We still ship clutter_get_show_fps() and clutter_get_debug_enabled() as
public entry points. Yet another case of missing API review prior to the
1.0 release, so really the bucket stops around my desk.
Let's deprecate these two useless functions, and reduce the API
footprint of Clutter.
This function should have never been made public in the first place; its
output depends on a configuration option of Clutter, and it's basically
useful only for internal debugging.
Make it consistent across the various build options (with or without
profiling enabled), and add a timestamp using the monotonic clock to
every debug message.
Setting the default frame rate does not do anything even remotely
useful, unless synchronization to the vertical refresh rate is also
disabled - which can only be done through environment variable or
through configuration file. Having a programmatic way to change the
default frame rate is, thus, completely pointless.
Changing the default frame rate through environment variable and
configuration file is still allowed.
Instead of defining new symbols for the windowing systems enabled at
configure time, we can reuse the same symbols for both the compile time
and run time checks, e.g.:
#ifdef CLUTTER_WINDOWING_X11
if (clutter_check_windowing_backend (CLUTTER_WINDOWING_X11))
/* use the clutter_x11_* API */
else
#endif
#ifdef CLUTTER_WINDOWING_WIN32
if (clutter_check_windowing_backend (CLUTTER_WINDOWING_WIN32))
/* use the clutter_win32_* API */
#endif
This scheme allows us to ensure that the input system namespace is free
for us to use and select at run time in later versions of Clutter.
Portable code should be allowed to check type backend currently being
used, so that it can use platform-specific API (not just Clutter's).
We don't want to go down the GDK route, with public types for
ClutterBackend and ClutterStageWindow implementations, and use the type
system, e.g.:
#ifdef GDK_WINDOWING_X11
if (GDK_IS_WINDOW_X11 (window))
use_x11_api (window);
else
#endif
#ifdef GDK_WINDOWING_WIN32
if (GDK_IS_WINDOW_WIN32 (window))
use_win32_api (window);
else
#endif
g_critical ("Unsupported backend");
This system would make us expose the backend system, and we want to
still reserve us the option to change the backend system to increase its
granularity — e.g. choosing different input event systems regardless of
the windowing system.
This commit adds a simple function that checks the backend type against
a symbolic constant — the same constant string that can be used to
select the backend at run-time through the CLUTTER_BACKEND environment
variable.
The Clutter backend split is opaque enough that should allow us to just
build all possible backends inside the same shared object, and select
the wanted backend at initialization time.
This requires some work in the build system, as well as the
initialization code, to remove duplicate functions that might cause
conflicts at build and link time. We also need to defer all the checks
of the internal state of the platform-specific API to run-time type
checks.
Since the Windows GUI system is assuming multithreadedness, initializing
locks after entering the GUI portion on Windows is likely to cause
problems[1][2], which results many Clutter programs to crash due to
releasing resources that they did not own.
[1]: Multi-threaded use of GTK+ on Win32 in README.win32 of GTK+
source package
[2]: Explanation of Windows GUI system regarding its multithreadness
assumptions-
http://mail.gnome.org/archives/gtk-list/2011-June/msg00005.html
Or, better, the fact that the behaviour of any Clutter function will be
undefined in case the initialization fails.
The value returned by clutter_init() and friends has to be handled
properly.
The CLUTTER_DEBUG class of debugging flags is meant for debugging notes,
while the CLUTTER_PAINT debugging flags are for changing the output of
the paint cycle. Painting the DeformEffect tiles should go in the latter.
On top of the existing "Settings" group in the settings.ini file we
should have two more groups:
Environment - contains all the configuration possible through
environment variables
Debug - contains all the possible debug variables
ClutterSettings should be able to load its initial state by using
configuration files in SYSCONFDIR and XDG_CONFIG_HOME. This allows
Clutter to have a system (and user) configuration on platforms that
do not have XSETTINGS bridges.
We already have two mechanisms for controlling the font rendering
quality on a per-application basis:
• ClutterSettings properties
• clutter_backend_set_font_options()
The font flags were always a stop-gap solution, and one that tried to
simplify a fairly complex issue beyond the point of actually being
useful.
https://bugzilla.gnome.org/show_bug.cgi?id=660786
The grab API is a relic of Clutter 0.6, and hasn't been through proper
vetting in a *long* time — mostly due to the fact that we don't really
like grabs, and point to the ::captured-event as a way to implement
"soft grabs" in toolkits and applications.
