Apache Geronimo passes J2EE 1.4 test suite

Discussions

News: Apache Geronimo passes J2EE 1.4 test suite

  1. Apache Geronimo has passed the J2EE 1.4.1 test compatibility kit, putting it well on its way to completing the entire certification process, after which it would become the third major open source appserver, and arguably the one with the most industry support (IBM, BEA, Apache), to reach J2EE compliancy.

    The tests were completed by the Geronimo team just 12 minutes before Geir Magnusson began his 9:30pm BOF on the Apache Geronimo project at Java One, and roughly 90 minutes before the writing of this news post.

    The next step before becoming certified is to pass the "non testable assertions", a set of conditions that cannot be tested with code but every appserver must pass through. Work has already started on the next phase.

    "It was a very long 22 months to complete the test passing. We hope no one will ever have to do this again, please use our code, we are the Apache Group", said Dain Sundstrom, Apache Geronimo member.

    "This is a fantastic milestone, but we're certainly not done", said Dain Sundstrom, "next focus for us will be on the user experience, tools and ease of use. It should work and be understandable right out of the box. When we get to a place where we are comfortable with 1.0, that's when 1.0 will be released."

    "This marks the point of the project where we stop being so internally focused, but externally focused. Now it’s about delivering to people and building the community", said David Blevins, another Geronimo committer.

    So what does the eventual certification of Geronimo mean for the software community? Due to the Apache community model, Geronimo can now become a central point of collaboration on behalf of many major groups in the industry. For example, persistence providers can take advantage of Geronimo to provide a complete J2EE solution. We will likely see appserver vendors around the world (there are a number of implementations outside of the typical companies we are familiar with) drop their own server in favour of Geronimo, and even other groups who would have built their own appserver instead start with Geronimo as a base. Combined with BEA and IBM’s announced support, this will mean that a strong community will emerge that will probably be contributing many bits and pieces of software to Geronimo, helping it grow and helping them grow.

    Congratulations to the Geronimo team for the hard work!

    Threaded Messages (73)

  2. Congratulations to the Apache Geronimo team. Eager to work with the AppServer.
  3. So what ?[ Go to top ]

    so what? Geronimo remains vaporware. I do not really understand why it got so much press (well maybe because of Apache). You do not really think to use it in production, do you ?? so till it is a production ready product let's give more
    press to more serious and solid applications servers such as JBoss and JONAS!
    BTW I am not associated to JBoss and Jonas whatsoever but I do run high volume mission critical J2EE applications with real software....
  4. So what ?[ Go to top ]

    Vaporware ?

    That's a tad harsh. I would certainly consider using it to find out if it's production ready. This deserves press, passing the compatibility kit is a major milestone. Congratulations to the team.
  5. So what ?[ Go to top ]

    so what? Geronimo remains vaporware. I do not really understand why it got so much press (well maybe because of Apache). You do not really think to use it in production, do you ?? so till it is a production ready product let's give more press to more serious and solid applications servers such as JBoss and JONAS!BTW I am not associated to JBoss and Jonas whatsoever but I do run high volume mission critical J2EE applications with real software....

    Would be interested in learning more about your high volume mission critical applications that use JOnAS. To add in my list of use cases :)
  6. So what ?[ Go to top ]

    so what? Geronimo remains vaporware. I do not really understand why it got so much press (well maybe because of Apache). You do not really think to use it in production, do you ?? so till it is a production ready product let's give more press to more serious and solid applications servers such as JBoss and JONAS!BTW I am not associated to JBoss and Jonas whatsoever but I do run high volume mission critical J2EE applications with real software....

    Because it's another free, open source solution from a respected software provider that many people are eagerly waiting for, that's why. Heaven forbid we talk about J2EE servers on a J2EE site. Why dont we start up our usual ".Net sucks!" and see 100+ posts about nothing whatsoever.

    And I'm not sure about everyone else here, but I'll be downloading and testing their first release about a week after they release it. Why not sooner? Cause traffic on the site will be crazy from everyone trying to d/l it, that's why :)

    Congrats! Very impressive.
  7. So what ?[ Go to top ]

    I'll be downloading and testing their first release about a week after they release it.

    Sure. :)) The question is WHEN they got a PRODUCTION release ?
    I think, by THAT time, you most likely reconsider :))
  8. So what ?[ Go to top ]

    Sure. :)) The question is WHEN they got a PRODUCTION release ?I think, by THAT time, you most likely reconsider :))

    Why is everyone so cynical about this announcement? I'm confused. Considering how the large majority of people on this site use at least one Apache product (whether directly or through inclusion in another app), why is everyone so pesimistic about Geronimo?

    And no, once they have a production release I'll test it, regardless of when it occurs. IMO, a production release is worthy of a few hours to see how it works and whether I can port my code to the app.

    Maybe I'm just naive, but I doubt they'll release this without being confident in the quality of their code and the ability of developers to produce production systems on top of it. It wont be perfect, but I think the app will be able to run my EJBs, JMS queues, and handle servlet requests. Something tells me it will be able to do at least that.

    I can tolerate bugs in production releases. And I'd be surprised to hear someone say they wouldn't, as every server out there has them. I have yet to see a J2EE server release with, "Added functionality x,y,z and we didnt fix any bugs cause none exist in our code."
  9. So what ?[ Go to top ]

    "You do not really think to use it in production, do you ?? so till it is a production ready product let's give more
    press to more serious and solid applications servers such as JBoss and JONAS!"

