We have a bunch of experimental convenience functions like
cogl_primitive_p2/p2t2 that have corresponding vertex structures but it
seemed a bit odd to have the vertex annotation e.g. "P2T2" be an infix
of the type like CoglP2T2Vertex instead of be a postfix like
CoglVertexP2T2. This switches them all to follow the postfix naming
style.
When determining the maximum number of layers we also need to take
into account GL_MAX_VERTEX_ATTRIBS on GLES2. Cogl needs one vertex
attrib for each texture unit plus two for the position and color.
This creates a material which users a layer to override the color of
the rectangle. A simple vertex shader is then created which just
emulates the fixed function pipeline. No fragment shader is
added. This demonstrates a bug where the layer state is getting
ignored when a vertex shader is in use.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2221
Previously the alpha component of the test texture data was always set
to 255 and the data was read back as RGB so that the alpha component
is ignored. Now the alpha component is set to a generated value and
the data is read back a second time as RGBA to verify that Cogl is not
doing any premult conversions when the internal texture and target
data is the same.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2414
The CSS Color Module 3, available at:
http://www.w3.org/TR/css3-color/
allows defining colors as:
rgb ( r, g, b )
rgba ( r, g, b, a)
along with the usual hexadecimal and named notations.
The r, g, and b channels can be:
• integers between 0 and 255
• percentages, between 0% and 100%
The alpha channel, if included using the rgba() modifier, can be a
floating point value between 0.0 and 1.0.
The ClutterColor parser should support this notation.
The Behaviour class and its implementations have been replaced by the
new animation framework API and by the constraints for layout-related
animations.
Currently, we need to make tests build, so we undef DISABLE_DEPRECATED
in specific test cases while they get ported.
The size of the texture used for test-cogl-npot-texture was only using
1 pixel of waste and the texture was scaled down so it would be quite
likely that the test would still pass if only the top left slice was
rendered. It also didn't test using non-default texture
coordinates. These problems made it fail to pick up bug 2398. The
texture is now using the maximum amount of waste and rendered in four
parts at 1:1 scale.
Previously in the tests/tools directory we build a disable-npots
library which was used as an LD_PRELOAD to trick Cogl in to thinking
there is no NPOT texture extension. This is a little awkward to use so
it seems much simpler to just define a COGL_DEBUG option to disable
npot textures.
Unparented actors are owned by the Script instance, and if that goes
away then the actors go away with it. The fact that we needed an
explicit destroy() before was a hint of a memory management issue that I
blissfully - and regretfully - ignored for the sake of a passing test
suite.
They are generated at configure time, so it's a good idea to have them
in the main ignore file instead of adding them to the built ignore files
under tests.
Re-order the units into a sensible list, with basic tests at the
beginning, and per-class tests at the end - with Cogl last.
Also, start renaming the unit functions from test_<foo> to <foo>,
so that the executable wrappers and the reports have sensible names.
The TODO() macro for adding new tests to the test suite has always meant
to be implemented like the TODO block in Test::More, i.e. a test that is
assumed to fail, and which warns if it unexpectedly succeeds.
Since GTest lacks the expressivity of Test::More, the implementation
just verifies that the tests marked as TODO actually fail, and will fail
if they happen to succeed - at which point the developer will have to
change the macro to SIMPLE or SKIP.
Even if gtester-report doesn't use that information (yet), we should
store the revision of Clutter that generated the report, and the date in
which the test suite was ran.
Instead of trying to run ./test-conformance with the -l option to
generate a list of available tests it now runs sed on the
test-conform-main.c file instead. Running the generated executable is
a pain for cross-compiling so it would be nice to avoid it unless it's
absolutely necessary. Although you could tell people who are cross
compiling to just disable the conformance tests, this seems a shame
because they could still be useful along with the wrappers for example
if the cross compile is built to a shared network folder where the
tests can be run on the actual device.
The sed script is a little more ugly than it could be because it tries
to avoid using the GNU extensions '\+' and '\|'.
The script ends up placing restrictions on the format of the C file
because the tests must all be listed on one line each. There is now a
comment to explain this. Hopefully the trade off is worth it.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2363
Instead of calling clutter_init immediately, test-conformance now only
calls it as part of test_conform_simple_fixture_setup. The conformance
tests assert that only one test is run per instance of
test-conformance so it should never end up calling clutter_init
twice. Delaying clutter_init has the advantage that calling
"test-conformance -l" will still work even on systems with no X
server. This could be useful for automated build systems.
The keysyms defines in clutter-keysyms.h are generated from the X11 key
symbols headers by doing the equivalent of a pass of sed from XK_* to
CLUTTER_*. This might lead to namespace collisions, down the road.
