2008-06-23 10:57:36 -04:00
|
|
|
/*
|
2009-04-27 10:48:12 -04:00
|
|
|
* Cogl
|
2008-06-23 10:57:36 -04:00
|
|
|
*
|
2014-02-21 20:28:54 -05:00
|
|
|
* A Low Level GPU Graphics and Utilities API
|
2008-06-23 10:57:36 -04:00
|
|
|
*
|
2010-04-14 14:41:08 -04:00
|
|
|
* Copyright (C) 2007,2008,2009,2010 Intel Corporation.
|
2008-06-23 10:57:36 -04:00
|
|
|
*
|
2014-02-21 20:28:54 -05:00
|
|
|
* Permission is hereby granted, free of charge, to any person
|
|
|
|
* obtaining a copy of this software and associated documentation
|
|
|
|
* files (the "Software"), to deal in the Software without
|
|
|
|
* restriction, including without limitation the rights to use, copy,
|
|
|
|
* modify, merge, publish, distribute, sublicense, and/or sell copies
|
|
|
|
* of the Software, and to permit persons to whom the Software is
|
|
|
|
* furnished to do so, subject to the following conditions:
|
2008-06-23 10:57:36 -04:00
|
|
|
*
|
2014-02-21 20:28:54 -05:00
|
|
|
* The above copyright notice and this permission notice shall be
|
|
|
|
* included in all copies or substantial portions of the Software.
|
2008-06-23 10:57:36 -04:00
|
|
|
*
|
2014-02-21 20:28:54 -05:00
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
|
|
|
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
|
|
|
|
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
|
|
|
|
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
|
|
|
|
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
|
|
* SOFTWARE.
|
2010-03-01 07:56:10 -05:00
|
|
|
*
|
|
|
|
*
|
2008-06-23 10:57:36 -04:00
|
|
|
*/
|
|
|
|
|
|
|
|
#ifdef HAVE_CONFIG_H
|
|
|
|
#include "config.h"
|
|
|
|
#endif
|
|
|
|
|
2008-12-04 08:45:09 -05:00
|
|
|
#include <string.h>
|
2009-11-04 14:42:17 -05:00
|
|
|
#include <math.h>
|
2009-05-08 11:32:01 -04:00
|
|
|
|
|
|
|
#include <glib.h>
|
|
|
|
|
2008-06-23 10:57:36 -04:00
|
|
|
#include "cogl-clip-stack.h"
|
2008-12-04 08:45:09 -05:00
|
|
|
#include "cogl-primitives.h"
|
2010-11-04 18:25:52 -04:00
|
|
|
#include "cogl-context-private.h"
|
2009-11-26 14:06:35 -05:00
|
|
|
#include "cogl-framebuffer-private.h"
|
2010-02-10 13:18:30 -05:00
|
|
|
#include "cogl-journal-private.h"
|
2010-02-17 10:58:32 -05:00
|
|
|
#include "cogl-util.h"
|
2010-04-22 09:33:47 -04:00
|
|
|
#include "cogl-matrix-private.h"
|
2010-11-02 13:15:06 -04:00
|
|
|
#include "cogl-primitives-private.h"
|
2011-09-14 07:17:09 -04:00
|
|
|
#include "cogl-private.h"
|
2011-10-01 12:55:41 -04:00
|
|
|
#include "cogl-pipeline-opengl-private.h"
|
2011-10-03 09:39:05 -04:00
|
|
|
#include "cogl-attribute-private.h"
|
|
|
|
#include "cogl-primitive-private.h"
|
2012-02-17 16:46:39 -05:00
|
|
|
#include "cogl1-context.h"
|
2012-02-17 20:19:17 -05:00
|
|
|
#include "cogl-offscreen.h"
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
#include "cogl-matrix-stack.h"
|
2008-06-23 10:57:36 -04:00
|
|
|
|
2011-10-01 12:55:41 -04:00
|
|
|
|
2009-11-04 14:42:17 -05:00
|
|
|
|
Switch use of primitive glib types to c99 equivalents
The coding style has for a long time said to avoid using redundant glib
data types such as gint or gchar etc because we feel that they make the
code look unnecessarily foreign to developers coming from outside of the
Gnome developer community.
