path: root/doc/api.texi

@menu
* Kernel::                      The fundamental components.
* Math::                        Linear algebra and more.
* Graphics::                    Eye candy.
* Scripting::                   Bringing the game world to life.
@end menu

@node Kernel
@section Kernel

At the very core of Chickadee, in the @code{(chickadee)} module, lies
an event loop.  This loop, or ``kernel'', is responsible for ensuring
that the game is updated at the desired interval, rendering the
current state of the game world, and handling errors if they occur.
The kernel implements what is known as a ``fixed timestep'' game loop,
meaning that the game simulation will be advanced by a fixed interval
of time and will never vary from frame to frame, unlike some other
styles of game loops.  The appropriately named @code{run-game} and
@code{abort-game} procedures are the entry and exit points to the
Chickadee game loop kernel.

On its own, the kernel does not do very much at all.  In order to
actually respond to input events, update game state, or render output,
the programmer must provide an engine.  But don’t worry, you don’t
have to start from scratch!  Chickadee comes with a simple engine that
uses SDL to create a graphical window and handle input devices, and
OpenGL to handle rendering.  This default engine is enough for most
users to get started writing games quickly.  More advanced users may
want to write a custom engine that uses a different I/O system.
Perhaps you are writing a text adventure or roguelike that reads from
and writes to a terminal instead of a graphical window.  The game loop
kernel makes no assumptions.

@deffn {Procedure} run-game [#:update] [#:render] [#:time] [#:error] @
       [#:update-hz 60]

Start the game loop.  This procedure will not return until
@code{abort-game} is called.

The core game loop is generic and requires four additional procedures
to operate:

@itemize
@item
@var{update}: Called @var{update-hz} times per second to advance the
game simulation.  This procedure is called with a single argument: The
amount of time that has passed since the last update, in milliseconds.
@item
@var{render}: Called each iteration of the loop to render the game to
the desired output device.  This procedure is called with a single
argument: A value in the range [0, 1] which represents how much time
has past since the last game state update relative to the upcoming
game state update, as a percentage.  Because the game state is updated
independent of rendering, it is often the case that rendering is
occuring between two updates.  If the game is rendered as it was
during the last update, a strange side-effect will occur that makes
animation appear rough or ``choppy''.  To counter this, the
@var{alpha} value can be used to perfrom a linear interpolation of a
moving object between its current position and its previous position.
This odd trick has the pleasing result of making the animation look
smooth again, but requires keeping track of previous state.
@item
@var{time}: Called to get the current time in milliseconds.  This
procedure is called with no arguments.
@item
@var{error}: Called when an error from the @var{update} or
@var{render} procedures reaches the game loop.  This procedure is
called with three arguments: The call stack, the error key, and the
error arguments.  If no error handler is provided, the default
behavior is to simply re-throw the error.
@end itemize

@end deffn

@deffn {Procedure} abort-game
Stop the currently running Chickadee game loop.
@end deffn

Since most users will want to write 2D/3D games with hardware
accelerated graphics rendering, controlled via keyboard, mouse, or
game controller, Chickadee comes with an easy to use engine just for
this purpose in the @code{(chickadee sdl)} module:
@code{run-game/sdl}.

@deffn {Procedure} run-game/sdl [#:window-title "Chickadee!"] @
       [#:window-width 640] [#:window-height 480] @
       [#:window-fullscreen? @code{#f}] [#:update-hz 60] @
       [#:load] [#:update] [#:draw] [#:quit] @
       [#:key-press] [#:key-release] [#:text-input] @
       [#:mouse-press] [#:mouse-release] [#:mouse-move] @
       [#:controller-add] [#:controller-remove] [#:controller-press] @
       [#:controller-release] [#:controller-move] [#:error]

Run the Chickadee game loop using the SDL engine.

A new graphical window will be opened with @var{window-width} x
@var{window-height} as its dimensions, @var{window-title} as its
title, and in fullscreen mode if @var{window-fullscreen?} is
@code{#t}.

@itemize
@item
@var{load}: Called with zero arguments when the game window has opened
but before the game loop has started.  Can be used to perform
initialization that requires an open window and OpenGL context such as
loading textures.

@item
@var{update}: Called @var{update-hz} times per second with one
argument: The amount of time to advance the game simulation.

@item
@var{draw}: Called each time a frame should be rendered with a single
argument known as the @code{alpha} value.  See the documentation for
@code{run-game} for an explanation of this value.

@item
@var{quit}: Called with zero arguments when the user tries to close
the game window.  The default behavior is to exit the game.

@item
@var{key-press}: Called with four arguments when a key is pressed on
the keyboard:

@enumerate
@item
@var{key}: The symbolic name of the ``virtual'' key that was pressed.
For example: @code{backspace}.  It's called a virtual key because the
operating system may map a physical keyboard key to another key
entirely, such as how the author likes to bind the ``caps lock'' key
to mean ``control''.

@item
@var{scancode}: The symbolic name of the physical key that was
pressed.

@item
@var{modifiers}: A list of the symbolic names of modifier keys that
were being held down when the key was pressed.  Possible values
include @code{ctrl}, @code{alt}, and @code{shift}.

@item
@var{repeat?}: @code{#t} if this is a repeated press of the same key.

@end enumerate

@item
@var{key-release}: Called with three arguments when a key is released
on the keyboard:

@enumerate
@item
@var{key}: The symbolic name of the ``virtual'' key that was released.

@item
@var{scancode}: The symbolic name of the physical key that was
released.

@item
@var{modifiers}: A list of the symbolic names of modifier keys that
were being held down when the key was released.

@end enumerate

@item
@var{text-input}: Called with a single argument, a string of text,
when printable text is typed on the keyboard.

@item
@var{mouse-press}: Called with four arguments when a mouse button is
pressed:
@enumerate

@item
@var{button}: The symbolic name of the button that was pressed, such
as @code{left}, @code{middle}, or @code{right}.

@item
@var{clicks}: The number of times the button has been clicked in a row.

@item
@var{x}: The x coordinate of the mouse cursor.

@item
@var{y}: The y coordinate of the mouse cursor.

@end enumerate

@item
@var{mouse-release}: Called with three arguments when a mouse button
is released:

@enumerate

@item
@var{button}: The symbolic name of the button that was released.

@item
@var{x}: The x coordinate of the mouse cursor.

@item
@var{y}: The y coordinate of the mouse cursor.

@end enumerate

@item
@var{mouse-move}: Called with five arguments when the mouse is moved:

@enumerate

@item
@var{x}: The x coordinate of the mouse cursor.

@item
@var{y}: The y coordinate of the mouse cursor.

@item
@var{dx}: The amount the mouse has moved along the x axis since the
last mouse move event.

@item
@var{dy}: The amount the mouse has moved along the y axis since the
last mouse move event.

@item
@var{buttons}: A list of the buttons that were pressed down when the
mouse was moved.

@end enumerate

@item
@var{controller-add}: Called with a single argument, an SDL game
controller object, when a game controller is connected.

@item
@var{controller-remove}: Called with a single argument, an SDL game
controller object, when a game controller is disconnected.

@item
@var{controller-press}: Called with two arguments when a button on a
game controller is pressed:

@enumerate

@item
@var{controller}: The controller that triggered the event.

@item
@var{button}: The symbolic name of the button that was pressed.
Possible buttons are:

@itemize
@item
@code{a}
@item
@code{b}
@item
@code{x}
@item
@code{y}
@item
@code{back}
@item
@code{guide}
@item
@code{start}
@item
@code{left-stick}
@item
@code{right-stick}
@item
@code{left-shoulder}
@item
@code{right-shoulder}
@item
@code{dpad-up}
@item
@code{dpad-down}
@item
@code{dpad-left}
@item
@code{dpad-right}

@end itemize

@end enumerate

@item
@var{controller-release}: Called with two arguments when a button on a
game controller is released:

@enumerate

@item
@var{controller}: The controller that triggered the event.

@item
@var{button}: The symbolic name of the button that was released.

@end enumerate

@item
@var{controller-move}: Called with three arguments when an analog
stick or trigger on a game controller is moved:

@enumerate

@item
@var{controller}: The controller that triggered the event.

