Re: [sca-j] ISSUE 107: RFC2119 Language is needed for the EJB BindingSpecification - proposal v3

[David Booz <booz@us.ibm.com>]

Comments below <dab> like this </dab>

And a new revision:

(See attached file: sca-ejbbinding-1.1-spec-wd-04+issue107_v4.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:booz@us.ibm.com

Mike Edwards ---07/30/2009 06:01:17 AM---Dave, Thanks for your updates.

From:	Mike Edwards <mike_edwards@uk.ibm.com>
To:	David Booz/Poughkeepsie/IBM@IBMUS
Cc:	sca-j@lists.oasis-open.org
Date:	07/30/2009 06:01 AM
Subject:	Re: [sca-j] ISSUE 107: RFC2119 Language is needed for the EJB Binding Specification - proposal v3

---

Dave, 

Thanks for your updates. 

I've had a go at improving section 3, but I get the feeling that more is needed in that section, relating to SCA <service/> elements 
that use the EJB binding: 



Also see my comments below. 

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

From:	David Booz <booz@us.ibm.com>
To:	sca-j@lists.oasis-open.org
Date:	29/07/2009 16:44
Subject:	[sca-j] ISSUE 107: RFC2119 Language is needed for the EJB Binding Specification - proposal v3

---

Following Mike's review of the Issue 107 progress to date, I have another revision (v3) for 107 ready for review.

Mike, there are a couple of your comments that I'd like to talk to you about (still in the document).

1) Comment attached to section 3 ("Layout of section 3 is confusing...". I restructured this section and re-worded things slightly, guessing at your concerns, but am not really sure I understand your comment. Please take a look at what I've done and let me know if I addressed the comment or not. 

No, your updates don't solve my issue.  There is a whole context that is missing for this section - without that context, I find 
it impossible to work out what the section is really saying.  I suspect you may think it so obvious that it does not need to be 
said, but coming to it cold, I felt lost. 
<dab> :-) Yep, you are probably right.  For EJB 2.x there are two kinds of interfaces of interest, the home interface and the remote/local interface of the bean.  Speaking from the SCA reference perspective, it seems a natural step to put the remote/local bean interface on the SCA reference when connecting the reference to the bean.  The problem with this is that the remote interface contains more than business methods, so the intent of this section (I believe) is to instruct the developer what is allowed (or not) when creating the reference interface.  For EJB3, the remote (or local) interface is purely a business interface and so no mapping is really needed.  

To deal with the EJB 2.x home interface, the SCA reference developer wants to know what has to be done with the home interface so that an SCA reference can connect to the bean.  There is one restriction that the home interface as a predictable create method so that the binding implementation can first create the bean before invoking a business method.  This bit on the home interface is in section 6. </dab>

When I talked about "two interfaces" I meant it - let's look at what I mean.  The simpler case is an SCA <reference/> 
wired to an EJB offering some service: 

<reference name="foo"> 
        <interface.java class="foo.myService"/> 
        <binding.ejb uri="corbaname:rir:#ejb/fooMyServiceBeanHome"/> 
<reference>

So, the interface of the reference is clear - it is whatever is in foo.myService interface class. 
The OTHER interface is the interface of the EJB - the interface represented by fooMyServiceBeanHome in this case. 
<dab> No, it's not the home interface.  That section is concerned about the remote/local interface of the bean, not so much the home interface. </dab>

So what this whole section is talking about is the compatibility of foo:myService interface with the fooMyServiceBeanHome interface. 
It needs to make that crystal clear, in my opinion. 

I note that there are NO compatibility rules relating to SCA services exposed over the EJB Binding.  I find this surprising. 
What happens for an SCA service that uses a "pure" SCA business interface, with no EJB related features?  What can the 
EJB ref interface look like?  I can't answer this question - I hope that you can. 
<dab> l also found this surprising (except there is one rule about the home interface BSB60001) and made a mental note to re-visit this section after I resolved your questions on the reference side. Essentially, I think it comes down to this:
1) EJB service binding must implement the EJBObject and EJBLocalObject methods so that the SCA service implementation doesn't have to.  This is hinted at in section 6 and then section 6.1 states what the behavior of the EJBObject and EJBLocalObject, methods should be for SCA services with the EJB binding.  Section 3.1 BSB30003 also states this. I think BSB30003 needs to be moved to section 6, which I did in v4 of the proposal.
2) Decide if we want the spec to state HOW an EJB client (2.x is the hard one) gets an EJB remote/local interface and home interface for an SCA service with EJB binding.  I'd rather leave the HOW up to vendors.
</dab>

<dab> Despite all my commentary above, I am happy with the changes you've made to section 3 because your changes don't mention home interfaces, and I will incorporate them into the next revision. </dab>

2) Comment attached to section 3.1 - I'd like to strike the sentence you highlighted in your comment. Would that resolve the concern, or is there still a concern? 

Striking the sentence is OK. 
However, I suspect that there is a need for a discussion section which talks about the problems of the "poorly suited" 
interfaces and describes possible approaches to dealing with those problems.
<dab> The first and second sentence in 3.1 explain it.  It's simply the presence of those pesky EJBObject and EJBLocalObject interface methods that are required to be present on your nice clean business interface. </dab>

I removed your comments where I was nearly certain that I had addressed it.

I know about the formatting problem is section 8 of the pdf document, I'll correct that on the next iteration. I found it after I had already uploaded the documents.

[1] http://www.oasis-open.org/committees/download.php/33580/sca-ejbbinding-1.1-spec-wd-04%20issue107_v3.pdf
[2] http://www.oasis-open.org/committees/download.php/33579/sca-ejbbinding-1.1-spec-wd-04%20issue107_v3.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:booz@us.ibm.com

---

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

[attachment "sca-ejbbinding-1.1-spec-wd-04 issue107_v3_mje.doc" deleted by David Booz/Poughkeepsie/IBM]

sca-ejbbinding-1.1-spec-wd-04+issue107_v4.doc