Package com.smartgwt.client.widgets.tree

Class ResultTree

java.lang.Object
com.smartgwt.client.core.BaseClass
com.smartgwt.client.data.RecordList
com.smartgwt.client.widgets.tree.Tree
com.smartgwt.client.widgets.tree.ResultTree
All Implemented Interfaces:
HasHandlers, HasDataChangedHandlers, HasDataChangedHandlers
public class ResultTree extends Tree
ResultTrees are an implementation of the Tree API, used to handle hierarchical data, whose nodes are DataSource records which are retrieved from a server.

Modifying ResultTrees

ResultTree nodes cannot be directly added or removed from a paged ResultTree via Tree APIs such as Tree.add() or Tree.remove(), since such trees are considered to be read-only by virtue of containing ResultSets, which are read-only data structures. Even in other FetchModes, calling such APIs will only update the local cache of the ResultTree, rather than triggering any server traffict to update the DataSource.

Use DataSource.addData()/removeData() to add/remove rows from the DataSource, and the ResultTree will reflect the changes automatically. Alternatively, the DataSource.updateCaches() method may be called to only update local caches of the DataSource in question, without generating any server traffic.

To create a locally modifiable cache of records from a DataSource, you can use DataSource.fetchData() to retrieve a List of records which can be modified directly, or you can create a client-only DataSource from the retrieved data to share a modifiable cache between several DataBoundComponents.

  • Constructor Details

    • ResultTree

      public ResultTree()

    • ResultTree

      public ResultTree(JavaScriptObject jsObj)

  • Method Details

    • getOrCreateRef

      public static ResultTree getOrCreateRef(JavaScriptObject jsObj)

    • getPaletteDefaults

      public Map getPaletteDefaults()
      This method returns a Map of config properties suitable for use as the "defaults" attribute of a PaletteNode. Use it when you need to work with PaletteNodes indirectly, such when setting up TileRecords that will be used in a TilePalette. See the dev tools overview for examples of how to assemble and acquire a suitable defaults object when you are creating a PaletteNode indirectly

    • create

      public JavaScriptObject create()
      Overrides:
      create in class Tree

    • setAutoOpen

      public ResultTree setAutoOpen(String autoOpen)
      Which nodes should be opened automatically - applied whenever setRoot() is called, including during initialization and as part of a re-fetch caused, for example, by duplicate() or invalidateCache().

      Options are:

      • "none" - no nodes are opened automatically
      • "root" - opens the top-level node - in databound trees, this node is always hidden
      • "all" - when loading data on demand, opens the top-level node and all of it's direct descendants - otherwise, opens all loaded nodes
      Parameters:
      autoOpen - New autoOpen value. Default value is "none"
      Returns:
      ResultTree instance, for chaining setter calls

    • getAutoOpen

      public String getAutoOpen()
      Which nodes should be opened automatically - applied whenever setRoot() is called, including during initialization and as part of a re-fetch caused, for example, by duplicate() or invalidateCache().

      Options are:

      • "none" - no nodes are opened automatically
      • "root" - opens the top-level node - in databound trees, this node is always hidden
      • "all" - when loading data on demand, opens the top-level node and all of it's direct descendants - otherwise, opens all loaded nodes
      Returns:
      Current autoOpen value. Default value is "none"

    • setAutoPreserveOpenState

      public ResultTree setAutoPreserveOpenState(PreserveOpenState autoPreserveOpenState)
      Controls what happens to the "open state" - the set of nodes opened or closed by the end user after tree data is loaded - when an entirely new tree of nodes is loaded from the server, as a consequence of calling invalidateCache() or of changing criteria such that the current cache of nodes is dropped.
      Parameters:
      autoPreserveOpenState - New autoPreserveOpenState value. Default value is "whenUnique"
      Returns:
      ResultTree instance, for chaining setter calls

    • getAutoPreserveOpenState

      public PreserveOpenState getAutoPreserveOpenState()
      Controls what happens to the "open state" - the set of nodes opened or closed by the end user after tree data is loaded - when an entirely new tree of nodes is loaded from the server, as a consequence of calling invalidateCache() or of changing criteria such that the current cache of nodes is dropped.
      Returns:
      Current autoPreserveOpenState value. Default value is "whenUnique"

    • setAutoUpdateSiblingNodesOnDrag

      public ResultTree setAutoUpdateSiblingNodesOnDrag(Boolean autoUpdateSiblingNodesOnDrag) throws IllegalStateException
      For multi-link trees, indicates that we should automatically update the position values of old and new sibling records after a drag reparent or reposition-within-parent operation. For example, say you have a tree like this (where the number in parentheses indicates the node's position value):
             A
         - B (1)
         - C (2)
         - D (3)
       E
         - F (1)
         - G (2)
      and you drag node C out and drop it between nodes F and G. This drag operation will spawn two update operations to the server: a "remove" to delete node C from parent A, and an "add" to re-add it under parent E. With autoUpdateSiblingNodesOnDrag in force, we also automatically issue two "update" operations to the server - one to change the position on node D to 2, and another to change the position on node G to 3. The end result of this is that node position values are kept correct.

      Please note the following:

      • As noted above, these automatic updates are persistent - we send a queue of actual update requests to the server. This is convenient, but it may not be terribly efficient, particularly if you have just dropped a node at the head of a list of several hundred siblings. This is why we do not default this setting to true
      • The automatic updates work by applying an integer delta value to the existing position value. So in the above example, we would compute a delta of negative 1 for node D and positive 1 for node G. The upshot of this is that autoUpdateSiblingNodesOnDrag only works well if your position values are consecutive integers
      Parameters:
      autoUpdateSiblingNodesOnDrag - New autoUpdateSiblingNodesOnDrag value. Default value is (see below)
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getAutoUpdateSiblingNodesOnDrag

      public Boolean getAutoUpdateSiblingNodesOnDrag()
      For multi-link trees, indicates that we should automatically update the position values of old and new sibling records after a drag reparent or reposition-within-parent operation. For example, say you have a tree like this (where the number in parentheses indicates the node's position value):
             A
         - B (1)
         - C (2)
         - D (3)
       E
         - F (1)
         - G (2)
      and you drag node C out and drop it between nodes F and G. This drag operation will spawn two update operations to the server: a "remove" to delete node C from parent A, and an "add" to re-add it under parent E. With autoUpdateSiblingNodesOnDrag in force, we also automatically issue two "update" operations to the server - one to change the position on node D to 2, and another to change the position on node G to 3. The end result of this is that node position values are kept correct.

      Please note the following:

      • As noted above, these automatic updates are persistent - we send a queue of actual update requests to the server. This is convenient, but it may not be terribly efficient, particularly if you have just dropped a node at the head of a list of several hundred siblings. This is why we do not default this setting to true
      • The automatic updates work by applying an integer delta value to the existing position value. So in the above example, we would compute a delta of negative 1 for node D and positive 1 for node G. The upshot of this is that autoUpdateSiblingNodesOnDrag only works well if your position values are consecutive integers
      Returns:
      Current autoUpdateSiblingNodesOnDrag value. Default value is (see below)

    • setCanReturnOpenFolders

      public ResultTree setCanReturnOpenFolders(boolean canReturnOpenFolders) throws IllegalStateException
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, this property specifies the default value assumed for the canReturnOpenSubfoldersProperty when no value for that property is provided for a node.
      Parameters:
      canReturnOpenFolders - New canReturnOpenFolders value. Default value is false
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getCanReturnOpenFolders

      public boolean getCanReturnOpenFolders()
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, this property specifies the default value assumed for the canReturnOpenSubfoldersProperty when no value for that property is provided for a node.
      Returns:
      Current canReturnOpenFolders value. Default value is false

    • setCanReturnOpenSubfoldersProperty

      public ResultTree setCanReturnOpenSubfoldersProperty(String canReturnOpenSubfoldersProperty) throws IllegalStateException
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, canReturnOpenSubfoldersProperty may be set on any folder to indicate whether child folders might be returned by the server already open. If the property is set to false on a folder then subfolders of that folder are never allowed to be returned already open. This enables the paging mechanism to be more efficient in the amount of data that it requests from the server.

      For example, setting the canReturnOpenSubfoldersProperty value to false on a node is appropriate if the server-side code determines that the the node's children consist of entirely leaf nodes.

      Parameters:
      canReturnOpenSubfoldersProperty - New canReturnOpenSubfoldersProperty value. Default value is "canReturnOpenSubfolders"
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getCanReturnOpenSubfoldersProperty

      public String getCanReturnOpenSubfoldersProperty()
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, canReturnOpenSubfoldersProperty may be set on any folder to indicate whether child folders might be returned by the server already open. If the property is set to false on a folder then subfolders of that folder are never allowed to be returned already open. This enables the paging mechanism to be more efficient in the amount of data that it requests from the server.

