Re: [soa-rm] On shared state

[Frank McCabe <frank.mccabe@us.fujitsu.com>]

Jeff:
  Well, I would be OK with the current wording. I simply suggested  
the new paragraph in order to help with explanation.
  As far as the other points are concerned:
1. coupling. in an RA discussion we had already decided on the  
'correct' interpretation of loose vs tight coupling: a system is  
loosely coupled if the constraints on their interoperation is at a  
logical minimum. A system is tightly coupled if there are additional  
constraints which do not directly contribute to their interoperation.  
As I see it, this does not directly impact the issue of shared state.
2. view. I can see the potential for confusion; but English is a  
blunt instrument that forces us to use  a mallet where a pin-hammer  
is appropriate.
3. shared state. I agree that this has been the subject of much  
discussion. However, in my observations of industry, I would be  
forced to conclude that a lot of people still 'don't get it' :)
Frank


On Apr 20, 2006, at 3:28 PM, Jeffrey A Estefan wrote:

> Frank,
>
> I must admit, my brain is starting to get a bit saturated with  
> respect to
> this issue surrounding "shared state."  I hope we can put it to bed  
> soon.
>
> Two things really worry me.  One is the temptation to introduce yet  
> another
> highly overloaded term such as "view" into the potential re-wording  
> of the
> draft spec.  I would encourage use to refrain from the use of view  
> (and
> "viewpoint") as these terms will influence our SOA-RA work in an  
> ANSI/IEEE
> 1471-2000 Std. context and use of such terms should be consistent  
> with such
> a normative reference.
>
> Second, shared state (including "distributed shared state") is  
> certainly not
> a new idea and a subject of extensive research in academia and  
> industry (see
> the various IEEE articles as an example).  If this is really what  
> we mean,
> then we should just leave the text as is because further  
> elaboration will
> likely result in more confusion.  What bothers me a bit is the  
> potential
> connection between shared state and coupling (i.e., tight coupling).
> Remember the issue of coupling has been the subject of our RA  
> discussions &
> debate as of late.  Many of the existing SOA pundits out there  
> believe that
> loose coupling and stateless interactions are at the heart of SOA
> architectural style interaction semantics, and they even go further  
> to say
> they are the heart of SOA best practices, period.  So how do we  
> address
> "shared state" without implying some element of coupling?  I do not  
> have the
> answer at this moment but hopefully, you and others on our team  
> might have
> that answer!
>
> Cheers...
>
>  - Jeff
>
>
>