ClutterActor should be able to hold all transitions, even the ones that
have been explicitly created.
This will allow to add new transitions types in the future, like the
keyframe-based one, or the transition group.
It should be possible to set up the delay of a transition, but since
we start the Transition instance before returning control to the caller,
we cannot use clutter_actor_get_transition() to do it without something
extra-awkward, like:
transition = clutter_actor_get_transition (actor, "width");
clutter_timeline_stop (transition);
clutter_timeline_set_delay (transition, 1000);
clutter_timeline_start (transition);
for each property involved. It's much easier to add a delay to the
easing state of an actor.
Clutter is meant to be, and I quote from the README, a toolkit:
for creating fast, compelling, portable, and dynamic graphical
user interfaces
and yet the default mode of operation for setting an actor's state on
the scene graph (position, size, opacity, rotation, scaling, depth,
etc.) is *not* dynamic. We assume a static UI, and then animate it.
This is the wrong way to design an API for a toolkit meant to be used to
create animated user interfaces. The default mode of operation should be
to implicitly animate every state transition, and only allow skipping
the animation if the user consciously decides to do so — i.e. the design
tenet of the API should be to make The Right Thing™ by default, and make
it really hard (or even impossible) to do The Wrong Thing™.
So we should identify "animatable" properties, i.e. those properties
that should be implicitly animated by ClutterActor, and use the
animation framework we provide to tween the transitions between the
current state and the desired state; the implicit animation should
happen when setting these properties using the public accessors, and not
through some added functionality. For instance, the following:
clutter_actor_set_position (actor, newX, newY);
should not make the actor jump to the (newX, newY) point; it should
tween the actor's position between the current point and the desired
point.
Since we have to maintain backward compatibility with existing
applications, we still need to mark the transitions explicitly, but we
can be smart about it, and treat transition states as a stack that can
be pushed and popped, e.g.:
clutter_actor_save_easing_state (actor);
clutter_actor_set_easing_duration (actor, 500);
clutter_actor_set_position (actor, newX, newY);
clutter_actor_set_opacity (actor, newOpacity);
clutter_actor_restore_easing_state (actor);
And we can even start stacking animations, e.g.:
clutter_actor_save_easing_state (actor);
clutter_actor_set_easing_duration (actor, 500);
clutter_actor_set_position (actor, newX, newY);
clutter_actor_save_easing_state (actor);
clutter_actor_set_easing_duration (actor, 500);
clutter_actor_set_easing_mode (actor, CLUTTER_LINEAR);
clutter_actor_set_opacity (actor, newOpacity);
clutter_actor_set_depth (actor, newDepth);
clutter_actor_restore_easing_state (actor);
clutter_actor_restore_easing_state (actor);
And so on, and so forth.
The implementation takes advantage of the newly added Transition API,
which uses only ClutterTimeline sub-classes and ClutterInterval, to cut
down the amount of signal emissions and memory management of object
instances; as well of using the ClutterAnimatable interface for custom
properties and interpolation of values.
The ::paint, ::queue-redraw, and ::queue-relayout signals should be
marked as no-recurse and no-hooks; these signals are emitted *a lot*
during each frame, and since GLib has a bunch of optimizations for
signals with no closures, we should try and squeeze every single CPU
cycle we can.
In theory, handlers connected to the ::allocation-changed signal may be
able to modify the actor's real allocation and allocation flags,
especially now that we use STATIC_SCOPE; let's avoid this, so that we
don't regret it later.
The ActorBox passed to the ::allocation-changed signal should be
annotated as STATIC_SCOPE, given that it's a pointer to a structure
inside ClutterActorPrivate - hence there's no risk of it actually being
freed from a signal handler. This allows the GSignal machinery to avoid
a costly copy/free for each signal emission.
If the redraw entry is not cleared, queueing a redraw from a signal
handler could reinsert the same object in the stage redraw list,
causing the segfault later (as the object is immediately freed)
https://bugzilla.gnome.org/show_bug.cgi?id=671173
We currently check for the IN_DESTRUCTION flag inside the
add_child_internal() function.
This check disallows calling methods that change the stacking order
within the destruction sequence, by triggering a critical warning first,
and leaving the actor in an undefined state, which then ends up being
caught by an assertion.
The reproducible sequence is:
- actor gets destroyed;
- another actor, linked to the first, will try to change the
stacking order of the first actor;
- changing the stacking order is a composite operation composed
by the following steps:
1. ref() the child;
2. remove_child_internal(), which removes the reference;
3. add_child_internal(), which adds a reference;
- the state of the actor is not changed between (2) and (3), as
it could be an expensive recomputation;
- if (3) bails out, then the actor is in an undefined state, but
still alive;
- the destruction sequence terminates, but the actor is unparented
while its state indicates being parented instead.
- assertion failure.
The obvious fix would be to decompose each set_child_*_sibling() method
into proper remove_child()/add_child(), with state validation; this may
cause excessive work, though, and trigger a cascade of other bugs in
code that assumes that a change in the stacking order is an atomic
operation.
Another potential fix is to just remove this check here, and let code
doing stacking order changes inside the destruction sequence of an actor
continue doing the work.
The third fix is to silently bail out early from every
set_child_*_sibling() and set_child_at_index() method, and avoid doing
work.
I have a preference for the second solution, since it involves the least
amount of work, and the least amount of code duplication.
See bug: https://bugzilla.gnome.org/show_bug.cgi?id=670647
Now that ClutterActor has a default paint volume, subclasses may wish
to retrieve it without chaining up to the parent's implementation of
the get_paint_volume() function.
The get_default_paint_volume() returns a ClutterPaintVolume pointer
to the paint volume as computed by the default implementation of the
get_paint_volume() virtual function; it can only be used immediately,
as it's not guaranteed to survive across multiple frames.
The experimental cogl_pipeline_new() api was recently changed so it
explicitly takes a CoglContext. This updates all calls to
cogl_pipeline_new() in clutter accordingly.
If we have N children and the user passes N (or a number beyond N) to
clutter_actor_insert_child_at_index, we should respond by adding the
child at the end, not silently doing nothing.
This should avoid trying to fix the origin of a paint volume set from
the allocation's origin, and thus breaking everything.
A PaintVolume for an actor is defined to be relative to the actor's
modelview unless specifically modified by internal functions; the origin
of an actor's allocation is, on the other hand, parent-relative.
There are times when we don't want to remove all children and count of
the reference count to drop to 0 to ensure destruction; there are cases,
such as managed environments, where it's preferable to ensure that the
children of an actor get actually destroyed.
A bunch of private symbols have escaped into the SO; let's rectify this
situation by using the '_' private prefix, or making them static as they
should have been.
Some of Cogl's experimental apis have changed so that the buffer apis
now need to be passed a context argument and some drawing apis have been
replaced with cogl_framebuffer_ drawing apis that take explicit
framebuffer and pipeline arguments.
These changes were made as part of Cogl moving towards a more stateless
api that doesn't rely on a global context.
This patch updates Clutter to work with the latest Cogl api and bumps
the required Cogl version to 1.9.5.
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Similar to the clutter_actor_iter_remove(), but it'll call destroy()
instead of remove_child().
We can also reimplement the ::destroy default handler using it, and make
it more compact.
