Xebia

Blog Xebia France

J2EE, Agility and SOA

********** Last minute ***********
Review of the papers Xebia
January 21st, 2008
Print this ticket

Review of the papers Xebia

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

Actuality editors / SSII

Agility

RIA

SOA

The corner of technology

Actuality editors / SSII

Oracle buys back GAPED

Massimo Pezinni du Gartner Group displays his point of view in an article appeared on LMI: he sees in this redemption a danger of disappearance of components Tuxedo GAPED except WLS and.
According to him Fusion plan is already well hired at Oracle's and the arrival of BEA should not change deal. He speaks about some products which would be in favour service From there to wonder about the future of the new products (ALSB, WLI) there is only one step.
Really, in some years the range of products GAPED considerably widened. 3 ranges are differentiated:

  • The historical range: Tuxedo.
  • The range facilities WebLogic-*: Weblogic Server, Weblogic Integration, Weblogic Portal and JVM JROCKIT
  • The range SOA: Aqualogic-*: ALSB, ALDSP, , ALRR, ALES, ALUI,

It is necessary to look at what is this day in great numbers sold to client GAPED: Weblogic Server, Tuxedo, WLI / WLP, ALSB.

  • Oracle is indeed very interested by Weblogic Server who is admitted as one of the best servers of applications.
  • WLI is surely the product which can resist and continue existing if he is seen as a tool of integration and not as an engine of BPM.
  • Concerning ALUI, ex Plumbtree, there is a big installed park. It is necessary to know that ALUI is on the whole independent of the server of application. For a reflexion on portal GAPED / ORACLE, you can read the ticket of Chris Bucchere: One Portal to Rule Them All.
  • For his/her part, Tuxedo does not evolve tremendously. It is already in mode service many years but with a park huge client.
  • JROCKIT under Linux should be also kept. It is a very good JVM, more competitive than that of Sun that concentrates its optimisations rather on version Solaris (nothing amazing in cel ). This product is also one window for BEA where it showed its technological know-how. If the annual investment is raisonable, Oracle could well keep this window.
  • ALSB (Drunk AquaLogic Service), ESB of BEA could as for resisting him because already well sold and admitted.

For all rest, the future seems darker: few clients and in doubloon with an already existent offer at Oracle (it is nevertheless Oracle who buys back not BEA GAPED, and who buys back the department middleware of oracle).

Agility

Mike Cohn gives practices of adoption of agile methods

Mike Cohn, founder member of Agile Alliance, gives some tracks on the way of injecting agile practices into a firm through the patterns of base.

  • Is it necessary to start agility on a small plan, or to transform all organisation of a blow?
  • Is it first necessary to set up the technical practices XP either first to try hard to develop by iterations, without setting up TDD (Test Driven Development), programming peer or integration goes on?
  • Is it necessary to set agility up in " stealthy mode " , or it is necessary to make advertising across all firm and even outside?

As always, it is necessary to be pragmatic! And choose according to context the practices which fit the best to the firm.

RIA

What FLEX?

Ted Patrick, from Adobe, answers basic question: what simple percentage Flex?.
Extract:

  • FLEX is a means to produce files SWF (FLASH)
  • FLEX is made for the developers, not for the graphic designers
  • FLEX allows to develop applications
  • Who are going to be carried out on the web (player Adobe FLASH)
  • Or on the Офис (ADOBE AIR)
  • Where on motives (it is not still case, and it is one of the big defects of FLEX!)

SOA

JEE 6: Extensibility, Profile Pruning and

InfoQ introduces us the main lines of Popular dance EE 6 from which one waits for Q4 2008 for the finished version. It will be the first modular version of Popular dance EE thanks to introduction of profile. Debatable debate  Tomcat is him a server Popular dance EE?  Tomcat will officially become null and void, the  Popular dance EE Web Profiles  will implement  .

Major novelties
Finished servers Popular dance EE monolythiques! Popular dance EE 6 becomes modular with introduction of profile.
This mechanism of profiles is already used in Mobile Popular dance [1] with some difficulties by interop rabilit and the promoters of OSGI defend their model of assemblage; they will remember the glowing report nevertheless that Rod Jonhson, the founder of Spring Framework, addressed in profile JavaEE 6 in July, 2007 (source Spring 2007/07: Popular dance EE 6 gets it right). They will also point out that the debate on OSGI is also present on specification JBI 2.0.