@item
@var{axis}: The symbolic name of the axis that was moved.  Possible
values are:

@itemize
@item
@code{left-x}
@item
@code{left-y}
@item
@code{right-x}
@item
@code{right-y}
@item
@code{trigger-left}
@item
@code{trigger-right}
@end itemize

@end enumerate

@item
@var{error}: Called with three arguments when an error occurs:

@enumerate

@item
@var{stack}: The call stack at the point of error.

@item
@var{key}: The exception key.

@item
@var{args}: The arguments thrown with the exception.

@end enumerate

The default behavior is to re-throw the error.

@end itemize

@end deffn

@node Math
@section Math

Chickadee contains data types and procedures for performing the most
common computations in video game simulations such as linear algebra
with vectors and matrices and axis-aligned bounding box collision
detection.

@menu
* Basics::                      Commonly used, miscellaneous things.
* Vectors::                     Euclidean vectors.
* Rectangles::                  Axis-aligned bounding boxes.
* Grid::                        Spatial partitioning for bounding boxes.
* Matrices::                    Transformation matrices.
* Quaternions::                 Rotations about an arbitrary axis.
* Easings::                     Easing functions for interesting animations.
* Bezier Curves::               Cubic Bezier curves and paths in 2D space.
@end menu

@node Basics
@subsection Basics

@defvar pi
An essential constant for all trigonometry.  @code{@U{03C0}} is the ratio
of a circle's circumferences to its diameter.  Since @code{@U{03C0}} is an
irrational number, the @var{pi} in Chickadee is a mere floating point
approximation that is ``good enough.''
@end defvar

@defvar pi/2
Half of @var{pi}.
@end defvar

@deffn {Procedure} cotan @var{z}
Return the cotangent of @var{z}.
@end deffn

@node Vectors
@subsection Vectors

Unlike Scheme's vector data type, which is a sequence of arbitrary
Scheme objects, Chickadee's @code{(chickadee math vector)} module
provides vectors in the linear algebra sense: Sequences of numbers
specialized for particular coordinate spaces.  As of now, Chickadee
provides 2D and 3D vectors, with 4D vector support coming in a future
release.

Here's a quick example of adding two vectors:

@example
(define v (vec2+ (vec2 1 2) (vec2 3 4)))
@end example

Since vectors are used so frequently, the reader macro @code{#v} is
used to cut down on typing:

@example
(define v (vec2+ #v(1 2) #v(3 4)))
@end example

@subsubsection A Note About Performance

A lot of time has been spent making Chickadee's vector operations
perform relatively efficiently in critical code paths where excessive
garbage generation will cause major performance issues.  The general
rule is that procedures ending with @code{!} perform an in-place
modification of one of the arguments in order to avoid allocating a
new vector.  These procedures are also inlined by Guile's compiler in
order to take advantage of optimizations relating to floating point
math operations.  The downside is that since these are not pure
functions, they do not compose well and create more verbose code.

@subsubsection 2D Vectors

@deffn {Procedure} vec2 @var{x} @var{y}
Return a new 2D vector with coordinates (@var{x}, @var{y}).
@end deffn

@deffn {Procedure} vec2/polar @var{r} @var{theta}
Return a new 2D vector containing the Cartesian representation of the
polar coordinate (@var{r}, @var{theta}).  The angle @var{theta} is
measured in radians.
@end deffn

@deffn {Procedure} vec2? @var{obj}
Return @code{#t} if @var{obj} is a 2D vector.
@end deffn

@deffn {Procedure} vec2-x @var{v}
Return the X coordinate of the 2D vector @var{v}.
@end deffn

@deffn {Procedure} vec2-y @var{v}
Return the Y coordinate of the 2D vector @var{v}.
@end deffn

@deffn {Procedure} vec2-copy @var{v}
Return a fresh copy of the 2D vector @var{v}.
@end deffn

@deffn {Procedure} vec2-magnitude @var{v}
Return the magnitude of the 2D vector @var{v}.
@end deffn

@deffn {Procedure} vec2-dot-product @var{v1} @var{v2}
Return the dot product of the 2D vectors @var{v1} and @var{v2}.
@end deffn

@deffn {Procedure} vec2-normalize @var{v}
Return the normalized form of the 2D vector @var{v}.
@end deffn

@deffn {Procedure} vec2+ @var{v} @var{x}
Add @var{x}, either a 2D vector or a scalar (i.e. a real number), to
the 2D vector @var{v} and return a new vector containing the sum.
@end deffn

@deffn {Procedure} vec2- @var{v} @var{x}
Subtract @var{x}, either a 2D vector or a scalar, from the 2D vector
@var{v} and return a new vector containing the difference.
@end deffn

@deffn {Procedure} vec2* @var{v} @var{x}
Multiply the 2D vector @var{v} by @var{x}, a 2D vector or a scalar,
and return a new vector containing the product.
@end deffn

@deffn {Procedure} set-vec2-x! @var{v} @var{x}
Set the X coordinate of the 2D vector @var{v} to @var{x}.
@end deffn

@deffn {Procedure} set-vec2-y! @var{v} @var{y}
Set the Y coordinate of the 2D vector @var{v} to @var{y}.
@end deffn

@deffn {Procedure} vec2-copy! @var{source} @var{target}
Copy the 2D vector @var{source} into the 2D vector @var{target}.
@end deffn

@deffn {Procedure} vec2-add! @var{v} @var{x}
Perform an in-place modification of the 2D vector @var{v} by adding
@var{x}, a 2D vector or a scalar.
@end deffn

@deffn {Procedure} vec2-sub! @var{v} @var{x}
Perform an in-place modification of the 2D vector @var{v} by
subtracting @var{x}, a 2D vector or a scalar.
@end deffn

@deffn {Procedure} vec2-mult! @var{v} @var{x}
Perform an in-place modification of the 2D vector @var{v} by
multiplying it by @var{x}, a 2D vector or a scalar.
@end deffn

@subsubsection 3D Vectors

@deffn {Procedure} vec3 @var{x} @var{y}
Return a new 2D vector with coordinates (@var{x}, @var{y}).
@end deffn

@deffn {Procedure} vec3? @var{obj}
Return @code{#t} if @var{obj} is a 3D vector.
@end deffn

@deffn {Procedure} vec3-x @var{v}
Return the X coordinate of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3-y @var{v}
Return the Y coordinate of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3-z @var{v}
Return the Z coordinate of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3-copy @var{v}
Return a fresh copy of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3-magnitude @var{v}
Return the magnitude of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3-dot-product @var{v1} @var{v2}
Return the dot product of the 3D vectors @var{v1} and @var{v2}.
@end deffn

@deffn {Procedure} vec3-normalize @var{v}
Return the normalized form of the 3D vector @var{v}.
@end deffn

@deffn {Procedure} vec3+ @var{v} @var{x}
Add @var{x}, either a 3D vector or a scalar (i.e. a real number), to
the 3D vector @var{v} and return a new vector containing the sum.
@end deffn

@deffn {Procedure} vec3- @var{v} @var{x}
Subtract @var{x}, either a 3D vector or a scalar, from the 3D vector
@var{v} and return a new vector containing the difference.
@end deffn

@deffn {Procedure} vec3* @var{v} @var{x}
Multiply the 3D vector @var{v} by @var{x}, a 3D vector or a scalar,
and return a new vector containing the product.
@end deffn

@deffn {Procedure} set-vec3-x! @var{v} @var{x}
Set the X coordinate of the 3D vector @var{v} to @var{x}.
@end deffn

@deffn {Procedure} set-vec3-y! @var{v} @var{y}
Set the Y coordinate of the 3D vector @var{v} to @var{y}.
@end deffn

@deffn {Procedure} set-vec3-z! @var{v} @var{z}
Set the Z coordinate of the 3D vector @var{v} to @var{z}.
@end deffn

@deffn {Procedure} vec3-copy! @var{source} @var{target}
Copy the 3D vector @var{source} into the 3D vector @var{target}.
@end deffn

@deffn {Procedure} vec3-add! @var{v} @var{x}
Perform an in-place modification of the 3D vector @var{v} by adding
@var{x}, a 3D vector or a scalar.
@end deffn

@deffn {Procedure} vec3-sub! @var{v} @var{x}
Perform an in-place modification of the 3D vector @var{v} by
subtracting @var{x}, a 3D vector or a scalar.
@end deffn

@deffn {Procedure} vec3-mult! @var{v} @var{x}
Perform an in-place modification of the 3D vector @var{v} by
multiplying it by @var{x}, a 3D vector or a scalar.
@end deffn

@node Rectangles
@subsection Rectangles

The @code{(chickadee math rect)} module provides an API for
manipulating axis-aligned bounding boxes (AABBs).  AABBs are often
used for collision detection in games.  Common use-cases are defining
``hitboxes'' in platformers or using them for the ``broad phase'' of a
collision detection algorithm that uses a more complex (and thus
slower) method of determining the actual collisions.

