← Prev in month ← Prev in thread
Next in thread → Next in month →

Fw: [sca-assembly-comment] Few comments from Siemens

From
Mike Edwards <>
Date
2009-06-12T15:15:46+00:00
ID
Thread
Fw: [sca-assembly-comment] Few comments from Siemens
Folks,

A comment received from Siemens regarding
the SCA Assembly public review...

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:  

----- Forwarded by Mike
Edwards/UK/IBM on 12/06/2009 16:13 -----

From:
"Konradi, Philipp" <>

To:
<>

Date:
12/06/2009 00:13

Subject:
[sca-assembly-comment] Few comments
from Siemens

Below are two
comments from Siemens on the SCA Assembly
Specification draft, Version 1.1 CD03, put for public review.

------------------------------------------

Comment 1

------------------------------------------

Chapter 13.2 "SCA Runtime", line
3874-3875:

"3. The implementation MUST support
and comply with at least one of the OpenCSA Member Section adopted implementation
types."

 

Siemens proposes to skip this conformance
requirement for the following reasons:

a) This conformance requirement limits SCA
implementations focusing on support of proprietary or domain-specific languages.

The following two cases fully apply to Siemens
and probably to many other companies:

- a lot of proprietary, exotic programming
& scripting languages exist and are
used for system engineering and development.

- creation of domain-specific languages
gets more and more important and is actively used for application
development 

Because of this language diversity it has
been of big value for Siemens to see that SCA offers one common, language-independent
model to assemble and wire components together. 

The conformance requirement mentioned above
means a considerable obstacle for users and implementers, who would like
to use the SCA model and provide lightweight
runtime implementations supporting SCA-Assembly, -Policy + a very specific
implementation type, because it requires:

- either to standardize the proprietary
language (which is often not an option since
the language specification is not even public, the
overhead & time required
to get
it through the standardization process
are too high)

- or to implement support for another language
to be
able to claim conformance
but which is actually not required
and used by
application.

At the same time for
such a lightweight
SCA runtime it's still important to be able to claim SCA conformance. Even
company-internal providers need to offer to users some extent of standard
compliance, with SCA assembly and policy model in this particular case,
to assure usage of proved, widely used concepts, ease of extension
as well as of migration.

The big advantage of SCA for our company
has always been to have a model for building service-oriented applications
which is totally independent of any programming language or communication
technology. We'd like to see SCA stay as open as possible and supporting
diversity where it's needed. We think that this conformance statement is
too restrictive and doesn't facilitate broad acceptance of SCA.

b) This conformance requirement limits customization
of an SCA runtime. 

SCA runtime implementations are often build
in a modular way and allow to install/uninstall implementation types or
bindings. This conformance requirement does not allow customization of
a runtime in a way where particular implementation types, which are currently
not required by application (but required to claim conformance), don't
get deployed (e.g. to decrease the memory footprint --> important for
industrial use).

c) This conformance requirement doesn't
really ensure portability if it has been the purpose of this statement.

A runtime which complies with this statement,
and thus can claim being SCA-conformant, is free to deviate from SCA specifications
when it comes to any
other implementation type it supports.

If ensuring portability is the goal then
we’d rather propose to rephrase the conformance
requirement to something
like: "If the implementation supports an implementation type adopted
by OpenCSA Member Section then the implementation MUST support it in a
compliant way."

------------------------------------------

Comment 2

------------------------------------------

Chapter 13.2 "SCA Runtime", line
3876-3877:

"4. The implementation MUST support
binding.sca and MUST support and conform to the SCA Web Service Binding
Specification v 1.1."

Siemens proposes to change this conformance
requirement to the following statement:

"4. The implementation MUST support
binding.sca."

for the following reason:

This conformance requirement considerably
limits usage of SCA runtimes in the Industry. For most of embedded or industrial
applications Web Services aren’t used at all. Other communication standards
like BACnet are used in such applications instead. Requiring runtimes to
bring support for Web Services will unnecessarily increase their footprint
and prevent their usage on embedded devices, which is a really important
use case for companies like Siemens.

Thanks,

Philipp

Unless stated otherwise above:

IBM United Kingdom Limited - Registered in England and Wales with number
741598. 

Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU
← Prev in month ← Prev in thread
Next in thread → Next in month →