archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity
This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
3316853323226afe7003a8638fa09600d9ba028d
blender-archive/source/blender/compositor/intern/COM_Debug.cpp

426 lines
15 KiB
C++
Raw Normal View History

Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
/*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
Cleanup: conform headers to have license first Also remove doxy comments for licenses and add missing GPL header.
2019-02-18 07:21:50 +11:00
*
* Copyright 2013, Blender Foundation.
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
*/
#include "COM_Debug.h"
#ifdef COM_DEBUG
#include <typeinfo>
#include <map>
#include <vector>
extern "C" {
Compositor: Fix compilation error with debug enabled
2017-10-02 15:12:04 +05:00
#include "BLI_sys_types.h"
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
#include "BLI_fileops.h"
#include "BLI_path_util.h"
#include "BLI_string.h"
Compositor: Fix compilation error with debug enabled
2017-10-02 15:12:04 +05:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
#include "DNA_node_types.h"
Compositor: Fix compilation error with COM_DEBUG enabled
2015-03-05 20:12:28 +05:00
#include "BKE_appdir.h"
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
#include "BKE_node.h"
}
#include "COM_Node.h"
#include "COM_ExecutionSystem.h"
#include "COM_ExecutionGroup.h"
#include "COM_ReadBufferOperation.h"
#include "COM_ViewerOperation.h"
#include "COM_WriteBufferOperation.h"
int DebugInfo::m_file_index = 0;
DebugInfo::NodeNameMap DebugInfo::m_node_names;
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
DebugInfo::OpNameMap DebugInfo::m_op_names;
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
std::string DebugInfo::m_current_node_name;
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
std::string DebugInfo::m_current_op_name;
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
DebugInfo::GroupStateMap DebugInfo::m_group_states;
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
std::string DebugInfo::node_name(const Node *node)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
NodeNameMap::const_iterator it = m_node_names.find(node);
if (it != m_node_names.end())
return it->second;
else
return "";
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
std::string DebugInfo::operation_name(const NodeOperation *op)
{
OpNameMap::const_iterator it = m_op_names.find(op);
if (it != m_op_names.end())
return it->second;
else
return "";
}
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
void DebugInfo::convert_started()
{
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_op_names.clear();
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::execute_started(const ExecutionSystem *system)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
m_file_index = 1;
m_group_states.clear();
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (ExecutionSystem::Groups::const_iterator it = system->m_groups.begin(); it != system->m_groups.end(); ++it)
m_group_states[*it] = EG_WAIT;
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::node_added(const Node *node)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
m_node_names[node] = std::string(node->getbNode() ? node->getbNode()->name : "");
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::node_to_operations(const Node *node)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
m_current_node_name = m_node_names[node];
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::operation_added(const NodeOperation *operation)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_op_names[operation] = m_current_node_name;
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::operation_read_write_buffer(const NodeOperation *operation)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_current_op_name = m_op_names[operation];
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::execution_group_started(const ExecutionGroup *group)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
m_group_states[group] = EG_RUNNING;
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::execution_group_finished(const ExecutionGroup *group)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
m_group_states[group] = EG_FINISHED;
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
int DebugInfo::graphviz_operation(const ExecutionSystem *system, const NodeOperation *operation, const ExecutionGroup *group, char *str, int maxlen)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
int len = 0;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
std::string fillcolor = "gainsboro";
if (operation->isViewerOperation()) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
const ViewerOperation *viewer = (const ViewerOperation *)operation;
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (viewer->isActiveViewerOutput()) {
fillcolor = "lightskyblue1";
}
else {
fillcolor = "lightskyblue3";
}
}
else if (operation->isOutputOperation(system->getContext().isRendering())) {
fillcolor = "dodgerblue1";
}
else if (operation->isSetOperation()) {
fillcolor = "khaki1";
}
else if (operation->isReadBufferOperation()) {
fillcolor = "darkolivegreen3";
}
else if (operation->isWriteBufferOperation()) {
fillcolor = "darkorange";
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "// OPERATION: %p\r\n", operation);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (group)
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\"O_%p_%p\"", operation, group);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
else
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\"O_%p\"", operation);
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, " [fillcolor=%s,style=filled,shape=record,label=\"{", fillcolor.c_str());
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
int totinputs = operation->getNumberOfInputSockets();
if (totinputs != 0) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "{");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
for (int k = 0; k < totinputs; k++) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperationInput *socket = operation->getInputSocket(k);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (k != 0) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "|");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<IN_%p>", socket);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
switch (socket->getDataType()) {
case COM_DT_VALUE:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Value");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
case COM_DT_VECTOR:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Vector");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
case COM_DT_COLOR:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Color");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
}
}
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "|");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "%s\\n(%s)", m_op_names[operation].c_str(), typeid(*operation).name());
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Compositor: Fix strict compilation warnings when debug is enabled
