our 30 bits into 22 bits for the index and 8 bits for the generation. This means that we support a maximum of 4 million simultaneous entities. It also means that we can only distinguish between 256 different entities created at the same index slot. If more than 256 entities are created at the same index slot, the generation value will wrap around and our new entity will get the same ID as an old entity.

To prevent that from happening too often we need to make sure that we don't reuse the same index slot too often. There are various possible ways of doing that. Our solution is to put recycled indices in a queue and only reuse values from that queue when it contains at least MINIMUM_FREE_INDICES = 1024 items. Since we have 256 generations, an ID will never reappear until its index has run 256 laps through the queue. So this means that you must create and destroy at least 256 * 1024 entities until an ID can reappear. This seems reasonably safe, but if you want you can play with the numbers to get different margins. For example, if you don't need 4 M entities, you can steal some bits from index and give to generation.

A nice thing about only having 8 bits in generation is that we just need 8 bits per entity in our lookup array. This saves memory, but also gives us better performance, since we will fit more in the cache. With this solution, the code for the EntityManager becomes:

class EntityManager
{
 Array<unsigned char> _generation;
 Deque<unsigned> _free_indices;

public:
 Entity create()
 {
  unsigned idx;
  if (_free_indices.size() > MINIMUM_FREE_INDICES) {
   idx = _free_indices.front();
   _free_indices.pop_front();
  } else {
   _generation.push_back(0);
   idx = _generation.size() - 1;
   XENSURE(idx < (1 << ENTITY_INDEX_BITS));
  }
  return make_entity(idx, _generation[idx]);
 }

 bool alive(Entity e) const
 {
  return _generation[e.index()] == e.generation();
 }

 void destroy(Entity e)
 {
  const unsigned idx = e.index();
  ++_generation[idx];
  _free_indices.push_back(idx);
 }
};

In the next post, we will take a look at the design of the component classes.

Wednesday, June 18, 2014

What Is In a Name?

Today I'd like to revisit one of the most basic questions when designing a resource system for a game engine:

How should resources refer to other resources?

It seems like a simple, almost trivial question. Yet, as we shall see, no matter what solution we choose, there are hidden pitfalls along the way.

To give some context to the question, let's assume we have a pretty typical project setup. We'll assume that our game project consists of a number of individual resources stored in a disk hierarchy that is checked into source control.

There are three basic ways of referring to resources that I can think of:

  • By path

  • By GUID

  • By "name"

By Path

texture = "textures/flowers/rose"

This is the most straightforward approach. To refer to a particular resource you just specify the path to that resource.

A word of warning: If you use paths as references I would recommend that you don't accept ridiculous things such as "./././models\../textures\FLOWers/////rose" even though your OS may think that is a perfectly valid path. Doing that will just lead to lots of headaches later when trying to determine if two paths refer to the same resource. Only use a canonical path format, from the root of the project, so that the path to same resource is always the same identical string (and can be hashed).

Path references run into problem when you want to rename a resource:

textures/flowers/rose -> textures/flowers/less-sweet-rose

Suddenly, all the references that used to point to the rose no longer works and your game will break.

There are two possible ways around this:

Redirects

You can do what HTML does and use a redirect.

I.e., when you move rose, you put a little placeholder there that notifies anyone who is interested that this file is now called less-sweet-rose. Anyone looking for rose will know by the redirect to go looking in the new place.

There are three problems with this, first the disk gets littered with these placeholder files. Second, if you at some point in the future want to create a new resource called rose, you are out of luck, because that name is now forever occupied by the placeholder. Third, with a lot of redirects it can be hard to determine when two things refer to the same resource.

Renaming tool

You can use a renaming tool that understands all your file formats, so that when you change the path of a resource, the tool can find all the references to that path and update them to point to the new location.

Such a tool can be quite complicated to write -- depending on how standardized your file formats are. It can also be very slow to run, since potentially it has to parse all the files in your project to find out which other resources might refer to your resource. To get decent performance, you have to keep an up-to-date cache of the referencing information so that you don't have to read it every time.

Another problem with this approach can occur in distributed workflows. If one user renames a resource while another creates references to it, the references will break when the changes are merged. (Note that using redirects avoids this problem.)

