Go to: Synopsis. Return value. Keywords. Flags. Python examples.
evaluationManager([asyncUpdate=boolean], [cycleCluster=string], [downstreamFrom=string], [enabled=boolean], [idleBuild=boolean], [invalidate=boolean], [manipulation=boolean], [mode=string], [nodeTypeGloballySerialize=boolean], [nodeTypeParallel=boolean], [nodeTypeSerialize=boolean], [nodeTypeUntrusted=boolean], [safeMode=boolean], [upstreamFrom=string])
Note: Strings representing object names and arguments must be separated by commas. This is not depicted in the synopsis.
evaluationManager is NOT undoable, queryable, and NOT editable.
Handles turning on and off the evaluation manager method of evaluating the DG.
Query the 'mode' flag to see all available evaluation modes. The special mode
'off' disables the evaluation manager.
The scheduling override flags 'nodeTypeXXX' force certain node types to use
specific scheduling types, even though the node descriptions might indicate
otherwise. Use with caution; certain nodes may not react well to
alternative scheduling types.
Only one scheduling type override will be in force at a time, the most
restrictive one. In order, they are untrusted, globally serialized, locally
serialized, and parallel. The node types will however remember all overrides.
For example, if you set a node type override to be untrusted, then to be
parallel it will continue to use the untrusted override. If you then turn off
the untrusted override, the scheduling will advance to the parallel one.
The actual node scheduling type is always superceded by the overrides. For example, a
serial node will still be considered as parallel if the node type has the
parallel override set, even though 'serial' is a more restrictive scheduling type.
See the 'dbpeek' command 'graph' operation with arguments 'evaluationGraph' and
'scheduling' to see what scheduling type any particular node will end up using
after the hierarchy of overrides and native scheduling types is applied.
| string[] | The names of all evaluation manager modes (querying without flags) |
| string[] | The names of all nodes involved in a cycle cluster with the selected one. |
| boolean | The success of activating of deactivating manipulation (with the 'manipulation' flag) |
| boolean | The manipulation active or inactive state (querying the 'manipulation' flag) |
| boolean | The state of simultaneous Viewport 2 and Evaluation Graph update (querying the 'asyncUpdate' flag) |
| boolean | The success of setting the evaluation manager mode (with the 'mode' flag) |
| boolean | The success of setting the evaluation manager idle refresh build mode (with the 'idleBuild' flag) |
| boolean | Is the idle refresh build mode active? (querying with the 'idleBuild' flag) |
| boolean | The success of setting the node type parallel scheduling mode (with the 'nodeTypeParallel' flag) |
| boolean[] | The parallel scheduling states of specified node types (querying the 'nodeTypeParallel' flag with object(s)) |
| string[] | The names of all node types in parallel scheduling mode (querying the 'nodeTypeParallel' flag alone) |
| boolean | The success of setting the node type serialized mode (with the 'nodeTypeSerialize' flag) |
| boolean[] | The serialized states of specified node types (querying the 'nodeTypeSerialize' flag with object(s)) |
| string[] | The names of all node types in serial scheduling mode (querying the 'nodeTypeSerialize' flag alone) |
| boolean | The success of setting the node type globally serialized mode (with the 'nodeTypeGloballySerialize' flag) |
| boolean[] | The globally serialized states of specified node types (querying the 'nodeTypeGloballySerialize' flag with object(s)) |
| string[] | The names of all node types in globally serialized scheduling mode (querying the 'nodeTypeGloballySerialize' flag alone) |
| boolean | The success of setting the node type untrusted mode (with the 'nodeTypeUntrusted' flag) |
| boolean[] | The untrusted of specified node types (querying the 'nodeTypeUntrusted' flag with object(s)) |
| string[] | The names of all node types in untrusted scheduling mode (querying the 'nodeTypeUntrusted' flag alone) |
| string | The evaluation manager mode (querying with the 'mode' flag) |
| string[] | The names of all nodes immediately downstream/upstream of the named one(s) (with the 'upstreamFrom' or 'downstreamFrom' flags) |
In query mode, return type is based on queried flag.
evaluation, manager, DG, runtime
asyncUpdate, cycleCluster, downstreamFrom, enabled, idleBuild, invalidate, manipulation, mode, nodeTypeGloballySerialize, nodeTypeParallel, nodeTypeSerialize, nodeTypeUntrusted, safeMode, upstreamFrom
| Long name (short name) |
Argument types |
Properties |
asyncUpdate(asu)
|
boolean
|
 
|
|
This flag turns on/off the concurrent update of the Viewport 2.0 database
and the evaluation in the Evaluation Graph. This can improve performance in
those cases when the viewport update takes a long time and is mostly serialized
due to the structure of the Evaluation Graph. It may help performance in those
specific circumstances though you will have to measure it on a case-by-case basis.
|
|
cycleCluster(ccl)
|
string
|
|
|
Returns a list of nodes that are stored together with the given one in
a cycle cluster. The list will be empty when the evaluation mode is not
active or the node is not in a cycle.
|
|
enabled(e)
|
boolean
|
|
|
Valid in query mode only. Checks to see if the evaluation manager is currently enabled.
