path: root/doc/api.texi

@menu
* Kernel::                      The fundamental components.
* Math::                        Linear algebra, spatial partitioning, and more.
* Graphics::                    Eye candy.
* Audio::                       Make some noise.
* Scripting::                   Bringing the game world to life.
@end menu

@node Kernel
@section Kernel

This section of the manual covers the foundation of Chickadee: The
game loop and desktop environment interaction.

@menu
* The Game Loop::               The core event loop.
* Input Devices::               Mouse, keyboard, controller input.
* Window Manipulation::         Inspect and modify the graphical window.
* Live Coding::                 Tips for building games from the REPL.
@end menu

@node The Game Loop
@subsection The Game Loop

At the very core of Chickadee there is an event loop.  This loop, or
``kernel'', is responsible for ensuring that the game is updated at
the desired interval, handling input devices, 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.

@deffn {Procedure} run-game [#: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.

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 key that was pressed.  For
example: @code{backspace}.

@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 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

To stop the game loop, simply call @code{abort-game}.

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

The above explanation of the game loop was partially a lie.  It's true
that there is a game loop at the center of Chickadee, but
@code{run-game} is not it's true entry point.  There exists an even
lower level procedure, @code{run-game*}, in the @code{(chickadee
game-loop)} module that @code{run-game} uses under the hood.

On its own, @code{run-game*} does not do very much at all.  In order
to actually respond to input events, update game state, or render
output, the developer must provide an engine.  @code{run-game} is such
an engine, and it's likely all a developer will need.  However, what
if a developer wanted to use all of the useful Chickadee features to
make a terminal roguelike game instead?  Chickadee doesn't come with a
terminal rendering engine, but the developer could write one without
having to write their own core game loop.

@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} elapsed-time
Return the number of milliseconds that have passed since the game loop
was started.
@end deffn

@node Input Devices
@subsection Input Devices

While @code{run-game} provides hooks for mouse/keyboard/controller
input events, it is often necessary to query input devices for their
current state.  For example, it could be desirable to query the state
of the arrow keys every time the update hook is called to determine
which direction the player should move that frame.

@deffn {Procedure} key-pressed? key
Return @code{#t} if @var{key} is currently being pressed.
@end deffn

@deffn {Procedure} key-released? key
Return @code{#t} if @var{key} is @emph{not} currently being pressed.
@end deffn

@deffn {Procedure} mouse-x
Return the current X coordinate of the mouse cursor.
@end deffn

@deffn {Procedure} mouse-y
Return the current Y coordinate of the mouse cursor.
@end deffn

@deffn {Procedure} mouse-button-pressed? button
Return @code{#t} if @var{button} is currently being pressed.
@end deffn

@deffn {Procedure} mouse-button-released? button
Return @code{#t} if @var{button} is @emph{not} currently being
pressed.
@end deffn

@deffn {Procedure} controller-axis controller axis
Return a floating point value in the range [-1, 1] corresponding to
how much @var{axis} (an analog stick or trigger) is being pushed on
@var{controller}.  0 is returned if @var{axis} is not being pushed at
all.
@end deffn

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

@deffn {Procedure} controller-button-pressed? controller button
Return @code{#t} if @var{button} on @var{controller} is currently
being pressed.
@end deffn

@deffn {Procedure} controller-button-released? controller button
Return @code{#t} if @var{button} on @var{controller} is @emph{not}
currently being pressed.
@end deffn

@node Window Manipulation
@subsection Window Manipulation

@deffn {Procedure} current-window
Return the currently active game window.
@end deffn

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

@deffn {Procedure} window-title window
Return the title of @var{window}.
@end deffn

@deffn {Procedure} window-width window
Return the width of @var{window} in pixels.
@end deffn

@deffn {Procedure} window-height window
Return the height of @var{window} in pixels.
@end deffn

@deffn {Procedure} window-x window
Retun the X coordinate of the upper-left corner of @var{window}.
@end deffn

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

@deffn {Procedure} hide-window! window
Hide @var{window}.
@end deffn

@deffn {Procedure} show-window! window
Show @var{window}.
@end deffn

@deffn {Procedure} maximize-window! window
Maximize @var{window}.
@end deffn

@deffn {Procedure} minimize-window! window
Minimize @var{window}.
@end deffn

@deffn {Procedure} raise-window! window
Make @var{window} visible over all other windows.
@end deffn

@deffn {Procedure} restore-window! window
Restore the size and position of a minimized or maximized
@var{window}.
@end deffn

@deffn {Procedure} set-window-border! window border?
Enable/disable the border around @var{window}.  If @var{border?} is
@code{#f}, the border is disabled, otherwise it is enabled.
@end deffn

@deffn {Procedure} set-window-title! window title
Change the title of @var{window} to @var{title}.
@end deffn

@deffn {Procedure} set-window-size! window width height
Change the dimensions of @var{window} to @var{width} x @var{height}
pixels.
@end deffn

@deffn {Procedure} set-window-position! window x y
Move the upper-left corner of @var{window} to pixel coordinates
(@var{x}, @var{y}).
@end deffn

@deffn {Procedure} set-window-fullscreen! window fullscreen?
Enable or disable fullscreen mode for @var{window}.  If
@var{fullscreen?} is @code{#f}, fullscreen mode is disabled, otherwise
it is enabled.
@end deffn


@node Live Coding
@subsection Live Coding

One of the biggest appeals of any Lisp dialect is the ability to use
the ``read-eval-print loop'' (REPL for short) to build programs
iteratively and interactively while the program is running.  However,
programs that run in an event loop and respond to user input (such as
a video game) require special care for this workflow to be pleasant.
Chickadee provides no built-in support for live coding, but it's
fairly easy to hook up a special kind of REPL yourself.

First, create a cooperative REPL server (It's important to use Guile's
cooperative REPL server instead of the standard REPL server in
@code{(system repl server)} to avoid thread synchronization issues).
Then, in the game loop's update procedure, call
@code{poll-coop-repl-server} and pass the REPL object.  Here is a
template to follow:

@example
(use-modules (chickadee)
             (system repl coop-server))

(define repl (spawn-coop-repl-server))

(define (update dt)
  (poll-coop-repl-server repl)
  ...)

(run-game #:update update ...)
@end example

To use the REPL, connect to it via port 37146.  Telnet will do the
trick, but using the @uref{https://www.nongnu.org/geiser/, Geiser}
extension for Emacs is by far the best way to develop at the REPL with
Guile.  Use @code{M-x connect-to-guile} to connect to the REPL server.

@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.
* 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.
* Path Finding::                Generic A* path finding.
* Grid::                        Spatial partitioning for bounding boxes.
@end menu

@node Basics
@subsection Basics

@defvar pi
An essential constant for all trigonometry.  Pi is the ratio of a
circle's circumferences to its diameter.  Since pi 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

@defvar 2pi
Twice @var{pi}.
@end defvar

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

@deffn {Procedure} clamp min max x
Restrict @var{x} to the inclusive range defined by @var{min} and
@var{max}.  This procedure assumes that @var{min} is actually less
than @var{max}.
@end deffn

@deffn {Procedure} lerp start end alpha
Linearly interpolate the numbers @var{start} and @var{end} using the
factor @var{alpha}, a number in the range [0, 1].
@end deffn

@deffn {Procedure} degrees->radians degrees
Convert @var{degrees} to radians.
@end deffn

@deffn {Procedure} radians->degrees radians
Convert @var{radians} to degrees.
@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

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

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

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

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

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

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

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

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

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

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

@deffn {Procedure} set-vec2! v x y
Set the X and Y coordinates of the 2D vector @var{v} to @var{x} and
@var{y}, respectively.
@end deffn

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

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

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

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

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

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

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

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

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

@deffn {Procedure} vec3-cross v1 v2
Return a new 3D vector containing the cross product of @var{v1} and
@var{v2}.
@end deffn

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

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

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

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

@deffn {Procedure} set-vec3! v x y z
Set the X, Y, and Z coordinates of the 3D vector @var{v} to @var{x},
@var{y}, and @var{z}, respectively.
@end deffn

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

@deffn {Procedure} vec3-add! v 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! v 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! v 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

@deffn {Procedure} vec3-cross! dest v1 v2
Compute the cross product of the 3D vectors @var{v1} and @var{v2} and
store the result in @var{dest}.
@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} rect x y width height
@deffnx {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? obj
Return @code{#t} if @var{obj} is a rectangle.
@end deffn

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

@deffn {Procedure} rect-clip rect1 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! rect x
Set the left X coordinate of @var{rect} to @var{x}.
@end deffn

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

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

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

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

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

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

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

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

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

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

@deffn {Procedure} rect-clamp! rect1 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! v 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 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.