Both these methods require renames to be done with a tool. If you just change the file name on disk, without going through the tool, the references will break.

By GUID

The problems with renaming can be fixed by using GUIDs instead of paths. With this approach, each resource specifies a GUID that uniquely identifies it:

guid = "a54abf2e-d4a1-4f21-a0e5-8b2837b3b0e6"

And other resources refer to it by using this unique identifier:

texture = "a54abf2e-d4a1-4f21-a0e5-8b2837b3b0e6"

In the compile step, we create an index that maps from GUIDs to compiled resources that we can use to look things up by GUID.

Now files can be freely moved around on disk and the references will always be intact. There is not even a need for a special tool, everything will work automatically. But unfortunately there are still lots of bad things that can happen:

  • If a file is copied on disk, there will be two files with the same GUID, creating a conflict that needs to be resolved somehow (with a special tool?)

  • Lots of file formats that we might want to use for our resources (.png, .wav, .mp4, etc) don't have any self-evident place where we can store the GUID. So the GUID must be stored in a metadata file next to the original file. This means extra files on disk and potential problems if the files are not kept in sync.

  • Referring to resources from other resources is not enough. We also need some way of referring to resources from code, and writing:

    spawn_unit("a54abf2e-d4a1-4f21-a0e5-8b2837b3b0e6")

    is not very readable.

  • If a resource is deleted on disk, the references will break. Also if someone forgets to check in all the required resources, the references will break. This will happen no matter what reference system we use, but with GUIDs, everything is worse because the references:

    texture = "a54abf2e-d4a1-4f21-a0e5-8b2837b3b0e6"

    are completely unreadable. So if/when something breaks we don't have any clue what the user meant. Was that resource meant to be a rose, a portrait, a lolcat or something else.

In summary, the big problem is that GUIDs are unreadable and when they break there is no clue to what went wrong.

By "Name"

Perhaps we can fix the unreadability of GUIDs by using human readable names instead. So instead of a GUID we would put in the file:

name = "garden-rose"

And the reference would be:

texture = "garden-rose"

To me, this approach doesn't have any advantages over using paths. Sure, we can move and rename files freely on disk, but if we want to change the name of the resource, we run into the same problems as we did before. Also, it is pretty confusing that a resource has a name and a file name and those can be different.

By Path and GUID?

Could we get the best of both worlds by combining a path and a GUID?

I.e., the references would look like:

texture = {
 path = "textures/flower/rose"
 guid = "a54abf2e-d4a1-4f21-a0e5-8b2837b3b0e6"
}

The GUID would make sure that file renames and moves were handled properly. The path would give us the contextual information we need if the GUID link breaks. We would also use the path to refer to resources from code.

This still has the issue with needing a metadata file to specify the GUID. Duplicate GUIDs can also be an issue.

And also, if you move a file, the paths in the references will be incorrect unless you run a tool similar to the one discussed above to update all the paths.

Conclusions

In the Bitsquid engine we refer to resources by path. Frustrating as that can be sometimes, to me it still seems like the best option. The big problem with GUIDs is that they are non-transparent and unreadable, making it much harder to fix stuff when things go wrong. This also makes file merging harder.

Using a (GUID, path) combination is attractive in some ways, but it also adds a lot of complexity to the system. I really don't like adding complexity. I only want to do it when it is absolutely necessary. And the (GUID, path) combination doesn't feel like a perfect solution to me. It would also require us to come up with a new scheme for handling localization and platform specific resources. Currently we do that with extensions on the file name, so a reference to textures/flowers/rose may open textures/flowers/rose.fr.dds if you are using French localization. If we switched to GUIDs we would have to come up with a new system for this.

We already have a tool (the Dependency Checker) that understands references and can handle renames by patching references. So it seems to me that the best strategy going forward is to keep using paths as references and just add caching of reference information to the tool so that it is quicker to use.

Friday, April 25, 2014

Building an Engine Plugin System

A plugin system is a useful way to allow developers to extend the capabilities of a game engine. Of course, an engine can also be extended by directly modifying the source code, but there are several drawbacks with that approach:

  • Changing the code requires you to recompile the engine. Anyone who wants to modify the engine must have the full source code, access to all the libraries and the build environment set up correctly.

