If we manually wait for the VBLANK with:
- SGI_video_sync
- Direct usage of the DRM ioctl
Then we should call glFinish() first, or otherwise the swap-buffers
may be delayed by pending drawing and cause a tear.
http://bugzilla.openedhand.com/show_bug.cgi?id=1636
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
commit e2c4a2a9f8 fixed one thing but broke many others things :-/
hopfully this fixes that.
It turned out that the journal was mistakenly setting the OVERRIDE_LAYER0
flush option for all entries, but some other logic errors were also
uncovered in _cogl_material_equal.
Added an internal clutter function, _clutter_master_clock_ensure_next_iteration
that ensures another iteration of the master clock, can be called from repaint
functions as well as other threads.
To help us handle sliced textures; When flushing materials there is an
override option that can be given to replace the texture name for layer0
so we may iterate the slices without needing to modify the material
in use.
Since improving the journal's ability to batch state changes we added a
_cogl_material_equals function that is used by the journal to compare
materials and identify when a state change is required, but this wasn't
correctly considering the layer0 override resulting in false positives that
meant the journal wouldn't update the GL state and the first texture name
was used for all slices.
The cost of glGetFloatv with Mesa is still representing a majority of our
time in OpenGL for some applications, and the last thing left using this is
the current-matrix API when getting the projection matrix.
This adds a matrix stack for the projection matrix, so all getting, setting
and modification of the projection matrix is now managed by Cogl and it's only
when we come to draw that we flush changes to the matrix to OpenGL.
This also brings us closer to being able to drop internal use of the
deprecated OpenGL matrix functions, re: commit 54159f5a1d
The clutter_actor_get_allocation_coords() is not used, and since
the switch to floats in the Actor's API, it returns exactly what
the get_allocation_box() returns.
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
Scanners like gtk-doc and g-ir-scanner get confused by:
typedef struct _Foo {
...
} Foo;
And expect instead:
typedef struct _Foo Foo;
struct _Foo {
...
};
CoglMatrix definition should be changed to avoid the former type.
The race we were experiencing in the X11 backends is apparently
back after the fix in commit 00a3c698.
This time, just delaying the setting of the SYNC_MATRICES flag
is not enough, so we can resume the use of a STAGE_IN_RESIZE
private flag.
This should also fix bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1668
In order to validate the sequence of:
XResizeWindow
ConfigureNotify
glViewport
that should happen on X11 we need to add debug annotations to the
calls to glViewport() done through COGL.
This avoids some calls to glGetFloatv, which have at least proven to be very
in-efficient in mesa at this point in time, since it always updates all derived
state even when it may not relate to the state being requested.
Fixes and adds a unit test for creating and drawing using materials with
COGL_INVALID_HANDLE texture layers.
This may be valid if for example the user has set a texture combine string
that only references a constant color.
_cogl_material_flush_layers_gl_state will bind the fallback texture for any
COGL_INVALID_HANDLE layer, later though we could explicitly check when the
current blend mode does't actually reference a texture source in which case
binding the fallback texture is redundant.
This tests drawing using cogl_rectangle, cogl_polygon and
cogl_vertex_buffer_draw.
Although we wouldn't recommend developers try and interleve OpenGL drawing
with Cogl drawing - we would prefer patches that improve Cogl to avoid this
if possible - we are providing a simple mechanism that will at least give
developers a fighting chance if they find it necissary.
Note: we aren't helping developers change OpenGL state to modify the
behaviour of Cogl drawing functions - it's unlikley that can ever be
reliably supported - but if they are trying to do something like:
- setup some OpenGL state.
- draw using OpenGL (e.g. glDrawArrays() )
- reset modified OpenGL state.
- continue using Cogl to draw
They should surround their blocks of raw OpenGL with cogl_begin_gl() and
cogl_end_gl():
cogl_begin_gl ();
- setup some OpenGL state.
- draw using OpenGL (e.g. glDrawArrays() )
- reset modified OpenGL state.
cogl_end_gl ();
- continue using Cogl to draw
Again; we aren't supporting code like this:
- setup some OpenGL state.
- use Cogl to draw
- reset modified OpenGL state.
When the internals of Cogl evolves, this is very liable to break.
cogl_begin_gl() will flush all internally batched Cogl primitives, and emit
all internal Cogl state to OpenGL as if it were going to draw something
itself.
The result is that the OpenGL modelview matrix will be setup; the state
corresponding to the current source material will be setup and other world
state such as backface culling, depth and fogging enabledness will be also
be sent to OpenGL.
Note: no special material state is flushed, so if developers want Cogl to setup
a simplified material state it is the their responsibility to set a simple
source material before calling cogl_begin_gl. E.g. by calling
cogl_set_source_color4ub().
Note: It is the developers responsibility to restore any OpenGL state that they
modify to how it was after calling cogl_begin_gl() if they don't do this then
the result of further Cogl calls is undefined.
This function should only need to be called in exceptional circumstances
since Cogl can normally determine internally when a flush is necessary.