There is a typo in the check for a negative index: the index variable
should be index_, not index - unfortunately, the latter can still be
resolved to index(3), so compiler and linker are perfectly happy.
https://bugzilla.gnome.org/show_bug.cgi?id=669730
ClutterActor stopped requiring to override the map and unmap virtual
functions some time ago.
Now that ClutterActor implements the Container interface, overriding map
and unmap to control the MAPPED state of the children is pretty much
going to be a source of bugs and misunderstandings.
Plus, the ordering of the unmap, destroy, dispose, and finalize calls
should be be documented properly.
The documentation should clarify all that.
When calling clutter_actor_destroy(), ClutterActor calls
update_map_state() on itself to unset the REALIZED and MAPPED states,
prior to running the dispose() implementation.
The default dispose() will call remove_child() (either directly or
through the Container implementation), which will check for the MAPPED
state and then run update_map_state() again. We use the previously set
MAPPED state to decide whether or not the parent should queue for a
relayout/redraw when removing a visible children.
If the MAPPED flag was cleared prior to remove_child(), though, it'll
always be unset by the time we get to remove_child(), and this will
cause missing redraws/relayouts; we were ignoring this prior the
post-First Apocalypse changes because we were doing:
if (was_mapped)
clutter_actor_queue_relayout (parent);
clutter_actor_queue_redraw (parent);
which is obviously wrong. Once I removed that glaring brain damage from
the remove_child() implementation, bugs started appearing — bugs that
were probably the reason why we introduced that brain damage in the
first place, instead of checking the source of those bugs.
The obvious fix is to avoid clearing up the actor's state on destroy()
until we remove the actor from its parent. This also reduces the amount
of work we do, and the code paths that can potentially go wrong.
Since the code dealing with ClutterShader is pretty self-contained, now,
we can safely move it outside of the main ClutterActor source file and
into its own. This will allow us to just drop a bunch of files when
branching for 2.0.
Iterating over children and ancestors of an actor is a relatively common
operation. Currently, you only have one option: start a for() loop, get
the first child of the actor, and advance to the next sibling for the
list of children; or start a for() loop and advance to the parent of the
actor.
These operations can be easily done through the ClutterActor API, but
they all require going through the public API, and performing multiple
type checks on the arguments.
Along with the DOM API, it would be nice to have an ancillary, utility
API that uses an iterator structure to hold the state, and can be
advanced in a loop.
https://bugzilla.gnome.org/show_bug.cgi?id=668669
Now that we reinstated Group to its "former glory", we need to ensure
that applications using the deprecated containers with the new DOM API
in ClutterActor can actually work - or, at least, not break horribly.
This actually means making sure that ClutterStage and ClutterGroup can
cope with the DOM, while retaining their old implementations, as well as
their bizarre idiosyncrasies and their utter, utter brokenness.
Making Group just a proxy to Actor broke some behaviour that application
and toolkit code was relying on. Let's keep Group around to fight
another day.
This commit fixes gnome-shell as far as I can test it.
The usual way to implement a container actor is to override the
allocate() virtual function, chain up, and then allocate the actor's
children.
Clutter now has the ability to delegate layout management to
ClutterLayoutManager directly; in the allocation, this is done by
checking whether the actor has children, and then call
clutter_layout_manager_allocate() from within the default implementation
of the ClutterActor::allocate() vfunc. The same vfunc that everyone, has
been chaining up to.
Whoopsie.
Well, we can check if there's a layout manager, and if it's NULL, we
bail out. Except that there's a default layout manager, and it's the
fixed layout manager, so that classes like Group and Stage work by
default.
Double whoopsie.
The fix for this scenario is a bit nasty; we have to check if the actor
class has overridden the allocate() vfunc or not, before actually
looking at the layout manager. This means that classes that override the
allocate() vfunc are expected to do everything that ClutterActor's
default implementation does - which I think it's a fair requirement to
have.
For newly written code, though, it would probably be best if we just
provided a function that does the right thing by default, and that
you're supposed to be calling from within the allocate() vfunc
implementation, if you ever chose to override it. This new function,
clutter_actor_set_allocation(), should come with a warning the size of
Texas, to avoid people thinking it's a way to override the whole "call
allocate() on each child" mechanism. Plus, it should check if we're
inside an allocation sequence, and bail out if not.
If we want to set a default layout manager, we need to do so inside
init(), as it's not guaranteed that people subclassing Actor and
overriding ::constructed will actually chain up as they should.
The default pick() behaviour does not take into consideration the
children of a ClutterActor because the existing containter actors
usually override pick(), chain up, and then paint their children.
With ClutterActor now a concrete class, though, we need a way to pick
its children without requiring a sub-class; we could simply iterate over
the children inside the default pick() implementation, but this would
lead to double painting, which is not acceptable.
A moderately gross hack is to check if the Actor instance did override
the pick() implementation, and if it is not the case, paint the children
in pick mode.
The hide_all() method is pretty much pointless, as hiding an actor will
automatically prevent its children from being painted. The show_all()
method would only be marginally useful, if actors weren't set to be
visible by default when added to another actor - which was the case when
we introduced show_all() and hide_all().
The concept of "internal child" only meant anything when we had a
separate API for containers and actors. Now that we plugged that
particular hole, we can drop all the hacks we used to have in place
to work around its design limitations.
It can be convenient to be able to set, or get, all the components of an
actor's margin at the same time; since we already have a boxed type for
storing a margin, an accessors pair based on it is not a complicated
addition to the API.
Inside the set_child_[above|below]_sibling() and set_child_at_index() we
should be using the internal API for mutating the children list, instead
of the delegate functions. This ensures that we go through a single,
well-defined code path for all operations on the list of children of
an actor.
We have a replacement in ClutterActor, now.
The old ClutterContainer API needs to be deprecated, and the raise() and
lower() virtual functions need a default implementation, so we can check
for implementations overriding them, by using the diagnostic mode like
we do for add(), remove(), and foreach().
The sort_depth_order() virtual function just doesn't do anything, as it
should have been made ages ago.
The Actor wrappers for the Container methods also need to be deprecated.
ClutterActor provides four methods for changing the paint sequence order
of its children:
raise_top()
raise()
lower()
lower_bottom()
The first and last one being just wrappers around raise() and lower(),
respectively. These methods have various issues: they omit the parent,
preferring to retrieve it from the actor passed as the first argument;
this does not match the new style of API introduced to operate on the
list of children of an actor.
Additionally, the raise() and lower() methods of ClutterActor call into
the Container interface, and are not really aptly named (raise() in
particular collides with the completely unrelated 'raise' keyword in
Python, and usually needs to be wrapped in order to be used at all).
Furthermore, we need public methods that Container can call from its
default implementation, as well as methods to port current Container
implementations.
Finally, since we have insert_child_at_index(), we should also have an
equivalent set_child_at_index() as well.
The internal versions of add_child() and remove_child() currently use
boolean arguments to control things like the ChildMeta instances and
the emissions of signals; using more than one boolean argument is an
indication that you need flags to avoid readability issues, as well as
providing a way to add new behaviours without a combinatorial explosion
of arguments, later on.
I don't feel comfortable with this feature, and its implementation
still has too many rough edges. We can safely punt it for now, and
introduce it at a later point, as it doesn't block existing features
or API.