Instead, we should use the CLUTTER_KEY_* namespace.
This commit includes the script, taken from GDK, that parses the X11
key symbols and generates two headers:
- clutter-keysyms.h: the default included header, with CLUTTER_KEY_*
- clutter-keysyms-compat.h: the compatibility header, with CLUTTER_*
The compat.h header file is included if CLUTTER_DISABLE_DEPRECATED is
not defined - essentially deprecating all the old key symbols.
This does not change any ABI and, assuming that an application or
library is not compiling with CLUTTER_DISABLE_DEPRECATED, the source
compatibility is still guaranteed.
This creates a 3D texture with different colors on all of the images
and renders it using a VBO to verify that the texture coordinates can
select all of the images.
This verifies that calling cogl_texture_get_data returns the same data
uploaded to the texture. The bottom quarter of the texture is replaced
using cogl_texture_set_region. It tries creating the texture with
different sizes and flags in the hope that it will hit different
texture backends.
This greatly speeds up running all the conformance tests by no longer
delaying many of the tests for a number of dummy frames to be painted.
We used to skip frames because we thought there was a problem with the
driver's glReadPixels implementation. Although we have seen driver
issues at times the real reason the delay was needed was because
resizing the stage usually happens asynchronously (because a non
synchronous X request is used by clutter_stage_set_size()). We now force
all X requests to be synchronized for the conformance tests so this is
no longer a problem and we can avoid these hacks.
This makes test-timeline get the default stage so there is at least one
stage instantiated. Without any stages the master clock will never run
which was causing this test to fail.
The report generation was broken by the split of the various test units;
also, we were using GTest in a way that's not really sanctioned by
upstream.
This commit tries to re-use the targets from GLib's Makefile rules while
compensating for our own set up.
Normally the asynchronous nature of X means that setting the clutter
stage size may really happen an indefinite amount of time later but
since the tests are so short lived and may only render a single frame
this is not an acceptable semantic.
This way we should be able to remove all the hacky sleeps and frame
count delays from our tests.
Since we now run every test in a separate process there is no need to
try and avoid state leakage between tests. This removes the code to
cleanup all children of the stage and disconnect handlers from the
stage paint signal. We now explicitly print a warning if the users tries
to run multiple tests in one process.
In the .json file used for the test, there is no null -> "base"
transition defined only a "clicked" -> "base", when the "clicked" state
is removed the "base" state will also disappear.
I was fed up to cd into the tests/conform or tests/interactive directories
to launch a specific test. Now, with the power the abs_ variants of
builddir and srcdir we can run specific test from any directory.
http://bugzilla.clutter-project.org/show_bug.cgi?id=2159
This allows you to tell Cogl that you are planning to replace all the
buffer's data once it is mapped with cogl_buffer_map. This means if the
buffer is currently being accessed by the GPU then the driver doesn't
have to stall and wait for it to finish before it can access it from the
CPU and can instead potentially allocate a new buffer with undefined
data and map that.
There was a missing '* 4' and '* i' in the for() loops that initialized
the first test buffer, so it was containing uninitialized data causing
the test to fail.
Since CoglPixelBuffer was renamed to CoglPixelArray the test entry point
was also renamed to test_cogl_pixel_array, but mistakenly the
corresponding test-conform-main.c change wasn't pushed at the same time.
This removes cogl_pixel_array_new which just took a size in bytes.
Without the image size and pixel format then the driver often doesn't
have enough information to allocate optimal GPU memory that can be
textured from directly. This is because GPUs often have ways to
spatially alter the layout of a texture to improve cache access patterns
which may require special alignment and padding dependant in the images
width, height and bpp.
Although currently we are limited by OpenGL because it doesn't let us
pass on the width and height when allocating a PBO, the hope is that we
can define a better extension at some point.
The usage hint should be implied by the CoglBuffer subclass type so the
public getter and setter APIs for manually changing the usage hint of a
CoglBuffer have now been removed.
This creates a path with an outer clockwise and two internal sub
paths, one clockwise and one counter-clockwise. The path is then
painted twice, once with each fill rule.
The -rdynamic linker option is specific to ELF so it was breaking
builds on systems with other object formats such as Windows and
Solaris. This patch replaces that option with -export-dynamic which is
a portable libtool option which should do the right thing on each
platform.
http://bugzilla.clutter-project.org/show_bug.cgi?id=1930
This adds a separate variable name "CLUTTER_SONAME_INFIX" to define the
infix for the clutter library that gets linked. Currently the WINSYS
corresponds to the directory we enter when building to compile the
window system and input support, but it is desirable to be able to
define multiple flavours that use the same WINSYS but should result in
different library names.