As an optimization Cogl drawing functions may batch up primitives
internally, so if you are trying to use raw GL outside of Cogl you stand a
better chance of being successful if you ask Cogl to flush any batched
geometry before making your state changes.
cogl_flush() ensures that the underlying driver is issued all the commands
necessary to draw the batched primitives. It provides no guarantees about
when the driver will complete the rendering.
This provides no guarantees about the GL state upon returning and to avoid
confusing Cogl you should aim to restore any changes you make before
resuming use of Cogl.
If you are making state changes with the intention of affecting Cogl drawing
primitives you are 100% on your own since you stand a good chance of
conflicting with Cogl internals. For example clutter-gst which currently
uses direct GL calls to bind ARBfp programs will very likely break when Cogl
starts to use ARBfb programs internally for the material API, but for now it
can use cogl_flush() to at least ensure that the ARBfp program isn't applied
to additional primitives.
This does not provide a robust generalized solution supporting safe use of
raw GL, its use is very much discouraged.
Previously we would call _cogl_material_pre_change_notify unconditionally, but
now we wait until we really know we are removing a layer before notifying the
change, which will require a journal flush.
Since the convenience functions cogl_set_source_color4ub and
cogl_set_source_texture share a single material, cogl_set_source_color4ub
always calls cogl_material_remove_layer. Often this is a NOP though and
shouldn't require a journal flush.
This gets performance back to where it was before reverting the per-actor
material commits.
This reverts commit 8cf42ea8ac5c05f6b443c453f9c6c2a3cd75acfa.
Since the journal puts material colors in the vertex array accumulated for
drawing we don't need to flush the journal simply due to color changes which
means using cogl_set_source_color4ub is no longer a concern.
This reverts commit 85243da382025bd516937c76a61b8381f6e74689.
Since the journal puts material colors in the vertex array accumulated for
drawing we don't need to flush the journal simply due to color changes
which means using cogl_set_source_color4ub is no longer a concern.
Before any cogl vertex buffer drawing we call
enable_state_for_drawing_buffer which sets up the GL state, but we weren't
disabling unsed client texture coord arrays.
This simplifies the vertex data uploading in the journal, and could improve
performance. Modifying a VBO mid-scene could reqire synchronizing with the
GPU or some form of shadowing/copying to avoid modifying data that the GPU
is currently processing; the buffer was also being marked as GL_STATIC_DRAW
which could have made things worse.
Now we simply create a GL_STATIC_DRAW VBO for each flush and and delete it
when we are finished.
Using cogl_rectangle (and thus the journal) in
_cogl_add_path_to_stencil_buffer means we have to consider all the state
that the journal may change in case it may interfer with the direct GL calls
used. This has proven to be error prone and in this case the journal is an
unnecissary overhead. We now simply call glRectf instead of using
cogl_rectangle.
For small runs of text like icon labels, we can get better performance
going through the Cogl journal since text may then be batched together
with other geometry.
For larger runs of text though we still use VBOs since the cost of logging
the quads becomes too expensive, including the software transform which
isn't at all optimized at this point. VBOs also have the further advantage
of avoiding repeated validation of vertices by the driver and repeated
mapping of data into the GPU so long as the text doesn't change.
Currently the threshold is 100 vertices/25 quads. This number was plucked
out of thin air and should be tuned later.
With this change I see ~180% fps improvment for test-text. (x61s + i965 +
Mesa 7.6-devel)
We were missing the simplest test of all: are the two CoglHandles equal and
are the flush option flags for each material equal? This should improve
batching for some common cases.
Whenever we modify a material we call _cogl_material_pre_change_notify which
checks to see if the material is referenced by the journal and if so flushes
if before we modify the material.
Since the journal logs material colors directly into a vertex array (to
avoid us repeatedly calling glColor) then we know we never need to flush
the journal when material colors change.
Since most Clutter actors aren't much more than textured quads; flushing the
journal typically involves lots of 'change modelview; draw quad' sequences.
The amount of overhead involved in uploading a new modelview and queuing
that primitive is huge in comparison to simply transforming 4 vertices by
the current modelview when logging quads. (Note if your GPU supports HW
vertex transform, then it still does the projective and viewport transforms)
At the same time a --cogl-debug=disable-software-transform option has been
added for comparison and debugging.
This change allows typical pick scenes to be batched into a single draw call
and I'm seeing test-pick run over 200% faster with this. (i965 + Mesa
7.6-devel)
Enabling this option makes Cogl trace how the journal is managing to batch
your rectangles. The journal staggers how it emmits state to the GL driver
and the batches will normally get smaller for each stage, but ideally you
don't want to be in a situation where Cogl is only able to draw one quad per
modelview change and draw call.
E.g. this is a fairly ideal example:
BATCHING: journal len = 101
BATCHING: vbo offset batch len = 101
BATCHING: material batch len = 101
BATCHING: modelview batch len = 101
This isn't:
BATCHING: journal len = 1
BATCHING: vbo offset batch len = 1
BATCHING: material batch len = 1
BATCHING: modelview batch len = 1
BATCHING: journal len = 1
BATCHING: vbo offset batch len = 1
BATCHING: material batch len = 1
BATCHING: modelview batch len = 1
<repeat>
When this option is used Cogl will print a trace of all quads that get
logged into the journal, and a trace of quads as they get flushed.