Like some of the other math modules, there exists a collection of
functions that do in-place modification of rectangles for use in
performance critical code paths.

@deffn {Procedure} make-rect @var{x} @var{y} @var{width} @var{height}
Create a new rectangle that is @var{width} by @var{height} in size and
whose bottom-left corner is located at (@var{x}, @var{y}).
@end deffn

@deffn {Procedure} rect? @var{obj}
Return @code{#t} if @var{obj} is a rectangle.
@end deffn

@deffn {Procedure} rect-within? @var{rect1} @var{rect2}
Return @code{#t} if @var{rect2} is completely within @var{rect1}.
@end deffn

@deffn {Procedure} rect-intersects? @var{rect1} @var{rect2}
Return @code{#t} if @var{rect2} overlaps @var{rect1}.
@end deffn

@deffn {Procedure} rect-contains? @var{rect} @var{x} @var{y}
Return @code{#t} if the coordinates (@var{x}, @var{y}) are within
@var{rect}.
@end deffn

@deffn {Procedure} rect-contains-vec2? @var{rect} @var{v}
Return @code{#t} if the 2D vector @var{v} is within the bounds of
@var{rect}.
@end deffn

@deffn {Procedure} rect-x @var{rect}
Return the X coordinate of the lower-left corner of @var{rect}.
@end deffn

@deffn {Procedure} rect-y @var{rect}
Return the Y coordinate of the lower-left corner of @var{rect}.
@end deffn

@deffn {Procedure} rect-left @var{rect}
Return the left-most X coordinate of @var{rect}.
@end deffn

@deffn {Procedure} rect-right @var{rect}
Return the right-most X coordinate of @var{rect}.
@end deffn

@deffn {Procedure} rect-bottom @var{rect}
Return the bottom-most Y coordinate of @var{rect}.
@end deffn

@deffn {Procedure} rect-top @var{rect}
Return the top-most Y coordinate of @var{rect}.
@end deffn

@deffn {Procedure} rect-center-x @var{rect}
Return the X coordinate of the center of @var{rect}.
@end deffn

@deffn {Procedure} rect-center-y @var{rect}
Return the Y coordinate of the center of @var{rect}.
@end deffn

@deffn {Procedure} rect-width @var{rect}
Return the width of @var{rect}.
@end deffn

@deffn {Procedure} rect-height @var{rect}
Return the height of @var{rect}.
@end deffn

@deffn {Procedure} rect-area @var{rect}
Return the surface area covered by @var{rect}.
@end deffn

@deffn {Procedure} rect-clamp-x @var{rect} @var{x}
Restrict @var{x} to the portion of the X axis covered by @var{rect}.
@end deffn

@deffn {Procedure} rect-clamp-y @var{rect} @var{y}
Restrict @var{y} to the portion of the Y axis covered by @var{rect}.
@end deffn

@deffn {Procedure} rect-clamp @var{rect1} @var{rect2}
Return a new rect that adjusts the location of @var{rect1} so that it
is completely within @var{rect2}.  An exception is thrown in the case
that @var{rect1} cannot fit completely within @var{rect2}.
@end deffn

@deffn {Procedure} rect-move @var{rect} @var{x} @var{y}
Return a new rectangle based on @var{rect} but moved to the
coordinates (@var{x}, @var{y}).
@end deffn

@deffn {Procedure} rect-move-vec2 @var{rect} @var{v}
Return a new rectangle based on @var{rect} but moved to the
coordinates in the 2D vector @var{v}.
@end deffn

@deffn {Procedure} rect-move-by @var{rect} @var{x} @var{y}
Return a new rectangle based on @var{rect} but moved by (@var{x},
@var{y}) units relative to its current location.
@end deffn

@deffn {Procedure} rect-move-by-vec2 @var{rect} @var{v}
Return a new rectangle based on @var{rect} but moved by the 2D vector
@var{v} relative to its current location.
@end deffn

@deffn {Procedure} rect-inflate @var{rect} @var{width} @var{height}
Return a new rectangle based on @var{rect}, but expanded by
@var{width} units on the X axis and @var{height} units on the Y axis,
while keeping the rectangle centered on the same point.
@end deffn

@deffn {Procedure} rect-union @var{rect1} @var{rect2}
Return a new rectangle that completely covers the area of @var{rect1}
and @var{rect2}.
@end deffn

@deffn {Procedure} rect-clip @var{rect1} @var{rect2}
Return a new rectangle that is the overlapping region of @var{rect1}
and @var{rect2}.  If the two rectangles do not overlap, a rectangle of
0 width and 0 height is returned.
@end deffn

@deffn {Procedure} set-rect-x! @var{rect} @var{x}
Set the left X coordinate of @var{rect} to @var{x}.
@end deffn

@deffn {Procedure} set-rect-y! @var{rect} @var{y}
Set the bottom Y coordinate of @var{rect} to @var{y}.
@end deffn

@deffn {Procedure} set-rect-width! @var{rect} @var{width}
Set the width of @var{rect} to @var{width}.
@end deffn

@deffn {Procedure} set-rect-height! @var{rect} @var{height}
Set the height of @var{rect} to @var{height}.
@end deffn

@deffn {Procedure} rect-move! @var{rect} @var{x} @var{y}
Move @var{rect} to (@var{x}, @var{y}) in-place.
@end deffn

@deffn {Procedure} rect-move-vec2! @var{rect} @var{v}
Move @var{rect} to the 2D vector @var{v} in-place.
@end deffn

@deffn {Procedure} rect-move-by! @var{rect} @var{x} @var{y}
Move @var{rect} by (@var{x}, @var{y}) in-place.
@end deffn

@deffn {Procedure} rect-move-by-vec2! @var{rect} @var{v}
Move @var{rect} by the 2D vector @var{v} in-place.
@end deffn

@deffn {Procedure} rect-inflate! @var{rect} @var{width} @var{height}
Expand @var{rect} by @var{width} and @var{height} in-place.
@end deffn

@deffn {Procedure} rect-union! @var{rect1} @var{rect2}
Modify @var{rect1} in-place to completely cover the area of both
@var{rect1} and @var{rect2}.
@end deffn

@deffn {Procedure} rect-clip! @var{rect1} @var{rect2}
Modify @var{rect1} in-place to be the overlapping region of
@var{rect1} and @var{rect2}.
@end deffn

@deffn {Procedure} rect-clamp! @var{rect1} @var{rect2}
Adjust the location of @var{rect1} in-place so that its bounds are
completely within @var{rect2}.  An exception is thrown in the case
that @var{rect1} cannot fit completely within @var{rect2}.
@end deffn

@deffn {Procedure} vec2-clamp-to-rect! @var{v} @var{rect}
Restrict the coordinates of the 2D vector @var{v} so that they are
within the bounds of @var{rect}.  @var{v} is modified in-place.
@end deffn

@node Grid
@subsection Grid

The @code{(chickadee math grid)} module provides a simple spatial
partitioning system for axis-aligned bounding boxes
(@pxref{Rectangles}) in 2D space.  The grid divides the world into
tiles and keeps track of which rectangles occupy which tiles.  When
there are lots of moving objects in the game world that need collision
detection, the grid greatly speeds up the process.  Instead of
checking collisions of each object against every other object (an
O(n^2) operation), the grid quickly narrows down which objects could
possibly be colliding and only performs collision testing against a
small set of objects.