For example we are planning to combine the eglx and eglnative window
systems into one "egl" winsys but we will need to preserve the current
library names for the eglx and eglnative flavours.
This redirects the legacy depth testing APIs through CoglMaterial and
adds a new experimental cogl_material_ API for handling the depth
testing state.
This adds the following new functions:
cogl_material_set_depth_test_enabled
cogl_material_get_depth_test_enabled
cogl_material_set_depth_writing_enabled
cogl_material_get_depth_writing_enabled
cogl_material_set_depth_test_function
cogl_material_get_depth_test_function
cogl_material_set_depth_range
cogl_material_get_depth_range
As with other experimental Cogl API you need to define
COGL_ENABLE_EXPERIMENTAL_API to access them and their stability isn't
yet guaranteed.
This test breaks out into raw OpenGL to create a RECTANGLE texture so it
needs to be careful not to trample on any state that may be cached by
Cogl internally.
Layout properties work similarly to child properties, with the added
headache that they require the 3-tuple:
( layout manager, container, actor )
to be valid in order to be inspected, parsed and applied. This means
using the newly added back-pointer from the container to the layout
manager and then rejigging a bit how the ScriptParser handles the
unresolved properties.
Similarly to the child properties, which use the "child::" prefix, the
layout manager properties use the "layout::" prefix and are defined with
the child of a container holding a layout manager.
This adds a boolean "pick-with-alpha" property to ClutterTexture and when
true, it will use the textures alpha channel to define the actors shape when
picking.
Users should be aware that it's a bit more expensive to pick textures like
this (so probably best not to blindly enable it on *all* your textures)
since it implies rasterizing the texture during picking whereas we would
otherwise just send a solid filled quad to the GPU. It will also interrupt
the internal batching of geometry for pick renders which can otherwise often
be done in a single draw call.
The drawing code from test_invalid_texture_layers which draws a
rectangle, a polygon and a vertex buffer has been split out to
separate function. test_using_all_layers now also uses this so that it
will also test the other two primitives. This causes the test to fail
when all of the layers are drawn using a vertex buffer.
http://bugzilla.openedhand.com/show_bug.cgi?id=2132
GLES 2 doesn't have GL_MAX_TEXTURE_UNITS. Instead the cogl backend
uses GL_MAX_TEXTURE_IMAGE_UNITS with a maximum limit of 16. The same
restriction is now used in the test.
This adds a simple test for ClutterCairoTexture that draws two
rectangles to the cairo surface in an idle callback and then verifies
that they appeared at the right colours in the paint callback. If that
succeeds then the second time the idle callback is invoked it will
replace one of the rectangles with a sub region update and the
following paint callback will again verify the rectangles.
Instead of asking gtester to run ./test-conformance directly we now tell
it to run a list of wrapper scripts. This results in each test being
spawned in a separate process avoiding leakage of state between tests
which has been a big problem with the conformance tests for quite a
while now.
Otherwise it seems that rounding errors will cause the fragments at
the edge of the quad to blend with neighbouring quarters of the
texture which cause the test to fail.
A few of the tests connected to the paint signal but never
disconnected it. Most of these handlers had a call to g_main_quit in
them which meant that it could sometimes cause subsequent tests to
exit after the first frame is painted. Most of the tests don't
validate any of the results until after a couple of frames have been
rendered so this ended up skipping out the test entirely.
To workaround this the test setup function now disconnects all
handlers for the paint signal on the default stage before the test is
run.
The on_paint function for test-cogl-readpixels tries to temporarily
set the projection, modelview and viewport to its own values. However
it was never restoring the saved values so it could affect the results
of subsequent tests.
This adds a test which creates a material using the maximum number of
layers. All of the layers are assigned a white texture except the last
which is set to red. The default combine mode is used for all of the
layers so the final fragment should end up red.
Currently Cogl doesn't provide a way to query the maximum number of
layers so it just uses glGetIntegerv instead. This might cause
problems on GLES 2 because that adds additional restrictions on the
number of layers.
http://bugzilla.openedhand.com/show_bug.cgi?id=2064
This changes the original tests so that it splits the original path
into two sub paths. When adding a new block to the copied path it also
adds another sub path. This further stresses the path copying
mechanism and exposes a bug.
It also tests intersections by drawing a self-intersecting path and a
path with two sub-paths that overlap. Where the path overlaps it
should be inverted.
