Translating call-by-reference from Fortran

Question

I have been asked to translate some Fortran code to Java with minimal redesign.
I believe Java uses call-by-value and not call-by-reference, right?
What are my best options for translating the many call-by-reference subroutines and functions?

Answer 1

If you insist, then you can wrap objects up in a sort-of container, which can get you PBR-type semantics, but it's clunky. Eg
class StringRef  {
  private String payload;
  public String toString() { return payload; }
  public void setPayload(String string) { payload = string; }
public void passMeAString(StringRef string) {
  // do stuff;
}You can reach into the StringRef object and change it's effective value. But then all code dealing with it, whether it needs PBR semantics or not, is either coupled to this StringRef class, or is peppered with code to lift the actual string in and out of StringRef objects. Or you could deal with single-element arrays everywhere, which isn't really any better.
Messy, a maintenance nightmare, a haven for bugs. I'd really be looking at solving the problem without using pass-by-reference, if you don't want to have a nervous breakdown.
Another thing to consider is that simply shoehorning your existing Fortran through a Java compiler means you're also inheriting all the bugs and the convoluted messy code from fixing bugs that you'd probably rather not. I seriously wouldn't try to do this, just infer the requirements from the Fortran code and start again. You might well even find that there already exist Java libraries that can replace some of your code.