In addition to checking for collisions, the grid also handles the
resolution of collisions.  Exactly how each collision is resolved is
user-defined.  A player bumping into a wall may slide against it.  An
enemy colliding with a projectile shot by the player may get pushed
back in the opposite direction.  Two players colliding may not need
resolution at all and will just pass through each other.  The way this
works is that each time an object (A) is moved within the grid, the
grid looks for an object (B) that may possibly be colliding with A.  A
user-defined procedure known as a ``filter'' is then called with both
A and B.  If the filter returns @code{#f}, it means that even if A and
B are colliding, no collision resolution is needed.  In this case the
grid won't waste time checking if they really do collide because it
doesn't matter.  If A and B are collidable, then the filter returns a
procedure that implements the resolution technique.  The grid will
then perform a collision test.  If A and B are colliding, the resolver
procedure is called.  It's the resolvers job to adjust the objects
such that they are no longer colliding.  The grid module comes with a
very simple resolution procedure, @code{slide}, that adjusts object A
by the smallest amount so that it no longer overlaps with B.  By using
this filtering technique, a game can resolve collisions between
different objects in different ways.

@deffn {Procedure} make-grid [@var{cell-size} 64]
Return a new grid partitioned into @var{cell-size} tiles.
@end deffn

@deffn {Procedure} grid? @var{obj}
Return @code{#t} if @var{obj} is a grid.
@end deffn

@deffn {Procedure} cell? @var{obj}
Return @code{#t} if @var{obj} is a grid cell.
@end deffn

@deffn {Procedure} cell-count @var{cell}
Return the number of items in @var{cell}.
@end deffn

@deffn {Procedure} grid-cell-size @var{grid}
Return the cell size of @var{grid}.
@end deffn

@deffn {Procedure} grid-cell-count @var{grid}
Return the number of cells currently in @var{grid}.
@end deffn

@deffn {Procedure} grid-item-count @var{grid}
Return the number of items in @var{grid}.
@end deffn

@deffn {Procedure} grid-add @var{grid} @var{item} @var{x} @var{y} @
                            @var{width} @var{height}

Add @var{item} to @var{grid} represented by the axis-aligned bounding
box whose lower-left corner is at (@var{x}, @var{y}) and is
@var{width} x @var{height} in size.
@end deffn

@deffn {Procedure} grid-remove @var{grid} @var{item}
Return @var{item} from @var{grid}.
@end deffn

@deffn {Procedure} grid-clear @var{grid}
Remove all items from @var{grid}.
@end deffn

@deffn {Procedure} grid-move @var{grid} @var{item} @var{position} @var{filter}
Attempt to move @var{item} in @var{grid} to @var{position} (a 2D
vector) and check for collisions.  For each collision, @var{filter}
will be called with two arguments: @var{item} and the item it collided
with.  If a collision occurs, @var{position} may be modified to
resolve the colliding objects.
@end deffn

@deffn {Procedure} for-each-cell @var{proc} @var{grid} [@var{rect}]
Call @var{proc} with each cell in @var{grid} that intersects
@var{rect}, or every cell if @var{rect} is @code{#f}.
@end deffn

@deffn {Procedure} for-each-item @var{proc} @var{grid}
Call @var{proc} for each item in @var{grid}.
@end deffn

@deffn {Procedure} slide @var{item} @var{item-rect} @
                         @var{other} @var{other-rect} @var{goal}

Resolve the collision that occurs between @var{item} and @var{other}
when moving @var{item-rect} to @var{goal} by sliding @var{item-rect}
the minimum amount needed to make it no longer overlap
@var{other-rect}.
@end deffn

@node Matrices
@subsection Matrices

The @code{(chickadee math matrix)} module provides an interface for
working with the most common type of matrices in game development: 4x4
transformation matrices.

@subsubsection Another Note About Performance

Much like the vector API, the matrix API is commonly used in
performance critical code paths.  In order to reduce the amount of
garbage generated and improve matrix multiplication performance, there
are many procedures that perform in-place modifications of matrix
objects.

@subsubsection Matrix Operations

@deffn {Procedure} make-matrix4 @var{aa} @var{ab} @var{ac} @var{ad} @
                                @var{ba} @var{bb} @var{bc} @var{bd} @
                                @var{ca} @var{cb} @var{cc} @var{cd} @
                                @var{da} @var{db} @var{dc} @var{dd}

Return a new 4x4 matrix initialized with the given 16 values in
column-major format.
@end deffn

@deffn {Procedure} make-null-matrix4
Return a new 4x4 matrix with all values initialized to 0.
@end deffn

@deffn {Procedure} make-identity-matrix4
Return a new 4x4 identity matrix.  Any matrix multiplied by the
identity matrix yields the original matrix.  This procedure is
equivalent to the following code:

@example
(make-matrix4 1 0 0 0
              0 1 0 0
              0 0 1 0
              0 0 0 1)
@end example

@end deffn

@deffn {Procedure} matrix4? @var{obj}
Return @code{#t} if @var{obj} is a 4x4 matrix.
@end deffn

@deffn {Procedure} matrix4* . @var{matrices}
Return a new 4x4 matrix containing the product of multiplying all of
the given @var{matrices}.

Note: Remember that matrix multiplication is @strong{not} commutative!
@end deffn

@deffn {Procedure} orthographic-projection @var{left} @var{right} @
                                           @var{top} @var{bottom} @
                                           @var{near} @var{far}

Return a new 4x4 matrix that represents an orthographic (2D)
projection for the horizontal clipping plane @var{top} and
@var{bottom}, the vertical clipping plane @var{top} and @var{bottom},
and the depth clipping plane @var{near} and @var{far}.
@end deffn

@deffn {Procedure} perspective-projection @var{fov} @
                                          @var{aspect-ratio} @
                                          @var{near} @var{far}

Return a new 4x4 matrix that represents a perspective (3D) projection
with a field of vision of @var{fov} radians, an aspect ratio of
@var{aspect-ratio}, and a depth clipping plane defined by @var{near}
and @var{far}.
@end deffn

@deffn {Procedure} matrix4-translate @var{x}
Return a new 4x4 matrix that represents a translation by @var{x}, a 2D
vector, a 3D vector, or a rectangle (in which case the bottom-left
corner of the rectangle is used).
@end deffn

@deffn {Procedure} matrix4-scale @var{s}
Return a new 4x4 matrix that represents a scaling along the X, Y, and
Z axes by the scaling factor @var{s}, a real number.
@end deffn

@deffn {Procedure} matrix4-rotate @var{q}
Return a new 4x4 matrix that represents a rotation about an arbitrary
axis defined by the quaternion @var{q}.
@end deffn

@deffn {Procedure} matrix4-rotate-z @var{theta}
Return a new 4x4 matrix that represents a rotation about the Z axis by
@var{theta} radians.
@end deffn

@deffn {Procedure} matrix4-identity! @var{matrix}
Modify @var{matrix} in-place to contain the identity matrix.
@end deffn

@deffn {Procedure} matrix4-mult! @var{dest} @var{a} @var{b}
Multiply the 4x4 matrix @var{a} by the 4x4 matrix @var{b} and store
the result in the 4x4 matrix @var{dest}.
@end deffn

@deffn {Procedure} matrix4-translate! @var{matrix} @var{x}
Modify @var{matrix} in-place to contain a translation by @var{x}, a 2D
vector, a 3D vector, or a rectangle (in which case the bottom-left
corner of the rectangle is used).
@end deffn

@deffn {Procedure} matrix4-scale! @var{matrix} @var{s}
Modify @var{matrix} in-place to contain a scaling along the X, Y, and
Z axes by the scaling factor @var{s}, a real number.
@end deffn

@deffn {Procedure} matrix4-rotate! @var{matrix} @var{q}
Modify @var{matrix} in-place to contain a rotation about an arbitrary
axis defined by the quaternion @var{q}.
@end deffn

