Hello,
This is the second proposal that I submitted to GSoC. Please feel free to leave any comments on it.
Thank you.
One thing I'd like to suggest is that, at least for me, it would be useful if this kind of work could be passed off to another machine to handle. I know this request drastically increases the work load, but I think it would be beneficial to have a setup where one box handles server-to-client and another could handle server-to-server and other tasks that don't involve user interaction. Of course I would still want to be able to do both operations on the same machine, too. Basically, I would like to see this scalable.
As a side note: I notice both of your proposals so far deal with distributed systems. My proposal deals with distributed databases and allowing rails to interact with them (Sorry for the shameless plug). I was wondering if I could get your impressions on my proposal since are projects are sort of similar (maybe loosly). Thanks.
http://groups.google.com/group/rubyonrails-core/browse_thread/thread/a724099bfa2b493c?hl=en
Thanks for the suggestion Allen. Considering the scalability of this solution, I'm thinking on use starling to handle the asynchronous calls. Starling by itself is a perfectly scalable solution that allows you to add separated servers to handle asynchronous messages. My plugin should provide an easy way to configure extra servers if necessary. You can check episode 128 of Railscasts for an example of starling.
I'll give a look to your proposal later on.
