We were checking the number of texture units against the GL enum that is
used in glGetInteger() to query that number. Let's abstract this in a
little function.
Took the opportunity to dig a bit on the usage of GL limits for the
number of texture (image) units and document our use of them. We'll need
something finer grained if we want to fully exploit texture image units
with a programmable pipeline.
The index field of CoglTextureUnit was never set, leading to the
creation of units with index set to 0. When trying to retrieve a texture
unit by its index (!= 0) with _cogl_get_texture_unit(), a new one was
created as it could not find it back in the list of textures units:
ctx->texture_units.
http://bugzilla.openedhand.com/show_bug.cgi?id=1958
The :opacity property is defined using a GParamSpecUchar. This usually
leads to issues with language bindings that don't have an 'unsigned
char' type and that need to explicitly handle the conversion between
G_TYPE_UCHAR and G_TYPE_INT or G_TYPE_UINT.
The property definition already specifies an interval size of [0, 255]
on the values; more importantly, GObject already implicitly transforms
between G_TYPE_UCHAR and G_TYPE_UINT (the GValue transformation
functions are registered at type system initialization time) so
switching between a GParamSpecUchar and a GParamSpecUint should not be
an ABI break.
I have tested a simple program using the opacity property before and
after the change and I cannot see any run-time warnings related to this
issue.
Be more drastic if the internal state is broken, and assert() if the
expected Alpha and Timeline instances we need are not valid. This
usually implies a library bug or a massive heap corruption.
The Animation code does transformation of values between type A and A'
after checking for compatibility using g_value_type_compatible(). This
is incorrect: compatibility means that the two types can be copied. The
correct conversion should follow:
if (compatible (type (A), type (A')))
copy (A, A');
else
if (transformable (type (A), type (A')))
transform (A, A');
else
error("Unable to trasform type A in A'");
The transformation might still fail, so we need to check for errors
there as well as a fall-through case.
We should not just check for compatibility, but also for the ability to
transform a GValue of type A into another of type A'.
Usually compatibility is enough, especially if types can be
introspected beforehand; some times, though, we also need to check for
transformability as a type can provide the transformation functions
necessary for the operation.
The commit 1c69c61745 which improved the
protection against timeline removals during the master clock advancement
was only doing half the job - and actually broke the chaining of
animations inside the ::completed signal.
We cannot simply take a reference on the timelines and still use the list
held by the master clock because the do_tick() might result in the
creation of a new timeline, which gets added at the end of the list with
no reference increase and thus gets disposed at the end of the iteration.
We also cannot steal the master clock timelines list because a timeline
might be removed as the direct result of do_tick() and remove_timeline()
would not find the timeline, failing and leaving a dangling pointer
behind.
For this reason we copy the list of timelines out of the one that the
Master Clock holds, take a reference on each timeline, advance them all,
release the reference and free the list.
The extension keyboard support in XInput 1.x is hopelessly broken.
Nevertheless, it's possible to use some bits of it, as we prefer the
core keyboard events to the XInput events, thus at least having proper
handling for X11 key events on the Stage window.
The XI 1.0 layer is complementary to the X11 core devices handling; this
means that core events will still be emitted for the core pointer and
keyboard devices, and that secondary (floating) devices should be
handled on top of that.
Thus, the XI event handling code should be executed (if explicitly
compiled in and enabled) if the core device events have not been parsed.
Note: this is going away with XI2, which completely replaces both core and
XI1 events.
Even with XInput support we should always register core devices. This
allows us to handle enter and leave events correctly on the Stage and
to have a working XInput 1.x support in Clutter.
Mostly lifted from the core pointer and keyboard X11 backend support.
The win32 backend registers two devices (a core pointer and a core
keyboard) and assigns them to the event structure when doing the
translation from native events to Clutter events.
Thanks to: Samuel Degrande <Samuel.Degrande@lifl.fr> for testing this
patch.
Instead of overloading the device id of 0 and 1 we should treat the core
devices as special, and have a pointer inside the X11 backend singleton
structure, for fast access.
