A Space to Think...

First let me say that I totally agree with anyone who states that Java != J EE. A truer word (probably) cannot be said. But the fact remains that a language is only as useful as the things you can *do* with it.

With that as a given, I have been and still am a strong advocate of Java. I've been happily in the Java camp for the past 10 or so years, right since pre-release stuff when things were bantered around as Oak. But that doesn't really matter. The fact is that I fell for Java because I could do things with it that could then be portable across other platforms with minimal work. After all, I'm a lazy developer-in the best possible sense-and I'd rather be developing new or improved code rather than wasting huge amounts of time getting things ported around.

So this now leads me to why Rails is attractive. Contrary to what I've heard, Rails is being used and readied for production environments; I've even got a couple of myself sorted out. Admittedly, things are not as easy there as they should be, but I don't think an estimate of four years to get the framework souped up and ready for prime-time is accurate. Months is a better estimate in my opinion, because things are closer than you think.

But I am getting off the point, the reason I fell for Java was because I could actually get things done in it. With as little pain as possible. I could concentrate on actually developing rather than the points of pain that came with other languages, including Java and J EE because I actually evaluated those first.

But with J EE 5, and J2EE in the soon-to-be-past, I am simply not finding things as easy to get going with. I am absolutely certain that if I was willing to spend a week getting things going in Glassfish then I could have come up with an article that showed you how to develop services using Glassfish. I could have gone into details on how to set up the config files, get the deployment descriptors working and then get the whole service up and running to be consumed by .NET. This would have probably resulted in a series of articles, but it's technically as doable as the Rails solution. ^{(In fact, I am planning just such an article for SOA Ranch, because the more articles that are out there the easier it'll be to get going with Glassfish to get things done)}

The difference is in the speed and ease of development

I've no doubt that J EE 5 and its ancestors have some amazing must-have features buried down in those descriptors(-from-hell as I find myself calling them because more pain comes from these things than anything else) and there's no doubt that for some environments those facilities are very important. However, I had a simple remit: Create a web service as quickly and easily as possible with open source tools that can then be inter operable with .Net and any other web service consumer.

That's it, period. Did you notice the important thing there ;) Course you did. "As quickly and easily as possible" was my reason for going for Rails. And yes, I do believe that the Rails target is going to be just fine production-wise in the near future.

If I'm wrong and Rails drops by the way-side then the time it takes me to create an application in Rails is going to be about the same amount of time it takes me to develop just a small part of a J EE solution (even though things are getting less complicated, this is all about real simplicity and getting things done!). And that's not because I am particularly good with Ruby or Rails[1], things are purely just simpler.

But that's where the true agility come in, and the true reason that I'm getting on with Rails so well. In Java, I could get things done quickly and easily 5 years ago. With J2EE and J EE 5 there is a big gap between wanting to get something done and actually kicking it out the door. With Rails I can get things done quickly because they are actually easier. And that, partnered with a platform that is going from strength to strength, is enough for me to get serious about Rails.

Bottom line: I wanted to create and deploy a web service that had an automatically generated user interface and that could integrate with an HTML application that I had also speedily created in Rails. I also wanted to show that the service could be consumed by a .Net application; after all, the main purpose with services is to allow people to use them as easily as possible.

Java simply wasn't up to the job, not in the time scales, and I am a pro-Java guy. This is where things have to change. C'mon Sun, make Java the best and easiest choice for us when it comes to web services and web front-ends. That's the challenge, and I won't rule Java out while those smart guys in the community could still make that happen...

[1] Quick caveat. I actually prefer the Java programming language over Ruby. It's personal preference, so please no flame war, but for me Java is a great language that is being let down with overly complicated answers to today's questions. Questions such as "I want to create and run up a web service" should be doable in about 10 lines of code. With J EE 5, when you factor in the descriptors, this is not going to be the case even for the most simplest of requirements. The only saving grace is I reckon that the tools market will probably grab the opportunity to *make J EE 5* easier. It's just a shame (although probably part of the commercial model for Java) that tools have to come to save the day at all when that work could be done as part of the J EE 5 offering.