On 11/12/2013, dunnlow <[email protected]> wrote: > Hmm, whelp, no, nothing empirical. I think in my head I am combining a few > things I've read. > > In the Developers Cookbook (pg130) there is a comment that entry points can > help reduce cross products and can be used to partition memory. I remember > reading that the entry points are built during compile time, so I read that > to mean that if I have 1000 rules, but only 1 rule with entry point "ep1" > that if I insert a fact to ep1, the engine has already built the structure > to only evaluate that 1 rule. I thought that was the main reasoning behind > entry points :-/ You are saying that my thinking here is incorrect?
Entry points (or streams) have been added for representing different event data sources. I would not think highly of such a feature if its foremost intent would be memory optimization, i.e., passing the implementor's buck to the user. Now entry points may have the effect you have concluded from these statements - your reasoning may very well be correct. But there's no doubt that entry points are a static, compile time issue, and you are reporting a problem due to dynamic variations in the data. Using an iffy optimisation and trying to come to grips with it illustrates Michael A Jackson's pair of rules for optimisation ;-) To summarize: I wouldn't go up this alley unless I had data that would show it's worth the effort. > > So, in the manual > (http://docs.jboss.org/drools/release/6.0.0.Final/drools-docs/html/DroolsComplexEventProcessingChapter.html#d0e9947) > during the banking example for entry points, there is little difference > between having an "ATM Stream" and "Branch Stream" as shown and adding a > "stream" attribute to the WithdrawRequest object and adding the constraint, > stream=="Branch"? Given the problems arising from changing the SW at the *source* of these streams to accomodate the additional property: there is a big difference. As for the efficiency of the rules, I'd hold my judgment until after the benchmark :-) Cheers Wolfgang > > Thanks again > > > > laune wrote >> What makes you think that using entry points "speeds things up"? The >> Engine still has to evaluate the entry-point attribute of a fact >> (handle), >> just like any other attribute. >> >> Or do you have conclusive evidence based on solid benchmarks? >> >> -W > > > > > > -- > View this message in context: > http://drools.46999.n3.nabble.com/Insert-to-non-existing-entry-point-causing-NPE-in-drools-server-tp4027204p4027226.html > Sent from the Drools: User forum mailing list archive at Nabble.com. > _______________________________________________ > rules-users mailing list > [email protected] > https://lists.jboss.org/mailman/listinfo/rules-users > _______________________________________________ rules-users mailing list [email protected] https://lists.jboss.org/mailman/listinfo/rules-users