      For example, setting the canReturnOpenSubfoldersProperty value to false on a node is appropriate if the server-side code determines that the the node's children consist of entirely leaf nodes.

      Returns:
      Current canReturnOpenSubfoldersProperty value. Default value is "canReturnOpenSubfolders"
      See Also:

    • setChildCountProperty

      public ResultTree setChildCountProperty(String childCountProperty) throws IllegalStateException
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, childCountProperty must be set for any folders that include only a partial list of children. For a deeper discussion see the Paging large sets of children section of the TreeDataBinding overview.
      Parameters:
      childCountProperty - New childCountProperty value. Default value is "childCount"
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getChildCountProperty

      public String getChildCountProperty()
      When using fetchMode:"paged" and providing multiple levels of the tree in one DSResponse, childCountProperty must be set for any folders that include only a partial list of children. For a deeper discussion see the Paging large sets of children section of the TreeDataBinding overview.
      Returns:
      Current childCountProperty value. Default value is "childCount"
      See Also:

    • setCriteria

      public ResultTree setCriteria(Criteria criteria)
      The filter criteria to use when fetching rows. For usage see setCriteria().

      If this method is called after the component has been drawn/initialized: Set the filter criteria to use when fetching rows.

      Depending on the result of compareCriteria() and setting for fetchMode, setting criteria may cause a trip to the server to get a new set of nodes, or may simply cause already-fetched nodes to be re-filtered according to the new criteria.

      For a basic overview on when server fetches are generally performed, see fetchMode. However, this is not the final determination of when server fetches occur. Criteria can be split into local criteria and server criteria by specifying serverFilterFields. Thus, even when using fetchMode:"local" a new server fetch will occur if the server criteria changes. For details on how the criteria is split, see DataSource.splitCriteria().

      Note: if criteria is being split to retrieve server criteria portion and the criteria is an AdvancedCriteria, the criteria must consist of a single "and" operator and one or more simple criteria below it. No other logical operators may be used. In other words, the AdvancedCriteria provided must be exactly representable by a simple criteria.

      Parameters:
      criteria - the filter criteria. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls

    • getCriteria

      public Criteria getCriteria()
      The filter criteria to use when fetching rows. For usage see setCriteria().
      Returns:
      Current criteria value. Default value is null

    • setData

      public ResultTree setData(TreeNode... data) throws IllegalStateException
      Optional initial data for the tree. If the fetchMode is "basic" or "local" then the format of this data is exactly the same parentId-linked list of tree nodes as documented on Tree.data (when the modelType is set to "parent"). If the fetchMode is "paged" then the format is extended to allow the childCountProperty to be set on folder nodes.

      Providing an initial set of nodes in this way does not affect the behavior of the ResultTree in its loading of unloaded folders. An equivalent result is achieved if the first fetch from the server returns this same data.

      If fetchMode is "paged" then you may make folder-by-folder choices as to whether to use paging for the childen of each folder. If you would like to use paging in a folder then you may include a partial list of that folder's children with the data, provided that you set the childCountProperty to the total number of children. Otherwise you will need to include either all children of the folder or none of the children. Open folders without any children provided will cause immediate, new fetches for the children, as usual.

      Because the initial data is treated exactly as though it were returned from the tree's first server fetch, the order of the initial data must match the initial sort order of the TreeGrid displaying the data or, if no such sort is specified, the native storage order on the server. For example, consider initial data containing n records having the parentId "X", meaning they are all in the same folder. These n records are the records at indices 0 through (n - 1) that are stored on the server under the parent node. If the childCountProperty set on the parent node indicates that there are m > n total rows under the parent node then the records at indices n to (m - 1) will be fetched from the server as the user scrolls the additional rows into view.

      Note : This is an advanced setting

      Overrides:
      setData in class Tree
      Parameters:
      data - New data value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getData

      public TreeNode[] getData()
      Optional initial data for the tree. If the fetchMode is "basic" or "local" then the format of this data is exactly the same parentId-linked list of tree nodes as documented on Tree.data (when the modelType is set to "parent"). If the fetchMode is "paged" then the format is extended to allow the childCountProperty to be set on folder nodes.

      Providing an initial set of nodes in this way does not affect the behavior of the ResultTree in its loading of unloaded folders. An equivalent result is achieved if the first fetch from the server returns this same data.

      If fetchMode is "paged" then you may make folder-by-folder choices as to whether to use paging for the childen of each folder. If you would like to use paging in a folder then you may include a partial list of that folder's children with the data, provided that you set the childCountProperty to the total number of children. Otherwise you will need to include either all children of the folder or none of the children. Open folders without any children provided will cause immediate, new fetches for the children, as usual.

      Because the initial data is treated exactly as though it were returned from the tree's first server fetch, the order of the initial data must match the initial sort order of the TreeGrid displaying the data or, if no such sort is specified, the native storage order on the server. For example, consider initial data containing n records having the parentId "X", meaning they are all in the same folder. These n records are the records at indices 0 through (n - 1) that are stored on the server under the parent node. If the childCountProperty set on the parent node indicates that there are m > n total rows under the parent node then the records at indices n to (m - 1) will be fetched from the server as the user scrolls the additional rows into view.

      Overrides:
      getData in class Tree
      Returns:
      Current data value. Default value is null
      See Also:

    • setDefaultIsFolder

      public ResultTree setDefaultIsFolder(Boolean defaultIsFolder) throws IllegalStateException
      Controls whether nodes are assumed to be folders or leaves by default.

      Nodes that have children or have the isFolderProperty set to true will always be considered folders. Other nodes will be considered folders or leaves by default according to this setting.

      If defaultIsFolder is unset, the ResultTree will automatically set it to match the value of loadDataOnDemand. This means that, when using folder-by-folder load on demand (loadDataOnDemand:true), by default a newly loaded node will be considered to be a folder that has not loaded its children yet.

      When not using folder-by-folder load on demand, by default a newly loaded node is considered a leaf. If you set defaultIsFolder:true explicitly, by default a newly loaded node is considered to be a folder with no children.

      See Tree.isFolder() for details on how to explicitly mark nodes as folders or leaves.

      Overrides:
      setDefaultIsFolder in class Tree
      Parameters:
      defaultIsFolder - New defaultIsFolder value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getDefaultIsFolder

      public Boolean getDefaultIsFolder()
      Controls whether nodes are assumed to be folders or leaves by default.

      Nodes that have children or have the isFolderProperty set to true will always be considered folders. Other nodes will be considered folders or leaves by default according to this setting.

      If defaultIsFolder is unset, the ResultTree will automatically set it to match the value of loadDataOnDemand. This means that, when using folder-by-folder load on demand (loadDataOnDemand:true), by default a newly loaded node will be considered to be a folder that has not loaded its children yet.

      When not using folder-by-folder load on demand, by default a newly loaded node is considered a leaf. If you set defaultIsFolder:true explicitly, by default a newly loaded node is considered to be a folder with no children.

      See Tree.isFolder() for details on how to explicitly mark nodes as folders or leaves.

      Overrides:
      getDefaultIsFolder in class Tree
      Returns:
      Current defaultIsFolder value. Default value is null
      See Also:

    • setDefaultNewNodesToRoot

      public ResultTree setDefaultNewNodesToRoot(Boolean defaultNewNodesToRoot)
      This attribute governs how to handle cache-synch when a new node is added to this dataSource with no explicit parentId.

      If set to true, when a new node is added to this dataSource via DataSource.addData(), with no explicit parentId, the node will be added as a child of the root node of this result tree. Otherwise it will be ignored.

      Similar logic applies to updated nodes - if this property is true and the parentId of an updated node is cleared, it will be moved to become a child of root, otherwise it will be dropped from the tree.

      Note : This is an advanced setting

      Parameters:
      defaultNewNodesToRoot - New defaultNewNodesToRoot value. Default value is false
      Returns:
      ResultTree instance, for chaining setter calls

    • getDefaultNewNodesToRoot

      public Boolean getDefaultNewNodesToRoot()
      This attribute governs how to handle cache-synch when a new node is added to this dataSource with no explicit parentId.

      If set to true, when a new node is added to this dataSource via DataSource.addData(), with no explicit parentId, the node will be added as a child of the root node of this result tree. Otherwise it will be ignored.

      Similar logic applies to updated nodes - if this property is true and the parentId of an updated node is cleared, it will be moved to become a child of root, otherwise it will be dropped from the tree.

      Returns:
      Current defaultNewNodesToRoot value. Default value is false

    • setDisableCacheSync

      public ResultTree setDisableCacheSync(Boolean disableCacheSync) throws IllegalStateException
      By default when the data of this ResultTree's dataSource is modified, the ResultTree will be updated to display these changes. Set this flag to true to disable this behavior.

