Server TF Minutes 2007-02-14

From MemberWiki

Jump to: navigation, search

Minutes stored at url: /member/wiki/Server_TF_Minutes_2007-02-14

Attendees

  • Ric Smith <richard.allen.smith(at)oracle.com>
  • Ted Thibodeau <tthibodeau(at)openlinksw.com>
  • Christophe Jolif <cjolif (at) ilog.fr>
  • Jon Ferraiolo <jferrai (at) us.ibm.com>
  • Ken Tam <kentam(at)bea.com>
  • Greg Murray <greg.murray(at)sun.com>

Original Agenda

    • Review jMaki's plugable server-side architecture
      • GOAL: To identify possible requirements for a standardized platform
    • Handle any outstanding business
    • Wrap up

Minutes

Greg: Contract with the server ends at the URL. JSF implementation has the most bindings at the JSF server.

Currently, trying to extend that to the JSP world.

The only thing we hope for in jMaki is data is handled with JSON. The hope is to eliminate XML.

I prefer a world where the view is handled purely in JavaScript.

Lookup different components based on method binding settings. Handled by PhaseListener.

How do you handle more complicated queries with combo boxes? Desire to do this without multiple URLs. This requires a server agnostic approach that has yet to do this. This requires looking to see if XMLHttpRequest can handle anything other than POSTs and GETS.

Jon: Ajax is always a post. But could be handled further in the header.

Greg: Rather encourage best practices, than require a specific implementation.

Jon: On the client-side we use a hub. Could this possibly be a concept ported to the server.

Greg: Yes. But, you need to centralize this on this server.

Jon: Two socket limitation in Comet.

Greg: I have a lot of resources in this area. And we are starting to focus more and more on this area.

Support for Facelets and Portlets has been added to jMaki in the past few weeks.

A question to ask is should every piece of data be addressed with a URL. Could be a security issue in terms of JSF where we need to embed the ManagedBean calls, opening up your model to the rest of the model. If we follow a model where expressions map to objects we will have to build those facilities for technologies other JSF.

Craig is doing Shale Remoting and jMaki has its own implementation, but we need to consolidate this.

Jon: There is a possibility of brining Joe Walker into the Server TF. We are getting a Zend member on the TF. Ric can you email Joe and see if he would be willing to join the TF. I need to talk to the Ruby guys and bring them into the mix. Should definitely tried to work with them.

Greg: Working on one beach head at a time but Ruby is on my list. Really want to be good at one thing and want to stop at the URL.

Jon: Craig has been wrapping our server side wrapper with more widget specific stuff on the Java side. Our tooling requires Java class to back the widgets.

Ken: How do you do tag completion?

Greg: Have not accomplished that yet, we can do it we simply have not implemented it yet.

Jon: The eclipse guys offer code completion.

Greg: The eclipse plugin for jMaki is available this week. We extended the WTP stuff not JSF.

Jon: ATF stuff is focused on the client side. On the IDE task force we want to focus both on client side and server side tooling. Need to accommodate both.

Ken: Greg, there is concern about jMaki integration with Portlets. We fixed a few bugs and can now work with it.

Greg: The take away from this is that we are starting use Prototype and Dojo which is great but we often end having to use one specific version of either framework.

Jon: In order to meet OpenAjax conformance you must be backward compatible. Greg: In theory, if you have your name space within in an existing namespace you could swap it out.

Jon: So you convert to REST on the fly?

Greg: In a sense yes but this may change in a few weeks.


Greg: I will keep you posted on how we progress. There are a lot of changes coming in March.

Jon: Hub Meeting Replaces Server Meeting next week.