Hi All,
I have to investigate switching back to JBoss today, and I was going to just do the switch and forget about things, but I thought that I'd do a post in the hope that my experiences can help out the glassfish team and other users. This is a bit of a negative post, but please take it in the constructive manner in which I intend it. Hopefully my experiences will assist the glassfish team to adapt and become more successful.
A few years ago, at a Sun conference targeting partners and certified architects, a Sun speaker requested that we start considering the Netbeans/Glassfish stack instead of Eclipse/Spring/Jboss which is by far the market share here in Australia.
Because I owe so much to Java and Sun (essentially my entire career), I started to experiment with the Netbeans/Glassfish/JEE5 combination, and initially saw much potential in it, and started to champion it.
It took a looooong time for clients to take me seriously. I finally got my way on a project, and over the last year development has been quite good, and Netbeans has been a good platform.
Unfortunately, during deployment into production we found a pretty big bug in Glassfish v2.1, the stable production release. A web tier object won't repeatedly connect over iiop to a remote EJB application. This is stock standard JEE, and core functionality that JEE developers have been using for years. But I understand that bugs exist and it's no ones fault, in fact it's an opportunity to make a product even better.
So I prepared a test project that reproduced the issue and submitted 2 P1 bugs 8589 and 8590, along with massive amounts of configuration information and an offer to help as much as is useful.
That was two weeks ago, and after querying why the bug hasn't even been triaged, I was told that "it's not clear when we'll have time to investigate"...I would understand if this was a P3 or P4, or if it involved something on the fringes like using a custom JCA connector...but this is a P1 stopping a web tier using iiop to connect to an EJB tier...that's bread and butter stuff to a JEE developer.
Anyway, since development has finished and we're preparing to move into production, I've just been given the call to switch to back to JBoss, and I can't stall any more.
I understand this is an open source project, and one could argue that I should buy a service agreement license if I want bugs addressed, but one could also argue that open source is a collaborative environment, and to be take seriously, you have to be willing to meet the expectations of both your paying and community partners.
Over the years I've posted many issues to the JBoss forums (and indeed many other open source projects)...and have been involved in submitting many patches to many different projects. JBoss has never taken more than a few hours to get back to me on anything greater than a P4...and while the answer is not always the solution I wanted, at least they investigated and addressed the issue and gave me a way to keep the team moving forward. Indeed I think the longest I've ever had to wait for significant feedback from an open source project is about 3 days...until now anyway.
I offered to try and fix this issue myself with a little direction, an offer that is usually met with gratitude from open source projects...but to no avail :(
Anyway, it is with a very heavy heart that today I slink off to the Jboss site defeated as we plan a migrated back to eclipse/jboss. I can't in good conscience recommend Glassfish to my corporate clients (most of which are investment banks, govt departments and insurance companies), not because of the product itself (which I really like, and think it generally easier to develop against than JBoss), but because of the uncertainty around it's ability to be used in a production environment without a license (many of the departments I consult with and teach at have multiple JBoss installations without licenses working in production)
Anyway, I just thought I'd share my concerns. Like they say, for every one person that complains there's a hundred that stayed silent. My suggestion would be to spend some time looking at how bugs are triaged, or to state as part of the community download in big bold letters that preference in bug fixing will be given to paying clients (if that's is the case, I have no visibility into things, so I'm making an assumption which is possibly incorrect).
Thanks for all the help from the list during development. I've learned a lot about Glassfish over the last year and will continue to track progress of the product, looking forward to the day I can use it in production.
Cheers
Adam
____________________________________________________________________________________
Access Yahoo!7 Mail on your mobile. Anytime. Anywhere.
Show me how:
http://au.mobile.yahoo.com/mail