On Fri, Sep 2, 2011 at 10:35 AM, John Zabroski <[email protected]>wrote:
> For your own goals, check out Montanari's work on representing concurrent > computations as contextual nets. Contextual nets are used to close off > feedback loops so that hard problems like "weakly" and "meagerly" specified > computations do not interfere with the overall goal of a program. > I assume you refer to Ugo Montanari's 'Contextual' Petri Nets, which basically extend petri-nets with guards or readers. I will read about them, but it is not immediately clear to me how they are relevant to my goals or RDP. Can you explain your thoughts? > I generally like the idea of representing state as a database of values. > Pushing state to the 'edges' of the programming model, such as into a database, is very valuable when you want resilience, persistence, easy extension (with new observers or agents), or to tweak behavior at runtime. Smalltalk is not so good for for live programming, IMO, because it is difficult to refactor or reorganize state between objects, and because messages-in-transit is itself a form of implicit state that is difficult to adjust smoothly. I would imagine that the code tends to get out-of-sync with the runtime, such that developers need to know the whole edit-history of the project in order to fully understand its current behavior. Ideally, we should be able to reason about a program's behavior in strict terms of its current code and some highly visible state - and this goal is met only by keeping relevant state in some sort of 'external' system (such as a database or abstract register machine). > Round 2 of my question would be, What can we learn from the problem solving > approaches used by each of these domains? Can the tools and techniques be > combined to solve design contradictions well-known in one domain but totally > shadowed by the taking the perspective from another problem domain? > Those are very open-ended questions. If you have particular design contradictions in mind, I'm sure you could find ways to resolve them. Dave
_______________________________________________ fonc mailing list [email protected] http://vpri.org/mailman/listinfo/fonc