2017-10-02 15:18:10 +05:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, " (%u,%u)", operation->getWidth(), operation->getHeight());
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
int totoutputs = operation->getNumberOfOutputSockets();
if (totoutputs != 0) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "|");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "{");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
for (int k = 0; k < totoutputs; k++) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperationOutput *socket = operation->getOutputSocket(k);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (k != 0) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "|");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<OUT_%p>", socket);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
switch (socket->getDataType()) {
case COM_DT_VALUE:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Value");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
case COM_DT_VECTOR:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Vector");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
case COM_DT_COLOR:
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Color");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
}
}
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}\"]");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\r\n");
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return len;
}
int DebugInfo::graphviz_legend_color(const char *name, const char *color, char *str, int maxlen)
{
int len = 0;
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<TR><TD>%s</TD><TD BGCOLOR=\"%s\"></TD></TR>\r\n", name, color);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return len;
}
Compositor: Code cleanup, debug mode is now fine with strict flags
2015-03-27 16:47:11 +05:00
int DebugInfo::graphviz_legend_line(const char * /*name*/, const char * /*color*/, const char * /*style*/, char *str, int maxlen)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
/* XXX TODO */
int len = 0;
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return len;
}
Compositor: Code cleanup, debug mode is now fine with strict flags
2015-03-27 16:47:11 +05:00
int DebugInfo::graphviz_legend_group(const char *name, const char *color, const char * /*style*/, char *str, int maxlen)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
int len = 0;
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<TR><TD>%s</TD><TD CELLPADDING=\"4\"><TABLE BORDER=\"1\" CELLBORDER=\"0\" CELLSPACING=\"0\" CELLPADDING=\"0\"><TR><TD BGCOLOR=\"%s\"></TD></TR></TABLE></TD></TR>\r\n", name, color);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return len;
}
int DebugInfo::graphviz_legend(char *str, int maxlen)
{
int len = 0;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "{\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "rank = sink;\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "Legend [shape=none, margin=0, label=<\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, " <TABLE BORDER=\"0\" CELLBORDER=\"1\" CELLSPACING=\"0\" CELLPADDING=\"4\">\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<TR><TD COLSPAN=\"2\"><B>Legend</B></TD></TR>\r\n");
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
len += graphviz_legend_color("NodeOperation", "gainsboro", str + len, maxlen > len ? maxlen - len : 0);
style cleanup
2013-09-18 23:08:23 +00:00
len += graphviz_legend_color("Output", "dodgerblue1", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_color("Viewer", "lightskyblue3", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_color("Active Viewer", "lightskyblue1", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_color("Write Buffer", "darkorange", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_color("Read Buffer", "darkolivegreen3", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_color("Input Value", "khaki1", str + len, maxlen > len ? maxlen - len : 0);
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "<TR><TD></TD></TR>\r\n");
len += graphviz_legend_group("Group Waiting", "white", "dashed", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_group("Group Running", "firebrick1", "solid", str + len, maxlen > len ? maxlen - len : 0);
len += graphviz_legend_group("Group Finished", "chartreuse4", "solid", str + len, maxlen > len ? maxlen - len : 0);
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "</TABLE>\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, ">];\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return len;
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
bool DebugInfo::graphviz_system(const ExecutionSystem *system, char *str, int maxlen)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
char strbuf[64];
int len = 0;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "digraph compositorexecution {\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "ranksep=1.5\r\n");
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "rankdir=LR\r\n");
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "splines=false\r\n");
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
#if 0
for (ExecutionSystem::Operations::const_iterator it = system->m_operations.begin();
Cleanup: indentation
2018-08-30 01:31:20 +10:00
it != system->m_operations.end(); ++it)
{
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperation *op = *it;
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "// OPERATION: %s\r\n", node->getbNode()->typeinfo->ui_name);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
#endif
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
int totops = system->m_operations.size();
int totgroups = system->m_groups.size();
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
std::map<NodeOperation *, std::vector<std::string> > op_groups;
for (int i = 0; i < totgroups; ++i) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
const ExecutionGroup *group = system->m_groups[i];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "// GROUP: %d\r\n", i);
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "subgraph cluster_%d{\r\n", i);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
/* used as a check for executing group */
if (m_group_states[group] == EG_WAIT) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "style=dashed\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
else if (m_group_states[group] == EG_RUNNING) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "style=filled\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "color=black\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "fillcolor=firebrick1\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
else if (m_group_states[group] == EG_FINISHED) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "style=filled\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "color=black\r\n");
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "fillcolor=chartreuse4\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (ExecutionGroup::Operations::const_iterator it = group->m_operations.begin(); it != group->m_operations.end(); ++it) {