@deffn {Procedure} matrix4-rotate-z! @var{matrix} @var{theta}
Modify @var{matrix} in-place to contain a rotation about the Z axis by
@var{theta} radians.
@end deffn

@deffn {Procedure} matrix4-2d-transform! @var{matrix} [#:origin] @
                                         [#:position] [#:rotation] @
                                         [#:scale] [#:skew]

Modify @var{matrix} in-place to contain the transformation described
by @var{position}, a 2D vector or rectangle, @var{rotation}, a scalar
representing a rotation about the Z axis, @var{scale}, a 2D vector,
and @var{skew}, a 2D vector.  The transformation happens with respect
to @var{origin}, a 2D vector.  If an argument is not provided, that
particular transformation will not be included in the result.
@end deffn

@deffn {Procedure} transform! @var{matrix} @var{v}
Modify the 2D vector @var{v} in-place by multiplying it by the 4x4
matrix @var{matrix}.
@end deffn

@node Quaternions
@subsection Quaternions

In game development, the quaternion is most often used to represent
rotations.  Why not use a matrix for that, you may ask.  Unlike
matrices, quaternions can be interpolated (animated) and produce a
meaningful result.  When interpolating two quaternions, there is a
smooth transition from one rotation to another, whereas interpolating
two matrices would yield garbage.

@deffn {Procedure} quaternion @var{x} @var{y} @var{z} @var{w}
Return a new quaternion with values @var{x}, @var{y}, @var{z}, and
@var{w}.
@end deffn

@deffn {Procedure} quaternion? @var{obj}
Return @code{#t} if @var{obj} is a quaternion.
@end deffn

@deffn {Procedure} quaternion-w @var{q}
Return the W component of the quaternion @var{q}.
@end deffn

@deffn {Procedure} quaternion-x @var{q}
Return the X component of the quaternion @var{q}.
@end deffn

@deffn {Procedure} quaternion-y @var{q}
Return the Y component of the quaternion @var{q}.
@end deffn

@deffn {Procedure} quaternion-z @var{q}
Return the Z component of the quaternion @var{q}.
@end deffn

@deffn {Procedure} make-identity-quaternion
Return the identity quaternion.
@end deffn

@node Easings
@subsection Easings

@deffn {Procedure} linear @var{t}
@end deffn

@deffn {Procedure} smoothstep @var{t}
@end deffn

@deffn {Procedure} ease-in-quad @var{t}
@end deffn

@deffn {Procedure} ease-out-quad @var{t}
@end deffn

@deffn {Procedure} ease-in-out-quad @var{t}
@end deffn

@deffn {Procedure} ease-in-cubic @var{t}
@end deffn

@deffn {Procedure} ease-out-cubic @var{t}
@end deffn

@deffn {Procedure} ease-in-out-cubic @var{t}
@end deffn

@deffn {Procedure} ease-in-quart @var{t}
@end deffn

@deffn {Procedure} ease-out-quart @var{t}
@end deffn

@deffn {Procedure} ease-in-out-quart @var{t}
@end deffn

@deffn {Procedure} ease-in-quint @var{t}
@end deffn

@deffn {Procedure} ease-out-quint @var{t}
@end deffn

@deffn {Procedure} ease-in-out-quint @var{t}
@end deffn

@deffn {Procedure} ease-in-sine @var{t}
@end deffn

@deffn {Procedure} ease-out-sine @var{t}
@end deffn

@deffn {Procedure} ease-in-out-sine @var{t}
@end deffn

@node Bezier Curves
@subsection Bezier Curves

The @code{(chickadee math bezier)} module provides an API for
describing cubic Bezier curves in 2D space.  These curves are notably
used in font description, vector graphics programs, and when it comes
to games: path building.  With Bezier curves, it's somewhat easy to
create a smooth looking path for an enemy to move along, for example.
Bezier curves become particularly interesting when they are chained
together to form a Bezier ``path'', where the end point of one curve
becomes the starting point of the next.

Currently, the rendering of Bezier curves is rather crude and provided
mostly for visualizing and debugging curves that would be unseen in
the final game.  See @xref{Lines and Shapes} for more information.

@deffn {Procedure} make-bezier-curve @var{p0} @var{p1} @var{p2} @var{p3}
Return a new Bezier curve object whose starting point is @var{p0},
ending point is @var{p3}, and control points are @var{p1} and
@var{p2}.  All points are 2D vectors.
@end deffn

@deffn {Procedure} bezier-curve? @var{obj}
Return @code{#t} if @var{obj} is a Bezier curve.
@end deffn

@deffn {Procedure} bezier-curve-p0 @var{bezier}
Return the starting point of @var{bezier}.
@end deffn

@deffn {Procedure} bezier-curve-p1 @var{bezier}
Return the first control point of @var{bezier}.
@end deffn

@deffn {Procedure} bezier-curve-p2 @var{bezier}
Return the second control point of @var{bezier}.
@end deffn

@deffn {Procedure} bezier-curve-p3 @var{bezier}
Return the end point of @var{bezier}.
@end deffn

@deffn {Procedure} bezier-path . @var{control-points}
Return a list of connected bezier curves defined by
@var{control-points}.  The first curve is defined by the first 4
arguments and every additional curve thereafter requires 3 additional
arguments.
@end deffn

@deffn {Procedure} bezier-curve-point-at @var{bezier} @var{t}
Return the coordinates for @var{bezier} at @var{t} (a value in the
range [0, 1] representing how far from the start of the curve to
check) as a 2D vector.
@end deffn

@deffn {Procedure} bezier-curve-point-at! @var{dest} @var{bezier} @var{t}
Modify the 2D vector @var{dest} in-place to contain the coordinates
for @var{bezier} at @var{t}.
@end deffn

@node Graphics
@section Graphics

Chickadee aims to make hardware-accelerated graphics rendering as
simple and efficient as possible by providing high-level APIs that
interact with the low-level OpenGL API under the hood.  Anyone that
has worked with OpenGL directly knows that it has a steep learning
curve and a lot of effort is needed to render even a single triangle.
The Chickadee rendering engine attempts to make it easy to do common
tasks like rendering a sprite while also providing all of the building
blocks to implement additional rendering techniques.

@menu
* Rendering Engine::            Rendering state management.
* Textures::                    2D images.
* Sprites::                     Draw 2D images.
* Tile Maps::                   Draw 2D tile maps.
* Lines and Shapes::            Draw line segments and polygons.
* Fonts::                       Drawing text.
* Blending and Depth Testing::  Control how pixels are combined.
* Vertex Arrays::               Create 2D/3D models.
* Shaders::                     Create custom GPU programs.
* Framebuffers::                Render to texture.
* Viewports::                   Restrict rendering to
@end menu

@node Rendering Engine
@subsection Rendering Engine

Chickadee defines rendering using a metaphor familiar to Scheme
programmers: procedure application.  A shader (@pxref{Shaders}) is
like a procedure for the GPU to apply.  Shaders are passed arguments:
A vertex array containing the geometry to render (@pxref{Vertex
Arrays}) and zero or more keyword arguments that the shader
understands.  Similar to how Scheme has @code{apply} for calling
procedures, Chickadee provides @code{gpu-apply} for calling shaders.

Additionally, there is some dynamic state that effects how
@code{gpu-apply} will behave.  Things like the current viewport,
framebuffer, and blend mode are stored as dynamic state because it
would be tedious to have to have to specify them each time
@code{gpu-apply} is called.

The following procedures and syntax can be found in the
@code{(chickadee render)} module.

@deffn {Syntax} gpu-apply @var{shader} @var{vertex-array} @
       [#:uniform-key @var{uniform-value} ...]
@deffnx {Syntax} gpu-apply* @var{shader} @var{vertex-array} @
        @var{count} [#:uniform-key @var{uniform-value} ...]

Render @var{vertex-array} using @var{shader} with the uniform values
specified in the following keyword arguments.

While @code{gpu-apply} will draw every vertex in @var{vertex-array},
@code{gpu-apply*} will only draw @var{count} vertices.