    Hope that you don't mind Johny if i quote from a posting that you did on a different thread in order to comment this?

    "People like you XXX [name removed] are killing the OSS. If you do not want to use it, just do not use it. We need more open minded developers and less lamers..
    While they are "wasting" their time writing code
    you are wasting your time posting useless comments."

    Taken from here.

    Cheers

    Eike
  10. read AND understand is a basic skill[ Go to top ]

    Eike,
    I am Johnny not Johny ;-)
    sure I do not mind!
    I did not say I am against Geronimo but that it got too much press comparing to other more solid and widely used application servers..
    Eike, you should not only read the post (and cut and paste)..but understand what you read ;-)
    ..hope you do not do the same when you code...... :))

    PS
    I will be glad if Geronimo guys show me one day they have a production ready AS I can use.....but till then...
  11. Ain't the web grand[ Go to top ]

    "You do not really think to use it in production, do you ?? so till it is a production ready product let's give morepress to more serious and solid applications servers such as JBoss and JONAS!"Hope that you don't mind Johny if i quote from a posting that you did on a different thread in order to comment this?"People like you XXX [name removed] are killing the OSS. If you do not want to use it, just do not use it. We need more open minded developers and less lamers..While they are "wasting" their time writing codeyou are wasting your time posting useless comments."Taken from here.CheersEike

    Sweet...I thought politicians needed to be careful.
  12. So what ?[ Go to top ]

    Hope that you don't mind Johny if i quote from a posting that you did on a different thread in order to comment this?"People like you XXX [name removed] are killing the OSS. If you do not want to use it, just do not use it. We need more open minded developers and less lamers..While they are "wasting" their time writing codeyou are wasting your time posting useless comments."Taken from here.CheersEike
    Sub thread "So What" ended here. Thanks Eike for the magic.
  13. Take it easy Johnny[ Go to top ]

    You cannot pretend Eike and Juan to understand what they read. It is fine they are not native americans, so take it easy with them, you already made a jackass out of Eike.

    However I see your point. Geronimo sucks big time and the code is a clear example of antipatterns. static code everywhere.
    So Apache or not, people should think twice before putting that crap in production.
  14. Take it easy Johnny[ Go to top ]

    they are not native americans, so take it easy with them

    What an irony. I'm not native american, and even I know that "take it easy WITH them" is incorrect. You should try "take it easy ON them". Or perhaps finish highschool; that would help. Not to learn english (you've demonstrated you're unable to) but in order to learn manners.
  15. Take it easy Johnny[ Go to top ]

    Not to learn english (you've demonstrated you're unable to) but in order to learn manners.
    Let me know which highschool teaches manners. :) I don't know of any that teach them. Sure doesn't seem so.
  16. Take it easy Johnny[ Go to top ]

    Hopefully, all of them.
  17. Geronimo, Tomcat etc...[ Go to top ]

    Hi

    Sorry to ask a basic question but:

    I understand that Tomcat is the apache JSP/Servlet container

    And that Geronimo is the server for EJBs from apache

    How (if at all) do they/will they integrate?

    Is Geronimo just an EJB container or is a super set of J2EE
    functionality including a JSP and Servlet container?

    I guess I'm looking at all this from a Vendor point of view
    where WAS or BEA or Sun's J2EE servers are all just bundled together in terms of containers...

    Thanks for your enlightening replies : )

    Mike
  18. http://chariotsolutions.com/geronimo/geronimo-book.pdf

    which tells me all I want to know, and far, far more...
  19. Final conclusions[ Go to top ]

    Reading this thread I clearly understand one simple thing.
    I could start coding randomly bunch of classes tonight to develop my "kernel" (of course including lifecycle interfaces ;-) for a week, then spend another week to select the open source products that implement the J2EE specs, put the code under the APACHE license (the key differentiator) and voila'
    before even releasing my code here the result:
    -support from the whole IT industry
    -make the majority of the Java community buy the idea I am revolutionizing the IT World
    -talk about high load/clustering/enterprise level with arrogance without even having deployed my product in a production environment
    -publish a book!

    I have learned a lot today!
  20. Final conclusions[ Go to top ]

    Reading this thread I clearly understand one simple thing.I could start coding randomly bunch of classes tonight to develop my "kernel" (of course including lifecycle interfaces ;-) for a week, then spend another week to select the open source products that implement the J2EE specs, put the code under the APACHE license (the key differentiator) and voila' before even releasing my code here the result:-support from the whole IT industry -make the majority of the Java community buy the idea I am revolutionizing the IT World-talk about high load/clustering/enterprise level with arrogance without even having deployed my product in a production environment-publish a book!I have learned a lot today!

    LOL!
  21. RE: Take it easy Johnny[ Go to top ]

    I think i understood what Johnny was trying to say.

    It just didn't make much sense.

    I can understand that somebody might be concerned about JOnAS getting not enough press coverage - but after all they are not native americans (just like me) and they maybe like to use different sites/channels for their press releases ;-).
    If Johnny is that worried about the lack of press coverage, he could just hit "Post a news item" and write about his experiences running "high volume mission critical J2EE applications" on Jonas.

    But to say that JBoss is not getting enough press coverage? - There are almost daily press releases by the JBoss Group (on the server side and various other).
    Maybe it would be worth to raise your concerns with Marc Fleury? ;-)
     

    cheers

    eike
  22. RE: Take it easy Johnny[ Go to top ]

    I am a native American and I think I can answer your questions:-

    Geronimo, wise leader also passed 1.4 test suite, not vapourware.
    JBoss don't knock it until you've tried it, then knock it.