When an InputDevice leaves a stage we set the stage member of
InputDevice to NULL. We should also unset the cursor_actor (as the
device is obviously not on an actor any more).
When the device re-enters the Stage the ENTER/LEAVE event generation
machinery will then be able to emit the ENTER event on the Stage.
If the user presses a button on a pointer device and then moves out the
Stage X11 will emit the following events:
LeaveNotify ➔ MotionNotify ... ➔ ButtonRelease ➔ LeaveNotify
The second LeaveNotify differs from the first by the state field.
Unfortunately, ClutterCrossingEvent doesn't have a modifier_state field
like other events, so we cannot provide a way for programmatically
distinguishing them from a Clutter perspective. This is also an X11-ism
we might not even want to replicate on every backend with sane
enter/leave semantics.
For this reason we should check inside the X11 event processing if the
pointer device has already left the Stage and ignore the second
LeaveNotify.
The Stage field of an InputDevice is set by the backend, whenever the
pointer enters or leaves the Stage. The Stage should not overwrite the
stage field for every event it processes.
The previous state for the device is used by the click count machinery
and we should not be overwriting it at every event; instead, we should
use a parallel storage for the current state coming from the windowing
system.
The device manager does not need to update the state of the devices
when the user has disabled the delivery of motion events to actors:
the events will always be delivered as they are to the stage.
The LEAVE/ENTER event pairs should be queued during the InputDevice
update process, when we change the actor under the device pointer.
This commit cleans up the event emission code inside clutter-main.c
and the logic of the event processing.
The InputDevice objects stores pointer coordinates, state, stage and
the actor under the cursor, so if the current backend provides us with
one attached to the Event structure then we want the InputDevice itself
to update its state and give us the ClutterActor underneath the
pointer's cursor.
Even when we are not using XInput we now have fallback devices; the
X11 backend should always assign the default devices when translating
the X events to Clutter events.
Use the device manager to store the input devices. Also, provide
two fallback devices when initializing the X11 backend: device 0
for the pointer and device 1 for the keyboard.
Previously the atlas textures were being created with whatever format
the first sub texture is in. Only three formats are supported so this
only matters if the first texture is a premultiplied alpha
texture. Instead it now masks out the premultiplied bit so that the
textures are always either RGB_888 or RGBA_8888.
The win32 backend now handles the WM_SETCURSOR message and sets a
fully transparent cursor if the cursor-visible property has been
cleared on the stage. The icon is stored in the library via a resource
file. The instance handle for the DLL is needed to load the resource
so there is now a DllMain function to grab the handle.
g_list_foreach has better protection against the current node being
removed. This will happen for example if someone calls
clutter_container_foreach(container, clutter_actor_destroy). This was
causing valgrind errors for the conformance tests which do just that.
When uploading texture data it was just calling cogl_texture_set_data
on the large texture. This would attempt to convert the data to the
format of the large texture. All of the textures with alpha channels
are stored together regardless of whether they are premultiplied so
this was causing premultiplied textures to be unpremultiplied
again. It now just uploads the data ignoring the premult bit of the
format so that it only gets converted once.
With the atlas texture backend ensuring the mipmaps can make it become
a completely different texture which will have different texture
coordinates or may even be sliced. Therefore we need to ensure the
mipmaps before deciding which quads to log in the journal. This adds a
new private function to cogl-material which ensures the mipmaps if
needed.
The sub texture backend doesn't work well as a completely general
texture backend because for example when rendering with cogl_polygon
it needs to be able to tranform arbitrary texture coordinates without
reference to the other coordintes. This can't be done when the texture
coordinates are a multiple of one because sometimes the coordinate
should represent the left or top edge and sometimes it should
represent the bottom or top edge. For example if the s coordinates are
0 and 1 then 1 represents the right edge but if they are 1 and 2 then
1 represents the left edge.