If you are seeing a bug with the geometry being drawn by Cogl this may give
some clues by letting you sanity check the numbers being logged vs the
numbers being emitted.
For testing the VBO fallback paths it helps to be able to disable the
COGL_FEATURE_VBOS feature flag. When VBOs aren't available Cogl should use
client side malloc()'d buffers instead.
Previously we only used the Cogl matrix stack API for indirect contexts, but
it's too costly to keep on requesting modelview matrices from GL (for
logging in the journal) even for direct rendering.
I also experimented with a patch for mesa to improve performance and
discussed this with upstream, but we agreed to consider the GL matrix API
essentially deprecated. (For reference the GLES 2 and GL 3 specs have
removed the matrix APIs)
CoglColors shouldn't be compared using memcmp since they may contain
uninitialized padding bytes.
The prototype is also suitable for passing to g_hash_table_new as the
key_equal_func.
_cogl_pango_display_list_add_texture now uses this instead of memcmp.
We now put the color of materials into the vertex array used by the journal
instead of calling glColor() but the number of requests for the material
color were quite expensive so we have changed the material color to
internally be byte components instead of floats to avoid repeat conversions
and added _cogl_material_get_colorubv as a fast-path for the journal to
copy data into the vertex array.
To improve batching of geometry in the Cogl journal we need to avoid modifying
materials midscene.
Currently cogl_set_source_color and cogl_set_source_texture simply modify a
single shared material. In the future we can improve this so they use a pool
of materials that gets recycled as the journal is flushed, but for now we
give all ClutterRectangles their own private materials for painting with.
Currently cogl_set_source_color uses a single shared material which means
each actor that uses it causes the journal to flush if the color changes.
Until we improve cogl_set_source_color to use a pool of materials that can
be recycled as the journal is flushed we avoid mid-scene material changes by
giving all actors a private material instead.
The number of material layers enabled when logging a quad in the journal
determines the stride of the corresponding vertex data (since we need a set
of texture coordinates for each layer.) By padding data in the case where we
have only one layer we can avoid a change in stride if we are mixing single
and double layer primitives in a scene (e.g. relevent for a composite
manager that may use 2 layers for all shaped windows) Avoiding stride
changes means we can minimize calls to gl{Vertex,Color}Pointer when flushing
the journal.
Since we need to update the texcoord pointers when the actual number of
layers changes, this adds another batch_and_call() stage to deal with
glTexCoordPointer and enabling/disabling the client arrays.
Previously the journal was always flushed at the end of
_cogl_rectangles_with_multitexture_coords, (i.e. the end of any
cogl_rectangle* calls) but now we have broadened the potential for batching
geometry. In ideal circumstances we will only flush once per scene.
In summary the journal works like this:
When you use any of the cogl_rectangle* APIs then nothing is emitted to the
GPU at this point, we just log one or more quads into the journal. A
journal entry consists of the quad coordinates, an associated material
reference, and a modelview matrix. Ideally the journal only gets flushed
once at the end of a scene, but in fact there are things to consider that
may cause unwanted flushing, including:
- modifying materials mid-scene
This is because each quad in the journal has an associated material
reference (i.e. not copy), so if you try and modify a material that is
already referenced in the journal we force a flush first)
NOTE: For now this means you should avoid using cogl_set_source_color()
since that currently uses a single shared material. Later we
should change it to use a pool of materials that is recycled
when the journal is flushed.
- modifying any state that isn't currently logged, such as depth, fog and
backface culling enables.
The first thing that happens when flushing, is to upload all the vertex data
associated with the journal into a single VBO.
We then go through a process of splitting up the journal into batches that
have compatible state so they can be emitted to the GPU together. This is
currently broken up into 3 levels so we can stagger the state changes:
1) we break the journal up according to changes in the number of material layers
associated with logged quads. The number of layers in a material determines
the stride of the associated vertices, so we have to update our vertex
array offsets at this level. (i.e. calling gl{Vertex,Color},Pointer etc)
2) we further split batches up according to material compatability. (e.g.
materials with different textures) We flush material state at this level.
3) Finally we split batches up according to modelview changes. At this level
we update the modelview matrix and actually emit the actual draw command.
This commit is largely about putting the initial design in-place; this will be
followed by other changes that take advantage of the extended batching.
Removed the G_PARAM_CONSTRUCT from the property registration of
"load-async" and "load-data-async". It made it impossible to use only
load-data-async, as the async loading state would be unset when
load-async got set it's default FALSE value.
Use signed integers while combining window space clip rectangles, so we avoid
arithmatic errors later resulting in glScissor getting negative width and
height arguments.
Previously this was RGBA_8888. It souldn't really make a difference but for
consistency we expect almost all textures in use to have an internaly
premultiplied pixel format.
_cogl_texture_download_from_gl needs to create transient CoglBitmaps when
downloading sliced textures from GL, and then copies these as subregions
into the final target_bitmap. _cogl_texture_download_from_gl also supports
target_bitmaps with a different format to the source CoglTexture being
downloaded.
