mutter-performance-source/README

Clutter - README
===============================================================================

Clutter is an open source software library for creating portable, fast,
visually rich and animated graphical user interfaces.

REQUIREMENTS
-------------------------------------------------------------------------------

Clutter currently requires:

  • GLib >= 2.18.0
  • Atk >= 1.17
  • Cairo >= 1.6
  • PangoCairo >= 1.20
  • OpenGL >= 1.3 or 1.2+multitexturing, OpenGL ES 1.1 or OpenGL ES 2.0
  • GLX, SDL, WGL or an EGL Implementation

Clutter also has optional dependencies:

  • JSON-GLib >= 0.10 (optional, see --with-json below)
  • GDK-Pixbuf >= 2.0 (optional, see --with-imagebackend below)

On X11, Clutter depends on the following extensions

  • XComposite
  • XDamage
  • XExt
  • XFixes
  • XInput 1.x (if --enable-xinput is passed to configure)

When running the OpenGL flavor, Clutter requires at least version 1.3
or 1.2 with the multitexturing extension. However to build Clutter
you will need the latest GL headers which can be obtained from:

  http://www.khronos.org

If you are building the API reference you will also need:

  • GTK-Doc >= 1.13

If you are building the additional documentation you will also need:

  • xsltproc
  • jw (optional, for generating PDFs)

If you are building the Introspection data you will also need:

  • GObject-Introspection >= 0.6.7

If you want support for profiling Clutter you will also need:

  • UProf >= 0.2

UProf is available from:

  git://git.moblin.org/uprof.git

RESOURCES
-------------------------------------------------------------------------------

The official Clutter website is:

   http://www.clutter-project.org/

The API reference for the latest stable release and unstable developers
snapshot are available at:

   http://www.clutter-project.org/docs/clutter/stable/
   http://www.clutter-project.org/docs/clutter/unstable/

New releases of Clutter are available at:

   http://www.clutter-project.org/sources/clutter/

The Clutter blog is at:

   http://www.clutter-project.org/blog/

To subscribe to the Clutter mailing list, send an email to:

   clutter+subscribe@o-hand.com

The official mailing list archive is:

   http://lists.o-hand.com/clutter/

New bug page on Bugzilla:

   http://bugzilla.o-hand.com/enter_bug.cgi?product=Clutter

Clutter is licensed under the terms of the GNU Lesser General Public
License, version 2.1 or (at your option) later.

BUILDING AND INSTALLATION
-------------------------------------------------------------------------------

See the INSTALL file.

Clutter has additional command line options for the configure script:

 --enable-debug=[no/minimum/yes]
	Controls Clutter debugging level:

	yes:
                All GLib asserts, checks and support for runtime Clutter
                debugging notes through CLUTTER_DEBUG. This is the default
                value for developers snapshots.

        minimum:
                Just GType cast checks and support for runtime Clutter
                debugging notes through CLUTTER_DEBUG. This is the default
                for stable releases.

        no:
                No GLib asserts or checks and no support for runtime Clutter
                debugging notes. Only use in extreme performance and/or size
                optimization cases, though it is strongly discouraged.

 --enable-cogl-debug=[no/minimum/yes]
        Controls COGL debugging level (default=minimum):

        yes:
                Support for COGL debugging notes through COGL_DEBUG and
                error checking for each GL primitive. This is useful mostly
                to debug COGL itself.

        minimum:
                Support for COGL debugging notes through COGL_DEBUG. This is
                the default for developers snapshots.

        no:
                Disable support for COGL runtime debugging notes.

 --enable-maintainer-flags=[no/yes]
        Use strict compiler flags. This defaults to 'yes' for developers
        snapshots and to 'no' for stable releases.

 --enable-gtk-doc
	use gtk-doc to build API documentation (default=no). Requires gtk-doc
	present on the target system.

 --enable-docs=[no/yes]
        Build additional documentation. Requires xsltproc for DocBook
        conversion, and optionally jw for PDF generation.

 --enable-gcov=[no/yes]
        Build Clutter with coverage report support, provided by gcov. This
        feature only works with the GNU Compiler Suite and gcov installed.

 --enable-profile=[no/yes]
        Build Clutter with profiling instrumentation. Requires the GNU
        C Compiler and the UProf library.

 --enable-conform=[yes/no]
        Build the Clutter conformance test suite.