    These facts became apparent while studying the entrails of a sparrow.

    Running Bear.
  23. Congrats[ Go to top ]

    This is what java is all about: cool choices. And commodization of fine, complete tools.
    It will be interesting to see if they can become the Tomcat of J2EE containers.
    & Hopefully with this step, Geronimo gets a little more momentum, they haven't exactly crossed the dev.speed limit until now. :-)
    Big hurray however. Great milestone. Keep up the good work.
    Phil
    (Ps For all those OSX users: it shines on Mac too! Try that with Bea, IBM WS, ... : no chance)
  24. Rewriting history[ Go to top ]

    It will be interesting to see if they can become the Tomcat of J2EE containers.

    Quoting Wikipedia: "Tomcat started off as a servlet specification implementation by James Duncan Davidson who worked as a software architect at Sun. He later helped in making the project open source and in its donation by Sun to the Apache Software Foundation."
  25. Hoew to spell Geronimo[ Go to top ]

    Hoew to spell Geronimo ?
  26. We are a 100+ person company in Seattle. We require 15 developers with 5 years or more of experience on Apache Geronimo and Java 1.5/1.6. Salary among the best.
    Please rush your CV to careers at mindstorm dot com.
  27. Congratulations to the Geronimo team! This is great news.

    The brief mentions that Geronimo would have lots of industry support, and specifically from vendors like IBM and BEA, which both have commercial application servers. Can anybody explain to me the rationale behind this? Why would they support a product that will infringe on their own market share?

    Cheers,
    GB
  28. why?[ Go to top ]

    Why IBM and BEA are supporting Geronimo??

    They can use Geronimo as a base and then fucus at things they can charge for. Today with Tomcat, Geronomi and JBoss it's more and more unlikey that anyone will pay for the basic application servers. Extras are not getting cheaper.
  29. why?[ Go to top ]

    Why IBM and BEA are supporting Geronimo?? They can use Geronimo as a base and then fucus at things they can charge for.

    IBM's strategy is here much more likely to try to kill JBoss, and BEA in the long run. Because they don't like the stanglehold JBoss Group retains over JBoss AS. It's not their understanding of open source. The way Big Blue does open source is by injecting cash and manpower in open source projects when they want to get rid of painfull low end competition. In Apache webserver to get rid of IIS. In Linux to get rid of Windows. In Eclipse to get rid of Visual Studio and in Geronimo to get rid of JBoss. BEA will be collateral damage, but they had no choice but abide by IBM's open source way. Else they would have felt very isolated...
  30. why?[ Go to top ]

    IBM's strategy is here much more likely to try to kill JBoss, and BEA in the long run. Because they don't like the stanglehold JBoss Group retains over JBoss AS. It's not their understanding of open source. The way Big Blue does open source is by injecting cash and manpower in open source projects when they want to get rid of painfull low end competition...
    Wouldn't it be better if IBM injected money in JONAS instead then?
  31. why?[ Go to top ]

    IBM's strategy is here much more likely to try to kill JBoss, and BEA in the long run.
    Wouldn't it be better if IBM injected money in JONAS instead then?

    Hmmm. I think there are several reasons:
    - IBM has been *very* supportive of ASF, OSDL, EF in the past, and much less of OW. Habits die hard.
    - "please use our code, we are the Apache Group", said Dain. Do I need ellaborate?
    - GlueCode was to sell. A pack with clever developers plus commit rights. Look hard, you won't find the equivalent for JonAS.
    - JOnAS has been around for over 6 years, it's certified for some time now, it has its ecosystem, there was no emergency in rescuing JOnAS. It's pretty high-end open-source (nice scalability, solid transaction management, and it already outperforms JBoss in many respects). Better buy GlueCode and end up with two nails in JBoss' coffin than invest in JOnAS and bear the risk of seeing GlueCode go out of business and Geronimo fade away. And now that Gero & JOnAS are alive and kicking, we may find synergies between the two projects. I mean, like real synergies.
    - ObjectWeb is definitely not organized to serve as a facade (I mean, membership is transparent, you know all corporate members by their names), and JOnAS is (so far) LGPL, a license that is not well suited to nurturing low-cost entrants on the software market. It allows people to actually innovate on top of it, not to recycle open source code into low-end proprietary offers, ie into offers positionned on the same market segment as JBoss.
  32. why?[ Go to top ]

    and JOnAS is (so far) LGPL, a license that is not well suited to nurturing low-cost entrants on the software market. It allows people to actually innovate on top of it, not to recycle open source code into low-end proprietary offers, ie into offers positionned on the same market segment as JBoss.

    What the heck are you talking about ? is JBOSS not LGPL ?
    ObjectWeb is definitely not organized to serve as a facade (I mean, membership is transparent, you know all corporate members by their names)

    How it is different from Eclipse foundation ? It is also transparent and you know the names of corporates behind it.

    Francois, you do not know why IBM chosed Geronimo over JONAS. :))
  33. why?[ Go to top ]

    and JOnAS is (so far) LGPL, a license that is not well suited to nurturing low-cost entrants on the software market. It allows people to actually innovate on top of it, not to recycle open source code into low-end proprietary offers, ie into offers positionned on the same market segment as JBoss.
    What the heck are you talking about ? is JBOSS not LGPL ?

