Title: Message
There is a model of SCA in UML at
(1) http://www.osoa.org/display/Main/SCA+Expressed+as+a+UML+Model
attributed to Mike Edwards
And I even found a link this week what
might be a critique of SCA terminology from a UML “purist” (sorry
Duane)
(2) http://technoracle.blogspot.com/2007/09/soa-anti-patterns-service-composition.html
that would change at least some of the arrows decorated with solid
diamonds of composition in (1) to hollow diamonds of aggregation! For example,
destroying a Composite in (1) would presumably not necessarily destroy the
Components.
A google of “SCA UML” already yields a number of
entries also!
So I second Martin that some issue be raised that calls for a TC
review of a sufficiently accurate SCA model in UML 2.0 for the group to have at
least as background. Unfortunately I don’t know exactly what in the
submission to tie this issue to.
From: Martin Chapman
[mailto:]
Sent: Thursday, September 27, 2007
11:20 AM
To: 'Michael Rowley'; 'Jeffrey A.
Estefan';
Subject: RE: [sca-assembly] SCA
Visual Modeling "Standard?"
Sounds like we should raise an issue and
chat about it there.
Martin.
-----Original Message-----
From: Michael Rowley
[mailto:]
Sent: Wednesday, September 26,
2007 8:54 PM
To: Jeffrey A. Estefan;
Subject: RE: [sca-assembly] SCA
Visual Modeling "Standard?"
I’m not a UML expert, but I believe
that UML’s concept of a component maps to SCA’s concept of an
“implementation”, not to SCA’s components.
For example, in UML, could a single Java
class be represented as multiple components in one component diagram (each
configured differently), as can be done in SCA?
Michael
From: Jeffrey
A. Estefan [mailto:]
Sent: Wednesday, September 26,
2007 3:30 PM
To:
Subject: [sca-assembly] SCA Visual
Modeling "Standard?"
Martin & Mike,
I would like to know why an visual modeling
industry standard such as OMG UML 2 was not used (and currently not being
used) to represent SCA artifacts; specifically, SCA Component and SCA
Composite diagrams. It seems a no-brainer to leverage the UML 2
component diagram to represent SCA components and UML 2 composite structure
diagrams to represent SCA composites. The current diagram formats used in
all SCA specs from Open SOA (and now under the auspices of OASIS) seem to use
custom diagram semantics to represent services, references, and properties when
UML 2 provided and required interfaces and the use of ports would suffice just
fine for both SCA components and composites.
I ask because there is probably some history to this
decision and since I did not participate in development of the Open SOA specs,
I'm curious as to why an industry standard such as UML 2 was not used and if it
is worth considering for use in the OASIS version of these specs;
particularly, since these are architecture-centric artifacts.
Regards...
- Jeff Estefan, JPL