 --with-flavour=[glx/eglx/eglnative/osx/win32/fruity]
        Select the Clutter backend: (default=glx)

	glx:
                Fully featured GLX backend. Using Open GL.

	eglx:
                EGL/Open GL ES backend for EGL on X windows implementations

	eglnative:
                EGL/Open GL ES backend on 'native windowing system' - i.e
		raw framebuffer. Expects the EGL implementation to provide
                a createNativeWindow() call. Also it optionally supports
		tslib for touchscreen events.

        osx:
                OS X backend. (EXPERIMENTAL)

        win32:
                Microsoft Windows(tm) WGL backend

        fruity:
                Apple iPod Touch(tm)/iPhone(tm) backend (EXPERIMENTAL)

 --with-imagebackend=[gdk-pixbuf/quartz/internal]
        Select the image loading backend used by COGL

        gdk-pixbuf:
                Depend on gdk-pixbuf-2.0 (default for the glx, eglx,
                eglnative, win32 flavours and recommended)

        quartz:
                Depend on CoreGraphics (default for the osx flavour)

        internal:
                Internal JPEG and PNG loader. Should only be used
                for testing on new platforms

 --with-gles=[1.1/2.0]
        Select the GLES version (for EGL backends) (default=1.1)

 --with-json=[internal/check/system]
        Select the JSON-GLib copy to use (default=check)

        internal:
                Use the internal copy of JSON-GLib for ClutterScript

        check:
                Check for the existence of a system copy of JSON-GLib
                and if it is available, make Clutter depend on it

        system:
                Only use the system copy of JSON-GLib and warn if not found

VERSIONING
-------------------------------------------------------------------------------

Clutter uses the common "Linux kernel" versioning system, where
even-numbered minor versions are stable and odd-numbered minor
versions are development snapshots.

Different major versions break both API and ABI but are parallel
installable. The same major version with differing minor version is
expected to be ABI compatible with other minor versions; differing
micro versions are meant just for bug fixing. On odd minor versions
the newly added API might still change.

The micro version indicates the origin of the release: even micro
numbers are only used for released archives; odd micro numbers are
only used on the Git repository.

HACKING
-------------------------------------------------------------------------------

If you want to hack on and improve Clutter, check the contained TODO file for
pending tasks, the HACKING file for general implementation guidelines, and the
HACKING.backends for backend-specific implementation issues.

The CODING_STYLE file contains the rules for writing code conformant to the
style guidelines used throughout Clutter. Remember: the coding style is
mandatory; patches not conforming to it will be rejected.

BUGS
-------------------------------------------------------------------------------

Bugs should be reported to the Clutter Bugzilla at:

  http://bugzilla.o-hand.com/enter_bug.cgi?product=Clutter

You will need a Bugzilla account.

In the report you should include:

  • what system you're running Clutter on;
  • which version of Clutter you are using;
  • which version of GLib and OpenGL (or OpenGL ES) you are using;
  • which video card and which drivers you are using, including output of
    glxinfo and xdpyinfo (if applicable);
  • how to reproduce the bug.

If you cannot reproduce the bug with one of the tests that come with Clutter
source code, you should include a small test case displaying the bad
behaviour.

If the bug exposes a crash, the exact text printed out and a stack trace
obtained using gdb are greatly appreciated.

CONTRIBUTING
-------------------------------------------------------------------------------

Patches should be submitted using Bugzilla. Patches fixing a bug should be
attached to the bug report; patches for new features or for fixing bugs not
yet reported should be attached to a newly opened bug. Patches should always
be in the unified diff format, using:

  diff -Nuarp clutter.source clutter.patched > clutter-patch.diff

If diffing against the Git repository, you should use:

  git diff > clutter-patch.diff

Or, better, commit locally and use `git format-patch` to generate a patch
containing authorship details, so that members of the Clutter development
team can credit your contribution properly.

When attaching a patch to Bugzilla you should also toggle the
'copyright-waived' attachment flag, after reading the Copyright Waiver notice
here:

  http://bugzilla.openedhand.com/waiver.html

If you do not intend to waive your copyright you should contact the Clutter
development team to arrange a suitable solution.

RELEASE NOTES
-------------------------------------------------------------------------------

Relevant information for developers with existing Clutter applications
wanting to port to newer releases (see NEWS for general information on new
features).

Release Notes for Clutter 1.4
-------------------------------------------------------------------------------

* ClutterLayoutManager sub-classes should chain up to the parent class's
  implementation.

Release Notes for Clutter 1.2
-------------------------------------------------------------------------------

* ClutterStageManager is now publicly available and documented API.

