- From: Assaf Arkin <arkin@intalio.com>
- Date: Fri, 11 Apr 2003 09:56:41 -0700
- To: Martin Chapman <martin.chapman@oracle.com>
- CC: "'Burdett, David'" <david.burdett@commerceone.com>, "'Cummins, Fred A'" <fred.cummins@eds.com>, jdart@tibco.com, public-ws-chor@w3.org
I agree. We have different opinion on how much capabilities we want in the language in terms of what it can describe. Some want a language that is only capable of expressing shared states that are general enough but satisfactory for most B2B use cases. Others want more capabilities that *in addition* to the above and not as replacement, also support more complex scenarios, e.g. the ones you would see in A2A or optionally exposing part of the white/black box. We can go either way, but if we fail to consider the importance of the "glue" requirement, I have the feeling we will end up with a superb specification that has no practical use. arkin Martin Chapman wrote: >Maybe the answer is staring us in the face: > >When we talk about external definitions we seem to be implying a shared >state machine. There is a common understanding between all parties about >who is playing what role, what states each role can get into, the >(shared) state of the process itself etc. [is this what is commonly >called a global model?] > >On the other hand an internal defintion seems to define a state machine >that is not shared (tho may or may not be visible i.e. could be black >box or white box). So what states you need, what tranistions you make, >how you name other parties is totally private. > >Obviously the two have to be glued together at some point. >As a group we have to decide whether we are working on shared state >machines vs priavte ones, or both. In all case we will have to look at >the "glue" requirements. > >Martin. > > > -- "Those who can, do; those who can't, make screenshots"
Received on Friday, 11 April 2003 12:58:25 UTC