UptoData, Inc.

gwt: initial [raw] thoughts..

Posted by Eitan Thu, 18 May 2006 01:16:00 GMT

so google finally shows its cards. they've recently announced gwt: the google web toolkit.

before i get going let me first say that i think it is wonderfully cool for a company like google to work out and develop a solution for the web stack, and then to turn around and open source it; i.e. to make it freely available for anyone to use.

ok, let's get going.

i find gwt very interesting, to say the least.

[a] it's simple. i was able to read their documentation and understand how to go about building web apps with gwt in very little time. +1.

i really think it's too early for many of us to realize exactly what gwt brings to the table and what implications this is going to have on the way we build web apps.

[b] i'm pleasantly surprised to see this solution leverage java.

i'm also surprised to see a solution that provides a way to build ajax applications without getting one's hands dirty with html and javascript. it's somewhat ominous. what does this say about the future of writing web apps?

implications? here's a quick brainstorm:

1. the end of markup? is the swing metaphor for putting together an app winning over the markup metaphor?
2. validates the approach of frameworks such as echo

here's a bigger one:

what does this do to struts and the other umpteen mvc java web frameworks out there? is gwt the death of the mvc java web framework era?

what does this do to spring?

observations?

• it's nice to see that some of the great features introduced by RoR are being used / adopted by other frameworks. i'm referring to gwt's scripts projectCreator and applicationCreator

i think gwt is terribly cool. its introduction is a terrific step forward.

stop for a moment and contrast the approach taken by google vs, for example, yui by yahoo. yui is very good. it doesn't attempt to tackle the whole ball of wax. it does ajax and has really no dependencies on any other parts of the stack.

google's solution is a much much much more involved one. they build a java to javascript compiler (!!!) to make this work. they've got a customized mozilla web browser for testing this thing. yet, the solution remains simple, from the point of view of the developer who has to construct a web app.

what a testament to css: of all the technologies that gwt swallows, css remains intact.

the introduction of gwt is the first step.

here's what i think the next step should be:

• provide a rich set/variety of widgets
• ensure that the mechanism for writing widgets for gwt is as simple as it can be, and standard. urge the community to contribute widgets, panels, and more. gwt can quickly grow into an ecosystem; its own platform.

i also have my ideas on subsequent steps but i'll reserve those for another time.