* Clutter now depends on the system copy of JSON-GLib, and will fall
  back to the internal copy only if JSON-GLib is not installed.

* ClutterActor:opacity is now defined using GParamSpecUint instead of
  GParamSpecUchar; the same interval of [ 0, 255 ] applies, and GValue
  has internal transformation functions for converting between G_TYPE_UINT
  and G_TYPE_UCHAR, so this change should be fully transparent to the
  user of the code.

* On X11 Clutter will emulate XKB's detectable key auto-repeat; this means
  that when holding down a key, Clutter will emit multiple CLUTTER_KEY_PRESS
  events and a single CLUTTER_KEY_RELEASE event instead of a list of
  CLUTTER_KEY_PRESS and CLUTTER_KEY_RELEASE pairs.

* On X11 and Win32 the default Stage is created when
  clutter_stage_get_default() is called for the first time, and not as
  part of the stage initialization.

Cogl API changes for Clutter 1.2
-------------------------------------------------------------------------------

* cogl_viewport is now deprecated in favour of cogl_set_viewport which
  accepts a viewport offset.

* cogl_clip_push() is now deprecated and new code should use
  cogl_clip_push_rectangle() instead. The old API wasn't consistent with other
  Cogl APIs that specify model space rectangles using (x0,y0)(x1,y1) pairs.

* cogl_clip_push_window_rect() is now deprecated and new code should use
  cogl_clip_push_window_rectangle(). The old API shouldn't have been defined
  to take floats and the abbreviation wasn't consistent with other Cogl API.

* cogl_clip_stack_save() and cogl_clip_stack_restore() are deprecated, as
  the functionality is redundant now that offscreen draw buffers own their
  clip state and switching to/from offscreen rendering will automatically
  save and restore the clip state.

* cogl_material_copy() was added. It is advised that developers use
  this instead of cogl_material_new() when creating a material that is in some
  way derived from another. This will allow Cogl to track material
  ancestries/similarities and reduce the cost of GPU state changes.

* cogl_push_draw_buffer, cogl_set_draw_buffer and cogl_pop_draw_buffer are now
  deprecated and new code should use the cogl_framebuffer_* API instead.
  Code that previously did:
    cogl_push_draw_buffer ();
    cogl_set_draw_buffer (COGL_OFFSCREEN_BUFFER, buffer);
    /* draw */
    cogl_pop_draw_buffer ();
  can now be re-written as:
    cogl_push_framebuffer (buffer);
    /* draw */
    cogl_pop_framebuffer ();

* All cogl_<type>_ref() and cogl_<type>_unref() functions have been
  deprecated, and superceded by cogl_handle_ref() and cogl_handle_unref()
  respectively.

* The cogl_check_extension() function has been deprecated. This function
  was never meant to be public, since it depends on calling glGetString()
  before its invocation. Users of this function can be replaced by the
  equivalent code:

    gl_ext = glGetString (GL_EXTENSIONS);
  - has_ext = cogl_check_extension (ext_name, gl_ext);
  + has_ext = strstr (gl_ext, ext_name) != NULL ? TRUE : FALSE;

Release Notes for Clutter 1.0
-------------------------------------------------------------------------------

* The clutter_actor_set_shader_param() function now takes a
  GValue, which can be set using the clutter_value_set_shader()
  family of functions. The floating point wrapper has been
  rename clutter_actor_set_shader_param_float() to match the newly
  added clutter_actor_set_shader_param_int().

* The Pango renderer API has been exposed as public API, after
  a full rename from PangoClutter to CoglPango, to avoid namespace
  collisions with upstream Pango. The Pango font map, renderer and
  glyph cache can be used by third party code and depend only on
  COGL.

* Both Clutter and COGL only allow including <clutter/clutter.h>
  and <cogl/cogl.h> directly, respectively. This allows avoiding
  breaking API every time a type definition is moved across
  headers, and improves the reliability of third party code against
  internal refactorings.

* COGL has an internal Color type, used to store a color definition
  that can be efficiently used with the least amount of conversions
  by both the GL and GLES implementations. The COGL API has been
  changed to drop the usage of ClutterColor in favour of CoglColor.

* The fixed point API implementation Clutter uses internally has been
  moved from the Clutter namespace to the COGL one.

* ClutterLabel and ClutterEntry have been removed from the API, as
  both have been superceded by the ClutterText actor.

