When Pipi 6 was finished, it was already overdue to be replaced by what I had learned by building it.
The most significant change was adopting a Microservice Architecture. Each microservice has a separate database, logic, workflow and web presentation layer. There were 300 microservices and lots of drawings inside 6-sided polygons.
I must have left a big smoke trail of my Google searches because Google started phoning me every two months. Initially, I ignored them, but they wanted to host Pipi on GCP.
So, I optimised the internal architecture for cloud hosting on GCP using their platform tools.
A rules engine was built to handle logic. CQRS messaging that included persistent data storage.
The self-documentation of the platform used ColdFusion.
Roger Sessions on complexity. Chris Richardson's writings on Microservices. Santa Fe Institute (SFI) on emergence. Quanta Magazine on materialist science. Perimeter (University of Waterloo) on philosophy of science.
It looked good, and then it was ready for a complete rebuild (not even a refactor). The microservices architecture wouldn't scale because of the growing complexity caused by the Covert cell simulator-derived components. Docker wouldn't work. It was now going to require a Virtual machine (VM).