@emph{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 3x3 Matrices

@deffn {Procedure} make-matrix3 aa ab ac ba bb bc ca cb cc
Return a new 3x3 initialized with the given 9 values in column-major
format.
@end deffn

@deffn {Procedure} make-null-matrix3
Return a new 3x3 matrix with all values initialized to 0.
@end deffn

@deffn {Procedure} make-identity-matrix3
Return a new 3x3 identity matrix.  Any matrix multiplied by the
identity matrix yields the original matrix.  This procedure is
equivalent to the following code:

@example
(make-matrix3 1 0 0
              0 1 0
              0 0 1)
@end example

@end deffn

@deffn {Procedure} matrix3? obj
Return @code{#t} if @var{obj} is a 3x3 matrix.
@end deffn

@deffn {Procedure} matrix3* . matrices
Return a new 3x3 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} matrix3-translate v
Return a new 3x3 matrix that represents a translation by @var{v}, a 2D
vector.
@end deffn

@deffn {Procedure} matrix3-scale s
Return a new 3x3 matrix that represents a scaling along the x and y
axes by the scaling factor @var{s}, a number or 2D vector.
@end deffn

@deffn {Procedure} matrix3-rotate angle
Return a new 3x3 matrix that represents a rotation by @var{angle}
radians.
@end deffn

@deffn {Procedure} matrix3-transform matrix v
Return a new 2D vector that is @var{v} as transformed by the 3x3
matrix @var{matrix}.
@end deffn

The following procedures perform in-place, destructive updates to 3x3
matrix objects:

@deffn {Procedure} matrix3-identity! matrix
Modify @var{matrix} in-place to contain the identity matrix.
@end deffn

@deffn {Procedure} matrix3-mult! dest a b
Multiply the 3x3 matrix @var{a} by the 3x3 matrix @var{b} and store
the result in the 3x3 matrix @var{dest}.
@end deffn

@deffn {Procedure} matrix3-translate! matrix v
Modify @var{matrix} in-place to contain a translation by @var{v}, a 2D
vector.
@end deffn

@deffn {Procedure} matrix3-scale! matrix s
Modify @var{matrix} in-place to contain a scaling along the x and y
axes by the scaling factor @var{s}, a number or 2D vector.
@end deffn

@deffn {Procedure} matrix3-rotate! matrix angle
Modify @var{matrix} in-place to contain a rotation by @var{angle}
radians.
@end deffn

@deffn {Procedure} matrix3-transform! matrix v
Modify the 2D vector @var{v} in-place to contain @var{v} as
transformed by the 3x3 matrix @var{matrix}.
@end deffn

@subsubsection 4x4 Matrices

@deffn {Procedure} make-matrix4 aa ab ac ad @
                                ba bb bc bd @
                                ca cb cc cd @
                                da db dc 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? obj
Return @code{#t} if @var{obj} is a 4x4 matrix.
@end deffn

@deffn {Procedure} matrix4* . 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 left right top bottom near 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 fov aspect-ratio near 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 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 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 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 theta
Return a new 4x4 matrix that represents a rotation about the Z axis by
@var{theta} radians.
@end deffn

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

@deffn {Procedure} matrix4-mult! dest a 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! matrix 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! matrix 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! matrix 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! matrix 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! 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! matrix 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 x y z w
Return a new quaternion with values @var{x}, @var{y}, @var{z}, and
@var{w}.
@end deffn

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

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

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

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

@deffn {Procedure} quaternion-z 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

Easing functions are essential for animation.  Each easing function
provides a different path to go from an initial value to a final
value.  These functions make an excellent companion to the
@code{tween} procedure (@pxref{Tweening}).  Experiment with them to
figure out which function makes an animation look the best.

Pro tip: @code{smoothstep} provides nice results most of the time and
creates smoother animation than using @code{linear}.

@deffn {Procedure} linear t
@end deffn

@deffn {Procedure} smoothstep t
@end deffn

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

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

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

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

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

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

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

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

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

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

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

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

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

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

@deffn {Procedure} ease-in-out-sine 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.

@deffn {Procedure} make-bezier-curve p0 p1 p2 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? obj
Return @code{#t} if @var{obj} is a Bezier curve.
@end deffn

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

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

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

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

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

@node Path Finding
@subsection Path Finding

Most game worlds have maps.  Often, these games have a need to move
non-player characters around in an unscripted fashion.  For example,
in a real-time strategy game, the player may command one of their
units to attack something in the enemy base.  To do so, the unit must
calculate the shortest route to get there.  It wouldn't be a very fun
game if units didn't know how to transport themselves efficiently.
This is where path finding algorithms come in handy.  The
@code{(chickadee math path-finding)} module provides a generic
implementation of the popular A* path finding algorithm.  Just add a
map implementation!

The example below defines a very simple town map and finds the
quickest way to get from the town common to the school.

@example
(define world-map
  '((town-common . (town-hall library))
    (town-hall . (town-common school))
    (library . (town-common cafe))
    (school . (town-hall cafe))
    (cafe . (library school))))
(define (neighbors building)
  (assq-ref town-map building))
(define (cost a b) 1)
(define (distance a b) 1)
(define pf (make-path-finder))
(a* pf 'town-common 'school neighbors cost distance)
@end example

In this case, the @code{a*} procedure will return the list
@code{(town-common town-hall school)}, which is indeed the shortest
route.  (The other possible route is @code{(town-common library cafe
school)}.)

The @code{a*} procedure does not know anything about about any kind of
map and therefore must be told how to look up neighboring nodes, which
is what the @code{neighbors} procedure in the example does. To
simulate different types of terrain, a cost procedure is used.  In
this example, it is just as easy to move between any two nodes because
@code{cost} always returns 1.  In a real game, perhaps moving from
from a field to a rocky hill would cost a lot more than moving from
one field to another.  Finally, a heuristic is used to calculate an
approximate distance between two nodes on the map.  In this simple
association list based graph it is tough to calculate a distance
between nodes, so the @code{distance} procedure isn't helpful and
always returns 1.  In a real game with a tile-based map, for example,
the heuristic could be a quick Manhattan distance calculation based on
the coordinates of the two map tiles.  Choose an appropriate heuristic
for optimal path finding!

@deffn {Procedure} make-path-finder
Return a new path finder object.
@end deffn

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

@deffn {Procedure} a* path-finder start goal neighbors cost distance

Return a list of nodes forming a path from @var{start} to @var{goal}
using @var{path-finder} to hold state.  @var{neighbors} is a procedure
that accepts a node and returns a list of nodes that neighbor it.
@var{cost} is a procedure that accepts two neighboring nodes and
returns the cost of moving from the first to the second as a real
number.  @var{distance} is a procedure that accepts two nodes and
returns an approximate distance between them.
@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 [cell-size 64]
Return a new grid partitioned into @var{cell-size} tiles.
@end deffn

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

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

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

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

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

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

@deffn {Procedure} grid-add grid item x y @
                            width 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 grid item
Return @var{item} from @var{grid}.
@end deffn

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

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

@deffn {Procedure} slide item item-rect other other-rect 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 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
* Colors::                      Such pretty colors...
* Textures::                    2D images.
* Sprites::                     Draw 2D images.
* Fonts::                       Drawing text.
* Vector Paths::                Draw filled and stroked paths.
* Particles::                   Pretty little flying pieces!
* Tile Maps::                   Draw 2D tile maps.
* 3D Models::                   Spinning teapots everywhere.
* Blending::                    Control how pixels are combined.
* Framebuffers::                Render to texture.
* Viewports::                   Restrict rendering to a particular area.
* Rendering Engine::            Rendering state management.
* Buffers::                     Send data to the GPU.
* Shaders::                     Create custom GPU programs.
@end menu

@node Colors
@subsection Colors

Merriam-Webster defines color as ``a phenomenon of light (such as red,
brown, pink, or gray) or visual perception that enables one to
differentiate otherwise identical objects.''  In this essay, I
will@dots{}

Okay, okay.  We all know what colors are.  Chickadee provides a data
type to represent color and some convenient procedures to work with
them in the @code{(chickadee graphics color)} module.  Colors are made
up of four components, or channels: red, green, blue, and alpha
(transparency.)  Each of these values is expressed as a uniform
floating point value in the range [0, 1].  0 means that color channel
is unrepresented in the resulting color, whereas 1 means that color
channel is at full intensity.

Making a color object is easy, and there's a few ways to do it
depending on what's most convenient.  The first is @code{make-color},
where you specify each channel exactly as described above.  This is
fully opaque magenta:

@example
(make-color 1.0 0.0 1.0 1.0)
@end example

Many people are used to representing colors as 6 or 8 digit
hexadecimal numbers, so Chickadee also allows that.  Here's magenta,
again:

@example
(rgba #xFF00FFFF)
(rgb #xFF00FF) ; equivalent to the above
@end example

@deffn {Procedure} make-color r g b a
Return a new color object with a red value of @var{r}, a green value
of @var{g}, a blue value of @var{b}, and an alpha (transparency) value
of @var{a}.  All values are clamped to the range [0, 1].
@end deffn

@deffn {Procedure} rgba color-code
Return a new color object using the values of the first 32 bits of
@var{color-code}.  Each channel occupies 8 bits.  Starting from the
most significant bit, red is first, followed by green, then blue, then
alpha.  Color codes are often represented as 6 or 8 digit hexadecimal
numbers in various other programs.
@end deffn

@deffn {Procedure} rgb color-code
Like @code{rgba}, but @var{color-code} is a 24 bit code with no alpha
channel.
@end deffn

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

@deffn {Procedure} color-r color
Return the red channel of @var{color}.
@end deffn

@deffn {Procedure} color-g color
Return the green channel of @var{color}.
@end deffn

@deffn {Procedure} color-b color
Return the blue channel of @var{color}.
@end deffn

@deffn {Procedure} color-a color
Return the alpha channel of @var{color}.
@end deffn

@deffn {Procedure} transparency alpha
Return a new color that is white (RGB channels set to 1) with an alpha
channel value of @var{alpha}.  This can be useful for creating a color
that can be multiplied against another color to make it more
transparent.
@end deffn

@deffn {Procedure} string->color s
Convert the hexadecimal color code in the string @var{s} to a color
object.  The following string formats are supported:

@example
(string->color "#FF00FFFF")
(string->color "FF00FFFF")
(string->color "#FF00FF")
(string->color "FF00FF")
@end example

@end deffn

@deffn {Procedure} color* a b
Multiply the color @var{a} with the color or number @var{b} and return
a new color with the result.
@end deffn

@deffn {Procedure} color+ a b
Add the color @var{a} to the color @var{b} and return a new color with
the result.
@end deffn

@deffn {Procedure} color- a b
Subtract the color @var{b} from the color @var{a} and return a new
color with the result.
@end deffn

@deffn {Procedure} color-inverse color
Invert the red, green, and blue channels of @var{color} and return a
new color with the result.
@end deffn

@deffn {Procedure} color-lerp start end alpha
Linearly interpolate the colors @var{start} and @var{end} using the
factor @var{alpha}, a number in the range [0, 1].
@end deffn

@subsubsection Stock Colors

For convenience, Chickadee comes with some basic colors predefined:

@defvar white
@end defvar

@defvar black
@end defvar

@defvar red
@end defvar

@defvar green
@end defvar

@defvar blue
@end defvar

@defvar yellow
@end defvar

@defvar magenta
@end defvar

@defvar cyan
@end defvar

For fun, there are also predefined colors for the classic
@url{https://en.wikipedia.org/wiki/Tango_Desktop_Project#Palette,
Tango color palette}.