We need to paint the background color in the default class handler for
two reasons: it's logically appropriate, and we don't want actor
subclasses overriding the ::paint class handler to change behaviour only
because somebody decided to set the background color.
Instead of making ClutterActor implement the basic add/remove/foreach
virtual functions of ClutterContainer, we can simply do that from
within the ClutterContainer implementation.
Given the size and scope of the changes in ClutterActor, we ought to
rewrite the overall description of what an actor is, what it does, and
how are you supposed to use it and subclass it.
This will make things interesting.
We have better replacements in ClutterActor, that do The Right Thing™
instead of deferring control and requiring reimplementation in every
single container actor.
The correct sequence of actions should be remove(old) → insert(new), not
insert(new) → remove(old). We can implement a simple delegate insertion
functions to insert the new child between the previous and next siblings
of the old child.
While we're at it, let's also add a unit test for replace_child().
Providing a default get_paint_volume() that takes into account the
children of an actor was a goal of the whole First Apocalypse; if we
make all the containers rely on it, and yet we return a FALSE value
(meaning: we don't have a valid paint volume) even when we do have it,
then we are going to break the whole machinery, though.
The insert_child_at_index, insert_below and insert_above messed up the
first and last child pointers in various cases. This commit fixes all
the instances of first and last child pointers being stale or set to
NULL.
Instead of requiring every consumer of the ClutterActor API that wishes
to iterate over the children of an actor to use the get_children()
method, we should provide an iteration API directly inside ClutterActor
itself.
Instead of storing the list of children, let's turn Actor inside a
proper node of a tree.
This change adds the following members to the Actor private data
structure:
first_child
last_child
prev_sibling
next_sibling
and removes the "children" GList from it; iteration is performed through
the direct pointers to the siblings and children.
This change makes removal, insertion before a sibling, and insertion
after a sibling constant time operations, but it still retains the
feature of ClutterActor.add_child() to build the list of children while
taking into account the depth set on the newly added child, to allow the
default painter's algorithm we employ inside the paint() implementation
to at least try and cope with the :depth property (albeit in a fairly
naïve way). Changes in the :depth property will not change the paint
sequence any more: this functionality will be restored later.
ClutterTransformInfo is a (private) ancillary data structure that
contains all the decomposed transformation data, i.e. rotation angles
and centers, scale factors and centers, and anchor point. This data
structure is stored in the GData of the actor instance instead of the
actor's private data. This change gives us:
• a smaller, cleaner private data structure;
• no size penalty for untransformed actors;
• static constant storage for the defaults, shared across all
instances;
• cache locality for all the decomposed transformation data,
given that the structure size is smaller.
At the end of the day, the only authoritative piece of information for
actor transformation is the CoglMatrix that we initialize in
apply_transform() from all the decomposed parameters, and that can stay
inside the private data structure of ClutterActor.
There are only two kinds of actors that allow underallocations,
according to the API contract:
• ClutterStage, as it is a viewport and it doesn't have an implicit
minimum size;
• Actors using the CLUTTER_ACTOR_NO_LAYOUT escape hatch, which allows
them to bail out from our layout management policies.
The warning about underallocations should take these two exceptions
under consideration.
ClutterActor now has all the API and capabilities for being a concrete
class:
- layout management, through delegation
- container implementation and API
- background color
This means that a simple scene can be built straight out of actors
without using subclasses except for the Stage.
This is the first step towards the deprecation of most of the Actor
subclasses provided by Clutter.
ClutterActor can do better by default than just giving up immediately.
An actor can check for the clip region, and for its children's paint
volume, for instance.
Just these two should give us a better default implementation for newly
written code.
Each actor should have a background color property, disabled by default.
This property allows us to cover 99% of the use cases for
ClutterRectangle, and brings us one step closer to being able to
instantiate ClutterActor directly.
And make sure that overriding Container and calling
clutter_actor_add_child() will result in the same sequence of operations
as the current set_parent()+queue_relayout()+signal_emit pattern.
Existing containers can continue using:
clutter_actor_set_parent (child, CLUTTER_ACTOR (container));
clutter_actor_queue_relayout (CLUTTER_ACTOR (container));
g_signal_emit_by_name (container, "actor-added", child);
and newly written containers overriding Container.add() can simply call:
clutter_actor_add_child (CLUTTER_ACTOR (container), child);
instead.
We need to queue a relayout when removing a visible child from a visible
parent.
We also need to insert the child at the right position (depending on the
depth) so that newly added actors will be painted on top.
Remove four more floats from ClutterActorPrivate.
The fixed minimum and natural sizes should be stored inside the
ClutterLayoutInfo structure, along with the fixed position.
Add a failsafe against a NULL parent, to avoid a segfault when calling
clutter_actor_allocate() on the Stage.
We also need to deal with floating point values: straight comparison is
not going to cut it.
ClutterActor has various properties controlling the allocation:
- x-align, y-align
- margin-top, margin-bottom, margin-left, margin-right
These properties should adjust the ClutterActorBox passed from the
parent actor to its children when calling clutter_actor_allocate(),
so that the child can just allocate its children at the right origin
with the right available size.
The actor class should be able to hold the margin offsets like it does
for expand and alignment flags.
Instead of filling the private data structure with data, we should be
able to use an ancillary data structure, given that all this data is
optional and might never be set in the first place.
In case no layout manager was set during construction, we fall back to a
FixedLayout. The FixedLayout has the property of making the fixed
positioning and sizing API, as well as the various Constraints, work
out of the box.
Now that ClutterActor implements the Container contract we can actually
defer the size negotiation to a ClutterLayoutManager directly from the
default implementation of the Actor's virtual functions.
We can provide most of the ClutterContainer implementation directly
within ClutterActor — basically removing the need of having the
Container interface in the first place. For backward compatibility
reasons we can keep the interface, but let Actor implement it directly.
Let's try and move away from the reverse implicit scene graph build API,
which we mutuated from GTK+, towards a more traditional node/child API.
The set_parent()/unparent() API is confusing, unless you know the
history; having a add_child()/remove_child() methods pair makes it more
explicit.
We can easily implement the old set_parent()/unparent() pair in terms of
the newly add_child()/remove_child() one.
Enclose the check inside a #ifdef CLUTTER_ENABLE_DEBUG ... #endif, so
that we can compile it out; also, use g_string_append() instead of the
g_string_append_printf() function, given that we're just concatenating
strings.
Currently, we're emitting the ClutterActor::destroy at the end of the
dispose implementation - right before we chain up to the parent
implementation.
The point of emission makes the ::destroy signal handlers able to just
use the actor pointer - as the actor state will have been mostly cleared
by the time application can run. This (undocumented) behaviour severely
limits the amount of things you can do inside a ::destroy signal
handler, thus making the ::destroy signal just a weird weak reference,
instead of a proper way to break application reference cycles.
Given that this change relaxes some of the conditions, this change
should be safe - obviously, if anything happens, we'll back it out; the
conformance and interactive tests confirm that, for common patterns of
usage, this change does not break existing code.
GLib has a nice, atomic object clearing function that allows us to drop
code looking like:
if (priv->object != NULL)
{
g_object_unref (priv->object);
priv->object = NULL;
}
from the ::dispose implementation.