    JBoss' positioning is that of a low-cost substitute to heavyweights. Their business model would work as well if JBoss AS was freeware instead of open source (except for PR). Apart from that, a single company basically steers the development, just the way it happens with proprietary software.
    ObjectWeb is definitely not organized to serve as a facade (I mean, membership is transparent, you know all corporate members by their names)
    How it is different from Eclipse foundation ? It is also transparent and you know the names of corporates behind it.

    The difference being that IBM created Eclipse.
    Francois, you do not know why IBM chosed Geronimo over JONAS. :))

    Nice try.
  34. why?[ Go to top ]

    JOnAS is the european answer to J2EE. IBM is an American Company. It's natural to select an American code... and JBOSS is private code, lgpl'ed, but private code.

    JOnAS is used by a leather cardholder to give service to 3 countries. JOnAS scaled better that JBoss and right now Geronimo is Vaporware.

    :))))))))))))))
  35. why?[ Go to top ]

    Why IBM and BEA are supporting Geronimo?? They can use Geronimo as a base and then fucus at things they can charge for.

    IBM's strategy is here much more likely to try to kill JBoss, and BEA in the long run.

    The truth is much less exciting, of course. IBM bought Gluecode because Gluecode had something IBM needed. Not because it married them into a lineage of kings, or allowed them to drip poison into the ear of JBoss. While the latter no doubt makes for better play writing, the former makes for better business.

    As far as Geronimo "killing" JBoss or BEA, how is that any more likely than it "killing" IBM? Come on, people ..

    The problem is that if one succeeds in convincing one's self that one is the center of the universe and the reason for its existence, then anything that anyone else does is obviously somehow about one's self. I won't name examples; I probably don't have to.

    Peace,

    Cameron Purdy
    Tangosol Coherence: Clustered Shared Memory for Java
  36. why[ Go to top ]

    Both IBM and BEA are loosing deals in favor jboss, which is free - at least the license is free - and of comparable quality. I see this with our customers: with our product we support wls, was & jboss and small clients tend to choose jboss.

    So what can IBM and BEA do to get a slice of this small business market? Push and open source app server and sell service arround it, of course. Because was or wls can't compete on the small business market with jboss because of their very high prices, but another open source server could.

    regards,
    e.
  37. why[ Go to top ]

    So what can IBM and BEA do to get a slice of this small business market? Push and open source app server and sell service arround it, of course.

    Was there something in JBoss's license that prevented IBM and BEA from commercially building on JBoss?
  38. LGPL[ Go to top ]

    Was there something in JBoss's license that prevented IBM and BEA from commercially building on JBoss?

    JBoss is LGPL. If IBM or BEA make changes to the server itself, they have to distribute the source for those changes.

    With Geronimo, for example, IBM can make changes to the server, and distribute them as closed-source components of Websphere.
  39. Congratulations from me, too.

    The pleasant surprise is still bigger, because I had sometimes looked at Geronimo's website and there is not too much progress to be seen (the last downloadable milestone 3 is from 2004-11-10).

    I am sure Apache Geronimo will be as successful as Apache Tomcat.
  40. Apache Geronimo has passed the J2EE 1.4.1 test compatibility kit, putting it well on its way to completing the entire certification process, after which it would become the third major open source appserver, and arguably the one with the most industry support (IBM, BEA, Apache), to reach J2EE compliancy.

    Congratulations to all Geronimo folks here!!! I like the "Tadam -- Geronimo is certified!" right in the middle of the BoF session ;-) Like what a surprise :)

    http://os3g.blogspot.com/2005/06/geronimo-is-certified.html
  41. The Bileblog[ Go to top ]

    Hani had the best take on this.
  42. Congratulations! It is very good.

    Our j2ee toolkit Super will support it.

    Wei Jiang
    Perfecting J2EE!
  43. "We are Apache ??? "[ Go to top ]

    "It was a very long 22 months to complete the test passing. We hope no one will ever have to do this again, please use our code, we are the Apache Group", said Dain Sundstrom, Apache Geronimo member. "This is a fantastic milestone, but we're certainly not done",

    Okay, how many people have chosen 1 open source over another just because it had the "Apache" label ? There are many high quality Apache projects, but I'm sure they also have their share of stinkers also. Each project needs to stand on it's own. Geronimo doesn't even have a release candidate yet. Why should anyone outside of IBM/Gluecode use this code right now ?

    BTW, I assume they will need to find an EJB3 implementation somewhere in the near future. Are they capable of developing this themselves ? Once the EJB3 stuff is integrated, a new round of certification will be necessary. Certification testing probably needs to be performed with every release.
  44. Completely agree[ Go to top ]

    with Steve. "We are Apache... everybody should use it"
    what an arrogant statement!!
    Better arrogant people that do deliver (i.e. JBoss) than arrogant people that take pieces of code here and there and "glue" them together pretending to have developed a cool product...
  45. "We are Apache ??? "[ Go to top ]

    "We hope no one will ever have to do this again, please use our code, we are the Apache Group"
    Okay, how many people have chosen 1 open source over another just because it had the "Apache" label ? There are many high quality Apache projects, but I'm sure they also have their share of stinkers also.

    The context of the quote was actually not arrogance (I was there). What Dain meant to say was that that Apache's license allows anyone in the world to just use and build off Geronimo. Up until now there have not been any open source servers with such a license.

    Dain actually did mean that he hoped no one else would need to go through the pain of putting a J2EE server through the TCK.