The problem was that in the case of slice textures we were always looking
at the format of the CoglTexture, not of the target_bitmap when setting
up the transient slice bitmap.
To allow for flushing of batched geometry within Cogl we can't support users
directly calling glReadPixels. glReadPixels is also awkward, not least
because it returns upside down image data.
All the unit tests have been swithed over and clutter_stage_read_pixels now
sits on top of this too.
I've found this is something I do quite often when debugging rendering
problems since its a simple way to wipe out lots of geometry and removes a
lot of unpredictable noise when logging geometry passing through the Cogl
journal.
We were calculating our vertex stride and allocating our vertex array
differently depending on whether the user passed TRUE for use_color or not.
The problem was that we were always writting color data to the array
regardless of use_color.
There was also a bug with _cogl_texture_sliced_polygon in that it was
writing byte color components but we were expecting float components. We
now use byte components in _cogl_multitexture_unsliced_polygon too and pass
GL_UNSIGNED_BYTE to glColorPointer.
Cogl already add similar defines but with the CLUTTER namespace
(CLUTTER_COGL_HAS_GL and CLUTTER_COGL_HAS_GLES). Let's just add two
similar defines with the COGL namespace. Removing the CLUTTER_COGL ones
could break applications silently for no real good reason.
While grepping through the public headers looking for invalid use of
private HAVE_* defines, I stumbled upon two out of sync comments. Yes
it's a very minor trivial change.
JSON-GLib provides simple accessors for basic types so that we
can avoid getting the JsonNode out of a complex type. This makes
the code simpler to understand.
Currently, Clutter depends on the internal copy of JSON-GLib for
the ClutterScript parser. This is done to allow building Clutter
on platforms that do not have the library installed on the system.
Just like we use the internal PNG/JPEG loader as a fallback in
case we don't have GdkPixbuf or CoreGraphics available, we should
use the internal copy of JSON-GLib only in case the system copy
is not present.
The change is simply to move the default for the --with-json
configure switch from "internal" to "check".
In order to allow stricter compliance, a third setting should
be present: "system", which fails if the system copy is not
available.
We should also change the introspection generation to avoid
breaking in case we require the installed Json-1.0.gir instead
of the generated ClutterJson.gir
The clutter_actor_pick() function just emits the ::pick signal
on the actor. Nobody should be using it, since the paint() method
is already context sensitive and will result in a ::pick emission
by itself. The clutter_actor_pick() is just confusing things.
In the int-to-float switch for actor properties, the ::size-change signal
was moved to use floats instead of integers. Sub-pixel precision for image
size is meaningless, though, so we should revert it back to ints.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1659
Instead of using a specific function to check whether the X
server supports the XInput extension we can use the generic
Xlib function XQueryExtension(). This cuts down the extra
checks inside the configure.ac and simplifies the code inside
clutter_x11_register_xinput().
Currently, XInput support requires a function call. In order to
make it easier for people to test it, we can also add a command
line switch that moves the pointer device detection and handling
to XInput. This should ensure that, at least for people building
Clutter with --enable-xinput, applications can be easily migrated
and regressions can be caught.
The StageManager singleton instance is already kept around
by the clutter_stage_manager_get_default() function; there is
no need to have it inside the main Clutter context as well.
Instead of using _clutter_context_get_default() and checking the
is_initialized flag, we should use the newly added private function
that does not cause side effects, especially for functions that have
to be called before any other Clutter function.
The _clutter_context_get_default() function will automatically
create the main Clutter context; if we just want to check whether
Clutter has been initialized this will complicate matters, by
requiring a call to g_type_init() inside the client code.
Instead, we should simply provide an internal API that checks
whether the main Clutter context exists and if it has been
initialized, without any side effect.
The input device API is split halfway thorugh the backends in a very
weird way. The data structures are private, as they should, but most
of the information should be available in the main API since it's
generic enough.
The device type enumeration, for instance, should be common across
every backend; the accessors for device type and id should live in the
core API. The internal API should always use ClutterInputDevice and
not the private X11 implementation when dealing with public structures
like ClutterEvent.
By adding accessors for the device type and id, and by moving the
device type enumeration into the core API we can cut down the amount
of symbols private and/or visible only to the X11 backends; this way
when other backends start implementing multi-pointer support we can
share the same API across the code.
HAVE_COGL_GLES2 is defined in config.h through the configure script and
should not be used in public headers.
The patch makes configure generate the right define that can be used
later in the header.
The clutter_context_get_default() function is private, but shared
across Clutter. For this reason, it should be prefixed by '_' so
that the symbol is hidden from the shared object.
ActorBox should have methods for easily extracting the X and Y
coordinates of the origin, and the width and height separately.
These methods will make it easier for high-level language bindings
to manipulate ActorBox instances and avoid the Geometry type.
The Vertex and ActorBox boxed types are meant to be used across
the API, but are fairly difficult to bind. Their memory management
is also unclear, and has to go through the indirection of
g_boxed_copy() and g_boxed_free().