Instead the sub-textures are now documented not to support coordinates
outside the range [0,1]. The coordinates for the sub-region are now
represented as integers as this helps avoid rounding issues. The
region can no longer be a super-region of the texture as this
simplifies the code quite a lot.
There are two new texture virtual functions:
transform_quad_coords_to_gl - This transforms two pairs of coordinates
representing a quad. It will return FALSE if the coordinates can
not be transformed. The sub texture backend uses this to detect
coordinates that require repeating which causes cogl-primitives
to use manual repeating.
ensure_non_quad_rendering - This is used in cogl_polygon and
cogl_vertex_buffer to inform the texture backend that
transform_quad_to_gl is going to be used. The atlas backend
migrates the texture out of the atlas when it hits this.
When calculating the next integer position for negative coordinates it
would not increment if the position is already a multiple of one so we
need to manually add one.
When try_creating_fbo fails it returns 0 to report the error and if it
succeeds it returns ‘flags’. However cogl_offscreen_new_to_texture
also passes in 0 for the flags as the last fallback to create the fbo
with nothing but the color buffer. In that case it will return 0
regardless of whether it succeeded so the last fallback will always be
considered a failure.
To fix this it now just returns a gboolean to indicate whether it
succeeded and the flags used for each attempt is assigned when passing
the argument rather than from the return value of the function.
Also if the only configuration that succeeded was with flags==0 then
it would always try all combinations because last_working_flags would
also be zero. To avoid this it now uses a separate gboolean to mark
whether we found a successful set of flags.
http://bugzilla.openedhand.com/show_bug.cgi?id=1873
Use the newly-added profiling timers inside the master clock dispatch
function to see how much time we spend:
• in the whole function
• in the event processing for each stage
• in the timeline advancement
Reading back the texture data in unrealize does not seem like a
desirable feature any more, clutter has evolved a lot since it was
implemented.
What's wrong with it now:
* It takes *a lot* of time to read the data back with glReadPixel(),
* When several textures share the same CoglTexture, the same data can
be read back multiple times,
* If the underlying material uses multiple texture units, only the
first one was copied back,
* In ClutterCairoTexture, we end up having two separate copies of the
data,
* GL actually manages texture memory accross system/video memory
for us!
For all the reasons above, let's get rid of the glReadPixel() in
Texture::unrealize()
Fixes: OHB#1842
Since 755cce33a7 the framebuffer code is using the GL enums
GL_DEPTH_ATTACHMENT and GL_DEPTH_COMPONENT16. These aren't available
directly under GLES except with the OES suffix so we need to define
them manually as we do with the other framebuffer constants.
These macros used to define Cogl wrappers for the GLenum values. There are
now Cogl enums everywhere in the API where these were required so we
shouldn't need them anymore. They were in the public headers but as
they are not neccessary and were not in the API docs for Clutter 1.0
it should be safe to remove them.
Using the ::event signal to match the CLUTTER_DELETE event type (and
block the stage destruction) can be costly, since it means checking
every single event.
The ::delete-event signal is similar in spirit to any other specialized
signal handler dealing with events, and retains the same semantics.
If a user supplied multiple groups of texture coordinates with
cogl_rectangle_with_multitexture_coords() then we would repeatedly log only
the first group in the journal. This fixes that bug and adds a conformance
test to verify the fix.
Thanks to Gord Allott for reporting this bug.
The Intel drivers in Mesa 7.6 (and possibly earlier versions) don't
support creating FBOs with a stencil buffer but without a depth
buffer. This reworks framebuffer allocation so that we try a number
of fallback options before failing.
The options we try in order are:
- the same options that were sucessful last time if available
- combined depth and stencil
- separate depth and stencil
- just stencil, no depth
- just depth, no stencil
- neither depth or stencil
Allow the user of the ClutterMedia interface to specify a Pango font
description to display subtitles. Even if the underlying implementation
of the interface does not natively use Pange, it must be capable of
parsing the grammar that pango_font_description_from_string() accepts.
Some source files should not be passed through the introspection parser,
as they are fully private and do not expose any valuable API.
