MDSplus Physical Subtrees

Creating and Using Physical Subtrees

The MDSplus data structure is meant to be understood as a tree, and logically, any branching node defines a subtree consisting of itself and all the nodes beneath it. However, for purposes of data management and for defining permissions/access control, some of the logical subtreees can also be located in separate files. These are called "physical subtrees" in this documentation. When examining an MDSplus tree with the Traverser application, green "tree" icons indicate the physical subtrees.

A physical subtree structure is very similar to a top-level experiment tree. It is contained in .TREE, .DATAFILE and .CHARACTERISTICS files in the file-system, and it may have physical subtrees of its own.

The physical subtree is known to MDSplus via the logical name (subtree)$DATA, just as the top-level tree is known by (experiment-name)$DATA, i.e. CMOD$DATA or NSTX$DATA. The 12-character node-name length limit carries over to the subtree name, since they must match. It is often convenient, but not necessary, to locate the files for the physical subtrees in a directory immediately below that for the files of the parent tree.

There are many I/O efficiency advantages to opening physical subtrees in order to read files, rather than opening the top-level tree and following a long path to read a node in a subtree. Whenever a tree is opened, all of its physical subtrees are also opened; so opening the lowest convenient subtree minimizes the number of files involved. It also saves a lot of typing.

To create a new physical subtree,

NOTE: It is NOT necessary to use the LINK command, which appears to have been superceded by the SET NODE /SUBTREE command.

The nodes of the subtrees are not available when editing the upper-level tree. (They are not indicated at all.) To verify that the connection has been successful, open the higher-level tree for readonly (using SET TREE in TCL, or TRAVERSER in normal mode).