Cairo stores the image data in ARGB native byte order so we need to
upload this as BGRA on little endian architectures and ARGB on big
endian. ClutterTexture doesn't currently expose any flags to describe
ARGB format so until we can fix the Clutter API it now uses the Cogl
API directly.
In order to chain up animations using clutter_actor_animate() and
friends you have to use an idle handler that guarantees that the
main loop spins at least once after the animation pointer has been
detached from the actor.
This has several drawbacks, first and foremost the fact that the
slice of the main loop for the idle handler might be starved by
other operations, like redrawing. This inevitably leads to tricks
with priorities and the like, contributing to the overall complexity.
Instead, we should guarantee that the animation instance created by
clutter_actor_animate() is valid for the ::completed signal until
it reaches its default handler; after that, the animation is detached
from the actor and destroyed. This means that it's possible to
create a new animation after the first is complete by simply using
g_signal_connect_after().
This unfortunately makes it impossible to keep a reference to the
animation pointer attached to the actor by using g_object_ref(); a
way to "fix" this would be to have a clutter_animation_attach()
and a clutter_animation_detach() pair of methods that allow attaching
any animation to an actor. This might overcomplicate what it is
the simple animation API, though, so it's currently not implemented
and left for future versions.
The test-easing interactive demo has been modified to show how
the animation queuing works by adding a command line switch that
recenters the animated actor once the first animation has ended.
Continuing in the tradition on making clutter_actor_animate() the
next g_object_connect(), here's the addition of the signal-after::
and signal-swapped:: modifiers for the automagic signal connection
arguments.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1646
In order to be ready for the next major version of GLib we need to
disable single header inclusion by using the G_DISABLE_SINGLE_INCLUDES
define in the build process.
We can't short-circuit the emission of ::queue-redraw for not-visible
actors, since ClutterClone uses that signal to know when things need
to be redrawn.
Calling clutter_actor_queue_redraw() out of clutter_actor_real_map() /
clutter_actor_real_unmap() was causing the flag state to get set
incorrectly from _clutter_actor_set_enable_paint_unmapped(), because
a paint queueing a redraw was not expected.
Moving queuing the redraw to clutter_actor_hide()/show() fixes this, and
also fixes a problem where showing a child of a cloned actor wouldn't
cause the clone to be repainted.
http://bugzilla.openedhand.com/show_bug.cgi?id=1484
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
If we have an not-visible texture pixmap, we need to:
- Still update it if it is realized, since it won't be
updated when shown. And it might be also be cloned.
- Queue a redraw if even if not visible, since it
it might be cloned.
http://bugzilla.openedhand.com/show_bug.cgi?id=1647
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
My patch to choose a premultiplied format when the user gives
COGL_PIXEL_FORMAT_ANY for the internal_format broke the case where the data
in question doesn't have and alpha channel.
This was accidentally missed when merging the premultiplication branch
since I merged a local version of the branch that missed this commit.
Although the underlying materials should allow layers with INVALID_HANDLES
it shouldn't be necissary to expose that via cogl_set_source_texture() and
it's easier to resolve a warning/crash here than odd artefacts/crashes later
in the pipeline.
Merge branch 'premultiplication'
[cogl-texture docs] Improves the documentation of the internal_format args
[test-premult] Adds a unit test for texture upload premultiplication semantics
[fog] Document that fogging only works with opaque or unmultipled colors
[test-blend-strings] Explicitly request RGBA_888 tex format for test textures
[premultiplication] Be more conservative with what data gets premultiplied
[bitmap] Fixes _cogl_bitmap_fallback_unpremult
[cogl-bitmap] Fix minor copy and paste error in _cogl_bitmap_fallback_premult
Avoid unnecesary unpremultiplication when saving to local data
Don't unpremultiply Cairo data
Default to a blend function that expects premultiplied colors
Implement premultiplication for CoglBitmap
Use correct texture format for pixmap textures and FBO's
Add cogl_color_premultiply()
Clarifies that if you give COGL_PIXEL_FORMAT_ANY as the internal format for
cogl_texture_new_from_file or cogl_texture_new_from_data then Cogl will
choose a premultiplied internal format.
The fixed function fogging provided by OpenGL only works with unmultiplied
colors (or if the color has an alpha of 1.0) so since we now premultiply
textures and colors by default a note to this affect has been added to
clutter_stage_set_fog and cogl_set_fog.
test-depth.c no longer uses clutter_stage_set_fog for this reason.
In the future when we can depend on fragment shaders we should also be
able to support fogging of premultiplied primitives.
We don't want to force texture data to be premultipled if the user
explicitly specifies a non premultiplied internal_format such as
COGL_PIXEL_FORMAT_RGBA_8888. So now Cogl will only automatically
premultiply data when COGL_PIXEL_FORMAT_ANY is given for the
internal_format, or a premultiplied internal format such as
COGL_PIXEL_FORMAT_RGBA_8888_PRE is requested but non-premultiplied source
data is given.
This approach is consistent with OpenVG image formats which have already
influenced Cogl's pixel format semantics.
The _cogl_unpremult_alpha_{first,last} functions which
_cogl_bitmap_fallback_unpremult depends on were incorrectly casting each
of the byte components of a texel to a gulong and performing shifts as
if it were dealing with the whole texel.
