Refactor JVM native code in terms of Jace?
For anyone who has taken a look at the JVM native code, it is clearly a mess of spaghetti code :) Here is why I think we should refactor in terms of Jace...
- Far more readable code
- C++ instead of C
- Better error checking. Currently we don't always check for errors when invoking JNI functions. Jace does this for you automatically.
- The most famous and important catch all: reduces cost of maintaining code
- Slight runtime overhead since we're interacting with JNI through an extra layer. But, from personal experience, this overhead is negligble as compared to the native computation itself.
- Overhead of Jace DLL (140k)
- Jace is a mature framework. It has been around for ages and is proven to work well.
- Jace is free and does not impose a restrictive license.
Personally I'm a huge fan of Jace because I've seen what a difference it makes when refactoring my own code in terms of it. The result is so much more readable and easier to maintain.
So what do you think?
PS: A Jace update release is due in the near future. So the project is still very much alive.