@end deffn

@deffn {Procedure} current-viewport
Return the currently bound viewport (@pxref{Viewports}).
@end deffn

@deffn {Procedure} current-framebuffer
Return the currently bound framebuffer (@pxref{Framebuffers}).
@end deffn

@deffn {Procedure} current-blend-mode
Return the currently bound blend mode (@pxref{Blending and Depth
Testing}).
@end deffn

@deffn {Procedure} current-depth-test
Return @code{#t} if depth testing is currently enabled (@pxref{Blending and Depth Testing}).
@end deffn

@deffn {Procedure} current-texture
Return the currently bound texture (@pxref{Textures}).
@end deffn

@deffn {Procedure} current-projection
Return the currently bound projection matrix (@pxref{Matrices}).
@end deffn

@deffn {Syntax} with-viewport @var{viewport} @var{body} ...
Evaluate @var{body} with the current viewport bound to @var{viewport} (@pxref{Viewports}).
@end deffn

@deffn {Syntax} with-framebuffer @var{framebuffer} @var{body} ...
Evaluate @var{body} with the current framebuffer bound to
@var{framebuffer} (@pxref{Framebuffers}).
@end deffn

@deffn {Syntax} with-blend-mode @var{blend-mode} @var{body} ...
Evaluate @var{body} with the current blend mode bound to
@var{blend-mode} (@pxref{Blending and Depth Testing}).
@end deffn

@deffn {Syntax} with-depth-test @var{depth-test?} @var{body} ...
Evaluate @var{body} with the depth-test disabled if @var{depth-test?}
is @code{#f}, or enabled otherwise (@pxref{Blending and Depth
Testing}).
@end deffn

@deffn {Syntax} with-texture @var{texture} @var{body} ...
Evaluate @var{body} with the current texture bound to @var{texture}
(@pxref{Textures}).
@end deffn

@deffn {Syntax} with-projection @var{projection} @var{body} ...
Evaluate @var{body} with the current projection matrix bound to
@var{projection} (@pxref{Matrices}).
@end deffn

@node Textures
@subsection Textures

@deffn {Procedure} load-image @var{file} [#:min-filter nearest] @
       [#:mag-filter nearest] [#:wrap-s repeat] [#:wrap-t repeat]

Load the image data from @var{file} and return a new texture object.

@var{min-filter} and @var{mag-filter} describe the method that should
be used for minification and magnification when rendering,
respectively.  Possible values are @code{nearest} and @code{linear}.

@var{wrap-s} and @var{wrap-t} describe how to interpret texture
coordinates that are greater than @code{1.0}.  Possible values are
@code{repeat}, @code{clamp}, @code{clamp-to-border}, and
@code{clamp-to-edge}.

@end deffn

@node Sprites
@subsection Sprites

For those who are new to this game, a sprite is a 2D rectangular
bitmap that is rendered to the screen.  For 2D games, sprites are the
most essential graphical abstraction.  They are used for drawing maps,
players, NPCs, items, particles, text, etc.  In Chickadee, bitmaps are
stored in textures (@pxref{Textures}) and can be used to draw sprites
via the @code{draw-sprite} procedure.

@deffn {Procedure} draw-sprite @var{texture} @var{position} @
       [#:origin] [#:scale] [#:rotation] [#:blend-mode alpha] @
       [#:rect] [#:shader]

Draw @var{texture} at @var{position}.

Optionally, other transformations may be applied to the sprite.
@var{rotation} specifies the angle to rotate the sprite, in radians.
@var{scale} specifies the scaling factor as a 2D vector.  All
transformations are applied relative to @var{origin}, a 2D vector,
which defaults to the lower-left corner.

Alpha blending is used by default but the blending method can be
changed by specifying @var{blend-mode}.

The area drawn to is as big as the texture, by default.  To draw to an
arbitrary section of the screen, specify @var{rect}.

Finally, advanced users may specify @var{shader} to change the way the
sprite is rendered entirely.
@end deffn

It's not uncommon to need to draw hundreds or thousands of sprites
each frame.  However, GPUs (graphics processing units) are tricky
beasts that prefer to be sent few, large chunks of data to render
rather than many, small chunks.  Using @code{draw-sprite} on its own
will involve at least one GPU call @emph{per sprite}, which will
quickly lead to poor performance.  To deal with this, a technique
known as ``sprite batching'' can be used.  Instead of drawing each
sprite immediately, the sprite batch will build up a large of buffer
of sprites to draw and defer rendering until the last possible moment.
Batching isn't a panacea, though.  Batching only works if the sprites
being drawn share as much in common as possible.  Every time you draw
a sprite with a different texture or blend mode, the batch will be
sent off to the GPU.  Therefore, batching is most useful if you
minimize such changes.  A good strategy for reducing texture changes
is to stuff many bitmaps into a single image file and create a
``texture atlas'' (@pxref{Textures}) to access the sub-images within.

Taking advantage of sprite batching in Chickadee is easy, just wrap
the code that is calling @code{draw-sprite} a lot in the
@code{with-batched-sprites} form.

@deffn {Syntax} with-batched-sprites @var{body} @dots{}
Use batched rendering for all @code{draw-sprite} calls within
@var{body}.
@end deffn

With a basic sprite abstraction in place, it's possible to build other
abstractions on top of it.  One such example is the ``nine patch''.  A
nine patch is a sprite that can be rendered at various sizes without
becoming distorted.  This is achieved by diving up the sprite into
nine regions:

@itemize
@item
the center, which can be scaled horizontally and vertically
@item
the four corners, which can never be scaled
@item
the left and right sides, which can be scaled vertically
@item
the top and bottom sides, which can be scaled horizontally
@end itemize

The one caveat is that the bitmap regions must be designed in such a
way so that they are not distorted when stretched along the affected
axes.  For example, that means that the top and bottom sides could
have varying colored pixels vertically, but not horizontally.

The most common application of this technique is for graphical user
interface widgets like buttons and dialog boxes.  By using a nine
patch, they can be rendered at any size without unappealing scaling
artifacts.

@deffn {Procedure} draw-nine-patch @var{texture} @var{rect} @
       [#:margin 0] [#:top-margin margin] [#:bottom-margin margin] @
       [#:left-margin margin] [#:right-margin margin] @
       [#:origin] [#:scale] [#:rotation] [#:blend-mode alpha] @
       [#:shader]

Draw a nine patch sprite.  A nine patch sprite renders @var{texture}
as a @var{width} x @var{height} rectangle whose stretchable areas are
defined by the given margin measurements @var{top-margin},
@var{bottom-margin}, @var{left-margin}, and @var{right-margin}. The
@var{margin} argument may be used to configure all four margins at
once.

Refer to @code{draw-sprite} (@pxref{Sprites}) for information about
the other arguments.
@end deffn

@node Tile Maps
@subsection Tile Maps

A tile map is a scene created by composing lots of small sprites,
called ``tiles'', into a larger image.  One program for editing such
maps is called @url{http://mapeditor.org,Tiled}.  Chickadee has native
support for loading and rendering Tiled maps in the @code{(chickadee
render tiled)} module.

@deffn {Procedure} load-tile-map @var{file-name}
Load the Tiled formatted map in @var{file-name} and return a new tile
map object.
@end deffn

@deffn {Procedure} draw-tile-map @var{tile-map} [#:layers] [#:region] @
       [#:origin] [#:position] [#:scale] [#:rotation]

Draw the layers of @var{tile-map}.  By default, all layers are drawn.
To draw a subset of the available layers, pass a list of layer ids
using the @var{layers} keyword argument.

Refer to @code{draw-sprite} (@pxref{Sprites}) for information about
the other arguments.
@end deffn

@node Lines and Shapes
@subsection Lines and Shapes

Sprites are fun, but sometimes simple, untextured lines and polygons
are desired.  That's where the @code{(chickadee render shapes)} module
comes in!

