Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[eclipselink-dev] Minutes: Committer Meeting - Apr 1 2010

== Agenda Apr 1, 2010  ==
[ 11 am EST, Thurs Apr 1st, 2010]

=== Ad Hoc ===
EclipseCon summary
* Runtime projects got more airtime
** Several well attended runtime sessions
*** JPA 
*** OSGI
*** DBWS

== Release Planning ==
* 2.0.2: Proposed date: April 21st
** Stream is now frozen.  Showstoppers approved by Doug/Peter only.
* [ 2.1] [[Helios]]

== QA ==

Mar 26 - April 1, 2010:

* Nightly Results
** 1.1.4 - JPA/Glassfish 2.0 jpql test failures - bug [ 307114]
** 2.0.2 - no recent builds
** 1.2.0 - no recent builds
** trunk - JPA/J2SE tests hanging - investigating
* Last week
** Fixed Bugs
*** 307404 - need to add JPA server testing on JBOSS EAP
** Ran SDO server tests on Websphere - passed
** EL 2.1 Advanced Query - FUNC support test spec 90%; Downcast - documents reviewed, reviewed example
** EL 2.1 Advanced JPA Config - reviewed feature docs, and the tests that have been added. 
** Triage - 7 bugs filed
* This week
** Change SDO server tests to run on any app server
** Continue with EL 2.1 test dev work - Advanced querying, Advanced JPA config
** JAXB 2.2 - enable in nightly automation
** JAXB nightly tests - email notification not working - will fix
** JAXB osgi testing - investigation
** Fix Bugs

== Build ==
* Milestone Releases
** M6 EclipseLink
** M6 Helios contribution (nightly build)
*** Due to an uncaught bug found by the Helios build system.
*** We need to modify our milestone strategy. Milestones should be promoted *after* the Helios contribution is made. 
* "target platform" work for Helios (bug 306188)
** RT is requesting we simplify how we are included in RT - siingle SDK
*** the one feature can and should simply include our other already existing features.
* As of M6 EclipseLink is no longer shipped with the Java EE Package
** As a team we need to decide:
*** What we should offer regarding IDE tooling that helps the developer use EclipseLink
*** If we should help another team provide it (Dali ?)
* Helios build integration (293034) (Blocked until Post M6)
** Need to do p2 consolidation work to manage multiple versions in a single update location.
*** investigation composite repositories
**** Have working prototype. 
**** Working on creating a Composite of all our releases - (completed locally).
***** Cannot generate on Download machine due to bug 303802 fix will be available in M6 helios release.
**** Still need to finalize automation.
***** Automation for Composite work still pending. Looks like a separate script will be necessary.
*** Migration from p2.metadata.generator to p2.publisher (Blocked by bug 303802).
**** Prototype complete. 
***** Awaiting confirmation of a bugfix before commit (Blocked by bug 303802).
***** Still need to finalize automation (Blocked by bug 303802).
* Test component build cleanup (Holding)
** Restructure to allow testing to *not* rebuild product jars (293032)
*** Will use "static targets" as discussed in meeting last week (297217).
*** Tests will not compile when run target called. 
**** merged partial work to update standards. 
**** migration work for test projects, currently priority 2 
** Need shippable testing artifacts (R2).
* jpa.test10 Suite (Holding)
** Explore possibility of using eclipselink.jpa.test from 1.1.3 as a way to certify javax.persistence 1.0 backward compatability in EclipseLink 2.0.x
*** All tests now compiling with trunk, 83% success rate on Test run.
*** Awaiting review by Tom or Peter of test results (still).
** Need "project name" assuming this gets running 100% 

Fixed Bugs:

== JPA  ==
Apr 1
* no bugs
** two bug fixes waiting on reopen of stream
*** Exteranious logging
*** Blob/Clob Length
* WDF tests broken by
* Advanced JPA config
** Virtual access
*** Dynamic persistence requirment
* Advanced JPA Querying
** Downcast queries implementation
** Fetch Group
* Bug Fixes
** Cache evict
** Native Case sensitivity

== DBWS  ==
[ DBWS Development]

* No remaining bugs

* 290156 Validate SOAP message elements
* Scoping out potential EclipseLink feature work
** Improved Sparse Merge support (an 'Update' operation currently requires the instance to be fully specified in the SOAP message)