This renders a texture using different combinations of wrap modes for
the s and t coordinates and then verifies that the expected wrapping
is acheived. The texture is drawn using rectangles, polygons and
vbos. There is also code to test a rectangle using an atlased texture
(which should test the manual repeating) however the validation for
this is currently disabled because it doesn't work.
http://bugzilla.openedhand.com/show_bug.cgi?id=2063
This tests various paths drawing rectangles and verifies that the
expected pixels are filled in. Some of the paths are drawn by copying
an existing path and modifying it which should test the copy-on-write
code.
The test creates a GL_TEXTURE_RECTANGLE_ARB texture using
cogl_texture_new_from_foreign and confirms that rendering it works
correctly. If the rectangle texture extension isn't available then
this test always succeeds.
http://bugzilla.openedhand.com/show_bug.cgi?id=2015
cogl_read_pixels() no longer asserts that the format passed in is
RGBA_8888 but instead accepts any format. The appropriate GL enums for
the format are passed to glReadPixels so OpenGL should be perform a
conversion if neccessary.
It currently assumes glReadPixels will always give us premultiplied
data. This will usually be correct because the result of the default
blending operations for Cogl ends up with premultiplied data in the
framebuffer. However it is possible for the framebuffer to be in
whatever format depending on what CoglMaterial is used to render to
it. Eventually we may want to add a way for an application to inform
Cogl that the framebuffer is not premultiplied in case it is being
used for some special purpose.
If the requested format is not premultiplied then Cogl will convert
it. The tests have been changed to read the data as premultiplied so
that they won't be affected by the conversion. Picking in Clutter has
been changed to use COGL_PIXEL_FORMAT_RGB_888 because it doesn't need
the alpha component. clutter_stage_read_pixels is left unchanged
because the application can't specify a format for that so it seems to
make most sense to store unpremultiplied values.
http://bugzilla.openedhand.com/show_bug.cgi?id=1959
Allow a ClutterModel to be constructed through the ClutterScript API.
Currently this allows a model to be generated like like this:
{
"id" : "test-model",
"type" : "ClutterListModel",
"columns" : [
[ "text-column", "gchararray" ],
[ "int-column", "gint" ],
[ "actor-column", "ClutterRectangle" ]
]
}
where 'columns' is an array containing arrays of column-name,
column-type pairs.
http://bugzilla.openedhand.com/show_bug.cgi?id=2007
The adds tests for the remaining layer combine functions, the 1 minus
value operator and the TEXTURE_N source. Note however that Cogl
currently fails when parsing a TEXTURE_N source so the test is
commented out.
We need to make the Stage set the MAPPED flag on itself if we want to
verify the MAPPED state. That was always the case - it just worked
before because the Stage was shown at least once.
Since all conformance tests share the same state we should not touch
stuff like the stage size; sharing is already fairly complex and adds a
lot of caveats on the implementation of a conformance test unit, and if
we make tests influence later ones then we might slip in bugs or false
negatives - thus defeating the whole point of a conformance test suite.
The g_assert_cmpint() macro prints out not just the assertion condition
but also the assertion contents; this is useful to catch wrong values
without incrementing the verbosity of the test itself.
Since the "internal" state is global, it will leak onto actors that you
didn't intend for it to, because it applies not just to the actors you
create, but also to any actors *they* create. Eg, if you have a dialog
box class, you might push/pop_internal around creating its buttons, so
that those buttons get marked as internal to the dialog box. But
ctx->internal_child will still be set during the *button*'s constructor
as well, and so, eg, the label and icon inside the button actor will
*also* be marked as internal children, even if that isn't what the
button class wanted.
The least intrusive change at this point is to make push_internal() and
pop_internal() two methods of the Actor class, and take a ClutterActor
pointer as the argument - thus moving the locality of the internal_child
counter to the Actor itself.
http://bugzilla.openedhand.com/show_bug.cgi?id=1990
When creating a Cogl sub-texture, if the full texture is also a sub
texture it will now just offset the x and y and reference the full
texture instead. This avoids one level of indirection when rendering
the texture which reduces the chances of getting rounding errors in
the calculations.
The test was calling g_object_get to fetch the "opacity-start" property
(unsigned int) into a guint8 local variable. It's a bit of a mean trap
given that the getter function returns guint8 values so this also adds a
comment explaining what's going on.
The whole point of having the Animator class is that the developer can
describe a complex animation using ClutterScript. Hence, ClutterAnimator
should hook into the Script machinery and parse a specific description
format for its keys.
This adds three new texture backends.
- CoglTexture2D: This is a trimmed down version of CoglTexture2DSliced
which only supports a single texture and only works with the
GL_TEXTURE_2D target. The code is a lot simpler so it has a less
overheads than dealing with slices. Cogl will use this wherever
possible.
