The Future of PFC

Once I have created a business object framework module with a well defined interface to the outside world I would work on plumbing. The glue section would be the interface module. The goal of the interface module is to translate the business object interface to the protocol used by the UI layer. As mentioned the UI layer was the PFC as it exists today, but I would probably choose to make it work with plain PowerBuilder, simply because it would not require any additional effort. But why stop there? Additional interface modules could be created to talk to thin client HTML, Java and even VB front-ends.

I would need to make sure that my Business Foundation Class would work with most common distributed protocols. Base PowerBuilder has that functionality in Component Factory. PowerBuilder NVO can be deployed as a JavaBean or a COM component. The challenge is in the data transfer. Get/SetFullState, Get/SetChanges deal with BLOBs . This would have to be customised for each protocol...

Wait a minute! Time for a wake-up call! I just noticed that in my utopian vision of the PFC's future I did not include any major enhancements to the PFC as it exists today. Instead I have chosen to add a separate class library. Oh my goodness! I myself have chosen to "freeze" PFC features.

The fact is that PFC is a mature class library with a well defined interface and methodology. Many third party vendors but more importantly thousands of PFC users have built upon PFC and used it as a foundation for their applications. Is there room for enhancements and new services? Definitely! Is there room for major architectural changes? I don't think so.

PowerBuilder is no longer just PowerBuilder. PowerBuilder today includes PowerSite, PowerDesigner Appmodeler, and Web Deployment Toolkit. The new or extended class library would have to encompass the new products. If I had limited resources I would spend them to design a class library that would provide the distributed server framework and all the glue logic to connect the pieces now included in PowerBuilder, specifically PowerSite, Distributed PowerBuilder, COM and JavaBean objects, and Jaguar.

This would bring me to another dilemma. There is an immense effort currently in progress of PowerBuilder and Jaguar integration. Both PowerBuilder and Jaguar are in a state of transit. The reality is that before a framework is designed the feature list of the base product must be frozen.

This is happening right now in PowerBuilder 7, which should be in beta by the end of the year.

What does that mean for PFC users? Since version 3.0 Datawindow has seen a number of enhancements, dwModify() changed to Modify(), Direct Data Manipulation was introduced, Get/SetChanges() Get/SetFullState() functions were added. But over 3 versions new functionality added to the datawindow probably constitutes less than 2% of total Datawindow features. They have added a few new interface functions but over the period of 4-5 years the bulk of Datawindow functionality has remained the same. Does that mean that Sybase was trying to phase out the datawindow? Probably not. A more plausible explanation is that datawindow was considered feature-complete for the purpose it was designed to serve. Enhancements were made elsewhere. Resources that could have been used on the datawindow enhancements have brought us a datastore instead.

Times have changed. Sybase is feeling pressure from other tool vendors. PowerBuilder users are nervous. However we have a responsibility to our employers and clients to take a deep cleansing breath, use common sense to examine the direction of Sybase tools, and make a proper recommendation. For open architecture, client / server, distributed, multi-tier, and Internet projects PowerBuilder and PowerJ are still my tools of choice. For a well documented, well designed, thoroughly tested, industry standard PowerBuilder Class Library PFC is still my first choice.

Conclusion

What does the future hold for PFC? Can the rumours be true? Given the effect it would have on their user community and in the end on their flagship product that is highly unlikely. There is only one way Sybase can prove the renewed dedication to PowerBuilder. PFC or a similar class library should be used to provide a roadmap to the distributed partitioned architecture and to help us manage the complexity of the new environment. The need for the roadmap class library has recently doubled with the addition of new technology and varying environment.

Some see the glass as half full, some see it as half empty. I see some water which was used to putout a small fire in order to save the next year's crops. Is there a future for PFC? I believe that future is guaranteed by the number of it’s users. Will I use PFC on my next project? Absolutely !

about the author