Lambda the Ultimate

Overloading is hard to avoid sometimes, the first try was ‘Synchronic Net’. A better search term might be ‘Synchronic Computation’. The details of SSD have not been published yet, a pdf summary of the work that led to SSD is here. An external scheduler for non-determinism would indeed be less awkward, I was trying to address the concerns of process algebraists who take non-determinism to be fundamental.

Although I was familiar with Kahn Process Networks (KPN), the other references were less so and am grateful for them. KPNs and other dataflow/datastream models are mid level structures whose nodes are understood to be Turing Machines or Von Neumann processors, and program control is based on data tokens. I felt their level of abstraction was too high and obliged coders to view every task as a stream based program, so developed a more general purpose bit level machine model called the Synchronic A-Ram, where streams can implemented as higher level constructs.

Instances of the motivating examples for static single assignments are not impossible in the interlanguage environment mentioned in the summary, but are rendered far less likely by encoding dataflows as interstrings, which are the basic program syntactic structures. (More overloading -- there is no connection to Selinker’s linguistics concept of an interlanguage as an intermediate form in second language acquisition.) My initial impression is that Lightweight Time Warp is an interesting optimization for running simulations on processor networks, and will need more time to absorb CRDTs, and determine their connections and overlaps with the restrictions in interlanguage.