@defvar tango-light-butter
@end defvar

@defvar tango-butter
@end defvar

@defvar tango-dark-butter
@end defvar

@defvar tango-light-orange
@end defvar

@defvar tango-orange
@end defvar

@defvar tango-dark-orange
@end defvar

@defvar tango-light-chocolate
@end defvar

@defvar tango-chocolate
@end defvar

@defvar tango-dark-chocolate
@end defvar

@defvar tango-light-chameleon
@end defvar

@defvar tango-chameleon
@end defvar

@defvar tango-dark-chameleon
@end defvar

@defvar tango-light-sky-blue
@end defvar

@defvar tango-sky-blue
@end defvar

@defvar tango-dark-sky-blue
@end defvar

@defvar tango-light-plum
@end defvar

@defvar tango-plum
@end defvar

@defvar tango-dark-plum
@end defvar

@defvar tango-light-scarlet-red
@end defvar

@defvar tango-scarlet-red
@end defvar

@defvar tango-dark-scarlet-red
@end defvar

@defvar tango-aluminium-1
@end defvar

@defvar tango-aluminium-2
@end defvar

@defvar tango-aluminium-3
@end defvar

@defvar tango-aluminium-4
@end defvar

@defvar tango-aluminium-5
@end defvar

@defvar tango-aluminium-6
@end defvar

@node Textures
@subsection Textures

Textures are essentially images: a 2D grid of color values.  The
@code{(chickadee graphics texture)} module provides an interface for
working with texture objects.

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

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}.

For color-keyed images (images where a specific color should be made
transparent), specify the appropriate @var{transparent-color}.

@end deffn

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

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

@deffn {Procedure} texture-parent texture
If @var{texture} is a texture region, return the full texture that it
is based upon.  Otherwise, return @code{#f}.
@end deffn

@deffn {Procedure} texture-min-filter texture
Return the minification filter for @var{texture}, either
@code{nearest} or @code{linear}.
@end deffn

@deffn {Procedure} texture-mag-filter texture
Return the magnification filter for @var{texture}, either
@code{nearest} or @code{linear}.
@end deffn

@deffn {Procedure} texture-wrap-s texture
Return the method used for wrapping texture coordinates along the X
axis for @var{texture}.

Possible wrapping methods:
@itemize
@item @code{repeat}
@item @code{clamp}
@item @code{clamp-to-border}
@item @code{clamp-to-edge}
@end itemize

@end deffn

@deffn {Procedure} texture-wrap-t texture
Return the method used for wrapping texture coordinates along the Y
axis for @var{texture}.
@end deffn

@deffn {Procedure} texture-width texture
Return the width of @var{texture} in pixels.
@end deffn

@deffn {Procedure} texture-height texture
Return the height of @var{texture} in pixels.
@end deffn

It is common practice to combine multiple bitmap images into a single
texture, known as a ``tile atlas'' or ``tile set'', because it is more
efficient to render many regions of a large texture than it is to
render a bunch of small textures.  Chickadee provides a tile atlas
data type for collecting texture regions into a single vector.

@deffn {Procedure} split-texture texture tile-width tile-height @
                                 [#:margin 0] [#:spacing 0]

Return a new texture atlas that splits @var{texture} into a grid of
@var{tile-width} by @var{tile-height} rectangles.  Optionally, each
tile may have @var{spacing} pixels of horizontal and vertical space
between surrounding tiles and the entire image may have @var{margin}
pixels of empty space around its border.

This type of texture atlas layout is very common for 2D tile maps.
@xref{Tile Maps} for more information.
@end deffn

@deffn {Procedure} load-tileset file-name tile-width tile-height @
                                [#:margin 0] [#:spacing 0] @
                                [#:min-filter nearest] @
                                [#:mag-filter nearest] [#:wrap-s repeat] [#:wrap-t repeat] @
                                [#:transparent-color]

Return a new texture atlas that splits the texture loaded from the
file @var{file-name} into a grid of @var{tile-width} by
@var{tile-height} rectangles.  See @code{load-image} and
@code{split-texture} for information about all keyword arguments.
@end deffn

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

@deffn {Procedure} texture-atlas-texture atlas
Return the texture that all texture regions in @var{atlas} have been created from.
@end deffn

@deffn {Procedure} texture-atlas-ref atlas index
Return the texture region in @var{atlas} at @var{index}.
@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, the @code{(chickadee graphics sprite)} module provides the
interface for working with sprites.  Bitmaps are stored in textures
(@pxref{Textures}) and can be used to draw sprites via the
@code{draw-sprite} procedure.

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

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.

@var{tint} specifies the color to multiply against all the sprite's
pixels.  By default white is used, which does no tinting at all.

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}.
@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}.  This is fine
for rendering a few dozen sprites, but will become a serious
bottleneck when rendering hundreds or thousands of sprites.  To deal
with this, a technique known as ``sprite batching'' is used.  Instead
of drawing each sprite immediately, the sprite batch will build up a
large of buffer of sprites to draw and send them to the GPU all at
once.  There is one caveat, however.  Batching only works if the
sprites being drawn share a common texture.  A good strategy for
reducing the number of different textures is to stuff many bitmaps
into a single image file and create a ``texture atlas''
(@pxref{Textures}) to access the sub-images within.

@deffn {Procedure} make-sprite-batch texture [#:capacity 256]
Create a new sprite batch for @var{texture} with initial space for
@var{capacity} sprites.  Sprite batches automatically resize when they
are full to accomodate as many sprites as necessary.
@end deffn

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

@deffn {Procedure} sprite-batch-texture batch
Return the texture for @var{batch}.
@end deffn

@deffn {Procedure} set-sprite-batch-texture! batch texture
Set texture for @var{batch} to @var{texture}.
@end deffn

@deffn {Procedure} sprite-batch-add! batch position @
                   [#:origin] [#:scale] [:rotation] @
                   [#:tint @code{white}] [#:texture-region]

Add sprite located at @var{position} to @var{batch}.

To render a subsection of the batch's texture, a texture object whose
parent is the batch texture may be specified as @var{texture-region}.

See @code{draw-sprite} for information about the other arguments.
@end deffn

@deffn {Procedure} sprite-batch-clear! batch
Reset size of @var{batch} to 0.
@end deffn

@deffn {Procedure} draw-sprite-batch batch [#:blend-mode @code{alpha}]
Render @var{batch} using @var{blend-mode}.  Alpha blending is used by
default.
@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 dividing 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 texture rect @
       [#:margin 0] [#:top-margin margin] [#:bottom-margin margin] @
       [#:left-margin margin] [#:right-margin margin] @
       [#:origin] [#:scale] [#:rotation] [#:blend-mode alpha] @
       [#:tint white]

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 Fonts
@subsection Fonts

Printing text to the screen is quite easy:

@example
(draw-text "Hello, world" (vec2 100.0 100.0))
@end example

Chickadee supports OpenType/TrueType fonts (via the FreeType library),
bitmap fonts in Angel Code bmfont format, and simple sprite sheet
bitmap fonts.  A default font named Inconsolata is used for all text
rendering operations where a font is not specified, as is the case in
the above example.

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

@deffn {Procedure} load-font file-name point-size [#:char-set]
Load the scalable (OpenType, TrueType, etc.) font in the file
@var{file-name} and display it at the given @var{point-size}.  By
default, all the characters in the ASCII character set are loaded.
This can be changed by passing a different character set
(@pxref{Character Sets,,, guile, GNU Guile Reference Manual}) using
the @var{char-set} keyword argument.
@end deffn

@deffn {Procedure} load-bitmap-font file
Load the Angel Code font (in either XML or FNT format) in @var{file}
and return a new font object.
@end deffn

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

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

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

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

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

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

@deffn {Procedure} draw-text text position
       [#:font] [#:color] [#: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}.  If @var{font} is not specified, a
built-in font is used.

@example
(draw-text "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 Vector Paths
@subsection Vector Paths

The @code{(chickadee graphics path)} module can be used to draw lines,
curves, circles, rectangles, and more in a scalable, resolution
independent manner.  It is @emph{not} an SVG compliant renderer, nor
does it intend to be.  However, those familiar with SVG and/or the
HTML5 Canvas API should find lots of similarities.

@emph{This API is considered to be experimental and may change
substantially in future releases of Chickadee.  There are many missing
features such as gradient fills and dashed strokes.}

The first step to rendering vector graphics is to create a
@emph{path}: A series of commands that can be thought of as moving a
pen around a piece of paper.  A path can be either open or closed.  A
closed path draws a straight line from the last point in the path to
the first.

@deffn {Procedure} path . commands
Return a new path that follows @var{commands}.