It now just uses array indexing to access the byte components without
needing to cast or manually shift any bits around.
Even though we used to depend on unpremult whenever we used a
ClutterCairoTexture, clutter_cairo_texture_context_destroy had it's own
unpremult code which worked which is why this bug wouldn't have been noticed
before.
Now that we typically have premultiplied data stored in Cogl
textures, when fetching a texture into local data for temporary
storage, use a premultiplied format to avoid an unpremultiply/
premultiply roundtrip.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Instead of unpremultiplying the Cairo data, pass it directly to Cogl
in premultiplied form; we now *prefer* premultiplied data.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Many operations, like mixing two textures together or alpha-blending
onto a destination with alpha, are done most logically if texture data
is in premultiplied form. We also have many sources of premultiplied
texture data, like X pixmaps, FBOs, cairo surfaces. Rather than trying
to work with two different types of texture data, simplify things by
always premultiplying texture data before uploading to GL.
Because the default blend function is changed to accommodate this,
uses of pure-color CoglMaterial need to be adapted to add
premultiplication.
gl/cogl-texture.c gles/cogl-texture.c: Always premultiply
non-premultiplied texture data before uploading to GL.
cogl-material.c cogl-material.h: Switch the default blend functions
to ONE, ONE_MINUS_SRC_ALPHA so they work correctly with premultiplied
data.
cogl.c: Make cogl_set_source_color() premultiply the color.
cogl.h.in color-material.h: Add some documentation about
premultiplication and its interaction with color values.
cogl-pango-render.c clutter-texture.c tests/interactive/test-cogl-offscreen.c:
Use premultiplied colors.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
cogl-bitmap.c cogl-bitmap-pixbuf.c cogl-bitmap-fallback.c cogl-bitmap-private.h:
Add _cogl_bitmap_can_premult(), _cogl_bitmap_premult() and implement
a reasonably fast implementation in the "fallback" code.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
RGBA data in X pixmaps and in FBOs is already premultiplied; use
the right format when creating cogl textures.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Since commit d743aeaa updated the internal copy of JSON-GLib and
added a new private header file, we need to fix the build to avoid
a distcheck failure.
Merge branch 'master-clock-updates'
* master-clock-updates: (22 commits)
Change the paint forcing on the Text cache text
[timelines] Improve marker hit check and don't fudge the delta
Revert "[timeline] Don't clamp the elapsed time when a looping tl reaches the end"
[tests] Don't add a newline to the end of g_test_message calls
[test-timeline] Add a marker at the beginning of the timeline
[timeline] Don't clamp the elapsed time when a looping tl reaches the end
[master-clock] Throttle if no redraw was performed
[docs] Update Clutter's API reference
Force a paint instead of calling clutter_redraw()
Fix clutter_redraw() to match the redraw cycle
Run the repaint functions inside the redraw cycle
Remove useless manual timeline ticking
Move elapsed-time calculations into ClutterTimeline
Limit the frame rate when not syncing to VBLANK
Decrease the main-loop priority of the frame cycle
Avoid motion-compression in test-picking test
Compress events as part of the frame cycle
Remove stage update idle and do updates from the master clock
Call g_main_context_wakeup() when we start running timelines
Remove unused msecs_delta member
...
Markers added at the start of the timeline need to be special cased
because we effectively check whether the marker time is greater than
the old frame time or less than or equal to the new frame time but we
never actually emit a frame for the start of the timeline.
If the timeline is looping then it adjusts the position to interpolate
a wrapped around position. However we do not emit a new frame after
setting this position so we need to check for markers again there.
clutter_timeline_get_delta should return the actual wall clock time
between emissions of the new-frame signal - even if the elapsed time
has been fudged at the end of the timeline. To make this work it no
longer directly manipulates priv->msecs_delta but instead passes a
delta value to check_markers.
Someone might hide the previously focused actor in the focus-out signal
handler and as key focus still appears to be on that actor we'd get
re-entrant call and get glib critical from g_object_weak_unref
This changes clutter_stage_get_key_focus() to return stage/NULL during
focus-out signal emission. It used to be the actor focus-out was being
emitted on.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1547
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
This reverts commit 9c5663d671.
The patch was causing problems for applications that expect the
elapsed_time to be at either end of the timeline when the completed
signal is fired. For example test-behave swaps the direction of the
timeline in the completed handler but if the time has overflowed the
end then the timeline would only take a short time to get back the
beginning. This caused the animation to just vibrate around the
beginning.
The new-frame signal of a timeline was previously guaranteed to be
emitted with the elapsed_time set to the end before it emits the
completed signal. This doesn't necessarily make sense for looping
timelines because it would cause the elapsed time to be clamped to a
slightly off value whenever the timeline restarts. This patch makes it
perform the wrap around before emitting the new-frame signal so that
the elapsed time always corresponds to the time elapsed since the
timeline was started.
Additionally it no longer fudges the msecs_delta property to make the
marker check work so clutter_timeline_get_delta will always return the
wall clock time since the last frame.