* Bug - Using DBWS in JRockit environment due to ASM generated class incompatibility
# ASM <br/> Each DBWS web service <tt>.war</tt> file contains an ASM-generated class that implements the JAX-WS <code></code> API
<source lang="java5">
public interface Provider<T> {

     T invoke(T request) {
         return ...
For DBWS, the generic <code><T></code> is bound to <code>javax.xml.soap.SOAPMessage</code>. In addition, the generated class requires a number annotations to indicate information about WSDL, target XML namespace, resource injection and lifecycle methods: 
<source lang="java5">
//Java extension libraries
import javax.annotation.PostConstruct;
import javax.annotation.PreDestroy;
import javax.annotation.Resource;
import javax.servlet.ServletContext;
import javax.xml.soap.SOAPMessage;
import static;

//EclipseLink imports
import org.eclipse.persistence.internal.dbws.ProviderHelper;

    wsdlLocation = "WEB-INF/wsdl/eclipselink-dbws.wsdl",
    serviceName = {SERVICE NAME},
    portName = {SERVICE NAME + "Port"},
    targetNamespace = {SERVICE NAMESPACE}
public class DBWSProvider extends ProviderHelper implements Provider<SOAPMessage> {
    public  DBWSProvider() {
    protected WebServiceContext wsContext;
    public void init() {
    public SOAPMessage invoke(SOAPMessage request) {
        if (wsContext != null) {
        return super.invoke(request);
    public void destroy() {
A <tt>.war</tt> file containing <code>_dbws.DBWSProvider</code> will deploy to a WebLogic server fine, but when the <code>invoke(request)</code> method is hit, a <tt>.dump</tt> file is generated for JRockit JVMs; the same <tt>.war</tt> file, deployed to the same WebLogic server running on the Sun JDK, works fine.

The underlying problem is the version of ASM in <code>o.e.p.internal.libraries.asm</code> is 1.5.3. When I re-implemented the one class <code></code> that does all the ASM work for the <code>DBWSBuilder</code>, the <tt>.war</tt> file works for both JRockit <b>and</b> Sun JVMs.

There are no runtime-dependencies to ASM (1.5 or 3.n, internal or external) for generated <code>DBWSProvider</code>'s - this is solely a design-time requirement.

The proposal is to include in the standalone install a copy of ASM 3.1 in ...\eclipse\utils\dbws (org.objectweb.asm_3.1.0.v200803061910.jar from Orbit) and generate a new jar file (eclipselink-dbwsutils-asm3.jar) that, if the user requires, can be placed <b>ahead</b> of the 'real' eclipselink-dbwsutils.jar file so that the new functionality can be used to generate a <tt>.war</tt> file that works with JRockit JVMs.

For the EclipseLink installer, the issue is whether or not we can get a CQ thru the system in time for Helios/2.1

== SDO / MOXy  ==
* All 2.0.2 bugs resolved

* Representation of JAXB Annotations and enhanced MOXy metadata in externalized 'eclipselink-oxm.xml' metadata format [ Design]
* MOXY Dynamic Persistence [ Design]
* JAXB2.2 Compliance
** JAXB 2.1.12, 2.2 API, Impl jars submitted to Orbit
** Need to determine whether new CQ is required for XJC, based on new Dynamic Persistence runtime dependency
* OSGi, Server testing
** Discussion with Dev, QA teams to improve the level of server and OSGi testing we have for MOXy and SDO components.

== Dynamic Persistence ==
[ Dynamic Persistence]

== EclipseLink Incubator  ==

* [[EclipseLink/Development/Incubator/Extensions]]
** NetWeaver platform promotion.

== Documentation ==
*Working on automated doc build process (Rick & Eric)<br />If not possible, Rick is working on an automated  localized doc build process, and would simply post the daily doc builds
*Updated [ Documentation Requirements and Build Proposal]
*Several "chapters" currently stored in XML. More to come once the production/build process is finalized.
*Need discussion about how the strategy we are using for User Documentation compares to the wiki-based strategy we are using for developer documentation and whether there is merit to changing the way we do developer documentation. This discussion should wait until we have used found all the issues with our User Documentation Strategy. 

Back to the top