The review of the papers of actuality Popular dance / weekly J2EE offered by Xebia.

Actuality editors / SSII

• Sun, Popular dance and JCP
• RIA, SOA, Tim Bray pulls in the heap

SOA

• Red Hat JBoss chooses to invest in CXF

The corner of technology

• Finally a roadmap itemised for JDK7
• The creation of bundles OSGI according to SpringSource
• Use the cloud to reduce the costs of uninterrupted integration.
• Tools Glassfish for Eclipse

Actuality editors / SSII

Sun, Popular dance and JCP

For some days, an agitation creates within the community Popular dance around the activities of Sun in relation to its plan OpenJDK 7. Several observations indeed allow to think that JDK 7 could well not be based on a standard of JCP as it is habitually case:

• Neil Bartlett points out that the Jigsaw plan, pushed by Sun to be the system of modules used by JDK, is not defined by JSR. This makes therefore a technology possessing legitimacy of which he doubts facing OSGI, a standard benefiting of 10 years of experience.
• Stephen Colebourne notes that Sun does not mention any more  Popular dance 7 for some time, but only of  JDK 7 which is only a realisation, where  Popular dance 7 would qualify a standard.
• Ann Thomas Manes making similar observations, compares OpenJDK with the position of no framework  not standard, "rebel " Open Source  , which holds Spring.
• Stephen Colebourne still, obviously very implicated by this debate, comes back in a series of tickets on the different who compare Sun to foundation Apache, , as well as on the problems of intellectual property within JCP in a general way.

If JDK 7 must be finalised without no standard supporting him, it would constitute a major event, since the future would become uncertain for JVMS given by other editors such as Oracle or IBM.

RIA, SOA, Tim Bray pulls in the heap

It is that Tim Bray co-author of the XML gives us his opinion on technologies RIA. According to him, RIA portrays Flex, Silverlight and JavaFx are aimed only at giving applications still more immersives and less and less useful. He softens his purposes however by leaning on examples of success as Google and Yahoo! who use in an elegant way Ajax and consort. In short, for him the advantages of interfaces web are simplicity and speed, Tim therefore does not believe RIA in longevity of. Nothing better to make of the heavy applicatif where from the heavy applicatif.
In his interview, he explains fast that the emergence of the Web Services and architectures of type SOA was made contrary to common sense. At the beginning each went there from its private format XML and this worked, but they wanted to generalise problem and to give a solution for all. This will to create a philosopher's stone to display its services caused only difficulties of realisation, of integration and especially of interop rabilit . In short, Tim is rather pessimistic as regards WS-* , he supports on the other hand REST who according to its terms, a reformulation of the good practices of the Web Oriented Architectures is only.
Finally, it remains very discreet on Sun and on the cloud computing by pointing out however that for the moment the department tries to find little. Is it necessary to give solutions and tools to implement clouds or to accommodate clouds directly and to become supplier?

The answer of Yakov Fain

SOA

Red Hat JBoss chooses to invest in CXF

Red Hat announced last week which she becomes one  " key contributor " of plan APACHE CXF.
Eventually, as explains it Sacha Labourey on the blog, it means that JBOSSWS who gives the choice of the realisation Web Service (today giving no that a layer of abstraction) will not give more than the realisation Apache CXF. It is damage, that the releases last of CXF had more rather painful and richer were in bugs.
On the other hand, the active participation of JBOSS teams in plan can only help to rectify shooting. Let us suppose therefore that the releases intendeds of APACHE CXF will be much better bills than those of 2008.

The corner of technology

Finally a roadmap itemised for JDK7

During the conference Devoxx of last December, a keynote had particularly held attention: Mark Reinhold introduced then the list of the functionality which would be included in future JDK 7, as well as a date for the finished version.

Mark Reinhold announces now the online bet of detailed calendar of the envisaged milestones and of linked functionality.