Also the clutter-profile.h header is private and should not be
installed.
We weren't taking a reference on the texture to be used as the color buffer
for offscreen rendering, so it was possible to free the texture leaving the
framebuffer in an inconsistent state.
This adds gives Cogl a dedicated UProf context which will be linked together
with Clutter's context during clutter_init_real().
Initial timers cover _cogl_journal_flush and _cogl_journal_log_quad
You can explicitly ask for a report of Cogl statistics by exporting
COGL_PROFILE_OUTPUT_REPORT=1 but since the context is linked with Clutter's
the statisitcs will also be shown in the automatic Clutter reports.
This suspends and resumes all uprof timers and counters except while dealing
with picking, so as to give more focused statistics.
Be aware that there are still some issues with this profile option since
there are a few special case counters and timers that shouldn't be
suspended; noteably the frame counters are incorrect so the per frame stats
can't be trusted.
As we have for debugging, this adds the ability to control profiling flags
either via the command line or an environment variable.
The first option added is CLUTTER_PROFILE=disable-report
This also changes the reporting to be opt-out so you don't need to export
CLUTTER_PROFILE_OUTPUT_REPORT=1 to see a report but you can use
CLUTTER_PROFILE=disable-report to disable it if desired.
UProf is a small library that aims to help applications/libraries provide
domain specific reports about performance. It currently provides high
precision timer primitives (rdtsc on x86) and simple counters, the ability
to link statistics between optional components at runtime and makes report
generation easy.
This adds initial accounting for:
- Total mainloop time
- Painting
- Picking
- Layouting
- Idle time
The timing done by uprof is of wall clock time. It's not based on stochastic
samples we simply sample a counter at the start and end. When dealing with
the complexities of GPU drivers and with various kinds of IO this form of
profiling can be quite enlightening as it will be able to represent where
your application is blocking unlike tools such as sysprof.
To enable uprof accounting you must configure Clutter with --enable-profile
and have uprof-0.2 installed from git://git.moblin.org/uprof
If you want to see a report of statistics when Clutter applications exit you
should export CLUTTER_PROFILE_OUTPUT_REPORT=1 before running them.
Just a final word of caution; this stuff is new and the manual nature of
adding uprof instrumentation means it is prone to some errors when modifying
code. This just means that when you question strange results don't rule out
a mistake in the instrumentation. Obviously though we hope the benfits out
weigh e.g. by focusing on very key stats and by having automatic reporting.
Since asking for ARGB by default is still somewhat experimental on X11
and not every toolkit or complex widgets (like WebKit) still do not like
dealing with ARGB visuals, we should switch back to RGB by default - now
that at least we know it works.
For applications (and toolkit integration libraries) that want to enable
the ClutterStage:use-alpha property there is a new function:
void clutter_x11_set_use_argb_visual (gboolean use_argb);
which needs to be called before clutter_init().
The CLUTTER_DISABLE_ARGB_VISUAL environment variable can still be used
to force this value off at run-time.
Currently, ClutterActor detects a relayout cycle (an actor causing a
relayout to be queued from within an allocate() function) and aborts
after printing out a warning. This might be a little bit too anal
retentive, and it currently breaks GTK+ embedding inside clutter-gtk
so we should probably relax the behaviour a bit. Now we just emit the
warning but we still go ahead with the relayout.
When Clutter tries to pick an ARGB visual it tried to set the
GLX_TRANSPARENT_TYPE attribute of the FBConfig to
GLX_TRANSPARENT_RGB. However the code to do this was broken so that it
was actually trying to set the non-existant attribute number 0x8008
instead. Mesa silently ignored this so it appeared as if it was
working but the Nvidia drivers do not like it.
It appears that the TRANSPARENT_TYPE attribute is not neccessary for
getting an ARGB visual anyway and instead it is intended to support
color-key transparency. Therefore we can just remove it and get all of
the FBConfigs. Then if we need an ARGB visual we can just walk the
list to look for one with depth == 32.