- CoglSubTexture: This is used to get a CoglHandle to represent a
subregion of another texture. The texture can be used as if it was a
standalone texture but it does not need to copy the resources.
- CoglAtlasTexture: This collects RGB and RGBA textures into a single
GL texture with the aim of reducing texture state changes and
increasing batching. The backend will try to manage the atlas and
may move the textures around to close gaps in the texture. By
default all textures will be placed in the atlas.
The coverage of the Behaviour sub-classes is currently abysmal. An
initial test suite for Behaviours should at least verify that the
accessors and the constructors are doing the right thing.
This initial test suite just verifies the BehaviourOpacity sub-class,
but it already bumps up the overall coverage by 2%.
It's very useful to see the actual number the reference value is
compared too when the test fails. GTest has g_assert_cmp$type()
functions for that, so make good use of them.
Otherwise the paint handler will still be run for the subsequent
tests. This ends up writing to the ‘state’ variable which used to be
on the stack so it will end up corrupting some stack variable. This
was causing test-cogl-premult to fail.
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.
This adds a test which renders a texture into a 1x1 pixel quad with
and without filters that use mipmaps. The pixel without mipmaps will
be one of the colors from the texture and the one with will be the
average of all the pixels in the texture.
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.
This tests creating a sub texture from a larger texture using various
different texture coordinates. It also tries to read back the texture
data using cogl_texture_get_data.
cogl_push_draw_buffer, cogl_set_draw_buffer and cogl_pop_draw_buffer are now
deprecated and new code should use the new cogl_framebuffer_* API instead.
Code that previously did:
cogl_push_draw_buffer ();
cogl_set_draw_buffer (COGL_OFFSCREEN_BUFFER, buffer);
/* draw */
cogl_pop_draw_buffer ();
should now be re-written as:
cogl_push_framebuffer (buffer);
/* draw */
cogl_pop_framebuffer ();
As can be seen from the example above the rename has been used as an
opportunity to remove the redundant target argument from
cogl_set_draw_buffer; it now only takes one call to redirect to an offscreen
buffer, and finally the term framebuffer may be a bit more familiar to
anyone coming from an OpenGL background.
cogl_clip_push, and cogl_clip_push_window_rect which are now deprecated were
used in various places internally so this just switches to using the
replacement functions.
The ClutterScript parser needs to be extended to parse child properties
and apply them after an actor has been added to a container. In order to
distinguish child properties from regular GObject properties we can use
the "child::" prefix, e.g.:
{
"type" : "ClutterRectangle",
"id" : "child-01",
"child::has-focus" : true,
...
}
Parsing child properties can be deferred to the ClutterScriptable
interface, just like regular properties.
All the ClutterColor parsing rules should be coalesced inside
clutter_script_parse_color(): object, array and string notations
are the canonical ways of defining a ClutterColor inside a
ClutterScript definition. Having a single function in charge of
the parsing cleans up the code.
ClutterScript is a very complicated piece of machinery, with a
parser that has custom variations on top of the basic JSON
format; it could also be extended in the future, so if we don't
want to introduce regressions or break existing ClutterScript
definitions, we'd better have a conformance test suite.
The units under the conformance test suite should be able to use
external files. Linking the files in tests/conform like the
interactive tests do seems like a hack piled on top of a hack, so
instead we should provide a programmatic way for a conformance
test unit to get the full path of a file, regardless of where the
tests/data directory is.
We can use a define to get the full path of tests/data and then
a function using g_build_filename() to construct the path to the
file we want.
Since offscreen rendering is internally forced to be upside down Cogl
needs to reverse the glFrontFace winding order so as not to interfere
with the use of cogl_set_backface_culling_enabled()
This ensures we test that mechanism.
Mostly this was written to verify that we don't flip the data read back from
an offscreen draw buffer. (since all offscreen rendering is done upside
down)
This adds a basic test to check that rendering a few colored rectangles
offscreen works and that the modelview gets restored when switching back to
the previous buffer.
Unlike OpenGL Cogl puts the origin of windows/viewports at the top left
instead of bottom left. This test verifies that we correctly translate Cogl
viewports to OpenGL viewports for the awkward cases where the given viewport
has an offset and/or the viewport has a different size to the current draw
buffer.
It helps to be able to quickly glance at the definition to see which
quadrant of the test actor should be which color, so when debugging a
problem and looking at the visual output you can easily verify if it's being
flipped upside down/left to right.
This contains four tests :-
- A regular onscreen source with a clone next to it
- An offscreen source with a clone. This is currently commented out
because it no longer works.
- An onscreen source with a rectangular clip and a clone.