Pruning
The process of ' Pruning ' (ie. at the end of life) exists for a long time for standards Popular dance but was seldom used. In break with the previous versions, Popular dance EE 6 opens the construction site of cleaning of the IPA - INTERNATIONAL PHONETIC ALPHABET outdated (Popular dance 7 follows the same approach). The first candidates for retirement are:

  • EJB-CMP: grieved for the firms which invested on EJB 1.x and 2.x. Abolition begins with the mechanism of persistence EJB-CMP and it will be needed will expect to see fast following the rest of these IPA - INTERNATIONAL PHONETIC ALPHABETS.
  • JAX-RPC and JAX-REGISTRY [2]: IPA - INTERNATIONAL PHONETIC ALPHABETS Web Services for Popular dance do not live longer than standards WS-*. These disappearances will have probably not enough impacts, XFIRE and Axis 1 were often preferred in JAX-RPC. As for JAX-REGISTRY, its use remained confidential.

The absentees

Two notable absentees to JavaEE 6 will be noted:

  • Portals popular dance and their management of contents [3]: the wave of intense marketing of the editors of portals popular dance passed; the promises of r utilisabilit and KING one seldom been in appointment. IPA - INTERNATIONAL PHONETIC ALPHABETS portlet are they going to make ' Pruner ' even before joining Popular dance EE? We sense rather the creation of one profile  popular dance portals  the community of which will be very restrained. Only a major break in the management of contents in Popular dance would allow to threaten the predominance of PHP on this subject.
  • The standards of integration JBI and SCA / SDO [4]: these subjects seem too polemical and too political to allow to ratify standards Popular dance. JBI 1.0 disappointed those who ventured (there ServiceMix, Female mule, etc) and the draught of version 2.0 always meets the hostility of IBM and of GAPED. These two editors prefer him standards SCA and SDO which meet an at least circumspect reception on behalf of Sun, perhaps because the gouvernance of these standards is shared.

[1] Mobile JSR 37 Information Device Profile, JSR 137 POPULAR DANCE GAME PROFILE, etc
[2] JSR-101 POPULAR DANCE IPA - INTERNATIONAL PHONETIC ALPHABETS for XML BASED RPC and JSR 93 POPULAR DANCE IPA - INTERNATIONAL PHONETIC ALPHABET for XML REGISTRIES
[3] JSR-168 PORTLET SPECIFICATION, JSR-286 PORTLET SPECIFICATION 2.0, JSR-301 Portlet Bridge Specification for JavaServer Faces and JSR-170 Content Repository for Popular dance technology IPA - INTERNATIONAL PHONETIC ALPHABET
[4] and JSR-235 DATA SERVICE OBJECTS

The corner of technology

A positive experience on Wicket

After last year to develop an application with Wicket, Julian Sinai, gives us its very positive return of experience. From the world of the  heavy client  (Swing), he assessed web at first different frameworks:

  • GWT: too much moved away from other frameworks web, of HTML, of WAR
  • Tapestry: big cost of entrance, difficult to develop new components
  • JSF: too complex

Among the positive points that he withdraws experience, he stresses clear separation especially between code of presentation and code job, and on the easiness with which it is possible to develop r utilisables components. Matt Raible commented fast on the article on his blog.

Key words :, , , Popular dance, , , , ,

Tickets on the same topic:

You can follow answers accepted by this article thanks to the thread of comments.

9 answers to "Revue de presse Xebia"

  • On January 21st, 2008 in 20:37 (), Alexis MP said :

     (JVM JROCKIT is) more competitive than that of Sun 
    Performances are relating and changeable in time.
    It is not apparently possible to say that the one is superior to second, even on Linux.

    I am curious, are there references of JROCKIT in production?
    Weblogic Server GAPED is him always configur to use JVM Sun by default?

  • On January 22nd, 2008 in 8:32 (), Thomas Recloux said :

    Alexis =>

    I have of my performed quoting some benchs comparing the performances of Weblogic with both JVM (Sun and JROCKIT) every time JROCKIT was very appreciably below.

    I know at least two big count which makes turn Weblogic with JROCKIT.

    On the contrary I systematically met bugs with JROCKIT, of bugs of stability or of bugs in IPA - INTERNATIONAL PHONETIC ALPHABETS.

  • On January 22nd, 2008 in 11:02 (), Manuel Eveno said :

    Small precision: JVM of Sun is offered by default in mode development but JROCKIT production is it for mode.

  • On January 22nd, 2008 in 15:14 (), Guillaume Carre said :

    Good morning Alexis and thanks to the attention which you turn in our blog.
    Here are some precision of our point of view on JROCKIT:

    References use of JROCKIT in production

    To name references is always difficult because of commitments of confidentiality.
    If we meet JROCKIT more often in production on servers Windows, here is some clients with whom I worked who use JROCKIT under Linux:
    * a French leader of Remote Sale uses WebLogic Server 8.1 with JROCKIT under Linux.
    * a leader of loaded electronics uses JBOSS with JROCKIT under Linux RHEL 4.