Add a public version of the clipped queue redraw, using a 2D clip. This
allows implementing actors with trackable 2D clipped regions, like the
ClutterX11TexturePixmap, outside of Clutter itself.
https://bugzilla.gnome.org/show_bug.cgi?id=660997
A lot of the example code in the cookbook and the API reference still
uses the default stage — sometimes as if it were a non-default one,
which once again demonstrates how the default stage was a flawed concept
that just confused people.
These methods are short-hands for accessing the position and size,
which are already shorthands for accessing the various dimensional
and positional attributes. Plus, they use ClutterGeometry, which is a
fairly bad data type for a rectangle.
There are a couple of gotchas in the 'mapped' flag that are not properly
documented, or are documented only in the actor_invariants.txt file; we
should have a proper description in the API reference as well, to avoid
confusion.
Out-of-band transforms are considered to be all actor transforms done
directly with the Cogl API instead of via ClutterActor::apply_transform.
By running with CLUTTER_DEBUG=oob-transform then Clutter will explicitly
try to detect when un-expected transforms have been applied to the
modelview matrix stack.
Out-of-band transforms can lead to awkward bugs in Clutter applications
because Clutter itself doesn't know about them and this can disrupt
Clutter's input handling and calculations of actor paint-volumes
which can lead to visual artifacts.
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Because we have had several reports about significant performance
regressions since we enabled offscreen redirection by default for
handling correct opacity we are now turning this feature off by default.
We feel that clutter should prioritize performance over correctness in
this case. Correct opacity is still possible if required but the
overhead of the numerous offscreen allocations as well as the cost of
many render target switches per-frame seems too high relative the
improvement in quality for many cases.
On reviewing the offscreen_redirect property so we have a way to
disable redirection by default we realized that it makes more sense for
it to take a set of flags instead of an enum so we can potentially
extend the number of things that might result in offscreen redirection.
We removed the ability to say REDIRECT_ALWAYS_FOR_OPACITY, since it
seems that implies you don't trust the implementation of an actor's
has_overlaps() vfunc which doesn't seem right.
The default value if actor::redirect_offscreen is now 0 which
effectively means don't ever redirect the actor offscreen.
Actually this change has two notable effects; firstly we no longer
perform culling during picking and secondly we avoid updating the
last-paint-volume of an actor when picking.
We shouldn't perform culling during picking until clutter-stage.c is
updated to setup the clipping planes appropriately.
Since the last-paint-volume is intended to represent the visible region
of the actor the last time it was painted on screen it doesn't make
sense to update this during off screen pick renders since we are liable
to end up with a last-paint-volume that maps to an actors new position
when we next come to paint for real.
This fixes a bug in gnome-shell with dragging dash icons leaving a
messy trail on the screen.
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
If the meta for the animation property is not found, the name of the
property to look for is still from the token, and we need to free the
memory allocated for it.
Instead of relying on C to round the floating point allocation to
integers by flooring the values we now use CLUTTER_NEARBYINT to round
the allocation's position and size to the nearest integers instead. Using
floor leads to rather unstable rounding for the width and height when
there may be tiny fluctuations in the floating point width/height.
Signed-off-by: Neil Roberts <neil@linux.intel.com>
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
If we're building on/for Windows, set 'win32' as the default flavour; if
we're building on OS X, set 'osx' as the default flavour. For everything
else, use 'glx'.
Commit 0ede622f51 inadvertently made it so that shaders are applied
during picking. This was making test-shader fail to respond to clicks.
The commit also makes it so that culling is applied during
picking. Presumably this is also unintentional because the commit
message does not mention it. However I think it may make sense to do
culling during picking so it might as well stay that way.
https://bugzilla.gnome.org/show_bug.cgi?id=653959
Although this patch doesn't make them public, it documents the
_clutter_actor_get/apply_relative_transform_matrix functions so they
could easily be made public if desired. I think these API could be
useful to have publicly, and I originally documented them because I
thought they would be needed in the MX toolkit.
Previously ClutterActor was using priv->propagated_one_redraw to
determine whether to pass CLUTTER_EFFECT_PAINT_ACTOR_DIRTY to the
paint method of the effect. This isn't a good idea because the
propagated_one_redraw flag is cleared whenever clutter_actor_paint is
called, even if the actor isn't actually painted because of the zero
opacity shortcut. Instead of this, ClutterActor now has a separate
flag called is_dirty that gets set whenever queue_redraw_full is
called or whenever the queue redraw signal is bubbled up from a child
actor. The flag is only cleared in clutter_actor_paint if the effects
are actually run. Therefore it will stay set even if the opacity is
zero or if the parent actor decides not to paint the child.
Previously there were two places set propagated_one_redraw to FALSE -
once if the opacity is zero and once just before we emit the paint
signal. Now that propagated_one_redraw is only used to determine
whether to pass on the queue redraw signal it seems to make sense to
just clear it in one place right at the start of clutter_actor_paint.
https://bugzilla.gnome.org/show_bug.cgi?id=651784
On reviewing the clutter-actor.c code using
_apply_modelview_transform_recursive I noticed various comments stating
that it will never call the stage's ->apply_transform vfunc to transform
into eye coordinates, but actually looking at the implementation that's
not true. The comments probably got out of sync with an earlier
implementation that had that constraint. This removes the miss-leading
comments and also updates various uses of the api where we were manually
applying the stage->apply_transform.
Instead of using the cogl_vertex_buffer API this uses the more concise
cogl_primitive API instead. The aim is to get rid of the
cogl_vertex_buffer API eventually so we should be trying out the
replacement API wherever possible.
This makes sure we don't try and draw paint-volumes or culling results
during a pick cycle since that results in us reading back invalid ids
from the pick-buffer.
This adds CLUTTER_PAINT=disable-offscreen-redirect to help diagnose
problems with the correct opacity changes. This just makes it so that
it never installs the flatten effect so it will never automatically
redirect an actor offscreen.
The G_CONST_RETURN define in GLib is, and has always been, a bit fuzzy.
We always used it to conform to the platform, at least for public-facing
API.
At first I assumed it has something to do with brain-damaged compilers
or with weird platforms where const was not really supported; sadly,
it's something much, much worse: it's a define that can be toggled at
compile-time to remove const from the signature of public API. This is a
truly terrifying feature that I assume was added in the past century,
and whose inception clearly had something to do with massive doses of
absynthe and opium — because any other explanation would make the
existence of such a feature even worse than assuming drugs had anything
to do with it.
Anyway, and pleasing the gods, this dubious feature is being
removed/deprecated in GLib; see bug:
https://bugzilla.gnome.org/show_bug.cgi?id=644611
Before deprecation, though, we should just remove its usage from the
whole API. We should especially remove its usage from Cally's internals,
since there it never made sense in the first place.
If an actor or the stage to which it belongs are being destroyed then
there is no point in queueing a redraw that will not be seen anyway.
Bailing out early also avoids the case in which a redraw is queued
during destruction wil cause redraw entries will be added to the Stage,
which will take references on it and cause the Stage never to be
finalized.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2652
With the instantiatable ClutterShaderEffect, the only reason for
ClutterShader to exist is to make the ClutterActor::paint implementation
miserable.
Yes, ClutterShader doesn't use a FBO, so it's "more efficient" on
ClutterTextures. It's also generally wrong unless you know *exactly* how
the actor's pipeline is set up — something we cannot even guarantee
internally unless we start doing lame type checks.