This is independent of the current mode.
|
|
idleBuild(ib)
|
boolean
|
|
|
This flag sets the rebuild option. If set to true then the evaluation
graph will rebuild on an idle event as soon as it is able to do so. If false then
it will only rebuild when required, for example at the start of playback.
Note: This only applies to the graph attached to the normal context. All
other graphs will be built according to their own rules.
The disadvantage of building on an idle event is that for some workflows that are
changing the graph frequently, or very large graphs, the graph build time
may impact the workflow. The default is to have idleBuild turned on. If
the build time is impacted, this flag can be turned off.
|
|
invalidate(inv)
|
boolean
|
|
|
This flag invalidates the graph. Value is used to control auto rebuilding on idle (false) or forced (true).
This command should be used as a last resort.
|
|
manipulation(man)
|
boolean
|
|
|
This flag is used to activate evaluation manager manipulation support.
|
|
mode(m)
|
string
|
|
|
Changes the current evaluation mode in the evaluation manager. Supported values are
"off", "serial", "serialUncached" and "parallel".
|
|
downstreamFrom(dst)
|
string
|
|
|
Find the DG nodes that are immediately downstream of the named one in
the evaluation graph. Note that the connectivity is via evaluation mode
connections, not DG connections.
In query mode the graph is walked and any nodes downstream of the named
one are returned. The return type is alternating pairs of values that
represent the graph level and the node name, e.g. if you walk downstream
from A in the graph A -> B -> C then the return will be the array of
strings ("0","A","1","B","2","C"). Scripts can deconstruct this
information into something more visually recognizable. Note that cycles
are likely to be present so any such scripts would have to handle them.
In query mode, this flag needs a value.
|
|
nodeTypeGloballySerialize(ntg)
|
boolean
|
|
|
This flag is used only when the evaluation manager is in "parallel" mode
but can be set at anytime. It activates or deactivates the override to force
global serial scheduling for the class name argument(s) in the evaluation manager.
Legal object values are class type names: e.g. "transform", "skinCluster", "mesh".
When queried without specified nodes, it returns the list of nodes with the
global serial scheduling override active.
Scheduling overrides take precedence over all of the node and node type
scheduling rules. Use with caution; certain nodes may not react well to
alternative scheduling types.
|
|
nodeTypeParallel(ntp)
|
boolean
|
|
|
This flag is used only when the evaluation manager is in "parallel" mode
but can be set at anytime. It activates or deactivates the override to force
parallel scheduling for the class name argument(s) in the evaluation manager.
Legal object values are class type names: e.g. "transform", "skinCluster", "mesh".
When queried without specified nodes, it returns the list of nodes with the
parallel scheduling override active.
Scheduling overrides take precedence over all of the node and node type
scheduling rules. Use with caution; certain nodes may not react well to
alternative scheduling types.
|
|
nodeTypeSerialize(nts)
|
boolean
|
|
|
This flag is used only when the evaluation manager is in "parallel" mode
but can be set at anytime. It activates or deactivates the override to force
local serial scheduling for the class name argument(s) in the evaluation manager.
Legal object values are class type names: e.g. "transform", "skinCluster", "mesh".
When queried without specified nodes, it returns the list of nodes with the
local serial scheduling override active.
Scheduling overrides take precedence over all of the node and node type
scheduling rules. Use with caution; certain nodes may not react well to
alternative scheduling types.
|
|
nodeTypeUntrusted(ntu)
|
boolean
|
|
|
This flag is used only when the evaluation manager is in "parallel" mode
but can be set at anytime. It activates or deactivates the override to force
untrusted scheduling for the class name argument(s) in the evaluation manager.
Legal object values are class type names: e.g. "transform", "skinCluster", "mesh".
When queried without specified nodes, it returns the list of nodes with the
untrusted scheduling override active.
Scheduling overrides take precedence over all of the node and node type
scheduling rules. Use with caution; certain nodes may not react well to
alternative scheduling types.
Untrusted scheduling will allow nodes to be evaluated in a critical section,
separately from any other node evaluation. It should be used only as a last resort
since the lost parallelism caused by untrusted nodes can greatly reduce performance.
|
|
upstreamFrom(ust)
|
string
|
|
|
Find the DG nodes that are immediately upstream of the named one in
the evaluation graph. Note that the connectivity is via evaluation mode
connections, not DG connections.
In query mode the graph is walked and any nodes upstream of the named
one are returned. The return type is alternating pairs of values that
represent the graph level and the node name, e.g. if you walk upstream
from C in the graph A -> B -> C then the return will be the array of
strings ("0","C","1","B","2","A"). Scripts can deconstruct this
information into something more visually recognizable. Note that cycles
are likely to be present so any such scripts would have to handle them.