Note: When we tried to find the historical rationale for the types we
just found that they were apparently only added for consistent syntax
highlighting which didn't seem that compelling.
Up until now we have been continuing to use some of the platform
specific type such as gint{8,16,32,64} and gsize but this patch switches
us over to using the standard c99 equivalents instead so we can further
ensure that our code looks familiar to the widest range of C developers
who might potentially contribute to Cogl.
So instead of using the gint{8,16,32,64} and guint{8,16,32,64} types this
switches all Cogl code to instead use the int{8,16,32,64}_t and
uint{8,16,32,64}_t c99 types instead.
Instead of gsize we now use size_t
For now we are not going to use the c99 _Bool type and instead we have
introduced a new CoglBool type to use instead of gboolean.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit 5967dad2400d32ca6319cef6cb572e81bf2c15f0)
2012-04-16 16:56:40 -04:00
|
|
|
static void *
|
2010-04-14 08:17:26 -04:00
|
|
|
_cogl_clip_stack_push_entry (CoglClipStack *clip_stack,
|
|
|
|
size_t size,
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStackType type)
|
2010-04-14 08:17:26 -04:00
|
|
|
{
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *entry = g_slice_alloc (size);
|
2010-04-14 08:17:26 -04:00
|
|
|
|
|
|
|
/* The new entry starts with a ref count of 1 because the stack
|
|
|
|
holds a reference to it as it is the top entry */
|
|
|
|
entry->ref_count = 1;
|
|
|
|
entry->type = type;
|
2010-11-01 15:52:45 -04:00
|
|
|
entry->parent = clip_stack;
|
2010-04-14 08:17:26 -04:00
|
|
|
|
|
|
|
/* We don't need to take a reference to the parent from the entry
|
2010-11-01 15:52:45 -04:00
|
|
|
because the we are stealing the ref in the new stack top */
|
2010-04-14 08:17:26 -04:00
|
|
|
|
|
|
|
return entry;
|
|
|
|
}
|
|
|
|
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
static void
|
|
|
|
get_transformed_corners (float x_1,
|
|
|
|
float y_1,
|
|
|
|
float x_2,
|
|
|
|
float y_2,
|
|
|
|
CoglMatrix *modelview,
|
|
|
|
CoglMatrix *projection,
|
|
|
|
const float *viewport,
|
|
|
|
float *transformed_corners)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
transformed_corners[0] = x_1;
|
|
|
|
transformed_corners[1] = y_1;
|
|
|
|
transformed_corners[2] = x_2;
|
|
|
|
transformed_corners[3] = y_1;
|
|
|
|
transformed_corners[4] = x_2;
|
|
|
|
transformed_corners[5] = y_2;
|
|
|
|
transformed_corners[6] = x_1;
|
|
|
|
transformed_corners[7] = y_2;
|
|
|
|
|
|
|
|
|
|
|
|
/* Project the coordinates to window space coordinates */
|
|
|
|
for (i = 0; i < 4; i++)
|
|
|
|
{
|
|
|
|
float *v = transformed_corners + i * 2;
|
|
|
|
_cogl_transform_point (modelview, projection, viewport, v, v + 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-04-22 09:33:47 -04:00
|
|
|
/* Sets the window-space bounds of the entry based on the projected
|
|
|
|
coordinates of the given rectangle */
|
|
|
|
static void
|
2010-11-01 15:52:45 -04:00
|
|
|
_cogl_clip_stack_entry_set_bounds (CoglClipStack *entry,