A flag in the master clock is now set whenever the dispatch caused an
actual redraw of a stage. If this flag is not set during the prepare
and check functions then it will resort to limiting the redraw
attempts to the default frame rate as if vblank syncing was
disabled. Otherwise if a timeline is running that does not cause the
scene to change then it would busy-wait with 100% CPU until the next
frame.
This fix was suggested by Owen Taylor in:
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
We do not need the whole redraw machinery inside
clutter_stage_read_pixels(): instead, we want Clutter to drop everything,
paint and call glReadPixels() with the current buffer.
The clutter_redraw() function is used by embedding toolkits to
force a redraw on a stage. Since everything is performed by
toggling a flag inside the Stage itself and then letting the
master clock advance, we need a ClutterStage method to ensure
that we start the master clock and redraw.
Instead of calculating a delta in the master clock, and passing that
into each timeline, make each timeline individually responsible for
remembering the last time and computing the delta.
This:
- Fixes a problem where we could spin infinitely processing
timeline-only frames with < 1msec differences.
- Makes timelines consistently start timing on the first frame;
instead of doing different things for the first started timeline
and other timelines.
- Improves accuracy of elapsed time computations by avoiding
accumulating microsecond => millisecond truncation errors.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
clutter-master-clock.c clutter-master-clock.h: When the
SYNC_TO_VBLANK feature is not available, wait for 1/frame_rate
seconds since the start of the last frame before drawing the next
frame. Add _clutter_master_clock_start_running() to abstract
the usage of g_main_context_wakeup()
clutter-stage.c: Add _clutter_master_clock_start_running()
clutter-main.c: Update docs for clutter_set_default_frame_rate()
clutter_get_default_frame_rate() to no longer talk about timeline
frame rates.
test-text-perf.c test-text.c: Set a frame rate of 1000fps so that
frame-rate limiting doesn't affect the result.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Change CLUTTER_PRIORITY_REDRAW to be lower than the GTK+ resize
and relayout priorities to avoid starving GTK+ when run in the
same process as clutter.
Remove the unused CLUTTER_PRIORITY_TIMELINE
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Instead of trying to guess about which motion events are
extraneous, queue up all events until we process a frame.
This allows us to look ahead and reliably compress consecutive
sequence of motion events.
clutter-main.c: Feed received events to the stage for queueing.
Remove old compression code. Remove clutter_get_motion_events_frequency()
clutter_set_motion_events_frequency()
clutter-stage.c: Keep a queue of pending events.
clutter-master-clock.c: Add processng of queued events to the
clock source dispatch function.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
When a redraw is queued on a stage, simply set a flag; then in
the check/prepare functions of the master clock source, check
for stages that need redrawing.
This avoids the complexity of having multiple competing sources
at the same priority and makes the update ordering more reliable and
understandable.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
If a timeline is added from a different thread, we need to
call g_main_context_wakeup() to wake the main thread up to
start updating the timeline.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Instead of keeping a list of all timelines, and connecting to
signals and weak notifies, simply keep a list of running timelines;
this greatly simplifies both the book-keeping, and also determining
if there are any running timelines.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Remove code to advance the master clock after drawing a frame; if
there are any running timelines the master clock will do another
frame by itself, and the clock will be advanced before running
that frame.
With this change, there is no point in queueing an extra frame
redraw after completing a timeline, since we are always advancing
the timeline *before* redrawing, so remove that code as well.
(This does mean that calling clutter_timeline_stop() won't implicitly
cause the stage to be redrawn; this doesn't seem like something
an app should rely on in any case.)
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The clutter_stage_fullscreen() and clutter_stage_unfullscreen() are
a GDK-ism. The underlying implementation is already using an accessor
with a boolean parameter.
This should take the amount of collisions between properties, methods
and signals to zero.
The :fullscreen property is very much confusing as it is implemented.
It can be written to a value, but the whole process might fail. If we
set:
g_object_set (stage, "fullscreen", TRUE, NULL);
and the fullscreen process fails or it is not implemented, the value
will be reset to FALSE (if we're lucky) or left TRUE (most of the
times).
The writability is just a shorthand for invoking clutter_stage_fullscreen()
or clutter_stage_unfullscreen() depending on a boolean value without
using an if.
The :fullscreen property also greatly confuses high level languages,
since the same symbol is used:
- for a method name (Clutter.Stage.fullscreen())
- for a property name (Clutter.Stage.fullscreen)
- for a signal (Clutter.Stage::fullscreen)
For these reasons, the :fullscreen should be renamed to :fullscreen-set
and be read-only. Implementations of the Stage should only emit the
StageState event to change from normal to fullscreen, and the Stage
will automatically update the value of the property and emit a notify
signal for it.
There have been changes in JSON-GLib upstream to clean up the
data structures, and facilitate introspection.
We still not use the updated JsonParser with the (private) JsonScanner
code, since it's a fork of GLib's GScanner.
Otherwise if there is an error before the slices are created it will
try to free the first_pixels array and crash.
