This looks great Pavel.
I've got an odd sort of setup that I've been messing around with. I have a Glassfish 3.1 with a Jersey 1.7 deployed, and I use HttpClient (for now.. going to use Jersey Client especially with this addition) sending RESt requests to another Glassfish server. Basically my first server is the API tier that does filtering, validation, logic, etc..then hands off the "work" to worker machines. Anyway, by using Rest between the servers, the back end worker servers can be any sort of implementation in any language or platform. However, I am using Jersey (again) to handle the REST calls from the front to back. On the back server, because it's a JEE6 container, I do have the jersey resource call an EJB 3.1 stateless session async method. This allows my request/response to be almost immediate between the servers. I am wondering if there is any benefit to me using this new async code on my front server since I have the back server call to an async method already? I
am guessing not, since my request/response times are so fast.. however if the back end server is not my own.. some other language/platform, then I suppose this new async code would be beneficial for the "just in case" scenario?
--- On Thu, 6/30/11, Pavel Bucek <pavel.bucek_at_oracle.com> wrote:
From: Pavel Bucek <pavel.bucek_at_oracle.com>
Subject: [Jersey] Jersey non blocking client
To: users_at_jersey.java.net
Date: Thursday, June 30, 2011, 1:44 AM
Hello all,
I merged one of our branches with trunk earlier this week and I would like to receive some feedback before we move it from experimental area to stable.
If you are interested, please see
http://blogs.oracle.com/PavelBucek/entry/jersey_non_blocking_client
Thanks!
Pavel