The implementation of full and device grabs uses weak references on
actors instead of using the ::destroy signal, which is meant exactly for
the case of releasing pointers to actors when they are disposed.
The API naming scheme is also fairly broken, especially for
device-related grabs.
Finally, keyboard device grabs are just not implemented.
We can, in one go, clean up this mess and deprecate a bunch of badly
named API by introducing generic device grab/ungrab methods on
ClutterInputDevice, and re-implement the current API on top of them.
GLib deprecated g_thread_init(), and threading support is initialized
by GObject, so Clutter already runs with threading support enabled. We
can drop the clutter_threads_init() call requirement, and initialize the
Big Clutter Lock™ on clutter_init(). This reduces the things that have
to be done when dealing with threads with Clutter, and the things that
can possibly go wrong.
The Big Clutter Lock™ can now be a static GMutex, since GLib supports
them. We can also drop a bunch of checks given the recent changes in
GLib threading API.
When testing the performance of an application, it's often useful to
force it to continuously redraw instead of going idle to help measure
the frame rate. This just adds a CLUTTER_PAINT=continuous-redraw which
causes the master clock to queue a redraw on all of the stages
just before it prepares its source.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
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>
The repaint functions list can (and should) be manipulated from
different threads, but it currently doesn't prevent multiple threads
from accessing it concurrently. We should have a simple lock and take it
when adding and removing elements from the list; the invocation is still
performed under the Big Clutter Lock™, so it doesn't require special
handling.
It is possible, by calling clutter_set_motion_events_enabled() prior to
the creation of any stage, to control the per-actor motion event
delivery flag on each newly created stage. Since we deprecated the
global accessor functions in favour of the per-Stage ones, we need to
remove the call to clutter_get_motion_events_enabled() inside the
ClutterStage instance initialization, and replace it with an internal
function.
This code will go away when we can finally break API and remove the
deprecated functions.
Complete the quest of commit bc548dc862
by making the ClutterStage methods for controlling the per-actor motion
and crossing event delivery public, and deprecating the global ones.
When using CLUTTER_PAINT=paint-volumes to visualize the paint-volumes of
actors we were already disabling clipped-redraws because we are drawing
extra geometry that the actors don't know about but we didn't disable
culling. This was resulting in actors disappearing while using this
debug option.
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.
In Cogl, cogl-pango.h has moved to <cogl-pango/cogl-pango.h>. When
using the experimental 2.0 API (which Clutter does) it is no longer
possible to include it under the old name of <cogl/cogl-pango.h> so we
need to update the include location.
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
Cogl has now been split out into a standalone project with a separate
repository at git://git.gnome.org/cogl. From now on the Clutter build
will now simply look for a cogl-1.0 pkg-config file to find a suitable
Cogl library to link against at build time.
This backend hasn't been used for years now and so because it is
untested code and almost certainly doesn't work any more it would be a
burdon to continue trying to maintain it. Considering that we are now
looking at moving OpenGL window system integration code down from
Clutter backends into Cogl that will be easier if we don't have to
consider this backend.
This makes it possible to build Clutter against a standalone build of
Cogl instead of having the Clutter build traverse into the clutter/cogl
subdirectory.
This moves the implementation of _clutter_do_pick to clutter-stage.c and
renames it _clutter_stage_do_pick. This function can be compared to
_clutter_stage_do_update/redraw in that it prepares for and starts a
traversal of a scenegraph descending from a given stage. Since it is
desirable that this function should have access to the private state of
the stage it is awkward to maintain outside of clutter-stage.c.
Besides moving _clutter_do_pick this patch is also able to remove the
following private state accessors from clutter-stage-private.h:
_clutter_stage_set_pick_buffer_valid,
_clutter_stage_get_pick_buffer_valid,
_clutter_stage_increment_picks_per_frame_counter,
_clutter_stage_reset_picks_per_frame_counter and
_clutter_stage_get_picks_per_frame_counter.
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.
Anything that is not CLUTTER_INIT_SUCCESS is to be considered an error.
This fixes the Clutter initialization sequence to actually error out
on pre-conditions and backend initialization failures.
Once upon a time, the land of Clutter had a stage singleton. It was
created automatically at initialization time and stayed around even
after the main loop was terminated. The singleton was content in
being all there was. There also was a global API to handle the
configuration of the stage singleton that would affect the behaviour
on other classes, signals and properties.
Then, an evil wizard came along and locked the stage singleton in his
black tower, and twisted it until it was possible to create new stages.
These new stages were pesky, and didn't have the same semantics of the
singleton: they didn't stay around when closed, or terminate the main
loop on delete events.
