Re: [sca-j] NEW ISSUE: ejb binding @uri has the wrong schema

From
David Booz <>
Date
2008-10-29T20:19:42+00:00
ID
Thread
Re: [sca-j] NEW ISSUE: ejb binding @uri has the wrong schema
Some things that Henning Blohm said once are starting to come back to me about EJB3.

Here's more details of the problem, hopefully stated more clearly. The URI I'm looking at has two '#' characters:

<reference name="ejbReference">

      <interface.java interface="packageName"/>

      <binding.ejb uri="corbaname:iiop:localhost:2809/NameServiceServerRoot#ejb/componentName#packageName"/>

</reference>

Where componentName is a name that is used to uniquely identify the deployed EJB, and packageName is the Java package name of the Business Interface.  The componentName is only needed if the Business interface is deployed twice.  '#' was chosen because it's a character that would never be present in an EJB name.

This problem is self inflicted because the EJB 3.0 specs don't describe how to solve the "deploy it twice" problem.  This is how IBM solved it.  However, EJB3.1 looks to be solving the problem in a similar fashion.

To make this worse, in the course of further investigation today, I re-discovered that corbaName addressability for EJB is no longer required, so EJB3.1 has no impetus to avoid standardizing the '#' as a delimeter. We probably want to address this in the spec as well since we explicitly call out the use of corbaName.

I would like to continue supporting corbaName URLs for remote EJB3s, but that is becoming nearly impossible now if the binding.ejb @uri is an xs:anyURI.  The assembly spec has a provision for bindings which can't use the @uri and suggests that such bindings would define their own mechanism for configuring endpoint information.  I'd like to assert that the EJB binding as now arrived at that spot.

Direction for a proposal:

Given that EJB3 has backed away from corbaName, vendors are now free to invent all kinds of interesting schemes for addressing EJB3s remotely.  If we think that anyURI might be too constraining (I do) then we should invent a new attribute or element to hold the endpoint configuration.  My initial suggestion is something of type xs:string, but I'm open to other ideas.

Dave Booz

STSM, BPM and SCA Architecture

Co-Chair OASIS SCA-Policy TC and SCA-J TC

"Distributed objects first, then world hunger"

Poughkeepsie, NY (845)-435-6093  or  8-295-6093

e-mail:

David Booz---10/29/2008 03:16:35 PM---I am willing to believe I jumped the gun.  Issues are easy to close if they don't get accepted.

From:

David Booz/Poughkeepsie/IBM@IBMUS

To:



Date:

10/29/2008 03:16 PM

Subject:

Re: [sca-j] NEW ISSUE: ejb binding @uri has the wrong schema

I am willing to believe I jumped the gun. Issues are easy to close if they don't get accepted.

Can a fragment identifier contain a '#'? I've got an absolute URI for an ejb that has two '#' characters in it (not back to back).

Dave Booz

STSM, BPM and SCA Architecture

Co-Chair OASIS SCA-Policy TC and SCA-J TC

"Distributed objects first, then world hunger"

Poughkeepsie, NY (845)-435-6093 or 8-295-6093

e-mail:

Anish Karmarkar ---10/29/2008 02:45:44 PM---Dave, xs:anyURI allows absolute URIs, relative URIs as well as URI references.

From:

Anish Karmarkar <>

To:

David Booz/Poughkeepsie/IBM@IBMUS

Cc:



Date:

10/29/2008 02:45 PM

Subject:

Re: [sca-j] NEW ISSUE: ejb binding @uri has the wrong schema

Dave,

xs:anyURI allows absolute URIs, relative URIs as well as URI references. 

 [1] explicit talks about fragment identifiers. It seems to me that 

xs:anyURI does allow the '#' character.

I'm sure schema experts on this list will correct me if I'm wrong.

-Anish

--

[1] http://www.w3.org/TR/xmlschema-2/#anyURI

David Booz wrote:

> TARGET: EJB Binding spec [1]

> 

> DESCRIPTION:

> 

> binding.ejb/@uri is typed as a anyURI. This is because the EJB binding 

> inherits this attribute from the base binding schema.

> The problem is that anyURI doesn't allow the '#' character. The EJB 

> binding spec builds on the CORBA location syntax

> which includes the '#' character. See lines 103-115 [1]

> 

> PROPOSAL:

> None.

> 

> 

> [1] 

> http://www.oasis-open.org/apps/org/workgroup/sca-j/download.php/25979/sca-jejbbinding-1.1-spec-wd-02.doc

> 

> Dave Booz

> STSM, BPM and SCA Architecture

> Co-Chair OASIS SCA-Policy TC and SCA-J TC

> "Distributed objects first, then world hunger"

> Poughkeepsie, NY (845)-435-6093 or 8-295-6093

> e-mail:

>