[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [wsbpel] Link semantics and control dependencies
Hello: > I believe that the following two fragments of the spec. > contradict each other and I would like to gather second > opinions about it: Most of your arguments revolve around using links that cross into a switch statement. My gut feeling is that this is an ambiguous construct and probably should be outlawed in WS-BPEL, or at least frowned upon. Like a previous post, and if I understand the example, I will argue that the switch can be replaced with its all link equivalent.I do this in order to illustrate how I think DPE comes into play. SW (T1 ->) A2, SW (T2 ->) A3 -> link (we don't care about its name) T1 and T2 are transitionConditions and for completeness A2 -> Dummy A3 -> Dummy Dummy joinCondition("True") In this scenario, if T2 evaluates to True (and T1 evaluates to False), A3 still cannot start because A1 is not finished. When A1 *finishes*, A3 can evaluate all of its links and start. Your mistake is: "the status of link x2 is then set to negative since branch A3 was not taken." You forget that A3 cannot start executing until A1 is complete and link x1 is evaluated. When A3 is finished and sets a value on link X2, A4 can start. Note the logical order is correct, A1 finished before A3, and A3 finished before A4. There are no contradictions. Marlon, the main thing I get out of your example is that the specification is not clear on how switch operates. I guess the correct behaviour of switch and sequence in these circumstances by converting them into their all link equivalents and working from there. That said, I find the DPE algorithm elegant. Cheers, Andrew
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]