In _clutter_actor_queue_redraw_with_clip, there was the possibility that
the actor will add itself to the stage's redraw queue without keeping track
of the allocated list member.
In clutter_actor_unparent, the redraw queue entry was being invalidated
before the mapped notify signal was being fired, meaning that queueing a
redraw of an unmapped actor in the mapped notification callback could
cause a crash.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2621
_clutter_stage_queue_actor_redraw returns a pointer to the
ClutterStageQueueRedrawEntry struct which it allocates. The actor is
expected to store a pointer to this so that it doesn't need to search
the list of queued redraws next time a queue redraw is called. However
_clutter_actor_queue_redraw_full wasn't storing this pointer which
meant that it thought every queue redraw was the first queue
redraw. That meant that queueing a redraw with a clip or an effect
would override any previous attempts to queue a redraw instead of
trying to combine them.
I think this happened because the old queue_redraw_with_clip also
didn't store the pointer and queue_redraw_full was based on that.
This adds a virtual to ClutterActor so that an actor subclass can
report whether it has overlapping primitives. ClutterActor uses this
to determine whether it needs to use ClutterFlattenEffect to implement
the opacity property. The default implementation of the virtual
returns TRUE which means that most actors will end up being redirected
offscreen when the opacity != 255. ClutterTexture and ClutterRectangle
override this to return FALSE because they should never need to be
redirected. ClutterClone overrides it to divert to the source.
The values for the ClutterOffscreenRedirect enum have changed to:
AUTOMATIC_FOR_OPACITY
The actor will only be redirected if has_overlaps returns TRUE and
the opacity is < 255
ALWAYS_FOR_OPACITY
The actor will always be redirected if the opacity < 255 regardless
of the return value of has_overlaps
ALWAYS
The actor will always be redirected offscreen.
This means that the property can't be used to prevent the actor from
being redirected but only to increase the likelihood that it will be
redirected.
ClutterActor now adds and removes the flatten effect depending on
whether flattening is needed directly in clutter_actor_paint(). There
are new internal versions of add/remove_effect that don't queue a
redraw. This means that ClutterFlattenEffect is now just a no-op
subclass of ClutterOffscreen. It is only needed because
ClutterOffscreen is abstract. Removing the effect also makes it so
that the cached image will be freed as soon as an actor is repainted
without being flattened.
This adds a property which can be used to redirect the actor through
an FBO before painting so that it becomes flattened in an image. The
image can be used as a cache to avoid having to repaint the actor if
something unrelated in the scene changes. It can also be used to
implement correct opacity even if the actor has overlapping
primitives. The property is an enum that takes three values:
CLUTTER_OFFSCREEN_REDIRECT_NEVER: The default behaviour which is to
never flatten the actor.
CLUTTER_OFFSCREEN_REDIRECT_ALWAYS: The actor is always redirected
through an FBO.
CLUTTER_OFFSCREEN_REDIRECT_ONLY_FOR_OPACITY: The actor is only
redirected through an FBO if the paint opacity is not 255. This
value would be used if the actor wants correct opacity. It will
avoid the overhead of using an FBO whenever the actor is fully
opaque.
The property is implemented by installing a ClutterFlattenEffect.
ClutterFlattenEffect is a new internal class which subclasses
ClutterOffscreen to redirect the painting to an FBO. When
ClutterOffscreen paints, the effect sets an opacity override on the
actor so that the image will always contain the actor at full
opacity. The opacity is then applied to the resulting image before
painting it to the stage. This means the actor does not need to be
redrawn while the opacity is being animated.
The effect has a high internal priority so that it will always occur
before any other effects and it gets hidden from the application.
When calling clutter_actor_clear_constraints the layout of the actor
may change so we need to queue a relayout. Similarly when the effects
are cleared we need to queue a redraw.
This adds a priority property to all ClutterActorMetas. The
ClutterMetaGroup keeps the list sorted so that higher priority metas
remain at the beginning of the list. The priority is a signed integer
with the default as zero. An actor meta can therefore be put before
all default metas with a positive number, or after with a negative
number.
There are constants to set an 'internal' priority. The intention is
that applications wouldn't be allowed to use these values so that we
can keep special internal metas to that are before or after all
application metas.
The property isn't a real GObject property because for now it is
completely internal and only used to implement the 'transparency'
property of ClutterActor. ClutterMetaGroup doesn't currently resort
the list if the property changes so if we wanted to make it public we
should either make it construct-only or make the meta group listen for
changes on the property and resort accordingly.
The methods in ClutterActor that get the list of metas now use a new
function that filters out internal metas from the meta
group. Similarly for clearing the metas, the internal metas are left
in.
This adds a new public function to queue a rerun of an effect. If
nothing else queues a redraw then when the effect's actor is painted
the effect will be run without the CLUTTER_EFFECT_RUN_ACTOR_DIRTY
flag. This allows parametrised offscreen effects to report that they
need to redraw the image without having to redraw the underlying
actor. This will be used to implement the 'transparency' effect of
ClutterActor.
If multiple redraws are queued with different effects then redrawing
is started from the one that occurs last in the list of effects.
Internally the function is a wrapper around the new function
_clutter_actor_queue_redraw_full. This is intended to be the sole
point of code for queuing redraws on an actor. It has parameters for
the clip and the effect. The other two existing functions to queue a
redraw (one with a clip and one without) now wrap around this function
by passing a NULL effect.
This adds a new virtual to ClutterEffect which is intended to be a
more flexible replacement for the pre and post_paint functions. The
implementation of a run virtual would look something like this:
void
effect_run (ClutterEffect *effect,
ClutterEffectRunFlags flags)
{
/* Set up state */
/* ... */
/* Chain to the next item in the paint sequence */
clutter_actor_continue_paint (priv->actor);
/* Clean up state */
/* ... */
}
ClutterActor now just calls this virtual instead of the pre_paint and
post_paint functions. It keeps track of the next effect in the list so
that it knows what to do when clutter_actor_continue_paint is
called. clutter_actor_continue_paint is a new function added just for
implementing effects.
The default implementation of the run virtual just calls pre_paint and
post_paint so that existing effects will continue to work.
An effect is allowed to conditionally skip calling
clutter_actor_continue_paint(). This is useful to implement effects
that cache the image of an actor. The flags parameter can be used to
determine if the actor is dirty since the last paint. ClutterActor
sets this flag whenever propagated_one_redraw is TRUE which means that
a redraw for this actor or one of its children was queued.
The id pool used for the actor's id should be a per-stage field. At some
point, we might have a Stage mapped to multiple framebuffers, or each
Stage mapped to a different framebuffer; also, on backend with low
color precision we don't want to exhaust the size of the available ids
with a global pool. Finally, it's yet another thing we can remove from
the global Clutter context.
Having the id pool allocated per-stage, and the pick id assigned on
Actor:mapped will make the whole pick-id more reliable and future proof.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2633https://bugzilla.gnome.org/show_bug.cgi?id=647876
Do not use the generic GType class name: we have a :name property on
ClutterActor that is generally used for debugging purposes — so we
should use it when creating debugging spew in a consistent way.
This extends visualization for CLUTTER_PAINT=redraws so it now also
draws outlines for actors to show how they are being culled. Actors get
a green outline if they are fully inside the clip region, blue if fully
outside and greeny-blue if only partially inside.