      Note : This is an advanced setting

      Parameters:
      disableCacheSync - New disableCacheSync value. Default value is false
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getDisableCacheSync

      public Boolean getDisableCacheSync()
      By default when the data of this ResultTree's dataSource is modified, the ResultTree will be updated to display these changes. Set this flag to true to disable this behavior.
      Returns:
      Current disableCacheSync value. Default value is false

    • setDiscardParentlessNodes

      public ResultTree setDiscardParentlessNodes(Boolean discardParentlessNodes) throws IllegalStateException
      When data is loaded from the server, should nodes with an explicit value for the Tree.parentIdField which doesn't map to a valid parent node be dropped? If set to false, for TreeGrid.loadDataOnDemand:false trees, parentless nodes will be added as children of the root node - for TreeGrid.loadDataOnDemand:true, they will be added as children of the folder currently requesting children.

      This effectively allows nodes to be loaded into the current (or root) folder without needing an explicit parentIdField value that matches the folder's ID or rootValue for the resultTree.

      Note: For loadDataOnDemand:false trees, if this property is unset at init time, it will default to true if an explicit rootNode has been specified. This ensures that if the data tree retrieved from the server includes ancestors of the desired root-node we don't display them. Otherwise this property always defaults to false.

      Note : This is an advanced setting

      Overrides:
      setDiscardParentlessNodes in class Tree
      Parameters:
      discardParentlessNodes - New discardParentlessNodes value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getDiscardParentlessNodes

      public Boolean getDiscardParentlessNodes()
      When data is loaded from the server, should nodes with an explicit value for the Tree.parentIdField which doesn't map to a valid parent node be dropped? If set to false, for TreeGrid.loadDataOnDemand:false trees, parentless nodes will be added as children of the root node - for TreeGrid.loadDataOnDemand:true, they will be added as children of the folder currently requesting children.

      This effectively allows nodes to be loaded into the current (or root) folder without needing an explicit parentIdField value that matches the folder's ID or rootValue for the resultTree.

      Note: For loadDataOnDemand:false trees, if this property is unset at init time, it will default to true if an explicit rootNode has been specified. This ensures that if the data tree retrieved from the server includes ancestors of the desired root-node we don't display them. Otherwise this property always defaults to false.

      Overrides:
      getDiscardParentlessNodes in class Tree
      Returns:
      Current discardParentlessNodes value. Default value is null

    • setFetchMode

      public ResultTree setFetchMode(FetchMode fetchMode) throws IllegalStateException
      Mode of fetching records from server.

      fetchMode:"local" implies that local filtering will always be performed. See keepParentsOnFilter for additional filtering details.

      fetchMode:"basic" or "paged" implies that if search criteria change, the entire tree will be discarded and re-fetched from the server. When retrieving the replacement tree data, the default behavior will be to preserve the openState for any nodes that the server returns which were previously opened by the user. Note that this implies that if loadDataOnDemand is enabled and the server returns only root-level nodes, open state will be preserved only for root-level nodes, and children of open root-level nodes will be immediately fetched from the server if they are not included in the server's initial response.

      fetchMode:"paged" enables paging for nodes that have very large numbers of children. Whenever the children of a folder are loaded, the resultTree will set DSRequest.startRow and endRow when requesting children from the DataSource, and will manage loading of further children on demand, similar to how a ResultSet manages paging for lists. For a deeper discussion see the Paging large sets of children section of the TreeDataBinding overview.

      Parameters:
      fetchMode - New fetchMode value. Default value is "basic"
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getFetchMode

      public FetchMode getFetchMode()
      Mode of fetching records from server.

      fetchMode:"local" implies that local filtering will always be performed. See keepParentsOnFilter for additional filtering details.

      fetchMode:"basic" or "paged" implies that if search criteria change, the entire tree will be discarded and re-fetched from the server. When retrieving the replacement tree data, the default behavior will be to preserve the openState for any nodes that the server returns which were previously opened by the user. Note that this implies that if loadDataOnDemand is enabled and the server returns only root-level nodes, open state will be preserved only for root-level nodes, and children of open root-level nodes will be immediately fetched from the server if they are not included in the server's initial response.

      fetchMode:"paged" enables paging for nodes that have very large numbers of children. Whenever the children of a folder are loaded, the resultTree will set DSRequest.startRow and endRow when requesting children from the DataSource, and will manage loading of further children on demand, similar to how a ResultSet manages paging for lists. For a deeper discussion see the Paging large sets of children section of the TreeDataBinding overview.

      Returns:
      Current fetchMode value. Default value is "basic"
      See Also:

    • setFirstPositionValue

      public ResultTree setFirstPositionValue(Integer firstPositionValue)
      If autoUpdateSiblingNodesOnDrag is in force, this is the value we will use to auto-update the position of a node when we cannot derive that value from the existing value of a neighbor. This happens when a node is dropped into the very first position below a parent (including the special case of the parent being previously empty)
      Parameters:
      firstPositionValue - New firstPositionValue value. Default value is 1
      Returns:
      ResultTree instance, for chaining setter calls

    • getFirstPositionValue

      public Integer getFirstPositionValue()
      If autoUpdateSiblingNodesOnDrag is in force, this is the value we will use to auto-update the position of a node when we cannot derive that value from the existing value of a neighbor. This happens when a node is dropped into the very first position below a parent (including the special case of the parent being previously empty)
      Returns:
      Current firstPositionValue value. Default value is 1

    • setImplicitCriteria

      public ResultTree setImplicitCriteria(Criteria implicitCriteria)
      Criteria that are never shown to or edited by the user and are cumulative with any criteria provided via DataBoundComponent.initialCriteria, setCriteria() etc.
      Parameters:
      implicitCriteria - New implicitCriteria value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls

    • getImplicitCriteria

      public Criteria getImplicitCriteria()
      Criteria that are never shown to or edited by the user and are cumulative with any criteria provided via DataBoundComponent.initialCriteria, setCriteria() etc.
      Returns:
      Current implicitCriteria value. Default value is null

    • setKeepParentsOnFilter

      public ResultTree setKeepParentsOnFilter(Boolean keepParentsOnFilter) throws IllegalStateException
      If set, tree-based filtering is performed such that parent nodes are kept as long as they have children that match the filter criteria, even if the parents themselves do not match the filter criteria. If not set, filtering will exclude parent nodes not matching the criteria, and all nodes below them in the tree.

      ResultTrees will default to fetchMode:"local" whenever keepParentsOnFilter is true, unless fetchMode was explicitly set to "paged" (see below). This allows the filtering logic to fetch a complete tree of nodes from the DataSource (or if loadDataOnDemand:true, a complete set of nodes under a given parent) and then filter the resulting data locally on the client.

      This means that the server does not need to implement special tree filtering logic to support looking up nodes that match the specified criteria as well as ancestor nodes that may not.

      If some criteria must be sent to the server in order to produce a valid tree of data, but keepParentsOnFilter is also required, the serverFilterFields attribute may be used to specify a list of field names that will be sent to the server whenever they are present in the criteria. Note that for the subset of criteria applied to these fields, keepParentsInFilter behavior will not occur without custom logic in the DataSource fetch operation.

      If FetchMode is explicitly set to "paged", it is not possible to implement keepParentsOnFilter, either by local filtering or with the automatic client-driven handling mentioned below. Support for keepParentsOnFilter for a paged ResultTree therefore also requires custom logic in the DataSource fetch operation. To support this a developer must ensure that their fetch operation returns the appropriate set of nodes - all nodes that match the specified criteria plus their ancestor nodes even if they do not match the specified criteria.

      keepParentsOnFilter with load-on-demand trees

      The combination of keepParentsOnFilter and loadDataOnDemand presents additional difficulties that require special handling. The problem is that in order to determine even the top-level folders, you have to examine every node in the entire tree. For example, say there is one top-level folder that has thousands of folders and nodes underneath it, and there is just one leaf node, 6 levels deep, that matches the filter criteria. You have to find out about that node, because it implies the top-level folder must be retained.

      So the server basically has to examine every node in the dataset to determine even what shows up at the top level of the tree. If it does not do this, parent nodes that don't match the filter criteria will be excluded from the tree, with the upshot that the child nodes that do match the criteria will be inaccessible because nodes in load-on-demand trees are only loaded when their parent node is opened

      By default, Smart GWT solves this with a client-driven implementation of this special handling. This algorithm involves finding the nodes that match the filter criteria - which we term matching leaves - and then recursively travelling back up the tree, determining the ancestors of the matching leaves - the so-called dangling parents. When we have traversed all the way back to the root node from every matching leaf, we have recorded every dangling parent and have what we term the skeleton of the tree. The skeleton is then added to fetch criteria whenever a load-on-demand fetch request is made, ensuring that we fetch both dangling parents and matching leaves.

