Hi Jon, Well written. I understand your pull vs. push model. What I have proposed is a push model no doubt. The push model simply requires the GUI designer to understand page layout. All of the data access, looping and conditional logic is still the responsibility of the java programmer AND his code IS tied to the layout. You are right about that.
The whole point of your article, unless I missed something, was that the pull model is superior because it allows the GUI designers to be "free from the Java engineers" by giving them complete control and responsibility for the UI. This is true. Except it essentially makes the GUI person an entry level programmer themselves by requiring them to understand data models, APIs and (in the case of JSPs), the Java programming language. If you have a UI person that can do that, you are lucky and the pull model may be better. If not, the Java programmer will have to write "dumb down" code so that the UI person can live up to their new responsibilities. How do you do this except to design beans that match the page they are to be used on. If you do that, you've defeated the whole purpose of the pull model and you're back to Java programmers coding for each web page. So, I don't think one way is obviously better than the other. It may just depend more on the team doing it than anything inherent in the design paradigm. The second problem I have with JSPs and scripting engines like Velocity is that the end result is a file with THREE different languages all mixed together (HTML, JavaScript and Java or some other templating language). That just spells "mess" to me. At least with what I was proposing you've seperated out the client side display description and control code from the server side data generation. Maybe I'm just being anal on this point. :) So, the whole point of this is that I think ECS could have a role in this push model beyond what it has now, or used to have. It already has the data model for HTML. If that model could be created from an HTML file and the API tweaked to work with that (such as find() methods), I think it would be exceedingly easy and productive to use. At least it would offer a workable alternative for people who don't think crowding everything into one ugly file is the way to go. JMHO. John -----Original Message----- From: Jon Scott Stevens [mailto:[EMAIL PROTECTED]] Sent: Wednesday, August 14, 2002 4:16 PM To: ecs-dev Subject: Re: html2ecs on 8/14/02 1:39 PM, "John Crater" <[EMAIL PROTECTED]> wrote: > Someone please tell me why this logic is flawed. > > John http://jakarta.apache.org/turbine/turbine-2/pullmodel.html -jon -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
