Ok -- any other comments or concerns on this, or have they all been addressed?
I'm not going to commit this patch until Brian Eaton's new OAuth stuff is committed, because this patch doesn't properly deal with any token or signing related code, and I don't want to break anyone depending on it. Brian has told me he'll have something soon -- so hopefully this patch can go in by this weekend (or early next week). Any objections to that schedule? On Thu, Mar 27, 2008 at 12:15 PM, Kevin Brown <[EMAIL PROTECTED]> wrote: > On Thu, Mar 27, 2008 at 6:39 AM, Michael Mahemoff <[EMAIL PROTECTED]> > wrote: > > > It's encouraging to see DI incorporated into Shindig's architecture in > > an explicit manner such as this. > > > > To accommodate the diverse needs of Shindig consumers, I'm hopeful > > Shindig will, over time, become flexible and plugin-based. Adoption of > > Guice would be a major step in that direction. > > > > Slightly OT, but I expect the Javascript code will also need a means for > > consumers to switch between different implementations, in order to make > > browser behaviour more flexible. It would probably work simply by > > redefining certain variables, but would need a clean separation of > > concerns in order to work. (Similar to JQuery's plugin architecture.) > > > You can already drop in a custom implementation of any js feature. As long > as the new code conforms to spec it's fine to replace it in your own > deployment. In fact, it's actually a requirement for supporting > opensocial-0.7 today. > > The other js (stuff in the javascript/) directory isn't nearly as robust, > but I wouldn't really endorse using that code on a production site anyway > unless you feel like doing a lot of work. Paul, Zhen, and Cassie have all > been doing some work to make it more useful, but I feel that it should be > scrapped and started from scratch to actually try to meet the needs of real > sites. The original version was only created as a stopgap solution > anticipating the creation of the gadget server. > > > > > > > Kevin Brown wrote: > > > Hi everyone, > > > > > > If you're not working with or interested in development of the Java > > > implementation, you can ignore this. > > > > > > After discussing this on several separate occasions, I've decided to > > go > > > ahead and give using Guice a shot. So far it's coming along pretty > > nicely, > > > though it does require a few small changes to some of the current > > behavior > > > (mostly cleanup of stuff that was known to be crufty and didn't really > > get > > > revisited in the last refactoring changes). For the most part, our > > code was > > > pretty DI friendly to start with, so the changes are minimal. Probably > > the > > > biggest change is that most configuration will be moving to an > > external > > > properties file, and some new classes will need to be introduced to > > handle > > > various url generation pieces of CrossServletState. > > > > > > When I get closer to something complete, I'll be doing the usual JIRA > > / > > > patch submission. > > > > > > If anyone has strong objections to Guice, feel free to speak up -- > > I've > > > personally never used it myself, but I've heard good things from other > > > people about it, and it seems to me to be a bit easier than Spring. > > > > > > For reference: > > > > > > http://code.google.com/p/google-guice/source/browse > > > > > > One minor annoyance -- I have had some trouble finding an up to date > > maven > > > repository, so I'm only using the initial 1.0 release that's available > > on > > > http://repo1.maven.org/maven2/com/google/code/guice/guice/1.0/ -- this > > > doesn't include some of the servlet integration work, so I still had > > to wire > > > up pieces by hand. > > > > > > > > > > -- > ~Kevin -- ~Kevin