      There are three ways this recursive traversal can be implemented:

      • For dataSources that support dynamic tree joins, we use the additionalOutputs feature to declare self-joins that fetch multiple levels of parent in one query (the number of levels is configurable, see matchingLeafJoinDepth). Of Smart GWT's built-in DataSource types, only SQLDataSource is currently capable of this approach
      • For server-side dataSources that do not support self-joins, we combine individual single-level fetches into a queue, using fieldValueExpressions with responseData "allRecords" so that each fetch in the queue uses the output of the previous fetch as its criteria (so the first fetch returns the parents of the matching nodes, the second fetch returns the parents of those nodes, and so on). Again, the number of fetches per queue can be configured with the matchingLeafJoinDepth property. This approach works for any server-side DataSource implementation, including your own custom implementations
      • For client-side dataSources, which support neither self-joins not queueing, the algorithm simply makes as many single-level requests as necessary to build the entire skeleton. Note, this is exactly what would happen with previously-mentioned queueing approach, if you set matchingLeafJoinDepth to 1
      If you want to disable the automatic handling of keepParentsOnFilter on load-on-demand trees, see serverKeepParentsOnFilter
      Parameters:
      keepParentsOnFilter - New keepParentsOnFilter value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getKeepParentsOnFilter

      public Boolean getKeepParentsOnFilter()
      If set, tree-based filtering is performed such that parent nodes are kept as long as they have children that match the filter criteria, even if the parents themselves do not match the filter criteria. If not set, filtering will exclude parent nodes not matching the criteria, and all nodes below them in the tree.

      ResultTrees will default to fetchMode:"local" whenever keepParentsOnFilter is true, unless fetchMode was explicitly set to "paged" (see below). This allows the filtering logic to fetch a complete tree of nodes from the DataSource (or if loadDataOnDemand:true, a complete set of nodes under a given parent) and then filter the resulting data locally on the client.

      This means that the server does not need to implement special tree filtering logic to support looking up nodes that match the specified criteria as well as ancestor nodes that may not.

      If some criteria must be sent to the server in order to produce a valid tree of data, but keepParentsOnFilter is also required, the serverFilterFields attribute may be used to specify a list of field names that will be sent to the server whenever they are present in the criteria. Note that for the subset of criteria applied to these fields, keepParentsInFilter behavior will not occur without custom logic in the DataSource fetch operation.

      If FetchMode is explicitly set to "paged", it is not possible to implement keepParentsOnFilter, either by local filtering or with the automatic client-driven handling mentioned below. Support for keepParentsOnFilter for a paged ResultTree therefore also requires custom logic in the DataSource fetch operation. To support this a developer must ensure that their fetch operation returns the appropriate set of nodes - all nodes that match the specified criteria plus their ancestor nodes even if they do not match the specified criteria.

      keepParentsOnFilter with load-on-demand trees

      The combination of keepParentsOnFilter and loadDataOnDemand presents additional difficulties that require special handling. The problem is that in order to determine even the top-level folders, you have to examine every node in the entire tree. For example, say there is one top-level folder that has thousands of folders and nodes underneath it, and there is just one leaf node, 6 levels deep, that matches the filter criteria. You have to find out about that node, because it implies the top-level folder must be retained.

      So the server basically has to examine every node in the dataset to determine even what shows up at the top level of the tree. If it does not do this, parent nodes that don't match the filter criteria will be excluded from the tree, with the upshot that the child nodes that do match the criteria will be inaccessible because nodes in load-on-demand trees are only loaded when their parent node is opened

      By default, Smart GWT solves this with a client-driven implementation of this special handling. This algorithm involves finding the nodes that match the filter criteria - which we term matching leaves - and then recursively travelling back up the tree, determining the ancestors of the matching leaves - the so-called dangling parents. When we have traversed all the way back to the root node from every matching leaf, we have recorded every dangling parent and have what we term the skeleton of the tree. The skeleton is then added to fetch criteria whenever a load-on-demand fetch request is made, ensuring that we fetch both dangling parents and matching leaves.

      There are three ways this recursive traversal can be implemented:

      • For dataSources that support dynamic tree joins, we use the additionalOutputs feature to declare self-joins that fetch multiple levels of parent in one query (the number of levels is configurable, see matchingLeafJoinDepth). Of Smart GWT's built-in DataSource types, only SQLDataSource is currently capable of this approach
      • For server-side dataSources that do not support self-joins, we combine individual single-level fetches into a queue, using fieldValueExpressions with responseData "allRecords" so that each fetch in the queue uses the output of the previous fetch as its criteria (so the first fetch returns the parents of the matching nodes, the second fetch returns the parents of those nodes, and so on). Again, the number of fetches per queue can be configured with the matchingLeafJoinDepth property. This approach works for any server-side DataSource implementation, including your own custom implementations
      • For client-side dataSources, which support neither self-joins not queueing, the algorithm simply makes as many single-level requests as necessary to build the entire skeleton. Note, this is exactly what would happen with previously-mentioned queueing approach, if you set matchingLeafJoinDepth to 1
      If you want to disable the automatic handling of keepParentsOnFilter on load-on-demand trees, see serverKeepParentsOnFilter
      Returns:
      Current keepParentsOnFilter value. Default value is null
      See Also:

    • setKeepParentsOnFilterMaxNodes

      public ResultTree setKeepParentsOnFilterMaxNodes(Integer keepParentsOnFilterMaxNodes)
      When a tree specifies the combination of keepParentsOnFilter and loadDataOnDemand, Smart GWT by default automatically fetches the "skeleton" of the filtered tree - see the keepParentsOnFilter overview for details, including the definition of "skeleton" and other relevant terminology

      A problem can arise with this approach if the user enters overly inclusive filter criteria. For example, say you have a 200,000 row dataset and the user chooses to apply a filter of "a". Chances are that is going to include the majority of the nodes in the tree, which would be OK because this is a load-on-demand tree. However, because we will build, cache and then pass around the list of id's of the dangling parents, this may become a performance issue. A lot depends on the nature of your data - this will be much less of an issue for shallow trees with lots of leaf nodes relative to parents, compared to deep trees with a lot of dangling parents to record.

      If the user tries to filter the TreeGrid such that there are more matching nodes than is allowed by this setting, the system will truncate the fetch and show the warning message defined in keepParentsOnFilterMaxNodesExceededMessage. Since the cached node-list is derived from bottom to top, this truncation of the fetch process will usually mean we have not yet derived any top-level nodes. This in turn means that the tree will appear to be empty.

      Setting this property to a suitable value for your specific use case is an application tuning exercise, finding the right balance between usability and performance. To remove the node limit altogether, set this property to -1. However, if you have a load-on-demand tree over a large dataset, we do not recommend that you remove the limit completely, as it can lead to serious problems on both the client and server, as the application tries to cope with criteria that contains huge numbers of id's.

      Parameters:
      keepParentsOnFilterMaxNodes - New keepParentsOnFilterMaxNodes value. Default value is 8000
      Returns:
      ResultTree instance, for chaining setter calls

    • getKeepParentsOnFilterMaxNodes

      public Integer getKeepParentsOnFilterMaxNodes()
      When a tree specifies the combination of keepParentsOnFilter and loadDataOnDemand, Smart GWT by default automatically fetches the "skeleton" of the filtered tree - see the keepParentsOnFilter overview for details, including the definition of "skeleton" and other relevant terminology

      A problem can arise with this approach if the user enters overly inclusive filter criteria. For example, say you have a 200,000 row dataset and the user chooses to apply a filter of "a". Chances are that is going to include the majority of the nodes in the tree, which would be OK because this is a load-on-demand tree. However, because we will build, cache and then pass around the list of id's of the dangling parents, this may become a performance issue. A lot depends on the nature of your data - this will be much less of an issue for shallow trees with lots of leaf nodes relative to parents, compared to deep trees with a lot of dangling parents to record.

      If the user tries to filter the TreeGrid such that there are more matching nodes than is allowed by this setting, the system will truncate the fetch and show the warning message defined in keepParentsOnFilterMaxNodesExceededMessage. Since the cached node-list is derived from bottom to top, this truncation of the fetch process will usually mean we have not yet derived any top-level nodes. This in turn means that the tree will appear to be empty.

      Setting this property to a suitable value for your specific use case is an application tuning exercise, finding the right balance between usability and performance. To remove the node limit altogether, set this property to -1. However, if you have a load-on-demand tree over a large dataset, we do not recommend that you remove the limit completely, as it can lead to serious problems on both the client and server, as the application tries to cope with criteria that contains huge numbers of id's.

      Returns:
      Current keepParentsOnFilterMaxNodes value. Default value is 8000

    • setLinkDataAddOperation

      public ResultTree setLinkDataAddOperation(String linkDataAddOperation)
      The operationId this ResultTree should use when performing add operations on its linkDataSource. Has no effect if this is not a multi-link tree.