* ClutterCloneTexture has been removed from the API; in its place,
  there is a generic ClutterClone actor which allows to "clone"
  any existing actors -- even composite ones -- without using
  frame buffer objects (FBOs).

* The ClutterEffectTemplate and clutter_effect_* functions have been
  superceded by ClutterAnimation and thus removed from the public API.

* The ClutterBehaviourBspline has been superceded by the usage of
  ClutterPath inside ClutterBehaviourPath, and thus removed from the
  public API.

* ClutterColor API has received a much needed review to increase its
  consistency. This has led to the following changes:

    - clutter_color_parse() has been renamed to clutter_color_from_string()
      and the order of the arguments has been changed

    - the factor argument of clutter_color_shade() has been swapped with
      the return location for the new color

    - the fixed point entry points have been removed

    - clutter_color_from_hls() and clutter_color_to_hls() do not
      normalize the values in the [ 0, 255 ] interval but use the
      correct HLS intervals:

         Hue: [ 0, 360 )
         Luminance: [ 0, 1 ]
         Saturation: [ 0, 1 ]

* The ClutterFixed symbols have been completely removed: fixed-point
  public entry points now take a CoglFixed.

* The -x and -u API have been removed. All the pixel-based API now
  takes a float to allow sub-pixel precision; this is true also for
  properties. WARNING: functions with variadic arguments (like
  g_object_set(), g_object_get() and clutter_actor_animate()) do not
  behave very well when dealing with integers instead of expected
  floating point values, and vice versa. On 32bit machines it will
  most likely lead to a crash. So:

    g_object_set (actor, "width", 100, NULL);

  is incorrect, and should be changed in:

    g_object_set (actor, "width", 100.0, NULL);

* Composite actors that do not implement the Container interface
  should implement the following virtual functions:

    - void map   (ClutterActor*)
    - void unmap (ClutterActor*)

  and chain up to the parent's implementation after calling
  clutter_actor_map() or clutter_actor_unmap() on their children.

* Actors implementing the Container interface that have private
  children that are not meant to be added/removed through the
  Container API should implement the:

    - void foreach_with_internals (ClutterContainer*,
                                   ClutterCallback,
                                   gpointer);

  virtual function.

* Actors that perform direct transformations using the COGL API inside
  their paint() implementation should override the apply_transform()
  virtual function instead. Implementations of the apply_transform()
  vfunc must chain up to the implementation of the parent class.

* The ClutterUnit type and the CLUTTER_UNITS_* conversion macros have
  been removed: all Actors are now using sub-pixel precision directly
  throughout the API. The new ClutterUnits type has been added as a
  generic opaque storage for logical distance values.

* Timelines are now fully time-based: all the frame-related properties
  and methods have been removed. ClutterTimeline::new-frame will provide
  the elapsed milliseconds since the beginning of the Timeline, instead
  of the current frame number. The clutter_timeline_new() constructor
  takes the duration of the timeline in milliseconds, and thus it replaces
  the clutter_timeline_new_for_duration() variant.

Cogl API changes for Clutter 1.0
-------------------------------------------------------------------------------

* All drawing functions now use a source material to determine how geometry is
  filled. The source material is set via cogl_set_source. Or the convenience
  functions cogl_set_source_color and cogl_set_source_texture.

  "drawing functions" include: cogl_rectangle, cogl_texture_rectangle,
  cogl_texture_multiple_rectangles, cogl_texture_polygon (though the
  cogl_texture_* funcs have been renamed; see below for details),
  cogl_path_fill/stroke and cogl_vertex_buffer_draw*.

  cogl_texture_rectangle, cogl_texture_multiple_rectangles and
  cogl_texture_polygon no longer take a texture handle; instead the current
  source material is referenced. The functions have also been renamed to:
  cogl_rectangle_with_texture_coords, cogl_rectangles_with_texture_coords
  and cogl_polygon respectively.

  Most code that previously did:
    cogl_texture_rectangle (tex_handle, x, y,...);
  needs to be changed to now do:
    cogl_set_source_texture (tex_handle);
    cogl_rectangle_with_texture_coords (x, y,....);

  In the less likely case where you were blending your source texture with a
  color like:
    cogl_set_source_color4ub (r,g,b,a); /* (where r,g,b,a isn't just white) */
    cogl_texture_rectangle (tex_handle, x, y,...);
  you will need your own material to do that:
    material = cogl_material_new ();
    cogl_material_set_color4ub (r,g,b,a);
    cogl_material_set_layer (material, 0, tex_handle));
    cogl_set_source_material (material);

  Code that uses the texture coordinates, 0, 0, 1, 1 don't need to use
  cogl_rectangle_with_texture_coords since these are the coordinates that
  cogl_rectangle will use.