@example
(path (move-to (vec2 50.0 50.0))
      (line-to (vec2 500.0 50.0))
      (line-to (vec2 400.0 200.0))
      (bezier-to (vec2 500.0 250.0) (vec2 380.0 300.0) (vec2 400.0 400.0))
      (line-to (vec2 300.0 400.0))
      (close-path))
@end example

@end deffn

Available drawing commands:

@deffn {Procedure} move-to point
Pick up the pen and move it to @var{point}.
@end deffn

@deffn {Procedure} line-to point
Draw a line from the current pen position to @var{point}.
@end deffn

@deffn {Procedure} bezier-to control1 control2 point
Draw a cubic bezier curve from the current pen position to
@var{point}.  The shape of the curve is determined by the two control
points: @var{control1} and @var{control2}.
@end deffn

@deffn {Procedure} close-path
Draw a straight line back to the first point drawn in the path.
@end deffn

@deffn {Procedure} arc center rx ry angle-start angle-end
Draw an elliptical arc spanning the angle range [@var{angle-start},
@var{angle-end}], centered at @var{center} with radii @var{rx} and
@var{ry} (set both to the same value for a circular arc.)
@end deffn

Included are some helpful procedures for generating common types of
paths:

@deffn {Procedure} line start end
Return a path that draws a straight line from @var{start} to @var{end}.
@end deffn

@deffn {Procedure} polyline . points
Return a path that draws a series of lines connecting @var{points}.
@end deffn

@deffn {Procedure} rectangle bottom-left width height
Return a path that draws a rectangle whose bottom-left corner is at
@var{bottom-left} and whose size is defined by @var{width} and
@var{height}.
@end deffn

@deffn {Procedure} square bottom-left size
Return a path draws a square whose bottom-left corner is at
@var{bottom-left} and whose size is defined by @var{size}.
@end deffn

@deffn {Procedure} rounded-rectangle bottom-left width height @
                   [#:radius 4.0] [#:radius-bottom-left] @
                   [#:radius-bottom-right] [#:radius-top-left] @
                   [#:radius-top-right]

Return a path that draws a rectangle with rounded corners whose
bottom-left corner is at @var{bottom-left} and whose size is defined
by @var{width} and @var{height}.  The argument @var{radius} is used to
define the corner radius for all corners.  To use a different radius
value for a corner, use @var{radius-bottom-left},
@var{radius-bottom-right}, @var{radius-top-left}, and/or
@var{radius-top-right}.
@end deffn

@deffn {Procedure} regular-polygon center num-sides radius
Return a path that draws a regular polygon with @var{num-sides} sides
centered on the point @var{center} with each vertex @var{radius} units
away from the center.
@end deffn

@deffn {Procedure} ellipse center rx ry
Return a path that draws an ellipsed centered on the point
@var{center} with radii @var{rx} and @var{ry}.
@end deffn

@deffn {Procedure} circle center r
Return a path that draws a circle centered on the point @var{center}
with radius @var{r}.
@end deffn

With one or more paths created, a @emph{painter} is needed to give the
path its style and placement in the final picture.  Painters can be
combined together to form arbitrarily complex pictures.

@deffn {Procedure} stroke . paths
Apply a stroked drawing style to @var{paths}.
@end deffn

@deffn {Procedure} fill . paths
Apply a filled drawing style to @var{paths}.
@end deffn

@deffn {Procedure} fill-and-stroke . paths
Apply a filled and stroked drawing style to @var{paths}.
@end deffn

@deffn {Procedure} transform matrix painter
Apply @var{matrix}, a 3x3 transformation matrix, to @var{painter}.
@end deffn

@deffn {Procedure} translate v painter
Translate @var{painter} by the 2D vector @var{v}.
@end deffn

@deffn {Procedure} rotate angle painter
Rotate @var{painter} by @var{angle} radians.
@end deffn

@deffn {Procedure} scale x painter
Scale @var{painter} by the scalar @var{x}.
@end deffn

@deffn {Procedure} pad pad-x pad-y painter
Add @var{pad-x} and @var{pad-y} amount of empty space around
@var{painter}.
@end deffn

@deffn {Procedure} superimpose . painters
Stack @var{painters} on top of each other.
@end deffn

@deffn {Procedure} beside . painters
Place @var{painters} next to each other in a row.
@end deffn

@deffn {Procedure} below . painters
Place @var{painters} next to each other in a column.
@end deffn

@deffn {Syntax} with-style ((style-name value) ...) painter
Apply all the given style settings to @var{painter}.

Possible style attributes are:

@itemize
@item blend-mode
@item fill-color
@item stroke-color
@item stroke-width
@item stroke-feather
@item stroke-cap
@end itemize

@example
(with-style ((stroke-color green)
             (stroke-width 4.0))
  (stroke (circle (vec2 100.0 100.0) 50.0)))
@end example

@end deffn

As in real life, a painter cannot paint anything without a canvas.
Once a painter has been associated with a canvas, it can finally be
rendered to the screen.

@deffn {Procedure} make-canvas painter [#:matrix]
Return a new canvas that will @var{painter} will draw on.  Optionally,
a 3x3 @var{matrix} may be specified to apply an arbitrary
transformation to the resulting image.
@end deffn

@deffn {Procedure} make-empty-canvas [#:matrix]
Return a new canvas that no painter is using. Optionally, a 3x3
@var{matrix} may be specified to apply an arbitrary transformation to
the image, should a painter later be associated with this canvas.
@end deffn

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

@deffn {Procedure} set-canvas-painter! canvas painter
Associate @var{painter} with @var{canvas}.
@end deffn

@deffn {Procedure} set-canvas-matrix! canvas matrix
Set the 3x3 transformation matrix of @var{canvas} to @var{matrix}.
@end deffn

@deffn {Procedure} draw-canvas canvas
Render @var{canvas} to the screen.
@end deffn

@node Particles
@subsection Particles

Effects like smoke, fire, sparks, etc. are often achieved by animating
lots of little, short-lived sprites known as ``particles''.  In fact,
all of these effects, and more, can be accomplished by turning a few
configuration knobs in a ``particle system''.  A particle system takes
care of managing the many miniscule moving morsels so the developer
can quickly produce an effect and move on with their life.  The
@code{(chickadee graphics particles)} module provides an API for
manipulating particle systems.

Below is an example of a very simple particle system that utilizes
nearly all of the default configuration settings:

@example
(use-modules (chickadee graphics particles))
(define texture (load-image "particle.png"))
(define particles (make-particles 2000 #:texture texture))
@end example

In order to put particles into a particle system, a particle
``emitter'' is needed.  Emitters know where to spawn new particles,
how many of them to spawn, and for how long they should do it.

Below is an example of an emitter that spawns 16 particles per frame
at the coordinates @code{(320, 240)}:

@example
(use-modules (chickadee math vector))
(define emitter (make-particle-emitter (vec2 320.0 240.0) 16))
(add-particle-emitter particles emitter)
@end example

To see all of the tweakable knobs and switches, read on!

@deffn {Procedure} make-particles capacity [#:blend-mode @code{alpha}] @
       [#:color white] [#:end-color transparent] [#:texture] @
       [#:animation-rows 1] [#:animation-columns 1] [#:width] [#:height] @
       [#:speed-range (vec2 0.1 1.0)] [#:acceleration-range (vec2 0.0 0.1)] @
       [#:direction-range (vec2 0 (* 2 pi))] [#:lifetime 30] [#:sort]

Return a new particle system that may contain up to @var{capacity}
particles.  Achieving the desired particle effect involves tweaking
the following keyword arguments as needed:

- @var{blend-mode}: Pixel blending mode.  @code{alpha} by default.
(@pxref{Blending} for more about blend modes).

- @var{start-color}: The tint color of the particle at the beginning of its
life.  White by default.

- @var{end-color}: The tint color of the particle at the end of of its
life.  Completely transparent by default for a fade-out effect.  The
color in the middle of a particle's life will be an interpolation of
@var{start-color} and @var{end-color}.

- @var{texture}: The texture applied to the particles.  The texture
may be subdivided into many animation frames.

- @var{animation-rows}: How many animation frame rows there are in the
texture.  Default is 1.

- @var{animation-columns}: How many animation frame columns there are
in the texture.  Default is 1.

- @var{width}: The width of each particle.  By default, the width of
an animation frame (in pixels) is used.

- @var{height}: The height of each particle.  By default, the height
of an animation frame (in pixels) is used.

- @var{speed-range}: A 2D vector containing the min and max particle
speed.  Each particle will have a speed chosen at random from this
range.  By default, speed ranges from 0.1 to 1.0.

- @var{acceleration-range}: A 2D vector containing the min and max
particle acceleration.  Each particle will have an acceleration chosen
at random from this range.  By default, acceleration ranges from 0.0
to 0.1.

- @var{direction-range}: A 2D vector containing the min and max
particle direction as an angle in radians.  Each particle will have a
direction chosen at random from this range.  By default, the range
covers all possible angles.

- @var{lifetime}: How long each particle lives, measured in
updates. 30 by default.

- @var{sort}: @code{youngest} if youngest particle should be drawn
last or @code{oldest} for the reverse.  By default, no sorting is
applied at all.
@end deffn

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

@deffn {Procedure} update-particles particles
Advance the simulation of @var{particles}.
@end deffn

@deffn {Procedure} draw-particles particles
Render @var{particles}.
@end deffn

@deffn {Procedure} draw-particles* particles matrix
Render @var{particles} with @var{matrix} applied.
@end deffn

@deffn {Procedure} make-particle-emitter spawn-area rate [duration]

Return a new particle emitter that spawns @var{rate} particles per
frame within @var{spawn-area} (a rectangle or 2D vector) for
@var{duration} frames.  If @var{duration} is not specified, the
emitter will spawn particles indefinitely.
@end deffn

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

@deffn {Procedure} particle-emitter-spawn-area emitter
Return the spawn area for @var{emitter}.
@end deffn

@deffn {Procedure} particle-emitter-rate emitter
Return the number of particles that @var{emitter} will spawn per
frame.
@end deffn

@deffn {Procedure} particle-emitter-life emitter
Return the number of frames remaining in @var{emitter}'s lifespan.
@end deffn

@deffn {Procedure} particle-emitter-done? emitter
Return @code{#t} if @var{emitter} has finished spawning particlces.
@end deffn

@deffn {Procedure} add-particle-emitter particles emitter
Add @var{emitter} to @var{particles}.
@end deffn

@deffn {Procedure} remove-particle-emitter particles emitter
Remove @var{emitter} to @var{particles}
@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
graphics tiled)} module.