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
float *transformed_corners)
|
2010-04-22 09:33:47 -04:00
|
|
|
{
|
|
|
|
float min_x = G_MAXFLOAT, min_y = G_MAXFLOAT;
|
|
|
|
float max_x = -G_MAXFLOAT, max_y = -G_MAXFLOAT;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < 4; i++)
|
|
|
|
{
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
float *v = transformed_corners + i * 2;
|
2010-04-22 09:33:47 -04:00
|
|
|
|
|
|
|
if (v[0] > max_x)
|
|
|
|
max_x = v[0];
|
|
|
|
if (v[0] < min_x)
|
|
|
|
min_x = v[0];
|
|
|
|
if (v[1] > max_y)
|
|
|
|
max_y = v[1];
|
|
|
|
if (v[1] < min_y)
|
|
|
|
min_y = v[1];
|
|
|
|
}
|
|
|
|
|
|
|
|
entry->bounds_x0 = floorf (min_x);
|
|
|
|
entry->bounds_x1 = ceilf (max_x);
|
|
|
|
entry->bounds_y0 = floorf (min_y);
|
|
|
|
entry->bounds_y1 = ceilf (max_y);
|
|
|
|
}
|
|
|
|
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *
|
2010-05-27 20:01:28 -04:00
|
|
|
_cogl_clip_stack_push_window_rectangle (CoglClipStack *stack,
|
2010-04-14 14:41:08 -04:00
|
|
|
int x_offset,
|
|
|
|
int y_offset,
|
|
|
|
int width,
|
|
|
|
int height)
|
2009-05-08 11:32:01 -04:00
|
|
|
{
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *entry;
|
2009-05-08 11:32:01 -04:00
|
|
|
|
2010-04-14 08:17:26 -04:00
|
|
|
entry = _cogl_clip_stack_push_entry (stack,
|
2010-11-01 15:52:45 -04:00
|
|
|
sizeof (CoglClipStackWindowRect),
|
2010-04-14 08:17:26 -04:00
|
|
|
COGL_CLIP_STACK_WINDOW_RECT);
|
2009-05-08 11:32:01 -04:00
|
|
|
|
2010-04-22 09:33:47 -04:00
|
|
|
entry->bounds_x0 = x_offset;
|
|
|
|
entry->bounds_x1 = x_offset + width;
|
|
|
|
entry->bounds_y0 = y_offset;
|
|
|
|
entry->bounds_y1 = y_offset + height;
|
2010-11-01 15:52:45 -04:00
|
|
|
|
|
|
|
return entry;
|
2009-05-08 11:32:01 -04:00
|
|
|
}
|
|
|
|
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *
|
2010-05-27 20:01:28 -04:00
|
|
|
_cogl_clip_stack_push_rectangle (CoglClipStack *stack,
|
2010-04-14 14:41:08 -04:00
|
|
|
float x_1,
|
|
|
|
float y_1,
|
2009-11-04 15:17:56 -05:00
|
|
|
float x_2,
|
2010-04-14 14:41:08 -04:00
|
|
|
float y_2,
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
CoglMatrixEntry *modelview_entry,
|
|
|
|
CoglMatrixEntry *projection_entry,
|
|
|
|
const float *viewport)
|
2008-06-23 10:57:36 -04:00
|
|
|
{
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStackRect *entry;
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
CoglMatrix modelview;
|
|
|
|
CoglMatrix projection;
|
|
|
|
CoglMatrix modelview_projection;
|
|
|
|
|
|
|
|
/* Corners of the given rectangle in an clockwise order:
|
|
|
|
* (0, 1) (2, 3)
|
|
|
|
*
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* (6, 7) (4, 5)
|
|
|
|
*/
|
|
|
|
float rect[] = {
|
|
|
|
x_1, y_1,
|
|
|
|
x_2, y_1,
|
|
|
|
x_2, y_2,
|
|
|
|
x_1, y_2
|
|
|
|
};
|
2008-12-04 08:45:09 -05:00
|
|
|
|
2008-06-23 10:57:36 -04:00
|
|
|
/* Make a new entry */
|
2010-04-14 08:17:26 -04:00
|
|
|
entry = _cogl_clip_stack_push_entry (stack,
|
2010-11-01 15:52:45 -04:00
|
|
|
sizeof (CoglClipStackRect),
|
2010-04-14 08:17:26 -04:00
|
|
|
COGL_CLIP_STACK_RECT);
|
|
|
|
|