  For cogl_texture_polygon; as well as dropping the texture handle, the
  n_vertices and vertices arguments were transposed for consistency. So
  code previously written as:
    cogl_texture_polygon (tex_handle, 3, verts, TRUE);
  need to be written as:
    cogl_set_source_texture (tex_handle);
    cogl_polygon (verts, 3, TRUE);

* The arguments to cogl_rectangle, cogl_path_rectangle and
  cogl_path_round_rectangle have been changed - for consistency - from
  x, y, width, height, to x1, y1, x2, y2.

* A CoglMatrix type and utility API has been added; this is currently used to
  support describing texture matrices.

* cogl_alpha_func has been removed, since this is now controlled using the
  material API via cogl_material_set_alpha_test_function ()

* A Cogl Vertex Buffer API has been added that allows you to efficiently
  manage arrays of vertex attributes in buffers that may be stored on
  the GPU. These allow you to avoid the costs of repeatedy validating
  vertex data and mapping it into the GPU.

* cogl_scale now supports scaling on the z axis

* cogl_clip_set* and cogl_clip_unset have been renamed to cogl_clip_push and
  cogl_clip_pop respectively so they self document their stacking semantics.

* cogl_paint_init was renamed to cogl_clear and no longer disables lighting and
  fogging. cogl_clear also now takes a mask of the auxiliary buffers you want
  to clear so you can avoid redundant clears of buffers you aren't using.

* cogl_fog_set was renamed to cogl_set_fog and it now takes a mode argument
  giving control over the fogging blend factor equation, so that the
  density argument isn't just ignored. A cogl_disable_fog function was
  also added.

* cogl_get_*_matrix were changed to use the CoglMatrix type instead of
  GLfloat m[16] arrays.

* cogl_offscreen_blit_region, cogl_offscreen_blit were removed since they were
  only implemnted for GL, not GLES, and it was assumed no one was using them.

* cogl_offscreen_new_multisample was removed since it only ever returned
  COGL_INVALID_HANDLE so it wasn't usefull.

* The COGL_MASK_BUFFER type was removed, since there should be nicer ways of
  exposing color mask if anyone wants it later. It was assumed that no one was
  using this currently.

* COGLenum, COGLint and COGLuint which were simply typedefs for
  GL{enum,int,uint} have been removed from the API and replaced with
  specialised enum typedefs, int and unsigned int. These were causing
  problems for generating bindings and also considered poor style.

* The cogl texture filter defines CGL_NEAREST and CGL_LINEAR etc are now
  replaced by a namespaced typedef 'CoglTextureFilter' so they should be
  replaced with COGL_TEXTURE_FILTER_NEAREST and COGL_TEXTURE_FILTER_LINEAR etc.

* The shader type defines CGL_VERTEX_SHADER and CGL_FRAGMENT_SHADER are handled
  by a CoglShaderType typedef and should be replaced with
  COGL_SHADER_TYPE_VERTEX and COGL_SHADER_TYPE_FRAGMENT.

* cogl_shader_get_parameteriv has been replaced by cogl_shader_get_type and
  cogl_shader_is_compiled. More getters can be added later if desired.

* cogl_enable_depth_test has been renamed to cogl_set_depth_test_enabled and
  a corresponding cogl_get_depth_test_enabled function has been added.

Release Notes for Clutter 0.8
-------------------------------------------------------------------------------

* The COGL GL wrapper API has been completely overhauled and now
  contains many new features including new greatly improved texture
  abstractions (slicing, mipmapping, deformations etc, greatly
  simplifying ClutterTexture), image loading and abstraction, path
  based primitive drawing, clipping, and improved FBO and shader
  support. It is now also fully documented.

* GL Texture Rectangle ext is no longer used, the regular 2D NPOTS
  extension is preffered instead but not required.

* Clutter now has basic suppport for multiple input devices assuming
  the backend supports it (currently X11 based via XInput and Fruity
  backends). New API supporting this includes
  clutter_event_get_device_id, clutter_get_input_device_for_id,
  clutter_grab_pointer_for_device & clutter_ungrab_pointer_for_device.

