Tapestry

There are some scripting language for fast web development emerging these days. However, there is still lots of people stick to the Java web programming because of its great collections of library available.

It may be feasible to develop web site by those scripting language when time is the FIRST priority, but for some apllication, it is still better to use Java as the development language.

If you happens to choose to use Java as the language to develop your website, Tapestry is a great tool to help you develop a website which is robust & easy to maintain.

About 3 years ago some developers in my shop, including me, had been working with JSF 1.2 for about 18 months and were waiting for 2.0 to come out. Meanwhile, other developers had been using Tapestry 4 for a couple years. Nobody was really thrilled with either, and we were all trying to decide whether JSF 2.0 or Tapestry 5 made the most sense for our future projects, or maybe something else. I was leaning heavily toward JSF for all the usual reasons -- standards, available books, experienced developers, lots of component libraries, etc. -- but I decided I had owed it to the guys I was arguing with to at least learn what Tapestry 5 was like. So I began working my way through the main Tapestry tutorial.

What I found there was a revelation. Tapestry 5 was worlds better than Tapestry 4, and it made JSF 1.2 feel like a broken toy. Tapestry was much easier to learn than expected, and best of all, it actually makes it *fun* to code again. I can't remember developing with JSF *ever* being fun. Live class reloading, Post-Redirect-Get, bookmarkable URLs, clean and simple conventions, easy AJAX, XML-free, with a great built-in IoC layer that makes everything little thing tailorable, the whole thing really comes together nicely.

Meanwhile, in the last 3 years, Tapestry has gotten even better -- but so has JSF. The JSF 2.0 improvements, particularly the inclusion of Facelets (which, by the way, was reportedly inspired by Tapestry 4) and the elimination of most XML configuration, have gone a long way into turning JSF into a *reasonable* choice. But it still isn't fun. And it still feels constrained and contrived, not elegant, not the way Tapestry is. I know lots of people use JSF, and some of them even like JSF, but hardly anybody *loves* it. Hardly anybody *enjoys* it.

Everyone needs to evaluate these frameworks for themselves. If you are just reading bulleted lists of features and trying to make technology decisions bases on the strong opinions you find in web forums then you're just gambling recklessly. Try the out these frameworks and see what fits you and your team the best. Each of them has a simple tutorial or quick-start. Don't cheat yourselves by assuming you can divine the real truth by reading some web site descriptions.

Well, in the end, I liked Tapestry 5 so much that I felt I ought to contribute back -- mostly in the form of documentation improvements at first, and then some patches. And now I have actually become a committer for the project. Hey... you should only spend your personal time on projects that really, truly, deserve it ...right?

Tapestry means a lot of things. Tapestry 4 is a different beast to Tapestry 5. Tapestry 4 is stable and mature and reasonably well used.

Tapestry 5 is fast changing, in beta and a tad unstable. Neither are strictly speaking enterprise ready.

The big question is why would you? When there is Spring MVC, Wicket, JBoss Seam why would you

A) Write a competing framework on your own?
B) Use such a framework?

I think as a framework it is OK. It has some nice ideas though the implementation leaves you jumping through hoops if you want non-default behaviour.

The Docs are poor. They cover a lit of ground conceptually but with very few examples. I get the impression they were written in a hurry by an engineer.

The community is small but pretty helpful as long as your question doesn't point up a deficiency in the Tapestry 5 design.

In general I would say IF you are looking for a framework to Proof Of Concept or prototype a Java Web App. then try Tapestry 5. The risks are just that it is unproven and there are few developers. Kudus to Howard Lewis Ship for seeing a gap where there is none and filling it with a framework that does a little less than the alternatives.

If you want fast dev and convention in Java use Groovy on Grails. If you want enterprise standards use Jboss Seam of Spring MVC. Or try this. It is intriguing and may well grow into a good 'un IF the project team let more people commit useful code.