This adds an internal _clutter_stage_get_active_framebuffer function
that can be used to get a pointer to the current CoglFramebuffer pointer
that is in use, in association with a given stage.
The "active" infix in the function name is there because we shouldn't
assume that a stage will always correspond to only a single framebuffer
so we aren't getting a pointer to a sole framebuffer, we are getting
a pointer to the framebuffer that is currently in use/being painted.
This API is now used for culling purposes where we need to check if we
are currently painting an actor to a framebuffer that is offscreen, that
doesn't correspond to the stage.
This renames the two internal functions _cogl_get_draw/read_buffer
as cogl_get_draw_framebuffer and _cogl_get_read_framebuffer. The
former is now also exposed as experimental API.
Previously we were applying the culling optimization to any actor
painted without considering that we may be painting to an offscreen
framebuffer where the stage clip isn't applicable.
For now we simply expose a getter for the current draw framebuffer
and we can assume that a return value of NULL corresponds to the
stage.
Note: This will need to be updated as stages start to be backed by real
CoglFramebuffer objects and so we won't get NULL in those cases.
To give quick visibility to the things going on relating to clipping and
culling this adds some more CLIPPING debug notes to clutter-actor.c and
clutter-stage.c
There's an optimisation in clutter-actor.c to avoid calculating the
last known paint volume whenever culling and clipped redraws are both
disabled. However there was a small thinko in the logic so that it
would also avoid calculating the paint volume whenever only one of the
debug flags is enabled. This fixes it to explicitly check that the two
flags are not both enabled before skipping the paint volume
calculation.
Previously each time we needed to retrieve the model transform for a
given actor we would call the apply_transform vfunc which would build up
a transformation matrix based on the actor's current anchor point, its
scale, its allocation and rotation. The apply_transform implementation
would repeatedly call API like cogl_matrix_rotate, cogl_matrix_translate
and cogl_matrix_scale.
All this micro matrix manipulation APIs were starting to show up in the
profiles of dynamic applications so this adds priv->transform matrix
cache which maintains the combined result of the actors scale, rotation
and anchor point etc. Whenever something like the rotation changes then
then the matrix is marked as dirty, but so long as the matrix isn't
dirty then the apply_transform vfunc now just calls cogl_matrix_multiply
with the cached transform matrix.
This implements a variation of frustum culling whereby we convert screen
space clip rectangles into eye space mini-frustums so that we don't have
to repeatedly transform actor paint-volumes all the way into screen
coordinates to perform culling, we just have to apply the modelview
transform and then determine each points distance from the planes that
make up the clip frustum.
By avoiding the projective transform, perspective divide and viewport
scale for each point culled this makes culling much cheaper.
OpenGL < 4.0 only supports integer based viewports and internally we
have a mixture of code using floats and integers for viewports. This
patch switches all viewports throughout clutter and cogl to be
represented using floats considering that in the future we may want to
take advantage of floating point viewports with modern hardware/drivers.
This is needed if an effect wants to temporarily override the paint
opacity. It needs to be able to restore the old opacity override in
the post_paint handler otherwise it would replace the effect of the
opacity override from any outer effects.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2541
The shader stack held by ClutterMainContext should only be accessed
using functions, and not directly.
Since it's a stack, we can use stack-like operations: push, pop and
peek.
Let's try and start reducing the size of ClutterActorPrivate by moving
some optional, out-of-band data from it to GObject data.
The ShaderData structure is a prime candidate for this migration: it
does not need to be inspected by the actor, and its relationship with an
actor is transient and optional.
By attaching it to the actor's instance through g_object_set_data() we
neatly tie its lifetime to the instance, and we don't have to care
cleaning it up in the finalize()/dispose() implementation of
ClutterActor itself.
In _clutter_actor_queue_redraw_with_clip it has a local variable to
mark when a new paint volume for the clip is created so that it can be
freed when the function returns. However the actual code to free the
paint volume went missing in 3b789490d2 so the variable did
nothing. This patch just adds the free back in.
This time, in Clutter core.
The ObjC standard library provides a type called 'id', which obviously
requires any library to either drop the useful shadowed variable warning
or stop using 'id' as a variable name.
Yes, it's almost unbearably stupid. Well, at least it's not 'index' in
string.h, or 'y2' in math.h.
The 'in_clone_paint' parameter of the private function
_clutter_actor_set_in_clone_paint() shadowed the private function
in_clone_paint(). Rename this parameter to 'is_in_clone_paint' to remove
a compiler warning.
Replace the opacity_parent with an opacity_override variable, to allow
direct overriding of the paint opacity and simplify this mechanism
somewhat.
This also required a new private flag, in_clone_paint, to maintain the
functionality of the public function clutter_actor_is_in_clone_paint()
Don't use ugly "#undef CLUTTER_DISABLE_DEPRECATED" inside source code
using deprecated symbols; we have the handy CLUTTER_COMPILATION define
that we can use as part of the "disable deprecated" conditional.
Use the internal child list for the default map/unmap vfuncs. This removes
the requirement for non-container composite actors to implement their own
map/unmap functions.
Unrealizing an actor is a recursive process that needs to traverse the
children of an actor to ensure they are also unrealized. This maintains
the invariant that if any given actor is marked as unrealized then you
know that all its children have also been unrealized.
The previous implementation would use the container interface's
foreach_with_internals vfunc to explicitly traverse the children of
container actors but this didn't consider composite actors that aren't
containers.
Since clutter-actor now maintains an explicit list of children we can
also handle composite actors that aren't containers using
_clutter_actor_traverse.
This makes it possible to choose the traversal order; either depth first
or breadth first and when visiting actors in a depth first order there
is now a callback called before children are traversed and one called
after. Some tasks such as unrealizing actors need to explicitly control
the traversal order to maintain the invariable that all children of an
actor are unrealized before we actually mark the parent as unrealized.
The callbacks are now passed the relative depth in the graph of the
actor being visited and instead of only being able to return a boolean
to bail out of further traversal it can now do one of: continue,
skip_children or break. To implement something like unrealize it's
desirable to skip children that you find have already been unrealized.
The last_paint_box for an actor represents its "normal" position - we
shouldn't update it or use it to cull drawing if we are painting
a clone of the actor. Tracking whether we are painting a clone is
done by adding _clutter_actor_push/pop_clone_paint() and a global
"clone paint level".
http://bugzilla.clutter-project.org/show_bug.cgi?id=2396
When transforming a paint-volume or transforming allocation vertices we
are transforming more than one point at a time so we can batch those
together with cogl_matrix_transform_points instead of
cogl_matrix_transform_point. Also in both of these cases we don't need
to do a projective transform so using cogl_matrix_transform_points also
lets us reduce the per-vertex computation.
The Behaviour class and its implementations have been replaced by the
new animation framework API and by the constraints for layout-related
animations.
Currently, we need to make tests build, so we undef DISABLE_DEPRECATED
in specific test cases while they get ported.
The paint volume structure is cached in the Actor it references, and
this causes a reference cycle.
The paint volume is going to be used when painting, so the actor must
still be valid - otherwise Clutter will bail out far before than
accessing the actor pointer in ClutterPaintVolume.