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

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

@deffn {Procedure} tile-map-orientation tile-map
Return the orientation of @var{tile-map}.
@end deffn

@deffn {Procedure} tile-map-width tile-map
Return the width of @var{tile-map} in tiles.
@end deffn

@deffn {Procedure} tile-map-height tile-map
Return the height of @var{tile-map} in tiles.
@end deffn

@deffn {Procedure} tile-map-tile-width tile-map
Return the width of tiles in @var{tile-map}.
@end deffn

@deffn {Procedure} tile-map-tile-height tile-map
Return the height of tiles in @var{tile-map}.
@end deffn

@deffn {Procedure} tile-map-tilesets tile-map
Return the tilesets for @var{tile-map}.
@end deffn

@deffn {Procedure} tile-map-layers tile-map
Return the layers of @var{tile-map}.
@end deffn

@deffn {Procedure} tile-map-properties tile-map
Return the custom properties of @var{tile-map}.
@end deffn

@deffn {Procedure} point->tile tile-map x y
Translate the pixel coordinates (@var{x}, @var{y}) into tile
coordinates.
@end deffn

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

Draw the layers of @var{tile-map} as viewed from @var{camera}, a 2D
vector offset.  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.
@end deffn

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

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

@deffn {Procedure} tileset-first-gid tileset
Return the starting GID of @var{tileset}.
@end deffn

@deffn {Procedure} tileset-size tileset
Return the number of tiles in @var{tileset}.
@end deffn

@deffn {Procedure} tileset-tile-width tileset
Return the width of tiles in @var{tileset}.
@end deffn

@deffn {Procedure} tileset-tile-height tileset
Return the height of tiles in @var{tileset}.
@end deffn

@deffn {Procedure} tileset-atlas tileset
Return the texture atlas for @var{tileset}.
@end deffn

@deffn {Procedure} tileset-tiles tileset
Return the tiles in @var{tileset}.
@end deffn

@deffn {Procedure} tileset-properties tileset
Return the custom properties of @var{tileset}.
@end deffn

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

@deffn {Procedure} tile-id tile
Return the ID of @var{tile}.
@end deffn

@deffn {Procedure} tile-animation tile
Return the animation for @var{tile}.
@end deffn

@deffn {Procedure} tile-properties tile
Return the custom properties of @var{tile}.
@end deffn

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

@deffn {Procedure} animation-frame-tile frame
Return the tile for @var{frame}.
@end deffn

@deffn {Procedure} animation-frame-duration frame
Return the duration of @var{frame}.
@end deffn

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

@deffn {Procedure} tile-layer-name layer
Return the name of @var{layer}.
@end deffn

@deffn {Procedure} tile-layer-width layer
Return the width in tiles of @var{layer}.
@end deffn

@deffn {Procedure} tile-layer-height layer
Return the height in tiles of @var{layer}.
@end deffn

@deffn {Procedure} tile-layer-tiles layer
Return the tile data for @var{layer}.
@end deffn

@deffn {Procedure} tile-layer-properties layer
Return the custom properties of @var{layer}.
@end deffn

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

@deffn {Procedure} object-layer-name layer
Return the name of @var{layer}.
@end deffn

@deffn {Procedure} object-layer-objects layer
Return the objects for @var{layer}.
@end deffn

@deffn {Procedure} object-layer-properties layer
Return the custom properties of @var{layer}.
@end deffn

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

@deffn {Procedure} map-object-id obj
Return the ID of @var{obj}.
@end deffn

@deffn {Procedure} map-object-name obj
Return the name of @var{obj}.
@end deffn

@deffn {Procedure} map-object-type obj
Return the type of @var{obj}.
@end deffn

@deffn {Procedure} map-object-shape obj
Return the shape of @var{obj}.
@end deffn

@deffn {Procedure} map-object-properties obj
Return the custom properties of @var{obj}.
@end deffn

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

@deffn {Procedure} polygon-points polygon
Return the list of points that form @var{polygon}.
@end deffn

@node 3D Models
@subsection 3D Models

@emph{Disclaimer: Chickadee is alpha software, but 3D model support is
even more alpha than that.  There are many missing features in both
the model loading and rendering components, so set your expectations
accordingly!}

Chickadee can load and render 3D models in the classic OBJ and more
modern glTF 2.0 formats.

Here's some basic boilerplate to render a 3D model:

@example
(use-modules (chickadee)
             (chickadee math)
             (chickadee math matrix)
             (chickadee graphics model))

(define model #f)
(define projection-matrix
  (perspective-projection (/ pi 3.0) (/ 4.0 3.0) 0.1 500.0))
;; Adjust these 2 matrices so that you can actually see the model.
(define view-matrix (make-identity-matrix4))
(define model-matrix (make-identity-matrix4))

(define (load)
  (set! model (load-obj "model.obj"))

(define (draw alpha)
  (with-projection projection-matrix
    (with-depth-test #t
      (draw-model model model-matrix view-matrix))))

(run-game #:load load #:draw draw)
@end example

@deffn {Procedure} load-obj file-name
Load the OBJ formatted model in @var{file-name} and return a 3D model
object.

OBJ models are rendered using a Phong lighting model, which is a
work-in-progress.
@end deffn

@deffn {Procedure} load-gltf file-name
Load the glTF 2.0 formatted model in @var{file-name} and return a 3D
model object.

glTF models are rendered using a physically based lighting model,
which is currently a stub to be implemented later.
@end deffn

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

@deffn {Procedure} draw-model model model-matrix view-matrix
Render @var{model} with the transformation matrices @var{model-matrix}
and @var{view-matrix} applied.
@end deffn

@node Blending
@subsection Blending

Rendering a scene often involves drawing layers of objects that
overlap each other.  Blending determines how two overlapping pixels
are combined in the final image that is rendered to the screen.
Chickadee provides the following blend modes:

@itemize

@item @code{replace}
Use the latest color, ignoring all others.

@item @code{alpha}
Blend pixels according to the values of their alpha channels.  This is
the most commonly used blend mode and thus is Chickadee's default
mode.

@item @code{add}
Add all pixel color values together.  The more colors blended
together, the more white the final color becomes.

@item @code{subtract}
Subtract all pixel color values.  The more colors blended together,
the more black the final color becomes.

@item @code{multiply}

@item @code{darken}

@item @code{lighten}

@item @code{screen}

@end itemize

@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 width height @
                                    [#:min-filter @code{linear}] @
                                    [#:mag-filter @code{linear}] @
                                    [#:wrap-s @code{repeat}] @
                                    [#:wrap-t @code{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? obj
Return @code{#t} if @var{obj} is a framebuffer.
@end deffn

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

@deffn {Procedure} framebuffer-viewport 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

A viewport represents a subset of the screen (or framebuffer).  When
rendering a frame, the resulting image will only appear within that
viewport.  These aren't often needed, and Chickadee's default viewport
occupies the entire screen, but there are certain situations where
they are useful.  For example, a split-screen multiplayer game may
render to two different viewports, each occupying a different half of
the screen.  For information about how to set the current viewport,
see @code{with-viewport} in @ref{Rendering Engine}.

The @code{(chickadee graphics viewport)} module provides the following
API:

@deffn {Procedure} make-viewport x y width height @
                                 [#:clear-color] [#:clear-flags]

Create a viewport that covers an area of the window starting from
coordinates (@var{x}, @var{y}) and spanning @var{width} @code{x}
@var{height} pixels.  Fill the viewport with @var{clear-color} when
clearing the screen.  Clear the buffers denoted by the list of symbols
in @var{clear-flags}.

Possible values for @var{clear-flags} are @var{color-buffer},
@var{depth-buffer}, @var{accum-buffer}, and @var{stencil-buffer}.
@end deffn

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

@deffn {Procedure} viewport-x viewport
Return the left edge of @var{viewport}.
@end deffn

@deffn {Procedure} viewport-y viewport
Return the bottom edge of @var{viewport}.
@end deffn

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

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

@deffn {Procedure} viewport-clear-color viewport
Return the clear color for @var{viewport}.
@end deffn

@deffn {Procedure} viewport-clear-flags viewport
Return the list of clear flags for @var{viewport}.
@end deffn

@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{Buffers}) 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 graphics)} module.

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

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 {Syntax} gpu-apply/instanced shader vertex-array @
       n [#:uniform-key uniform-value @dots{}]
@deffnx {Syntax} gpu-apply/instanced shader vertex-array @
        count n [#:uniform-key uniform-value @dots{}]

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

Instanced rendering is very beneficial for rendering the same object
many times with only small differences for each one.  For example, the
particle effects described in @ref{Particles} use instanced rendering.

While @code{gpu-apply/instanced} 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}).
@end deffn

@deffn {Procedure} current-depth-test
Return @code{#t} if depth testing is currently enabled (@pxref{Blending}).
@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 viewport body @dots{}
Evaluate @var{body} with the current viewport bound to @var{viewport} (@pxref{Viewports}).
@end deffn

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

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

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

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

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

@node Buffers
@subsection Buffers

Alright, let's brush aside all of those pretty high level abstractions
and discuss what is going on under the hood.  The GPU exists as a
discrete piece of hardware separate from the CPU.  In order to make it
draw things, we must ship lots of data out of our memory space and
into the GPU.  The @code{(chickadee graphics buffer}) module provides an
API for manipulating GPU buffers.

In OpenGL terminology, a chunk of data allocated on the GPU is a
``vertex buffer object'' or VBO.  For example, here is a bytevector
that could be transformed into a GPU buffer that packs together vertex
position and texture coordinates:

@example
(use-modules (chickadee graphics buffer) (srfi srfi-4))
(define data
  (f32vector -8.0 -8.0 ; 2D vertex
             0.0 0.0   ; 2D texture coordinate
             8.0 -8.0  ; 2D vertex
             1.0 0.0   ; 2D texture coordinate
             8.0 8.0   ; 2D vertex
             1.0 1.0   ; 2D texture coordinate
             -8.0 8.0  ; 2D vertex
             0.0 1.0)) ; 2D texture coordinate
@end example

This data represents a textured 16x16 square centered on the
origin. To send this data to the GPU, the @code{make-buffer} procedure
is needed:

@example
(define buffer (make-buffer data #:stride 16))
@end example

The @code{#:stride} keyword argument indicates how many bytes make up
each element of the buffer.  In this case, there are 4 floats per
element: 2 for the vertex, and 2 for the texture coordinate.  A 32-bit
float is 4 bytes in length, so the buffer's stride is 16.