    Floyd
  46. "We are Apache ??? "[ Go to top ]

    The context of the quote was actually not arrogance (I was there). What Dain meant to say was that that Apache's license allows anyone in the world to just use and build off Geronimo. Up until now there have not been any open source servers with such a license.Dain actually did mean that he hoped no one else would need to go through the pain of putting a J2EE server through the TCK. Floyd

    AFAIK anyone who would like to build off Geronimo (be it only by changing a single line in the code) and claim J2EE compliance would have to get a J2EE license and... go through the aforementioned pain. The Apache license if of no help here.

    As for arrogance, I don't know Dain much, but for what I know him I would rather imagine him excited than arrogant.
  47. Goodlucl Geronimo[ Go to top ]

    My heart felt congrats to the Geronimo team, for I understand that its not easy to work with consistency against all odds and roll out a product which will serve the community for the times to come.The near future would see Geronimo attract all the attention & respect just like its ancestor the Vanilla king !
  48. I don't know if Suns recent release of GlassFish is consider 1.4 compliant or not (considering it's based on version 8, you'd like to think "yes", but...).

    It would be interesting to see a benchmark between Suns release and Geronimo.

    It's also an interesting question if you want to create a "new" app server based on an OSS solution, would you base it on Geronimo, or perhaps Suns.

    Of course, the detail is that even if you do create a new appserver based on either of these, you still have to go through the qualifiaction and compliance tests in the long run.

    Anyway, not to steal their thunder, but Suns new announcement shouldn't be tossed aside out of hand.
  49. It would be interesting to see a benchmark between Suns release and Geronimo.

    I honestly could not care less about that.

    It is very unlikely that the app. server will be the bottleneck in anything you deploy. But surely, your application will be the problem.
  50. It would be interesting to see a benchmark between Suns release and Geronimo.
    I honestly could not care less about that.It is very unlikely that the app. server will be the bottleneck in anything you deploy. But surely, your application will be the problem.

    Hum !
    I don't care either on generic benchmarks. Having said that, I DO care in specific bench I can do for my own application.
    We did one on top of major app servers stacks, - with the same app - : results : almost an order of magnitude between the first and the last.

    Laurent.
  51. Hum !I don't care either on generic benchmarks. Having said that, I DO care in specific bench I can do for my own application.We did one on top of major app servers stacks, - with the same app - : results : almost an order of magnitude between the first and the last.Laurent.

    Considering the maturity of J2EE servers, I doubt, unless your application hit a very particular spot rather than a large spectrum, that it can bring a significant difference assuming effort has been done to invest time in finding out how to tune each one.

    For instance, coming down to the servlet container level, historically, Tomcat was always perceived very slow compared to Jetty, but how many people did really stretch their application and the container to the limit so that the container was the bottleneck. I know only of one at a given time.

    Each app. server (or servlet container) has unique settings out of the box, which is never appropriate for production use but development, and is more or less slow or bound to crash in the medium term in production use. Unfortunately as pre-production testing is rarely done because it is considers by many (including customers) like a waste of time and money, what is deployed in rarely adequate.

    I have had customers requesting training on XXXXCache because they thought for sure that it was the problem with EJB entity caching in XXXX, while serving static pages was already taking ages...

    Another one complained to me that YYYY was unstable and that support provided by ZZZZ was rubbish...what was the problem ? YYYY ships as a default with unlimited size pool and the customer did not know they could tune it.

    Unless you're doing something very specific where one shines, as I said, I doubt there will be such a significant difference if you take the time to tune each of them properly.

    But knowing 2 app servers well is very difficult because most of the time, documentation is sparse...or you are simply drowning under the 25000 possible flags with hundred of useless services enabled...and time is very limited for testing.

    It is already quite an achievement if you are able to get people understand that testing is mandatory for the app. server you plan to deploy...let alone convince them to finance a comparison between two. And to find really good testers able to isolate the problems.

    For the records, in a previous experience I have had a QA guy telling me that the application was performing miserably under load and was degrading until a point where the machine was crawling and stopping...

    - "Is there any message in the logs ?"
    - "I don't know, I can't read them, the file is 4GB, that's also why I did not attach it to the bug tracker"
    As you can guess it was being load tested with full debug mode with everything (not rolled nor anything), dumped into a single log file...and it did not strike him that there was something slightly weird in doing performance testing with such a configuration.


    Thus in reality, because of human nature and time constraints, unfortunately the way the app. server is configured 'as a default' should be:
    - flexible enough to be used as-is during development
    - fast and reliable enough to be used as-is in production

    That's a difficult compromise, but the impact is important for adoption.
  52. Thus in reality, because of human nature and time constraints, unfortunately the way the app. server is configured 'as a default' should be:- flexible enough to be used as-is during development- fast and reliable enough to be used as-is in productionThat's a difficult compromise, but the impact is important for adoption.
    I beg to differ: app servers should be packed with the most secure and stable config, all in order to make production deployment safer and faster (just take MS' IIS history as an example). Developers would have time to change configs to suit development use, and can cope with less secure or stable servers, but production shouldn't be impacted just to make developers' life easier.

    My 0.00002c
    Henrique Steckelberg
  53. I beg to differ: app servers should be packed with the most secure and stable config, all in order to make production deployment safer and faster (just take MS' IIS history as an example).

    Well, that's the ideal world, you are assuming you are having a team of people perfectly knowledgeable with the product which is both nicely and concisely documented.

    And that is assuming documentation is being read...
    Developers would have time to change configs to suit development use, and can cope with less secure or stable servers, but production shouldn't be impacted just to make developers' life easier.