2009-11-04 15:17:56 -05:00
|
|
|
entry->x0 = x_1;
|
|
|
|
entry->y0 = y_1;
|
|
|
|
entry->x1 = x_2;
|
|
|
|
entry->y1 = y_2;
|
2008-06-23 10:57:36 -04:00
|
|
|
|
2012-11-20 12:08:43 -05:00
|
|
|
entry->matrix_entry = cogl_matrix_entry_ref (modelview_entry);
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
|
2012-11-20 12:08:43 -05:00
|
|
|
cogl_matrix_entry_get (modelview_entry, &modelview);
|
|
|
|
cogl_matrix_entry_get (projection_entry, &projection);
|
2010-04-22 09:33:47 -04:00
|
|
|
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
cogl_matrix_multiply (&modelview_projection,
|
|
|
|
&projection,
|
|
|
|
&modelview);
|
|
|
|
|
|
|
|
/* Technically we could avoid the viewport transform at this point
|
|
|
|
* if we want to make this a bit faster. */
|
|
|
|
_cogl_transform_point (&modelview, &projection, viewport, &rect[0], &rect[1]);
|
|
|
|
_cogl_transform_point (&modelview, &projection, viewport, &rect[2], &rect[3]);
|
|
|
|
_cogl_transform_point (&modelview, &projection, viewport, &rect[4], &rect[5]);
|
|
|
|
_cogl_transform_point (&modelview, &projection, viewport, &rect[6], &rect[7]);
|
|
|
|
|
|
|
|
/* If the fully transformed rectangle isn't still axis aligned we
|
|
|
|
* can't handle it using a scissor.
|
2010-11-09 07:01:04 -05:00
|
|
|
*
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
* We don't use an epsilon here since we only really aim to catch
|
|
|
|
* simple cases where the transform doesn't leave the rectangle screen
|
|
|
|
* aligned and don't mind some false positives.
|
2010-11-09 07:01:04 -05:00
|
|
|
*/
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
if (rect[0] != rect[6] ||
|
|
|
|
rect[1] != rect[3] ||
|
|
|
|
rect[2] != rect[4] ||
|
|
|
|
rect[7] != rect[5])
|
2010-11-09 07:01:04 -05:00
|
|
|
{
|
|
|
|
entry->can_be_scissor = FALSE;
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
|
2010-11-09 07:01:04 -05:00
|
|
|
_cogl_clip_stack_entry_set_bounds ((CoglClipStack *) entry,
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
rect);
|
2010-11-09 07:01:04 -05:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
CoglClipStack *base_entry = (CoglClipStack *) entry;
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
x_1 = rect[0];
|
|
|
|
y_1 = rect[1];
|
|
|
|
x_2 = rect[4];
|
|
|
|
y_2 = rect[5];
|
2010-11-09 07:01:04 -05:00
|
|
|
|
|
|
|
/* Consider that the modelview matrix may flip the rectangle
|
|
|
|
* along the x or y axis... */
|
|
|
|
#define SWAP(A,B) do { float tmp = B; B = A; A = tmp; } while (0)
|
|
|
|
if (x_1 > x_2)
|
|
|
|
SWAP (x_1, x_2);
|
|
|
|
if (y_1 > y_2)
|
|
|
|
SWAP (y_1, y_2);
|
|
|
|
#undef SWAP
|
|
|
|
|
|
|
|
base_entry->bounds_x0 = COGL_UTIL_NEARBYINT (x_1);
|
|
|
|
base_entry->bounds_y0 = COGL_UTIL_NEARBYINT (y_1);
|
|
|
|
base_entry->bounds_x1 = COGL_UTIL_NEARBYINT (x_2);
|
|
|
|
base_entry->bounds_y1 = COGL_UTIL_NEARBYINT (y_2);
|
|
|
|
entry->can_be_scissor = TRUE;
|
|
|
|
}
|
2010-11-01 15:52:45 -04:00
|
|
|
|
|
|
|
return (CoglClipStack *) entry;
|
2009-11-04 15:17:56 -05:00
|
|
|
}
|
|
|
|
|
2011-10-03 09:39:05 -04:00
|
|
|