Within a VBO, one or more ``attributes'', as OpenGL calls them, may be
present.  Attributes are subregions within the buffer that have a
particular data type.  In this case, there are two attributes packed
into the buffer.  To provided a typed view into a buffer, the
@code{make-buffer-view} procedure is needed:

@example
(define vertices
  (make-buffer-view #:buffer buffer
                    #:type 'vec2
                    #:component-type 'float
                    #:length 4))
(define texcoords
  (make-buffer-view #:buffer buffer
                    #:type 'vec2
                    #:component-type 'float
                    #:length 4
                    #:offset 8))
@end example

To render a square, the GPU needs to draw two triangles, which means
we need 6 vertices in total.  However, the above buffer only contains
data for 4 vertices.  This is because there are only 4 unique vertices
for a square, but 2 of them must be repeated for each triangle.  To
work with deduplicated vertex data, an ``index buffer'' must be
created.

@example
(define index-buffer
  (make-buffer (u32vector 0 3 2 0 2 1)
               #:target 'index)
(define indices
  (make-buffer-view #:type 'scalar
                    #:component-type 'unsigned-int
                    #:buffer index-buffer))
@end example

Note the use of the @code{#:target} keyword argument.  It is required
because the GPU treats index data in a special way and must be told
which data is index data.

Now that the buffer views representing each attribute have been
created, all that's left is to bind them all together in a ``vertex
array object'', or VAO.  Vertex arrays associate each buffer view
with an attribute index on the GPU.  The indices that are chosen must
correspond with the indices that the shader (@pxref{Shaders}) expects
for each attribute.

@example
(define vertex-array
  (make-vertex-array #:indices indices
                     #:attributes `((0 . ,vertices)
                                    (1 . ,texcoords))))
@end example

With the vertex array created, the GPU is now fully aware of how to
interpret the data that it has been given in the original buffer.
Actually rendering this square is left as an exercise to the reader.
See the @ref{Shaders} section and the @code{gpu-apply} procedure in
@ref{Rendering Engine} for the remaining pieces of a successful draw
call.  Additionally, consider reading the source code for sprites,
shapes, or particles to see GPU buffers in action.

Without further ado, the API reference:

@deffn {Procedure} make-buffer data [#:name "anonymous"] @
       [#:length] [#:offset 0] [#:stride 0] [#:target @code{vertex}] @
       [#:usage @code{static}]

Upload @var{data}, a bytevector, to the GPU.  By default, the entire
bytevector is uploaded.  A subset of the data may be uploaded by
specifying the @var{offset}, the index of the first byte to be
uploaded, and @var{length}, the number of bytes to upload.

If @var{data} is @code{#f}, allocate @var{length} bytes of fresh GPU
memory instead.

@var{target} and @var{usage} are hints that tell the GPU how the
buffer is intended to be used.

@var{target} may be:

@itemize
@item @code{vertex}
Vertex attribute data.

@item @code{index}
Index buffer data.

@end itemize

@var{usage} may be:

@itemize
@item @code{static}
The buffer data will not be modified after creation.

@item @code{stream}
The buffer data will be modified frequently.

@end itemize

@var{name} is simply an arbitrary string for debugging purposes that
is never sent to the GPU.
@end deffn

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

@deffn {Procedure} index-buffer? buffer
Return @code{#t} if @var{buffer} is an index buffer.
@end deffn

@defvar null-buffer
Represents the absence of a buffer.
@end defvar

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

@deffn {Procedure} buffer-length buffer
Return the length of @var{buffer}.
@end deffn

@deffn {Procedure} buffer-stride buffer
Return the amount of space, in bytes, between each element in
@var{buffer}.
@end deffn

@deffn {Procedure} buffer-target buffer
Return the the intended usage of @var{buffer}, either @code{vertex} or
@code{index}.
@end deffn

@deffn {Procedure} buffer-usage buffer
Return the intended usage of @var{buffer}, either @code{static} for
buffer data that will not change once sent to the GPU, or
@code{stream} for buffer data that will be frequently updated from the
client-side.
@end deffn

@deffn {Procedure} buffer-data buffer
Return a bytevector containing all the data within @var{buffer}.  If
@var{buffer} has not been mapped (see @code{with-mapped-buffer}) then
this procedure will return @code{#f}.
@end deffn

@deffn {Syntax} with-mapped-buffer buffer body @dots{}
Evaluate @var{body} in the context of @var{buffer} having its data
synced from GPU memory to RAM.  In this context, @code{buffer-data}
will return a bytevector of all the data stored in @var{buffer}.  When
program execution exits this form, the data (including any
modifications) is synced back to the GPU.

This form is useful for streaming buffers that need to update their
contents dynamically, such as a sprite batch.
@end deffn

@deffn {Procedure} make-buffer-view #:buffer #:type @
       #:component-type #:length [#:offset @code{0}] [#:divisor @code{1}] @
       [#:name @code{"anonymous"}]

Return a new buffer view for @var{buffer} starting at byte index
@var{offset} of @var{length} elements, where each element is of
@var{type} and composed of @var{component-type} values.

Valid values for @var{type} are:

@itemize
@item @code{scalar}
single number

@item @code{vec2}
2D vector

@item @code{vec3}
3D vector

@item @code{vec4}
4D vector

@item @code{mat2}
2x2 matrix

@item @code{mat3}
3x3 matrix

@item @code{mat4}
4x4 matrix
@end itemize

Valid values for @var{component-type} are:

@itemize

@item @code{byte}
@item @code{unsigned-byte}
@item @code{short}
@item @code{unsigned-short}
@item @code{int}
@item @code{unsigned-int}
@item @code{float}
@item @code{double}

@end itemize

@var{divisor} is only needed for instanced rendering applications (see
@code{gpu-apply/instanced} in @ref{Rendering Engine}) and represents
how many instances each vertex element applies to.  A divisor of 0
means that a single element is used for every instance and is used for
the data being instanced.  A divisor of 1 means that each element is
used for 1 instance.  A divisor of 2 means that each element is used
for 2 instances, and so on.
@end deffn

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

@deffn {Procedure} buffer-view->buffer buffer-view
Return the buffer that @var{buffer-view} is using.
@end deffn

@deffn {Procedure} buffer-view-name buffer-view
Return the name of @var{buffer-view}.
@end deffn

@deffn {Procedure} buffer-view-offset buffer-view
Return the byte offset of @var{buffer-view}.
@end deffn

@deffn {Procedure} buffer-view-type buffer-view
Return the data type of @var{buffer-view}.
@end deffn

@deffn {Procedure} buffer-view-component-type buffer-view
Return the component data type of @var{buffer-view}
@end deffn

@deffn {Procedure} buffer-view-divisor buffer-view
Return the instance divisor for @var{buffer-view}.
@end deffn

@deffn {Syntax} with-mapped-buffer-view buffer-view body @dots{}

Evaluate @var{body} in the context of @var{buffer-view} having its
data synced from GPU memory to RAM.  See @code{with-mapped-buffer} for
more information.
@end deffn

@deffn {Procedure} make-vertex-array #:indices #:attributes @
       [#:mode @code{triangles}]

Return a new vertex array using the index data within the buffer view
@var{indices} and the vertex attribute data within @var{attributes}.

@var{attributes} is an alist mapping shader attribute indices to typed
buffers containing vertex data:

@example
`((1 . ,buffer-view-a)
  (2 . ,buffer-view-b)
  @dots{})
@end example

By default, the vertex array is interpreted as containing a series of
triangles.  If another primtive type is desired, the @var{mode}
keyword argument may be overridden.  The following values are
supported:

@itemize
@item @code{points}
@item @code{lines}
@item @code{line-loop}
@item @code{line-strip}
@item @code{triangles}
@item @code{triangle-strip}
@item @code{triangle-fan}
@end itemize

@end deffn

@defvar null-vertex-array
Represents the absence of a vertex array.
@end defvar

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

@deffn {Procedure} vertex-array-indices vertex-array
Return the buffer view containing index data for @var{vertex-array}.
@end deffn

@deffn {Procedure} vertex-array-attributes vertex-array
Return the attribute index -> buffer view mapping of vertex attribute
data for @var{vertex-array}.
@end deffn

@deffn {Procedure} vertex-array-mode vertex-array
Return the primitive rendering mode for @var{vertex-array}.
@end deffn

@node Shaders
@subsection Shaders

Shaders are programs that the GPU can evaluate that allow the
programmer to completely customized the final output of a GPU draw
call.  The @code{(chickadee graphics shader)} module provides an API for
building custom shaders.

Shaders are written in the OpenGL Shading Language, or GLSL for short.
Chickadee aspires to provide a domain specific language for writing
shaders in Scheme, but we are not there yet.

Shader programs consist of two components: A vertex shader and a
fragment shader.  A vertex shader receives vertex data (position
coordinates, texture coordinates, normals, etc.) and transforms them
as desired, whereas a fragment shader controls the color of each
pixel.

Sample vertex shader:

@example
@verbatim
#version 130

in vec2 position;
in vec2 tex;
out vec2 fragTex;
uniform mat4 mvp;

void main(void) {
    fragTex = tex;
    gl_Position = mvp * vec4(position.xy, 0.0, 1.0);
}
@end verbatim
@end example

Sample fragment shader:

@example
@verbatim
#version 130

in vec2 fragTex;
uniform sampler2D colorTexture;

void main (void) {
    gl_FragColor = texture2D(colorTexture, fragTex);
}
@end verbatim
@end example

This manual will not cover GLSL features and syntax as there is lots
of information already available about this topic.

One way to think about rendering with shaders, and the metaphor
Chickadee uses, is to think about it as a function call: The shader is
a function, and it is applied to some ``attributes'' (positional
arguments), and some ``uniforms'' (keyword arguments).

@example
(define my-shader (load-shader "vert.glsl" "frag.glsl"))
(define vertices (make-vertex-array @dots{}))
(gpu-apply my-shader vertices #:color red)
@end example

@xref{Rendering Engine} for more details about the @code{gpu-apply}
procedure.

