[Checkers] Javari checker possibly ready for release

[Matt Papi]

Telmo,

Thanks, that's good news!

I'll take a look at the Javari checker and let you know later this 
morning if there's anything that urgently needs to be done before the 
release.

- Matt

Telmo wrote:
> After this last night of debugging, the Javari checker passes the 3 basic 
> tests; this version has been checked in on the repository.
> 
> The last changes refer to "RoMaybe variables"; while it is not clear to me 
> that they are added on the specification, they must be supported, such 
> that variables may be passed around as follows:
> 
> @RoMaybe Node foo(@RoMaybe Node arg) {
>     @RoMaybe next = arg.getNext();
>     // print information about next
>     return next;
> }
> 
> Since those checks may depend on context, if this-mutable variables are 
> thrown in the mix, some assignment of annotations to variables must be 
> done with dependencies on the VisitorState, which keeps information about 
> the receiver and class type. However, none of the forms of 
> commonAssignmentCheck accepts AnnotatedTypeMirror parameters both as 
> variables and values, so the functionality of commonAssignmentCheck has 
> been partially reimplemented on the overriding method at JavariVisitor, 
> for now, and no calls to the super method are made.
> 
> Constructors still do not behave properly with RoMaybe; I tried touching 
> the code that deals with it (visitNewClass, on JavariVisitor, is commented 
> out and has a stub functionality active at its end), but the old code 
> causes many tests to fail when activated. I don't understand yet what is 
> happening there, so I do not plan on changing it before the release today.
> (In fact, unless told otherwise I don't plan on touching the code again 
> before the release.)
> 
> -Telmo
> 
> _______________________________________________
> checkers mailing list
> checkers at lists.csail.mit.edu
> https://lists.csail.mit.edu/mailman/listinfo/checkers