The analysis of this information leads to some comments:

• The period announced last December for the arrival of a finished version, to know  at the beginning of 2010 , is slightly brought forward since a final milestone is envisaged for 2/18/2010 monitoring of a stage of release candidate from one to two months. This would carry therefore the arrival of a finished version in April or May, 2010.
• The list of novelties remains the whole unchanged with one exception: it is not any more a matter of MVM (Multiple Virtual Machines) who had to allow to carry several applications out within same VM. This functionality, particularly interesting in the context of the rebound of interest for Applets Popular dance and other JavaFX applications loaded into the browser, was annotated of  "Maybe" two months ago and was obviously too ambitious to resist the pressure of the calendar.
• has not been named in December and am not him still. The cry for help of the spec lead Stephen Colebourne will have been vain therefore. Indeed, as he then pointed it out, in the case of an inclusion in JDK, one and on NIO2 would have an impact on JDBC to take into account new IPA - INTERNATIONAL PHONETIC ALPHABET. Count was held by this impact, it was essential that this JSR could be fast finalised. And his status, always announced as  "inactive" by the site of JCP, tends to make us think that the class java.util.Date still has "nice" days in front of her.

The creation of bundles OSGI according to SpringSource

SpringSource announces the arrival of Bundlor, its tool of creation of bundles OSGI, currently in version 1.0M2.

In a plan OSGI it is indeed necessary that all the modules of application and third bookshops are given in form of bundles. The creation of bundles for the modules control of which there is is not a problem it am, for the third bookshops there, is there very in another way.

Two possibilities present themselves then to the developer: get the bookshop that he searches packag e in form of a BUNDLE OSGI on a repository devoted (that of SpringSource for example) or create this BUNDLE by himself. In the latter case, it will be necessary to analyse all the dependency of the bookshop to create an adapted manifest. It is there that the tools of creation automated by bundles intervene. the available only one until then was, creates by Peter Kriens, Technical Director d' OSGi. SpringSource brings therefore from now on the own tool.

Bundlor is justified by SpringSource by some not present functionality in Bnd such as the incr mentale creation of manifests and integration with Eclipse via a devoted plugin.

In spite of the visible effectiveness of this type of tool, it is necessary to apply it only to third bookshops. Indeed one of the interests of OSGI domiciles in his capacity to display only a part carefully chosen by the interfaces of a module, what only a human intervention will be in even to perform correctly.

Use the cloud to reduce the costs of uninterrupted integration.

It is generally known that the more the end of a sprint approaches, the more the resources of the server of integration are solicited. Besides, it is for these periods when the developers need the quickest feedbacks. While, to make? Give itself a high-powered machine, devoted to uninterrupted integration, but which will be only seldom used in full capacity?
Atlassian, the editor of the server of uninterrupted integration Bamboo, offers to resolve this problem in the version 2.2 of his product, by giving the possibility of using "Elastic Bamboo" , a distant agent complying on the cloud Amazon EC2. So, availability and power are maximum when the team needs it, while limiting costs.
And as Atlassian uses Bamboo inside, some plasters are already wiped:

• Builds must indeed be independent of environment (to you to construct your Builds aware of this problem).
• In standard case, dependency must be downloaded in every build (the authority of build disappearing from the cloud when she ends). A solution can be to use an image disc safeguarded on the cloud.

All smartness of Edwin Wong to define your server of uninterrupted integration on the cloud is on the blog of Atlassian.

Tools Glassfish for Eclipse

Sun has just taken out version 0.9 (version 1.0 soon arrives) from his BUNDLE.
This one includes:

• Eclipse 3.4.1
• Servers Glassfish v 2.1 and v3 Be a prelude
• (Optional) JDK 1.6 UPDATE 12
• A series of plugins allowing to configurer your server J2EE, as well as your Web services (Soap or REST) since IDE. (More details on plugins, here).

This BUNDLE gives advantage to give a tout-en-un packet. He is intended for the developers and for the organisations which do not use NetBeans, but Eclipse. This one being most spread IDE. Sun enlarges his public so and gets closer to the big community of Eclipse.