The fbconfig is now stored in a single variable instead of having a
separate variable for the rgb and rgba configs because the old code
only ever retrieved one of them anyway.
Previously when the markup property is set it would generate an
attribute list from the markup and then merge it with the attributes
from the attribute property and store it as the effective
attributes. The markup attributes and the marked up text would then be
forgotten. This breaks if the application then later changes the
attributes property because it would try to regenerate the effective
attributes from the markup text but the stored text no longer contains
any markup. If the original markup text happened to contain entities
like '<' they would end up causing parse errors because they would
be converted to the actual symbols.
To fix this the attributes from the markup are now stored
independently from the effective attributes. The effective attributes
are now regenerated if either set of attributes changes right before a
layout is created.
http://bugzilla.openedhand.com/show_bug.cgi?id=1940
Destroy the dummy XImage we create even on success.
http://bugzilla.openedhand.com/show_bug.cgi?id=1918
Based on a patch by: Carlos Martín Nieto <carlos@cmartin.tk>
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
PropertyInfo should store a copy of the JsonNodes it references, so
that property_info_free() can safely dispose them, and we can reference
values across different UI definition data.
The implicit timeline parsing code is not copying the JsonNode; this
leads to a double free in some cases, which is masked by the GSlice
allocator and produces a heap corruption later on.
Allow the user of the ClutterMedia interface to specify an external (as
in not multiplexed with the audio/video streams) location of a subtitle
stream.
Both the ::insert-text and ::delete-text are "action" signals, that is
signals that are safe to (and should) be emitted using g_signal_emit()
directly.
A timeline advancement might cause another timeline to be
destroyed, which will likely lead to a segmentation fault.
Before advancing the timelines we should take a reference
on them - just like we do for the stages before doing
event processing. This will prevent dispose() from running
until the end of the advancement.
http://bugzilla.openedhand.com/show_bug.cgi?id=1854
Apparently, calling g_set_prgname() multiple times is not allowed
anymore, and hence clutter_init_* calls should not do that. Though this
is really GLib's fault - and a massive nuisance for us - we should
prolly comply to avoid the test suite dying on us.
* animate-layout-manager:
layout-manager: Document the animation support
layout-manager: Rewind the timeline in begin_animation()
box-layout: Remove the allocations hash table
docs: Clean up the README file
layout: Let begin_animation() return the Alpha
box-layout: Add knobs for controlling animations
box-layout: Animate layout properties
layout: Add animation support to LayoutManager
Add ActorBox animation methods
Add a section inside the LayoutManager class API reference documenting,
with examples, how to implement animation support inside a layout
manager sub-class.
If the default implementation begin_animation() is called twice then we
should rewind the timeline, as well as updating its duration and the
easing mode of the alpha.
The BoxLayout uses a HashTable to map the latest stable allocation of
each child, in order to use that as the initial value during an
animation; this in spite of already having a perfectly valid per-child
storage as part of the layout manager: ClutterBoxChild.
The last stable allocation should be stored inside the ClutterBoxChild
instead of having it in the private data for ClutterBoxLayout. The
access remains O(1), since there is a 1:1 mapping between child and
BoxChild instances, but we save a little bit of memory and we avoid
keeping aroud allocations for old children.
* stage-use-alpha:
tests: Use accessor methods for :use-alpha
stage: Add accessors for :use-alpha
tests: Allow setting the stage opacity in test-paint-wrapper
stage: Premultiply the stage color
stage: Composite the opacity with the alpha channel
glx: Always request an ARGB visual
stage: Add :use-alpha property
materials: Get the right blend function for alpha
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
When beginning a new animation for a LayoutManager, the implementation
should return the ClutterAlpha used. This allows controlling the
timeline and/or modifying the animation parameters on the fly.
ClutterLayoutManager does not have any state associated with it, and
defers all the state to its sub-classes.
