The pipeline private data is accessed both from the private data set
on a CoglPipeline and the destroy notify function of a weak material
that the vertex buffer creates when it needs to override the wrap
mode. However when a CoglPipeline is destroyed, the CoglObject code
first removes all of the private data set on the object and then the
CoglPipeline code gets invoked to destroy all of the weak children. At
this point the vertex buffer's weak override destroy notify function
will get invoked and try to use the private data which has already
been freed causing a crash.
This patch instead adds a reference count to the pipeline private data
stuct so that we can avoid freeing it until both the private data on
the pipeline has been destroyed and all of the weak materials are
destroyed.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2544
In cogl_pipeline_set_layer_combine_constant it was comparing whether
the new color is the same as the old color using a memcmp on the
constant_color parameter. However the combine constant is stored in
the layer data as an array of four floats but the passed in color is a
CoglColor (which is currently an array of four guint8s). This was
causing valgrind errors and presumably also the check for setting the
same color twice would always fail.
This patch makes it do the conversion to a float array upfront before
the comparison.
Instead of just setting the input device pointer in the private event
data, it should also set the field in the event sub-types, so that
direct access to the structures still works.
We should force building the cookbook when releasing a tarball of
Clutter, so that users of packaged tarballs can actually build the
cookbook themselves.
Try to make the cookbook pass the distcheck phase, so that we can run
distcheck with --enable-docs, and make sure that a tarballed clutter
release can actually build the cookbook.
cogl_matrix_project_points and cogl_matrix_transform_points had an
optimization for the common case where the stride parameters exactly
match the size of the corresponding structures. The code for both when
generated by gcc with -O2 on x86-64 use two registers to hold the
addresses of the input and output arrays. In the strided version these
pointers are incremented by adding the value of a register and in the
packed version they are incremented by adding an immediate value. I
think the difference in cost here would be negligible and it may even
be faster to add a register.
Also GCC appears to retain the loop counter in a register for the
strided version but in the packed version it can optimize it out and
directly use the input pointer as the counter. I think it would be
possible to reorder the code a bit to explicitly use the input pointer
as the counter if this were a problem.
Getting rid of the packed versions tidies up the code a bit and it
could potentially be faster if the code differences are small and we
get to avoid an extra conditional in cogl_matrix_transform_points.
* elliot/cookbook-actors-composite:
docs: Add reference to useful GObject tutorial
docs: Explain why destroy() is implemented
docs: Implement destroy() rather than dispose()
docs: Don't use clutter_stage_get_default()
docs: Change text on button
docs: Add a note about other state variables
docs: Complete composite actor recipe
docs: Change order of functions in example to match docs
docs: Add more comments on how allocate() works
docs: Include code examples in the recipe
docs: Explain enums for properties and signals
docs: Don't set explicit size on button
docs: Add example of preferred_height() and preferred_width()
docs: Add recipe for creating a custom ClutterActor with composition
docs: Add more comments on code example for composite actor
docs: Improve example code formatting
docs: Add some gtk-doc annotations to example
docs: Add custom ClutterActor example which uses composition
Use a DeviceManager sub-class similar to the Win32 backend one, which
creates two InputDevices: a core pointer and a core keyboard.
The event translation code then uses these two devices to fill out the
.device field of the events.
Throw in enter/leave tracking, given that we need to update the device's
state.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2490
Implementation of event loop which works with GLib events, native OS X
events and Clutter events.
The event loop source code comes from the equivalent code in the Quartz
GDK backend from GTK+ 2.22.1, which is LGPL v2.1+ and thus compatible
with Clutter's licensing terms.
The code has been tested with libsoup, which did not work before together
with Clutter.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
http://bugzilla.clutter-project.org/show_bug.cgi?id=2490
Remove the dispose() implementation and replace
with destroy().
This should be promoted as the standard approach
for implementing a composite actor, as it emits a
signal when instances of the actor subclass are destroyed.
Add some extra detail to the Discussion section of the
composite actor recipe, concentrating on the pros and
cons of this approach.
Also explain more about the Clutter parts of the implementation.
Also general tidy up of language and style.
Add some extra description to the allocate() function,
explaining how the allocation has to be adjusted to
coordinates relative to the actor as a whole, before
applying to the single child actor it is composed from.
Include all the code examples inline as part of the recipe.
Remove sections around each code example, as these are
unnecessary; leave full discussion for the Discussion section
instead of trying to cram it in around the code example.
As most actor subclasses will probably want to implement
size requisition, give a simple example of how to do this
on the basis of the composed actor's size, plus some padding.
Wayland visuals refer to a pixel's bytes in order from
most significant to least significant, while the
one-byte-per-component Cogl formats refer to the order
of increasing memory addresses, so converting between
the two depends on the system's endianness.
The height was being set from the ClutterGeometry in some parts
and from the stage in others. And since both callers of this
function pass &stage_wayland->allocation as the geometry anyway,
the stage argument isn't really even needed.