In query mode, this flag needs a value.
|
|
safeMode(sfm)
|
boolean
|
|
|
This flag activates/deactivates parallel evaluation safe mode. When
enabled, parallel execution will fall back to serial when evaluation
graph is missing dependencies. Detection is happening on scheduling
of parallel evaluation, which means potential fallback will happen at
the next evaluation.
WARNING: This mode should be disabled with extreme caution. It will
prevent parallel mode from falling back to serial mode when an invalid
evaluation is detected. Sometimes the evaluation will still work
correctly in those situations and use of this flag will keep the peak
parallel performance running. However since the safe mode is used to
catch invalid evaluation disabling it may also cause problems with
evaluation, anything from invalid values, missing evaluation, or even
crashes.
|
|
|
Flag can appear in Create mode of command
|
Flag can appear in Edit mode of command
|
|
Flag can appear in Query mode of command
|
Flag can have multiple arguments, passed either as a tuple or a list.
|
import maya.cmds as cmds
import maya.cmds as cmds
# Enable evaluation manager in serial mode
cmds.evaluationManager( mode="serial" )
# Result: True #
# Confirm that the evaluation manager is currently enabled
cmds.evaluationManager( query=True, enabled=True )
# Result: True #
# Does nothing since it was already in serial mode
cmds.evaluationManager( mode="serial" )
# Result: True #
# Return the current evaluation manager mode and graph state
cmds.evaluationManager( query=True, mode=True )
# Result: [u'serial'] #
# Return all legal evaluation modes
cmds.evaluationManager( query=True )
# Result: [u'off', u'serial'] #
# Disable evaluation manager
cmds.evaluationManager( mode="off" )
# Result: True #
# Activate parallel scheduling override on transform nodes
cmds.evaluationManager( "transform", nodeTypeParallel=True, on=True )
# Result: True #
# Query for node types with the parallel scheduling override
cmds.evaluationManager( query=True, nodeTypeParallel=True )
# Result: ["transform"] #
# Query the parallel scheduling override of a particular node type
cmds.evaluationManager( "transform", query=True, nodeTypeParallel=True )
# Result: True #
# Activate locally serial scheduling override on transform nodes
cmds.evaluationManager( "transform", nodeTypeSerialize=True, on=True )
# Result: True #
# Query for node types with the locally serial scheduling override
cmds.evaluationManager( query=True, nodeTypeSerialize=True )
# Result: ["transform"] #
# Query the locally serial scheduling override of a particular node type
cmds.evaluationManager( "transform", query=True, nodeTypeSerialize=True )
# Result: True #
# Activate globally serial scheduling override on transform nodes
cmds.evaluationManager( "transform", nodeTypeGloballySerialize=True, on=True )
# Result: True #
# Query for node types with the globally serial scheduling override
cmds.evaluationManager( query=True, nodeTypeGloballySerialize=True )
# Result: ["transform"] #
# Query the globally serial scheduling override of a particular node type
cmds.evaluationManager( "transform", query=True, nodeTypeGloballySerialize=True )
# Result: True #
# Activate untrusted scheduling override on transform nodes
cmds.evaluationManager( "transform", nodeTypeUntrusted=True, on=True )
# Result: True #
# Query for node types with the untrusted scheduling override
cmds.evaluationManager( query=True, nodeTypeUntrusted=True )
# Result: ["transform"] #
# Query the untrusted scheduling override of a particular node type
cmds.evaluationManager( "transform", query=True, nodeTypeUntrusted=True )
# Result: True #
# Force rebuild of evaluation graph for normal context
cmds.evaluationManager( invalidate=True )
# Result: True #
# Test if evaluation graph is valid for normal context
cmds.evaluationManager( query=True, invalidate=True )
# Result: False #
# Activate safe mode for parallel executor
cmds.evaluationManager( safeMode=True )
# Result: True #
# Query safe mode for parallel executor
cmds.evaluationManager( query=True, safeMode=True )
# Result: True #
# Enable manipulation with evaluation manager
cmds.evaluationManager( manipulation=True )
# Result: True #
# Query manipulation with evaluation manager
cmds.evaluationManager( query=True, manipulation=True )
# Result: True #
# Show downstream nodes
cmds.evaluationManager( downstreamFrom='myNode' )
# Result: ['myDownstreamNode'] #
# Show all downstream nodes
cmds.evaluationManager( query=True, downstreamFrom='myRoot' )
# Result: ['0', 'myRoot', '1', 'myNode', '2', 'myDownstreamNode' ] #
# Show upstream nodes
cmds.evaluationManager( upstreamFrom='myDownstreamNode' )
# Result: ['myNode'] #
# Show cycle cluster
cmds.evaluationManager( cycleCluster='myNodeInCycle' )
# Result: ['myNode', 'myNodeInCycle', 'myNode'] #
# Activate async update
cmds.evaluationManager( asyncUpdate=True )
# Result: True #
# Query async update
cmds.evaluationManager( query=True, asyncUpdate=True )
# Result: True #
# Deactivate idle build
cmds.evaluationManager( idleBuild=False )
# Result: True #
# Query idleBuild
cmds.evaluationManager( query=True, idleBuild=True )
# Result: False #