On Thursday 19 January 2006 05:12, Martijn Dashorst wrote: > I would like some input from Timothy Bennett, our oscar/felix/osgi > guy, and see whether he has some objections from that > framework/container point of view.
Sorry, for being a bit late on this... Tim contacted me to see if I could figure out the consequences (besides the factual change of API) that this would have in highly dynamic setups. Personally, I am putting all Wicket stuff into a single bundle, which then picks up the "business logic" via OSGi service registrations. In that context, there is no impact on the feasibility of the proposed change. And I personally prefer constructor injection over setters/add mechanisms. Interesting to note that Eclipse/SWT uses the "parent in constructor" which initially confuses Swing developers and one often hear "I can't do X" early in the adoption cycle (myself included) but after groking the new patterns, I liked this approach a lot better and felt it reduced coding errors. Now, Tim has been considering using an additional approach to Wicket/OSGi integration, where he envisions a page or a group of pages to reside in a bundle, which can be loaded and unloaded any time in runtime. By itself a very interesting idea... However, I think we are possibly looking at 2 issues; 1. Runtime manipulation of links to dynamic pages. 2. Classloading issues between bundles. As I don't have any experience with Wicket over multiple bundles, I have nothing much to add at this point in time, but I think the constructor injection doesn't change the underlying issues. Since it is an interesting idea, I may spend some cycles on it. Now, for people who uses Spring, I can't see that this would change much either, as Spring nowadays supports constructor injection, both by value as well as by reference. Someone raised a usecase of extending the binary runtime, and that CDI would make things very messy. I would need a bit more info on that to see if there are any validity in the claim, or just the "fear of change" before "emergence of new patterns", just like a Swing->SWT transition do. Finally, as for when this change is to happen, I think it is important not to listen to people on the list, but put yourself in the shoes of those who have commercial stakes in progress. I am not one of them (yet), but I think you owe them a 1.2 release. The other issue is whether 1.2 and a new 2.0 can co-exist within the same application (like Hibernate2 and Hibernate3). If that could be catered for, I think everyone should be really happy. Cheers Niclas ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642 _______________________________________________ Wicket-develop mailing list Wicket-develop@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wicket-develop