- An onscreen source with a clip from a path and a clone.
The sources are all a 2x2 grid of colors. Each clone is tested that it
either contains the color that should be at that grid position or that
the stage color is showing through if the source is clipped.
The additional check draws another front facing rectangle but this time with
the texture coords flipped on the x axis. The code that handles sliced
textures in cogl-primitives.c makes some suspicious changes to the geometry
when the texture coords are inverted.
As part of an incremental process to have Cogl be a standalone project we
want to re-consider how we organise the Cogl source code.
Currently this is the structure I'm aiming for:
cogl/
cogl/
<put common source here>
winsys/
cogl-glx.c
cogl-wgl.c
driver/
gl/
gles/
os/ ?
utils/
cogl-fixed
cogl-matrix-stack?
cogl-journal?
cogl-primitives?
pango/
The new winsys component is a starting point for migrating window system
code (i.e. x11,glx,wgl,osx,egl etc) from Clutter to Cogl.
The utils/ and pango/ directories aren't added by this commit, but they are
noted because I plan to add them soon.
Overview of the planned structure:
* The winsys/ API is the API that binds OpenGL to a specific window system,
be that X11 or win32 etc. Example are glx, wgl and egl. Much of the logic
under clutter/{glx,osx,win32 etc} should migrate here.
* Note there is also the idea of a winsys-base that may represent a window
system for which there are multiple winsys APIs. An example of this is
x11, since glx and egl may both be used with x11. (currently only Clutter
has the idea of a winsys-base)
* The driver/ represents a specific varient of OpenGL. Currently we have "gl"
representing OpenGL 1.4-2.1 (mostly fixed function) and "gles" representing
GLES 1.1 (fixed funciton) and 2.0 (fully shader based)
* Everything under cogl/ should fundamentally be supporting access to the
GPU. Essentially Cogl's most basic requirement is to provide a nice GPU
Graphics API and drawing a line between this and the utility functionality
we add to support Clutter should help keep this lean and maintainable.
* Code under utils/ as suggested builds on cogl/ adding more convenient
APIs or mechanism to optimize special cases. Broadly speaking you can
compare cogl/ to OpenGL and utils/ to GLU.
* clutter/pango will be moved to clutter/cogl/pango
How some of the internal configure.ac/pkg-config terminology has changed:
backendextra -> CLUTTER_WINSYS_BASE # e.g. "x11"
backendextralib -> CLUTTER_WINSYS_BASE_LIB # e.g. "x11/libclutter-x11.la"
clutterbackend -> {CLUTTER,COGL}_WINSYS # e.g. "glx"
CLUTTER_FLAVOUR -> {CLUTTER,COGL}_WINSYS
clutterbackendlib -> CLUTTER_WINSYS_LIB
CLUTTER_COGL -> COGL_DRIVER # e.g. "gl"
Note: The CLUTTER_FLAVOUR and CLUTTER_COGL defines are kept for apps
As the first thing to take advantage of the new winsys component in Cogl;
cogl_get_proc_address() has been moved from cogl/{gl,gles}/cogl.c into
cogl/common/cogl.c and this common implementation first trys
_cogl_winsys_get_proc_address() but if that fails then it falls back to
gmodule.
When computing the pixels value of a ClutterUnits value we should
be caching the value to avoid recomputing for every call of
clutter_units_to_pixels(). We already have a flag telling us to
return the cached value, but we miss the mechanism to evict the
cache whenever the Backend settings affecting the conversion, that
is default font and resolution, change.
In order to implement the eviction we can use a "serial"; the
Backend will have an internal serial field which we retrieve and
put inside the ClutterUnits structure (we split one of the two
64 bit padding fields into two 32 bit fields to maintain ABI); every
time we call clutter_units_to_pixels() we compare the units serial
with that of the Backend; if they match and pixels_set is set to
TRUE then we just return the stored pixels value. If the serials
do not match then we unset the pixels_set flag and recompute the
pixels value.
We can verify this by adding a simple test unit checking that
by changing the resolution of ClutterBackend we get different
pixel values for 1 em.
http://bugzilla.openedhand.com/show_bug.cgi?id=1843
Current parsing of units has a number of shortcomings:
* a number followed by trailing space (without any unit specified) was
not recognized,
* "5 emeralds" was parsed as 5em,
* the way we parse the digits after the separator makes us lose
precision for no good reason (5.0 is parsed as 5.00010014...f which
makes g_assert_cmpfloat() fail)
Let's define a stricter grammar we can recognize and try to do so. The
description is in EBNF form, removing the optional <> which is a pain
when having to write DocBook, and using '' for the terminal symbols.