  XInput support needs to be explicitly enabled at runtime by calling
  clutter_x11_enable_xinput () before clutter_init. clutter_x11_has_xinput ()
  can then be called after to check if XInput extension present and use-able.

* The functions that return the transformed position of an actor have
  been renamed to be more explicit about it:

    clutter_actor_get_abs_position - clutter_actor_get_transformed_position
    clutter_actor_get_abs_size     - clutter_actor_get_transformed_size

  Their behaviour has not been changed.

* To increase portability, Clutter does not strictly depend on
  GdkPixbuf anymore; this means that you will have to include
  <gdk-pixbuf/gdk-pixbuf.h> yourself if you are operating with
  GdkPixbuf objects and not including that header. The GdkPixbuf-based
  API has been removed from Clutter core:

    clutter_texture_new_from_pixbuf
    clutter_texture_set_pixbuf
    clutter_texture_get_pixbuf

  are all deprecated functions. It is still possible to load a GdkPixbuf
  into a ClutterTexture with this sample code:

    clutter_texture_set_from_rgb_data (texture,
                                       gdk_pixbuf_get_pixels (pixbuf),
                                       gdk_pixbuf_get_has_alpha (pixbuf),
                                       gdk_pixbuf_get_width (pixbuf),
                                       gdk_pixbuf_get_height (pixbuf),
                                       gdk_pixbuf_get_rowstride (pixbuf),
                                       gdk_pixbuf_get_has_alpha (pixbuf) ? 4
                                                                         : 3,
                                       0,
                                       &error);

  ClutterTexture also now has a new filename property and
  clutter_texture_new_from_file which is intended as an alternate to
  common previous GdkPixbuf primary usage (i.e loading images from
  disk).

  To read texture data back into a pixbuf or system memory use a combination
  of clutter_texture_get_cogl_texture & cogl_texture_get_data.

  The clutter-gtk integration library has API for using GdkPixbuf with
  ClutterTextures (among others).

  ClutterTexture now supports a keep-aspect property which is set to FALSE
  by default.

* clutter_texture_from_actor will now reparent source actors if they
  are not parented. This behaviour may change in future releases.
  There are known not yet fixed issues with source actors that set
  depth or use clipping.

* The size negotiation API has been completely changed in order to allow
  the creation of non-fixed layout managers. These functions have been
  removed:

    clutter_actor_request_coords()
    clutter_actor_query_coords()

  from the ClutterActor API, as well as their virtual functions inside
  the ClutterActorClass. The size of an actor is now split into two
  different concepts: the preferred size (width and height
  separatedly) and the size that has been allocated by the container
  to which that actor belongs. Clutter still defaults to the fixed
  layout management (i.e ClutterGroup) of the actors, but supports
  fluid layout managers written using the new API.

  Composite actors (actors with 'private' children not implementing the
  container interface) now need to implement an allocate method here pass
  an allocation to any composite children.

* Clutter now depends on PangoCairo for the font rendering. The PangoClutter
  API is still considered semi-private and no guarantees are made on its
  stability.

* ClutterShader API has been slightly changed: the ClutterStage:bound
  property, clutter_shader_bind() and clutter_shader_is_bound() have
  been renamed to :compiled, clutter_shader_compile() and
  clutter_shader_is_compiled(), respectively. The previously semi-private
  clutter_shader_release_all() function is now not exported anymore.

* ClutterStage is not an abstract type anymore: it can be instantiated
  using clutter_stage_new() and it can be properly subclassed. If the
  backend supports multiple stages, every stage will be a new window,
  whose lifetime will have to be managed by the developer. Clutter will
  still create the default stage, and guarantees that every call to
  clutter_stage_get_default() will return exactly the same pointer.

* Actors now have a new 'show-on-set-parent' property set to TRUE by
  default. With this property set to TRUE, actors will automatically
  have clutter_actor_show() called on them when a parent is set (i.e
  added to a container).

* Clutter now features support for multiple stages assuming supported
  by the backend. See test-multistage.c for example of usage. This
  does not change the automatic creation of the default stage.
  A single GL context is used across all stages meaning GL resources
  are shared.

* There is now an experimental native iPod Touch/iPhone UIKit backend
  named 'fruity'.

* There is now an experimental native Win32 WGL backend.

* COGL (and therefor Clutter) now optionally features initial support for
  OpenGL ES 2.0. It features support for shaders.