The BoxLayout is thus in charge of controlling:
• whether or not animations should be used
• the duration of the animation
• the easing mode of the animation
By adding three new properties:
• ClutterBoxLayout:use-animations
• ClutterBoxLayout:easing-duration
• ClutterBoxLayout:easing-mode
And their relative accessors pairs we can make BoxLayout decide whether
or not, and with which parameters, call the begin_animation() method of
ClutterLayoutManager.
The test-box-layout has been modified to reflect this new functionality,
by checking the key-press event for the 'a' key symbol to toggle the use
of animations.
Use the newly added animation support inside LayoutManager to animate
between state changes of the BoxLayout properties.
The implementation is based on equivalent code from Mx, written by:
Thomas Wood <thomas.wood@intel.com>
In order to animate a fluid layout we cannot use the common animation
code paths as they will override the size request and allocation paths
that are handled by the layout manager itself.
One way to introduce animations in the allocation sequence is to use a
Timeline and an Alpha to compute a progress value and then use that
value to interpolate an ActorBox between the initial and final states of
the animation - with the initial state being the last allocation of the
child prior to the animation start, and the final state the allocation
of the child at the end; for every frame of the Timeline we then queue a
relayout on the layout manager's container, which will result in an
animation.
ClutterLayoutManager is the most likely place to add a generic API for
beginning and ending an animation, as well as the place to provide a
default code path to create the ancillary Timeline and Alpha instances
needed to drive the animation.
A LayoutManager sub-class will need to:
• call clutter_layout_manager_begin_animation() whenever it should
animate between two states, for instance: whenever a layout property
changes value;
• eventually override begin_animation() and end_animation() in case
further state needs to be set up, and then chain up to the default
implementation provided by LayoutManager;
• if a completely different implementation is required, the layout
manager sub-class should override begin_animation(), end_animation()
and get_animation_progress().
Inside the allocate() implementation the sub-class should also
interpolate between the last known allocation of a child and the newly
computed allocation.
ClutterActorBox should have an interpolate() method that allows to
compute the intermediate values between two states, given a progress
value, e.g.:
clutter_actor_box_interpolate (start, end, alpha, &result);
Another utility method, useful for layout managers, is a modifier
that clamps the members of the actor box to the nearest integer
value.
When getting signals from higher level toolkits, occasionally
one wants access to the underlying event; say for a Button
widget's "clicked" signal, to get the keyboard state.
Rather than having all of the highlevel widgets emit
ClutterEvent just for the more unusual use cases,
add a global function to access the event state.
http://bugzilla.openedhand.com/show_bug.cgi?id=1888
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Old-style X11 terminals require that even modern X11 send KeyPress
and KeyRelease pairs when auto-repeating. For this reason modern(-ish)
API like XKB has a way to detect auto-repeat and do a single KeyRelease
at the end of a KeyPress sequence.
The newly added check emulates XKB's detectable auto-repeat by peeking
the next event after a KeyRelease and checking if it's a KeyPress for
the same key and timestamp - and then ignoring the KeyRelease if it
matches.
If a Stage has been set to use a foreign Window then Clutter should not
be managing it; calling XWithdrawWindow and XMapWindow should be
reserved to the windows we manage ourselves.
Some actor implementation might avoid imposing any layout on their
children. The Actor base class usually assumes some sort of layout
management is in place, so it will queue relayouts when, for instance,
an actor is shown or is hidden. If the parent of the actor does not
impose any layout, though, showing or hiding one of its children will
not affect the layout of the others.
An example of this kind of container is ClutterGroup.
By adding a new Actor flag, CLUTTER_ACTOR_NO_LAYOUT, and by making
the Group actor set it on itself, the Actor base class can now decide
whether or not to queue a relayout. The flag is not meant to be used
by application code, and should only be set when implementing a new
container.
http://bugzilla.openedhand.com/show_bug.cgi?id=1838
When the texture is in the atlas, ensuring the mipmaps can effectively
make it become a completely different texture so we should do this
before getting the GL handle.