Shaders are incredibly powerful tools, and there's more information
about them than we could ever fit into this manual, so we highly
recommend searching the web for more information and examples.  What
we can say, though, is how to use our API:

@deffn {Procedure} strings->shader vertex-source fragment-source
Compile @var{vertex-source}, the GLSL code for the vertex shader, and
@var{fragment-source}, the GLSL code for the fragment shader, into a
GPU shader program.
@end deffn

@deffn {Procedure} load-shader vertex-source-file fragment-source-file
Compile the GLSL source code within @var{vertex-source-file} and
@var{fragment-source-file} into a GPU shader program.
@end deffn

@deffn {Procedure} make-shader vertex-port fragment-port
Read GLSL source from @var{vertex-port} and @var{fragment-port} and
compile them into a GPU shader program.
@end deffn

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

@defvar null-shader
Represents the absence shader program.
@end defvar

@deffn {Procedure} shader-uniform shader name
Return the metadata for the uniform @var{name} in @var{shader}.
@end deffn

@deffn {Procedure} shader-uniforms shader
Return a hash table of uniforms for @var{shader}.
@end deffn

@deffn {Procedure} shader-attributes shader
Return a hash table of attributes for @var{shader}.
@end deffn

@deffn {Procedure} shader-uniform-set! shader uniform value
@end deffn

@subsubsection Attributes

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

@deffn {Procedure} attribute-name attribute
Return the variable name of @var{attribute}.
@end deffn

@deffn {Procedure} attribute-location attribute
Return the binding location of @var{attribute}.
@end deffn

@deffn {Procedure} attribute-type attribute
Return the data type of @var{attribute}.
@end deffn

@subsubsection Uniforms

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

@deffn {Procedure} uniform-name uniform
Return the variable name of @var{uniform}.
@end deffn

@deffn {Procedure} uniform-type uniform
Return the data type of @var{uniform}.
@end deffn

@deffn {Procedure} uniform-value uniform
Return the current value of @var{uniform}.
@end deffn

@subsubsection User-Defined Shader Types

The shader examples in this manual thus far have only shown uniforms
defined using primitive types.  However, GLSL shaders support
user-defined compound structs, such as this one:

@example
@verbatim
struct DirectionalLight {
    vec3 direction;
    vec3 ambient;
    vec3 diffuse;
    vec3 specular;
};

uniform DirectionalLight light;
@end verbatim
@end example

While @code{light} is declared as a single uniform in the shader code,
OpenGL translates this into @emph{four} uniforms in this case: One
uniform each member of the @code{DirectionalLight} struct.  This poses
a problem for sending Scheme data to the GPU.  How can compound Scheme
data translate into compound uniform data on the GPU?  The answer is
with shader types.  Shader types are a special kind of Guile struct
that provide a one-to-one mapping between a Scheme data structure and
a shader struct.

Some example code will explain this concept best.  Here is the Scheme
equivalent of the @code{DirectionalLight} struct:

@example
(define-shader-type <directional-light>
  make-directional-light
  directional-light?
  (float-vec3 direction directional-light-direction)
  (float-vec3 ambient directional-light-ambient)
  (float-vec3 diffuse directional-light-diffuse)
  (float-vec3 specular directional-light-specular)
  (float shininess directional-light-shininess))
@end example

The macro @code{define-shader-type} closely resembles the familiar
@code{define-record-type} from SRFI-9, but with one notable
difference: Each struct field contains type information.  The type
must be one of several primitive types (documented below) or another
shader type in the case of a nested structure.

It is important to note that the names of the shader type fields
@emph{must} match the names of the struct members in the GLSL code,
otherwise Chickadee will be unable to perform the proper translation.

As of this writing, this interface is new and experimental.  It
remains to be seen if this model is robust enough for all use-cases.

Primitive data types:

@defvar bool
Either @code{#t} or @code{#f}.
@end defvar

@defvar int
An integer.
@end defvar

@defvar unsigned-int
An unsigned integer.
@end defvar

@defvar float
A floating point number.
@end defvar

@defvar float-vec2
A 2D vector (@pxref{Vectors}.)
@end defvar

@defvar float-vec3
A 3D vector (@pxref{Vectors}.)
@end defvar

@defvar float-vec4
A color.
@end defvar

@defvar mat4
A matrix (@pxref{Matrices}.)
@end defvar

@defvar sampler-2d
A texture (@pxref{Textures}.)
@end defvar

@defvar local-field
A special type that means that the data is for the client-side
(Scheme-side) only and should not be sent to the GPU.  Any object may
be stored in a local field.
@end defvar

@deffn {Syntax} define-shader-type <name> constructor predicate @
                (field-type field-name [field-getter] [field-setter]) @dots{}

Define a new shader data type called @var{<name>}.

Instances of this data type are created by calling the
@var{constructor} procedure.  This procedure maps each field to a
keyword argument.  A shader data type with the fields @code{foo},
@code{bar}, and @code{baz} would have a constructor that accepts the
keyword arguments @code{#:foo}, @code{#:bar}, and @code{#:baz}.

A procedure named @var{predicate} will test if an object is a
@var{<name>} shader data type.

Fields follow the format @code{(field-type field-name [field-getter]
[field-setter])}.  @var{field-type} and @var{field-name} are required
for each field, but @var{field-getter} and @var{field-setter} are
optional.

@end deffn

@deffn {Procedure} shader-data-type? obj
Return @code{#t} if @var{obj} is a shader data type object.
@end deffn

@node Audio
@section Audio

A game isn't complete without sound.  Most games play some cool
background music to set the mood and have many sound effects to play
when events happen.  The @code{(chickadee audio)} module provides a
robust audio API backed by the OpenAL 3D audio system.

@menu
* Audio Files::                 Not audiophiles.
* Sources::                     Audio emitters.
* The Listener::                The player's ears.
* The Environment::             Global sound model settings.
@end menu

The basics of playing audio are very simple.  Just load an audio file
in the load hook (or anywhere else once the game loop is running) and
play it!

@example
(use-modules (chickadee audio))

(define audio #f)

(define (load)
  (set! audio (load-audio "neat-sound-effect.wav"))
  (audio-play audio))

(run-game #:load load)
@end example

For more advanced usage, check out the full API reference in the
following sections.

@node Audio Files
@subsection Audio Files

Sound data is represented by a special @code{<audio>} data type that
stores not only the audio samples themselves, but metadata such as
sample rate, number of channels, and how many bits are used for each
sample.

@deffn {Procedure} load-audio file-name [#:mode @code{static}]
Load audio within @var{file-name}.  The following file formats are
currently supported:

@itemize
@item WAV
@item MP3
@item Ogg Vorbis
@end itemize

Audio files can be loaded in two different ways, as indicated by
@var{mode}:

@itemize
@item static:
Load the entire audio file into memory.
@item stream:
Load chunks of the audio file as needed.
@end itemize

Generally speaking, sound effects don't take up much space and should
be loaded statically, but music files are much larger and should use
streaming.  Static loading is the default.
@end deffn

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

@deffn {Procedure} streaming-audio? @var{audio}
Return @code{#t} if @var{audio} uses stream loading.
@end deffn

@deffn {Procedure} static-audio?
Return @code{#t} if @var{audio} uses static loading.
@end deffn

@deffn {Procedure} audio-mode audio
Return the loading mode for @var{audio}, either @code{static} or
@code{stream}.
@end deffn

@deffn {Procedure} audio-duration audio
Return the duration of @var{audio} in seconds.
@end deffn

@deffn {Procedure} audio-bits-per-sample audio
Return the number of bits per sample in @var{audio}.
@end deffn

@deffn {Procedure} audio-channels audio
Return the number of channels in @var{audio}.
@end deffn

@deffn {Procedure} audio-sample-rate audio
Return the sample rate of @var{audio}.
@end deffn

@deffn {Procedure} audio-play audio [#:pitch 1.0] [#:volume 1.0] @
                   [#:min-volume 0.0] [#:max-volume 1.0] [#:max-distance] @
                   [#:reference-distance 0.0] [#:rolloff-factor 1.0] @
                   [#:cone-outer-volume 0.0] [#:cone-inner-angle 0.0] @
                   [#:cone-outer-angle] @
                   [#:position @code{(vec3 0.0 0.0 0.0)}] @
                   [#:velocity @code{(vec3 0.0 0.0 0.0)}] @
                   [#:direction @code{(vec3 0.0 0.0 0.0)}] @
                   [#:relative? @code{#f}]

Play @var{audio}.  There are many, many knobs to tweak that will
affect the sound that comes out of the player's speakers.:

@itemize
@item @var{pitch}:
Pitch multiplier.  The default value of 1.0 means no change in pitch.
@item @var{volume}:
Volume multiplier.  The default value of 1.0 means no change in volume.
@item @var{min-volume}:
Minimum volume.
@item @var{max-volume}:
Maximum volume.
@item @var{max-distance}:
Used with the inverse clamped distance model (the default model) to
set the distance where there will no longer be any attenuation of the
source.
@item @var{reference-distance}:
The distance where the volume for the audio would drop by half (before
being influenced by the rolloff factor or maximum distance.)
@item @var{rolloff-factor}:
For slowing down or speeding up the rate of attenuation.  The default
of 1.0 means no attenuation adjustment is made.
@item @var{cone-outer-volume}:
The volume when outside the oriented cone.
@item @var{cone-inner-angle}:
Inner angle of the sound cone, in radians.  The default value is 0.
@item @var{cone-outer-angle}:
Outer angle of the sound cone, in radians.  The default value is 2pi
radians, or 360 degrees.
@item @var{position}:
The source of the sound emitter in 3D space.
@item @var{velocity}:
The velocity of the sound emitter in 3D space.
@item @var{direction}:
The direction of the sound emitter in 3D space.
@item @var{relative?}:
A flag that determines whether the position is in absolute coordinates
or relative to the listener's location.  Absolute coordinates are used
by default.
@end itemize

For games with basic sound needs (that is to say they don't need 3D
sound modeling), the only things that really matter are @var{volume}
and @var{pitch}.

@end deffn

@node Sources
@subsection Sources

While the @code{audio-play} procedure provides a quick and convenient
way to play audio, it has some limitations.  What if the audio is a
long piece of music that might need to be paused or stopped later?
What if the audio should be looped?  What if the volume or pitch needs
to be altered?  For manipulating audio in these ways, a ``source'' is
required.  Sources can be thought of like a boombox: They sit
somewhere in the room and play sound.  The pause or stop buttons can
be pressed; it can be moved somewhere else; the volume knob can be
adjusted; the CD can be changed.