    I know I'm working in a country where software development process is mostly at the Precambrian level, but hell, I know there are still a lot of people out there unfortunately doing server-side development using startup/shutdown and copying ear/war with the file explorer. So I guess you can imagine the productivity level and the code shortcuts they have to take to 'make things work' in a given time.

    Now I don't advocate that the shipped app. server is configured in a totally broken way for production like XXXX is doing with its messy non-scoped classloader configuration as a default. But go figure, because people have been embracing it and most ignore that it is totally flawed...and production people are screaming over and over when things are suddenly going berserk after upgrading app. or adding app to the server.

    I have been reading different stories about this design decision. Either it was designed to be wrong in order to be fast , or it was designed to be wrong to hide classcast, classnotfound, invalidversion at the early age of it,... I frankly don't care but this is totally flawed as a default and is driving people into a thick wall.

    Which is why I said, it must be a compromise between production and development, because the lazy path will be taken in the traditional Paretto rule.
  54. Benchmarking an appServer is pretty much marketing stuff : "Our appServer is X times faster than others..."

    What we need is an appServer which is able to handle, let say, 100 req/s. No more. And this is a huge number. I've worked on really big organization, and even if they wanted to deal with more than this number, the fact is that they didn't reached more than 25 req/s, at peak, as they overevaluated their needs. So catch the reality : having 25 req/s is already a big number.

    If you are switching to an SOA architecture, this number will dramatically increase : let's say up to 200 req/s. At peak.

    So the main problem is - and I agree fully with Stephane on that - the code you write. It's so easy to mess with the code, in so many different ways, that any appServer will sucks if you put this kind of code in production. And it will be an easy excuse to say "this appServer is pure crap". But it won't help a lot.

    Be humble. Accept the fact : writing good code is an incredibly difficult task. So focus on what you can improve a lot : your code.

    PS : some will obviously say that their number of requests will be higher, and sure they will be right. But it covers mostly 99.99 % of all the application. And I think that I'm pretty conservative with 99.99% ;-)
  55. Please correct me if I am wrong, but I believe that Apache recognizes CDDL'ed code as acceptable for inclusion (as opposed to LGPL'ed code). GlassFish and Geronimo can comingle as much as they want to.

    --John
  56. Look on the specJ site SPEC and at the SpecJ app server numbers posted by Sun for the PE (free) and SE versions of the app server. They are very compelling numbers for the V8.1 product. Only IBM has been able to post comparable numbers (look at the app tier results)for a J2EE 1.4 app server. Since V9 is based on 8.1, you should expect that the open source version to be similar.
  57. I don't know if Suns recent release of GlassFish is consider 1.4 compliant or not (considering it's based on version 8, you'd like to think "yes", but...)

    Glassfish isn't J2EE / Java EE anything compliant yet (the spec. is not complete) but it will implement Java EE 5. But, if you want to start looking at Java EE 5 technologies you can start doing that today via binary distributions or buildable source.

    Congratulations to the Geronimo team on reaching this important milestone.

    Rich Sharples
    Sun Microsystems
    http://blogs.sun.com/roller/page/sharps
  58. This will change things...[ Go to top ]

    yes, i will now see job postings seeking experience with yet another app server.
  59. Glassfish is irrelevant[ Go to top ]

    Open source projects are successful when they are built around novel architectural ideas or to meet pressing needs for developers. Struts for example satisfied the need for MVC framework. Spring filled in the need for highly productive environment for building J2EE applications. Jboss started with a new architecture for application servers and Geronimo with its unique Gbean architecture presents interesting perspective for container pluggability. However, a failed corporate effort rubbed as open source was never and will never be successful. Glassfish is a classic example of this. After unsuccessfully trying to sell application server in various incarnations, Sun finally decided to open source their application server. To serve whose interests – I don’t know. It is time for Sun to stop doing me too products and support one of the successful open source projects or move on to some other interesting products in the larger interest of the community and of course in Suns own interest.
  60. Glassfish is irrelevant (not)[ Go to top ]

    It is time for Sun to stop doing me too products and support one of the successful open source projects or move on to some other interesting products in the larger interest of the community and of course in Suns own interest.

    My irony-meter just blew another fuse. Let's step back a bit. The thread started out congratulating the Geronimo team on achieving a major milestone - passing the J2EE 1.4 CTS; J2EE 1.4.

    Sun released it's FREE, production quality J2EE 1.4 application server in November 2003, we've shipped 2 significant updates since then, registered approximately 3 million downloads and submitted the *only* SPECjAppServer2004 result for a FREE product. How exactly is that a me too product ?

    Sun are now committed to releasing the next version under an Open Source license and in an environment that will hopefuly foster greater involvement from others. We hope this will broaden the market and further accelerate adoption of Java EE. What exactly do you have to fear from this ? If GlassFish is so irrelevent - why do you feel you have to go out of your way to post FUD ?

    Rich Sharples
    Sun Microsystems
    http://blogs.sun.com/roller/page/sharps
  61. Glassfish is Irrelevant[ Go to top ]

    Rich. No FUD here. Just waking you up to reality. Download numbers do not mean anything. How many places is the Suns app server used in production? How many people are excited about Glassfish ? On the other hand Geronimo generated lot of interest in the community even though it has just passed J2EE certification. (Partly due to Apache web server Success and partly because of its unique architecture). It is in better interest of community if Sun backs one of the open source application server and help them instead of producing another product which hardly generates any excitement. Better still if Sun can produce product which is under served in the open source like for example a JBI (JSR 206) based product.