CoglClipStack *
|
|
|
|
_cogl_clip_stack_push_primitive (CoglClipStack *stack,
|
|
|
|
CoglPrimitive *primitive,
|
|
|
|
float bounds_x1,
|
|
|
|
float bounds_y1,
|
|
|
|
float bounds_x2,
|
|
|
|
float bounds_y2,
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
CoglMatrixEntry *modelview_entry,
|
|
|
|
CoglMatrixEntry *projection_entry,
|
|
|
|
const float *viewport)
|
2011-10-03 09:39:05 -04:00
|
|
|
{
|
|
|
|
CoglClipStackPrimitive *entry;
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
CoglMatrix modelview;
|
|
|
|
CoglMatrix projection;
|
|
|
|
float transformed_corners[8];
|
2011-10-03 09:39:05 -04:00
|
|
|
|
|
|
|
entry = _cogl_clip_stack_push_entry (stack,
|
|
|
|
sizeof (CoglClipStackPrimitive),
|
|
|
|
COGL_CLIP_STACK_PRIMITIVE);
|
|
|
|
|
|
|
|
entry->primitive = cogl_object_ref (primitive);
|
|
|
|
|
2012-11-20 12:08:43 -05:00
|
|
|
entry->matrix_entry = cogl_matrix_entry_ref (modelview_entry);
|
2011-10-03 09:39:05 -04:00
|
|
|
|
|
|
|
entry->bounds_x1 = bounds_x1;
|
|
|
|
entry->bounds_y1 = bounds_y1;
|
|
|
|
entry->bounds_x2 = bounds_x2;
|
|
|
|
entry->bounds_y2 = bounds_y2;
|
|
|
|
|
2012-11-20 12:08:43 -05:00
|
|
|
cogl_matrix_entry_get (modelview_entry, &modelview);
|
2013-12-02 13:18:39 -05:00
|
|
|
cogl_matrix_entry_get (projection_entry, &projection);
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
|
|
|
|
get_transformed_corners (bounds_x1, bounds_y1, bounds_x2, bounds_y2,
|
|
|
|
&modelview,
|
|
|
|
&projection,
|
|
|
|
viewport,
|
|
|
|
transformed_corners);
|
|
|
|
|
2011-10-03 09:39:05 -04:00
|
|
|
/* NB: this is referring to the bounds in window coordinates as opposed
|
|
|
|
* to the bounds above in primitive local coordinates. */
|
|
|
|
_cogl_clip_stack_entry_set_bounds ((CoglClipStack *) entry,
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
transformed_corners);
|
2011-10-03 09:39:05 -04:00
|
|
|
|
|
|
|
return (CoglClipStack *) entry;
|
|
|
|
}
|
|
|
|
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *
|
|
|
|
_cogl_clip_stack_ref (CoglClipStack *entry)
|
|
|
|
{
|
|
|
|
/* A NULL pointer is considered a valid stack so we should accept
|
|
|
|
that as an argument */
|
|
|
|
if (entry)
|
|
|
|
entry->ref_count++;
|
|
|
|
|
|
|
|
return entry;
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
_cogl_clip_stack_unref (CoglClipStack *entry)
|
2010-04-14 08:17:26 -04:00
|
|
|
{
|
|
|
|
/* Unref all of the entries until we hit the root of the list or the
|
|
|
|
entry still has a remaining reference */
|
|
|
|
while (entry && --entry->ref_count <= 0)
|
|
|
|
{
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *parent = entry->parent;
|
2010-04-14 08:17:26 -04:00
|
|
|
|
|
|
|
switch (entry->type)
|
|
|
|
{
|
|
|
|
case COGL_CLIP_STACK_RECT:
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
{
|
|
|
|
CoglClipStackRect *rect = (CoglClipStackRect *) entry;
|
2012-11-20 12:08:43 -05:00
|
|
|
cogl_matrix_entry_unref (rect->matrix_entry);
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
g_slice_free1 (sizeof (CoglClipStackRect), entry);
|
|
|
|
break;
|
|
|
