Re: Simple Choreography composition suggestion

On Thursday, July 17, 2003, at 12:46  AM, Haugen Robert wrote:

>
> Andrew Berry wrote:
>> I'd urge caution whenever you use the terms
>> "global" and "state" in the same phrase.
>> If you're talking about modelling then it's probably OK,
>> but run-time state shared between participants
>> connected via the Internet is a recipe for software
>> that will never perform or scale.
> [...]
>> There are co-ordination approaches that can operate
>> on a local, partial view of the process state.
>> HP have done this in their labs with some
>> of their workflow tools.  My PhD thesis provides
>> another solution.
>
> It is feasible to *align* (not *share*) state among
> participants in Internet business transactions.
> Several standardized protocols exist including
> OASIS-Business Transaction Protocol (BTP),
> ebXML BPSS, UNCEFACT-BCP, and RosettaNet.
>
> At the end of a business transaction, one can say
> with a high degree of reliability that the participants
> will agree on a common state of (for example) an order.

Agreed and understood.  The reason for caution is that a choreography 
language could be used to describe interaction in a generic form rather 
than using a pre-defined protocol with known predicates (e.g. "end of 
transaction") that can be implied from the completed interactions in 
those examples.  If the underlying model for the language relies on 
global state, you still have a serious problem because there is 
potential for requiring all-party synchronisation at every step in the 
execution of a program expressed in that langauge.

Ciao,

AndyB

Received on Wednesday, 16 July 2003 18:10:33 UTC