      Note, this property wll be used by internal update operations when you drag-move or drag-reparent nodes in a multi-link tree. Do not use it when adding records from application code by directly calling addData() on the linkDataSource; instead just use the regular operationId property in your add request. Also note, because this property is intended to allow your code to influence the operationId used by internal methods, and those methods never directly update link data (moved and re-parented links are always removed and then re-added), there is no corresponding linkDataUpdateOperation property.

      Parameters:
      linkDataAddOperation - New linkDataAddOperation value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls

    • getLinkDataAddOperation

      public String getLinkDataAddOperation()
      The operationId this ResultTree should use when performing add operations on its linkDataSource. Has no effect if this is not a multi-link tree.

      Note, this property wll be used by internal update operations when you drag-move or drag-reparent nodes in a multi-link tree. Do not use it when adding records from application code by directly calling addData() on the linkDataSource; instead just use the regular operationId property in your add request. Also note, because this property is intended to allow your code to influence the operationId used by internal methods, and those methods never directly update link data (moved and re-parented links are always removed and then re-added), there is no corresponding linkDataUpdateOperation property.

      Returns:
      Current linkDataAddOperation value. Default value is null

    • setLinkDataFetchMode

      public ResultTree setLinkDataFetchMode(LinkDataFetchMode linkDataFetchMode) throws IllegalStateException
      The fetch mode for this tree's link data; ignored if this is not a multi-link tree
      Parameters:
      linkDataFetchMode - New linkDataFetchMode value. Default value is "separate"
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getLinkDataFetchMode

      public LinkDataFetchMode getLinkDataFetchMode()
      The fetch mode for this tree's link data; ignored if this is not a multi-link tree
      Returns:
      Current linkDataFetchMode value. Default value is "separate"

    • setLinkDataFetchOperation

      public ResultTree setLinkDataFetchOperation(String linkDataFetchOperation)
      The operationId this ResultTree should use when performing fetch operations on its linkDataSource. Has no effect if this is not a multi-link tree

      Note, this value can be overridden by DSRequest.linkDataFetchOperation when calling fetchData() on the component (e.g. TreeGrid.fetchData) directly from application code.

      Parameters:
      linkDataFetchOperation - New linkDataFetchOperation value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls

    • getLinkDataFetchOperation

      public String getLinkDataFetchOperation()
      The operationId this ResultTree should use when performing fetch operations on its linkDataSource. Has no effect if this is not a multi-link tree

      Note, this value can be overridden by DSRequest.linkDataFetchOperation when calling fetchData() on the component (e.g. TreeGrid.fetchData) directly from application code.

      Returns:
      Current linkDataFetchOperation value. Default value is null

    • setLinkDataRemoveOperation

      public ResultTree setLinkDataRemoveOperation(String linkDataRemoveOperation)
      The operationId this ResultTree should use when performing remove operations on its linkDataSource. Has no effect if this is not a multi-link tree.

      See linkDataAddOperation for more information on how this property is intended to be used.

      Parameters:
      linkDataRemoveOperation - New linkDataRemoveOperation value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls

    • getLinkDataRemoveOperation

      public String getLinkDataRemoveOperation()
      The operationId this ResultTree should use when performing remove operations on its linkDataSource. Has no effect if this is not a multi-link tree.

      See linkDataAddOperation for more information on how this property is intended to be used.

      Returns:
      Current linkDataRemoveOperation value. Default value is null

    • setLinkDataSource

      public ResultTree setLinkDataSource(DataSource linkDataSource) throws IllegalStateException
      This property allows you to specify the dataSource to be used for fetching link information in a databound multilink tree. A multilink tree is one where the same node is allowed to appear in multiple places in the tree, and it is achieved by providing the node data and the link data separately. Nodes are provided via the normal dataSource; linkDataSource is only used for fetching and updating link information.

      The linkDataSource is an ordinary DataSource that you implement just like any other. However, for correct operation as a linkDataSource, it must have the following:

      Providing node data and link data

      Consider a structure for the components of a bicycle, greatly simplified:
                Frame
         /    \
      Wheel   Wheel
     /  \     /  \
   Hub Tire  Hub Tire
      Here, the two wheels are the same assembly, so really it should look like this:
              Frame
         | |               
        Wheel
        /   \     
      Hub  Tire
      Normal Smart GWT trees cannot model this arrangement accurately because this is not really a tree, it is a graph; trees do not contain multiple paths to a given node. The only way to handle this arrangement of nodes in a formal tree would be to make two copies of the "Wheel" node, at which point they are no longer the same thing. Either way, in a TreeGrid, we would have to visualise it like this:
       
    Frame
       Wheel
          Hub
          Tire
       Wheel
          Hub
          Tire
      But if we use copies so that the the two wheels are no longer the same thing, changing one of them will no longer change the other, which is a fundamental problem because in this scenario, the two wheels really are the same thing. Now, changing the name of the "Hub" in one "Wheel" would not change it in the other; adding a "Spokes" node to the second item would not also add it to the first. Drag-reordering child nodes in one "Wheel" would not re-order them in the other. All of these things are incorrect, because the two wheels are the same thing.

      Multilink trees provide a way to handle this arrangement without physical copying of the duplicate nodes, preserving the sameness of them and thus fixing all the problems described above.

      The node data for the above tree, simplified, would be a flat list something like this:

          [
       { id: 1, description: "Frame" },
       { id: 2, description: "Wheel" },
       { id: 3, description: "Hub" },
       { id: 4, description: "Tire" }
    ]
      The link data would look like this:
          [
       { linkId: 1, parentId: 1, id: 2, position: 1 },
       { linkId: 2, parentId: 2, id: 3, position: 1 },
       { linkId: 3, parentId: 2, id: 4, position: 2 },
       { linkId: 4, parentId: 1, id: 2, position: 2 }
    ]
      Or, if you were using linkDataFetchMode "single", you would combine the node and link data into a single dataset like this:
          [
       { id: 1, position: 0, description: "Frame" },
       { parentId: 1, id: 2, position: 1, description: "Wheel", linkId: 1 },
       { parentId: 2, id: 3, position: 1, description: "Hub", linkId: 2 },
       { parentId: 2, id: 4, position: 2, description: "Tire", linkId: 3 },
       { parentId: 1, id: 2, position: 2, description: "Wheel", linkId: 4 }
    ]

      NOTE: It is also possible to create an unbound multilink tree - see Tree.linkData.

      Parameters:
      linkDataSource - New linkDataSource value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getLinkDataSource

      public DataSource getLinkDataSource()
      This property allows you to specify the dataSource to be used for fetching link information in a databound multilink tree. A multilink tree is one where the same node is allowed to appear in multiple places in the tree, and it is achieved by providing the node data and the link data separately. Nodes are provided via the normal dataSource; linkDataSource is only used for fetching and updating link information.

      The linkDataSource is an ordinary DataSource that you implement just like any other. However, for correct operation as a linkDataSource, it must have the following:

      Providing node data and link data

      Consider a structure for the components of a bicycle, greatly simplified:
                Frame
         /    \
      Wheel   Wheel
     /  \     /  \
   Hub Tire  Hub Tire
      Here, the two wheels are the same assembly, so really it should look like this:
              Frame
         | |               
        Wheel
        /   \     
      Hub  Tire
      Normal Smart GWT trees cannot model this arrangement accurately because this is not really a tree, it is a graph; trees do not contain multiple paths to a given node. The only way to handle this arrangement of nodes in a formal tree would be to make two copies of the "Wheel" node, at which point they are no longer the same thing. Either way, in a TreeGrid, we would have to visualise it like this:
       
    Frame
       Wheel
          Hub
          Tire
       Wheel
          Hub
          Tire
      But if we use copies so that the the two wheels are no longer the same thing, changing one of them will no longer change the other, which is a fundamental problem because in this scenario, the two wheels really are the same thing. Now, changing the name of the "Hub" in one "Wheel" would not change it in the other; adding a "Spokes" node to the second item would not also add it to the first. Drag-reordering child nodes in one "Wheel" would not re-order them in the other. All of these things are incorrect, because the two wheels are the same thing.

      Multilink trees provide a way to handle this arrangement without physical copying of the duplicate nodes, preserving the sameness of them and thus fixing all the problems described above.

      The node data for the above tree, simplified, would be a flat list something like this:

          [
       { id: 1, description: "Frame" },
       { id: 2, description: "Wheel" },
       { id: 3, description: "Hub" },
       { id: 4, description: "Tire" }
    ]
      The link data would look like this:
          [
       { linkId: 1, parentId: 1, id: 2, position: 1 },
       { linkId: 2, parentId: 2, id: 3, position: 1 },
       { linkId: 3, parentId: 2, id: 4, position: 2 },
       { linkId: 4, parentId: 1, id: 2, position: 2 }
    ]
      Or, if you were using linkDataFetchMode "single", you would combine the node and link data into a single dataset like this:
          [
       { id: 1, position: 0, description: "Frame" },
       { parentId: 1, id: 2, position: 1, description: "Wheel", linkId: 1 },
       { parentId: 2, id: 3, position: 1, description: "Hub", linkId: 2 },
       { parentId: 2, id: 4, position: 2, description: "Tire", linkId: 3 },
       { parentId: 1, id: 2, position: 2, description: "Wheel", linkId: 4 }
    ]

      NOTE: It is also possible to create an unbound multilink tree - see Tree.linkData.

