(Continued from
the previous post)
Howard Smith posted a
revision of the "Workflow is just a pi-process" paper today. It is a very worthwhile paper to read, but it needs a different title ("Pi-C as the Basis for BPMS"?), as the main assertion in the title is still useless (except for generating debate).
While pi-c and its variants may have the potential to form the basis for BPMSs, there is a lot of work left to be done. I am not aware of a single piece of 'rigorous' work on how pi-c can be applied to BPM addressing all aspects - completeness in terms of ability to model BPM concepts, modeling efficiency, reasoning and execution complexity, analytical capabilities, and others (I wonder why no professor has taken up this challenge). Any pointers to such works will be highly appreciated.
Meanwhile, vendors will build their 'process virtual machines' in their own ways to support the chosen standard language such as BPEL. There is no compelling reason to base one's process virtual machine on pi-c without understanding the benefits of doing so.