  • Every time you pull changes from upstream you will have to merge your changes with the incoming patches. Over time, this adds up to a significant chunk of work.

  • Since you work directly in the source code, instead of against a published API, refactoring of engine systems might force you to rewrite your code from scratch.

  • There is no easy way to share the modifications you have made with other people.

A plugin system solves all these issues. Plugins can be distributed as compiled DLLs. They are easily shared and you can install them by just putting them in the engine's plugin folder. Since the plugins use an explicit API, they will continue to work with new versions of the engine (unless backwards compatibility is explicitly broken).

Of course, the plugin API can never cover everything, so there will always be things you can do by modifying the engine that you can't do through the plugin API. Nevertheless, it is a good complement.

A Tale of Two APIs

When building a plugin system, there are actually two APIs that you need to think about.

The first, and most obvious one, is the API that the plugin exposes to the engine: a set of exported functions that the engine will call at predefined times. For a very basic system, it could look something like this:

__declspec(dllexport) void init();
__declspec(dllexport) void update(float dt);
__declspec(dllexport) void shutdown();

The other API, which usually is a lot more complicated, is the API that the engine exposes to the plugin.

In order to be useful, the plugin will want to call on the engine to do stuff. This can be things like spawning a unit, playing a sound, rendering some meshes, etc. The engine needs to provide some way for plugins to call on these services.

There are a number of ways of doing this. One common solution is to put all the shared functionality in a common DLL and then link both the engine application and the plugin against this DLL.

The drawback of this approach is that the more functionality that the plugins need access to, the more must go in the shared DLL. Eventually you end up with most of the engine in the shared DLL, which is pretty far from the clean and simple APIs that we strive for.

This creates a very strong coupling between the engine and the plugins. Every time we want to modify something in the engine, we will probably have to modify the shared DLL and thus likely break all of the plugins.

As anyone who has read my previous articles know I really don't like these kinds of strong couplings. They prevent you from rewriting and refactoring your systems and thus eventually cause your code to stagnate.

Another approach is to let the engine's scripting language (in our case Lua) work as the engine's API. With this approach, any time a plugin wanted the engine to do something it would use a Lua call.

For lots of applications I think this can be a really good solution, but in our case it doesn't seem like a perfect fit. First, the plugins will need access to a lot of stuff that is more "low level" than what you can access from Lua. And I'm not to keen on exposing all of the engine's innards to Lua. Second, since both the plugins and the engine are written in C++, marshalling all the calls between them through Lua seems both overly complicated and inefficient.

I prefer to have an interface that is minimalistic, data-oriented and C-based (because of C++ ABI compatibility issues and also because of... well... C++).

Interface Querying

Instead of linking the plugin against a DLL that provides the engine API. We can send the engine API to the plugin when we initialize it. Something like this (a simplified example):

plugin_api.h:

typedef struct EngineApi
{
 void (*spawn_unit)(World *world, const char *name, float pos[3]);
 ...
} EngineApi;

plugin.h

#include "plugin_api.h"

__declspec(dllexport) void init(EngineApi *api);
__declspec(dllexport) void update(float dt);
__declspec(dllexport) void shutdown();

This is pretty good. The plugin develeoper does not need to link against anything, just include the header file plugin_api.h, and then she can call the functions in the EngineApi struct to tell the engine to do stuff.

The only thing that is missing is versioning support.

At some point in the future we probably want to modify the EngineApi. Perhaps we discover that we want to add a rotation argument to spawn_unit() or somehting else. We can achieve this by introducing versioning in the system. Instead of sending the engine API directly to the plugin, we send the plugin a function that lets it query for a specific version of the engine API.

With this approach, we can also break the API up into smaller submodules that can be queried for individually. This gives us a cleaner organization.

plugin_api.h

#define WORLD_API_ID    0
#define LUA_API_ID      1

typedef struct World World;

typedef struct WorldApi_v0 {
 void (*spawn_unit)(World *world, const char *name, float pos[3]);
 ...
} WorldApi_v0;

typedef struct WorldApi_v1 {
 void (*spawn_unit)(World *world, const char *name, float pos[3], float rot[4]);
 ...
} WorldApi_v1;

typedef struct lua_State lua_State;
typedef int (*lua_CFunction) (lua_State *L);

typedef struct LuaApi_v0 {
 void (*add_module_function)(const char *module, const char *name, lua_CFunction f);
 ...
} LuaApi_v0;

typedef void *(*GetApiFunction)(unsigned api, unsigned version);

When the engine instances the plugin, it passes along get_engine_api(), which the plugin can use to get hold of different engine APIs.