Mipmaps don't work very well in the current atlas because there is not
enough padding between the textures. If ensure_mipmaps is called it
will now create a new texture and migrate the atlased texture to
it. It will use the same blit mechanism as when migrating so it will
try to use an FBO for a fast blit. However if this is not possible it
will end up downloading the data for the entire atlas which is not
ideal.
When reorganizing the textures, we can avoid downloading the entire
texture data if we bind the source texture in a framebuffer object and
copy the destination using glCopyTexSubImage2D. This is also
implemented using a much faster path in Mesa.
Currently it is calling the GL framebuffer API directly but ideally it
would use the Cogl offscreen API. However there is no way to tell Cogl
not to create a stencil renderbuffer which seems like a waste in this
situation.
If FBOs are not available it will fallback to reading back the entire
texture data as before.
This adds a 'dump-atlas-image' debug category. When enabled, CoglAtlas
will use Cairo to create a png which visualizes the leaf rectangles of
the atlas.
This adds an 'atlas' category to the COGL_DEBUG environment
variable. When enabled Cogl will display messages when textures are
added to the atlas and when the atlas is reorganized.
When space can't be found in the atlas for a new texture it will now
try to reorganize the atlas to make space. A new CoglAtlas is created
and all of the textures are readded in decreasing size order. If the
textures still don't fit then the size of the atlas is doubled until
either we find a space or we reach the texture size limits. If we
successfully find an organization that fits then all of the textures
will be migrated to a new texture. This involves copying the texture
data into CPU memory and then uploading it again. Potentially it could
eventually use a PBO or an FBO to transfer the image without going
through the CPU.
The algorithm for laying out the textures works a lot better if the
rectangles are added in order so we might eventually want some API for
creating multiple textures in one go to avoid reorganizing the atlas
as far as possible.
This adds a CoglAtlas type which is a data structure that keeps track
of unused sub rectangles of a larger rectangle. There is a new atlased
texture backend which uses this to put multiple textures into a single
larger texture.
Currently the atlas is always sized 256x256 and the textures are never
moved once they are put in. Eventually it needs to be able to
reorganise the atlas and grow it if necessary. It also needs to
migrate the textures out of the atlas if mipmaps are required.
clutter_actor_get_preferred_width/height currently caches only one size
requests, for a given height / width.
It's common for a layout manager to call get_preferred_width with 2
different heights during the same allocation cycle. Typically once in
the size request, once in the allocation. If
clutter_actor_get_preferred_width is called
alternatively with 2 different for_height, the cache is totally
inefficient, and we end up always querying the actor size even
when the actor does not need a re-allocation.
http://bugzilla.openedhand.com/show_bug.cgi?id=1876
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Fix a copy-and-paste thinko where the cell size was computed using the
minimum size instead of the natural size. For actors with a minimum size
of zero, like Textures, this implied always a zero allocation.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
This is an optimised version of CoglTexture2DSliced that always deals
with a single texture and always uses the GL_TEXTURE_2D
target. cogl_texture_new_from_bitmap now tries to use this backend
first. If it can't create a texture with that size then it falls back
the sliced backend.
cogl_texture_upload_data_prepare has been split into two functions
because the sliced backend needs to know the real internal format
before the conversion is performed. Otherwise the converted bitmap
will be wasted if the backend can't support the size.
This provides a way to upload the entire data for a texture without
having to first call glTexImage and then glTexSubImage. This should be
faster especially with indirect rendering where it would needlessy
send the data for the texture twice.
new_from_data and new_from_file can be implemented in terms of
new_from_bitmap so it makes sense to move these to cogl-texture rather
than having to implement them in every texture backend.
This adds a new texture backend which represents a sub texture of a
larger texture. The texture is created with a reference to the full
texture and a set of coordinates describing the region. The backend
simply defers to the full texture for all operations and maps the
coordinates to the other range. You can also use coordinates outside
the range [0,1] to create a repeated version of the full texture.
A new public API function called cogl_texture_new_from_sub_texture is
available to create the sub texture.