Otherwise, we could have used dispose() to check for a valid actor and
remove a reference if the actor field is !NULL; it feels less clean,
though, since we're effectively managing an extra reference on
ourselves.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2431
For internal usage, writing:
clutter_actor_get_name (actor) != NULL
? clutter_actor_get_name (actor)
: G_OBJECT_TYPE_NAME (actor)
is overly verbose and does two type checks. A simple, internal method
for getting the same result without type checks would be much more
appreciated.
Instead of waiting until clutter_actor_paint to check if there are any
handlers connected to the "paint" signal, we now do the check whenever
the paint-volume is requested in _actor_get_paint_volume_mutable().
Previously we checked in clutter_actor_paint(), but at that time we may
already be using a stage clip that could be derived from an invalid
paint-volume. We used to try and handle that by queuing a follow up,
unclipped, redraw but anyway there was an additional problem with the
previous approach because the checking wasn't enough to always catch
invalid volumes involved in culling (considering that containers may
derive their volume from children that haven't yet been painted)
By moving the check to _get_paint_volume time not only do we now
correctly check children in cases where a container derives its volume
from its children's volumes but we no longer need to queue follow up
redraws to cover up artefacts.
Since we now never queue follow up redraws, this in turn means we should
no longer clobber redraws queued with an explicit clip which was
something affecting gnome-shell since it connects a handler to the paint
signal of the stage.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2388
Once an actor had _clutter_stage_queue_redraw_entry_invalidate()
called on it once, then priv->queue_redraw_entry would point to
an entry with entry->actor NULL. _clutter_stage_queue_actor_redraw()
doesn't handle this case and no further redraws would be queued.
To fix this, NULL out priv->queue_redraw_entry() and then make sure
we free the invalidated entry in
_clutter_stage_maybe_finish_queue_redraws() just as we do for
still valid entries.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2389
We need to make sure that redraws queued for actors on a stage are for
actors actually in the stage. So in clutter_actor_unparent() descend
through the children and remove redraws. Just removing the actor itself
isn't good enough since an entire hierarchy can be removed from the
stage without breaking it up into individual actors.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2359
This is based on an original patch from Owen Taylor who debugged the
root cause of this bug; thanks.
Instead of delegating the check for the ActorMeta:enabled property to
the sub-classes of ClutterActorMeta, ClutterActor can do the check prior
to using the ClutterActorMeta instances.
In all the changes made recently to how we handle redraws and adding
support for paint-volumes we stopped looking at explicit clip regions
passed to _clutter_actor_queue_redraw_with_clip.
In _clutter_actor_finish_queue_redraw we had started always trying to
clip the redraw to the paint-volume of the actor, but forgot to consider
that the user may have already determined the clip region for us!
Now we first check if the given clip != NUll and if so we don't need to
calculate the paint-volume of the actor.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2349
Since re-working how redraws are queued it is no longer necessary to
dirty the pick buffer in _clutter_actor_real_queue_redraw since this
should now reliably be handled in _clutter_stage_queue_actor_redraw.
This adds two internal functions relating to explicit traversal of the
scenegraph:
_clutter_actor_foreach_child
_clutter_actor_traverse
_clutter_actor_foreach_child just iterates the immediate children of an
actor, and with a new ClutterForeachCallback type it allows the
callbacks to break iteration early.
_clutter_actor_traverse traverses the given actor and all of its
decendants. Again traversal can be stopped early if a callback returns
FALSE.
The first intended use for _clutter_actor_traverse is to maintain a
cache pointer to the stage for all actors. In this case we will need to
update the pointer for all descendants of an actor when an actor is
reparented in any way.
This adds a private getter to query the number of children an actor has.
One use planned for this API is to avoid calling get_paint_volume on
such actors. (It's not clear what the best semantics for
get_paint_volume are for actors with children, so we are considering
leaving the semantics undefined for the initial clutter 1.4 release)
We now explicitly track the list of children each actor has in a private
GList. This gives us a reliable way to know how many children an actor
has - even for composite actors that don't implement the container
interface. This also will allow us to directly traverse the scenegraph
in a more generalized fashion. Previously the scenegraph was
more-or-less represented implicitly according the implementation of
paint methods.
When using the CLUTTER_PAINT=paint-volumes debug option we try and show
when a paint volume couldn't be determined by drawing a blue outline of
the allocation instead. There was a typo though and instead we were
drawing an outline the size of the stage instead of for the given actor.
This fixes that and removes a FIXME comment relating to the blue outline
that is now implemented.
This reverts commit ca44c6a7d8abe9f2c548bee817559ea8adaa7a80.
In reality there are probably lots of actors that depend on the exact
semantics as they are documented so this change isn't really acceptable.
For example when the font changes in ClutterText we only queue a
relayout, and since it's possible that the font will have the same size
and the actor won't get a new allocation it wouldn't otherwise queue a
redraw.
Since queue_redraw requests now get deferred until just before a paint
run it is actually no longer a problem to queue the redraw here.
Instead of immediately, recursively emitting the "queue-redraw" signal
when clutter_actor_queue_redraw is called we now defer this process
until all stage updates are complete. This allows us to aggregate
repeated _queue_redraw requests for the same actor avoiding redundant
paint volume transformations. By deferring we also increase the
likelihood that the actor will have a valid paint volume since it will
have an up to date allocation; this in turn means we will more often be
able to automatically queue clipped redraws which can have a big impact
on performance.
Here's an outline of the actor queue redraw mechanism:
The process starts in clutter_actor_queue_redraw or
_clutter_actor_queue_redraw_with_clip.
These functions queue an entry in a list associated with the stage which
is a list of actors that queued a redraw while updating the timelines,
performing layouting and processing other mainloop sources before the
next paint starts.
We aim to minimize the processing done at this point because there is a
good chance other events will happen while updating the scenegraph that
would invalidate any expensive work we might otherwise try to do here.
For example we don't try and resolve the screen space bounding box of an
actor at this stage so as to minimize how much of the screen redraw
because it's possible something else will happen which will force a full
redraw anyway.
When all updates are complete and we come to paint the stage (see
_clutter_stage_do_update) then we iterate this list and actually emit
the "queue-redraw" signals for each of the listed actors which will
bubble up to the stage for each actor and at that point we will
transform the actors paint volume into screen coordinates to determine
the clip region for what needs to be redrawn in the next paint.
Note: actors are allowed to queue a redraw in reseponse to a
queue-redraw signal so we repeat the processing of the list until it
remains empty. An example of when this happens is for Clone actors or
clutter_texture_new_from_actor actors which need to queue a redraw if
their source queues a redraw.
This makes clutter_actor_queue_redraw simply bail out early if the actor
isn't a descendant of a ClutterStage since the request isn't meaningful
and it avoids a crash when trying to queue a clipped redraw against the
stage to clear the actors old location.
This splits out all the clutter_paint_volume code from clutter-actor.c
into clutter-paint-volume.c. Since clutter-actor.c and
clutter-paint-volume.c both needed the functionality of
_fully_transform_vertices, this function has now been moved to
clutter-utils.c as _clutter_util_fully_transform_vertices.
There are too many examples where the default assumption that an actor
paints inside its allocation isn't true, so we now return FALSE in the
base implementation instead. This means that by default we are saying
"we don't know the paint volume of the actor", so developers need to
implement the get_paint_volume virtual to take advantage of culling and
clipped redraws with their actors.
This patch provides very conservative get_paint_volume implementations
for ClutterTexture, ClutterCairoTexture, ClutterRectangle and
ClutterText which all explicitly check the actor's object type to avoid
making any assumptions about subclasses.