      Returns:
      Current linkDataSource value. Default value is null

    • setLinkDataSource

      public ResultTree setLinkDataSource(String linkDataSource) throws IllegalStateException
      This property allows you to specify the dataSource to be used for fetching link information in a databound multilink tree. A multilink tree is one where the same node is allowed to appear in multiple places in the tree, and it is achieved by providing the node data and the link data separately. Nodes are provided via the normal dataSource; linkDataSource is only used for fetching and updating link information.

      The linkDataSource is an ordinary DataSource that you implement just like any other. However, for correct operation as a linkDataSource, it must have the following:

      Providing node data and link data

      Consider a structure for the components of a bicycle, greatly simplified:
                Frame
         /    \
      Wheel   Wheel
     /  \     /  \
   Hub Tire  Hub Tire
      Here, the two wheels are the same assembly, so really it should look like this:
              Frame
         | |               
        Wheel
        /   \     
      Hub  Tire
      Normal Smart GWT trees cannot model this arrangement accurately because this is not really a tree, it is a graph; trees do not contain multiple paths to a given node. The only way to handle this arrangement of nodes in a formal tree would be to make two copies of the "Wheel" node, at which point they are no longer the same thing. Either way, in a TreeGrid, we would have to visualise it like this:
       
    Frame
       Wheel
          Hub
          Tire
       Wheel
          Hub
          Tire
      But if we use copies so that the the two wheels are no longer the same thing, changing one of them will no longer change the other, which is a fundamental problem because in this scenario, the two wheels really are the same thing. Now, changing the name of the "Hub" in one "Wheel" would not change it in the other; adding a "Spokes" node to the second item would not also add it to the first. Drag-reordering child nodes in one "Wheel" would not re-order them in the other. All of these things are incorrect, because the two wheels are the same thing.

      Multilink trees provide a way to handle this arrangement without physical copying of the duplicate nodes, preserving the sameness of them and thus fixing all the problems described above.

      The node data for the above tree, simplified, would be a flat list something like this:

          [
       { id: 1, description: "Frame" },
       { id: 2, description: "Wheel" },
       { id: 3, description: "Hub" },
       { id: 4, description: "Tire" }
    ]
      The link data would look like this:
          [
       { linkId: 1, parentId: 1, id: 2, position: 1 },
       { linkId: 2, parentId: 2, id: 3, position: 1 },
       { linkId: 3, parentId: 2, id: 4, position: 2 },
       { linkId: 4, parentId: 1, id: 2, position: 2 }
    ]
      Or, if you were using linkDataFetchMode "single", you would combine the node and link data into a single dataset like this:
          [
       { id: 1, position: 0, description: "Frame" },
       { parentId: 1, id: 2, position: 1, description: "Wheel", linkId: 1 },
       { parentId: 2, id: 3, position: 1, description: "Hub", linkId: 2 },
       { parentId: 2, id: 4, position: 2, description: "Tire", linkId: 3 },
       { parentId: 1, id: 2, position: 2, description: "Wheel", linkId: 4 }
    ]

      NOTE: It is also possible to create an unbound multilink tree - see Tree.linkData.

      Parameters:
      linkDataSource - New linkDataSource value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getLinkDataSourceAsString

      public String getLinkDataSourceAsString()
      This property allows you to specify the dataSource to be used for fetching link information in a databound multilink tree. A multilink tree is one where the same node is allowed to appear in multiple places in the tree, and it is achieved by providing the node data and the link data separately. Nodes are provided via the normal dataSource; linkDataSource is only used for fetching and updating link information.

      The linkDataSource is an ordinary DataSource that you implement just like any other. However, for correct operation as a linkDataSource, it must have the following:

      Providing node data and link data

      Consider a structure for the components of a bicycle, greatly simplified:
                Frame
         /    \
      Wheel   Wheel
     /  \     /  \
   Hub Tire  Hub Tire
      Here, the two wheels are the same assembly, so really it should look like this:
              Frame
         | |               
        Wheel
        /   \     
      Hub  Tire
      Normal Smart GWT trees cannot model this arrangement accurately because this is not really a tree, it is a graph; trees do not contain multiple paths to a given node. The only way to handle this arrangement of nodes in a formal tree would be to make two copies of the "Wheel" node, at which point they are no longer the same thing. Either way, in a TreeGrid, we would have to visualise it like this:
       
    Frame
       Wheel
          Hub
          Tire
       Wheel
          Hub
          Tire
      But if we use copies so that the the two wheels are no longer the same thing, changing one of them will no longer change the other, which is a fundamental problem because in this scenario, the two wheels really are the same thing. Now, changing the name of the "Hub" in one "Wheel" would not change it in the other; adding a "Spokes" node to the second item would not also add it to the first. Drag-reordering child nodes in one "Wheel" would not re-order them in the other. All of these things are incorrect, because the two wheels are the same thing.

      Multilink trees provide a way to handle this arrangement without physical copying of the duplicate nodes, preserving the sameness of them and thus fixing all the problems described above.

      The node data for the above tree, simplified, would be a flat list something like this:

          [
       { id: 1, description: "Frame" },
       { id: 2, description: "Wheel" },
       { id: 3, description: "Hub" },
       { id: 4, description: "Tire" }
    ]
      The link data would look like this:
          [
       { linkId: 1, parentId: 1, id: 2, position: 1 },
       { linkId: 2, parentId: 2, id: 3, position: 1 },
       { linkId: 3, parentId: 2, id: 4, position: 2 },
       { linkId: 4, parentId: 1, id: 2, position: 2 }
    ]
      Or, if you were using linkDataFetchMode "single", you would combine the node and link data into a single dataset like this:
          [
       { id: 1, position: 0, description: "Frame" },
       { parentId: 1, id: 2, position: 1, description: "Wheel", linkId: 1 },
       { parentId: 2, id: 3, position: 1, description: "Hub", linkId: 2 },
       { parentId: 2, id: 4, position: 2, description: "Tire", linkId: 3 },
       { parentId: 1, id: 2, position: 2, description: "Wheel", linkId: 4 }
    ]

      NOTE: It is also possible to create an unbound multilink tree - see Tree.linkData.

      Returns:
      Current linkDataSource value. Default value is null

    • setLoadDataOnDemand

      public ResultTree setLoadDataOnDemand(Boolean loadDataOnDemand) throws IllegalStateException
      Does this resultTree load data incrementally as folders within the tree are opened, or is it all loaded in a single request? Must be true if fetchMode is "paged"

      See the keepParentsOnFilter overview for special considerations when filtering a load-on-demand tree

      Parameters:
      loadDataOnDemand - New loadDataOnDemand value. Default value is true
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getLoadDataOnDemand

      public Boolean getLoadDataOnDemand()
      Does this resultTree load data incrementally as folders within the tree are opened, or is it all loaded in a single request? Must be true if fetchMode is "paged"

      See the keepParentsOnFilter overview for special considerations when filtering a load-on-demand tree

      Returns:
      Current loadDataOnDemand value. Default value is true
      See Also:

    • setMatchingLeafJoinDepth

      public ResultTree setMatchingLeafJoinDepth(Integer matchingLeafJoinDepth)
      This property allows you to specify the number of ancestor levels Smart GWT attempts to retrieve with each request, when using the built-in support for keepParentsOnFilter on loadDataOnDemand trees. See the keepParentsOnFilter overview for details.
      Parameters:
      matchingLeafJoinDepth - New matchingLeafJoinDepth value. Default value is 3
      Returns:
      ResultTree instance, for chaining setter calls

    • getMatchingLeafJoinDepth

      public Integer getMatchingLeafJoinDepth()
      This property allows you to specify the number of ancestor levels Smart GWT attempts to retrieve with each request, when using the built-in support for keepParentsOnFilter on loadDataOnDemand trees. See the keepParentsOnFilter overview for details.
      Returns:
      Current matchingLeafJoinDepth value. Default value is 3

    • setModelType

      public ResultTree setModelType(TreeModelType modelType)
      Selects the model used to construct the tree representation. See TreeModelType for the available options and their implications.

      If the "parent" modelType is used, you can provide the initial parent-linked data set to the tree via the Tree.data attribute. If the "children" modelType is used, you can provide the initial tree structure to the Tree via the Tree.root attribute.

      Note : This is an advanced setting

      Overrides:
      setModelType in class Tree
      Parameters:
      modelType - New modelType value. Default value is "parent"
      Returns:
      ResultTree instance, for chaining setter calls
      See Also:

    • getModelType

      public TreeModelType getModelType()
      Selects the model used to construct the tree representation. See TreeModelType for the available options and their implications.

