Why we stopped using Virgo/OSGi in our project [message #997180] |
Fri, 04 January 2013 09:55 |
Erik Vande Velde Messages: 82 Registered: September 2012 |
Member |
|
|
After a few months of experimenting with OSGi/Virgo we decided to give up on it for these reasons:
(1) It's impossible to use RMI remoting (see http://www.eclipse.org/forums/index.php?t=rview&goto=981292#msg_981292)
(2) When using spring remoting (instead of rmi) our bundle automatically became a war instead of a jar, and other bundles had to depend on that war. Maven doesn't like dependencies on wars.
(3) Using AOP in a modular world gives trouble too (http://www.eclipse.org/forums/index.php?t=rview&goto=994544#msg_994544).
(4) For JPA you have to list all Entity types in the persistence.xml, because scanning for Entity annotated classes doesn't work.
(5) Using transactions with JPA in an OSGi world is more complex then in the traditional java + spring world. Still it's possible to do it as explained by Dmitry here: http://www.eclipse.org/forums/index.php?t=rview&goto=991427#msg_991427
(6) Maybe even the most important argument is that the sample application doesn't work: http://www.eclipse.org/forums/index.php?t=rview&goto=936829#msg_936829. More generally: there aren't many books/internet resources available explaining how you should structure your application, and address typical enterprise concerns like AOP, persistence, transactions, remoting, etc ...
Briefly: enterprise application developers like me encounter too many hurdles on the way because they use OSGi + spring (Virgo), and the productivity drops significantly because you have to post your questions on the forum, and hope that someone replies. I gladly admit that the experts were very helpful in answering my questions, but in a non OSGi world these issues simply don't exist.
Conclusion: The OSGi and spring combination is indeed very interesting, but today it isn't mature yet. This post is certainly not intended to offend the people working on Virgo and OSGi, but maybe it can help the experts in determining what needs to be addressed to make more people use their technologies ...
|
|
|
|
Powered by
FUDForum. Page generated in 0.03464 seconds