The CoglTextureSliceCallback function pointer now takes const pointers
for the texture coordinates. This makes it clearer that the callback
should not modify the array and therefore the backend can use the same
array for both sets of coords.
Given a region of texture coordinates this utility invokes a callback
enough times to cover the region with a subregion that spans the
texture at most once. Eg, if called with tx1 and tx2 as 0.5 and 3.0 it
it would invoke the callback with:
0.5,1.0 1.0,2.0 2.0,3.0
Manual repeating is needed by all texture backends regardless of
whether they can support hardware repeating because when Cogl calls
the foreach_sub_texture_in_region method then it sets the wrap mode to
GL_CLAMP_TO_EDGE and no hardware repeating is possible.
In _cogl_multitexture_quad_single_primitive we use a wrap mode of
GL_CLAMP_TO_EDGE if the texture coordinates are all in the range [0,1]
or GL_REPEAT otherwise. This is to avoid pulling in pixels from either
side when using GL_LINEAR filter mode and rendering the entire
texture. Previously it was checking using the unconverted texture
coordinates. This is ok unless the texture backend is radically
transforming the texture coordinates, such as in the sub texture
backend where the coordinates may map to something completely
different. We now check whether the coordinates are in range after
converting them.
Most of the fields that were previously in CoglTexture are specific to
the implementation of CoglTexture2DSliced so they should be placed
there instead. For example, the 'mipmaps_dirty' flag is an
implementation detail of the ensure_mipmaps function so it doesn't
make sense to force all texture backends to have this function.
Other fields such as width, height, gl_format and format may make
sense for all textures but I've added them as virtual functions
instead. This may make more sense for a sub-texture backend for
example where it can calculate these based on the full texture.
The CoglTexture struct previously contained some fields which are only
used to upload data such as the CoglBitmap and the source GL
format. These are now moved to a separate CoglTextureUploadData struct
which only exists for the duration of one of the cogl_texture_*_new
functions. In cogl-texture there are utility functions which operate
on this new struct rather than on CoglTexture directly.
Some of the fields that were previously stored in the CoglBitmap
struct are now copied to the CoglTexture such as the width, height,
format and internal GL format.
The rowstride was previously stored in CoglTexture and this was
publicly accessible with the cogl_texture_get_rowstride
function. However this doesn't seem to be a useful function because
there is no need to use the same rowstride again when uploading or
downloading new data. Instead cogl_texture_get_rowstride now just
calculates a suitable rowstride from the format and width of the
texture.
Commit 558b17ee1e added support for rectangle textures to the
framebuffer code. Under GLES there is no GL_TEXTURE_RECTANGLE_ARB
definition so this was breaking the build. The rest of Cogl uses
ifdef's around that constant so we should do the same here.
• The debug flags are pre-processor ones, so they should be listed
inside AM_CPPFLAGS.
• Clutter's publicly exported symbols match the following regular
expression:
^(clutter|cogl|json)_*
The old one also listed "pango" as a possible prefix, but the
Pango API is now under the Cogl namespace.
We need to add the row-spacing value when calculating the y position for lines
of actors in horizontal flowing layouts.
Similarly we need to add the col-spacing value when calculating the x posution
for actors in vertical flowing layouts.
When requesting the GLXFBConfig for creating the GLX context, we should
always request one that links to an ARGB visual instead of a plain RGB
one.
By using an ARGB visual we allow the ClutterStage:use-alpha property to
work as intended when running Clutter under a compositing manager.
The default behaviour of requesting an ARGB visual can be disabled by
using the:
CLUTTER_DISABLE_ARGB_VISUAL
Environment variable.
The ClutterStage:use-alpha property is used to let a stage know that it
should honour the alpha component of the ClutterStage:color property.
If :use-alpha is set to FALSE the stage always uses the full opacity
when clearing itself before a paint(); otherwise, the alpha value is
used.
The correct blend function for the alpha channel is:
GL_ONE, GL_ONE_MINUS_SRC_ALPHA
As per bug 1406. This fix was dropped when the switch to premultiplied
alpha was merged.