Last step, add more ClutterUnits unit test to get a better coverage of
the grammar we want to parse.
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Parse #rgb and #rrggbb in addition to forms with the alpha channel
specified. This allows conversion of colour strings from documents such as
CSS where the alpha channel is not specified when using '#' notation.
This patch also adds the relevant conformance test.
This unit verifies that an Actor class will invoke the get_preferred_*
virtual functions unless the caching is in effect; it also verifies
that the cached values are correctly evicted.
The size requisition and allocation mechanisms should be thoroughly
tested to avoid unwanted regressions.
For starters, we can test the explicit size setting and the side
effects of calling clutter_actor_set_size().
We need to test that the depth sorting of ClutterGroup works correctly
in case we wish to change the data structure that stores the children,
and do so without changing the default behaviour.
We should follow the convention for boxed types initializers of:
<type_name>_from_<another_type> (boxed, value)
For ClutterUnits as well; so:
clutter_units_pixels -> clutter_units_from_pixels
clutter_units_em -> clutter_units_from_em
...
We should still keep the short-hand version as a macro, though.
The ClutterColor conformance test should have a unit for verifying
the RGB<->HLS conversion code, especially the ability to roundtrip
between the two colorspaces.
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.
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.
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.
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()
cogl_texture_new_from_data lets you specify a source format for the users given
data, and an internal format which the user wants the GPU to see. This unit
test verifies that the users data is premultiplied, un-premultiplied or
left alone for a number of (source format, internal format) pairs.
cogl_texture_set_region allows specifying a source format, and the internal
format is determined from the texture being updated. As above we test
a number of format pairs and check Cogl is converting data correctly.
The test verifies that if the user allows COGL_FORMAT_ANY for the
internal_format then by default Cogl will choose a premultipled format for
RGBA textures.
Note: Currently this only tests cogl_texture_new_from_data and
cogl_texture_set_region, we should also test cogl_texture_new_from_file,
cogl_texture_new_from_bitmap and cogl_texture_new_from_foreign.
This test assumes that the textures will be stored internally with exactly
the color given so that specific texture combining arithmetic can be
tested. Using COGL_PIXEL_FORMAT_ANY allows Cogl to internally premultiply
the textures, so we have to explicitly request an unmultiplied format.
The changes in the master clock and the repaint cycle have been
changed, and broke the way the test for the Text actor cache of
PangoLayouts forces a redraw.
We have to call clutter_actor_paint() on the Stage embedding the
Text actor we want to test; this is kinda fugly because if the
Layout has changed it will end up causing a reallocation cycle
in the middle of the Text actor paint. Since it's a test case,
and since forcing redraws is a bit of a hack as well, we can
close both our eyes on that.
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.
The master clock now works fine whether or not there are any stages,
so in the timeline conformance tests don't need to set up their
own times.
Set CLUTTER_VBLANK=none for the conformance tests, which in addition
to removing an test-environment dependency, will result in the ticking
for timeline tests being throttled to the default frame rate.
http://bugzilla.openedhand.com/show_bug.cgi?id=1637
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
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>
The test was quiting after the 2nd frame but it should be the third frame because
the test doesn't actually check results until the third frame due to the workaround
for drivers with broken glReadPixels.
(When first written the test would have been verified with the
clutter_main_quit() commented out which gives visual feedback of what the
test does, so the off by one would have snuck in just before uncommenting
and pushing.)
The texture filters are now a property of the material layer rather
than the texture object. Whenever a texture is painted with a material
it sets the filters on all of the GL textures in the Cogl texture. The
filter is cached so that it won't be changed unnecessarily.
The automatic mipmap generation has changed so that the mipmaps are
only generated when the texture is painted instead of every time the
data changes. Changing the texture sets a flag to mark that the
mipmaps are dirty. This works better if the FBO extension is available
because we can use glGenerateMipmap. If the extension is not available
it will temporarily enable automatic mipmap generation and reupload
the first pixel of each slice. This requires tracking the data for the
first pixel.
The COGL_TEXTURE_AUTO_MIPMAP flag has been replaced with
COGL_TEXTURE_NO_AUTO_MIPMAP so that it will default to
auto-mipmapping. The mipmap generation is now effectively free if you
are not using a mipmap filter mode so you would only want to disable
it if you had some special reason to generate your own mipmaps.
ClutterTexture no longer has to store its own copy of the filter
mode. Instead it stores it in the material and the property is
directly set and read from that. This fixes problems with the filters
getting out of sync when a cogl handle is set on the texture
directly. It also avoids the mess of having to rerealize the texture
if the filter quality changes to HIGH because Cogl will take of
generating the mipmaps if needed.
