--- In [email protected], "Anne Thomas Manes" <[EMAIL PROTECTED]> wrote: > > I think BPEL is fundamentally flawed. I don't think an execution > language is the right way to manage orchestration. It leads to > centralized orchestration engines. I much prefer a distributed model > in which the next orchestration step is determined by the current > state of the process rather than a predefined sequence of steps.
I have been reading Keith's book, Human Interactions, and it would seem that this sort of model would be entirely in line with Human Interaction Management Systems (HIMS) on the basis that people do not always undertake a series of actions in a rigid, pre-ordained sequence. This would apply to many business processes in practice, given human nature. Gervas > > In any case, BPEL is so constrained in its capabilities that almost > all vendors have had to extend it to make it do more than just simple > orchestrations, and these extensions destroy any potential value that > BPEL might provide as a "portable" process execution language. These > issues will be somewhat mitigated in BPEL 2.0, but not enough to make > it portable. But I still don't see a time where a single BPEL process > might be executed by multiple distributed orchestration engines. The > whole model is just wrong. > > But given that we have it, just what value does it give us? It doesn't > enable distributed execution. The only real value that it might provide > is portability of processes across process engines. But given that > everyone extends BPEL thta value goes away. > > And no one in their right mind would attempt to write BPEL manually. > The typical development approach is to use a modeling tool, such as > UML, BPMN, or a domain-specific modeling language to define the > process. This model then gets rendered into extended BPEL, which > in turn often gets compiled into something native to the execution > engine. So I really don't see any value to BPEL. > > Would would be more useful would be to define a standard XMI > representation of BPMN. > > Anne > > On 1/11/07, Peter Walker <[EMAIL PROTECTED]> wrote: > > > > --- In [email protected]<service-orientated-architecture%40yahoogroups.com>, > > "Anne Thomas > > Manes" <atmanes@> wrote: > > > > > > Not from me ... > > > > > > My advice it to avoid BPEL. > > > > Anne, > > > > Please say why. Are you saying avoid in *all* circumstances? > > > > Peter. > > > > > > > > On 1/10/07, Steve Jones <jones.steveg@> wrote: > > > > I completely agree that its got nothing to do with SOA, unfortunately > > > > there is the fact that lots of companies will now be looking to buy > > > > one of these new shiny products for their SOA implementations. I'm > > > > already seeing people say things like "I have to be using BPEL in my > > > > SOA", because they've read about it from an analyst. > > > > > > > > > > > > <snip> > > > > > > >
