About Temp-Tables, a memory database, such as H2, could be used in order to reproduce TT fast access.
The point is not "can you imitate this in Java?" but, what is the best way to utilize OOABL? I.e., learn from standard OO best practices, but just the same way that one does things a bit differently in Java, C#, and C++, so should you do them differently in OOABL. Otherwise, you are just doing a sort of worst-case line-by-line translation and that isn't going to give you good code.
I managed to recreate a mechanism corresponding to shared variables...
Where is the smiley? Surely you aren't serious about wanting to imitate one of ABL's most unfortunate residual features? Do you write your Java code with public data members too?
And for the moment, I am not stucked with a syntax without matching in Java...
Stuck? How stuck? Seems like what you have going here is a way to write ugly ABL *and* ugly Java. This is not a virtue.
Exactly. The only way the "FIND FIRST" worked was when I did not specify a condition...
No, what I was suggesting as a test to make sure you are seeing what you are thinking you are seeing is to define a local variable for the return value of each of the get methods, assign the local variable using the method, *then* use the local variables in the FIND and, presumably, also display them so that you see what you actually have. I am not recommending this as best programming practice, but rather just as a diagnostic to make sure the problem you think you are seeing is actually the problem and not something else.
In fact, I also got a clue when Architect refused to compile my code because of a type mismatch between a date-type field and one of the getters, supposed to retrieve a date too......
Which is a clue that something isn't as you think it is ... not a clue that there is some kind of deferred execution going on.