    Performances

    The comparative benchmarks is a delicate science which it is necessary to handle with the biggest alertness.
    We kept the comparative of JVM SPEC JBB 2005 (Popular dance Server Benchmark) on which GAPED, IBM and Sun seem to agree since they publish the results of their realisation there.

    If it is impossible to compare abreast JVM of the different editors because the conditions of tests change (used server, etc), we will point out the very good performances of JROCKIT (cf here).

    It takes nothing away from the merits of the realisation of IBM and of Sun. Very important earnings of performances between versions 5 and 6 of realisation of Sun will be greeted in passage.

    Tools and exploitabilit

    The tools of JVM are a key element of the exploitabilit of an application Popular dance.

    JROCKIT always turned close attention to this subject and a distinctive argument is surely at the time of choosing a realisation.

    Concerning the importance of the tools of JVM, they will remember:
    - of the interview of the Dr Holly Cummins (IBM HURSLEY LABS) on The Server Side who learnt us that IBM put currently the focus on the tools of its JVM.

    - of the exhaustive list of options of JVM 6 Sun where Joseph D supports. Mocker.

  • On January 24th, 2008 in 8:22 (), Francois said :

    I want to bring a precision: the version of assessed Tapestry is not specified, and I am surprised of comments if they carry 5 on. Tapestry 5 does not have thing big to be seen with Tapestry 4 (if, there are the same author, the same concepts, but this stops there). He would have is possible be had to have another name, to avoid confusions.
    I think (after valuation of Wicket 1.3 and Tapestry 5, and after one year of development with T5) that T5 is much simpler than Wicket (and therefore surely simpler than quite other framework Web Popular dance).

    The cost of entrance in T5 became practically no one: there is one application which works in an order thanks to an archetype Maven, its first pages and components at the end of some minutes are made
    Especially, components are POJO, who only has to follow some conventions. There is not complex inheritance in Wicket. Where from my questioning on comments concerning complexity to create new components in T5.

    Nevertheless, I have two tracks:
    * it is obvious that there from Swing the author is at home with Wicket. And it is is possible be so why I am more comfortable with T5 :) :)
    * T5 uses a framework of IOC (very good, little in Guice, and who integrates in a transparent way with Spring). Idea is the same that for framework of IOC (manage the cycle of life of classes and their dependency for them). This allows between other one it can inject dependency into our classes without caring of the way the class is found, and to separate services very apparently (without states) from the code of presentation: in the final, there is less code without report with the purpose of the class. In short, this framework of IOC is copiously used, and if the author is not familiar with the concept of IOC, he could find it strange and complex. And however, it is so practical to have an annotation @Inject instead of 4 or 5 lines of access to factories and authorities of services

    By rereading me, my comment makes little  preach for its parish  . Therefore especially: do not believe me on word, test, and see me if he is suitable for you.

    Last comments: The part  interface web  from InterLDAP (http://interldap.org) is under construction since vicinity one year.

    T5 allowed me to develop in a completely elegant way a framework of generic components relating to our job, which it is possible to overload, to put together or to spread to specialise them for a client. And I begin being rather happy with result (code is free, and soon a site of d mo will be set up ;)

    Good day,
    Francois Armand

  • On January 30th, 2008 in 21:16 (), Alexis MP said :

    To run quickly without air bag ca seems to be a common practice in benchmarks objective of which is not to test stability but only performance

  • On March 3rd, 2008 in 19:32 (), Review of the papers Xebia by J2EE, Agility and SOA :  The blog of Xebia France said :

    [] planes recalled the Popular dance EE Web Profiles in January. Roberto Chinnici reveals us today in Profile the Popular dance EE 6 Platform [...]

  • On May 19th, 2008 in 18:42 (), Blog Xebia France - Review of the papers Xebia said :

    [] the framework of Apache, Tapestry. More recently, Fran ois Armand commented on our article A positive experience on Wicket surprised at the cost of entrance raised for Tapestry 5. Renat Zubairov and Igor [...]

  • On September 1st, 2008 in 18:48 (), Blog Xebia France - Review of the papers Xebia said :

    [] the announcement of the redemption of BEA by Oracle, each and every one had a good time betting Oracle [...] on the respective survivals of the software bricks

Leave a comment

********** Last minute ***********
Review of the papers Xebia

Xebia it Architects lock France
Defence Colis e - 10/12, avenue of The Ark
92419 Courbevoie Cedex

T l : +33 (0) 1 46 91 76 16
Fax : +33 (0) 1 46 91 88 00
E-mail : info@xebia.fr