opencsa-liaison — archive

[Mike Edwards <mike_edwards@uk.ibm.com>]

Folks,

I'm happy with either:

a) Everything in one "SCA"
namespace

or

b) Each binding in its own namespace,
with "blessed" versions in the SCA namespace also

I believe that b) is in effect the approach
suggested for new, initially non standardized bindings
(eg binding.json from the Tuscany project),
which are subsequently standardized.

The mechanics of b) do need investigation,
but I think that at worst a complete repeat of the
definitions in the two namespaces is
the answer.



Yours,  Mike.


Strategist - Emerging Technologies, SCA & SDO.

Co Chair OASIS SCA Assembly TC.

IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain.

Phone & FAX: +44-1962-818014    Mobile: +44-7802-467431  

Email:  mike_edwards@uk.ibm.com



"Michael Rowley"
<mrowley@bea.com> 
25/03/2008 21:35




To
"Anish Karmarkar" <Anish.Karmarkar@oracle.com>,
<opencsa-liaison@lists.oasis-open.org>


cc



Subject
[opencsa-liaison] Namespace for bindings
and other extension points (was: Latest/This Version URI for Schema/WSDL
files)






 

Good point Anish.  I suspect that one of us was indeed supposed to
bring

this up (I don't recall who, if anyone, was identified).  So, how
about

me.


Dear Liason Committee,


The Bindings TC would like guidance on the namespace to use for the

various <binding.xxx> elements that it is in charge of defining.

Specifically, the question is whether the bindings should always use the

same namespace as SCA assembly, or whether they should each use

different namespaces.


The Bindings TC debated this question for a while at its F2F, but agreed

that the approach taken should follow a generally agreed approach that

would also apply to all of the extensibility points in SCA assembly

(such as implementation elements <implementation.xxx> and interface

elements <interface.xxx>).  As such, we think this is an appropriate

issue for the Liason group to tackle.


Argument Kickstart:


At the F2F, we discussed the pros and cons of a few approaches.


Each binding gets its own namespace:

- This approach allows each binding definition to evolve independently

from other binding definitions and independent of SCA as a whole.


Everything in one "SCA" namespace:

- This approach gives the user of SCA a set of technologies that are

known to work together.  If each binding/implementation/etc evolved

independently, then the user would be hard pressed to figure out which

collection of them actually worked together.

- Having one namespace means that there are fewer prefixes to define at

the top of the various SCDL files (this seemed to carry less weight than

the previous point).


Both:

- Perhaps it is possible to define bindings/implementations/etc in their

own namespace, but then also create a overarching namespace that brings

together "blessed" versions of each candidate technology.  XML
Schema

may not have good ways of doing this (I don't know), but in the

worst-case, the element definitions could be repeated in a different

namespace.


No decision was made, but it was my impression that the last of these

approaches carried the greatest appeal, if the details could be worked

out.


Michael