Units as they have been implemented since Clutter 0.4 have always been
misdefined as "logical distance unit", while they were just pixels with
fractionary bits.
Units should be reworked to be opaque structures to hold a value and
its unit type, that can be then converted into pixels when Clutter needs
to paint or compute size requisitions and perform allocations.
The previous API should be completely removed to avoid collisions, and
a new type:
ClutterUnits
should be added; the ability to install GObject properties using
ClutterUnits should be maintained.
Timelines no longer work in terms of a frame rate and a number of
frames but instead just have a duration in milliseconds. This better
matches the working of the master clock where if any timelines are
running it will redraw as fast as possible rather than limiting to the
lowest rated timeline.
Most applications will just create animations and expect them to
finish in a certain amount of time without caring about how many
frames are drawn. If a frame is going to be drawn it might as well
update all of the animations to some fraction of the total animation
rather than rounding to the nearest whole frame.
The 'frame_num' parameter of the new-frame signal is now 'msecs' which
is a number of milliseconds progressed along the
timeline. Applications should use clutter_timeline_get_progress
instead of the frame number.
Markers can now only be attached at a time value. The position is
stored in milliseconds rather than at a frame number.
test-timeline-smoothness and test-timeline-dup-frames have been
removed because they no longer make sense.
All the underlying implementation and the public entry points have
been switched to floats; the only missing bits are the Actor properties
that deal with positioning and sizing.
This usually means a major pain when dealing with GValues and varargs
functions. While GValue will warn you when dealing with the wrong
conversions, varags will simply die an horrible (and hard to debug)
death via segfault. Nothing much to do here, except warn people in the
release notes and hope for the best.
cogl_enable_depth_test and cogl_enable_backface_culling have been renamed
and now have corresponding getters, the new functions are:
cogl_set_depth_test_enabled
cogl_get_depth_test_enabled
cogl_set_backface_culling_enabled
cogl_get_backface_culling_enabled
Setting up layer combine functions and blend modes is very awkward to do
programatically. This adds a parser for string based descriptions which are
more consise and readable.
E.g. a material layer combine function could now be given as:
"RGBA = ADD (TEXTURE[A], PREVIOUS[RGB])"
or
"RGB = REPLACE (PREVIOUS)"
"A = MODULATE (PREVIOUS, TEXTURE)"
The simple syntax and grammar are only designed to expose standard fixed
function hardware, more advanced combining must be done with shaders.
This includes standalone documentation of blend strings covering the aspects
that are common to blending and texture combining, and adds documentation
with examples specific to the new cogl_material_set_blend() and
cogl_material_layer_set_combine() functions.
Note: The hope is to remove the now redundant bits of the material API
before 1.0
The CoglTexture constructors expose the "max-waste" argument for
controlling the maximum amount of wasted areas for slicing or,
if set to -1, disables slicing.
Slicing is really relevant only for large images that are never
repeated, so it's a useful feature only in controlled use cases.
Specifying the amount of wasted area is, on the other hand, just
a way to mess up this feature; 99% the times, you either pull this
number out of thin air, hoping it's right, or you try to do the
right thing and you choose the wrong number anyway.
Instead, we can use the CoglTextureFlags to control whether the
texture should not be sliced (useful for Clutter-GST and for the
texture-from-pixmap actors) and provide a reasonable value for
enabling the slicing ourself. At some point, we might even
provide a way to change the default at compile time or at run time,
for particular platforms.
Since max_waste is gone, the :tile-waste property of ClutterTexture
becomes read-only, and it proxies the cogl_texture_get_max_waste()
function.
Inside Clutter, the only cases where the max_waste argument was
not set to -1 are in the Pango glyph cache (which is a POT texture
anyway) and inside the test cases where we want to force slicing;
for the latter we can create larger textures that will be bigger than
the threshold we set.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Signed-off-by: Neil Roberts <neil@linux.intel.com>
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
The units in the Timeline test suite just rely on the timeline
being a timeout automatically advanced by the main loop. This
is not the case anymore, since the merge of the master-clock.
To make the test units work again we need to "emulate" the master
clock without effectively having a stage to redraw; we do this
by creating a frame source and manually advancing the timelines
we create for test purposes, using the advance_msecs() "protected"
method.
Setting the wrap mode on the PangoLayout seems to have disappeared
during the text-actor-layout-height branch merge so this brings it
back. The test for this in test-text-cache no longer needs to be
disabled.
We also shouldn't set the width on the layout if there is no wrapping
or ellipsizing because otherwise it implicitly enables wrapping. This
only matters if the actor gets allocated smaller than its natural
size.
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