      If the "parent" modelType is used, you can provide the initial parent-linked data set to the tree via the Tree.data attribute. If the "children" modelType is used, you can provide the initial tree structure to the Tree via the Tree.root attribute.

      Overrides:
      getModelType in class Tree
      Returns:
      Current modelType value. Default value is "parent"
      See Also:

    • setProgressiveLoading

      public ResultTree setProgressiveLoading(Boolean progressiveLoading)
      Sets progressive loading mode for this ResultTree. The ResultTree will copy this setting onto the DSRequests that it issues, overriding the OperationBinding- and DataSource-level settings, in cases where the use of progressive loading does not affect the correctness of the tree's paging algorithm.

      This setting is applied automatically by DataBoundComponents that have their own explicit setting for progressiveLoading.

      Note: This property only has an effect for fetchMode:"paged" ResultTrees.

      Parameters:
      progressiveLoading - New progressiveLoading value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      See Also:

    • getProgressiveLoading

      public Boolean getProgressiveLoading()
      Sets progressive loading mode for this ResultTree. The ResultTree will copy this setting onto the DSRequests that it issues, overriding the OperationBinding- and DataSource-level settings, in cases where the use of progressive loading does not affect the correctness of the tree's paging algorithm.

      This setting is applied automatically by DataBoundComponents that have their own explicit setting for progressiveLoading.

      Note: This property only has an effect for fetchMode:"paged" ResultTrees.

      Returns:
      Current progressiveLoading value. Default value is null
      See Also:

    • setResultSize

      public ResultTree setResultSize(Integer resultSize) throws IllegalStateException
      How many tree nodes to retrieve at once from each large set of children in the tree.

      Applicable only with fetchMode: "paged". When a paged ResultTree is asked for rows that have not yet been loaded, it will fetch adjacent rows that are likely to be required soon, in batches of this size.

      Note : This is an advanced setting

      Parameters:
      resultSize - New resultSize value. Default value is 75
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getResultSize

      public Integer getResultSize()
      How many tree nodes to retrieve at once from each large set of children in the tree.

      Applicable only with fetchMode: "paged". When a paged ResultTree is asked for rows that have not yet been loaded, it will fetch adjacent rows that are likely to be required soon, in batches of this size.

      Returns:
      Current resultSize value. Default value is 75
      See Also:

    • setSendNullParentInLinkDataCriteria

      public ResultTree setSendNullParentInLinkDataCriteria(Boolean sendNullParentInLinkDataCriteria) throws IllegalStateException
      For multi-link trees only, should we send up the parentId in fetch criteria if the criteria value is null? If false, we remove the parentId from the criteria when fetching link data, if the criteria value is null (as it will be by default when fetching the direct child nodes of the tree's root).

      Ignored for non-multiLink trees.

      Parameters:
      sendNullParentInLinkDataCriteria - New sendNullParentInLinkDataCriteria value. Default value is true
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getSendNullParentInLinkDataCriteria

      public Boolean getSendNullParentInLinkDataCriteria()
      For multi-link trees only, should we send up the parentId in fetch criteria if the criteria value is null? If false, we remove the parentId from the criteria when fetching link data, if the criteria value is null (as it will be by default when fetching the direct child nodes of the tree's root).

      Ignored for non-multiLink trees.

      Returns:
      Current sendNullParentInLinkDataCriteria value. Default value is true

    • setServerFilterFields

      public ResultTree setServerFilterFields(String... serverFilterFields) throws IllegalStateException
      For fetchMode:"local" ResultTrees, this property lists field names that will be sent to the server if they are present in the criteria.

      This property may be used to ensure a dataSource receives the necessary criteria to populate a ResultTree's data, and also support keepParentsOnFilter.

      Note that for some AdvancedCriteria it will not be possible to extract the subcriteria that apply to certain fields. See DataSource.splitCriteria() for details on how serverFilterFields-applicable subcriteria are extracted from the specified criteria for the tree.

      Parameters:
      serverFilterFields - New serverFilterFields value. Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getServerFilterFields

      public String[] getServerFilterFields()
      For fetchMode:"local" ResultTrees, this property lists field names that will be sent to the server if they are present in the criteria.

      This property may be used to ensure a dataSource receives the necessary criteria to populate a ResultTree's data, and also support keepParentsOnFilter.

      Note that for some AdvancedCriteria it will not be possible to extract the subcriteria that apply to certain fields. See DataSource.splitCriteria() for details on how serverFilterFields-applicable subcriteria are extracted from the specified criteria for the tree.

      Returns:
      Current serverFilterFields value. Default value is null

    • setServerKeepParentsOnFilter

      public ResultTree setServerKeepParentsOnFilter(Boolean serverKeepParentsOnFilter)
      If true, indicates that your own server code will handle the complexities associated with the combination of keepParentsOnFilter and loadDataOnDemand. If this flag is true and your server code does not handle those complexities, the results are undefined, but most likely you will simply exclude non-matching parents if your tree is load-on-demand, which effectively means that filtering will be broken.

      If this flag is not set, Smart GWT will use its own automatic client-driven algorithm to ensure that keepParentsOnFilter is honored on load-on-demand trees. See the keepParentsOnFilter overview for details

      Parameters:
      serverKeepParentsOnFilter - New serverKeepParentsOnFilter value. Default value is See Below
      Returns:
      ResultTree instance, for chaining setter calls

    • getServerKeepParentsOnFilter

      public Boolean getServerKeepParentsOnFilter()
      If true, indicates that your own server code will handle the complexities associated with the combination of keepParentsOnFilter and loadDataOnDemand. If this flag is true and your server code does not handle those complexities, the results are undefined, but most likely you will simply exclude non-matching parents if your tree is load-on-demand, which effectively means that filtering will be broken.

      If this flag is not set, Smart GWT will use its own automatic client-driven algorithm to ensure that keepParentsOnFilter is honored on load-on-demand trees. See the keepParentsOnFilter overview for details

      Returns:
      Current serverKeepParentsOnFilter value. Default value is See Below

    • setUpdateCacheFromRequest

      public ResultTree setUpdateCacheFromRequest(Boolean updateCacheFromRequest) throws IllegalStateException
      When a successful Add, Update or Remove type operation fires on this ResultTree's dataSource, if DSResponse.data is unset, should we integrate the submitted data values (from the request) into our data-set?

      Note : This is an advanced setting

      Parameters:
      updateCacheFromRequest - New updateCacheFromRequest value. Default value is true
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getUpdateCacheFromRequest

      public Boolean getUpdateCacheFromRequest()
      When a successful Add, Update or Remove type operation fires on this ResultTree's dataSource, if DSResponse.data is unset, should we integrate the submitted data values (from the request) into our data-set?
      Returns:
      Current updateCacheFromRequest value. Default value is true

    • setUseSimpleCriteriaLOD

      public ResultTree setUseSimpleCriteriaLOD(boolean useSimpleCriteriaLOD)
      Whether or not we should skip promotion of a simple criteria to an AdvancedCriteria when sending the DSRequest to load the children of a node in a loadDataOnDemand or fetchMode:"paged" ResultTree. If the DSRequest.textMatchStyle is not "exact", we normally convert the simple criteria to an AdvancedCriteria for correctness in matching the node name, but setting this property to true will allow that to be skipped for backcompat with older releases.

      Note : This is an advanced setting

      Parameters:
      useSimpleCriteriaLOD - New useSimpleCriteriaLOD value. Default value is false
      Returns:
      ResultTree instance, for chaining setter calls
      See Also:

    • getUseSimpleCriteriaLOD

      public boolean getUseSimpleCriteriaLOD()
      Whether or not we should skip promotion of a simple criteria to an AdvancedCriteria when sending the DSRequest to load the children of a node in a loadDataOnDemand or fetchMode:"paged" ResultTree. If the DSRequest.textMatchStyle is not "exact", we normally convert the simple criteria to an AdvancedCriteria for correctness in matching the node name, but setting this property to true will allow that to be skipped for backcompat with older releases.
      Returns:
      Current useSimpleCriteriaLOD value. Default value is false
      See Also:

    • applyFilter

      public Tree applyFilter(Tree tree, Criteria criteria, TreeFilterMode filterMode, DataSource dataSource)
      The ResultTree will call applyFilter() when it needs to locally filter the tree using the current filter criteria.

      Default behavior is to call Tree.getFilteredTree() to obtain a new, filtered tree.

      Override this method or Tree.getFilteredTree() to implement your own client-side filtering behavior. Note that the original tree should not be affected.

