Thread: breaking the java habit...

Fundamentally you should probably think more about the task you are trying to accomplish than the actual code behind the task. There will be a point where you will have to begin thinking about code but that should be simple(r) since you already have a design. So if you find yourself thinking in Java then I'm afraid if you use C++ you will also find yourself thinking in C++ instead of thinking in concepts. Fact is you don't think in any language when you are approaching a specific problem. This will keep you concentrated on how to approach and solve the problem more than how you will code A with B.

Once you have your design and approach then you could pretty much code it in any language. At this point you will have to wrap your brain around C++ a bit but it should not be too hard - especially when the compiler starts yelling at you for this or that. Once that happens you will get a good feel for what is allowed in Java and what is allowed in C++. I don't think the adjustment should be that difficult. And your Java knowledge will not be 100% useless in C++ so I agree that you shouldn't throw out the baby with the bathwater here. Think of it as both and instead of one or the other.

Your problem as I see it may be you are thinking in relation to a specific language or a specific implementation rather than a general overall view of the underlying concepts.

Thanks everyone for the input.
So it looks like the general consensus is not forget to java, but remember that C is not Java and focus on the on design.

Although I dont see it as being useful outside web applications.

Alright... well... basically my entire Ph.D. project is written in Java. I wrote an english syntax parser, and vector analysis package, and a bunch of other stuff to do natural language processing in it. It has a lot of libraries by which you don't have to re-impliment a lot of very useful data structures (I actually wrote a tree-map, but then scratched it after I realized that java already had it! such wasted time but at least I learned something). It was good because I wanted to focus more on dealing with the grammar, and the algorithms, and not nitpick around making data structures for everything.

Also, it's cross-platform compatible so you can run your code on your windows box, but also on a unix box.

So, yeah...!

Oh, and before anyone mentions anything about performance, my simulations take six days already, so it doesn't really matter. I swear, the other Ph.D.'s always .......... about their sims taking half an hour, or like two hours, go get some lunch and come back, try a week... pfft!

So... yeah!

Oh, and before anyone mentions anything about performance, my simulations take six days already, so it doesn't really matter.

Err ... now if your simulation took a minute, then I'd say don't worry. But if it takes 6 days, then even making it run just 1% faster would shave a full hour and more off.
The longer something takes, the more you should worry about performance.

Well, that's true that the longer it takes the more the percentage-wise improvement would help. However, there are multiple modules to the project, so usually when I'm running simulation on a particular part, I'm coding or improving other parts, or writing new code for something else. If I were to re-write this in C++, it would probably take more work, though I have re-written this entire project from the ground up several times.