Simple Choreography composition suggestion

Dear Colleagues,
 
I suggested a simple choreography composition scenario during last week's
conference call.   I think I remember Steve making a general plea for a
write up -if only for the archives!
 
Thinking about it some more I can think of two possibly distinctive simple
situations where composing two (or more) choreographies into a larger single
one may usefully apply - or the reverse of taking a choreography and
breaking it into smaller 'component' choreographies.  These are illustrated
in the attached slide.
 
The top picture illustrates the situation where we start with two
choreographies and make them into one by adding extra parties and using at
least one of the parties as a 'hub' in which events triggered by one
choreography cause a message to flow in another.
(Note:  I have labelled them as System A, B and C with B being the 'hub' in
this case.  The use of the word system is partially justified as it when a
message actually exits or enters a system that we are interested, but it is
clearly not completely correct and should be replaced by party / role (where
B, for instance, is one party that plays two roles, one part of the Order
choreography and the other part of the Stock inquiry choreography - or
whatever terms we decide upon - is this an issue or one that already has a
resolution???)
 
The bottom picture illustrates the 'extension of a choreography.  It
introduces new messages, but no new parties or roles. In this example we
have a quote choreography, which is then extended with an order choreography
(or an order choreography, which is extended with a quote choreography).
 
Further thoughts:
1)  I do not necessarily expect there to be any neat 'maths' to allow us to
automatically compose or decompose choreographies - there may be in some
cases, but in general the process may involve some human manipulation (and
common sense).
 
2)  I take it that the smallest choreography involves just two parties
(roles, or whatever we call them) and one message flowing between them.  It
should be possible to build up any other choreography from this 'molecule'
(molecule rather than atom  as there are two parties c.f. Hydrogen atoms (~
BPEL) and Hydrogen molecules?)
 
3)  If this is true then it means we already have a shed load of use cases -
take any of the existing use cases and decompose them - then re-compose
them.

Best Regards,

Tony                           


 <http://www.choreology.com/> 

Tony Fletcher

Technical Advisor 
Choreology Ltd.
13, Austin Friars, London EC2N 2JX UK


Phone:  

+44 (0) 20 7670 1787


Mobile: 

+44 (0) 7801 948 219


Fax:    

+44 (0) 20 7670 1785


Web:

 <http://www.choreology.com/> www.choreology.com


CohesionsT


Business transaction management software for application coordination



Work: tony.fletcher@choreology.com 


Home: amfletcher@iee.org