* Some more focused timeline unit tests have been added and some tweaks to
  timeline semantics were made; E.g. For a 10 frame timeline here are some
  points about the semantics:
  - When you create a timeline it starts with current_frame_num == 0
  - After starting a timeline, the first timeout is for current_frame_num == 1
    (Notably it isn't 0 since there is a delay before the first timeout signals
    so re-asserting the starting frame (0) wouldn't make sense.)
  - For a non looping timeline the last timeout would be for
    current_frame_num == 10
  - For a looping timeline the timeout for current_frame_num == 10 would be
    followed by a timeout for current_frame_num == 1 and frame 0 is considered
    == frame 10.
  - Asking for a timeline of N frames might better be described as asking for
    a timeline of _length_ N.

* The behaviour of clutter_actor_get_opacity() has been slightly changed;
  instead of returning the composited opacity of the entire parents chain
  of an actor, clutter_actor_get_opacity() does what you mean, and returns
  the opacity set with clutter_actor_set_opacity(). The composited
  opacity value is now returned by clutter_actor_get_paint_opacity().

* Until 0.6, clutter_label_get_color() would have returned a ClutterColor
  with the alpha component equal to the composited opacity of the label.
  Now, clutter_label_get_color() returns a copy of the exact color set
  with clutter_label_set_color().

* The ClutterEntry actor will automatically handle key events inside
  a key-press-event handler. This deprecates the usage of
  clutter_entry_handle_key_event() to update the contents of the
  entry using the ClutterKeyEvent.

* The Clutter X11 and GLX backends feature support for wrapping external
  X Drawables (such as windows as pixmaps) via the 'texture_from_pixmap' GLX
  extension if available and fallback to slower XGetImage copies if not.

* ClutterContainer can have per child custom properties via ClutterChildMeta.

Release Notes for Clutter 0.6
-------------------------------------------------------------------------------

* Now that every actor has events, the class signal handlers have been
  removed from ClutterStageClass and moved into ClutterActorClass.

* The CLUTTER_2BUTTON_PRESS and CLUTTER_3BUTTON_PRESS event types have been
  removed in favour of the ClutterButtonEvent:click_count counter

* X11 related called for glx and eglx backends are now shared and moved into
  a clutter_x11 prefix.

* Scaling with gravity functionality was replaced by more generic and
  powerful anchor point.

* The ClutterLayout interface, the ClutterBox and its implementations have
  been moved outside Clutter core.

* The Effects API has been simplified, and it only requires the final value
  of the effect instead of both the start and end value.

* The per-axis rotation API has been deprecated in favour of a single
  pair of accessors, clutter_actor_set_rotation() and
  clutter_actor_get_rotation().

* Every actor sub-class that is overriding the ClutterActor::request_coords()
  virtual function *must* chain up to the parent's implementation in order
  to store the bounding box inside the ClutterActor private data.

* It is now impossible to call clutter_actor_destroy() on the stage. Backends
  will have to unset the CLUTTER_ACTOR_IS_TOPLEVEL private flag to actually
  destroy the stage.

* The default value of the ClutterLabel:wrap property has been changed from
  TRUE to FALSE.

* All the behaviours properties have been renamed to match the $PROPERTY-start
  and $PROPERTY-end convention.

* The clutter_group_add() function has been "undeprecated" and reimplemented
  as a commodity macro; a clutter_stage_add() commodity macro has also been
  added. Both macros just safely wrap clutter_container_add_actor().

* The ClutterContainer::find_child_by_id() virtual function has been removed;
  Clutter keeps track of every actor, so there's no need to recurse into
  containers anymore.

* The unused ClutterActor::set_depth() and ClutterActor::get_depth() virtual
  functions have been removed.

* It is now possible to roundtrip the string created by
  clutter_color_to_string() to clutter_color_parse().

* The amount of motion events is now being throttled using the default frame
  rate setting as the base value to compute the default and maximum frequency
  of motion events to be delivered to every actor.

* ClutterAngle usage has been removed from the public API: it is an internal
  optimization, which can be used for faster angle computations; users of
  ClutterAngle now take a fixed point number in form of a ClutterFixed.

* ClutterGravity usage when scaling has been superceded by the anchor point.

* The precision of the clutter_sqrti() algorithm has been improved for small
  values.

* ClutterBehaviourScale constructor, scale properties and accessors have
  been changed to accomodate the scaling factors on both the X and Y axis,
  matching the clutter_actor_set_scale() function. This also changed the
  clutter_effect_scale() function, which now accepts the final scale
  factor on both the X and Y axis. The gravity property has also been
  removed, as it behaved incorrectly with regards to the anchor point. When
  scaling an actor using a BehaviourScale, the anchor point should be set
  before applying the behaviour (remember to adjust the positioning as
  well by factoring in the anchor point).