It now also checks whether first_pixels has been created before using
it to update the mipmaps. This should only happen for
cogl_texture_new_from_foreign and doesn't matter if the FBO extension
is available. It would be better in this case to fetch the first pixel
using glGetTexImage as Owen mentioned in the last commit.
tex->first_pixels was never set for foreign textures, leading
to a crash when the texture object is freed.
As a quick fix, simply set to NULL. A more complete fix would
require remembering if we had ever seen the first pixel uploaded,
and if not, doing a glReadPixel to get it before triggering the
mipmap update.
http://bugzilla.openedhand.com/show_bug.cgi?id=1645
Signed-off-by: Neil Roberts <neil@linux.intel.com>
It's very common that there's no reasonable fallback to do if the
blend or combine string you set isn't supported. So, rather than
requiring everybody to pass in a GError purely to catch syntax erorrs,
automatically g_warning() if a parse error is encountered and @error
is NULL.
http://bugzilla.openedhand.com/show_bug.cgi?id=1642
Signed-off-by: Robert Bragg <robert@linux.intel.com>
When we complete a timeline, we clamp the elapsed_time variable
to the range of the timeline. We need to adjust msecs_delta so that
when we check for hit markers we have the correct interval.
http://bugzilla.openedhand.com/show_bug.cgi?id=1641
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The Animation should be referenced during the notification of the
alpha value, since the callback is invoked depending on the Alpha
and it won't vivify the Animation instance for us.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1537
ClutterEvent is not really gobject-introspection friendly because
of the whole discriminated union thing. In particular, if you get
a ClutterEvent in a signal handler, you probably can't access the
event-type-specific fields, and you probably can't call methods
like clutter_key_event_symbol() either, because you can't cast the
ClutterEvent to a ClutterKeyEvent.
The cleanest solution is to turn every accessor into ClutterEvent
methods, accepting a ClutterEvent* and internally checking the event
type.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1585
According to clutter_texture_set_cogl_texture you should unref the handle as
the texture takes its own.
Signed-off-by: Robert Bragg <robert@linux.intel.com>
The OpenGL spec states that if you create a pixmap using glXCreatePixmap you
should use glXDestroyPixmap to destroy it.
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Setting the pixmap for an unrealized ClutterGLXTexturePixmap should
not cause it to be realized, and certainly shouldn't cause the the
REALIZED flag to be set without using clutter_actor_realize().
This patch uses the simple approach that;
- pixmap changes on an unrealized ClutterGLXTexturePixmap
are ignored
- when the ClutterGLXTexturePixmap is realized, we then create
the GLXPixmap and the corresponding texture.
The call to clutter_glx_texture_pixmap_update_area() is moved
from create_cogl_texture() to
clutter_glx_texture_pixmap_create_glx_pixmap() since
create_cogl_texture() is only called from one place, and updating
the area is really something we do *after* creating the texture,
not part of creating the texture.
clutter_glx_texture_pixmap_create_glx_pixmap() is reorganized a
bit to avoid debug-logging confusingly if it's called before a pixmap
has been set, and for readability.
http://bugzilla.openedhand.com/show_bug.cgi?id=1635
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
An implementaton of realize() never needs to set the
CLUTTER_ACTOR_REALIZED flag, though it can unset the flag if
things fail unexpectedly. (Previously, stage backend implementations
had to do this since clutter_actor_realize() wasn't used; this
is no longer the case.)
http://bugzilla.openedhand.com/show_bug.cgi?id=1634
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Due to the accumulation of floating point errors, natural_width
and min_width can diverge significantly even if the math for
computing them is correct. So just clamp natural_width to
min_width instead of warning about it.
http://bugzilla.openedhand.com/show_bug.cgi?id=1632
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
If we use float temporaries when computing the bounds of
a group, then, depending on what variables are kept in registers
and what stored on the stack, the accumulated difference between
natural_width and min_width can be more than FLOAT_EPSILON.
Using double temporaries will eliminate the difference in most
cases, or, very rarely, reduce it to a last-bit error.
http://bugzilla.openedhand.com/show_bug.cgi?id=1632
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
If we are cloning an source actor with an unmapped parent, then when
we temporarily map the source actor:
- We need to skip the check that a mapped actor has a mapped
parent.
- We need to realize the actor's parents before mapping it,
or we'll get an assertion failure in clutter_actor_update_map_state()
because an actor with an unmapped parent is !may_be_realized.
http://bugzilla.openedhand.com/show_bug.cgi?id=1633
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Setting the stage size using clutter_actor_set_size() is almost always
wrong: the X11 stage implementation should save the size and queue a
relayout -- like it does when receiving a ConfigureNotify. The same
should happen when setting it to be full screen.
Since we build the Cogl GIR inside /clutter/cogl we should be looking
there when building the Clutter GIR. Otherwise g-ir-scanner will look
inside the gir directory -- and if you never built Clutter before it
will error out.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1638
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The load-finished signal has a GError* argument which is meant to
signify whether the loading was successful. However many of the
places in ClutterTexture that emit this signal directly pass their
'error' variable which is a GError** and will be NULL or not
completely independently of whether there was an error. If the
argument was dereferenced it would probably crash.
The test-texture-async interactive test case should also verify
that the ::load-finished signal is correctly emitted.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1622
