Has anyone tried BlueJ?

It looks really interesting. It is a Java development environment, but it starts in a UML-like class diagram mode for defining classes, properties, methods etc. and is much more visual than traditional development environments.

If it was strict UML, it would be a very nice tool, but from what I've read so far, it looks like it is only pseudo-UML.

http://www.bluej.org/

Looking at the documentation, you can interact with specific classes and methods etc. in the development environment, to test them, without having to build the whole project. It sounds very interesting. I'll be downloading it later to have a closer look.

It's widely used in education to teach OOP, and it's complete sh*t.

I mean really, really sh*t.

I would agree and disagree with Nick at the same time

I started learning Java using BlueJ, and for the way I was taught (writing UML first then going to do code later) BlueJ was an excellent starting tool. Plus there was an add-on which animated the creation of objects so you could see how you program was running visually.

For larger projects, I'm afraid BlueJ doesn't really cut it and trying to find a UML with code creation plug-in for Eclipse would probably be a better bet...

Thanks for the feedback.

If you weren't using UML, what were you using to design the code, before starting to program?

This last year they tried getting people used to Perl before getting them started on Java. I think they used flow-charts to design the program.
I know a friend in Bournemouth who was never taught how to design a program...So I'm not sure how common learning to design is...

How the frack can you write a program, if you don't design it first? You need to do some sort of design and modelling, before getting started on the code, if it has any chance of being useful at the end of the process...

I've had to rescue so many projects over the years, where little or no design was done, before they started on coding a solution. It is pitiful, that students aren't being taught the basics of programming, just told to get stuck into the coding - probably the part of a project which, generally, requires the least time. Analysis, design and testing probably account for 80+% of a typical project, the coding generally less than 10%...

It depends, I thought it was a little heavy writing out UML diagrams just for a "Hello World" type program, but that's how I was taught...
Much of the work I did on my own has very little design, where possible I prefer a type prototyping where I code something that does what I need, then I reimplement the project with good coding practices. It doesn't work for huge projects, but smallish projects it works fine.

Design is important, but must remain flexible. In this last year we were given a simple programming assignment, a week or so in we were given is different brief. The new brief described a very similar program which did most of the things the first brief said, but missed out some features and added in some other ones. A similar situation happened a week or so later.
Many people who had cracked on and done the assignment had to completely re-do it, many people hadn't started yet and could work in the changes as they went along. The assignment taught us to be flexible, but warned that in many coding houses flexible design doesn't really happen.

We had to buy the "Objects first with BlueJ" book in first year to get exercises out of it. Java was our second language after Python, which was used to teach basic concepts, and personally I really disliked BlueJ. In our lectures we'd been using UML to took about OOP, and it just wasn't necessary in BlueJ. I much preferred to use a text editor to write my java in, and design by whatever other means I would normally (typically a scribble pad and pen for small-ish problems - combined with OmniGraffle for bigger programs). I think my view of it may be tainted by the fact that I had to shell out for the book in order to complete 4 or 5 of our weekly labs.

(Amended for accuracy)

As opposed to 5% analysis, design and testing for a student project...

Back on topic, I never did get into the world of Java

That's not how we run projects!

My little experience of "real life" programming (in house) has shown me that why bother spending money and time designing, we'll just apply hack upon hack upon hack!

Then spend 3 times longer and more money developing it

Designing it first means it only takes twice as long

Mind you I'm going to push to get the people I work with to start using comments, currently it's dire (but I want to wait a bit first)

I know. And I arrived when the guy who first hacked this bit of software turned around and said "enough is enough! The next release is going to be a rewrite and is not going to be a huge collection of hacks". After two years I believe they are still waiting for that "stable" version

As to comments, I firmly believe that it doesn't matter how much design and how many documents you have, code without GOOD comments is like navigating a plate of spaghetti...

I was taught to use JavaDoc comments, blocks of comments at the beginning of each class and every method detailing what the class/method does, how and why. This could then be used to compile a set of documentation that is in a consistent style of the Java API documentation.
Much of the real world code I've seen (e.g. OpenSolaris, Mozilla, Adium) doesn't (or didn't) do anything like this!

I can show you an MS-BASIC financial information gathering tool, which was written by FORTRAN programmers and maintained by COBOL programmers, never had a spec and was just hacked every quarter, as new features were needed... They had never heard of a for...next loop, let alone while...wend, it was all add 1 to count and go to statements - and half the go tos were computed!

It got so big and unwieldy, I tried removing some of the code which had been commented out years ago, the program stopped working, because the computed go tos were jumping into the middle of commented out blocks of code!

On another occassion, we had a suite of Lotus 1-2-3 spreadsheets, which were so big (about 2MB, with around 60 worksheets linked together, we are talking DOS versions here, so they were huge, considering the PCs had 512KB memory!). The system started doing unexpected things, but it was a hack built on a hack, added to a hack... Debugging it, everything worked, letting it run, it didn't. In the end, we sent it to Lotus to look at, their reply? "1-2-3 was never designed to something so complex, please re-write it in C!"