* The clutter_do_event() is now public; it can be used to feed an event to
  Clutter and let it generate the capture-bubble signal emissions; it should
  not be used by applications.

* In the X11-based backends, the DestroyNotify event is not propagated to
  Clutter core if the stage is using a foreign window.

* To avoid method name collisions between ClutterActor and ClutterEntry
  in high-level languages, the clutter_entry_set_position() and
  clutter_entry_get_position() functions have been renamed to
  clutter_entry_set_cursor_position() and clutter_entry_get_cursor_position()
  respectively.

Release Notes for Clutter 0.4.0
-------------------------------------------------------------------------------

* clutter_actor_show_all does not recurse for groups at least (this is to
  match the original group_show_all behaviour). This is like 0.3 but was
  never noted.

* ClutterBox API has changed: clutter_box_pack_start() and
  clutter_box_pack_end() have been removed in favour of the clutter_box_pack()
  API.

* Both clutter_threads_enter() and clutter_threads_leave() have been
  removed from the API, as they just created confusion and the wrong
  idea that Clutter is either thread-safe or thread-aware. Full
  thread-awareness is arriving in the next revision (see bug #429).

* The ClutterBehaviourRotate and ClutterBehaviourEllsipse APIs have been
  overhauled.

Release Notes for Clutter 0.3.1
-------------------------------------------------------------------------------

* clutter_actor_apply_transform_to_point() parameters changed to use
  ClutterVertices.

* New 'Native API' backend expects EGL implementation to provide a
  CreateNativeWindow() API call.

* Exisiting X11 based egl backend public API calls now prefixed eglx.

Release Notes for Clutter 0.3
-------------------------------------------------------------------------------

* ClutterTexture changes:
  + clutter_texture_set_pixbuf() now takes a GError paremeter.
  + clutter_texture_upload_data has been split into two new and
    more functional versions; clutter_texture_set_from_rgb_data(),
    clutter_texture_set_from_yuv_data().

* The GLX specific API has been moved to the GLX backend code and
  it's now under the ClutterGlx namespace.

* The priority of the various users of the GLib main loop has been
  reviewed and changed were necessary. Every paint is queued with a
  priority of G_PRIORITY_DEFAULT + 10; timelines are allocated with
  a G_PRIORITY_DEFAULT + 30 priority; events are processed with a
  G_PRIORITY_DEFAULT priority. This ensures that events are processed
  before the paints take place.

* The ClutterActor::allocate_coords() has been renamed to
  ClutterActor::query_coords(), in order to clarify its usage.

* Actors overriding ClutterActor::request_coords() and
  ClutterActor::query_coords() must convert sizes obtained from the
  public API from pixels to ClutterUnits, using the conversion
  macros found in clutter-units.h. The conversion will be necessary
  until the public API will switch over to returning the generic
  units too.

* Users of Intel video cards should find that disabling sync-to-vblank
  is no longer necessary. In case Clutter applications take really
  long times for redrawing and appear stuck and unresponsive, the
  sync-to-vblank feature might still be the culprit; in that case, use
  "export CLUTTER_VBLANK=none" to disable it and file a bug reporting the
  video card model, the driver version and the X server version.

* ClutterTimeline objects now share the same timeout pool (see the
  ClutterTimeoutPool API). This might cause problems with heavily
  threaded libraries without integration with the GLib main loop.
  If an application experiences bad behaviours during animations
  use "export CLUTTER_TIMELINE=no-pool" to disable the timeout pool.

* clutter_color_parse() now handles color definitions with alpha. Alpha
  will default to fully Opaque rather than fully transparent if not specified.

* The Clutter examples/ directory has been removed and replaced with a
  tests/ directory.

* The ClutterEvent API has been changed, and specific functions have been
  removed in favour of event-agnostic ones.

* The ClutterStage::input-event signal has been removed. All the events now
  emit the ClutterStage::event and ClutterStage::event-after signals, for
  generic event handling.

* Runtime options now dependant on backend.

* ClutterGroup API to add, remove and list children has been deprecated in
  favour of ClutterContainer API. The ClutterGroup::add and
  ClutterGroup::remove signals have been deprecated:
  ClutterContainer::actor-added and ClutterContainer::actor-removed should
  be used instead.