NodeOperation *operation = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
sprintf(strbuf, "_%p", group);
op_groups[operation].push_back(std::string(strbuf));
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += graphviz_operation(system, operation, group, str + len, maxlen > len ? maxlen - len : 0);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
// len += snprintf(str+len, maxlen>len ? maxlen-len : 0, "// OUTPUTOPERATION: %p\r\n", group->getOutputOperation());
// len += snprintf(str+len, maxlen>len ? maxlen-len : 0, " O_%p\r\n", group->getOutputOperation());
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
/* operations not included in any group */
for (int j = 0; j < totops; ++j) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperation *operation = system->m_operations[j];
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (op_groups.find(operation) != op_groups.end())
continue;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
op_groups[operation].push_back(std::string(""));
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += graphviz_operation(system, operation, 0, str + len, maxlen > len ? maxlen - len : 0);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
for (int i = 0; i < totops; i++) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperation *operation = system->m_operations[i];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
if (operation->isReadBufferOperation()) {
ReadBufferOperation *read = (ReadBufferOperation *)operation;
WriteBufferOperation *write = read->getMemoryProxy()->getWriteBufferOperation();
std::vector<std::string> &read_groups = op_groups[read];
std::vector<std::string> &write_groups = op_groups[write];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
for (int k = 0; k < write_groups.size(); ++k) {
for (int l = 0; l < read_groups.size(); ++l) {
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\"O_%p%s\" -> \"O_%p%s\" [style=dotted]\r\n", write, write_groups[k].c_str(), read, read_groups[l].c_str());
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
}
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int i = 0; i < totops; i++) {
NodeOperation *op = system->m_operations[i];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (NodeOperation::Inputs::const_iterator it = op->m_inputs.begin(); it != op->m_inputs.end(); ++it) {
NodeOperationInput *to = *it;
NodeOperationOutput *from = to->getLink();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (!from)
continue;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
std::string color;
switch (from->getDataType()) {
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
case COM_DT_VALUE:
Cleanup: spelling grey -> gray
2015-03-19 14:14:48 +11:00
color = "gray";
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
break;
case COM_DT_VECTOR:
color = "blue";
break;
case COM_DT_COLOR:
color = "orange";
break;
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperation *to_op = &to->getOperation();
NodeOperation *from_op = &from->getOperation();
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
std::vector<std::string> &from_groups = op_groups[from_op];
std::vector<std::string> &to_groups = op_groups[to_op];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "// CONNECTION: %p.%p -> %p.%p\r\n",
from_op, from, to_op, to);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
for (int k = 0; k < from_groups.size(); ++k) {
for (int l = 0; l < to_groups.size(); ++l) {
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\"O_%p%s\":\"OUT_%p\":e -> \"O_%p%s\":\"IN_%p\":w",
from_op, from_groups[k].c_str(), from, to_op, to_groups[l].c_str(), to);
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, " [color=%s]", color.c_str());
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "\r\n");
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
}
}
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += graphviz_legend(str + len, maxlen > len ? maxlen - len : 0);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
style cleanup
2013-09-18 23:08:23 +00:00
len += snprintf(str + len, maxlen > len ? maxlen - len : 0, "}\r\n");
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
return (len < maxlen);
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::graphviz(const ExecutionSystem *system)
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
{
char str[1000000];
style cleanup
2013-09-18 23:08:23 +00:00
if (graphviz_system(system, str, sizeof(str) - 1)) {
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
char basename[FILE_MAX];
char filename[FILE_MAX];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
BLI_snprintf(basename, sizeof(basename), "compositor_%d.dot", m_file_index);
Refactor: BLI_path_util (part 2) Use BKE_appdir/tempdir naming prefix for functions extracted from BLI_path_util
2014-11-23 15:54:29 +01:00
BLI_join_dirfile(filename, sizeof(filename), BKE_tempdir_session(), basename);
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
++m_file_index;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
FILE *fp = BLI_fopen(filename, "wb");
fputs(str, fp);
fclose(fp);
}
}
#else
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
std::string DebugInfo::node_name(const Node * /*node*/) { return ""; }
std::string DebugInfo::operation_name(const NodeOperation * /*op*/) { return ""; }
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
void DebugInfo::convert_started() {}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
void DebugInfo::execute_started(const ExecutionSystem * /*system*/) {}
void DebugInfo::node_added(const Node * /*node*/) {}
void DebugInfo::node_to_operations(const Node * /*node*/) {}
void DebugInfo::operation_added(const NodeOperation * /*operation*/) {}
void DebugInfo::operation_read_write_buffer(const NodeOperation * /*operation*/) {}
void DebugInfo::execution_group_started(const ExecutionGroup * /*group*/) {}
void DebugInfo::execution_group_finished(const ExecutionGroup * /*group*/) {}
void DebugInfo::graphviz(const ExecutionSystem * /*system*/) {}
Cleanup and improvements of the compositor debug output. Debug code for graphviz output moved to a dedicated file COM_Debug.h/cpp. The DebugInfo class has only static functions, which are called from a number of places to keep track of what is happening in the compositor. If debugging is disabled these are just inline stubs, so we don't need #ifdefs everywhere and don't get any overhead. The graphviz output is much more useful now. DebugInfo keeps track of node names in a static string map for meaningful names. It uses a number of colors for various special operation classes. ExecutionGroups are indicated in graphviz with clusters. Currently the graphviz .dot files are stored in the BLI_temporary_dir() folder. A separate dot file is generated for each stage of the ExecutionGroup scheduling, this is intended to give some idea of the compositor progress, but could still be improved.
2013-09-13 13:36:47 +00:00
#endif
Copy Permalink