    I am not associated with Geronimo or JBoss or Jonas (so I do not have any vested interest here). I am just trying to see how Sun can be more relevant and be more helpful to the community. Surely open sourcing an unsuccesful product does not serve the community. Sun can better focus else where for greater good of the Java community instead re-branding under various incarnations. Show me one example where a commercial organization has turned its unsuccesful product into ahugely succesful open source project.

    Peace...
  62. Qiwei Yin[ Go to top ]

    Qiwei Yin
  63. Glassfish is Irrelevant[ Go to top ]

    Rich. No FUD here. Just waking you up to reality. Download numbers do not mean anything. How many places is the Suns app server used in production? How many people are excited about Glassfish ?

    1. Honestly, I'm not personally excited about Glassfish.

    2. However, I do see Sun's app server used in production, about as much as I see Oracle's for example. It's not as popular as JBoss, but some pretty big name sites run on the Sun server.

    Besides, Tomcat is probably the most popular app server, and Sun wrote it, and (AFAIK) the Sun app server is still built with Tomcat as its Servlet container.

    Peace,

    Cameron Purdy
    Tangosol Coherence: Clustered Shared Memory for Java
  64. Glassfish is irrelevant[ Go to top ]

    Cameron

    Yes Sun was the initial contributor to Tomcat and contributes even today. Calling Tomcat an app server is a strech though. Besides my point is that of relevance. You personify the general mood of the community in stating that you are not excited about Glassfish. Tomcat was relevant when it was released. It served a need not fulfilled by anyone then. Glassfish does not serve any need now. Sun being a benevolent grandfather of Java community is better off spending its resources for greater good of the community and that is not surely in open sourcing a failed product.
  65. Glassfish is Irrelevant[ Go to top ]

    I disagree that GlassFish is irrelevant. It's as relevant and any other OSS app server project.

    Do we have a lot of OSS App Servers now? Yup, we do. Does it risk fragmenting the developer base? Yes it does, but OSS is (in)famous for that anyway.

    If I were a developer looking to tightly integrate an app server into a custom application, (vs just writing to EJBs and the J2EE spec), I think you'd be foolish to not look at GlassFish along with the others.

    But then, when Sun released PE 8.0, I was one (of the few, apparently) that was excited to see it come out.

    I agree with the others about at least at a certain level, Appservers are quite similar, and for many applications the performance between one container over another is minor. I'd say that of the modern servlet containers. Most likely the application itself will wash over any of the real differences.

    But what I liked about Suns appserver were several things.

    1) It was free for deployment.
    2) Well documented (not as good as Weblogic, but a far cry from other projects).
    3) Decent admin interface through the command line, and the current version has a decent Web GUI for it.
    4) It wasn't substantially different from the higher models. Not a lot of Bait and Switch going on.

    Mind, I was focused on smaller department size, single server applications vs monster things like eBay.

    Because I felt that it was important for something like PE 8 to be available to ISVs writing custom solutions for customers and being able to keep costs down, and being productive on the server.

    I've never been a fan of JBoss. I've never liked trying to navigate their documentation to get the thing running, so I felt the Sun appserver filled a void that helps promote J2EE.

    I only wish it was more obviously able to run behind IIS, rather than in lieu of it. (This may have been fixed in 8.1, I haven't looked).

    With the new integration with NetBeans, you pretty much have a double-click end-to-end package for development and deployment, which makes entry that much easier. So, instead of a .NET solution with VSS, you have a viable Java solution, and it's free to all comers, which just promotes the platform even more so.

    The only thing it lacks is "Street Cred", and why that is I have no idea. I haven't see anything that documents what's so damningly horrible about their server compared to other options.

    And finally, it just continues along the front where Sun is opening up more and more of their core infrastructure. Sure must be encouraging to them to go through the headache of vetting their code for opening up only to have every release flamed down, but that seems to be the hobby de jour.

    If they ever release Java, they'll be condemned for that as well. Sun can do nothing right.
  66. Glassfish is Irrelevant[ Go to top ]

    Rich. No FUD here. Just waking you up to reality. Download numbers do not mean anything. How many places is the Suns app server used in production? How many people are excited about Glassfish ?

    OK, last post on this subject from me; the thread is about Geronimo and I don't want to distract people from their good news.

    I'm just interested in what motivates you to go out of your way to post to TSS (which you haven't done before - at least not by your current name) to bash Sun for doing something that will only benefit the Java community.

    To answer, some of your questions :

    Download numbers are a great way to track a project's success - they demonstrate the level of interest and relevance - we're hardly alone in using this metric.

    Sun's App Server is used in many places - big and small, worldwide - from small departmental deployments to some of the largest J2EE deployments on the planet. It's delivered as part of every major Software release from Sun - Java Enterprise System, Tools, Solaris.

    I can't tell you how many people are excited about Glassfish but if you went along to the GlassFish BOF @ JavaOne this week you would have no doubt come away with the feeling that GlassFish is relevant (it was packed, standing room only). The JavaOne pod had a steady stream of people asking hard questions from people who were interested. The click-through stats. and page hit counts for the GlassFish site have exceeded our expectations so far.

    The great thing about the world of Java is that vendors (and OSS projects) can compete. We can collaborate on specs. and standards but compete fiercely on implementation. That competition is what drives the evolution and success of Java.

    Let a thousand flowers bloom.


