Plans for the Development of CASCADE

This document outlines many of the features of CASCADE, both those currently in operation and those in the planning phase. It is important to note that while CASCADE has reached a state where it is viewed by many as an operational platform. In reality, it primary function is still as a testbed for conducting research on collaborative authoring, document structuring, network protocols, and agent design. Related to the research testbed focus, there are three important things that should be made explicit:

• The system currently has two forms. An ``X/Unix'' form and an ``SQL/RPC'' form. The X/Unix form is a large executable based on more than 200 modules that does all of the CASCADE work based on information stored in the Unix file system and a series of flat files. The interface is X/Motif and is served from the application across the network to a workstation. This form of CASCADE is relatively easy for developers to work with, can be quickly modified, and can have new features installed in minutes. At the same times, it does not scale well, nor does it work well over congested networks. For that purpose, CASCADE is in the process of being redesigned as a true client server application. The code has been split into client and server portions and the server in being redesigned piece by piece to replace the Unix and flat file information with information from an SQL compliant database (Oracle) residing on the server. Instead of processing all of the information on the server and shipping X to the clients, the SQL version will use RPC protocols. The clients will be manifest to the users as interfaces written in the native form of the platform they use -- Windows, X, Mac. While the X/Unix prototype is useful for demonstrating general functionality, it will be abandoned in the future for two reasons:
  • its performance suffers over a wide area network
  • even though and PC or Mac can serve as an X client, most users prefer a familiar client interface

• Because CASCADE has been developed through the contributions of graduate students who come and go, there are some pieces of it that are less well refined than others. As a matter of fact, there are numerous places where work abruptly stopped and the research team has failed to identify an interested graduate student to go back and finish the work. The Mail system is a good example of this. A reasonable interface has been set up, and the core functions have been worked out, but a number of little pieces are simply stubs with no functionality attached. The interface is less than optimal in other ways. For example, color is very important in CASCADE and we have built in significant color capabilities that are controllable in a variety of ways by the user. In the final analysis, research will be done on which color combinations are recommended for what purposes. At the current time, the color scheme reflects personal preferences, or random assignments, etc. As you go through CASCADE you may see some of the options set to arbitrary values. While these are under the control of the end user where that is important -- care has been taken to identify these places -- it may not be immediately apparent that this is the case.

• CASCADE is still under construction. For that reason it may not match the exact description you find here or in other documents. In addition, as in any building project, there are construction tools laying around the site. We have tried to collect these tools under the development menu, but some of them creep out into other places. In addition, despite our best efforts at modularity, one minor improvement in the editor may have unexpected effects elsewhere.