      Parameters:
      tree - the source tree to be filtered
      criteria - the filter criteria
      filterMode - mode to use for filtering
      dataSource - dataSource for filtering if the Tree does not already have one
      Returns:
      the filtered tree (copy)

    • applyFilter

      public Tree applyFilter(Tree tree, Criteria criteria, TreeFilterMode filterMode, DataSource dataSource, DSRequest requestProperties)
      The ResultTree will call applyFilter() when it needs to locally filter the tree using the current filter criteria.

      Default behavior is to call Tree.getFilteredTree() to obtain a new, filtered tree.

      Override this method or Tree.getFilteredTree() to implement your own client-side filtering behavior. Note that the original tree should not be affected.

      Parameters:
      tree - the source tree to be filtered
      criteria - the filter criteria
      filterMode - mode to use for filtering
      dataSource - dataSource for filtering if the Tree does not already have one
      requestProperties - Request properties block. This allows developers to specify properties that would impact the filter such as DSRequest.textMatchStyle
      Returns:
      the filtered tree (copy)

    • compareCriteria

      public int compareCriteria(Criteria newCriteria, Criteria oldCriteria)
      Default behavior is to call DataSource.compareCriteria() to determine whether new criteria is equivalent to the old criteria (returns 0) or not.

      See DataSource.compareCriteria() for a full explanation of the default behavior. The CriteriaPolicy used is "dropOnChange".

      Override this method or DataSource.compareCriteria() to implement your own client-side filtering behavior.

      Parameters:
      newCriteria - new filter criteria
      oldCriteria - old filter criteria
      Returns:
      0 if the criteria are equivalent, -1 if the criteria are different
      See Also:

    • compareCriteria

      public int compareCriteria(Criteria newCriteria, Criteria oldCriteria, DSRequest requestProperties)
      Default behavior is to call DataSource.compareCriteria() to determine whether new criteria is equivalent to the old criteria (returns 0) or not.

      See DataSource.compareCriteria() for a full explanation of the default behavior. The CriteriaPolicy used is "dropOnChange".

      Override this method or DataSource.compareCriteria() to implement your own client-side filtering behavior.

      Parameters:
      newCriteria - new filter criteria
      oldCriteria - old filter criteria
      requestProperties - dataSource request properties
      Returns:
      0 if the criteria are equivalent, -1 if the criteria are different
      See Also:

    • dataArrived

      public void dataArrived(TreeNode parentNode)
      This callback fires whenever the resultTree receives new nodes from the server, after the new nodes have been integrated into the tree.
      Parameters:
      parentNode - The parentNode for which children were just loaded

    • get

      public TreeNode get(int pos)
      Get the item in the openList at a particular position.
      Overrides:
      get in class RecordList
      Parameters:
      pos - position of the node to get
      Returns:
      node at that position

    • getCombinedCriteria

      public Criteria getCombinedCriteria()
      Returns a copy of all explicit and implicit criteria currently applied to this ResultTree.
      Returns:
      combined criteria

    • getCombinedCriteriaAsAdvancedCriteria

      public AdvancedCriteria getCombinedCriteriaAsAdvancedCriteria()
      Returns a copy of all explicit and implicit criteria currently applied to this ResultTree.
      Returns:
      combined criteria

    • getOpenState

      public String getOpenState()
      Returns a snapshot of the current open state of this tree's data as a TreeGridOpenState object.

      This object can be passed to setOpenState() or TreeGrid.setOpenState() to open the same set of folders within the tree's data (assuming the nodes are still present in the data).

      Returns:
      current open state for the grid. See TreeGridOpenState
      See Also:

    • getRange

      public TreeNode[] getRange(int start, int end)
      Get a range of items from the open list
      Overrides:
      getRange in class RecordList
      Parameters:
      start - start position
      end - end position (NOT inclusive)
      Returns:
      list of nodes in the open list

    • invalidateCache

      public void invalidateCache()
      Manually invalidate this ResultTree's cache.

      Generally a ResultTree will detect and incorporate updates to the DataSource that provides its records, but when this is not possible, invalidateCache() allows manual cache invalidation.

      Components bound to this ResultTree will typically re-request the currently visible portion of the dataset, causing the ResultTree to re-fetch data from the server.

    • setChildren

      public void setChildren(TreeNode parent, List newChildren)
      Replaces the existing children of a parent node. This leaves the node in the loaded state (unless a partially loaded set of children is specified using the optional totalChildren argument). The supplied array of children may be null or empty to indicate there are none, but if present must be in the standard format as would be sent from the server, as described by TreeDataBinding.

      In particular, note that for a paged ResultTree, each child node:

        :
      • can have nested children spcified under the Tree.childrenProperty (but not via TreeNode.id/TreeNode.parentId linking)
      • cannot be open unless it includes either a complete set of children, or partial set of children and a childCount
      Overrides:
      setChildren in class Tree
      Parameters:
      parent - parent of children
      newChildren - children to be set
      See Also:

    • setChildren

      public void setChildren(TreeNode parent, List newChildren, Integer totalChildren)
      Replaces the existing children of a parent node. This leaves the node in the loaded state (unless a partially loaded set of children is specified using the optional totalChildren argument). The supplied array of children may be null or empty to indicate there are none, but if present must be in the standard format as would be sent from the server, as described by TreeDataBinding.

      In particular, note that for a paged ResultTree, each child node:

        :
      • can have nested children spcified under the Tree.childrenProperty (but not via TreeNode.id/TreeNode.parentId linking)
      • cannot be open unless it includes either a complete set of children, or partial set of children and a childCount
      Parameters:
      parent - parent of children
      newChildren - children to be set
      totalChildren - number of total children (if not all have been provided as newChildren); only allowed if paging
      See Also:

    • setOpenState

      public void setOpenState(String openState)
      Reset the set of open folders within this tree's data to match the TreeGridOpenState object passed in.

      Used to restore previous state retrieved from the tree by a call to getOpenState().

      Parameters:
      openState - Object describing the desired set of open folders. See TreeGridOpenState
      See Also:

    • unloadChildren

      public void unloadChildren(TreeNode node)
      Deprecated.
      It's recommended that you instead use Tree.reloadChildren() to reload the children of a folder, or Tree.removeChildren() if you need to clear the cached children of a folder to add specific local data.
      Unload the children of a folder, returning the folder to the "unloaded" state.
      Overrides:
      unloadChildren in class Tree
      Parameters:
      node - folder in question

    • willFetchData

      public Boolean willFetchData(Criteria newCriteria)
      Will changing the criteria for this resultTree require fetching new data from the server or can the new criteria be satisfied from data already cached on the client?

      This method can be used to determine whether TreeGrid.fetchData() or TreeGrid.filterData() will cause a server side fetch when passed a certain set of criteria.

      Parameters:
      newCriteria - new criteria to test.
      Returns:
      true if server fetch would be required to satisfy new criteria.

    • setDefaultProperties

      public static void setDefaultProperties(ResultTree resultTreeProperties)
      Class level method to set the default properties of this class. If set, then all existing and subsequently created instances of this class will automatically have default properties corresponding to the properties set on the SmartGWT class instance passed to this function before its underlying SmartClient JS object was created. This is a powerful feature that eliminates the need for users to create a separate hierarchy of subclasses that only alter the default properties of this class. Can also be used for skinning / styling purposes.

      Note: This method is intended for setting default attributes only and will affect all instances of the underlying class (including those automatically generated in JavaScript). This method should not be used to apply standard EventHandlers or override methods for a class - use a custom subclass instead. Calling this method after instances have been created can result in undefined behavior, since it bypasses any setters and a class instance may have already examined a particular property and not be expecting any changes through this route.

      Parameters:
      resultTreeProperties - properties that should be used as new defaults when instances of this class are created
      See Also:

    • setDataSource

      public ResultTree setDataSource(DataSource dataSource) throws IllegalStateException
      What DataSource is this resultTree associated with?
      Overrides:
      setDataSource in class Tree
      Parameters:
      dataSource - dataSource Default value is null
      Returns:
      ResultTree instance, for chaining setter calls
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created
      See Also:

    • getDataSource

      public DataSource getDataSource()
      What DataSource is this resultTree associated with?
      Overrides:
      getDataSource in class Tree
      Returns:
      DataSource
      See Also:

    • setRootNode

      public void setRootNode(JavaScriptObject rootNode) throws IllegalStateException
      This attribute may be used to specify a root value for the parentIdField of this resultTree. This overrides the default rootValue for this tree, allowing a component to navigate a tree starting at a specific node.

      May be overridden via treeRootValue for ResultTrees generated by a TreeGrid component.

      Parameters:
      rootNode - rootNode Default value is null
      Throws:
      IllegalStateException - this property cannot be changed after the underlying component has been created

    • getRootNode

      public JavaScriptObject getRootNode()
      This attribute may be used to specify a root value for the parentIdField of this resultTree. This overrides the default rootValue for this tree, allowing a component to navigate a tree starting at a specific node.

      May be overridden via treeRootValue for ResultTrees generated by a TreeGrid component.

      Returns:
      Object