From: Bryan Aupperle
[mailto:aupperle@us.ibm.com]
Sent: Monday, August 11, 2008 7:54
AM
To: sca-c-cpp@lists.oasis-open.org
Subject: RE: [sca-c-cpp] ISSUE 17:
Should C++ -> WSDL Roundtrip with WSDL -> C++
I've looked at the JAX-WS spec more fully and I think there are some
areas we need to look at in a bit more detail.
1) A
header file with no Web service annotations
If I
read the JAX-WS specification correctly, then a file with no @WebService annotations
would not lead to any portTypes being defined. I think we want to be a
bit more liberal than this. If a file is used in an interface.cpp element
then clearly the intent is for at least one class in the file to define an
interface and if the interface is marked as remotable in the SCDL, then the
interface has to map to WSDL. So some options are:
- Each
class in the file maps to a portType
- Each
class with only virtual methods maps to a portType
- A
class matching the name of the service maps to a portType
- A
class identified on either a class or callbackClass attribute of an
interface.cpp method is mapped to a portType
The
last option is probably the best from an SCA standpoint, but it requires having
the SCDL at the time of the mapping. The second option is perhaps the
most generally useful.
2)
Default parameter names when mapping from C++ to WSDL.
JAX-WS
says that without @WebParam annotations, the names parts/elements corresponding
to method parameters are named arg1, arg2, ... I realize that this covers the
most general case where in a virtual method declaration parameters are not
themselves named, only typed. However, in the case where parameters are
named in the declaration, I think it would be desirable to use the name in the
parameter mapping.
3) We
already identified that we will need examples parameter mapping, but some
explanation may be needed as well. Section 3.6.2 of the JAX-WS
specification talks about wrapper beans, request beans, and response beans.
We clearly need to talk about how C++ is different here. I don't
have any specific thought just yet, but I hope to before our next call.
Bryan Aupperle, Ph.D.
STSM, WebSphere Enterprise Platform Software Solution Architect
Master Inventor
Research Triangle Park,
NC
+1 919-254-7508 (T/L 444-7508)
Internet Address: aupperle@us.ibm.com
|
"David Haney"
<haney@roguewave.com>
08/05/2008
08:35 PM
|
|
To
|
Bryan
Aupperle/Raleigh/IBM@IBMUS, <sca-c-cpp@lists.oasis-open.org>
|
|
cc
|
|
|
Subject
|
RE: [sca-c-cpp] ISSUE 17: Should C++ -> WSDL
Roundtrip with WSDL -> C++
|
|
I’ve taken a second pass at providing a
C++->WSDL / WSDL->C++ mapping, however this time instead of attempting to
completely define the mapping in our spec, I’ve referenced the JAX-WS
spec, and have provided guidance for applying that specification to C++.
There are some areas where the behavior is different enough that JAX-WS
doesn’t apply and we still have to define the behavior ourselves, however
much of the work around defining the mapping of WSDL
portTypes/bindings/operations to Java classes/methods applies equally well to
C++ classes/member functions.
Pros:
- We don’t have to reinvent the wheel and define a complete
mapping ourselves. We can rely on an established mapping, augmenting it
as necessary so that it can be applied to C++.
- We provide some level of interoperability between the SCA C++ and
SCA Java specifications, as SCA Java also relies on the JAX-WS binding for
defining their mapping.
Cons:
- JAX-WS only defines a mapping to WSDL 1.1, so we also will only be
defining a mapping to WSDL 1.1. This limitation also applies to the SCA
Java spec, so we can follow their lead in terms of how we’ll address a WSDL
2.0 mapping.
- Our C++ implementation is relying on a Java specification, which
seems a little odd.
The current draft is still pretty rough, and needs examples
throughout, however I think it provides a more complete mapping than the previous
proposal. Main areas of work will be the incorporation of examples (which
will probably result in minor changes to the described APIs), as well as
cleanup around fault/exception handling. We also need to provide some
clarification around how the SDO data bindings will be handled in this system,
and how we’ll roundtrip between WSDL and C++ if parameters are defined in
terms of complexTypes (which will map to commonj::sdo::DataObjectPtrs, which
lack the type information to roundtrip without code generation of those types).
We can discuss this at this week’s meeting, and see if
we want to pursue this approach, or if we should revert back to enhancing the
previous proposal.
Thanks.
David.
From: Bryan Aupperle [mailto:aupperle@us.ibm.com]
Sent: Wednesday, May 07, 2008 11:41 AM
To: sca-c-cpp@lists.oasis-open.org
Subject: Re: [sca-c-cpp] ISSUE 17: Should C++ -> WSDL Roundtrip
with WSDL -> C++
As promised...
Upon further study, the real distinction seems to be based on if messages have
one part with a type defined by a schema element or potentially multiple parts
each with a type defined by a schema type (simple or otherwise).
As far as I can tell the difference between doc-literal and wrapped doc-literal
has to do with the rules for the name of the schema element describing the
request message at least as far as the WSDL goes. If I am missing
something here, I would love to be enlightened. I do not suggest
distinguishing between these.
In any event, I still believe we have to address both ways of describing
messages and this document has revised mapping rules based on those David
described last week as well as the WSDL fragments promised below.
Bryan Aupperle, Ph.D.
STSM, WebSphere Enterprise Platform Software Solution Architect
Research Triangle Park,
NC
+1 919-254-7508 (T/L 444-7508)
Internet Address: aupperle@us.ibm.com
|
Bryan Aupperle/Raleigh/IBM@IBMUS
05/05/2008
04:30 PM
|
|
To
|
sca-c-cpp@lists.oasis-open.org
|
|
cc
|
|
|
Subject
|
Re: [sca-c-cpp] ISSUE 17: Should C++ -> WSDL
Roundtrip with WSDL -> C++
|
|
As I think about this some more, we are going to have to address mapping from WSDL
for at least rpc-literal, doc-literal and wrapped doc-literal. We could
try and limit this, but the reality is that all three styles are in use.
In the reverse direction I believe that again, we cannot ignore doc-literal or
wrapped doc-literal since these are popular, and interoperable styles. We
can discuss if we need annotations to control which style is to be generated or
leave it to be specified in an implementation dependent manner (e.g. tool
preference).
David has stated the basic principles but now we need to apply them to each
style.
I'll try to work on some WSDL fragments of each style that illustrate the
principles.
Bryan Aupperle, Ph.D.
STSM, WebSphere Enterprise Platform Software Solution Architect
Research Triangle Park,
NC
+1 919-254-7508 (T/L 444-7508)
Internet Address: aupperle@us.ibm.com
|
"David Haney"
<haney@roguewave.com>
04/29/2008
07:53 PM
|
|
To
|
<sca-c-cpp@lists.oasis-open.org>
|
|
cc
|
|
|
Subject
|
[sca-c-cpp] ISSUE 17: Should C++ -> WSDL Roundtrip
with WSDL -> C++
|
|
Attached is some initial work around redefining the C++ -> WSDL and WSDL
-> C++ mapping. It still needs quite a bit of work, but there may be
enough there to start discussing whether it's a direction we want to
take.
Thanks.
David.
-- David Haney
-- Director of Architecture
-- Rogue Wave Software
-- http://www.roguewave.com/
---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail. You may a link to this group and all your TCs in
OASIS
at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail. You may a link to this group and all your TCs in
OASIS
at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
[attachment "wsdl-cpp-mapping-2.doc" deleted by Bryan
Aupperle/Raleigh/IBM]