@deffn {Procedure} draw-line @var{start} @var{end} @
       [#:thickness 0.5] [#:feather 1.0] [#:cap round] [#:color] @
       [#:shader]

Draw a line segment from @var{start} to @var{end}. The line will be
@var{thickness} pixels thick with an antialiased border @var{feather}
pixels wide.  The line will be colored @var{color}. @var{cap}
specifies the type of end cap that should be used to terminate the
lines, either @code{none}, @code{butt}, @code{square}, @code{round},
@code{triangle-in}, or @code{triangle-out}.  Advanced users may use
the @var{shader} argument to override the built-in line segment
shader.
@end deffn

@deffn {Procedure} draw-bezier-curve @var{bezier} [#:segments 32] @
                                     [#:control-points?] [#:tangents?] @
                                     [#:control-point-size 8] @
                                     [#:control-point-color yellow] @
                                     [#:tangent-color yellow] @
                                     [#:thickness 0.5] [#:feather 1.0] @
                                     [#:matrix]

Draw the curve defined by @var{bezier} using a resolution of N
@var{segments}.  When @var{control-points?} is @code{#t}, the control
points are rendered as squares of size @var{control-point-size} pixels
and a color of @var{control-point-color}.  When @var{tangents?} is
@code{#t}, the tangent lines from terminal point to control point are
rendered using the color @var{tangent-color}.

All line segments rendered use @code{draw-line}, and thus the
arguments @var{thickness} and @var{feather} have the same effect as in
that procedure.

A custom @var{matrix} may be passed for applications that require more
control over the final output.
@end deffn

@deffn {Procedure} draw-bezier-path @var{path} [#:segments 32] @
                                    [#:control-points?] [#:tangents?] @
                                    [#:control-point-size 8] @
                                    [#:control-point-color yellow] @
                                    [#:tangent-color yellow] @
                                    [#:thickness 0.5] [#:feather 1.0] @
                                    [#:matrix]

Render @var{path}, a list of bezier curves.  See the documentation for
@code{draw-bezier-curve} for an explanation of all the keyword
arguments.
@end deffn

@node Fonts
@subsection Fonts

Unlike the traditional TrueType font format that many are accustomed
to, Chickadee loads and renders bitmap fonts in the
@url{http://www.angelcode.com/products/bmfont/doc/file_format.html,
Angel Code format}.  But why use this seemingly obscure format?  It's
easy to find TTFs but not easy to find FNTs (the canonical file
extension used for Angel Code fonts) and bitmap fonts don't scale
well.  The reason is efficiency.

If all of the glyphs of a font are pre-rendered and packed into an
image file then it becomes possible to use a texture atlas
(@pxref{Textures}) and a sprite batch (@pxref{Sprites}) when
rendering, which is a more efficient way to render fonts than using,
say, @url{https://www.libsdl.org/projects/SDL_ttf/, SDL_ttf} or other
solutions that involve using the FreeType library directly.

Now what about scaling?  In libraries that use TTF fonts, one must
choose the size that the glyphs will be rasterized at up front.  To
use @code{n} sizes of the same font, one must load @code{n} variants
of that font.  If the size of the text is dynamic, some kind of
texture scaling algorithm must be used and the text will inevitably
look blurry.  At first glance, using bitmap fonts seem to have an even
worse issue.  Instead of just loading the same font @code{n} times at
different sizes, one would need to generate @code{n} image files for
each font size needed.  This is where the ``signed distance field''
rendering technique comes in.  Introduced by
@url{http://www.valvesoftware.com/.../2007/SIGGRAPH2007_AlphaTestedMagnification.pdf,
Valve} in 2007, signed distance field fonts can be efficiently stored
in a bitmap and be rendered at arbitrary scale factors with good
results.

While Chickadee does not yet offer a tool for converting TTF fonts
into FNT fonts, tools such as
@url{https://github.com/libgdx/libgdx/wiki/Hiero, Hiero} may be used
in the meantime.

The following procedures can be found in the @code{(chickadee render
font)} module.

@deffn {Procedure} load-font @var{file}
Load the Angel Code formatted XML document in @var{file} and return a
new font object.
@end deffn

@deffn {Procedure} font? @var{obj}
Return @code{#t} if @var{obj} is a font object.
@end deffn

@deffn {Procedure} font-face @var{font}
Return the name of @var{font}.
@end deffn

@deffn {Procedure} font-line-height @var{font}
Return the line height of @var{font}.
@end deffn

@deffn {Procedure} font-line-height @var{font}
Return the line height of @var{font}.
@end deffn

@deffn {Procedure} font-bold? @var{font}
Return @code{#t} if @var{font} is a bold font.
@end deffn

@deffn {Procedure} font-italic? @var{font}
Return @code{#t} if @var{font} is an italicized font.
@end deffn

@deffn {Procedure} draw-text @var{font} @var{text} @var{position}
       [#:origin] [#:scale] [#:rotation] [#:blend-mode]
       [#:start 0] [#:end @code{(string-length text)}]

Draw the string @var{text} with the first character starting at
@var{position} using @var{font}.

@example
(draw-text font "Hello, world!" (vec2 128.0 128.0))
@end example

To render a substring of @var{text}, use the @var{start} and @var{end}
arguments.

Refer to @code{draw-sprite} (@pxref{Sprites}) for information about
the other arguments.
@end deffn

@node Blending and Depth Testing
@subsection Blending and Depth Testing

@node Vertex Arrays
@subsection Vertex Arrays

@node Shaders
@subsection Shaders

Shaders are programs for the GPU to evaluate.  They are written in the
OpenGL Shading Language, or GLSL.  Chickadee does not currently
provide a Scheme-like domain specific language for writing shaders.
Since shaders must be written in GLSL and not Scheme, they are
considered an advanced feature.

@node Framebuffers
@subsection Framebuffers

A framebuffer is a chunk of memory that the GPU can render things
onto.  By default, the framebuffer that is used for rendering is the
one belonging to the game window, but custom framebuffers can be used
as well.  A common use-case for custom framebuffers is applying
post-processing effects: The entire scene is rendered to a
framebuffer, and then the contents of that framebuffer are applied to
a post-processing shader and rendered to the game window.  The
post-processing shader could do any number of things: scaling,
antialiasing, motion blur, etc.

@deffn {Procedure} make-framebuffer @var{width} @var{height} [#:min-filter 'linear] [#:mag-filter 'linear] [#:wrap-s 'repeat] [#:wrap-t 'repeat]

Create a new framebuffer that is @var{width} pixels wide and @var{height} pixels high.

@var{min-filter} and @var{mag-filter} determine the scaling algorithm
applied to the framebuffer when rendering.  By default, linear scaling
is used in both cases. To perform no smoothing at all, use
@code{nearest} for simple nearest neighbor scaling.  This is typically
the best choice for pixel art games.
@end deffn

@deffn {Procedure} framebuffer? @var{obj}
Return @code{#t} if @var{obj} is a framebuffer.
@end deffn

@deffn {Procedure} framebuffer-texture @var{fb}
Return the texture backing the framebuffer @var{fb}.
@end deffn

@deffn {Procedure} framebuffer-viewport @var{fb}
Return the default viewport (@pxref{Viewports}) used by the
framebuffer @var{fb}.
@end deffn

@deffn {Procedure} null-framebuffer
The default framebuffer.
@end deffn

@node Viewports
@subsection Viewports

@node Scripting
@section Scripting

Game logic is a web of asynchronous events that are carefully
coordinated to bring the game world to life.  In order to make an
enemy follow and attack the player, or move an NPC back and forth in
front of the item shop, or do both at the same time, a scripting
system is a necessity.  Chickadee comes with an asynchronous
programming system in the @code{(chickadee scripting)} module.
Lightweight, cooperative threads known as ``scripts'' allow the
programmer to write asynchronous code as if it were synchronous, and
allow many such ``threads'' to run concurrently.

But before we dig deeper into scripts, let's discuss the simple act
of scheduling tasks.

@menu
* Agendas::                     Scheduling tasks.
* Scripts::                     Cooperative multitasking.
* Tweening::                    Animations.
* Channels::                    Publish data to listeners.
@end menu

@node Agendas
@subsection Agendas

To schedule a task to be performed later, an ``agenda'' is used.
There is a default, global agenda that is ready to be used, or
additional agendas may be created for different purposes.  The
following example prints the text ``hello'' when the agenda has
advanced to time unit 10.