The evil wizard also started moving all the stage-related API from the
global context into class-specific methods.
Finally, the evil wizard cast a spell, and the stage singleton was
demoted to creation on demand - and until somebody called the
clutter_stage_get_default() function, the singleton remained in a limbo
of NULL pointers and undefined memory areas.
There was a last bit - literally - of information still held by the
global API; a tiny, little flag that disabled per-actor motion events.
The evil wizard added private accessors for it, and stored it inside the
stage private structure, in preparation for a deprecation that would
come in a future development cycle.
The evil wizard looked down upon the land of Clutter from the height of
his black tower; the lay of the land had been reshaped into a crucible
of potential, and the last dregs of the original force of creation were
either molted into new, useful shapes, or blasted away by the sheer fury
of his will.
All was good.
The clutter-id-pool.h header is private and not installed; yet, all the
clutter_id_pool_* symbols are public. Let's correct this oversight we've
been stringing along since forever.
Only allow access to the ClutterMainContext through the private
_clutter_context_get_default() function, so we can easily grep
it and remove the unwanted usage of the global context.
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.
The _clutter_do_redraw() function should really be moved inside
ClutterStage, since all it does is calling private stage and
backend functions. This also allows us to change a long-standing
issue with a global fps counter for all stages, instead of a\
per-stage one.
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.
As the prelude to deprecation of the function in 1.8, let's move the
implementation to an internal function, and use that instead of the
public facing one.
The GQueue that stores the global events queue is handled all over the
place:
• the structure is created in _clutter_backend_init_events();
• the queue is handled in clutter-event.c, clutter-stage.c and
clutter-backend.c;
• ClutterStage::dispose cleans up the events associated with
the stage being destroyed;
• the queue is destroyed in ClutterBackend::dispose.
Since we need to have access to it in different places we cannot put it
inside ClutterBackendPrivate, hence it should stay in ClutterMainContext;
but we should still manage it from just one place - preferably by the
ClutterEvent API only.
Clutter should just require that the windowing system used by a backend
adds a device to the stage when the device enters, and removes it from
the stage when the device leaves; with this information, we can
synthesize every crossing event and update the device state without
other intervention from the backend-specific code.
The generation of additional crossing events for actors that are
covering the stage at the coordinates of the crossing event should be
delegated to the event processing code.
The x11 and win32 backends need to be modified to relay the enter and
leave events from the windowing system.
This adds a transparent optimization to cogl_read_pixels for when a
single pixel is being read back and it happens that all the geometry of
the current frame is still available in the framebuffer's associated
journal.
The intention is to indirectly optimize Clutter's render based picking
mechanism in such a way that the 99% of cases where scenes are comprised
of trivial quad primitives that can easily be intersected we can avoid
the latency of kicking a GPU render and blocking for the result when we
know we can calculate the result manually on the CPU probably faster
than we could even kick a render.
A nice property of this solution is that it maintains all the
flexibility of the render based picking provided by Clutter and it can
gracefully fall back to GPU rendering if actors are drawn using anything
more complex than a quad for their geometry.
It seems worth noting that there is a limitation to the extensibility of
this approach in that it can only optimize picking a against geometry
that passes through Cogl's journal which isn't something Clutter
directly controls. For now though this really doesn't matter since
basically all apps should end up hitting this fast-path. The current
idea to address this longer term would be a pick2 vfunc for ClutterActor
that can support geometry and render based input regions of actors and
move this optimization up into Clutter instead.
Note: currently we don't have a primitive count threshold to consider
that there could be scenes with enough geometry for us to compensate for
the cost of kicking a render and determine a result more efficiently by
utilizing the GPU. We don't currently expect this to be common though.
Note: in the future it could still be interesting to revive something
like the wip/async-pbo-picking branch to provide an asynchronous
read-pixels based optimization for Clutter picking in cases where more
complex input regions that necessitate rendering are in use or if we do
add a threshold for rendering as mentioned above.
Move the private Backend API to a separate header.
This also allows us to finally move the class vtable and instance
structure to a separate file and plug the visibility hole that left
the Backend class bare for everyone to poke into.
When we don't use a window system drawable, we can't query the color
masks at context initialization time. Do it lazily so we're sure to have
a current context with a valid framebuffer.
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 adds a private ->relayout_pending boolean similar in spirit to
redraw_pending. This will allow us to queue a relayout without
implicitly queueing a redraw; instead we can depend on the actions
of a relayout to queue any necessary redraw.