Sources are a great fit for handling background music, among other
things.  For quick, one-off sound effects, @code{audio-play} is a
better fit.

@deffn {Procedure} make-source audio loop? [#:pitch 1.0] [#:volume 1.0] @
                   [#:min-volume 0.0] [#:max-volume 1.0] [#:max-distance] @
                   [#:reference-distance 0.0] [#:rolloff-factor 1.0] @
                   [#:cone-outer-volume 0.0] [#:cone-inner-angle 0.0] @
                   [#:cone-outer-angle] @
                   [#:position @code{(vec3 0.0 0.0 0.0)}] @
                   [#:velocity @code{(vec3 0.0 0.0 0.0)}] @
                   [#:direction @code{(vec3 0.0 0.0 0.0)}] @
                   [#:relative? @code{#f}]

Return a new audio source.  @var{audio} is the audio data to use when
playing.  @var{loop?} specifies whether or not to loop the audio
during playback.

Refer to @code{audio-play} (@pxref{Audio Files}) for information about
the optional keyword arguments.
@end deffn

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

@deffn {Procedure} streaming-source? source
Return @code{#t} if @var{source} contains streaming audio.
@end deffn

@deffn {Procedure} static-source? source
Return @code{#t} if @var{source} contains static audio.
@end deffn

@deffn {Procedure} source-playing? source
Return @code{#t} if @var{source} is currently playing.
@end deffn

@deffn {Procedure} source-paused? source
Return @code{#t} if @var{source} is currently paused.
@end deffn

@deffn {Procedure} source-stopped? source
Return @code{#t} if @var{source} is currently stopped.
@end deffn

@deffn {Procedure} source-pitch source
Return the pitch multiplier of @var{source}.
@end deffn

@deffn {Procedure} source-volume source
Return the volume multiplier of @var{source}.
@end deffn

@deffn {Procedure} source-min-volume source
Return the minimum volume of @var{source}.
@end deffn

@deffn {Procedure} source-max-volume source
Return the maximum volume of @var{source}.
@end deffn

@deffn {Procedure} source-max-distance source
Return the maximum distance of @var{source}.
@end deffn

@deffn {Procedure} source-reference-distance source
Return the reference distance of @var{source}.
@end deffn

@deffn {Procedure} source-rolloff-factor source
Return the rolloff factor of @var{source}.
@end deffn

@deffn {Procedure} source-cone-outer-volume source
Return the volume of @var{source} when outside the oriented cone.
@end deffn

@deffn {Procedure} source-cone-inner-angle source
Return the inner angle of the sound cone of @var{source} in radians.
@end deffn

@deffn {Procedure} source-cone-outer-angle source
Return the outer angle of the sound cone of @var{source} in radians.
@end deffn

@deffn {Procedure} source-position source
Return the position of @var{source} as a 3D vector.
@end deffn

@deffn {Procedure} source-velocity source
Return the velocity of @var{source} as a 3D vector.
@end deffn

@deffn {Procedure} source-direction source
Return the direction of @var{source} as a 3D vector.
@end deffn

@deffn {Procedure} source-relative? source
Return @code{#t} if the position of @var{source} is relative to the
listener's position.
@end deffn

@deffn {Procedure} source-play source
Begin/resume playback of @var{source}.
@end deffn

@deffn {Procedure} source-pause source
Pause playback of @var{source}.
@end deffn

@deffn {Procedure} source-toggle source
Play @var{source} if it is currently paused or pause @var{source} if
it is currently playing.
@end deffn

@deffn {Procedure} source-stop [source]
Stop playing @var{source} or, if no source is specified, stop playing
@emph{all} sources.
@end deffn

@deffn {Procedure} source-rewind source
Rewind @var{source} to the beginning of the audio stream.
@end deffn

@deffn {Procedure} set-source-audio! source audio
Set the playback stream for @var{source} to @var{audio}.
@end deffn

@deffn {Procedure} set-source-loop! source loop?
Configure whether or not @var{source} should loop the audio stream.
@end deffn

@deffn {Procedure} set-source-pitch! source pitch
Set the pitch multiplier of @var{source} to @var{pitch}
@end deffn

@deffn {Procedure} set-source-volume! source volume
Set the volume of @var{source} to @var{volume}.  A value of 1.0 is
100% volume.
@end deffn

@deffn {Procedure} set-source-min-volume! source volume
Set the minimum volume of @var{source} to @var{volume}.
@end deffn

@deffn {Procedure} set-source-max-volume! source volume
Set the maximum volume of @var{source} to @var{volume}.
@end deffn

@deffn {Procedure} set-source-max-distance! source distance
Set the distance where there will no longer be any attenuation of
@var{source} to @var{distance}.
@end deffn

@deffn {Procedure} set-source-reference-distance! source distance
Set the reference distance of @var{source} to @var{distance}.
@end deffn

@deffn {Procedure} set-source-rolloff-factor! source factor
Set the rolloff factor for @var{source} to @var{factor}.
@end deffn

@deffn {Procedure} set-source-cone-outer-volume! source volume
Set the volume for @var{source} when outside the sound cone to @var{volume}.
@end deffn

@deffn {Procedure} set-source-cone-inner-angle! source angle
Set the inner angle of the sound cone of @var{source} to @var{angle} radians.
@end deffn

@deffn {Procedure} set-source-cone-outer-angle! source angle
Set the outer angle of the sound cone of @var{source} to @var{angle} radians.
@end deffn

@deffn {Procedure} set-source-position! source position
Set the position of @var{source} to the 3D vector @var{position}.
@end deffn

@deffn {Procedure} set-source-velocity! source velocity
Set the velocity of @var{source} to the 3D vector @var{velocity}.
@end deffn

@deffn {Procedure} set-source-direction! source direction
Set the velocity of @var{source} to the 3D vector @var{direction}.
@end deffn

@deffn {Procedure} set-source-relative! source relative?
If @var{relative?} is @code{#t}, the position of @var{source} is
interpreted as relative to the listener's position.  Otherwise, the
position of @var{source} is in absolute coordinates.
@end deffn

@node The Listener
@subsection The Listener

The listener is a collection of global state that represents the
player within the 3D sound model.  For games that do not need 3D sound
modeling, manipulating the listener's master volume is the only
interesting thing to do here.

@deffn {Procedure} listener-volume
Return the current master volume of the listener.
@end deffn

@deffn {Procedure} listener-position
Return the current position of the listener.
@end deffn

@deffn {Procedure} listener-velocity
Return the current velocity of the listener.
@end deffn

@deffn {Procedure} listener-orientation
Return the current orientation of the listener.
@end deffn

@deffn {Procedure} set-listener-volume! volume
Set the listener's master volume to @var{volume}, a value in the range [0,
1].
@end deffn

@deffn {Procedure} set-listener-position! position
Set the listener's position to the 3D vector @var{position}.
@end deffn

@deffn {Procedure} set-listener-velocity! velocity
Set the listener's velocity to the 3D vector @var{velocity}.
@end deffn

@deffn {Procedure} set-listener-orientation! at up
Set the listener's orientation to the 3D vectors @var{at} and
@var{up}.
@end deffn

@node The Environment
@subsection The Environment

The environment defines global parameters that govern how sound is
processed within the 3D modeling space.

@deffn {Procedure} doppler-factor
Return the current doppler factor.
@end deffn

@deffn {Procedure} speed-of-sound
Return the current speed of sound.
@end deffn

@deffn {Procedure} distance-model
Return the current distance model.

Possible return values are:

@itemize
@item @code{none}
@item @code{inverse-distance}
@item @code{inverse-distance-clamped} (the default)
@item @code{linear-distance}
@item @code{linear-distance-clamped}
@item @code{exponent-distance}
@item @code{exponent-distance-clamped}
@end itemize

@end deffn

@deffn {Procedure} set-doppler-factor! doppler-factor
Change the doppler factor to @var{doppler-factor}.
@end deffn

@deffn {Procedure} set-speed-of-sound! speed-of-sound
Change the speed of sound to @var{speed-of-sound}.
@end deffn

@deffn {Procedure} set-distance-model! distance-model
Change the distance model to @var{distance-model}.  Valid distance
models are:

@itemize
@item @code{none}
@item @code{inverse-distance}
@item @code{inverse-distance-clamped}
@item @code{linear-distance}
@item @code{linear-distance-clamped}
@item @code{exponent-distance}
@item @code{exponent-distance-clamped}
@end itemize

@end deffn

@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 procedure
(@pxref{Kernel}) to 1 unit of agenda time, like so:

@example
(define (update 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? obj
Return @code{#t} if @var{obj} is an agenda.
@end deffn

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

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

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

@deffn {Procedure} schedule-every interval thunk [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 time body @dots{}
Schedule @var{body} to be evaluated at @var{time}.
@end deffn

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

@deffn {Syntax} every interval body @dots{}
@deffnx {Syntax} every (interval n) 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

It is also possible to schedule things that are not dependent on how
much time passes.  The agenda will periodically poll to see if any
registered conditions are met.

@deffn {Procedure} call-when pred thunk
Call @var{thunk} sometime in the future when @var{pred} is satisfied
(returns a value other than @code{#f}.)
@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 thunk
Apply @var{thunk} as a script and return a handle to it.
@end deffn

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

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

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

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

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

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

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

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

@deffn {Syntax} wait-until condition
Wait until @var{condition} is met before resuming the current script.

@example
(script
  (wait-until (key-pressed? 'z))
  (display "you pressed the Z key!\n"))
@end example

@end deffn

@deffn {Syntax} forever 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 duration start end proc @
                         [#:step @code{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? obj
Return @code{#t} if @var{obj} is a channel.
@end deffn

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

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

A low-level API also exists for using channels outside of a script via
callback procedures:

@deffn {Procedure} channel-get! channel proc
Asynchronously retrieve a value from @var{channel} and call @var{proc}
with that value.
@end deffn

@deffn {Procedure} channel-put! channel data [thunk]
Asynchronously send @var{data} to @var{channel} and call @var{thunk}
after it has been received.
@end deffn