The plugin will typically set up the APIs in the init() function:

static WorldApi_v1 *_world_api = nullptr;
static LuaApi_v0 *_lua_api = nullptr;

void init(GetApiFunction get_engine_api)
{
 _world_api = (WorldApi_v1 *)get_engine_api(WORLD_API, 1);
 _lua_api = (LuaApi_v0 *)get_engine_api(LUA_API, 0);
}

Later, the plugin case use these APIs:

_world_api->spawn_unit(world, "player", pos);

If we need to make a breaking change to an API, we can just introduce a new version of that API. As long as get_engine_api() can still return the old API version when requested for it, all existing plugins will continue to work.

With this querying system in place for the engine, it makes sense to use the same approach for the plugin as well. I.e. instead of exposing individual functions init(), update(), etc, the plugin just exposes a single function get_plugin_api() which the engine can use in the same way to query APIs from the plugin.

plugin_api.h

#define PLUGIN_API_ID 2

typedef struct PluginApi_v0
{
 void (*init)(GetApiFunction get_engine_api);
 ...
} PluginApi_v0;

plugin.c

__declspec(dllexport) void *get_plugin_api(unsigned api, unsigned version);

Since we now have versioning on the plugin API as well, this means we can modify it (add new required functions, etc) without breaking existing plugins.

Putting It All Together

Putting all this together, here is a complete (but very small) example of a plugin that exposes a new function to the Lua layer of the engine:

plugin_api.h

#define PLUGIN_API_ID       0
#define LUA_API_ID          1

typedef void *(*GetApiFunction)(unsigned api, unsigned version);

typedef struct PluginApi_v0
{
 void (*init)(GetApiFunction get_engine_api);
} PluginApi_v0;

typedef struct lua_State lua_State;
typedef int (*lua_CFunction) (lua_State *L);

typedef struct LuaApi_v0
{
 void (*add_module_function)(const char *module, const char *name, lua_CFunction f);
 double (*to_number)(lua_State *L, int idx);
 void (*push_number)(lua_State *L, double number);
} LuaApi_v0;

plugin.c

#include "plugin_api.h"

LuaApi_v0 *_lua;

static int test(lua_State *L)
{
 double a = _lua->to_number(L, 1);
 double b = _lua->to_number(L, 2);
 _lua->push_number(L, a+b);
 return 1;
}

static void init(GetApiFunction get_engine_api)
{
 _lua = get_engine_api(LUA_API_ID, 0);

 if (_lua)
  _lua->add_module_function("Plugin", "test", test);
}

__declspec(dllexport) void *get_plugin_api(unsigned api, unsigned version)
{
 if (api == PLUGIN_API_ID && version == 0) {
  static PluginApi_v0 api;
  api.init = init;
  return &api;
 }
 return 0;
}

engine.c

// Initialized elsewhere.
LuaEnvironment *_env = 0;

void add_module_function(const char *module, const char *name, lua_CFunction f)
{
 _env->add_module_function(module, name, f);
}

void *get_engine_api(unsigned api, unsigned version)
{
 if (api == LUA_API_ID && version == 0 && _env) {
  static LuaApi_v0 lua;
  lua.add_module_function = add_module_function;
  lua.to_number = lua_tonumber;
  lua.push_number = lua_pushnumber;
  return &lua;
 }
 return 0;
}

void load_plugin(const char *path)
{
 HMODULE plugin_module = LoadLibrary(path);
 if (!plugin_module) return;
 GetApiFunction get_plugin_api = (GetApiFunction)GetProcAddress(plugin_module, "get_plugin_api");
 if (!get_plugin_api) return;
 PluginApi_v0 *plugin = (PluginApi_v0 *)get_plugin_api(PLUGIN_API_ID, 0);
 if (!plugin) return;
 plugin->init(get_engine_api);
}