|
}
|
2010-04-14 08:17:26 -04:00
|
|
|
case COGL_CLIP_STACK_WINDOW_RECT:
|
2010-11-01 15:52:45 -04:00
|
|
|
g_slice_free1 (sizeof (CoglClipStackWindowRect), entry);
|
2010-04-14 08:17:26 -04:00
|
|
|
break;
|
2011-10-03 09:39:05 -04:00
|
|
|
case COGL_CLIP_STACK_PRIMITIVE:
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
{
|
|
|
|
CoglClipStackPrimitive *primitive_entry =
|
|
|
|
(CoglClipStackPrimitive *) entry;
|
2012-11-20 12:08:43 -05:00
|
|
|
cogl_matrix_entry_unref (primitive_entry->matrix_entry);
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
cogl_object_unref (primitive_entry->primitive);
|
|
|
|
g_slice_free1 (sizeof (CoglClipStackPrimitive), entry);
|
|
|
|
break;
|
|
|
|
}
|
2010-04-14 08:17:26 -04:00
|
|
|
default:
|
|
|
|
g_assert_not_reached ();
|
|
|
|
}
|
|
|
|
|
|
|
|
entry = parent;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *
|
2010-05-27 20:01:28 -04:00
|
|
|
_cogl_clip_stack_pop (CoglClipStack *stack)
|
2008-06-23 10:57:36 -04:00
|
|
|
{
|
2010-11-01 15:52:45 -04:00
|
|
|
CoglClipStack *new_top;
|
2008-12-04 08:45:09 -05:00
|
|
|
|
2011-10-13 17:34:30 -04:00
|
|
|
_COGL_RETURN_VAL_IF_FAIL (stack != NULL, NULL);
|
2008-12-04 08:45:09 -05:00
|
|
|
|
2010-04-14 08:17:26 -04:00
|
|
|
/* To pop we are moving the top of the stack to the old top's parent
|
|
|
|
node. The stack always needs to have a reference to the top entry
|
|
|
|
so we must take a reference to the new top. The stack would have
|
|
|
|
previously had a reference to the old top so we need to decrease
|
|
|
|
the ref count on that. We need to ref the new head first in case
|
|
|
|
this stack was the only thing referencing the old top. In that
|
|
|
|
case the call to _cogl_clip_stack_entry_unref will unref the
|
|
|
|
parent. */
|
2010-11-01 15:52:45 -04:00
|
|
|
new_top = stack->parent;
|
|
|
|
|
|
|
|
_cogl_clip_stack_ref (new_top);
|
|
|
|
|
|
|
|
_cogl_clip_stack_unref (stack);
|
|
|
|
|
|
|
|
return new_top;
|
2008-06-23 10:57:36 -04:00
|
|
|
}
|
|
|
|
|
2011-01-12 14:13:45 -05:00
|
|
|
void
|
|
|
|
_cogl_clip_stack_get_bounds (CoglClipStack *stack,
|
|
|
|
int *scissor_x0,
|
|
|
|
int *scissor_y0,
|
|
|
|
int *scissor_x1,
|
|
|
|
int *scissor_y1)
|
|
|
|
{
|
|
|
|
CoglClipStack *entry;
|
|
|
|
|
|
|
|
*scissor_x0 = 0;
|
|
|
|
*scissor_y0 = 0;
|
|
|
|
*scissor_x1 = G_MAXINT;
|
|
|
|
*scissor_y1 = G_MAXINT;
|
|
|
|
|
|
|
|
for (entry = stack; entry; entry = entry->parent)
|
|
|
|
{
|
|
|
|
/* Get the intersection of the current scissor and the bounding
|
|
|
|
box of this clip */
|
2012-10-02 06:44:00 -04:00
|
|
|
_cogl_util_scissor_intersect (entry->bounds_x0,
|
|
|
|
entry->bounds_y0,
|
|
|
|
entry->bounds_x1,
|
|
|
|
entry->bounds_y1,
|
|
|
|
scissor_x0,
|
|
|
|
scissor_y0,
|
|
|
|
scissor_x1,
|
|
|
|
scissor_y1);
|
2011-01-12 14:13:45 -05:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-06-23 10:57:36 -04:00
|
|
|
void
|
2011-08-03 12:06:10 -04:00
|
|
|
_cogl_clip_stack_flush (CoglClipStack *stack,
|
|
|
|
CoglFramebuffer *framebuffer)
|
2008-06-23 10:57:36 -04:00
|
|
|
{
|
Re-design the matrix stack using a graph of ops
