I have written an application in Java using Eclipse IDE and I now need to know the minimum JRE version that is required to run the application! I know that certain methods are only available under later JREs, but I was wondering what the easiest way to find out the highest requirement of my application would be, so any suggestions would be appreciated...
Also whilst I am on the topic of requirements, I would appreciate any advice or methods for determining the minimum system requirements for my software in general - i.e minimum amount of RAM...
Thanks in advance
As far as RAM, that's easy. When the JVM starts it sets a 'maximum' amount of RAM (I believe the default may be 128MB), and that's a hard limit that your application cannot exceed without crashing. Profile your app over time, tweaking the memory settings on the JVM, and find out what the minimum amount of RAM is that you'll need for your app to run both (a) with acceptable performance, and (b) without throwing an OutOfMemoryError, and you're done.
Ref: How to configure JVM options and memory?
For other requirements such as CPU req., things get a little fuzzier. There are a lot of CPUs out there, and the throughput that a given system produces can vary not just based on CPU speed, but the speed of the hard drive, the amount of RAM installed in the system, the speed of the network interface (if you're writing a network app), and other things. For requirements such as that, you'll want to just test it on a variety of systems and sort of draw a line somewhere, and say, "You can expect acceptable performance if you have hardware that is at least as powerful as X, Y, Z".
The other thing you could do is build in a benchmark, or some kind of performance logging, and have that performance data sent back to you. Lots of apps do this. You know that "May we send anonymous usage data back to the mothership?" question you get when installing some software? Well, common among that data are system-specific details such as RAM, CPU, hard drive model, and other hardware details (whatever data you determine is relevant to your app), along with performance logging data. By taking that kind of approach, what you get is a lot of performance data from lots of different system configurations without needing to have a huge number of differently configured machines in-house.
You can do the same thing for program crashes and bugs - have the stack traces, system info, and other relevant data dumped to a log file that is sent back to you - but of course, only if your users have said it's okay to send that data back to you.