@example
(at 10 (display "hello\n"))
@end example

Most of the time it is more convenient to schedule tasks relative to
the current time.  This is where @code{after} comes in handy:

@example
(after 10 (display "hello\n"))
@end example

Time units in the agenda are in no way connected to real time.  It's
up to the programmer to decide what agenda time means.  A simple and
effective approach is to map each call of the update hook
(@pxref{Kernel}) to 1 unit of agenda time, like so:

@example
(add-hook! update-hook (lambda (dt) (update-agenda 1)))
@end example

It is important to call @code{update-agenda} periodically, otherwise
no tasks will ever be run!

In addition to using the global agenda, it is useful to have multiple
agendas for different purposes.  For example, the game world can use a
different agenda than the user interface, so that pausing the game is
a simple matter of not updating the world's agenda while continuing to
update the user interface's agenda.  The current agenda is dynamically
scoped and can be changed using the @code{with-agenda} special form:

@example
(define game-world-agenda (make-agenda))

(with-agenda game-world-agenda
  (at 60 (spawn-goblin))
  (at 120 (spawn-goblin))
  (at 240 (spawn-goblin-king)))
@end example

@deffn {Procedure} make-agenda
Return a new task scheduler.
@end deffn

@deffn {Procedure} agenda? @var{obj}
Return @code{#t} if @var{obj} is an agenda.
@end deffn

@deffn {Procedure} current-agenda
@deffnx {Procedure} current-agenda @var{agenda}
When called with no arguments, return the current agenda.  When called
with one argument, set the current agenda to @var{agenda}.
@end deffn

@deffn {Syntax} with-agenda @var{agenda} @var{body} @dots{}
Evaluate @var{body} with the current agenda set to @var{agenda}.
@end deffn

@deffn {Procedure} agenda-time
Return the current agenda time.
@end deffn

@deffn {Procedure} update-agenda @var{dt}
Advance the current agenda by @var{dt}.
@end deffn

@deffn {Procedure} schedule-at @var{time} @var{thunk}
Schedule @var{thunk}, a procedure of zero arguments, to be run at
@var{time}.
@end deffn

@deffn {Procedure} schedule-after @var{delay} @var{thunk}
Schedule @var{thunk}, a procedure of zero arguments, to be run after
@var{delay}.
@end deffn

@deffn {Procedure} schedule-every @var{interval} @var{thunk} [@var{n}]
Schedule @var{thunk}, a procedure of zero arguments, to be run every
@var{interval} amount of time.  Repeat this @var{n} times, or
indefinitely if not specified.
@end deffn

@deffn {Syntax} at @var{time} @var{body} @dots{}
Schedule @var{body} to be evaluated at @var{time}.
@end deffn

@deffn {Syntax} after @var{delay} @var{body} @dots{}
Schedule @var{body} to be evaluated after @var{delay}.
@end deffn

@deffn {Syntax} every @var{interval} @var{body} @dots{}
@deffnx {Syntax} every (@var{interval} @var{n}) @var{body} @dots{}
Schedule @var{body} to be evaluated every @var{interval} amount of
time.  Repeat this @var{n} times, or indefinitely if not specified.
@end deffn

@node Scripts
@subsection Scripts

Now that we can schedule tasks, let's take things to the next level.
It sure would be great if we could make procedures that described a
series of actions that happened over time, especially if we could do
so without contorting our code into a nest of callback procedures.
This is where scripts come in.  With scripts we can write code in a
linear way, in a manner that appears to be synchronous, but with the
ability to suspend periodically in order to let other scripts have a
turn and prevent blocking the game loop.  Building on top of the
scheduling that agendas provide, here is a script that models a child
trying to get their mother's attention:

@example
(script
  (while #t
    (display "mom!")
    (newline)
    (sleep 60))) ; where 60 = 1 second of real time
@end example

This code runs in an endless loop, but the @code{sleep} procedure
suspends the script and schedules it to be run later by the agenda.
So, after each iteration of the loop, control is returned back to the
game loop and the program is not stuck spinning in a loop that will
never exit.  Pretty neat, eh?

Scripts can suspend to any capable handler, not just the agenda.
The @code{yield} procedure will suspend the current script and pass
its ``continuation'' to a handler procedure.  This handler procedure
could do anything.  Perhaps the handler stashes the continuation
somewhere where it will be resumed when the user presses a specific
key on the keyboard, or maybe it will be resumed when the player picks
up an item off of the dungeon floor; the sky is the limit.

Sometimes it is necessary to abruptly terminate a script after it has
been started.  For example, when an enemy is defeated their AI routine
needs to be shut down.  When a script is spawned, a handle to that
script is returned that can be used to cancel it when desired.

@example
(define script (script (while #t (display "hey\n") (sleep 60))))
;; sometime later
(cancel-script script)
@end example

@deffn {Procedure} spawn-script @var{thunk}
Apply @var{thunk} as a script and return a handle to it.
@end deffn

@deffn {Syntax} script @var{body} @dots{}
Evaluate @var{body} as a script and return a handle to it.
@end deffn

@deffn {Procedure} script? @var{obj}
Return @code{#t} if @var{obj} is a script handle.
@end deffn

@deffn {Procedure} script-cancelled? @var{obj}
Return @code{#t} if @var{obj} has been cancelled.
@end deffn

@deffn {Procedure} script-running? @var{obj}
Return @code{#t} if @var{obj} has not yet terminated or been
cancelled.
@end deffn

@deffn {Procedure} script-complete? @var{obj}
Return @code{#t} if @var{obj} has terminated.
@end deffn

@deffn {Procedure} cancel-script @var{co}
Prevent further execution of the script @var{co}.
@end deffn

@deffn {Procedure} yield @var{handler}
Suspend the current script and pass its continuation to the
procedure @var{handler}.
@end deffn

@deffn {Procedure} sleep @var{duration}
Wait @var{duration} before resuming the current script.
@end deffn

@deffn {Procedure} channel-get @var{channel}
Wait for a message from @var{channel}.
@end deffn

@deffn {Syntax} forever @var{body} @dots{}
Evaluate @var{body} in an endless loop.
@end deffn

@node Tweening
@subsection Tweening

Tweening is the process of transitioning something from an initial
state to a final state over a pre-determined period of time.  In other
words, tweening is a way to create animation.  The @code{tween}
procedure can be used within any script like so:

@example
(define x 0)
(script
  ;; 0 to 100 in 60 ticks of the agenda.
  (tween 60 0 100 (lambda (y) (set! x y))))
@end example

@deffn {Procedure} tween @var{duration} @var{start} @var{end} @var{proc} [#:step 1 #:ease @code{smoothstep} #:interpolate @code{lerp}]
Transition a value from @var{start} to @var{end} over @var{duration},
sending each succesive value to @var{proc}.  @var{step} controls the
amount of time between each update of the animation.

To control how the animation goes from the initial to final state, an
``easing'' procedure may be specified.  By default, the
@code{smoothstep} easing is used, which is a more pleasing default
than a simplistic linear function.  @xref{Easings} for a complete list
of available easing procedures.

The @var{interpolate} procedure computes the values in between
@var{start} and @var{end}.  By default, linear interpolation (``lerp''
for short) is used.
@end deffn

@node Channels
@subsection Channels

Channels are a tool for communicating amongst different scripts.  One
script can write a value to the channel and another can read from it.
Reading or writing to a channel suspends that script until there is
someone on the other end of the line to complete the transaction.

Here's a simplistic example:

@example
(define c (make-channel))

(script
 (forever
  (let ((item (channel-get c)))
    (pk 'got item))))

(script
 (channel-put c 'sword)
 (channel-put c 'shield)
 (channel-put c 'potion))
@end example

@deffn {Procedure} make-channel
Return a new channel
@end deffn

@deffn {Procedure} channel? @var{obj}
Return @code{#t} if @var{obj} is a channel.
@end deffn

@deffn {Procedure} channel-get @var{channel}
Retrieve a value from @var{channel}.  The current script suspends
until a value is available.
@end deffn

@deffn {Procedure} channel-put @var{channel} @var{data}
Send @var{data} to @var{channel}.  The current script suspends until
another script is available to retrieve the value.
@end deffn