Choose 'better' or more familiar technologies for a new project?

Question

I am looking to start work on a brand-new project, something I've been thinking about for a while as my first independent sellable project. It's broadly speaking a web-based service application, and my first choice, server-language is quite easy... I know Java pretty well from working on Java web-apps in the past.

However my experience doing web-apps involved JSP, Servlets and JSTL... I know the ideas behind newer technologies like Hibernate/Spring but have never used them. So we wrote our own DAOs, handled AJAX by writing special mini-JSP pages that generated XML/JSON pages, etc.

I'm not hugely into the idea that Spring/Hibernate are the 'only' or 'right' way to do any Java web-project, but they are widely used. On the other hand, not only would trying to learn these increase initial development time, but I'd be using my learning attempts to build a production system.

I remember one of Joel's early articles said (I'll paraphrase since I can't find it)

"regardless what's cool, always use the technologies that the lead developer (or dev team?) knows best"

I wondered what people thought about that?

ps: should this be CW?

Answer 1

Don't try everything all at once - take on one new technology at a time.

Answer 2

It depends what the customer wants (in the world of consultancy). You have to learn new technologies. Does the customer wants to pay for that?

Not all the caveats of the new ones are known, whereas the older ones are proven a lot more.

Of course, if everybody thought like this we would all be stuck with VB these days. You have to look for the right balance, and learn a lot yourself too so you can get an objective view on the technologies available, their up and downsides.

Answer 3

i personally would definitely recommend looking into spring, i've found it's saved me countless hours. hibernate is also useful if you need an ORM layer (and spring has nice integrations with hibernate too). they're certainly not the 'only' or even the 'right' way to do things (that's quite subjective) but they have both saved me time and effort, especially spring.

Answer 4

Beware the lure of cool new frameworks! I'm currently hacking on a tiny little web app that just has a login, a few mostly static pages, and a few forms to request some information by email. It would have taken me maybe two days to do as traditional Servlet/JSP in MVC style. Instead, since there was slack in the schedule, I decided to use this project to get up to speed in Spring, Spring MVC, and Spring WebFlow. While it's quite possible that I'm just dense, it took me several weeks to get my head around the right way of doing things, I'm still not totally confident that I'm doing everything correctly, and the application is still not done. Fortunately, due to slack, I'm not in danger of the overall project schedule slipping, but I'm always asking myself if I'm going to have to scrap it and start over.

I have learned my lesson, though: next time, I won't be the one pushing a new framework unless its one I've used for production projects before. That said, I'm glad I now understand Spring (or at least I think I do) and will not hesitate to use it again next time.

So how would I learn a new framework next time? If there's a project lead (in this case I'm a project lead of a team of one, no help there) I'd use the framework that they put in place. If there isn't, or if I want to learn a framework that the project lead isn't using, I'd use it for a side project on my own time. Learning is good. Putting company work at risk by throwing untested technology at it is not so good.

Answer 5

There is one major trap with any unknown technology. You do not know where the dragons are, and you do not know how to rub the new technology "with the hairs".

Learning that will take time, and you need to have that in your estimates. Also your estimates will most likely be too low...

Answer 6

I can say for sure that Spring is worth considering. It gives you as much as you can take, but it doesn't bother you with things you don't need.

For example, at the very beginning you probably need dependency injection only. Then you'll need help with database interactions and transaction management. Then you'll decide to apply MVC patter to your web-application. After that may be you'll realize that components of your system are to send JMS to each other. And so on and so forth.

For all this cases Spring has it's own simple, intuitive, light-weight solution.

Answer 7

I work as a consultant, and I've seen a lot of projects where the devs started out with servlets+JSP because that's what they knew, and it's pretty simple to get started with. However, it gives the team an opportunity/excuse to write a platform of their own, which is more fun than using someone else's and just writing an application.

As the project grows, the team reinvents more and more wheels, quite a few of which end up square. That's where I enter the picture - adding new stuff to this semi-flexible platform has become so complicated that the devs can't keep up adding features and fixing bugs without calling in reinforcements. Just to add insult to injury, the internal devs are usually the ones who get assigned to do the boring bug fixes because bug fixes require more knowledge of the gory entrails of what has become the team's proprietary persistence-and-web framework, and so those gosh-danged consultants get to do the new, fun stuff.

Now, you shouldn't use a framework just because people have been regurgitating each other's blog posts about the awesomeness of it, but you should also realize that there are very good reasons why those frameworks exist (and why they're used). If you haven't used any web frameworks at all, I'd recommend you to take Spring MVC, Wicket or whatever for a test drive. They don't solve all problems, and they do cause some of their own, but the grand total is usually a productivity increase, especially if you're making advanced user interfaces.

I have been on projects where plain JDBC has been quite sufficient for persistence, and where no more advanced web frameworks than servlets+JSP have been needed, but those projects are a minority. Without having used a framework or two, you'll never whether your project is part of that minority that doesn't need one, or if it is part of the grand majority that does.

Answer 8

When starting a new project limit the number of unfamiliar technologies / frameworks to use. Every framework takes time to learn and every framework has issues especially if not implemented correctly.

If you can I would recommend you look into the Play framework. It is a web framework for Java that focuses on developer productivity. You can choose to use Spring / Hibernate if you want but you are not bound by that. It has a very easy to learn implementation and you should be able to get a good idea within a day of playing around with it if it is what you are looking for.