I've been trying very hard to use BRITS (SpringSource bundle repository) as much as possible and to avoid using the "temporary" Spring-DM repo on Amazon S3. But I keep running into things that lure me back to the temporary repository.
I thought I'd bring them up as items to discuss in this forum...just to see if I'm overlooking something that someone can help me with. And, I'll create JIRA issues against BRITS to see if these problems can be addressed.
Specifically, I have three issues at the moment:
First, the Catalina bundles in BRITS have *required* dependencies on javax.ejb, javax.mail, and several other similar packages that I'm pretty sure should be optional. This means that I can't use Tomcat and the web extender from BRITS unless I also install a ton of other bundles that my app doesn't really need. This seems to violate the spirit of modularity, as you don't really need EJB packages to server web pages with Tomcat.
Second, if I use the web extender from BRITS, it defaults to using Tomcat--as documented--but there isn't a web extender Jetty fragment that switches the web extender to use Jetty. There *is* such a fragment in the temporary repository, but that fragment's Fragment-Host references a symbolic name that is different from the web extender in BRITS. It seems that if the web extender in BRITS is going to have a different symbolic name than the one in the temporary repository, then there should also be a corresponding Jetty fragment in BRITS.
Finally, I submit that if the temporary Spring-DM repository is truly temporary and if the Spring-DM bundles in BRITS are going to have different symbolic names, then there is a fairly significant inconsistency in the Spring-DM documentation. Perhaps the Spring-DM documentation needs to be updated to reflect the current selection of symbolic names.