1

Topic: Re: Reactive Programming and Bayesian programming

Hello, Tyomchik, you wrote: Those> the code written by the colleague Got to me. There RxJava it is used. I a little  and smoked it to dock on RxJava a little. This piece is pleasant! This introduction. Now thought - in the theory there Observable and push. Whether It is possible easily and to construct easy behavior tree where observable - it is a node? For example, the selector and at it child nodes - is subscribers. Whether it is possible besides, easily and easy to construct a tree where nodes-selectors provide conditional probability, and all it on lambdas, with a code minimum? It is possible. Moreover - and it is necessary to do.  changes, behind it all branches on which are subscriptions. Truth thus suffers performance and the storage expenditure.

2

Re: Re: Reactive Programming and Bayesian programming

Hello, Tyomchik, you wrote: I>> It is possible. Moreover - and it is necessary to do.  changes, behind it all branches on which are subscriptions. Truth thus suffers performance and the storage expenditure. Those> About performance not  it is clear, for it behavior tree - it does 1 step on each tic. There 1 controlling flow which sends "tics". And about rxjava, how many tried to cope - with it at it an unresolved problem with synchronization of a source of events and the receiver. If the receiver in other flow also is not in time to process - queue from events expands, does not fall down yet with OutOfMemory. For a method to force a source to wait, while the receiver rakes beklog - is not present. Tics can be sent on everyone, depends from . Performance suffers at the expense of horse costs for a portion of the data. That is, RX it is necessary for binding. backpressure like as it is already implemented, but I not strongly track for RX.