This re-designs the matrix stack so we now keep track of each separate
operation such as rotating, scaling, translating and multiplying as
immutable, ref-counted nodes in a graph.
Being a "graph" here means that different transformations composed of
a sequence of linked operation nodes may share nodes.
The first node in a matrix-stack is always a LOAD_IDENTITY operation.
As an example consider if an application where to draw three rectangles
A, B and C something like this:
cogl_framebuffer_scale (fb, 2, 2, 2);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_translate (fb, 10, 0, 0);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_rotate (fb, 45, 0, 0, 1);
cogl_framebuffer_draw_rectangle (...); /* A */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_draw_rectangle (...); /* B */
cogl_framebuffer_pop_matrix(fb);
cogl_framebuffer_push_matrix(fb);
cogl_framebuffer_set_modelview_matrix (fb, &mv);
cogl_framebuffer_draw_rectangle (...); /* C */
cogl_framebuffer_pop_matrix(fb);
That would result in a graph of nodes like this:
LOAD_IDENTITY
|
SCALE
/ \
SAVE LOAD
| |
TRANSLATE RECTANGLE(C)
| \
SAVE RECTANGLE(B)
|
ROTATE
|
RECTANGLE(A)
Each push adds a SAVE operation which serves as a marker to rewind too
when a corresponding pop is issued and also each SAVE node may also
store a cached matrix representing the composition of all its ancestor
nodes. This means if we repeatedly need to resolve a real CoglMatrix
for a given node then we don't need to repeat the composition.
Some advantages of this design are:
- A single pointer to any node in the graph can now represent a
complete, immutable transformation that can be logged for example
into a journal. Previously we were storing a full CoglMatrix in
each journal entry which is 16 floats for the matrix itself as well
as space for flags and another 16 floats for possibly storing a
cache of the inverse. This means that we significantly reduce
the size of the journal when drawing lots of primitives and we also
avoid copying over 128 bytes per entry.
- It becomes much cheaper to check for equality. In cases where some
(unlikely) false negatives are allowed simply comparing the pointers
of two matrix stack graph entries is enough. Previously we would use
memcmp() to compare matrices.
- It becomes easier to do comparisons of transformations. By looking
for the common ancestry between nodes we can determine the operations
that differentiate the transforms and use those to gain a high level
understanding of the differences. For example we use this in the
journal to be able to efficiently determine when two rectangle
transforms only differ by some translation so that we can perform
software clipping.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
(cherry picked from commit f75aee93f6b293ca7a7babbd8fcc326ee6bf7aef)
2012-02-20 10:59:48 -05:00
|
|
|
CoglContext *ctx = framebuffer->context;
|
2008-12-04 08:45:09 -05:00
|
|
|
|
2012-09-19 15:37:32 -04:00
|
|
|
ctx->driver_vtable->clip_stack_flush (stack, framebuffer);
|
2010-11-02 10:28:12 -04:00
|
|
|
}
|