This ensures that clipped redraws are disabled when using
CLUTTER_PAINT=redraws. This may seem unintuitive given that this option
is for debugging clipped redraws, but we can't draw an outline outside
the clip region and anything we draw inside the clip region is liable to
leave a trailing mess on the screen since it won't be cleared up by
later clipped redraws.
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.
We have an optimization to track when there are multiple picks per
frame so we can do a full render of the pick buffer to reduce the
number of pick renders for a static scene.
There was a problem though in that we were tracking this information in
the ClutterMainContext, but conceptually this doesn't really make sense
because the pick buffer is associated with a stage framebuffer and there
can be multiple stages for one context.
This patch moves the state tracking to ClutterStage.
This reverts commit d7e86e2696.
This was a half baked patch that was pushed a bit early since it broke
test-texture-pick-with-alpha + the commit message refers to a change on
the wip/paint-box branch that hasn't happened yet.
We have an optimization to track when there are multiple picks per
frames so we can do a full render of the pick buffer to reduce the
number of pick renders for a static scene.
There were two problems with how we were tracking this state though.
Firstly we were tracking this information in the ClutterMainContext, but
conceptually this doesn't really make sense because the pick buffer is
associated with a stage framebuffer and there can be multiple stages for
one context. Secondly - since the change to how redraws are queued - we
weren't marking the pick buffer as invalid when a queuing a redraw, we
were only marking the buffer invalid when signaling/finishing the
queue-redraw process, which is now deferred until just before a paint.
This meant using clutter_stage_get_actor_at_pos after a scenegraph
change could give a wrong result if it just read from an existing (but
technically invalid) pick buffer.
This patch moves the state tracking to ClutterStage, and ensures the
buffer is invalidated in _clutter_stage_queue_actor_redraw.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2283
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
When building with --enable-profile we now depend on the uprof-0.3
developer release which brings a few improvements:
» It lets us "fix" how we initialize uprof so that instead of using a shared
object constructor/destructor (which was a hack used when first adding
uprof support to Clutter) we can now initialize as part of clutter's
normal initialization code. As a side note though, I found that the way
Clutter initializes has some quite serious problems whenever it
involves GOptionGroups. It is not able to guarantee the initialization
of dependencies like uprof and Cogl. For this reason we still use the
contructor/destructor approach to initialize uprof in Cogl.
» uprof-0.3 provides a better API for adding custom columns when reporting
timer and counter statistics which lets us remove quite a lot of manual
report generation code in clutter-profile.c.
» uprof-0.3 provides a shared context for tracking mainloop timer
statistics. This means any mainloop based library following the same
"Mainloop" timer naming convention can use the shared context and no
matter who ends up owning the final mainloop the statistics will always
be in the same place. This allows profiling of Clutter with an
external mainloop such as with the Mutter compositor.
» uprof-0.3 can export statistics over dbus and comes with an ncurses
based ui to vizualize timer and counter stats live.
The latest version of uprof can be cloned from:
git://github.com/rib/UProf.git
When building actor relative transforms, instead of using the matrix
stack to combine transformations and making assumptions about what is
currently on the stack we now just explicitly initialize an identity
matrix and apply transforms to that.
This removes the full_vertex_t typedef for internal transformation code
and we just use ClutterVertex.
ClutterStage now implements apply_transform like any other actor now
and the code we had in _cogl_setup_viewport has been moved to the
stage's apply_transform instead.
ClutterStage now tracks an explicit projection matrix and viewport
geometry. The projection matrix is derived from the perspective whenever
that changes, and the viewport is updated when the stage gets a new
allocation. The SYNC_MATRICES mechanism has been removed in favour of
_clutter_stage_dirty_viewport/projection() APIs that get used when
switching between multiple stages to ensure cogl has the latest
information about the onscreen framebuffer.
In 965907deb3 the picking was changed to render the full stage
instead of a single pixel whenever picking is performed more than once
between paints. However the condition in the if-statement was
backwards so it would end up always doing a full stage render.
The idea is that if we see multiple picks per frame then that implies
the visible scene has become static. In this case we can promote the
next pick render to be unclipped so we have valid pick values for the
entire stage. Now we can continue to read from this cached buffer until
the stage contents do visibly change.
Thanks to Luca Bruno on #clutter for this idea!
The P_() macro adds a context for the property nick and blurb. In order
to make xgettext recognize it, we need to drop glib-gettexize inside the
autogen.sh script and ship a modified Makefile.in.in with Clutter.
