Python NetWorkSpaces and Parallel Programs

The Web Interface

NetWorkSpaces and the variables they contain exist independently of particular client programs. Examining the contents can be helpful when debugging or understanding a program, or monitoring the progress of an application. NWS includes a web interface that displays the workspaces and their contents. Figure 1 is a list of NetWorkSpaces; if you clicked on "test," you'd see something like Figure 2. By clicking on a variable, you can see a list of the current values (Figure 3).

[Click image to view at full size]

Figure 1: List of NetWorkSpaces.

Figure 2: Results of clicking "test".

Figure 3: Clicking on variables displays a list of current values.

Since values are serialized objects not easily interpreted by Python, how can the web interface display values that come from other languages (R vectors, octave matrices, and so on)? The translation is performed by language-specific "babelfish" services that understand how to deserialize and display their own datatypes. Babelfish are just ordinary clients that get translation tasks and return results via NWS.

As you can see, the interface provides a limited capability to alter the NWS state by deleting workspaces and variables. Appropriately designed code can make good use of this as a signaling mechanism. We use it, for example, to provide a web-intermediated process initialization for certain deployments of the Sleigh system.