    Rich Sharples
    Sun Microsystems
    http://blogs.sun.com/sharps/
  67. Glassfish is Irrelevant[ Go to top ]

    ...I am just trying to see how Sun can be more relevant and be more helpful to the community. Surely open sourcing an unsuccesful product does not serve the community. Sun can better focus else where for greater good of the Java community instead re-branding under various incarnations. Show me one example where a commercial organization has turned its unsuccesful product into ahugely succesful open source project.

    Possibly IBM's Eclipse platform?

    Glassfish is a big project, but with the CCDL licensing, there is a number of J2EE 5 specifications under the hood.

    Everyone was buzzing about JSF, and Sun's reference implementation is the best implementation out there. Sun's even gone as far as making the distribution licensing more flexible in order to appeal to other developers that have felt the need to pursue MyFace's development version.

    Also, within the Web Tier, there's the new Grizzly connector, which already other developers are contributing to in order to make it even more performant. Also, much of the Jasper compiler was written by Sun employees. Now with the J2EE 5 webtier alignment, JSP 2.1, the EL-API, and JSF-RI are all available under Glassfish.

    I don't think Sun's telling you to use Glassfish in production right now, but it's worth developing projects on for J2EE 5 features right now.

    -- Jacob Hookom
  68. Glassfish is irrelevant (not)[ Go to top ]

    I completely agree that Glassfish *is* relevant. Sun has lots of potential to push ahead with J2EE 5 as an appserver vendor. Not only is Sun involved with many of the JSRs that form the J2EE 5 platform, but they have been developing reference implementations of those specs for over a year in some cases.

    The fact that Glassfish is now opensource is somewhat moot, but the potential to be even better is there with community contributions-- JSF, Grizzly (!), EJB 3, EL, etc. It will be a long time before the other vendors get up to the J2EE 5 specification and here Sun is making their J2EE 5 container readily available for download.

    For those looking for ease of use in an appserver, the project uses a couple maven targets to get up and running and their admin console is very easy to use. I know that doesn't mean much to some, but it's still worth a peek :-)

    With the opensourcing of Glassfish on Java.net along with other JSR specs, *anyone* has the possibility to add/write code for a J2EE 5 appserver. It's pretty cool when you think about it...

    -- Jacob Hookom
  69. EJB 3[ Go to top ]

    Congrats for getting J2EE 1.4 certified, Are you thinking of Support of EJB 3 when it will be released.
  70. Goodbye BEA[ Go to top ]

    Somebody better buy BEA soon (for their customers). Let's face it, app servers are a small market opportunity now.

    Apache rocks!
  71. Goodbye BEA[ Go to top ]

    Somebody better buy BEA soon (for their customers). Let's face it, app servers are a small market opportunity now.Apache rocks!

    Funny, but I don't think all that accurate. There is room for competition from vendors too. I haven't seen any benchmarks or quality reviews of Geronimo yet so it is way too soon to say whether it will become the de facto application server that steals market share from the big guns.

    Plus, BEA and IBM seem to be putting in much more effort on their add-ons: portal, soa, gui/ide, etc. Something that Geronimo will lack for awhile. It is these flashy add-ons that sell contracts to the managers, not server specs.

    I do hope that people take Geronimo and run with it. I'm tired of working with containers that are only documented skin-deep (I'm looking at you IBM), where you have to spend a day digging through google and blogs to find information on how to use a product's feature. Don't tell me to use JBoss, etc. I'd love to give it a try but I got to use what my clients use.

    At least if the container is open-source I can look at the code and figure out why a feature doesn't work as advertised. Of course, until then a good decompiler helps.
  72. Well, unlike the posters that go why Geronimo and not JBOSS/JONAS who seem to miss the point of the post (Geronimo passed the TCK) I would like to say a big CONGRATS to the Geronimo team for making that happen. Another pieace of super quality open source code out there :-D.

    I'd like to respond to the posters who say "why Geronimo ?" with a "because that's what I like". It's simply matter of taste to choose Geronimo. If you go a little bit underneath and see the whole Geronimo philosophy (all that "building the server from the modules you want" and Spring/Whatever integration) you'll be more than excited and wil ldefinately want to bring the FUN into your day to day development. No more the PROFESSIONAL :-)). Not that Geronimo wouldn't be professional. But because most important is that it is FUN and professional, it stimulates creativity and opens up ways that were closed until now.

    So go have fun with it.

    Another warm congrats to the Geronimo team. I'm not wondering if they will EVER have a production quality app server, now that they passed the TCK, but I am wondering WHEN.
  73. WHEN?[ Go to top ]

    I do not care about "Apache label" and I do not care about GBean architecture.
    I have been running the whole stack of my J2EE applications on top of an open source applicaton server.
    There are few choices out there about open source application servers I could give a try
    but definitely Geronimo is not one of them. Let's be realistic and stop playing politics.
    How can it be production ready if they do not even have a RC ?
    You guys gotta be joking!
    Yes, the question is WHEN ? Well I need to run my business today not in 2010.
    I can also see how the Java community has been brainwashed about that.
  74. yesterday's J2EE architecture[ Go to top ]

    By the time Geronimo gets fully 1.4 certified and becomes production ready, I'll be moving server-side development into an EJB3 embrace. New software architectures to be developed will be founded on EJB3. The last thing will want to do is start creating new systems today that are based on yesterday's J2EE architecture (which of course had numerous problems - such as the CMP entity bean model).

    The only way would possibly pay any attention to Geronimo is if it could offer EJB3 compliance but it won't anytime soon - so it's irrelevant.