We were always explicitly checking priv->needs_allocation in
_clutter_actor_queue_redraw_with_clip, but we only need to do that if
the CLUTTER_REDRAW_CLIPPED_TO_ALLOCATION flag is used.
This initializes priv->last_paint_box with a degenerate box, so a newly
allocated actor added to the scenegraph and made visible only needs to
trigger a redraw of its initial position. If we don't have a valid
last_paint_box though we would instead trigger a full stage redraw.
To make comparing the performance with culling/clipped redraws
enabled/disabled fairer we now avoid querying the paint box when they
are disabled, so that results should reflect how the cost of
transforming paint volumes into screen space etc gets offset against the
benefit of culling.
We have bent the originally documented semantics a bit so now where we
say "Queueing a new layout automatically queues a redraw as well" it
might be clearer to say "Queuing a new layout implicitly queues a redraw
as well if anything in the layout changes".
This should be close enough to the original semantics to not cause any
problems.
Without this change then we we fail to take advantage of clipped redraws
in a lot of cases because queuing a redraw with priv->needs_allocation
== TRUE will automatically be promoted to a full stage redraw since it's
not possible to determine a valid paint-volume.
Also queuing a redraw here will end up registering a redundant clipped
redraw for the current location, doing quite a lot of redundant
transforms, and then later when re-allocated during layouting another
queue redraw would happen with the correct paint-volume.
This uses actor paint volumes to perform culling during
clutter_actor_paint.
When performing a clipped redraw (because only a few localized actors
changed) then as we traverse the scenegraph painting the actors we can
now ignore actors that don't intersect the clip region. Early testing
shows this can have a big performance benefit; e.g. 100% fps improvement
for test-state with culling enabled and we hope that there are even much
more compelling examples than that in the real world,
Most Clutter applications are 2Dish interfaces and have quite a lot of
actors that get continuously painted when anything is animated. The
dynamic actors are often localized to an area of user focus though so
with culling we can completely avoid painting any of the static actors
outside the current clip region.
Obviously the cost of culling has to be offset against the cost of
painting to determine if it's a win, but our (limited) testing suggests
it should be a win for most applications.
Note: we hope we will be able to also bring another performance bump
from culling with another iteration - hopefully in the 1.6 cycle - to
avoid doing the culling in screen space and instead do it in the stage's
model space. This will hopefully let us minimize the cost of
transforming the actor volumes for culling.
This makes clutter_actor_queue_redraw transparently use an actor's paint
volume to queue a clipped redraw.
We save the actors paint box each time it is painted so that when
clutter_actor_queue_redraw is called we can determine the old and new
location of the actor so we know the full bounds of what must be redrawn
to clear its old view and show the new.
This makes _clutter_actor_transform_and_project_box a static function
and removes the prototype from clutter-private.h since it is no longer
used outside clutter-actor.c
The base implementation for the actor queue_relayout method was queuing
an implicit redraw, but there shouldn't be anything implied from the
mere process of queuing a redraw that should force us to queue a redraw.
If actors are moved as a part of relayouting later then they will queue
a redraw. Also clutter_actor_queue_relayout() still also explicitly
queues a redraw so I think this may have been doubly redundant.
If clutter_actor_allocate finds it necessary to update an actors
allocation then it now also queue a redraw of that actor. Currently we
queue redraws for actors very early on when queuing a relayout instead
of waiting to determine the final outcome of relayouting to determine if
a redraw is really required. With this in place we can move away from
preemptive queuing of redraws.
clutter_actor_queue_relayout currently queues a relayout and a redraw,
but the plan is to change it to only queue a relayout and honour the
documentation by assuming that the process of relayouting will
result queuing redraws for any actors whos allocation changes.
This doesn't make that change it just adds an internal
_clutter_actor_queue_only_relayout function which
clutter_actor_queue_relayout now uses as well as calling
clutter_actor_queue_redraw.
There is an internal _clutter_actor_queue_redraw_with_clip API that gets
used for texture-from-pixmap to minimize what we redraw in response to
Damage events. It was previously working in terms of a ClutterActorBox
but it has now been changed so an actor can queue a redraw of volume
instead.
The plan is that clutter_actor_queue_redraw will start to transparently
use _clutter_actor_queue_redraw_with_clip when it can determine a paint
volume for the actor.
This adds a debug option to visualize the paint volumes of all actors.
When CLUTTER_PAINT=paint-volumes is exported in the environment before
running a Clutter application then all actors will have their bounding
volume drawn in green with a label corresponding to the actors type.
This is a fairly extensive second pass at exposing paint volumes for
actors.
The API has changed to allow clutter_actor_get_paint_volume to fail
since there are times - such as when an actor isn't a descendent of the
stage - when the volume can't be determined. Another example is when
something has connected to the "paint" signal of the actor and we simply
have no way of knowing what might be drawn in that handler.
The API has also be changed to return a const ClutterPaintVolume pointer
(transfer none) so we can avoid having to dynamically allocate the
volumes in the most common/performance critical code paths. Profiling was
showing the slice allocation of volumes taking about 1% of an apps time,
for some fairly basic tests. Most volumes can now simply be allocated on
the stack; for clutter_actor_get_paint_volume we return a pointer to
&priv->paint_volume and if we need a more dynamic allocation there is
now a _clutter_stage_paint_volume_stack_allocate() mechanism which lets
us allocate data which expires at the start of the next frame.
The API has been extended to make it easier to implement
get_paint_volume for containers by using
clutter_actor_get_transformed_paint_volume and
clutter_paint_volume_union. The first allows you to query the paint
volume of a child but transformed into parent actor coordinates. The
second lets you combine volumes together so you can union all the
volumes for a container's children and report that as the container's
own volume.
The representation of paint volumes has been updated to consider that
2D actors are the most common.
The effect apis, clutter-texture and clutter-group have been update
accordingly.
An Effect implementation might override the paint volume of the actor to
which it is applied to. The get_paint_volume() virtual function should
be added to the Effect class vtable so that any effect can get the
current paint volume and update it.
The clutter_actor_get_paint_volume() function becomes context aware, and
does the right thing if called from within a ClutterEffect pre_paint()
or post_paint() implementation, by allowing all effects in the chain up
to the caller to modify the paint volume.
An actor has an implicit "paint volume", that is the volume in 3D space
occupied when painting itself.
The paint volume is defined as a cuboid with the origin placed at the
top-left corner of the actor; the size of the cuboid is given by three
vectors: width, height and depth.
ClutterActor provides API to convert the paint volume into a 2D box in
screen coordinates, to compute the on-screen area that an actor will
occupy when painted.
Actors can override the default implementation of the get_paint_volume()
virtual function to provide a different volume.
This reorganizes the loop for clutter_actor_contains so that it is a
for loop rather than a while loop. Although this is mostly just
nitpicking, I think this change could make the loop slightly faster if
not optimized because it doesn't perform the self == descendant check
twice and it is clearer.
The documentation for clutter_actor_contains didn't specify what
happens when self==descendant. A strict reading of it might lead you
to think that it would return FALSE because in that case the
descendant isn't an immediate child or a deeper descendant. The code
actually would return TRUE. I think this is more useful so this patch
fixes the docs rather than the code.
